University of Southern California Dissertations and Theses

The independence of binding and intensification

(USC Thesis Other)

The independence of binding and intensification

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE INDEPENDENCE OF BINDING AND INTENSIFICATION VOLUME I by Uffe Bergeton A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (LINGUISTICS) May 2004 Copyright 2004 Uffe Bergeton Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3140435 Copyright 2004 by Bergeton, Uffe All rights reserved. INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. UMI UMI Microform 3140435 Copyright 2004 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dedication Til mine fo rsid re Henning A. V. Larsen og Inger J. Larsen Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ill Acknowledgements This dissertation is built on a large body o f accum ulated insight into the workings o f hum an language in general and reflexives and intensifiers in particular found in the very rich literature on these topics. W hile I have tried to give credit where credit is due I fear that, in many cases, I may have failed to explicitly state in w hat respect and to w hat extent I depend on the research o f others. M y hope is that these om issions may be forgiven. In addition to the existing linguistic literature I am also profoundly indebted to a large num ber o f people who have provided me w ith invaluable support and assistance, both personal and professional, throughout the process o f w riting this dissertation. Since this acknow ledgem ent section is too b rief to adequately express m y gratitude to everybody, I hope that those not m entioned here will be able to forgive m e and trust that the lack o f m ention should not be translated into lack o f gratitude. N eedless to say, while all o f the people m entioned here have contributed to the dissertation in one way or another, directly or indirectly, and share the credit o f the better parts, only 1 am responsible for any inconsistencies, errors and other shortcom ings it may still contain. First o f all I’d like to thank my faculty advisors at USC, R oum yana Pancheva, Philippe Schlenker, and Jean-R oger Vergnaud. Beginning in chronological order, Jean-R oger V ergnaud has been an invaluable source o f both academ ic and adm inistrative support ever sin ce I ca m e to U S C . S in ce he g u id ed m e as facu lty ad visor during the initial sta g es o f m y research the im pact o f his input cannot not be overestim ated. Philippe Schlenker, who was my advisor from 1999 to 2001, and co-chair o f my Ph.D. com m ittee from 2002 to 2004, has been pivotal to the developm ent and refinem ent o f my ideas. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV W ithout his sharp reasoning m ost o f the ideas in the dissertation would still be entangled in conceptual confusion and unclarity. Roum yana Pancheva, who accepted the task o f being my com m ittee chair after Philippe left USC in 2001, has been no less im portant in helping me turn vague intuitions into coherent and explicit analyses. The late-insertion analysis o f reflexives in chapter 4 and the analysis o f the evolution o f English reflexives in chapter 7 were developed in close collaboration w ith her. M any other parts o f the analysis o f both reflexives and intensifiers were worked out during our many m eetings over the last few years w hich have, w ithout a doubt, contributed the m ost to m aking the final product into a readable and fairly coherent whole. In addition to my advisors, a num ber o f other faculty m em bers at USC have also had a large im pact on m y work. A m ong them are, o f course, m y Ph.D. Com m ittee M em bers, Joseph Aoun, Jam es H igginbotham , Audrey Li, and Edward Slingerland. I thank them for their helpful com m ents and encouragem ent and not the least for having had the patience to read through earlier drafts o f the dissertation. Special thanks also go to Jack Hawkins, Barry Schein, and M aria Luisa Zubizarreta for being in m y screening paper com m ittee and for their classes at USC w hich have contributed to form ing my thinking as a linguist. Finally, 1 also w ant to thank Jean Lowenstam m , m y faculty advisor at the U niversity o f Paris VII for his instruction. W ithout his suggestion that I apply for graduate school in the Unites States I w ould never have com e to USC in the first place. A nd w ithout his inspiring approach to the syntagm atic study o f the sound structure o f human language (i.e. G overnm ent Phonology) I w ould never have decided to pursue a career in formal linguistics. The student com m unity in the D epartm ent o f Linguistics and m y circle o f friends (including Shadi Ganjavi, Aaron Sonnenschein, Junichi Hayashishta, Barbara Lohse, Karine Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M egerdoom ian, Lina Choueiri, and many others) have provided the social setting w ithout w hich I would not have thrive in Los Angeles. Last but not least 1 am deeply indebted to m y girlfriend Iris Yim for having had the patience to live w ith a graduate student w orking on his dissertation. W ithout her unfailing support and encouragem ent 1 could never have com pleted this work. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VI Table of Contents Dedication ii Acknowledgements iii Abstract xvi Volume I Chapter 1 Introduction 1.1 The failure o f m ost current binding theories to correctly 2 separate binding and intensification 1.2 Independence o f binding and intensification in Danish 5 1.3 Revisiting the analysis o f sim ple and com plex reflexives 9 in D anish 1.4 Adnom inal intensification 12 1.5 Predicate m eaning and intensification o f reflexives 15 1.6 Intensification and the m eaning difference between o f 18 sim ple and com plex reflexives 1.7 Logophoricity and intensified pronouns 20 1.8 Possessive reflexives, pronouns and intensifiers 22 1.9 R efuting anti-locality 26 1.10 Self-nom inalizations 30 1.11 U nification o f all uses o f adnom inal expressions 30 1.12 Extending the analysis to M odern English 31 1.13 Historical evolution o f English reflexives and intensifiers 34 1.14 Extending the analysis to M andarin Chinese 35 1.15 C onclusion 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Vll Chapter 2 Intensification 2.1 Introduction 41 2.2 A dnom inal intensification 45 2.2.1 The m orphology o f adnom inal intensification 45 2.2.2 The syntax o f adnom inal intensification 48 2.2.2.1 Possible syntactic positions o f adnom inally intensified 48 nom inal expressions 2.2.2.2 The syntactic relation betw een intensifier and its focus: 50 adjunction 2.2.2.3 U nified analysis o f adnom inally intensified expressions 54 2.2.3 Sem antic and pragm atic aspects o f adnom inal intensifiers 55 2.2.3.1 The sem antic contribution the adnom inal intensifier selv 56 ‘h im se lf 2.2.3.2 Sem antic and pragm atic constraints on adnom inal 60 intensification 2.2.3.2.1 U nique identifiability 60 2.2.3.2.2 Prom inence, centrality or contrastiveness 62 2.2.4 Is adnom inal intensification o f PRO possible? 67 2.2.5 Selv used as q-floated adnom inal intensifier 72 2.2.6 Fronted selv 85 2.2.7 Adnom inal intensification: unification o f different sub- 88 cases 2.3 S e lv in secon d ary predication con stru ction s 89 2.4 5e/v used as scalar additive focus particle m e a n in g ‘even’ 91 2.4.1 Syntax o f the scalar, additive focus particle selv ‘even’ 95 2.4.2 Sem antics o f the scalar, additive focus particle selv ‘even’ 97 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Vlll 2.5 5'e/v used as a noun 102 2.6 Adjectival form s o f selv. selve and selveste 104 2.7 & / v ‘s e lf in nom inalizations 108 2.8 Idiom atic uses o f the elem ent selv 109 2.9 C ross-linguistic perspective on the range o f uses o f 110 intensifiers 2.10 C onclusion 111 Chapter 3 Binding: Reflexives and Pronouns in Modern Danish 3.1 Introduction. Binding o f reflexives and pronouns in 115 Danish 3.2 Basic properties o f Danish reflexives 119 3.2.1 M orphological characteristics o f reflexives in Danish 119 3.2.2 Sem antics o f reflexives 121 3.2.3 Syntactic structure o f reflexives 124 3.2.4 C onsequences o f relegating binding and intensification to 129 different m odules: a syntactic approach to binding o f reflexives: principle A 3.3 Binding o f reflexives in different contexts: testing the 141 independence o f binding and intensification 3.3.1 The independence o f intensification and locality 141 constraints 3.3.2 Locally bound reflexives in argum ent position: com plex 146 reflexives = a sub-type o f adnom inal intensification 3.3.2.1 A nti-reflexives predicates: predicates which are 148 incom patible with reflexive scenarios 3.3.2.2 Predicates w hich are sem antically incom patible w ith non- 153 reflexive scenarios 3.3.2.3 N eutral predicates: predicates w hich are sem antically 155 com patible w ith reflexive scenarios and w hich are not sem antically incom patible with non-reflexive scenarios Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX 3.3.2.4 Summary: Presuppositions triggered by predicate 159 m eaning 3.3.2.5 Com plex reflexives as argum ents o f proto-typical neutral 160 predicates 3.3.2.6 U sing inalienable possession to test w hether a verb is 163 neutral or anti-reflexive 3.3.2.7 Linking doppel-ganger-effects and anti-reflexivity 166 3.3.2.7.1 & /v as m arker o f statue-readings 167 3.3.2.7.2 Q ua-sentences 168 3.3.2.7.3 Strict readings o f reflexives in VP-ellipsis 171 3.3.2.8 Com plex reflexives and distributivity 173 3.3.2.9 Sum m ary 175 3.3.3 Reflexives in resultative constructions 176 3.3.4 Reflexives in ECM constructions 179 3.3.5 C om plex reflexives and pronouns in possessor position 180 3.3.6 Binding o f sim ple and com plex anaphors in PPs and 185 prepositional predicates: phonological factors affecting intensification o f reflexives 3.3.6.1 Stressability o f reflexives: prosodic uses o f selv = stress 186 carrying elem ent 3.3.6.2 Intensifier-adjunction to reflexives in PPs and the 188 interaction w ith sentential stress 3.3.6.3 C an intensification o f reflexives be reduced to lack o f 203 stressability? 3.3.6.3.1 Stressable reflexives in French 203 3.3.6.3.2 Stressable reflexives in G erm an 204 3.3.6.3.3 Stressable pronouns in D anish 205 3.3.6.3.4 Reducing intensification o f reflexives to unstressability 206 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3.7 Sum m ary 208 3.4 D istribution o f sim ple and com plex pronouns 209 3.4.1 B inding o f pronouns: principle B 210 3.4.2 Intensified object pronouns 212 3.4.2 Intensified subject pronouns: Intensified pronouns as 220 subjects o f em bedded clauses in English and Danish 3.4.3 Logophors and logophoricity vs. intensification and 224 intensified pronouns 3.4.4 Sum m ary 226 3.5 Conclusion: The independence o f binding and 227 intensification Chapter 4 The syntactic status of sig 4.1 Introduction 231 4.2 Testing the lexical/syntactic status o f sig with neutral 237 verbs 4.2.1 Testing for agentivity 239 4.2.2 Testing for com patibility w ith form ation o f agentive 245 nouns in ~er 4.2.3 N on-com positional m eaning change and the question test 247 4.2.4 Testing for com patibility w ith im personal passivization 249 4.2.5 A uxiliary-selection 252 4.2.6 Lodrups tests 254 4 .2 .7 G erm an s ic h is not an ti-local 2 5 5 4.2.8 C onstructions involving coordination o f verbs 258 4.2.9 The status o f sig as indirect object o f ditransitive verbs 261 4.2.10 N on-co-argum ent bound local sig 262 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XI 4.2.11 Summary: Falsifying the standard analyses o f sig as anti- 263 local 4.3 The status o f sig w ith inherently reflexive verbs 263 4.3.1 Inherently reflexive predicates w ith non-them atic sig 267 4.3.2 “Inherently reflexive verbs” with them atic sig 268 4.3.3 G ram m atical uses o f inherently reflexive verbs with s/g 273 selv 4.3.4 C onclusion 274 4.4 Late-insertion analysis o f reflexives and pronouns in 275 D anish 4.4.1 Lexical and syntactic properties o f pronouns and 275 reflexives 4.4.2 Binding theory in L* and 2"“ ^ person: the im plications o f 279 person asym m etries 4.5 Reflexive and non-reflexive uses o f reflexive elements: 281 How m any sig's do we have? 4.5.1 Use o f reflexive elem ents in reflexive constructions 287 4.5.2 Use o f reflexive elem ents in reciprocal constructions 288 4.5.3 The use o f reflexive elem ents in m iddle constructions 293 4.5.4 U se o f reflexive elem ents in passive constructions 295 4.5.5 Deponent verbs 297 4.6 C onclusion 301 V olum e II Chapter 5 Independence of intensification and binding in English: synchronic and diachronic perspectives 5.1 Introduction 304 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XII 5.2 Intensifiers, 0-reflexives and intensified pronouns in 308 M odern English and the independence o f intensification and binding 5.2.1 Binding o f 0-reflexives in different contexts: testing the 309 independence o f binding and intensification in M odern English 5.2.1.1 Locally bound 0-reflexives in argum ent position: 309 com plex reflexives = a sub-type o f adnom inal intensification 5.2.1.1.1 A nti-reflexive predicates 311 5.2.1.1.2 Inherently reflexive predicates 312 5.2.1.1.3 N eutral predicates 314 5.2.1.1.4 N on-optionality o f intensifier-adjunction to 0-reflexives 318 w ith proto-typical neutral predicates: focus and doppelganger-effects 5.2.1.1.5 D itransitives 319 5.2.1.1.6 Linking Doppelganger-effects and anti-reflexivity 320 5.2.1.1.6.1 X -self as m arker o f statue-readings 320 5.2.1.1.6.2 Q ua-sentences 321 5.2.1.1.6.3 VP-ellipsis and the sloppy vs. strict reading o f reflexives 321 5.2.1.1.7 Sum m ary 322 5.2.1.2 Resultatives and the 0-reflexive analysis o f English 324 5.2.1.3 ECM constructions the 0-reflexive analysis o f English 325 5.2.1.4 Adnom inal intensifiers in possessive constructions 326 5 .2 .1 .4 .1 A b se n c e o f in alien ab le p o sse ssio n in E n glish 3 2 8 5.2.1.5 0-reflexives in PPs and prepositional predicates in 329 English 5.2.1.6 Potential problem : stressed and unstressed forms o f 332 him self Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Xlll 5.2.2 N on-reflexive uses o f 0-reflexives in English 335 5.2.2.1 Reciprocals 335 5.2.2.2 M iddles 336 5.5.2.3 Inherently reflexive/deponent verbs 337 5.2.2.4 Sum m ary 337 5.2.3 Intensified and unintensified pronouns in English 338 5.2.3.1 Intensified subject pronouns 339 5.2.3.2 Intensified object pronouns 339 5.2.3.3 Summary: binding o f intensified and unintensified 340 pronouns 5.2.4 C onclusion 340 5.3 H istorical developm ent o f Danish and English intensifiers 341 5.3.1 The historical developm ent o f the G erm anic/M odern 341 Danish intensifier selv 5.3.2 Previous account o f the evolution o f the M odem English 343 self-foxms 5.3.3 A new account o f the evolution o f the M odem English 348 5e^-form s 5.3.3.1 Stage one: intensification and binding in Old English 348 5.3.3.2 Stage two: intensification and binding in Early M odem 354 English 5.3.3.3 Stage three: intensification and binding in M odern 363 English 5 .3 .3 .4 A d d ition al ev id en ce from the ev o lu tio n o f pronom inal 363 reciprocals into 0-reciprocaIs 5.3.4 Sum m ary 364 5.4 C onclusion 367 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XIV Chapter 6 Independence of Binding and Intensification in Mandarin Chinese 6.1 Introduction 369 6.2 Setting the stage: evidence supporting the assum ptions 369 that ziji is always an intensifier and Chinese has 0 - reflexives 6.3 Internal structure o f adnom inal intensifiers in Chinese 380 6.3.1 Is zi }7 really m ono-m orphem ic? 380 6.3.2 Ta ziji: com plex intensifier, com plex reflexive 0 ta ziji, or 383 intensified pronoun ta zijil 6.4 A nalyzing w hat appears to be uses o f as reflexive 387 anaphors as adnom inal intensifiers 6.4.1 Locally bound 0-reflexives in argum ent position: 387 com plex reflexives = a sub-type o f adnom inal intensification 6.4.1.1 A nti-reflexive predicates 387 6.4 .1.2 Neutral predicates 389 6.4 .1.3 Inherently reflexive predicates 394 6.4.1.4 Predicate types: sum m ary 395 6.4.2 LD -bound zijv. com plex pronoun or com plex reflexive? 396 6.5 Intensifier analysis o f local and LD ta ziji 396 6 .5 .1 Locally bound ta ziji: com plex pronoun or com plex 397 reflexive? 6.5.2 LD -bound ta ziji: com plex pronoun or com plex reflexive? 398 6.6 Intensified and unintensified pronouns in M andarin 400 Chinese 6.6.1 Sim ple/unintensified pronouns in M andarin Chinese: 400 Principle B 6.6. L I O verlapping distribution o f simple ta and zi}7/ta zz/7 401 6 .6 .1.1.1 Possessive constructions 401 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XV 6.6.1.1.2 Em bedded subject pronom inals 402 6.7 C onclusion 405 Chapter 7 The element selv ‘seir in nominalizations 7.1 Introduction 407 7.2 D ifferent uses o f the m orphem e selv ‘s e lf in derived 409 nouns and adjectives 7.2.1 W ord-internal selv ‘s e lf appearing to behave as a 409 “reflexive” while it really is an adnom inal intensifier 7.2.2 W ord-internal selv ‘s e lf behaving as an adverbial 413 7.2.3 Problem atic cases 414 7.3 C ross-linguistic perspective on use o f intensifiers in 416 derived words 7.3.1 French 416 7.3.2 Chinese 418 7.3.3 Sum m ary 420 7.4 O utline o f analysis o f selv ‘s e lf in derived nouns and 420 adjectives Chapter 8 Conclusion 424 References 431 Appendix I Danish xe/v-compounds 447 Appendix II Intensified and unintensified pronouns and reflexives in Danish 469 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XVI Abstract A new approach to the interaction o f binding and intensification is advanced. W hile m ost current approaches to (e.g. Reinhart& Reuland 1993, Huang& Liu 2001, am ong others) take the selv elem ent in so-called SELF-anaphors (e.g., Dan. sig selv) to be itself a reflexive, we argue that it is in fact an adnom inal intensifier (cf. Baker 1995, and K onig& Siem und 2000). Thus we achieve a unified account o f all types o f intensified nominal expressions, be they intensified DPs (e.g., Dan. Peter selv ‘Peter h im self), intensified reflexives (e.g., Dan. Peter hader sig selv ‘Peter hates REFL s e lf), or intensified pronouns (e.g., Dan. Peter sagde at Mary dansede med alle andre end ham selv “Peter said that M ary danced with everyone except him self.’) - som ething m ost current approaches to binding and intensification are unable to do. In contrast to predicate-based approaches to binding (e.g. Reinhart& Reuland 1993), we call for a nominal approach in w hich binding is defined as the interaction o f intrinsic properties o f nominal expressions and syntactic locality constraints. We also argue that the Danish reflexive sig (= Dutch zich, N orw egian seg, etc.) is not an anti-local anaphor specialized for long-distance binding, thus falsifying those binding theories which crucially rely on this assum ption. Finally, concerning intensification, we argue that the sem antic contribution o f adnom inal intensifiers (e.g. D anish selv, Eng. himself) is not sim ilar to that o f scalar focus particle (cf. Eckardt 2001) or reducible to centrality-effects (cf. Konig (1997) and Siem und (2000)), but rather very sim ilar to that o f contrastive focus, w hich - depending on the context - may or may not involve scalar ordering o f focus-generated alternatives. The analysis is extended to English and Chinese w here him self and ziji ‘self-self are argued always to be intensifiers (~ Dan. selv) m odifying 0-reflexives Dan. sig), e.g. Peter, shaved 0 , him self We show that this analysis provides answers to hitherto unanswered questions related to the evolution o f M odern English Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XVI1 reflexives and intensifiers from Old English which had a m onom orphem ic xe//intensifier and no reflexive pronouns. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 1 Introduction This dissertation presents an approach to adnom inal intensifiers and sim ple and com plex reflexives and pronouns based on the idea that intensification and binding constitute separate m odules o f the gram m ar. W hile the separation o f intensification and binding into different m odules is hardly controversial, the question o f w here exactly to draw the dividing line is far from trivial, especially in the case o f com plex reflexives and com plex pronouns. It will be argued that m ost current accounts o f binding (e.g. Reinhart and Reuland (1993), H uang and Liu (2001), am ong others) are m isguided in that they often take w hat are really adnom inal intensifiers for reflexive elements. In addition to achieving a descriptively and explanatorily m ore adequate account o f intensified reflexives and pronouns, the approach defended here also has the advantage o f being able to achieve a unified account o f all types o f intensified nominal expressions, be they intensified DPs (e.g. Dan. Peter selv ‘Peter h im self), intensified reflexives (e.g. Dan. Peter hader sig selv ‘Peter hates REFL s e lf), or intensified pronouns (e.g. Dan. Peter sagde at Mary dansede m ed alle andre end ham selv “Peter said that Mary danced w ith everyone except him self.’) - som ething current approaches to binding and intensification are unable to do. Though m ost o f the data is from Danish, in chapter 5 and 6 it will be shown that the analysis presented here can be extended to English and Chinese. Finally, besides proposing a new articulation o f binding and intensification the present proposal also has consequences for each o f these m odules individually. In contrast to predicate-based approaches to binding (e.g. Reinhart & Reuland 1993), we call for a nominal approach to binding in w hich binding is defined as the interaction o f intrinsic properties o f nom inal expressions and syntactic locality constraints. We also argue that the Danish reflexive sig (= Dutch zich, N orw egian seg, etc.) is not an anti-local anaphor specialized for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. long-distance binding, thus falsifying those binding theories w hich crucially rely on this assum ption. Finally, concerning intensification, we argue that the sem antic contribution o f adnom inal intensifiers (e.g. Danish selv, Eng. himself) is not sim ilar to that o f scalar focus particle (cf. Eckardt 2001) or reducible to centrality-effects (cf. by Konig (1997) and Siem und (2000)), but rather very sim ilar to that o f contrastive focus which - depending on the context - may or may not involve scalar ordering o f focus-generated alternatives. 1.1 The failure of most current binding theories to correctly separate binding and intensification The main problem with m ost existing analyses o f binding is that they do not take into account the fact that in m any languages the elem ents w hich serve as reflexives are either identical to the elem ents serving as adnom inal intensifiers, e.g. English him self (1), Chinese |=| (3 ziji ‘s e lf (2), and French lui-meme ‘him -sam e’ (3), or partially overlaps with adnom inal intensifiers, e.g. the Danish com plex reflexive sig selv ‘REFL s e lf (4a) which is com posed o f the simple reflexive sig ‘R E FL ’ (4b) plus the adnom inal intensifier selv ‘s e lf (4c). In the exam ples in (1- 4) bold has been used for expository reasons to highlight the elem ents functioning as adnom inal intensifiers/reflexives’. (1) a. Peter hated him self. (/u'w5'e/f= reflexive) b. Peter h im self attended the conference. {himself = intensifier) (2) M andarin C hinese: a. ° (zz)7 = reflexive) Zhangsan hen ziji. Zhangsan hate self ‘Zhangsan hates h im se lf’ ' See chapters 5-6 for analyses of English and Chinese based on the assumption that him self and ziji li] H ‘self-self always function as intensifiers. Reproduced with permission otth e copyright owner. Further reproduction prohibited without permission. b. ° (zzy/= intensifier) Zhangsan ziji canjia huiyi. Zhangsan self attend conference. Zhangsan him self attended the conference. (3) French: a. Pierre estjaloux de lui-meme. {lui-meme = reflexive) Peter is jealous o f him -sam e ‘Peter is jealous o f him self.’ b. Pierre lui-meme a assiste d la conference. {lui-meme = reflexive) Peter him -sam e has participated to the conference ‘Peter him self participated in the m eeting.’ (4) D anish: a. Peter vaskede sig selv. {sig selv = com plex reflexive) Peter w ashed REFL self ‘Peter w ashed him self.’ b. Peter vaskede sig. (5 /g = simple reflexive) Peter w ashed REFL ‘Peter w ashed (him self).’ c. Peter selv deltogi modet. (.se/v = intensifier) Peter self participated in m eeting-the ‘Peter him self participated in the m eeting.’ From the beginning o f m odem formal linguistics till the present day, m ost binding theoretical accounts o f reflexives have tended to ignore the formal identity o f intensifiers (l-3 b ) and reflexives (l-3 a), or in the case o f Danish, the identity o f intensifiers (4c) and a sub-part o f com plex reflexives (4a). If m entioned at all, intensifiers have usually been assum ed to be totally unrelated to reflexives. That is, the fact that reflexives and intensifiers often have the same (or partially identical) phonetic realization has in m any cases either been overlooked or tacitly assum ed to be a historical accident not worth investigating. In the m id-nineties a num ber o f researchers began to (re-)discover^ the close links b etw een in ten sifiers and reflex iv es and th e im portant interactions o f in ten sification and ^ Already in the 1971 article “Reflexive and Emphatic” J. A. Moyne observed that one of main problems besetting generative treatments of reflexives is “the general neglect of the emphatic, often confusing it with the reflexive.” (Moyne 1971:145). While the analysis presented here is very different from Moyne’s we agree with him that intensifiers and reflexives are elements of very different nature which should not be confused and that a number of problems o f current approaches to binding are due to the failure to properly identify when a given element is used as an intensifier and when it is a reflexive. Unfortunately, Moyne’s call for more research on intensifiers seems to have gone unheeded until recently. See also the early typological study by Moravcsik (1972 ). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. binding evident in the peculiar behavior o f com plex reflexives. A m ong these, M cKay (1991), Baker (1995), Zribi-H ertz (1995), Veraart (1996), Konig and Siem und (1999), and Siem und (2000) have provided the inspiration for many o f the insights on which the analysis presented here is based. Baker (1995) proposes to analyze w hat looks like locally free reflexives in English as intensified object pronouns. In a sim ilar vein, M cKay (1991) had suggested to analyze com plex pronom inal form s like he himself, she herself, etc. as intensified subject pronouns rather than as a special kind o f anaphoric expression, as proposed by Bickerton (1987) and latridou (1986). W e agree that adopting these proposals m akes it possible to correctly separate intensification and binding into independent m odules o f the grammar. That is, a syntactic principle B accounts for the distribution o f pronom inals (e.g. he, him) while the distribution o f intensifiers (e.g. himself) is subject to the sem antic and pragm atic principles o f intensification. At the end o f his paper Baker (1995) suggests that this analysis may be extended to the analysis o f com plex reflexives in G erm an. Stated in the term inology o f this dissertation, his idea is basically that the com plex reflexive sich selbst ‘REFL s e lf is form ed from its sim ple counterpart sich ‘R EFL’ by the same process o f adnom inal intensification w hich is responsible for the form ation o f intensified DPs, e.g. der Konig selbst the king self ‘the king h im se lf (see Konig and Siem und (1999) for an analysis o f Germ an along those lines). One o f the goals o f the present dissertation is to show that by adopting this approach it becom es possible to unify the account o f all adnom inally intensified expressions, be they reflexives, pronouns or DPs - som ething previous accounts o f binding and intensification are unable to do. The present dissertation is thus in part an attem pt to flesh out the research project outlined at the end o f Baker (1995): “The aim o f such work would be to gain a much clearer picture than now exists o f the essential nature o f locality, discourse prom inence, and contrast, considered as individual subsystem s o f gram m ar, and also a better understanding o f Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the ways in w hich they interact to produce the extrem ely com plex phenom ena that we encounter in actual language use.” (Baker 1995:99). That is, we argue that the so-called com plex reflexives (e.g. D anish sig selv ‘REFL s e lf are best analyzed as adnom inally intensified counterparts o f the sim ple reflexive (e.g. Dan. sig ‘R EFL’). By relegating reflexives anaphors and intensifiers to different m odules it becom es possible to adopt a nom inal approach to binding theory in which the distribution o f anaphors (e.g. sig) and pronouns (Dan. ham ‘him ’) follow from the interaction o f lexical properties o f nominal expressions and syntactic principles (e.g. LGB-style principles A and B). In contrast, the distribution o f the adnom inal intensifier selv ‘s e lf is accounted for by the sem antic/pragm atic principles o f intensification. That is, adnom inal intensifiers (e.g. Dan. selv ‘s e lf as used in Peter selv ‘Peter h im self), will be given a Roothian focus-based analysis (cf. Rooth 1992). In brief, the main idea consists in analyzing ALL the elem ents highlighted in bold in the exam ples in (1-4) as adnom inal intensifiers. 1.2 Independence of binding and intensifleation in Danish In this section, we show that due to its m orphologically transparent articulation o f intensification and binding, Danish is particularly well suited to test the proposals defended here. As illustrated in (1-3), in m any languages, including English, the mutual independence o f binding and intensification is obscured by w hat appears to be a total m orphological overlap betw een the elem ents uses as intensifiers and reflexives, see (5). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (5) System o f intensification o f nominal expressions in English. Sim ple/U nintensified Com plex/Intensified a. Reflexive (himself) h im self b. Subj.pronoun he he h im self c. O bj.pronoun him h im self {< h im h im selff d. DP the king the king h im self O ther languages, e.g. the M ainland Scandinavian languages, Dutch, G erm an etc., have m orphologically distinct realizations o f intensifiers and reflexives, e.g. the Danish intensifier selv ‘s e lf and the sim ple reflexive sig ‘REEL’, see (6). (6) System o f intensification o f nom inal expressions in Danish. Sim ple/U nintensified C om plex/Intensified a. Reflexive sig ‘him self/herself sig selv ‘him self/herself b. Subj. pronoun han ‘he’ han selv ‘he h im self c. Obj. pronoun ham ‘him ’ ham selv ‘him him self/him self d. DP kongen ‘the king’ kongen selv ‘the king him self As shown in (5-6) the English system o f nominal expressions is m ore opaque that the Danish system. The English form /i/m se//'translates into four different expressions in Danish, viz. the sim ple/unintensified reflexive sig ‘R EFL’, the com plex/intensified reflexive sig selv ‘REEL s e lf, the intensified pronoun ham selv ‘(him ) h im self, and the adnom inal intensifier selv ‘h im self. Figure 1 below further illustrates the mutual independence o f the reflexive sig ‘REEL’ and the intensifier selv ‘s e lf in D anish as well as their ‘overlap’ in the so-called com plex, viz. intensified, reflexive sig selv ‘REEL s e lf. The exam ples in figure 1 illustrate m ost o f the different uses o f sig ‘R EEL’ and selv ‘s e lf in figure 1 and (6a-d). The system in (5) represents the standard analyses o f nominal expressions in English in which h im se lf m (5a) is assumed to be an anaphor (e.g. Peterj washed himself) and him self in (5c) to be a logophor/Locally Free Reflexive(LFR) (e.g. Peterj said that Mary danced with everybody except himself). In contrast, as indicated in (5c), we follow Baker 1995 in assuming the latter use of him self to derive from an underlying intensified pronoun him himself. See chapter 5 for more detailed discussion of intensified pronominals in English. We also analyze him self in (5a) and (la) as an intensified 0 reflexive, Peterj hated 0, himself. See chapter 5 for more detailed discussion of intensified 0 reflexives. See also section 1.12 where this alternative analysis English is summarized in the table in (53). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 1. The intersection o f the intensifier selv ‘s e lf and the reflexive sig ‘R EFL’ and the independence o f intensification and binding. selv sig focus part. “middles’ (iii) excl.adv ((a)+(i)) int. refl. deagentive inchoatives adnom.int. refl. (iv) incl.adv inherently reflexives logophor Exam ples illustrating the different uses o f selv and sis in Figure 1: (i) Peter selv l0ste opgaven. Peter self solved problem -the ‘Peter himself solved the problem .’ (ii) Selv Peter loste opgaven. self Peter solved problem -the ‘Even Peter solved the problem .’ (iii) Peter loste opgaven selv. Peter solved problem -the self ‘Peter solved the problem himself.’ (iv) Peter havde selv lost opgaven. Peter had self solved problem -the ‘Peter had himself solved the problem .’ (v) Peter sagde at Marie dansede med alle andre end ham selv. Peter said that M arie danced w ith all other than him self ‘Peter said that M arie danced w ith everybody except (him ) himself.’ ((a)-l-(i)) Peter vaskede sig selv. Peter w ashed REFL self ‘Peter w ashed himself.’ (a) Peter vaskede sig. Peter w ashed REFLE ‘Peter w ashed.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (b) Lceberne l0snede sig, munden blev st0rre. lips-the loosen REFL m outh-the becam e bigger ‘The lips parted, the m outh grew bigger.’ (c) Peter slog sig. Peter hit REFL ‘Peter got hurt (accidentally).’ (d) Peter shammer sig. Peter sham es REFL ‘Peter is asham ed.’ A fter setting up the independence o f intensification and binding as the main hypothesis to be tested, we w alk trough all the cases o f overlap between these modules (e.g. the different uses o f intensified reflexives sig selv, intensified pronouns ham selv ‘him s e lf, and intensified R- expressions Peter selv ‘Peter h im se lf) as well as all the different uses o f selv and sig on their own. The core function o f the elem ent selv ‘s e lf is as a m arker o f adnom inal intensification, e.g. sentence (i) in figure 1. The core function o f the elem ent sig ‘R E FL ’ is as a reflexive pronoun, e.g. sentence (a) in figure 1. The behavior o f selv ‘s e lf and sig ‘R EFL’ is thus determ ined by different parts o f the gram m ar, viz. the m odules o f intensification and binding respectively. As figure 1 indicates, even when sig and selv come together to form the so- called com plex reflexive sig selv, see ((a)+(i)) above, these tw o elem ents still fall under two different m odules o f the grammar. Besides their core functions, both selv ‘s e lf and sig ‘R E FL ’ have a num ber o f other uses. The sentences in (ii)-(iv) illustrate the use o f selv ‘s e lf as an additive scalar focus particle sim ilar in m eaning to English even (ii), as a m anner adverbial m eaning ‘by oneself, w ith ou t h e lp ’ (iii), and as a q u antifier-floated adnom inal in ten sifier w h ich is u sually interpreted sim ilarly to an inclusive particle m eaning ‘also, too’ (iv). As will be shown in chapter 2, one o f the advantages o f the focus-based analysis presented here is that it allows for the potential unification o f both core- and non-core uses o f the elem ent selv ‘s e lf. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The sentences in (b)-(d) illustrate the non-reflexive uses o f the reflexive sig ‘R EFL’ in m iddle constructions, deagentive unaccusative predicates and inherently reflexive predicates. The main distinction betw een reflexive and non-reflexives uses lies in their them atic status: reflexive sig is a theta-m arked argum ent m erging in argum ent position o f a transitive predicate, see (a) in Figure 1, w hile non-reflexive sig (b-d) is non-them atic since it merges with the small v o f unaccusative (forms of) predicates. Reflexive uses o f sim ple reflexives, e.g. Dan. sig ‘R EFL’, will be treated in chapter 3 w here we also argue that sentences like (v) support analyzing so-called “logophors” (e.g. Dan. ham selv ‘him h im se lf) as adnom inally intensified pronouns (cf. M cKay 1991). N on-them atic uses o f reflexive elem ents will be discussed in m ore detail in chapter 4. 1.3 Revisiting the analysis of simple and complex reflexives in Danish Since Faltz’s (1977,1985) observation that com plex reflexives have a tendency to be locally bound w hile simple reflexives allow for long-distance binding, m ost accounts o f reflexives have aim ed to explain the correlation betw een the m orphological form o f reflexives and locality constraint within the binding theory. Faltz’s difference betw een sim ple and com plex reflexives has been claim ed to be found in Danish, see (7). (7)'' a. Peter, aflaste *sig, / sig, selv. Peter replaced REFL / REFL self ‘Peter, replaced him self,’ b. Peter, had Hans I , qflase sig,*^ / sig*, k selv. Peter asked Hans replace REFL / REFL self ‘ Peter; asked Hans to replace him;. ’ In the binding literature, exam ples like (7a,b) have been adduced as evidence that the simple reflexive sig is anti-local and specialized for LD-binding w hile the com plex reflexive sig selv requires a local antecedent. That is, Danish has been used to support the theory that the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 m orphological com plexity o f anaphors interacts directly w ith locality. Indeed, m ost current theories o f binding assum e that the correlation between m orphological com plexity and locality should be encoded in the binding principles them selves. V ikner (1985) proposes that simple sig is an anti-local anaphor w hile sig selv is a com plex anaphor in w hich presence o f selv has the direct effect o f restricting the binding dom ain o f the anaphor. A ccording to the proposal defended here, V ikner’s view is m isguided in several ways. First, his approach does not take the uses o f selv ‘s e lf as an intensifier, e.g. (4c) into account. Second, there is strong evidence in favor o f assum ing theta-m arked sim ple sig to be a locally bound theta-role receiving reflexive - thus falsifying claim s that sig is an anti-local anaphor specialized for long-distance binding, see (8a,b). (8) a. Peter, vaskede sig, / sig, selv. Peter w ashed REFL / REFL self ‘Petep washed (him selfi).’ b. Peter, bad Hansk vaske sig„k / sig*, k selv. Peter asked Hans wash REFL / REFL self ‘Petep asked HanSk to wash himi/himselfk.’ In Reinhart & R euland’s (1993) term inology, the sim ple vs. com plex distinction is referred to as a distinction between (sim ple) SE-anaphors and (com plex) SELF-anaphors. They propose a predicate-centered binding theory in w hich SELF is a reflexivizing elem ent whose function is to reflexive-m ark predicates that are not lexically reflexive. That is, they argue that it is selv ‘s e lf rather than sig ‘R EFL’ which is the true reflexive element. In order to m aintain this view, they are forced to assum e that sim ple sig can be locally bound only when occurring as argum ent o f inherently reflexive predicates, i.e. predicates w hich are lexically m arked for reflexivity. A s a consequence, they have to stipulate the existence o f double entries for all the predicates w hich allow both locally bound sim ple reflexives and non-coreferential DPs as internal argum ents, e.g. vaske ‘w ash’ in (8a,b). In other words, Reinhart & R euland’s (1993) * We adopt the convention o f using the star symbol “*” to indicate the an example is ungrammatical. For examples Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 analysis relies on an unw arranted stipulation o f lexical ambiguity. To account for the fact that both the sim ple reflexive sig and the com plex reflexive sig selv can he found in sentences like (8a) they need to stipulate the existence o f tw o synonym ous verbs m eaning ‘w ash’ differing only in the feature [+/-inherently reflexive]. In section 1.5 we argue that no such stipulation is necessary if sig selv is analyzed as the adnom inally intensified version o f sig. To account for the LD-binding o f sig we adopt a Pica-style analysis based on successive cyclic m ovem ent o f sig, first to the m ost local TP and then, potentially, to higher TPs at LF (cf. Pica 1984, 1986, and Jakubow icz 1994). However, in contrast to Pica who argues that it is the m orpho-syntactic properties o f sig selv w hich makes LD-binding im possible we suggest that the apparent locality constraint on sig selv is an epiphenom enal result o f adnom inally intensifying sig. That is, we follow Konig and Siem und (1999:63-4) in assum ing that sig .selv is an intensified reflexive whose sem antic and syntactic properties can be explained as deriving com positionally from the properties o f its constituent parts, i.e. the simple reflexive sig ‘R EFL’ and the adnom inal intensifier selv ‘s e lf ^ The sentence in ((a)+(i)) in figure 1 above thus illustrated the overlap o f the m odules o f binding and intensification w hich gives rise to the so-called com plex reflexive sig selv ‘REFL self, h im self. Unlike Reinhart & R euland’s (1993) predicate-centered approach to binding we propose a nom inal approach in w hich binding is defined as the interaction o f intrinsic properties o f nominal expressions and syntactic locality constraints (cf. Chom sky 1981), see (9-10). (9) a. An anaphor is bound in a local domain, b. A pronom inal is free in a local domain. which are merely pragmatically or semantically deviant/infelicitous the sharp symbol is used. “[l]n many European languages (i.e. Norwegian seg selv, Spanish si mismo, Russian sain sebja, Italian se stesso, etc.) [SELF anaphors] are combinations of an SE anaphor and an adnominal intensifier, whose overall meaning is simply a function of that of the two components.” (Konig and Siemund 1999:63-4). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 (10) B inding D om ain: A lpha is a binding dom ain for beta if and only if alpha is the m inimal category (i.e. the sm allest DP or IP/S) containing beta, a case-licensor o f beta, and a SUBJECT accessible to beta. The binding theory outlined here will be developed in m ore detail in chapter 3. 1.4 Adnominal intensification Before we m ove on to showing how predicate m eaning influences intensifier-adjunction to simple reflexives (see section 1.5), we need a theory o f adnominal intensification. We argue that the sem antic contribution o f adnom inal intensifiers (e.g. Danish selv, Eng. himself) neither always involves scalar ordering o f focus alternatives (cf. Eckardt 2001) nor is reducible to centrality-effects (cf. Konig (1997) and Siem und (2000), but is rather very sim ilar to that o f contrastive focus which - depending on the context - may or may not involve scalar ordering o f focus-generated alternatives. Unlike the focus particle selv ‘even’, the adnom inal intensifier selv ‘h im self does not presuppose that the proposition is true for at least one other elem ent in the focus-generated set o f alternatives, see (11a) vs. ( llh ) . This presupposition is som etim es referred to as the existential presupposition/im plicature. At this point no attem pt has been m ade to distinguish betw een presupposition and im plicature. The tw o term s are used interchangeably w ithout difference in m eaning in the rest o f this dissertation. (11) a. *Det var selv biskoppen der holdt gudstjenesten. (selv ~ even) It was self bishop-the who held service-the ‘*lt was even the bishop who held the service.’ b. Det var biskoppen selv der holdt gudstjenesten. (selv ~ himself) It was bishop-the self who held service-the ‘It was the bishop him self who held the service.’ The exam ple in (11a) shows that the focus particle selv ‘even’ cannot occur in clefts, w hich imply uniqueness, because that would lead to a contradiction o f the so-called existential Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 im plicature. Since the adnom inal intensifier selv ‘h im self can occur in clefts, see (11b), we therefore conclude that it does not evoke this existential im plicature w hich is m andatory with the focus particle even. The exam ples in (12)-(13), w hich contain predicates presupposing uniqueness, further confirm this difference betw een even and the adnom inal intensifier. (12) a. #Selv Dronningen gav nytdrstalen. {selv-eve n ) even queen-the gave new -years-speech-the ‘#Even the Queen delivered the N ew Y ear’s Speech.’ b. #Even the President delivered the inaugural speech. (13) a. Dronningen selv gav nytdrstalen. {selv ~ himself) queen-the self gave new -years-speech-the ‘The Q ueen herself delivered the N ew Y ear’s Speech.’ b. The President him self delivered the inaugural speech. A ccording to standard analysis o f focus particles, the sem antic contribution o f even involves both the existential im plicature illustrated ( 11) and a scalar im plicature w hich orders the focus alternatives on a scale o f expectedness. That is, in a sentence like Even the King came to the meeting the scalar im plicature w ould rank the K ing lower on a scale o f likelihood than any o f its focus-generated alternatives, e.g. the Queen, the prince, the bishop, etc. Eckardt (2000) suggests that the sem antic contribution o f adnom inal intensifiers is identical to that o f the focus particle even except for the absence o f existential implicature. We argue that this analysis o f the sem antic contribution o f adnom inal intensifiers in on the w rong track. W hile it is true that DPs adnom inally intensified by selv often refer to entities w hich are rem arkable/unexpected in a given context, this is by no m eans always the case, see (14-15). (14) A very powerful earthquake struck the center o f Rome and the Vatican. a. The Pope him self perished in the rubble. b. #Even the Pope perished in the rubble. (15) We do not live in the suburbs any more. a. We live in Paris itself. b. #We even live in PARIS. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 G iven that an earth quake does not differentiation betw een people o f high and low status or prom inence there is no reason to expect the Pope to be less likely to perish in an earth quake than any other person living close to the epicenter, see (14a). Likewise, in (15a) the place name Paris is not intensified because it denotes an unexpected or rem arkable location, but simply because it is contrasted w ith another location, viz. the suburbs. N otice that in both cases the adnom inal intensifier is perfectly acceptable while the focus particle even, w hich always carries with it a scalar im plicature, see (14b) and (15b), is not. We take this as evidence that the scalar im plicature o f even is N O T an integral part o f the sem antic contribution o f adnom inal intensifiers'^. We therefore conclude that the sem antic contribution o f the adnom inal intensifier involves neither existential nor scalar im plicatures and therefore is sim ilar to plain focus, see (16-17). The form ula highlighted in bold indicates a proposition. (16) a. [The king himself]i.- came to the meeting. b. [The kingjf came to the meeting. (17) a. O rdinary sem antic value o f (16a): [[ [The king himselfjp came to the meeting]]® = 1 iff the king cam e to the meeting, b. Focus sem antic value o f (16a): [[ [The king himself]F came to the m eeting ]] ^ = {cam e-to-the-m eeting(x) \ x e D,} ‘the queen cam e to the m eeting’, ‘the prince cam e to the m eeting’, etc. Like plain focus, the basic sem antic contribution o f adnom inal intensification consists in contrasting the referent o f the focused/intensified expression with a contextually determ ined set o f alternatives, see (17b). The contrastiveness condition on intensification in (18) captures th e fact that in ten sification can o n ly occu r in con texts in w here the gen eration o f contrast sets o f focus-alternatives is possible. ® Notice that the existential implicature may be responsible for the unacceptability of (15b). The focus particle selv ‘even’ presupposes that the proposition is true for at least one other member o f the set of alternatives. That is, in this case, the presupposition is that we lives in one or more places in addition to Paris (e.g. Rome, London, etc.). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (18) Contrastiveness condition on adnom inal intensification: A nom inal expression (DP, pronoun, reflexive) can only be intensified adnom inally if it can be contrasted with other expressions in the context in which it is found. A s shown in section 1.5, unlike the conditions on intensification proposed by Baker (1995:79- 80), Konig (1997), Siem und (2000:154), the existence o f (18) is supported by evidence from the distribution o f com plex reflexives. 1.5 P re d ic a te m ean in g a n d in tensification o f reflexives In this section, we outline how the analysis o f intensification proposed above can account for the interaction between predicate m eaning and adnom inal intensification o f reflexives. One o f the im portant contributions o f the present dissertation consists in proposing a new view o f the interaction betw een predicate m eaning and the choice between sim ple/unintensified and com plex/intensified reflexives. Early versions o f the binding theory (e.g. Chom sky 1981) usually had very little to say about the interaction o f predicate m eaning and reflexives. Binding theories w hich do take predicate m eaning into account usually encode the different behavior o f predicates with binary lexical features (e.g. Reinhart & Reuland 1993, and Zribi- Hertz 1995). Zribi-H ertz (1995) showed that non-contrastive prepositional predicates in French differ w ith respect to w hether the elem ent used as adnominal intensifier (lui)-meme ‘(him -)sam e’ is or is not adjoined to the locally bound pronoun in sentences like (19-21). (19) Pierre est fier de lui/ lui-meme/Marie. (neutral predicate) Pierre is proud o f him /him -sam e/M arie ‘Peter is proud o f him self/M arie.’ (20) Pierre est jalonx de *lui/lui-meme/Marie. (anti-reflexive predicate) Pierre is jealous o f him /him -sam e/M arie ‘Peter is jealous o f him self/M arie.’ This presupposition clashes with the expectation that, under normal circumstances, most people live in only one place. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 (21) Pierre est hors de lui/*lui-meme/*Marie. (inherently reflexive predicate) Pierre is outside him /him -sam e/M arie ‘Peter is besides him self/M arie (o f joy/anger/etc.).’ Zribi-H ertz (1995) suggests that the different behavior o f predicates is encoded lexically with the feature [+/-disjoint reference]. R& R (1993) also resorts to lexical m arking to account for the different behavior o f predicates, i.e. [+/-inherently reflexive]. In contrast, the proposal defended here assum es no binding-specific lexical features. Instead, adnom inal intensification o f reflexives is assum ed to be determ ined by pragm atic and sem antic features o f the governing predicate plus those o f the discourse context in w hich it occurs. That is, predicate m eaning is ju st one am ong m any contextual factors which may influence adnom inal intensification o f reflexives. As m entioned in section 1.4, we propose a focus-based analysis o f adnom inal intensification. The intensifier selv ‘s e lf is associated w ith focus and evokes a contextually defined set o f alternatives to its associate, i.e. the host DP to which it is adjoined. Unlike the focus particle selv ‘even’, the adnom inal intensifier selv ‘s e lf does not obligatorily involve scalar ordering o f alternatives. That is, it is the surrounding context w hich determ ines w hether or not the alternatives are ordered (with the associate D P being defined as the end-point o f a contextually defined scale) or w hether the associate is simply singled out as ‘prom inent’. W hen adjoined to the reflexive sig, the intensifier, because o f its meaning, will be com patible with some predicates (22), obligatory w ith others (23), and im possible w ith still others (24). Descriptively, predicates can thus be divided into three types determ ined by the distribution o f different kinds o f nominal expressions in argum ent position: (i) those w hich can take as direct object the unintensified reflexive sig, the intensified reflexive sig selv and DPs, (ii) those w hich can take sig selv and DPs but not unintensified sig, and (iii) those w hich can only take unintensified sig, see (22-24). The term s “neutral”, “anti-reflexive” and “inherently reflexive” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 does not im ply that we assum e the existence o f lexically determ ined predicate-classes. That is, these three term s denote distributional classes rather than lexically defined classes. As discussed in chapter 3 and 4, in certain predicates can change from an anti-reflexive to a neutral predicate depending on the larger context. K onig & Siemund (1999) also distinguish between different predicates classes. W hat he calls “other-directed” verbs corresponds more or less to w hat is referred to as “anti-reflexive” verbs here. See also Larsen (1997) who uses sim ilar term s to refer to different verb-types. (22) “N eutral” predicates: a. Peter vasker sig / Peter washes REFL / ‘Peter washes him self / the car.’ b. Peter tarrer sig / Peter dries REFL / ‘Peter dries him self / M ary.’ sig selv / REFL-SELF / sig selv / REFL-SELF / bilen. car-the Marie. Mary (23) (24) “A nti-reflexive” nredicates: a. Peter misunde *sig / sig selv / Marie. Peter envies *REFL / REFL-SELF I M ary ‘Peter envies him self / M ary.’ b. Peter mistcenker *sig / sig selv / Marie. Peter suspects *REFL / REFL-SELF / M ary ‘Peter suspects him self / M ary.’ “Inherentlv reflexive” predicates: a. Peter dukkede sig / *sig selv / *Marie. Peter ducked RELF / *REFL-SELF / *Mary ‘Peter ducked (*him selt) / *M ary.’ b. Peter tog en kniv med sig / *sig selv / *Marie Peter took a knife w ith REFL / *REFL self / *Mary ‘Peter took a knife w ith him /*him self/*M ary.’ We argue that it is the m eaning o f the predicates in (22-24) - in com bination with pragm atic factors and w orld-know ledge - w hich determ ines w hether sig can occur alone in object position or w hether intensification by selv is necessary. The predicates in (24) are sem antically “inherently reflexive” in that they are only com patible w ith reflexive scenarios, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 i.e. one cannot duck anybody but oneself’. The predicates in (23) are “anti-reflexive” in that they carry with them the presupposition that their argum ents refer to (representationally) different entities, i.e. under norm al circum stances it is highly unexpected for anyone to suspect him self o f a crim e or to envy him self. Finally, the predicates in (22) are “neutral” in the sense that they evoke no such presuppositions and consequently allow all types o f direct objects. Thus, these exam ples illustrate that xe/v-intensification o f reflexives follows the pattern o f 5g/v-intensified DPs by occurring only in contexts w hich allow for the generation o f contrast sets o f alternatives, thus satisfying the contrastiveness condition on adnom inal intensification (18). The sentences in (24) involve predicates w hose sem antics excludes anything but the simple reflexive sig as internal object. Flence, since they violate (18), adnom inal intensification is im possible. This analysis o f com plex reflexives as intensified nominal expression allow s for the unification o f the analyses o f adnom inally intensified DPs, e.g. Peter selv ‘Peter h im self (4c), and intensified reflexives like sig selv in (22-23a,b). 1.6 In ten sificatio n an d th e m eaning difference betw een sim ple a n d com plex reflexives Since Jack en d o ff s (1992) paper “M adam e Tussaud m eets the binding theory” called attention to the fact that nom inal expressions can be used to refer to statues or other types o f representations o f their normal referents, it has often been noted that sim ple and com plex reflexives differ w ith respect to w hether or not they can be used to refer to representations o f the individual denoted by the antecedent. First consider the instances o f locally bound sig and sig selv in (25). ’ The question whether the sig in dukke sig in (24a) above is a theta-role receiving argument of a transitive verb or a non-thematic sig marking unaccusativity of the predieate will be discussed in more detail in chapter 4. That is, are “inherently reflexive” verbs true transitive verbs, assigning two theta-roles, whose selection restrictions disallow anything but locally bound anaphors as internal arguments, or are they intransitive/unaccusative predicates. While most previous accounts assume that all “inherently reflexive” verbs are intransitive/unaccusative predicates, we propose that there are both transitive and unaccusative “inherently reflexive” verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 (25) Context: Im agine Bill Clinton visiting the w ax museum. He notices a statue o f him self w ith an unshaven face. Since he doesn’t like the look o f the statue he takes out a razor and starts to shave it. a. Bill Clinton, barberede sig/. b. Bill Clinton, barberede sigi selv. Bill C linton shaved RELF/REFL self ‘Bill C linton shaved (him self).’ The sentence with the simple reflexive sig in (25a) can only have the interpretation in w hich the real C linton shaves him self (i.e. the real Clinton), see (26a). It cannot have the so-called statue-reading, see (26b), in w hich the real Clinton shaves a statue o f Clinton. In contrast, the sentence w ith the com plex reflexive sig selv in (25b) can have both reading (26a) and (26b). (26) a. Clinton<real> shaves him self<real>. b. Clinton<real> shaves him self<statue>. We argue that the sem antic difference betw een sim ple/unintensified and com plex/intensified reflexives is a consequence o f intensifier-adjunction. As described in section 1.4, adnom inal intensification autom atically generates a set o f alternative referents for the associate o f the intensifier. In the case o f the com plex reflexive sig selv it is the focus-generated set o f alternative sem antic values for s'/g-(triggered by adnom inal intensification) that licenses statue- readings in w ax-m useum contexts, e.g. (25b) (see also chapter 3, section 3.3.2.7.1 for more discussion o f the link between adnominal intensification o f reflexives and the availability o f statue-readings). In its unintensified/sim ple form sig behaves as a variable w hich has to be referentially identical to its binder. In other words, adding selv to sig allows for the latter to refer to a statue/representation o f its antecedent, see (25b). Thus the present approach achieves a (sem antically and m orphologically) fully com positional analysis o f com plex reflexives. N ote that a sim ilar proposal is found in Konig & Siem und (1999:48). Furtherm ore since the anti-reflexive predicates, w hich require com plex reflexives, see (23), all presuppose the non-identity (or at least the non-representational identity) o f their argum ents, w e are now able to unify the account o f anti-reflexivity (23) and so-called doppelganger- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 effects as illustrated by the availability o f statue-readings o f sig selv in (25). This unification o f the analyses o f anti-reflexivity and doppelganger-effects in term s o f adnom inal intensitication is one o f the m ain contributions o f tbe present dissertation. 1.7 Logophoricity and intensified pronouns In this section we outline how the com positional analysis o f reflexives based on the independence o f intensification and binding (see sections 1.3 to 1.6) may be extended to the analysis o f sim ple and com plex subject and object pronouns. L et’s begin by looking at sim ple/unintensified and com plex/intensified subject pronouns. M cKay (1991) argues that he him self is not a new kind o f anaphor (pace Bickerton 1987). U sing our term inology, M cKay essentially proposes that he him self is an adnom inally intensified pronoun. Syntactically, it behaves like its unintensified counterparts he and him in that it is subject to Principle B o f the binding theory. Pragm atically, it is subject to the same sem antic/pragm atic condition, i.e. (18), as other intensified nominal expressions. In this sense, M cK ay (1991) is the forerunner o f Baker (1995) as well as the present dissertation. We thus share M cK ay’s conclusion that him self is gram m atical w henever he is gram m atical, and the same range o f antecedent relationships is gram m atically possible, though he him self requires a relevant contrast or com parison for pragm atic appropriateness” (M cK ay 1991:370- 371). M cK ay uses the exam ple in (27) to show that he him self is not a special kind o f anaphoric expression, but sim ply an intensified pronoun which is subject to principle B. (27) A: Mary, has been concerned about her friends. Susans said that several were going to fa il the course, and Susans might be right. But Mary, should think more about her, own work. How will Mary, do on the exam? B: I d o n ’ t know, but Susant says that she/she fsh e t herself/she^ herself will pass. M ore precisely, M cK ay’s exam ple in (27) falsifies B ickerton’s (1987) claim that he him self behaves like an anaphor in that it “cannot have an antecedent outside the sentence if there is a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 possible antecedent Inside”. M cK ay’s m ain contribution is thus to show that, given the right context, he him self can be bound by a sentence-external antecedent in spite o f the presence o f a potential antecedent inside the sentence. That is, the pronom inal nature o f subject pronouns is not affected by adnom inal intensification. In other words, binding and adnominal intensification are independent o f each other. So far we have seen that adnom inal intensification o f reflexives and subject pronouns is independent o f the binding properties o f these elem ents. Taking the independence o f binding and intensification to its logical conclusion we argue that it can be extended to object pronouns as well. That is we propose that the com plex pronoun ham selv ‘him self/him h im self in the Danish sentences (28-29) should be analyzed as intensified versions o f their simple counterpart, i.e. ham ‘him ’. (28) a. JoHi vil giftes med en kvinde som er stolt a f hamt / Hham, selv. Jon w ants m arry-PASS w ith a w om an who is proud o f him / him self ‘Jon wants to get m arried to a w om an w ho is proud o f him .’ b. Kongen troede at ingen kunne lide ham, / #hami selv. king-the thought that nobody could like him / him self ‘The king thought that nobody liked him .’ (29) a. Joni plejede at hade fo lk som var anderledes end ham/ / ham/ selv. Jon used to hate people who w ere different than him / him self ‘Jon used to hate people who w ere different from h im se lf’ b. Carlj sagde at Marie havde snakket med alle andre end ham, / hamf selv. Carl said that M arie had talked to all others than him / him self ‘Carl said that M arie had talked to all others than h im se lf’ W hile the distribution o f the pronom inal ham ‘him ’ is determ ined by the syntactic principle B o f the binding theory, i.e. pronouns have to be free from binding by a subject w ithin the tensed clause, see (9b), the distribution o f the elem ent selv is determ ined by the sem antic-pragm atic principle o f the m odule o f intensification, i.e. (18). In (29) both the unintensified pronoun ham ‘him ’ and the intensified pronouns ham selv ‘him s e lf are allowed. In contrast, the sentences in (28) do not seem to license Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 intensifier-adjunction o f the pronom inal ham ‘him ’ in an out-of-the-blue context. A dopting sim ilar proposals by Zribi-H ertz (1989), Safir (1992), and Konig and Siem und (1999) we suggest that this differenee is determ ined by tbe properties o f tbe governing expressions. Since tbe expressions anderledes end ‘different th an ’ and andre end ‘other th an ’ in (29) are inherently eontrastive they are able to license intensifier-adjunction to tbe pronoun. Lacking this ‘inherently contrastive’ quality tbe expressions stolt a / ‘proud o f and lide ‘like’ in (28) are unable to do so^ Once again we show that tbe distribution o f different kinds o f nominal expressions (e.g. reflexive anaphors sig ‘R E F F ’, pronouns ham ‘him ’, and r-expressions Peter) follows from syntaetie prineiples o f tbe binding theory, in this case prineiple B, while tbe distribution o f tbe elem ent selv is determ ined by tbe sem antie-pragm atic principles o f tbe m odule o f intensification. 1.8 Possessive reflexives, p ro n o u n s a n d in tensifiers One o f tbe m ain strengths o f tbe analysis proposed here is that it can be straightforw ardly extended to intensified possessive reflexives and pronouns, w hich constitute a m ajor problem for predicate-based approaebes to binding, e.g. Reinhart & Reuland (1993). Tbe Danish system o f nominal expressions in possessor position is given in (30). N otice that except for tbe suppletive form o f tbe intensifier, see (31), and tbe different case forms (assum ing sin ‘PO SSR EFL’ to be tbe genitive form o f tbe accusative/dative form sig ‘R EFL’) this system is exactly tbe same as tbe one found in argum ent position, com pare (30) with the system o f argum ent DPs given in (6) in section 1.2. * As noted by Zribi-Hertz (1989) among others, there are a number of sentences which do not appear to have “inherently contrastive” predicates but still require “logophors” (i.e. himself) rather than pronominal (him). These and other related issues (e.g. the definition of what constitutes an “inherently contrastive predicate”) will be discussed in more detail in chapter 3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 (30) D anish nominal expressions in possessor position: Sim ple/unintensified C om plex/intensified a. Reflexive sin ‘his/her/one’s ’ sin egen ‘his/her/one’s ow n’ b. Pronoun hans ‘h is’ hans egen ‘his ow n’ c. DP kongens ‘the king’s ’ kongens egen ‘the king’s ow n’ (31) Suppletive variants o f the adnom inal intensifier in Danish: a. A rgum ent position: selv ‘h im self b. Possessor position: egen ‘ow n’ Unlike English, Danish has a possessive reflexive, i.e. sin ‘his/her/one’s’, in addition to the possessive pronouns hans ‘h is’. The distribution o f simple and com plex possessive reflexives and possessive pronouns follow the sam e general pattern as the distribution o f sim ple and com plex reflexives and pronouns in argum ent position. T hat is, like their argum ent position counterparts (i.e. sig ‘him, her, one’ and ham ‘him ’) the distribution o f sim ple forms, e.g. sin “PO SSR ELF’ and hans ‘h is’, is constrained by the principles A and B o f the binding theory. Sin (like sig) is a reflexive and m ust be bound by a subject in a local dom ain (38a) and hans ‘his’ (like ham ‘him ’) is a pronoun w hich m ust be free in its local domain, i.e. the minimal tensed clause, here indicated w ith square brackets (32b). (32) a. Peterj sagde [at John/i vaskede sin*, 4 * 2 tegnebog]. Peter said that John w ashed PO SSREFL w allet ‘Petep said that John^ w ashed his.,/k/*z w a lle t’ b. Peter, sagde [ at John/, vaskede hans; *t z tegnebog]. Peter said that John w ashed his w allet ‘Petep said that John^ w ashed his,/.k/z w a lle t’ The sentences in (33a,b) illustrate the fact that both sim ple/unintensified and com plex/intensified possessive reflexives may be LD -bound out o f infinitival clauses. (33) a. Peter, bad Hans/, vaske sin,,/, hund. Peter asked H ans w ash P O SS R E F L d og ‘Petep asked Hansi to wash hisi/k dog.’ b. Peteri lad forceldrenez save i sin, * z egen seng mens hani selv sov pa sofaen i stuen. Peter let parents-the sleep in POSSREFL own bed w hile he self slept on sofa- the in living room -the ‘Petep let hisi parents sleep in his own bed while hcj him self slept on the sofa in the living room .’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 Though LD -binding o f the simple possessive reflexive sin is more com m on, LD -binding o f intensified possessive reflexives is not excluded, see (33b). That is, the adjunction o f the intensifier egen ‘ow n’ does not affect the LD-potential o f the possessive reflexive sin. Once again, adnom inal intensification o f reflexives does not affect locality constraints. The exam ples in (34-36) illustrate that Danish possessive reflexives sin and sin egen exhibit the same overall distributional pattern as the argum ent reflexives sig and sig selv with respect to “neutral”, “anti-reflexive” and “inherently reflexive” constructions/sentences. (34) “N eutral” constructions: a. Peter vasker sin /sin egen / Jo h n ’ s tegnebog. Peter washes POSREF /PO SREF ow n/ John’s wallet. ‘Peter washes his /his own / John’s w allet.’ b. Peter hader sin /sin egen /J o h n ’ s mar. Peter hates POSREF /PO SREF ow n/ John’s mother. ‘Peter hates his / his own / John’s m other.’ (35) “A nti-reflexive” constructions: a. Peter er *sin /sin egen /J o h n ’ s fjende. Peter is *POSREF /PO SR EFL ow n/ John’s enemy. ‘Peter is his / his own / his enem y.’ b. Peter stjal ’ ^??sin / sin egen /J o h n ’ s tegnebog. Peter stole *??PO SR E F/PO SSR E F own / John’s wallet. ‘Peter stole his / his own / John’s w allet.’ (36) “Inherentlv reflexive” constructions: a. P[an var ved at gaa ud a f sit /? * sit eget/*Peters gode skind a f glcede. he was about to go out o f PO SREF/?*PO SREF ow n/Peter’s good skin o f happiness ‘He nearly jum ped out o f his good skin o f sheer happiness.’ b. Plan hyttede sit / ??sit eget /*Peters skin. (cont. bet. a/b) he savede PO SREF/ ??PO SREF ow n/*Peter’s skin ‘He saved his own life.’ The contrast between (34) and (35) sbows that it is the sem antic/pragm atic m ake-up o f the sentence w hich triggers intensifier-adjunction to sin. Based on the m eanings o f the predicates vaske ‘w ash’ and stjcele ‘steal’ and world knowledge, native speakers know that for a given individual x, the proposition x washes x ’ s wallet is pragm atically OK.. The proposition x steals Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 X’ s wallet, however, is pragm atically odd; stealing is per definition an “anti-reflexive” activity; people do not consciously steal their own belongings. Indeed, “anti-reflexivity” or “other- directedness” is an integral part o f the m eaning o f the predicate stjcele ‘steal’, i.e. given normal circum stances the expectation is that people would not steal their own things. Hence, since anti-reflexive sentences o f the type DPi stole POSSREFL,’ s DP presuppose the (representational) non-identity o f the referents o f the possessive reflexive and its antecedent, intensification o f the possessive reflexive sin is necessary to mark the (representational) non identity. The exam ples in (36) show that Danish also have “inherently reflexive” possessive constructions (usually o f idiom atic nature) w hich - due to their sem antics - disallow all but the sim ple unintensified reflexive possessive sin, see (18). W e therefore conclude that the distribution o f Danish simple and com plex possessives {sin and sin egen) in “neutral”, “anti-reflexive” and “inherently reflexive” constructions follow the sam e general pattern as the sim ple and com plex argum ent reflexives sig and sig selv. This constitutes a m ajor obstacle for analyses o f binding w hich are based on the notions o f co- argum enthood, e.g. Reinhart and Reuland (1993). It seems rather im plausible to m aintain that the difference between (34a) and (35b) should derive from the a lexical feature, e.g. Reinhart and R euland’s [+/-inherently reflexive], or Z ribi-H erts’s [+/- Disjoint reference], distinguishing betw een the predicates vaske ‘w ash’ and .stjcele ‘steal’. The structural relationship betw een the reflexives (i.e. sin and .sin egen) and their antecedent has to be assum ed to be the sam e all the sentences in (34-36). A nd yet, the sentences differ as to w hether they allow the sim ple possessive reflexive or not. W hile this difference can be accounted for in term s o f focus and presuppositions as illustrated above, it seem s im possible to account for it in term s o f any kind o f predicate-centered ‘reanalysis’ or ‘lexical am biguity’ account w hich only applies to co-argum ent binding. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 1.9 Refuting anti-Iocatity The account o f the interaction betw een predicate/sentence m eaning and intensification o f reflexives outlined in section 1.5 and 1.8 above leads us to refute the standard account o f sig in Danish (Norwegian, Dutch, etc.) w hich is based on the assum ption that them atic sig is an anti local anaphor w hich can never be locally bound (cf. Holm berg, (1984), V ikner (1985), etc.). That is, the binding theory proposed here does allow sim ple reflexives to be locally bound. Rather than being due to any inherent anti-locality o f the simple reflexive sig, it is sem antic or pragm atic factors w hich conspire to prevent local binding o f unintensified reflexives in certain cases, e.g. anti-reflexive predicates (23) and (35). As shown above, intensifier-adjunction to simple reflexives is determ ined not by the binding theory but by the m odule o f intensification and follows directly from sem antic and pragm atic properties o f the predicates and the larger context. In other words, anti-locality as an intrinsic property o f sim ple reflexives does not exist; or, at least not in Dutch, Danish and the other M ainland Scandinavian languages. A lthough we do reject the idea that sig is anti-local, we still acknow ledge the existence in Danish o f tw o different uses o f unintensified sig: (i) sig used as a them atic reflexive pronoun as in (37), (ii) sig used as a non-them atic gram m atical m arker o f unaccusativity as in (38-40). (37) a. Peter vaskede sig. b. Peter vaskede sig selv. c. Peter vaskede Hans. Peter vaskede REFL / REFL self / Hans ‘Peter w ashed (him self)/H ans.’ (3 8 ) D ep on en t .s/g-verb s: a. Peter opfarte sig / *sig selv / *Hans godt. Peter behaved REFL/REFL self/Hans well ‘Peter behaved w e ll’ b. Peter skammer sig / *sig selv / *Hans. Peter sham es REFL/REFL self/Hans ‘Peter is asham ed.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 (39) D eagentive unaccusatives (inanim ate subject) w ith transitive counter-parts: a. D0ren abnede sig / *sig selv / *vinduet. (com pare w. (41 a)) door-the opened R EFL/REFL self/w indow -the ‘The door opened.’ b. D0ren lukkede sig/*sig selv / *vinduet. (com pare w. (41 b)) door-the closed REFL/REFL self/w indow -the ‘The door closed.’ (40) D eagentive unaccusatives (anim ate subject) w ith transitive counterparts: a. Peter slog sig (pd bordkanten). Peter hit REFL on table-edge-the ‘Peter got hurt (on the edge o f the table).’ b. Peter brcendte sig (pd ovnen). Peter burned REFL on oven-the ‘Peter got burned on the oven.’ The sig in (37a) is them atic. That is, it m erges in object position o f the transitive verb vaske ’w ash’ from w hich it receives the TH EM E theta-role. In this respect it is no different from the direct object DP Hans in (37c) or the intensified reflexive sig selv in (37b). In contrast, the sig in (38-40) does not receive any theta-role. The verbal predicates in (38-40) are either lexically unaccusative, e.g. skamme ’sham e’ and skynde ’hurry’ in (38), or unaccusative predicates with transitive counterparts, e.g. dbne ’open, lukke close’, sld ’h it’, and brcende in (39-40). The transitive counterparts o f these verbs are given in (41). (41) a. Peter abnede doren. (cf. (39a)) Peter opened door-the ’ Peter opened the door. ’ b. Peter lukkede daren. (cf. (39b)) Peter closed door-the ’Peter closed the door.’ c. Peter slog sig/sig selv/Hans med en kolle. (cf. (40a)) Peter hit REFL/REFL self/Hans w ith a club Peter hit him self/H ans w ith a club.’ d. Peter brcendte sig/sig selv/Hans med en cigaret. (cf. (40b)) P eter burned R E F L /R E F L self/H a n s w ith a cigarette ‘Peter burned him self/H ans w ith a cigarette.’ Rather than assum ing the existence o f tw o different sig’s we attribute the difference between (37) vs. (38-40) to differences w ith respect to w here in the derivation sig initially m erges. As m entioned above, the sig in (37) m erges in object position o f the verb w here it receives the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 internal theta-role. This option is not available for the unaccusative predicates in (38-40) w here it is the subject w hich initially m erges in object position receiving the unique theta-role o f the predicate before it m oves up to the subject position. We propose that in these cases sig does not receive any theta-role since it m erges directly with the small v o f the vP projection responsible for the voice properties o f the predicate. W hile we do acknow ledge that tw o different uses o f sig m ust be recognized we strongly disagree with the claim that them atic sig is anti-local (cf. V ikner (1985), Lidz (2001), etc.) or that it may only be locally bound be predicates which are lexically m arked as [+inherently reflexive] (cf. Reinhart & Reuland (1993)). That is, we argue that them atic sig can be locally bound with certain predicates, e.g. neutral predicates like vaske ‘w ash’ in (37). Both V ikner’s and Reinhart & R euland’s approaches run into problems with (37a). In order to account for the absence o f overt reflexive m arking (by the elem ent selv ‘self) in (37a) R&R have to assum e that vaske ‘w ash’ is lexically [+inherently reflexive] in (37a) but not in (37b,c). Since there is no independent evidence for the existence o f double entries for the verb vaske ‘w ash’ such a stipulation is best avoided. Indeed, as shown in (42) the only difference in m eaning betw een (37a) and (37b) is due to the adnom inal intensification o f sig, see (42a) vs. (42b). (42) a. Peter vaskede sig. ([+agentive], intentional event, [-contrast]) b. Peter vaskede sig selv. ([+agentive], intentional event, [-^contrast]) c. Peter vaskede Hans. ([+agentive], intentional event, [-contrast]) d. Peter vaskede Hans selv. ([+agentive], intentional event, [+contrast]) Peter w ashed REFL/REFL self/Flans/FIans self ‘ Peter w ashed (him self)/H ans (himself). ’ Since in our analysis the difference betw een (42a) vs. (42b) and (42c) vs. (42d) follows from the theory o f adnom inal intensification, there is no need for the ad hoc stipulation that Danish has double lexical entries for neutral predicates like vaske ‘w ash’. V ikner’s anti-locality approach does not fare m uch better. Basically, his proposal consists in assum ing that local sig Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 is “som e sort o f detransitivising elem ent that som ehow prevents the assignm ent o f the theta- role that would otherw ise have been assigned to the subject” (V ikner 1985:50, footnote 8). This analysis correctly predicts the non-agentive nature o f the sentences with sim ple sig in (38-40), i.e. these sentences are not com patible w ith purpose clauses (e.g. ‘in order to ’) or adverbials like deliberately. However, it has trouble explaining why both purpose clauses, e.g. ‘in order to ’, and adverbials like deliberately are ok in (37a). In contrast, the analysis proposed here correctly predicts that (37a) and (42a) should have agentive subjects since vaske is assum ed to rem ain the sam e transitive verb in all o f (37a-c) and assign both an A GENT theta-role (to the external argum ent Peter) and a TH EM E theta-role (to the internal object sig). In chapter 4 a num ber o f additional syntactic and sem antic tests will be adduced to support our proposal that them atic sig is neither anti-local nor needs to be reflexive-m arked by selv ‘s e lf to be locally bound. Finally, notice that in addition to avoiding unnecessary proliferation o f lexical entries for neutral predicates it also m akes it possible to assum e the existence o f only one sig w hich may merge in different positions in the derivation: (i) in object position in the case o f them atic sig, see (43a), and (ii) in the v o f the voice projection vP in the case o f non- them atic sig, see (43b). (43) a. Them atic s is ' . VP / \ .... V ’ / \ V sig b. N on-them atic x/g: vP / \ v’ / \ / \ VP sig V / \ .... V ’ / \ V .... Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 1.10 Self-nominalizations In chapter 7 the use o f the elem ent selv ‘s e lf in nom inalization will be discussed. It has been argued that selv ‘s e lf functions as a reflexive in nom inalizations such as selvrespekt ‘self- re sp e c f. That selv ‘s e lf should function as a reflexive in nom inalizations (rather than the sim ple reflexive sig ‘R EFL’, e.g. *sig-respekt ‘R E F L -respecf) has been taken as evidence in favor o f analyzing selv ‘s e lf as a reflexivizing elem ent (cf. Reinhart and Reuland (1993)) and could thus potentially be construed as an argum ent against the analysis defended here, namely that binding o f reflexives (i.e. sig) and the distribution o f intensifiers (i.e. selv) are controlled by tw o separate m odules o f the grammar. In chapter 7 these issues will be discussed and an alternative analysis will be proposed in which the elem ent selv ‘s e lf is consistently assum ed to be an intensifier even in its word-internal uses. 1.11 Unification of all uses of adnominal expressions Before m oving on to exploring the consequences o f extending the present analysis to English and M andarin Chinese, let us first sum m arize some its m ain advantages. First o f all, unlike current versions o f binding theory (e.g. R einhart & Reuland (1993), Lidz (2001), H uang & Liu (2001), etc.) the present approach provides a m orphologically fully transparent articulation o f interaction between intensification and binding: nominal expressions (be they DPs, reflexives, or pronouns) to w hich selv ‘s e lf has been adjoined, see (44) are all intensified in the sam e way and have to obey the contrastiveness condition on intensification (18). (4 4 ) S ystem o f in ten sification o f nom inal ex p ressio n s in D anish. U nintensified Intensified a. Reflexive sig ‘him self/herself/’ sig selv ‘him self/herself b. Subj. pronoun han ‘h e’ han selv ‘he h im self c. ObJ. pronoun ham ‘him ’ ham selv ‘him him self/him self d. DP kongen ‘the king’ kongen selv ‘the king him self Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 Furtherm ore, the present analysis avoids unw arranted proliferation o f lexical entries. M ost other approaches to binding and intensification all rely on the assum ption o f either (!) double lexical entries for neutral predicates (e.g. Reinhart & Reuland (1993)), or (ii) double/m ultiple lexical entries for intensifiers (e.g. V ikner (1985), Reinhart & Reuland (1993), or both. In the present approach selv is A LW A Y S an intensifier and thus falls under the binding-independent m odule o f intensification. The different subparts o f com plex reflexives (e.g. sig selv ‘REFL s e lf) and pronouns {han selv ‘he h im self) thus fall under different m odules, see (45). (45) Independence o f binding and intensification o f intensified nom inal expressions: B inding: Intensification: a. Intensified reflexives: sig (principle A (9a)) selv (contrastiveness condition (18)) b. Intensified pronouns: ham (principle B (9b)) selv (contrastiveness condition (18) c. Intensified DPs: F’eter (principle C) se/v (contrastiveness condition (18)) Finally, unlike Reinhart & R euland’s (1993) predicate-centered approach to binding we propose a nom inal approach in w hich binding is defined as the interaction o f intrinsic properties o f nom inal expressions and syntactic locality constraints, e.g. principle A and principle B, see (9a,b) and (45a,b). 1.12 Extending the analysis to Modern English The cross-linguistic applicability o f the analysis defended here is dem onstrated by showing how it can be extended to reflexives, pronouns and intensifiers in M odem English. We argue that the proposal in (46) make it possible to analyze the system o f intensified and unintensified nom inal expressions as having the same degree o f m orphological transparency as the Danish system in (44), see (47). (46) a. English x-self (e.g. himself, herself, etc.) are not reflexive anaphors but A LW A Y S adnom inal intensifiers (e.g. the king himself). b. M odem English has 0-reflexives, e.g. Peter, washes 0,. c. W hat looks like locally bound reflexives, e.g. Peter replaces him self are really locally bound adnom inally intensified 0-reflexives, e.g. Peter, replaces 0 , him self (cf. Siem und 2000 for an outline o f a sim ilar analysis) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 d W hat looks like locally free reflexives (also called “logophors”) are really intensified pronom inals whose pronom inal part is not realized phonetically (due to a phonological deletion rule), e.g. Peter, said that Mary danced with everyone but [himJhimself. (The square brackets contains the phonetically unrealized pronominal). (47) System o f intensification o f nominal expressions in M odern English. U nintensified Intensified Reflexive a. R eflexive 0 b. Subj. pronoun he c. Obj. pronoun him d. DP Peter 0 himself he himself [him] himself {<him himself) Peter himself We propose to consistently analyze him self as an intensifier everywhere it occurs. Such an analysis becom es possible if we assum e the existence o f a 0 reflexive in English. By assum ing that the internal argum ent position o f the verb wash in (48) is filled by a 0 reflexive in (48a(ii)) and an intensified 0 reflexive, i.e. 0 him self in (48a(iii)), we are able to unify the account o f intensified and unintensified reflexives in D anish and English: in both languages the intensified form s o f the reflexive pronouns are form ed by the sam e process o f intensifier adjunction w hich is responsible for intensification o f DPs like Peter selv ‘Peter h im self and which is subject to the contrastiveness condition (18). (ii) unint. refl. (48) N eutral verbs: (i) pron.: a. Pie, washes *himi / 0 , b Hani vasker *ham; / sig, (49) A nti-reflexive verbs: a. Hei suspects *him, / *0 , b. Han, mistcenker *hami / *sig, (50) Inherentlv reflexive verbs: a. He, rests *himi / 0 , b. Hani hviler *hami / sig. (iii) int. refl. 0 , him self sigi selv. 0 , himself, sigi selv. *0 , himself. *sigi selv. W e argue that the data in (48-50) show that English 0-reflexives, e.g. Peter, washed 0,, correspond to unintensified reflexives in Danish, e.g. Peter, vaskede sig, ‘Peter w ashed’. The English x-self form reflexives can thus be view ed as intensified versions o f the unintensified 0-reflexives, e.g. Peter washed 0 , him self ~ (D an.) Peter vaskede sig selv ‘Peter washed Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 REFL s e lf. Furtherm ore, the fact that, to a large extent, the distribution o f /2 /m.ve^ follow that o f the adnom inal intensifier selv in Danish also supports our claim that English x-self, e.g. himself, herself, ourselves, etc., are always intensifiers - never reflexives; ju st like selv ‘s e lf in M odern Danish, and self'm Old English. In a sim ilar vein, the neutralization o f the m orphological distinction between intensified pronouns and intensified reflexives in English (both appearing to be realized as himself), can be explained by assum ing the existence o f a deletion rule w hich erases redundant pronom inal elem ents. Since the English intensifier him self \s m orphologically specified for the same features as the pronoun him, the intensified pronoun resulting from the adjoining the form er to the latter w ould end up being overtly m arked tw ice for the same features^. With such a rule it now becom es possible to analyze locally free instances o f him self (also called “logophors”) as intensified pronouns, e.g. [him jhim self on a par with their Danish counterparts, e.g. ham selv ‘him h im self, see (51) and (52). The pronom inal enclosed in square brackets e.g. [him jhim self indicates the m orphem e deleted by the m orphological reduction rule. This notation will be used hereafter to refer to the m orphologically reduced intensified object pronouns. (51) Peter, said that Mary danced with everyone except [h im j himself. (52) Peteri sagde at Marie dansede med alle andre end hami selv. Peter said that M ary danced w ith all others than him self ‘Peter said that M ary danced w ith everyone except h im se lf’ The fact that reduction of intensified pronouns applies to accusative/dative forms (e.g. him him self > [him] himself) but not to nominative forms (e.g. he himself) may be taken as indication that: (i) case features matters, or (ii) phonological identity (i.e. haplology) is at play. Both of hypothesis (i) and (ii) run into problems with forms like [us] ourselves and [you] yourself where the pronouns differ from the adjoined intensifiers in both case (us/you are accusative vs. our/your which are genitive, etc.) and phonological realization. As these examples illustrate, the relevant reduction rule is not fully understood at this point. See chapter 5 for more discussion of English intensifiers. Note also that Mandarin Chinese, which does not distinguish overtly between subject and object pronouns (e.g. Ta, hen ta^ ‘he hates him’), does not allow subject pronouns to be intensified by complex intensifiers (e.g. *Ta tazi]i hen wo ‘he himself hates me’). This may be construed as evidence supporting both (i) and (ii). See chapter 6 for more discussion of intensifiers in Mandarin Chinese. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 In both languages the intensified object pronouns are com posed o f a pronom inal + the adnom inal intensifier. The only difference is that the pronominal part o f the English intensified pronoun is rendered inaudible by the m orphological reduction rule (51), while it is phonetically realized in Danish (52). See chapter 5 for further discussion o f these and other issues related to the analysis o f M odem English. 1.13 Historical evolution of English reflexives and intensifiers In the second half o f chapter 5 we argue that the account o f M odem English proposed above makes it possible to understand hitherto unexplained facts in the evolution o f intensifiers and reflexives from Old English into M odern English, see (53). (53) a. The Old English intensifier 5e /f was fused with pronouns to form the com pound x-.se//intensifiers, e.g. him + self» himself. b. Old English lost locally bound pronouns and developed 0-reflexives to replace them , e.g. Peter, washed him, » Peter, washed 0,. b. The form ation o f \-s e lf forms also took place in the and 2"‘ ‘ persons (this is unexpected in m ost traditional accounts which assum e that the initial m otivation for adnom inal intensification o f locally bound pronouns in Old English was to disam biguate between disjoint reference and coreference readings). As shown in (54) Old English had a m orphologically simple intensifier self but lacked m orphologically specialized reflexives, using locally bound pronom inals instead (54a). (54) Stage one: system o f intensification o f nom inal expressions in Old E nglish'”. Unintensified Intensified a. Locally bound pron. him him s e lf b. Subject pronoun he he s e lf c. Locally free pronoun him him s e lf d. DP the king the king s e lf For ease of exposition, the forms in this table are adapted forms which have been made to be as similar as possible to Modern English forms by ignoring certain aspects o f Old English morphology and spelling. The Old English intensifier self was overtly marked for case agreement. Though it has been argued that case plays an important role in the historical evolution of intensifiers and reflexives in English (cf. Farr (1905), and E. van Gelderen (2000), among others) case has not been included here. See chapter 5, section 5.3 for more discussion of the role of case in the evolution o f intensifiers and reflexives in English. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 (55) System o f intensification o f nom inal expressions in M odern English (=(47)). U nintensified Intensified a. Reflexive 0 0 h im self b. Subject pronoun he he h im self c. O bject pronoun him [him] himself[<him himself) d. DP Peter Peter h im self The sentences in (56) illustrate local binding o f pronom inals in Old English. They also illustrate how predicate m eaning affects intensifier-adjunction (killing is an other- directed/anti-reflexive activity while defending is neutral). (5 6 )'' a. Hine, he/beweradh m id wcepnum. (O ld English) him he defended with weapons ‘Hcj defended himselfj w ith w eapons.’ b. Hannibal/ ... hine/ selfne mid atre acwealde. (O ld English) Hannibal him self w ith poison killed ‘Hannibal killed him self with poison.’ In chapter 5 we argue that assum ing the system o f intensified and unintensified nominal expressions in M odern English to be as outlined in (55), allows us to explain the evolution o f the O ld English (54) into M odern English in a rather straightforw ard way as a consequence o f replacem ent o f locally bound pronouns w ith 0 -reflexives at a certain point in history, com pare (54a) and (55a). As a result o f the substitution o f locally bound pronouns w ith 0-reflexives, in Old English the form him s e lf was reanalyzed as an intensifier. 1.14 E x ten d in g th e analysis to M a n d a rin C hinese In chapter 6 we argue that the analysis based on the independence o f intensification and binding can be extended to reflexives and intensifiers in M andarin Chinese. The Chinese intensifier elem ent ziji g ‘self-self has a great deal in com m on with M odern English him self and it will be shown that these sim ilarities can be used to argue that the 0 -reflexive analysis o f M odern English proposed in chapter 5 may be extended to Chinese. First, like the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 English adnom inal intensifier him self (57) the Chinese adnominal intensifier § ziJi also appears to be able to function as a reflexive anaphor, see (58). (57) 1 = 1 T ° (ziji = adnom inal intensifier) Huang-di ziji lai-ie. em peror self-self com e-PERF ‘The em peror him self cam e.’ (him self = adnom inal intensifier) (58) M ^ 'E l = [ t 3 ° (zi}7 = reflexive) Huang-di hen ziji. em peror hate self-self ‘The em peror hates him self.' (him self = reflexive) However, as discussed above, the peculiar properties o f ziji and him self when allegedly used as reflexives, com pared with reflexive anaphors in other languages, e.g. Danish sig, D utch zich, French se, follow from the fact that they are A LW A Y S adnom inal intensifiers, w hich have long been mistakenly assum ed to be reflexive anaphors. The differences between, elem ents like Chinese ziji and English him self (w hich are both A LW AYS intensifiers) and true reflexives like Danish sig and D utch zich are illustrated by the fact is that ziji and him self do not have any o f the non-reflexive uses (e.g. reciprocal, m iddles, m edio-passives, unaccusatives, deponent verb, etc.) displayed by sim ple reflexive elem ents many other languages, e.g. Dan. sig, Ger. sich, see (59b-e). " Examples (56a-b) are from Zupitza, J. (ed.) (1966) Aelfrics Grammatik und Glossar, Berlin: Wiedmannsche Verlagsbuchhandlung, 96.11, and Bately, J. (1980) The Old English Orosius, Oxford University Press, 4 11.110.2 respectively. The same examples are also discussed in Siemund 2000:25-26. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 (59) Typological survey o f reflexive and non-reflexive uses o f reflexive elem ents'^: C hinese E nglish F re n c h G e rm a n D anish a. reflexive ziji him self se/soi sich sig b. reciprocal - se sich -s d. middle - se sich - d. m edio-pass. - - se sich -s e. deponent - se sich -s/sig f. intensifier ziji him self - - - Furtherm ore, as show n in (59f), both English him self and Chinese ziji can function as intensifiers, som ething the r e f l e x e s in Danish, French, Germ an and Russian are unable to do, see (59f). So except for the' reflexive uses, Chinese ziji and English him self are basically in com plem entary distribution with the Danish reflexive sig (as well as with Ger. sich, Fr. se, Russian -sja). On the other hand, as discussed in chapter 2, section 2.9, Chinese ziji and English him self display m ost o f the uses o f the D anish intensifier selv, see (60). (60) Typological survey o f different uses o f adnom inal intensifiers: Chinese English French Danish a. Adnominal intensifier guo-wang (ta) ziji the king himself le roi (luijmeme kongen selv b. Complex reflexive 0 ziji 0 himself soi-meme/lui-meme sig selv c. Doppelganger-marker 0 ziji 0 himself soi-meme sig selv d. Intensified pron,/logop. 7to7z//iV to z/ y7 [him jhim self [lui] lui-meme ham selv e. Exclusive adverb, inten. ziji himself lui-meme selv We take the com plem entary distribution o f ziji and him self v?, . sig, see (59), as well as the sim ilar distribution o f ziji and him self vs. selv, see (60), as strong support o f the assum ption that both ziji and him self always are intensifiers rather than reflexive anaphors. The apparent reflexive uses o f ziji and him self m (58)/(59a) can be explained if we assum e that Chinese and English both have 0-reflexives w hich can be intensified by the adnom inal intensifiers ziji and /z/ffive//respectively, see 0 him self m A 0 ziji in (60a). In other words, the absence o f non- them atic uses o f ziji and him self plus their consistent behavior as intensifiers provide strong Danish has two reflexive elements: (i) the ‘free form’ reflexive sig ‘REFL’, and (ii) the bound morpheme which is a medio-passive/reciprocal suffix derived historically from an encliticized form of the Old Norse reflexive pronouns sik ‘REFL’. A similar situation obtains in Russian which also have two reflexive elements: (i) the ‘free form’ reflexive sebja ‘REFL’,. and (ii) the bound morpheme -sja which is also a middle/reciprocal suffix derived historically from an encliticized form of sebja ‘REFL’. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 support for the analysis o f intensified and unintensified nominal expressions in Chinese, outlined in (61). (61) Systems o f intensified and unintensified nominal expressions in Chinse, Danish and English: Sim nle/unintensified Com plex/intensified a. Reflexives (i) Chinese 0 0 ziji (ii) Danish sig sig selv (iii) English 0 0 himself b. Obj. Pron. (i) Chinese ta (ta) ziji (ii) Danish ham ham selv (iii) English him [him]‘^ himself c. Subj. Pron. (i) Chinese ta ta ziji (ii) Danish han han selv (iii) English he he himself d. DPs (i) Chinese huang-di huang-di ziji (ii) Danish kejseren kejseren selv (iii) English the emperor the emperor himself The analysis o f Chinese outlined in (61), w hich yields a m orphological transparent system o f adnom inal intensification, is based on the hypotheses listed in (62). (62) a. Chinese ziji is not a reflexive anaphor but rather an adnom inal intensifier (e.g. huangdi ziji ‘the em peror himself). b. Chinese has 0-reflexives, e.g. Peteri shu 0,- [de] tou ‘Peter com bed c. W hat looks like locally bound reflexives, e.g. Peter, hen ziji, ‘Peter hates h im se lf, is really locally bound adnom inally intensified 0 -reflexives, e.g. Peter i hen 0,- ziji ‘Peter hates &, h im se lf. d. W hat looks like locally free reflexives (also som etim es called “logophors”) is really adnom inally intensified pronom inals, e.g.: Peter M ary ) g “ Peter; shuo Mary gen chu-le (tUi) ziji yi-wai suo-you de ren tiao-wu P say M w ith except he self-self INST-outside that-have DE people dance ‘Petep said that M ary danced with everyone except [himi] h im self.’ The material enclosed in square brackets is assumed to have been rendered phonologically zero by a morphological deletion rule which will be discussed in more detail in chapter 5 and 6. As discussed in chapter 6, the clearest evidence for the existence of 0-reflexive in Mandarin Chinese are found in inalienable constructions in which the 0-reflexive occurs in possessor position: (i) a. b. w P M 'm p-m m » Zhangsart; shu xiao haizi^ Zhangsan comb small child Zhangsart; shu 0 , ziji Zhangsan comb self-self WPjfM - Zhangsan; shu 0, Zhangsan comb 0 de tou. DE head de tou. DB head [de] tou. head ‘Zhangsan combs the child’s/his (own) head.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 In addition to exploring the pros and cons o f adopting the analysis outlined above for M andarin Chinese, chapter 6 will also contain discussion o f w hether Chinese allows for adnom inal intensification o f PRO or if sentences like (62d) should be analyzed as having intensified 0 subject pronouns. 1.15 Conclusion The m ain proposals o f the approach to binding and intensification advocated in this dissertation are sum m arized in (63). (63) M ain proposals: a. Binding and intensification belong to separate modules o f the grammar. b. A dnom inal intensification is subject to pragm atic/sem antic factors, should be given a focus-based account, see (18). c. A nom inal rather than a predicate-based approach to binding is necessary. The distribution o f nom inal expressions is determ ined by the interaction o f the feature properties o f nom inal expressions and syntactic locality constraints (principles A and B), see (9)-(10). d. Com plex reflexives are intensified nom inal expressions on a par with intensified DPs. That is, we propose a unified analysis o f all uses o f adnom inal intensifiers (e.g. Dan. kongen selv ‘the king h im se lf, and sig selv ‘REFL s e lf = Eng. the king himself, 0 h im self etc.). e. Intensification o f sim ple reflexives is determ ined by a com bination of: (i) pragm atics, i.e. background knowledge o f participants in the utterance situation, and (ii) predicate meaning. f. (le ) leads to a tripartite typology o f predicates: (i) anti-reflexive predicates, which presuppose (representational) non-identity o f their arguments, (ii) inherently reflexive predicates, w hich presuppose identity o f their argum ents, and (iii) neutral predicates, which evoke no such presuppositions. Additional results o f the analysis proposed here are listed in (64). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 (64) a. Contrary to w hat is often proposed in the literature, the sim ple reflexive sig ‘R E FL ’ in D anish Dutch zich, Nor. seg, etc.) is not-anti-local. b. Local sig can be a true theta-role receiving anaphor. c. Unification o f reflexive and non-reflexive uses o f sig: the sam e elem ent is used in both cases but it m erges in different places in the derivation: reflexive, theta-m arked sig m erges as internal argum ent o f the predicate, non-reflexive sig m erges w ith the head o f the vP projection responsible for the diathesis o f the sentence. In term s o f its feature com position it is the sam e sig in both cases. As m entioned throughout this introductory chapter m any o f the ideas presented in this dissertation have been proposed before. H owever, they have rarely been elaborated in any detail and they have never before been put together to form a coherent and com prehensive framework. Furtherm ore, the cross-linguistic validity o f the analysis o f intensification and binding presented here has been illustrated by outlining how it can be extended to both M odern English and M andarin Chinese. Finally, the 0-refiexive analysis o f m odem English proposed here receives further support from the fact that it makes it possible to explain the historical evolution o f intensifiers and reflexives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 Chapter 2 Intensiflcation 2.1 Introduction This chapter is dedicated to the theory o f intensification. Basing the discussion m ainly on Danish and English data, we propose a focus-based analysis o f adnom inal intensifiers, e.g. him self and selv ‘s e lf in (1)'. (1) Intensified D Ps: a. Peter selv deltog i modet. Peter self partook in m eeting-the ‘Peter him self attended the m eeting. b. Peter himself attended the meeting. The sem antic contribution o f adnom inal intensifiers has been previously thought to involve scalar ordering o f focus alternatives (as argued by Eckardt 2001), or to be reducible to central ity-effects (as argued by K onig (1997) and Siem und (2000)). Instead, we propose that it is sim ilar to that o f focus, w hich - depending on the context - may or may not involve scalar ordering o f focus-generated alternatives. In certain respects, the present analysis is close to Baker (1995), whose analysis o f adnom inal intensification is based on prom inence and contrastiveness. But unlike Baker (1995) w hose analysis includes both a prom inence and a contrastiveness condition on intensification, we propose to reduce the analysis o f intensification to a single condition, viz. the contrastiveness condition in (2). (2) Contrastiveness condition on adnom inal intensification: A nom inal expression (DP, pronoun, reflexive) can only be intensified adnom inally if it can be contrasted with other expressions in the context in w hich it is found. ’ As noted by Siemund (2000:16), adnominal intensifiers such as English himself, Danish selv, German selbst, etc., are not the only expressions used as intensifiers in these languages. Indeed, these languages all have a number of expressions which can be used for the purpose of emphasis or intensification, see (i), (ii), which are based on Siemund (2000:16, (2.28) and (2.29)), and (iii). (i) English: personally, in person, in itself, x-self, etc. (ii) German: personlich, hochstpersonlich, in Person, an sich, leibhaftig, in sich, von sich aus, etc. (iii) Danish: personlig ‘personally’, i egen person ‘personally, in person’, i sig selv ‘in itself, etc. Needless to say, a comprehensive theory of intensification would have to include all these expressions. However, this would far exceed the scope of the present dissertation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 The analysis o f the m orphology, syntax and sem antics o f adnominal intensifiers is presented in section 2.2. The advantage o f the analysis o f adnom inal intensification proposed here is that it can be extended to uses o f selv ‘s e lf w hich have usually been considered to be different from adnom inal intensifiers, see (3a, ii-iv). (3) D ifferent uses o f selv. Examples: Section: a. adnom inal intensification i. intensified DPs (1) (sect. 2.2) ii. intensified reflexives (4) (chap. 3, section 3.3) iii. intensified pronouns (5) (chap. 3, section 3.4) iv. q-floated intensifier (6) (sect. 2.2.5) b. selv in secondary predication constructions (7) (sect. 2.3) c. scalar additive focus particle ‘even’ (8) (sect. 2.4) d. selv as a noun (9) (sect. 2.5) e. adjectival form s o f the intensifier (10) (sect. 2.6) f. selv in nom inalizations (11) (sect. 2.7, chap. 7) A s indicated in (3a), we propose to unify as cases o f adnom inal intensification a num ber o f uses o f selv ‘s e lf hitherto analyzed as different phenom ena. M ost im portantly, in chapter 3 we argue that the so-called com plex reflexives (3a,ii), e.g. Dan sig selv ‘REFL s e lf, see (4), and com plex pronouns (3a,iii), e.g. Dan. ham selv ‘(him ) h im self, see (5), should be analyzed as intensified nom inal expressions on a par with intensified DPs (3a,i), e.g. Dan. Peter selv ‘Peter h im se lf, see (1). (4) Intensified reflexives: Peter mistcenkte sig selv. Peter suspected REFL self ‘Peter suspected h im se lf’ (5) Intensified pronouns: [..] pa onsdagflyver Jakobsen til Malaysia fo r at forhandlei hvad der ifolge hamj selv kan blive til kontrakter pd 300 km kystsikring. (JP July 20, 2003) on W ednesday flies Jakobsen to M alaysia for to negotiate w hat there according-to him self can becom e to contracts on 300 km coast securing ‘W ednesday Jacobsen flies to M alaysia to negotiate what according to (him ) him self could turn into contracts for 300 km coast securing.’ Since intensified reflexives and pronouns are dealt w ith in detail in chapter 3, they will only be m entioned cursorily in the rem ainder o f this chapter, w hich focuses m ore on the other uses o f Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 intensifiers. For exam ple, in section 2.2.5 we argue that what has been referred to as an “inclusive adverbial intensifier” (cf. Konig (1997)) m eaning ‘also, to o ’, com pare (6a-b) and (6c), should be analyzed as a quantifier-floated adnom inal intensifier (3a,iv). (6) Q uantifier-floated adnom inal intensifiers: a. Peter, var [t, selv] blevet nomineret til en Oscar. Peter was self becom e nom inated to an O scar ‘Peter had him self been nom inated for an O scar.’ b. Peter, had [t, himselfl been nominated fo r an Oscar. c. Peter had also been nominated fo r an Oscar. In addition to show ing that the phenom ena in (3a,i-iv) are all instances o f adnom inal intensification falling under the contrastiveness condition in (2), we propose that the analysis o f intensification presented here has the potential to be extended to all the non-adnom inal uses o f selv ‘s e lf, see (3b-f). C ross-linguistically the elem ents used as adnom inal intensifiers are also frequently found in a num ber o f other uses, e.g. as so-called “exclusive adverbial intensifiers” m eaning ‘by oneself, w ithout help’ (3b), see Eng. him self and Dan. selv in (7), and as additive scalar focus particles m eaning ‘even’ (3c), see Fr. meme ‘sam e’ and Dan. selv in (8). (7) Intensifiers in secondarv predication constructions (“exclusive adverbial selV Y a. Peter skrev stilen selv. Peter w rote essay-the self ‘Peter w rote the essay h im se lf’ b. Peter wrote the essay himself. (8) Intensifiers used as scalar additive focus particles ‘even’: a. Selv Peter kom. self Peter came ‘ Even Peter came. ’ b. Meme Pierre est venu. sam e Peter is come ‘Even Peter cam e.’ In section 2.3 we propose that w hat has som etim es been referred to as an exclusive adverbial intensifier m eaning ‘by oneself, alone’ (7a-b) is in fact the intensifier selv occurring in a secondary predication construction (3b). In section 2.4 we propose to analyze the intensifier Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 4 selv ‘h im se lf and the focus particle selv ‘even’ (8a-b) as the sam e elem ent w hich assum es different properties w hen occurring in different positions. In sections 2.5 and 2.6, the nom inal and adjectival uses o f selv (3d-e), illustrated in (9)-(10), will be argued to be m orphological variants o f the sam e intensifier elem ent selv ‘s e lf. (9) N om inal uses o f intensifiers: Selvet er en central del a f psyken. self-the is a central part o f psyche-the ‘The self is a central part o f the psyche.’ (10) Adiectival uses o f intensifiers: Selve huset er ikke meget vcerd. se lf house-the is not m uch worth ‘The house itself isn’t worth m uch.’ Detailed treatm ent o f the use o f the elem ent selv in nom inalizations (3f), briefly described in section 2.7, will be postponed till chapter 7, where we argue that word-internal uses o f selv follow the sam e pattern as the adnom inal intensifier (3a), see (11a), or as selv used in secondary predication constructions (3b), see (11b). (11) Intensifiers in nom inalizations: a. selvrespekt self-respect ‘self-respecf b. selvbygger self-builder ‘person building a house by him self/herself Section 2.8, com pletes the survey by listing and briefly discussing a few idiom atic uses o f the elem ent selv ‘s e lf. Finally, in section 2.9 a mini typological survey o f the ranges o f uses o f intensifier elem ents in Danish, English, French, and Chinese, is given. In brief, as sum m arized in section 2.10, the contribution o f this chapter is threefold. First, in section 2.2, we propose a focus-based analysis o f adnominal intensification based on the contrastiveness principle in (2). Second, in sections 2.2.5-6, we claim that many Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 phenom ena w hich have hitherto been considered unrelated, i.e. (3a,i-iv), can be accounted for w ithin this analysis o f adnom inal intensification. Third, in sections 2.3-8, we suggest that the lexicon contains only one elem ent selv ‘s e lf w hich can take on different syntactic functions (and yield different sem antic contributions) when found in different positions in the derivation, i.e. (3b-c). This proposal makes it possible to unify the analysis o f both adnom inal (3a,i-iv) and the so-called non-adnom inal uses (3b-f) o f the intensifier selv ‘s e lf in Danish. 2.2 Adnominal intensification In this section the m ain focus will be on adnom inal intensification o f full lexical DPs. A fter going through the m orphological, syntactic and sem antic characteristics o f adnom inal intensifiers, a focus-based analysis o f adnom inal intensification will be proposed. The rest o f this section is organized as follows. First in section 2.2.1 and 2.2.2 the m orphology and syntax o f adnom inal intensification will be discussed. Then, in section 2.2.3, a sem antic analysis o f adnom inal intensifiers will be outlined. 2.2.1 The morphology of adnominal intensification This section provides a b rief description o f the basic m orphological properties o f adnominal intensifiers in Danish and English. English intensifiers are in association with a DP with w hich they display agreem ent in the phi-features, i.e. person, num ber and gender, see (12). ( 12) English Danish a. / myself jegselv b. you yourself du selv c. we ourselves vi selv d. them themselves de selv e. the king himself kongen selv f. Maria herself Maria selv g- etc. etc. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 6 In Danish, the intensifier selv ‘s e lf is m orphologically invariable and it does not decline to m atch the features o f the nom inal elem ent that it is intensifying, as show n in (12). O f the Scandinavian languages, N orw egian (bokm al) behaves like Danish in that it has uninflected intensifiers o f the G erm an type. In contrast, Swedish and Icelandic inflect their intensifiers. See the Swedish exam ples in (13) in w hich the adnom inal intensifier sjdlv ‘s e lf displays agreem ent in gender and number. (13)^ a. Hon dr godtheten sjdlv. she is goodness-the self-SIN G -CO M M O N G EN D ER ‘She is kindness its e lf’ b. Barnet sjdlv-t sa ingenting. child-the self-SlN G -N EU TR said nothing ‘The child him self said nothing.’ c. Barnen sjdlv-a bestdmde. children-the self-PLU R decided ‘The children them selves decided.’ A ccording to typological surveys (cf. Siem und 2000:10-11), intensifiers with overtly m arked agreem ent (e.g. English himself, French lui-meme, Russian sam-, etc.) are more frequent than m orphologically invariable intensifiers (e.g. D anish selv, French -meme, G erm an selbst, etc.)^. The above discussion o f the m orphology o f intensifiers naturally leads us to the question w hat class o f linguistic expressions adnom inal intensifiers belong to. W hile in the Slavic and Rom ance languages adnom inal intensifiers tend to display adjectival behavior, intensifiers in the G erm anic languages primarily'* behave as adnom inal adjuncts (e.g. Ger. selbst, Dan. selv). These differences are w itnessed by the fact that, G erm anic intensifiers tend ^ These Swedish examples are adapted from Siemund, 2000:8 and pp. 17-22. ’ In Old English the intensifier se o //‘self displayed overt case and number agreement. See chapter 5, section 5.3. for more discussion of intensifiers in Old English, and the (hotly debated) role of case agreement in the evolution of intensifiers and reflexives. ‘ 'N ote, however, that, as discussed in section 2.6, Danish also has pre-nominal, inflected adjectival intensifiers, e.g. (i): (i) Selve dronningen self-e queen-the ‘the queen herself Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 to allow nom inal uses*, see the Danish exam ples in (14), while R om ance intensifiers tend to allow adjectival uses but disallow nom inal uses, see the Spanish exam ples in (15). (14) a. Selvets self-the-PO SS concept ‘the conception o f self (/the ego)’ b. Selvet er en central del a f psyken. self-the is a central part o f psyche-the ‘The self is a central part o f the psyche.’ c. Kun levende vcesner kan have et selv. only living creatures can have a self ‘Only living creatures can have a s e lf ’ (15) a. el mismo coche the sam e car ‘the sam e car’ b. la misma casa the same house ‘the sam e house’ c. *e/ mismo/*la misma the self In (14c) the D anish intensifier selv occurs as a noun preceded by the indefinite article et ‘a (n e u t, sing.)’. The sentences in (14a-b) further illustrate that selv behaves ju st like any other noun by taking the suffixal definite article ~et ‘the (neut., sing.)’, see (14a-b), and the suffixal - s genitive, see (14a). These facts have som etim es been adduced as evidence supporting the classification o f selv as a nom inal expression. In contrast, the Spanish exam ples in (15a-b) have been taken to show that Spanish intensifiers are a special kind o f adjectives w hich cannot be used as nouns m eaning ‘the s e lf, see (15c). However, rather than entering the som ew hat futile debate about w hat word-class the elem ent selv ‘s e lf belongs to, we sim ply take the ^ In Modern English it is a subpart of the adnominal intensifier, i.e. -self, rather than the entire complex intensifier, e.g. himself, which is used independently as a noun, see (1). See section 2.3 for more discussion intensifiers used as nouns. (i) The *oneself/*himself/self is a central part o f the psyche. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 elem ent selv ‘s e lf to be a m orphem e/root w hich can assum e nominal, adjectival, or adverbial properties depending on w here in the derivation it is inserted'". 2.2.2 The syntax of adnominal intensification In this section the syntax o f the use o f the adnom inal intensifier selv ‘h im self with full lexical DPs will he described in some detail. The discussion o f adnominal intensification will he structured around the questions in (16). (16) The svntax o f adnom inal intensification: a. A re there any constraints as to w hat syntactic positions intensified nominal expressions may occur in? h. W hat is the exact nature o f the syntactic relation (i.e. com plem entation, adjunction, etc.) betw een the adnom inal intensifier selv ‘h im se lf and its associate, i.e. the DP w hich it is intensifying? 2.2.2.1 Possible syntactic positions of adnominally intensified nominal expressions Let us now turn to the question o f w hether there are any constraints as to w hat syntactic positions intensified nom inal expressions may occur in. The answ er seem s to he no: apparently nominal expressions intensified by adnom inal selv can occur in the same range o f syntactic positions as their unintensified counterparts, see the Danish exam ples in (17)-(23): (17) Intensified nom inal expressions in suhiect position: Dronningen selv holdt talen. queen-the self held speech-the ‘The Queen herself held the speech.’ (18) Intensified expressions in direct object position: Jeg kom ikke til at interviewe kansleren selv; je g kom kun til at tale med hans kone. 1 cam e not to to interview chancellor-the self; 1 cam e only to to talk to his wife ‘1 did not get to interview the chancellor him self; 1 only got to talk to his w ife.’ * In this respect we differ from Safir (1996) who argues that the distinction between SELF- and SAME-type intensifiers is crucial to understanding the semantic differences between, for example, Germanic type SELF- intensifiers, which tend to behave as nouns, and Romance type SAME-intensifiers which tend to behave as adjectives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 9 (19) Intensified nom inal expressions in indirect object position: Jeg gav ikke Dronningen selv blomsterne; je g gov dem til hendes kammerpige. 1 gave not Q ueen-the self flowers-the; 1 gave them to her cham ber-m aid ‘1 d idn’t give the Queen herself the flowers; I gave them to her m aid.’ (20) Intensified nominal expressions as objects o f prepositions: Jeg gav blomsterne til Dronningen selv. 1 gave flow ers-the to Q ueen-the self ‘I gave the flowers to the Q ueen h e rs e lf’ (21) Intensified nom inal expressions in clefts: Det var Dronningen selv, som modtog os. it was Q ueen-the self who received us ‘It was the Queen herself who received us.’ (22) Intensified nominal expressions in topicalizations: D ronningen selv sa vi ikke. Vi sa kun prinsene. Q ueen-the self saw we not we saw only princes-the ‘The Queen herself we didn’t see. We only saw the princes. For DPs in possessor position the intensifier selv ‘h im self assumes the suppletive form egen ‘ow n’, see (23). (23) A rgum ent DPs : Possessor D Ps: a. Form o f intensifier: .selv egen ‘s e lf h im self ‘ow n’ b. Exam ple kongen selv kongens egen hund king-the self king-the-PO SS own dog ‘the king h im self ‘the king’s own dog.’ Adnom inal intensification in possessive constructions will be discussed in m ore detail in chapter 3, section 3.3.5. Though there are subtle differences between egen ‘ow n’ and selv ‘s e lf, we will argue that they are not due to lexical differences betw een these tw o forms, but rather are parasitic on the specific syntactic and sem antic properties o f possessive constructions. The only argum ents inside nom inalizations w hich are intensified are 0 - elem ents, see discussion in chapter 7 w here we argue that nom inalizations such as selvrespekt ‘self-respecf in (1 la) above contain a 0-reflexive intensified by the adnom inal intensifier selv ‘s e lf. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 As the above descriptive survey shows, selv may intensify different types o f nominal expressions’ occurring in a wide range o f syntactic positions, e.g. subject, direct object, indirect object, prepositional com plem ent, possessor position, etc. This leads us to conclude that there are no significant syntactic constraints as w hat positions intensified nom inal expressions may occur in (cf. Siem und (2000))*. 2.2.2.2 The syntactic relation between intensifier and its focus: adjunction Adnom inal selv forms a constituent with the DP it m odifies. This can be shown by fronting the w hole constituent in a cleft construction as in (24). (24) Det var Kongen selv, der loftede klaveret. It was king-the self who lifted the piano ‘It was the King him self who lifted the piano.’ Secondly, D anish is a V2 language. This m eans that in root sentences, only one constituent can precede the verb. Dronningen ‘the Q ueen’ and selv in (19a) in section 2.2.2.1 must therefore belong to the same constituent, the subject DP. This kind o f evidence leads us to conclude that selv m ust be som e kind o f adnom inal m odifier form ing a syntactic constituent with its associate. But it still does not tell us w hether it is an adjectival m odifier’, an adjunct or som ething else. Unlike regular adjectives, w hich in D anish are always pre-nom inal, see (25), selv m ust occur after the full DP w hich it is m odifying, see (26). ’ We have already seen examples of adnominally intensified full lexical DPs (17-23), pronouns (5,12), and reflexives (4). Intensification of traces was illustrated in (6a-b) and will be discussed in more detail in section 2,2.5 below. The only nominal expression which appear to resist intensification is PRO. In section 2.2.4 we suggest that this may be due to either semantic or syntactic constraints (i.e. the PRO theorem). *Note however, that some exceptions to this generalization do exist. Edmondson and Plank (1978) observed that Intensifiers cannot felicitously adjoin to DPs which are vocative, subject of imperatives, or predicate nominals in Identity sentences. While we agree with these facts, we argue that they follow from semantic/pragmatics factors, viz. the contrastiveness condition on intensification discussed in section 2.2.3.2.2, rather than syntactic constraints. ^ See L. Sanchez (1995:167) who analyzes Spanish intensifiers, e.g. mismo, as adjectives “that may select DPs or VPs as their syntactic complements and receive an adnominal or an adverbial interpretation.” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 (25) a. den gamle m aud (26) a. *den selv mand the old m an the man self ’the old m an’ b. * den mand gamle b. (??)denm and selv‘^ ’ the man old that man self ‘that man h im self c. *manden gamle c. manden selv m an-the old m an-the self ‘the man h im self Furtherm ore, notice that selv, unlike adjectives, can m odify definite DPs, proper names, pronouns and even reflexives. All this clearly shows that selv does not behave like an adjective in Danish. In Danish, restrictive relative clauses and FPs are placed after the DP they modify, see (27a-b), as are DP com plem ents, see (27c). (27) a. skuespilleren som havde modtaget prisen actor-the w ho had received aw ard-the ‘the actor who had received the aw ard’ b. dronningen med den gyldne krone queen-the with the golden crow n ‘the queen with the golden crow n.’ c. dronningen a f Sverige queen-the o f Sweden ‘the queen o f Sw eden’ As illustrated in (28), the different post-nom inal m odifiers in (27) do not all display the same behavior w ith respect to adnom inal selv. (28) a ’'. 1 1 skuespilleren som havde modtaget prisen selv actor-the who had received aw ard-the self ‘the actor who had received the aw ard h im self In Danish the definite article can be realized in two ways: (1) as a suffixal determiner -en/-et ‘the’, or (11) as a free standing determiner den/del ‘the’. The determiner den/det ‘the’ is always used when the noun is modified by an adjective, see (25a). Unmodified head nouns and head nouns modified by relative clauses or PPs take the suffixal form of the determiner, see (27a-c). When a simple unmodified noun is preceded by the determiner den/det, the latter is usually interpreted as a demonstrative ‘this/thaf as in (26b). In an out-of-the-blue context, like (26), a noun preceded by a demonstrative is usually interpreted deictically. Given the fact that intensifiers usually only modify DPs whose referent has been introduced earlier in the same situation/discourse, the sentence in (26b) sounds rather odd. ’’ Needless to say, this example has an acceptable reading in which the intensifier selv functions as an ‘exclusive’ adverbial intensifier modifying the VP of the relative clause: ‘The actor who had received the award all by himself, without help from anybody else’. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 b. 77/*dronningen med den gyldne krone selv queen-the w ith the golden crow n self ‘the queen with the golden crown h erself c'^. dronningen a f Sverige selv queen-the o f Sweden self ‘the queen o f Sweden h erself If a DP is already m odified by a restrictive relative clause or a PP o f the with-type then it cannot felicitously be intensified by selv, see (28a-b). However, if a DP takes a com plem ent o f the of-typQ then it can be intensified by selv, see (28c). The exam ples in (27) and (28) indicate that selv is in com plem entary distribution with post-nom inal prepositional adjuncts o f the med-iypQ and restrictive relative clauses*^. On the basis o f such facts it seems plausible to suggest that selv also has the syntactic status o f a post-nom inal adjunct m odifying D Ps'''. G iven that selv form s a constituent w ith the DP, it m ust be the highest elem ent in the DP, see (29). (29)'^ The structure o f DPs intensified by selv: DP / \ DP selv I \ self N D ‘h im self hong -en king the ‘the king’ '^T his example, as well as the one in (28b), is ambiguous. In principle, the intensifier selv ‘self could be interpreted as taking Sverige ‘Sweden’ rather than dronningen a f Sverige ‘the queen of Sweden’ as its associate. Due to overt morphological agreement between intensifiers and their associates, English usually avoids this kind of ambiguity, e.g. the people o f Iraq themselves (from Business Week Feb 2003). Ambiguity only occurs when the embedded DP and the matrix DP share the same features, e.g. the discussion o f the issue itself (continued fo r several hours) (from Siemund (2000:157) example (6.94a)). Cf. Everaert (1986:37-38) who observed the same pattern in Dutch: “it appears that a post-nominal quantifier like zelf occupies the same position as met/zorrrfer-PP-complements and restrictive relatives”. Note that PPs and relative clauses do not behave alike with respect to adjunct-stacking. As shown in (i) sequences of PP+relative clause are ok while sequences of relative clause+PP are not. (i) a. The man with glasses who was bald. b. *The man who was bald with glasses. Some researchers now believe that natural languages never use right-adjunction of the type illustrated in (29). The question whether right-adj unction is among the structures available to natural languages seems to be orthogonal Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 W hile stacking o f adjuncts yields unacceptable structures in (28a-b) w here the intensifier selv is the last in a series o f adjuncts, it appears to be possible to construct acceptable exam ples involving adnom inal intensifiers and adjunct-stacking, provided that the intensifier precedes the other adjunct, see (30). (30) Dronningen selv sammen m edprinserne loftede klaveret. Q ueen-the self together w ith princes-the liftede piano-the ‘The Queen herself, together w ith the princes, lifted the piano.’ Notice, however, that in this respect the PP sammen med ‘together w ith’ behave differently from the other constructions involving PPs and relative clauses in (27), see (31). (31) a. * skuespilleren selv som havde modtaget prisen actor-the self who had received award-the ‘the actor him self who had received the aw ard’ b. dronningen selv med den gyldne krone queen-the self w ith the golden crown ‘the queen herself w ith the golden crow n.’ c. * dronningen selv a f Sverige queen-the self o f Sweden ‘the queen herself o f Sw eden’ The contrast betw een (30) and (31), w hich are all unacceptable, indicates that the PP sammen med ‘together w ith’ is different from the adjuncts in (31a-b) and the prepositional com plem ent in (31c). This difference is further confirm ed by the exam ples in (32) and (33). (32) Dronningen selv loftede klaveret sammen med prinserne, Q ueen-the self liftede piano-the together with princes-the ‘The Queen herself lifted the piano together with the princes.’ (33) a. * skuespilleren selv kom til modet som havde modtaget prisen actor-the self came to m eeting-the w ho had received award-the ‘the actor him self came to the m eeting who had received the aw ard’ b. * dronningen selv satte sig p d tronen med den gyldne krone queen-the self sat REFL on throne-the with the golden crown ‘the queen herself sat dow n on the throne w ith the golden crow n.’ to the issues discussed here. The reader is therefore free to translate the structure in (29) to ieft-adjunction plus movement, if he/she is so inclined. '^Notice that this sentence is ok when the PP with the golden crown is interpreted as a secondary predication construction on a par with, for example The Queen sat down on the throne wearing her regal robe. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 c. * dronningen selv satte sig p d tronen a f Sverige queen-the self sat REFL on throne-the o f Sweden ‘the queen herself sat down on the throne o f Sw eden’ Based on the above exam ples we conclude that the PP sammen med ‘together w ith’ in (30) and (32) is used as a VP-adverbial or secondary predicate (w hich can either precede or follow the VP) rather than as an adnom inal adjunct. Hence, it does not constitute a counter-exam ple to the generalization that PPs adjunct and relative clauses are in com plem entary distribution with the adnom inal intensifier .se/v ‘s e lf. 2.2.2.3 Unified analysis of adnominally intensified expressions Adopting a m odified version o f Longobardi’s (1994) analysis o f nominal expressions we assum e that all nom inal argum ents are projections o f a head D constituent. This approach makes it possible to arrive at a unified account o f the syntax o f nominal expressions, see (3 4 )'\ a. [dp [n kong ] [d -en ] ] ‘the king’ b. [dp [n konger ] [d 0 ] ] ‘kings’ c. [dp [d ham ] ] ‘him ’ d. [dp [d sig ] ] ‘R EFL’ The structures in (34a) and (34b) illustrate the syntax o f DPs with overt and null determ iner respectively. Pronouns and reflexives are assum ed to behave as determ iners, i.e. they are assum ed to head their own DPs, see (34c) and (34d)'*. In addition to m aking it possible to As mentioned above, Danish has both post-nominal and pre-nominal articles, see (i-ii). (i) Hus-et tiouse-the (ii) Det store hus the big house The question of which order is the basic order o f article and noun and which is the derived order seems unrelated to the issues discussed here. Hence, although it may be more correct to assume the post-nominal articles in (34) to be derived via movement from a pre-nominal determiner position, for ease of exposition we have decided to ignore these complications in the structures given here. As work by Postal (1966), Abney (1987), Longobardi (1994), and Uriagereka (1995) have shown, there is evidence that pronouns may function as determiners, e.g. the examples in (i-ii) below (from Radford (1997:154,(79a,b))). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 arrive at a unified analysis o f the syntax o f nom inal expressions, this approach has the advantage o f enabling us to unify the account o f adnom inal intensification. As discussed above, intensified nom inals such as kongen selv ‘the king h im self are best analyzed as simple adjunction structures, see (35a). a. [dp [dp [n kong ] [d -en ] [selv ]] ‘the king h im self b. [dp [dp [n konger ] [d 0 ] [selv ]] (??)‘kings them selves’'® c. [dp [dp [d ham ] ] [selv ]] ‘him s e lf d. [dp [dp [d sig ] ] [selv ]] ‘REFL s e lf As shown in (35c) and (35d) the com plex pronoun ham selv ‘him s e lf and the com plex reflexive sig selv ‘REFL s e lf can be analyzed as intensified DPs having the same syntax as intensified R-expressions. In other words, syntactically all types o f nom inal expressions, e.g. full lexical DPs, subject pronouns, object pronouns, possessive pronouns, and reflexive pronouns, behave alike with respect to adnom inal intensification^®. 2.2.3 Semantic and pragmatic aspects of adnominal intensifiers The literature on intensifiers contains a num ber o f different proposals o f how to best capture the sem antic and pragm atic properties o f intensifiers some o f w hich are listed in (36). (I) [IP [DP [D We] [N psychologists]] [P [I do n ’ t] [VP [V trust] [DP \Dyou] [N linguists]]]]] (ii) [IP [DP [D We] ] [P [I do n ’ t] [VP [V trust] [DP [Dyou] ]]]] ‘“ ’ Due to the unique indentifiabiliy requirement on adnominal intensification, discussed in section 2.2.3.2.1, bare plural nominals like konger ‘kings’ usually cannot be felicitously intensified by adjunction of the intensifier selv ‘self. As observed in Edmondson & Plank (1978:381), only if the indefinite DP is given a specific interpretation is intensification possible in such cases. This generalization begs the question whether non-overt nominal expressions like, for example, traces and PRO can be adnominally intensified. Traces do seem to be able to undergo intensification, see discussion o f q-floated intensifiers in section 2.2.5, and fronted intensifiers in section 2.2.6. For discussion of the absence of adnominally intensified PRO, see section 2.4, as well as chapter 5 (for English), and chapter 6 (for Chinese). See also chapters 5 and 6 for discussion of intensification of 0-reflexives (i.e. phonologically unrealized reflexive pronouns) in English and Chinese. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (36) a. 56 a. em phasis D irven (1973) b. expectedness/rem arkability Edm ondson & Plank (1978), Eckardt (2001) c. centrality K onig (1991) d. (discourse) prom inence Baker (1995) e. obviation Baker (1995) f. focus Konig (1991), Ferro (1993), Sanchez (1994) g- contrastiveness M cK ay (1991), B aker (1995) In section 2.2.3.1 a focus-based, non-scalar account o f the sem antic contribution o f adnominal intensifiers will be proposed. In section 2.2.3.2 sem antic constraints on intensified DPs will be discussed. 2.2.3.1 The semantic contribution the adnominal intensifier selv ‘him self In Danish, as well as in G erm an and French, the elem ent used as adnom inal intensifier, i.e. Dan. selv, Ger. selbst, and Fr. -meme, can also be used as an additive, scalar focus particle sim ilar to the particle even in English. In all three languages the adnom inal intensifier is post- nom inal while the focus particle ‘even’ is pre-nom inal, com pare the (a) and (b) exam ples in (37)-(39). (37) Danish: a. b. Kongen selv holdt en tale. king-the self held a speech ‘The king him self delivered a speech.’ Selv visevcerten holdt en tale. self janitor held a speech ‘Even the janitor delivered a speech.’ (38) G erm an: a. Der Konig selbst hielt eine Rede^'. the king self held a speech ‘The king him self delivered a speech.’ b. Selbst der Hausmeister hielt eine Rede. even the janitor held a speech ‘Even the janitor gave a speech.’ (intensifier) (focus particle) (intensifier) (focus particle) The German and Freneh examples in (38a-b) and (39a-b) are adapted from Eckhardt (2000:1(1), (2)). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 (39) French: a. Le roi (lui-)meme^^ a prononce un discours. (intensifier) the king him self has pronounced a discourse ‘The king him -self has delivered a speech.’ b. M im e le concierge a prononce un discours. (focus particle) even the janitor has pronounced a discourse ‘Even the janitor has delivered a speech.’ Inspired by the overt sim ilarity betw een the tw o uses o f the intensifier elem ents illustrated in (37-39) a num ber o f researchers have proposed to analyze the adnom inal intensifier as a scalar focus particle sim ilar to ‘even’. In the following, we argue that such analyses are m isguided and that adnom inal intensifiers should be given a non-scalar, focus-based analysis. Unlike the focus particle selv ‘even’, the adnom inal intensifier selv ‘h im se lf does not presuppose that the proposition is true for at least one other elem ent in the focus-generated set o f alternatives, see (40a) vs. (40b). (40) a. *Det var selv biskoppen der holdt gudstjenesten. (focus particle) It was self bishop-the who held service-the ‘*It was even the bishop who held the service.’ b. D et var biskoppen selv der holdt gudstjenesten. (intensifier) It was bishop-the self who held service-the ‘It was the bishop him self w ho held the service.’ The exam ple in (40a) shows that the focus particle selv ‘even’ cannot occur in clefts w hich presuppose uniqueness because that would lead to a contradiction o f the so-called existential presupposition it always evokes, see (44b). Since the adnom inal intensifier selv ‘h im se lf can occur in clefts, see (40b), we therefore conclude that it does not evoke this existential presupposition w hich is m andatory w ith the focus particle even. The exam ples in (41)-(42), w hich contain predicates presupposing uniqueness, further confirm this difference between even and the adnom inal intensifier. In certain conservative registers of French, the complex adnominal intensifier lui-meme can be found to alternate with the simple intensifier -meme. That is, this example can also be realized without lui-, i.e. Le roi meme a prononce le discours. See Martin (1975) and Anscombre (1973) for more discussion of lui-meme and -meme as intensifiers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 (41) a. USelv Dronningen gav nytdrstalen. even queen-the gave new -years-speech-the ‘#Even the Queen delivered the N ew Y ear’s Speech.’ b. Even the President delivered the inaugural speech. (42) a. Dronningen selv gav ny tar stolen. queen-the self gave new -years-speech-the ‘The Q ueen herself delivered the N ew Y ear’s Speech.’ b. The President him self delivered the inaugural speech. A ccording to Eckardt (2000) the sem antic contribution o f adnominal intensifiers can be captured by the analysis given in (43) which is identical to the analysis she proposes for the focus particle even, except for the absence o f the existential presupposition/im plicature, com pare (43) with the analysis o f even in (44). As m entioned in chapter 1, at this point no attem pt has been m ade to distinguish between presupposition and im plicature. The tw o term s are used interchangeably w ithout difference in m eaning in the rest o f this dissertation (43) Analysis o f the adnom inal intensifier himself. him self + S a. A ssertion: [[ S ]]° b. Scalar im plicature: (i) Vp g {[[ S ] ] ^ \[[ S ]]°}: [[ S ]]° <c p (44) Analysis o f the additive focus particle even: even + S a. A ssertion: [[ S ]]° b. Scalar im plicature (i) Vp g {[[ S ] ] ^ [[ S ]]°}: [[ S ]]° <, p Existential im plicature(ii) 3p g {[[ S ]]*^\ [[ S ]]”}: p = 1 Follow ing Rooth (1992), [[ S ]]° and [[ S ]]* ^ are used to distinguish betw een the ordinary sem antic value o f a sentence, i.e. its truth value, see (48a), and the so-called focus sem antic value o f a sentence. As illustrated in (48b), the focus sem antic value o f a sentence is a set o f propositions obtained by replacing the associate o f the intensifier with elem ents from a contextually defined set o f relevant alternatives. In (43-44b) the scalar ordering o f alternative propositions on a scale o f likelihood is expressed by the com parative operator <„ that is ‘p <c q ’ should be read as ‘the proposition p is more likely to be true than the proposition q ’. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 W e argue that the analysis o f adnom inal intensifiers in (43) is on the w rong track. W hile it is true that DPs adnom inally intensified by selv often refer to entities occurring in propositions which are unexpected or rem arkable in a given context, this is by no means always the case, see (45-46). (45) A very powerful earthquake struck the center o f Rome and the Vatican. a. The Pope him self perished in the rubble. b. UEven the Pope perished in the rubble. (46) We do not live in the suburbs any more. a. We live in Paris itself. b. #We even live in PARIS. Given that an earth quake does not differentiation between people o f high and low status or prom inence there is no reason to expect the Pope to be less likely to perish in an earth quake than any other person living close to the epicenter. Likewise, in (46) the place name Paris is not intensified because it denotes an unexpected or rem arkable location, but sim ply because it is contrasted with another location, viz. the suburbs. N otice that in both cases the adnom inal intensifier is perfectly acceptable while the focus particle even, which always carries with it a scalar im plicature, see (44b(i)), is not. We take this as evidence that the scalar im plicature in (43b) is N O T an integral part o f the sem antic contribution o f adnom inal intensifiers. Based on the exam ples discussed above, we conclude that the sem antic contribution o f the adnom inal intensifier involves neither a scalar im plicature (44b(i)), nor an existential im plicature, (44b(ii)), and is thus different from the focus particle even, w hich has both these im plicatures. The sem antics o f adnom inal intensification is sim ilar to that o f focus, see (47a)- (48). (47) a. [The king himself]t,- came to the meeting. b. The king came to the meeting. (48) a. O rdinarv sem antic value o f t47ak [[ [The king himselfjp cam e to the m eeting]]” = 1 iff the king came to the meeting. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 b. Focus sem antic value o f (47a): [[ [The king himself]F cam e to the m eeting ]] ^ = {came-to-the-meeting(x) \ x e D e) ‘the queen cam e to the m eeting’, ‘the prince cam e to the m eeting’, Like focus, the basic sem antic contribution o f adnom inal intensification consists in contrasting the referent o f the focused/intensified expression w ith a contextually determ ined set o f alternatives, see (48b). That is, in addition to having the same truth conditions as (47b), the sentences in (47a) also tells us that the king is contrasted with a contextually defined set o f alternatives who also did (or m ight have) com e to the meeting. The generation o f a set o f alternatives gives rise to contrastiveness condition on adnom inal intensification w hich will be discussed in section 2.2.3.2.2. 2.2.3.2 Semantic and pragmatic constraints on adnominal intensification In section 2.2.2.1-2 syntactic constraints on adnom inal intensification w ere discussed. In this section, sem antic/pragm atic constraints on adnom inal intensification will be discussed, namely the “unique identifiability” (section 2.2.3.2.1) and “contrastiveness” (section 2.2.3.2.2) conditions. 2.2.3.2.1 Unique identifiability Since M oravscik (1972) and Edm ondson and Plank (1978) it has been generally known that a DP m odified by an adnom inal intensifier m ust have certain referential properties. However, the exact form ulation o f this constraint still seems to elude the com m unity o f researchers w orking on intensification. Exam ples like (49) and (50) seem to indicate the existence o f a constraint requiring adnom inally intensified DPs to be definite. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 (49) a. *The LA philharmonic is looking fo r a truly gifted violinist him self b. *A truly gifted violinist him self would never miss note. (50) a. The Queen herself showed up to the conference. b. 1 meet the Queen herself at the conference. However, as Edm ondson and Plank (1978:382) observed, things are not that simple. In eertain cases form ally definite DPs, i.e. DPs preceded by the definite article the, cannot be felicitously intensified either, see (5Ic). (51)^^ a. We wanted to call the doctor. b. We wanted to call the doctor himself. e. * We wanted to call the doctor him self but we d id n ’ t know any. In (51a) the DP the doctor is am biguous betw een a specific reading (referring to a contextually identifiable individual) and a non-specific reading (referring to any doctor). By adding the adnom inal intensifier, as in (51b), the reading o f the doctor as non-specific is excluded. Hence, in (51c) adnom inal intensification is not felicitous. These exam ples clearly show that formal definiteness alone is not enough to qualify a DP for adnom inal intensification. In the preceding we have showed that certain definite DPs cannot be intensified. As the exam ple in (52) shows, in certain eases, even indefinite DPs can be intensified. (52) A B A A ll Cretans lie. Where did you hear that? a. A Cretan him self told me. b. Cretans themselves told me. Based on this and sim ilar exam ples Edm ondson and Plank (1978) concluded that indefinite DPs may allow adnom inal intensification if they are interpreted as having a specific reference. Does this m ean that specificity is the correct generalization? W hile a large num ber o f exam ples can be adduced in support o f this hypothesis, e.g. (53a), the existence o f exam ples like (53b), from Siem und (2000:162(6.105)) seems to refute it. For discussion of this example, as well as its consequences for the theory of intensification, see Edmondson and Plank (1978:382) and Siemund (2000:156(6.90)). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 (53) a. *Unicorns themselves exist. b. Man him self is a product o f his environment. It may be possible to explain the difference betw een (53a) and (53b) as follow ing from a sem antic difference betw een the DP unicorns, w hich is an indefinite plural DP, and the DP man, w hich can be argued to refer to a specific species (i.e. hom o sapiens) w hich can be contrasted w ith other species (e.g. cat, dogs, spiders, etc.)^'*. Siem und (2000), who discusses these problem s in great detail, concludes that the relevant constraint on adnom inal intensification should be form ulated as in (54). (54) U nique Identifiability C ondition: In order to be successfully intensified by an adnom inal intensifier a DP “m ust denote a uniquely identifiable referent w here referent can be understood in the broadest sense o f the w ord.” (Siem und (2000:154, 170(6.137)) As it stands this constraint is still form ulated in rather vague terms. It is thus clear that more works needs to be done to arrive a more precise understanding o f the linguistic principles involved^^ A t the end o f the follow ing section we argue that the condition in (54) m ay not be necessary since it can be shown to follow from the contrastiveness condition on intensification. 2.2.3.2.2 Prominence, centrality or contrastiveness In the previous section referential constraints on DPs undergoing adnom inal intensification were discussed. In this section, sem antic and pragm atic prom inence conditions on intensification will be investigated in some detail and it will be shown that different constraints form ulated in term s o f prom inence, centrality and contrastiveness, as well as the Unique Indentifiability Condition discussed above, see (54), can be subsum ed under one contrastiveness requirem ent. As suggested to me by R. Pancheva (p.c.), the choice of verb might also be (partially) responsible for the unacceptability of (53a). Hence, although it also includes the expression unicorns themselves, the sentences in (i) below does seem better than (53a): (i) Unicorns themselves are but a product o f our imagination. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 It has long been noticed that there is a strong tendency for adnom inal intensifiers to m odify DPs w hich denote prom inent individuals o f high status^*’, see (55). (55) a. dronningen selv queen-the self ‘the Queen h erself b. biskoppen selv bishop-the self ‘the bishop h im self c. statsministeren selv prim e m inister-the self ‘the prim e m inister h im self d. prcesidenten selv president-the self ‘the president h im self Data such as (55) has lead to the form ulation o f different prom inence or centrality requirem ents on adnom inal intensification. Baker (1995) proposes the follow ing condition on the use o f intensive NPs. (56) B aker’s (1995) Condition o f Relative D iscourse Prom inence: Intensive N Ps can only be used to m ark a character in a sentence or discourse who is relatively m ore prom inent or central than others. T h e n o tio n o f p ro m in e n c e is in ten d ed to c o v e r b o th cases o f lex ical p ro m in e n c e , e.g. (55), an d case o f c o n te x tu a lly d efin ed p ro m in e n c e , e.g. (57). A c c o rd in g to B ak er (1 9 9 5 :7 9 ), th e se n ten ces in (5 7 a) an d (5 7 b ) b o th in v o lv e c o n tra stin g th e c u sto d ia n s w ith o th e r e m p lo y ees. In sp ite o f th e se sim ilarities, in ten sificatio n o f th e D P the custodians is o n ly ac c e p tab le in (5 7 a), b u t n o t in (57b). (57) a. A ll o f the employees o f that company will have to appear before the grand jury, where they will be asked what they know about the alleged illegal trash disposal. The custodians (themselves) will testify late Thursday afternoon, the other employees on Friday. For more detailed discussion of these issues, see Edmondson & Plank (1978) and Siemund (2000). ^’ ’ Cf. Keenan (1994) who observes this tendency for the adnominal intensifier se(o)lfm Old English. Cf. also Moravcsik (1972). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 b. A ll o f the employees o f that company will have to appear before the grand jury, where they will be asked what they know about the alleged check-kiting scheme. The custodians {??themselves) will testify late Thursday afternoon, the other employees on Friday. The exam ples in (55) exem plify cases o f expressions which, by their m eaning alone, tend to refer to high-status individuals (e.g. kings, presidents, etc.) who, under normal circum stances, are prom inent in m ost contexts. In contrast. B aker’s exam ple in (57) contains an intensified DP, i.e. the custodians, denoting low status individuals. N evertheless, the exam ple in (57a) is ok since, in the given context, the D P the custodians can be construed as referring to a contextually prom inent group o f individuals. A ccording to Baker (1995:80), “the much greater naturalness o f the intensive in [(57a)] as com pared to that in [(57b)] follows directly from our perception that custodians are much m ore likely to play a central role in an episode o f illegal trash disposal than in an episode o f check-kiting” . Baker therefore claims that these exam ples strongly support the existence o f his prom inence condition on intensification. K onig’s (1998) notion o f centrality, see (58), is sim ilar to B aker’s (1995) prom inence condition. (58)” K onig’s ( 1998) Conditions for the use o f adnom inal intensifiers: A dnom inal intensifiers relate a center X (=referent o f the focus) to a periphery o f alternative values, such that: a. X has higher rank than Y in a real-w orld hierarchy. b. X is m ore im portant that Y in a specific situation. c. Y is identified relative to X (kinship term s, part-whole, etc..) d. X is the subject o f consciousness, center o f observation, etc.. The exam ples in (59a-d) illustrate the different kinds o f centrality defined in (59a-d). (59) a. The Pope him self is against this view. b. Nobody cared about the fans when the fire broke out, hut the rock .star him self was quickly whisked away. c. L u cy’ s sister is more intelligent then Lucy herself. d. Jemina guessed that Pompey had chivalrous doubts about leaving her in the gaunt building, with only Tiger, now in a restless mood, as company. She herself had no such fears. ' The centrality condition in (58) as well as the examples in (59) are from Konig (1997). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 Siem und (2000:154) w ho adopts a version o f K onig’s centrality requirem ent, adm its that the existence o f this constraint is very difficult, if not im possible, to test since “given a sufficient am ount o f adequate context or som e im agination on the part o f the decoder, alm ost any referent can be thought o f as being central and hence associated with another referent which forms its periphery” (Siem und (2000:154)). The elusive nature o f the prom inence and centrality requirem ents in (56), (58), as well as the virtual im possibility o f constructing any exam ple which clearly violate these constraints indicate that we would be well advised to look for a better form ulation o f the constraint or perhaps a different constraint altogether. M cKay (1991) offers a possible solution, see (60). (60) M cK av’s (1991:368) constraints on adnom inal intensification: In order to be successfully intensified by an adnom inal intensifier a DP m ust have a. “a clearly indicated referent”, and b. “a relevant contrast or com parison class” The constraint in (60a) corresponds to the unique identifiability requirem ent discussed in the previous section. The contrastiveness requirem ent in (60b) offers an interesting alternative to the prom inence and centrality requirem ents proposed by Baker (1995) and Konig (1997). Baker (1995) suggests that we need both the prom inence requirem ent in (56) and a contrastiveness requirem ent w hich he form ulates as in (61). (61) B akers’ (1995:77) contrastiveness condition: “ Intensive N Ps are appropriate only in contexts in which em phasis or contrast is desired” In the following, we argue that only M cK ay’s contrastiveness condition (60b) is needed. Furtherm ore, using evidence from inherently reflexive constructions we show that, unlike the prom inence and centrality requirem ents, solid evidence for the existence o f a contrastiveness requirem ent, as form ulated in (62), can be found. (62) C ontrastiveness condition on adnom inal intensification: A nom inal expression (DP, pronoun, reflexive) can only be intensified adnom inally if it can be contrasted with other expressions in the context in w hich it is found. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 6 Zribi-H ertz (1995:349-350) uses the incom patibility o f intensification with inherently reflexive constructions such as (63-64) as evidence supporting a scalar analysis o f intensifiers. In contrast, we propose to account for the absence o f intensified DPs in inherently reflexive constructions as a consequence o f the contrastiveness condition (62). (63) a. Bill, is out o f his 1 * 2 mind. b. ^Bill is out o f Jo h n ’ s mind. c. *Bill is out o f your mind. d. *Bill is out o f his own mind. (64) a. Billj took a knife with himj * 2 . b. *Bill took a knife with John. c. *5/7/ took a knife with you. d. *5/7/ took a knife with himself. Due to the inherently reflexive m eaning o f the constructions in (63-64), only expressions w hich are coreferential w ith the subjects are allowed. That is, the pronouns his and him, which have to be coreferntial with the subject, see (63-4a), cannot be replaced by any other expressions, see (63-64b-d). Hence no contrast set can be generated and, as a consequence, these constructions are unable to satisfy the contrastiveness condition in (62). Since it appears to be im possible to construct solid, unam biguous violations o f K onig’s centrality condition and B aker’s prom inence condition, we conclude that they are not needed. O n the other hand, the existence o f data like (63-4) clearly supports the assum ption o f a contrastiveness constraint as (62). Indeed, it seems possible to derive prom inence and centrality phenom ena from the contrastiveness condition in (62). Intensification o f a DP is thus very sim ilar to focus. Just like focus it triggers the generation o f a set o f alternatives which are contrasted with the intensified expression. D epending on the context, the relationship between the associate o f the intensifier and the focus-generated contrast set may or may not be ordered in term s o f prom inence or centrality. In other words, centrality and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 prom inence are not essential parts o f the process o f intensification itself, but sim ply different interpretations w hich intensification may get in different contexts. A t this point we have identified tw o conditions on adnom inal intensification: the unique identifiability condition (54) and the contrastiveness condition (62). It seem s very likely that these tw o conditions are connected^* and that they may potentially be further reduced to even m ore general underlying principles. D etailed com parison o f focus and adnom inal intensification reveals a link between unique identifiability and contrastiveness. Intensification always involves the generation o f a contrast set o f alternatives to the referent o f the associate o f the intensifier. In order to generate such a contrast set one m ust be able to (uniquely) identify the referent o f the associate. However, while is possible to subsum e the unique identifiability condition (54) under the contrastiveness condition (62) in this way, for ease o f exposition we will still continue to refer to the condition in (54) when testing different uses o f intensifiers. 2.2.4 Is adnominal intensification of PRO possible? A s show n in (65) PRO resists adnom inal intensification^^. In this section we discuss three possible explanations o f this fact. Cf. Siemund (2000:170) who also speculates that unique identifiability and centrality may somehow be “intimately connected.” At first glance, Danish sentences like (i) may appear to falsify the claim that PRO cannot be adnominally intensified: (I) Peter lovede os selv at skrive en artikel am skolesystemet. a. Peteri promised us t| self [PRO to write an article about school-system-the] ‘Peter himself promised us to write an article about the school system.’ b. Peter promised [us self] [PRO to write an article about school-system-the] ‘Peter promised us ourselves to write an article about the school system.’ c. Peter promised us [ PRO [VP self [VP to write an article about school-system-the]]] ‘Peter promised us to write an article about the school system him self’ d. *Peter promised us [ [PRO self] to write an article about school-system-the] However, as the glosses in (ia-d) shows, the fact that (i) is grammatical is not necessarily proof that PRO can be intensified. The sentence in (i) is multiply ambiguous. The element selv can be either: (1) a q-floated adnominal intensifier intensifying the matrix subject (ia), (II) an adnominal intensifier intensifying the pronoun os ‘us’ (ib), or, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 8 (65)^° a. Bill hopes PRO to win the election. b. *Bill hopes [PRO himselfj to win the election If, as argued in 2.2.2.3, ail nom inal expressions (DP, pronouns, reflexives, traces, etc.) can be intensified in the same way by adjunction o f the adnom inal intensifier, one has to answ er the question why PRO disallows intensification. That is, why is (65b) ungram m atical? Based on our analysis o f intensification in (66) we have to argue that, with respect to the syntax o f intensifier-adjunction, intensification o f PRO is ok. That is, by itself, (67) should be syntactically w ell-form ed. (67) [dp [dp [d P R O ] ] [selv]\ ‘PRO s e lf Furtherm ore, since our analysis o f English reflexives proposed in chapter 5 is based on the assum ption that phonologically zero elem ents may be adnom inally intensified, we cannot explain (65b) sim ply by assum ing the existence o f a contraint banning adnom inal intensification o f phonologically unrealized elements^'. Since it not the syntax o f intensfier-adjunction which is responsible for the ungram m aticality o f (65), it has to be due to som ething else. In the follow ing, three different hypotheses will be discussed: (i) an explanation based on the contrastiveness condition on adnom inal intensification, (ii) an explanation based on the assum ptions about the lexical properties o f PRO, and (iii) an explanation based on the PRO-theorem. It seem s very tem pting to explain the absence o f intensified PRO in term s o f the contrastiveness condition on adnom inal intensification (62). As the data in (69-71) show, the (III) a pre-verbal exclusive adverbial intensifier m ean in g ‘w ith o u t help, all by oneselT w hich is adjoined to the V P (ic). It might be claimed that a sentence like (i), shows that PRO may be intensified. (i) Bill promised PRO to write the essay himself. We argue that such an account of (i) is not possible and that the occurrence of him self in (i) is an exclusive adverbial intensifier with the meaning ‘by himself, without help form others, etc.’ which attaches to the VP of the embedded clause rather than to the DP containing PRO. See 2.3 for more discussion of the use o f intensifiers as ‘exclusive’ manner adverbials. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 PRO constructions in (69) appear to violate the contrastiveness condition in the same way as inherently reflexive constructions (70-71) by not allow ing for the creation o f a contrast set. (69) a. Bill hopes PRO to win the election. b. *Bill hopes John to win the election. c. *Bill hopes you to win the election. *Bill hopes [PRO himself] to win the election. (70) a. Billj is out o f his mind. b. UBill is out o f Jo h n ’ s mind. c. #Bill is out o f your mind. d. #Bill is out o f his own mind. (71) a. Billj took a knife with himj n ^ . b. #Bill took a knife with John. c. #Bill took a knife with you. d. #Bill took a knife with himself. W hile the unacceptability o f (69) and (70-71) appear to be due to violations o f the contrasitiveness condition in (62), the ultim ate reasons why these sentences fail to generate contrast sets are probably different. In the case o f the inherently reflexive expressions in (70- 71) the failure to generate contrast sets, and thus m eet the requirem ents o f (62), is due to the sem antics o f these expressions. Syntactically there is nothing wrong w ith (70b-d) and (71 b-d). It is the m eaning o f these sentences w hich exclude the possibility o f com m utation o f his and him w ith any other expressions. Since it is sem antic rather than syntactic factors w hich are responsible for the unacceptability o f (70-7Ic-d), these sentences have been m arked with the symbol indicating sem antic or pragm atic anom aly, rather than w hich is reserved for syntactic anomaly. In the case o f the PRO sentences in (69), however, the failure to generate contrast sets, and thus m eet the requirem ents o f (62), may be due to syntactic contraints. O vert DPs have to The analysis of “ inclusive selv" as a a-floated adnominal intensifier, proposed in section 2.2.5, is also based on the assumption that DP-traces can be adnominally Intensified. See also discussion of fronted selv in section 2.2.6. .12 Note that this sentence may be acceptable to certain speakers when him self vs analyzed as a q-iloated adnominal intensifier modifying the matrix subject DP Bill, see section 2.2.5 for more discussion o f q-floated intensifiers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 have case, while the PRO-theorem requires PRO to be caseless (or have null-case, depending on analyses). As show n in (72-73), the ungram m atical sentences in (69b-d) can be saved by providing case-assigners for the subjects o f the em bedded clauses. (72) a. Bill hopes that he will win the election. b. Bill hopes that John will win the election. c. Bill hopes that you will win the election. d. Bill hopes that he him self will win the election. (73) a. Bill hopes fo r him to win the election. b. Bill hopes fo r John to win the election. c. Bill hopes fo r you to win the election. d. Bill hopes fo r him self to win the election. In (72) nom inative case is assigned to the em bedded subject by the finite T projection, in (73) accusative case is provided by the preposisiton for. Thus the im possibility o f generating a set o f alternatives to the referent o f PRO in (65b/69d) is due to the fact that case-constraints require overt DPs and PRO to be in com plem entary distribution. However, regardless o f the different reasons for failure to satisfy (62), the ungram m atical ity o f (69d) and (70-7 Id) all appear to be reduced to violations o f this contrastiveness principle. O n closer inspection, however, this account may not be as succesful as it seems. Since the contrastiveness condition is o f sem antic nature, there is no reason why generation o f a constrast in (69) should not possible. A fter all, the sentences in (72-73), clearly show that sem antically the contrast sets can be created once the syntax allows it. Let us now turn to another, perhaps m ore prom ising, account o f the absence o f intensified PRO, w hich is based on the assum ption that PRO is incom patible with intensification. As noted by C hierchia (1989) and others, PRO is always given a DE SE reading and is always interpreted as a bound variable. That is, while (72a) and (73a) can be given both DE SE and DE RE readings, (69a) can only be DE SE. The difference between DE SE and DE RE readings can be captured by the sem antic paraphrases in (74). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 (74) a. John hopes that hc[x wins the election] (D E SE, PRO, (69-70)) b. John; hopes that hcj wins the election (D E RE, non-PRO, (72-73)) The DE SE reading in (74a) establishes a relation betw een John and a property, in this case the property o f w inning the election. In contrast, since the paraphrase in (74b) does not imply self-ascription o f any property, it is am biguous between DE SE and DE RE readings. A ssum ing PRO to be lexically specified to require a bound-variable interpretation may then explain the incom patibility o f PRO and adnom inal intensification. Note, however, that in order to avoid the false prediction that PRO can have a DE RE reading when bound by a quantified expression (e.g. [Every candidate], hopes PRO; to be elected) it is necessary to assum e that PRO m ust be bound by the A.-operator in the closest Comp. Finally, an even sim pler way to explain the apparent absence o f adnom inally intensified PRO, consists in assum ing that adnom inal intensification o f phonological zero elem ents turns these elem ents into overt DPs w hich m ust have non-null case. Since PRO is per definition in a case-less (or null-case) position, intensification o f PRO via intensifier- adjunction would necessarily lead to violations o f case-requirem ents. See also discussion o f the possibility o f adnom inally intensifying PRO in English and Chinese in chapters 5 and 6 respectively. N otice that this account would not run into any problem s with 0-reflexives in English and Chinese, since these all occur in case-positions. Furtherm ore, this account needs no extra stipulations, but follows directly from the PRO-theorem . If this last, case-based account o f the absence o f intensified PRO is correct then it constitutes a potentially powerful argum ent against current analyses o f PRO as receiving zero Case. A t this point the third explanation seems to be the m ost straightforw ard in that it does not require any ad hoc stipulations. In the long run it may be worth while exploring the potential relationship between DE SE and PRO. However, doing so is beyond the scope o f the present dissertation. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 72 2.2.5 Selv used as q-floated adnominal intensifier In this section we dem onstrate how the analysis o f intensification proposed above can be used to unify the account o f two uses o f selv w hich have hitherto been assum ed to be o f very different nature, viz. the adnom inal intensifier and w hat has been called the “inclusive adverbial intensifier”. The term s “inclusive” and “exclusive” adverbial uses o f selv were introduced by Ekkehart K onig to refer to w hat he takes to be different adverbial uses o f intensifiers. The use o f the intensifier h im s e lf m the sentence Peter had him self written an essay is “inclusive” in the sense that it presupposes the inclusion o f Peter into the set o f individuals who have written essays. The use o f h im se lf m the sentence Peter had written the essay him self\s exclusive in that it excludes all agents but Peter, i.e. Peter wrote the essay w ithout help/all alone. W hen the intensifier selv is placed sentence-m edially, i.e. right after the inflected verb or auxiliary, it is usually interpreted as the so-called inclusive adverbial selv ‘too, also’, see (75). (75) Lcereren havde selv skrevet en stil. teacher-the had self written an essay ‘The teacher had him self written an essay. ’ A lthough selv in this usage has the appearance o f an adverbial elem ent w e propose to analyze it as being derived from an adnom inally intensified DP. T hat is, w e propose that the selv in (75) is ‘floated’ in the sam e sense that the quantifier alle ‘a ll’ in (76) is stranded by quantifier-floating (cf. Sportiche (1988)), see (77) w hich gives the syntactic tree for the sentences in (75-76). (76) Eleverne havde alle skrevet en stil. Pupils-the had all written an essay ‘The pupils had all w ritten an essay.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 73 (77) CP / \ Lcereren, C ’ / \ havde^ IP / \ DP r I / \ ti 4 VP / \ DP V ’ / \ / \ DP 5e/v V DP I I 1 4 skrevet en stil As shown in (77) q-floated 5e/v is syntactically derived from adnom inal selv. The existence o f this derivational relationship is supported by the fact that the interpretation o f selv in this usage is closely related to the m eaning o f adnom inal selv. The adjunction o f selv to a DP, e.g. the teacher him self had written an essay, m ay serve to em phasize the teacher’s involvem ent in the event and to contrast it with the other potential agents (i.e. his students, his wife, etc.). The quantifier-floated selv in (75)/(77) appears to have the same sem antic contribution. As m entioned in section 2.2.3.2.1, the distribution o f adnom inal selv is limited the unique identifiability constraint on the D P w hich is being intensified, see (54)^^. The exam ples in (78- 83) illustrate that quantifier-floated selv also displays the sam e identifiability effects and thus has m ore in com m on w ith adnom inal selv than w ith selv used as an “exclusive” m anner adverbial. (78) *lngen selv pudsede skoene. (adnom inal selv) nobody self polished shoes-the ‘*Nobody him self polished the shoes.’ (79) Statsministeren selv pudsede skoene. (adnom inal selv) prim e m inister-the self polished shoes-the ‘The prim e m inister him self polished the shoes.’ As mentioned in section 2.2.3.2.2 the unique identifiability condition (54) can be subsumed under the contrastiveness condition (62). In other words, the unique identifiability condition does not exist. If we continue to refer to the unique identifiability condition, it is therefore only for ease o f exposition. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 74 (80) Ingen pudsede skoene selv. (“exclusive” selv) nobody polished shoes-the self ‘N obody polished the shoes all by him self.’ (81) Statsministeren pudsede skoene selv. (“exclusive” ^e/v) prim e m inister-the polished shoes-the self ‘The prim e m inister polished the shoes all by h im se lf’ W hile adnom inal selv displays specificity effects in that it can only adjoin to a DP denoting a specific or uniquely identifiable referent, see (78) vs. (79), selv used as an “exclusive” adverbial does not, see (80) and (81) w hich are both fine. The question is now w hether or not quantifier-floated selv is subject to the same unique identifiability constraint. On the basis o f the proposed analysis o f quantifier-floated selv, see (77), we predict that sentences with q- floated selv should display som e sort o f identifiability effects. The exam ples in (82) and (83) offer evidence in favor o f this hypothesis. (82) a. ??/*Ingen pudsede selv skoene. (q-floated .se/v) nobody polished self shoes-the ‘*Nobody polished him self the shoes.’ b. Statsministeren pudsede selv skoene. (q-floated selv) prim e m inister-the polished self shoes-the ‘The prim e m inister him self polished the shoes.’ (83) a. *Ingen er selv en klovn. (q-flloated selv) nobody is self a clown ‘*Nobody is him self a clow n.’ b. Statsministeren er selv en klovn. {selv q-floated) prim e m inister-the is self a clown ‘The prim e m inister is him self a clow n.’ The fact that the sentences in (82a) and (83a) are both infelicitous indicates that q-floated selv is subject to the unique identifiability constraint. We take this as evidence that q-floated selv is an a d n o m in a l in te n s ifie r w h ic h is stra n d e d w h e n its a ss o c ia te m o v e s to a h ig h e r p o sitio n , se e (77). That (82a) seems slightly m ore acceptable to some speakers than (85a), is due to the nature o f the predicate. In (82) the predicate pudse sko ‘polish shoes’ is com patible with agent-oriented adverbs, w hile in (83) the predicate vcere en klovn ‘be a clow n’ is not. This R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 75 m eans that in (82) selv can be forced to be read as an exclusive adverbial, w hich is com patible with indefinite subjects, while such a reading is excluded by tbe non-agentive predicate in (83). The im possibility o f com bining exclusive adverbial selv witb non-agentive predicates is illustrated by the sentence in (84). (84) ??Statsministeren er en klovn selv. (Com pare w ith (81)) prim e m inister-the is a clown self As already m entioned exclusive adverbial selv is a m anner adverbial which usually occurs predicate-finally. W hile such a predicate-final exclusive adverbial selv is ok with the agentive predicate pudse sko ‘polish shoes’, as show n by (81), it is quite infelicitous in (84). U nlike the exclusive m anner adverbial selv ‘ all by o n eself, the q-floated selv ‘too, also’ does not im ply uniqueness o f the agent, see the exam ple in (83b) which contains a predicate w hich is incom patible w ith m anner adverbials. The m eaning o f (83b) is not incom patible w ith scenarios in which people other than the prime m inister are clowns. In fact, (83b) seem s to im ply the existence o f a set o f propositions o f the form ‘x is a clow n’, i.e. the focus generated contrast set, in w hich at least one proposition other than ‘The prime m inister is a clow n’ (i.e. the foreign m inister is a clown, the secretary is a clown, etc.) is also true. In other words, (83b) implies the inclusion o f the proposition ‘the prim e m inister is a clow n’ into the non-em pty set o f true propositions o f the form ‘x is a clow n’. In this respect, the m eaning o f the inclusive selv ‘also, to o ’ seem s to be sim ilar to the m eaning o f the additive particle ogsd ‘also’, com pare (83b) and (85). (85) Statsministeren er ogsaa en klovn. p rim e m in is te r-th e is a lso a c lo w n ‘Peter is also a clow n.’ Like q-floated se/v ‘too, also’, the additive focus particle ogsd ‘also’ is also usually placed im m ediately after conjugated verb or auxiliary (unlike the exclusive m anner adverbial selv ‘all by o n e se lf w hich is VP final), see (86). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 76 (86) *Statsministeren er en klovn ogsaa. (com pare w ith (85) and (87)) prim e m inister-the is a clown too Note, however, that inclusive selv and the additive focus particle ogsa ‘also’ can co-occur in the same sentence, see (87). (87) Statsministeren er ogsaa selv en klovn. prim e m inister-the is also self a clown ‘The Prime m inister is also a clown h im s e lf’ The fact that inclusive selv and the additive focus particle ogsd ‘also’ can co-occur in the sam e sentence is a strong indication that they do not fulfill exactly the same sem antic function(s). Otherwise, such co-occurrence w ould be im possible. The ‘additive’ aspect o f the m eaning o f inclusive selv seems to indicate that it m ight som ehow be related to the pre-nom inal additive focus particle selv ‘even’, discussed in the previous section, w hich also evokes additivity. However, unlike the additive focus particle selv ‘even’, q-floated selv does not necessarily evoke a scale o f expectedness. N or does it seem to involve the same kind o f additivity. The presence o f additivity can be tested using clefts, see (88). (88) a. Det var dronningen selv, som var blevet nomineret til en Oscar. it was queen-the self who was becom e nom inated to an O scar “It was the Queen herself who had been nom inated for an O scar.” b. {l)D et var dronningen, som selv var blevet nomineret til en Oscar. it was queen-the self who was becom e nom inated to an O scar It was the Queen who had herself been nom inated for an Oscar. c. *Det var selv dronningen, som var blevet nomineret til en Oscar. it was queen-the self who was becom e nom inated to an O scar *lt was even the Queen who had been nom inated for and Oscar. C left-constructions presuppose uniqueness and are thus incom patible with additive focus particles like selv ‘even’ in (88c). In contrast, since the adnom inal intensifier selv does not presuppose additivity, it is fine in cleft-constructions like (88a). If q-floated selv is basically a stranded adnom inal intensifier, then it should also be com patible with cleft-constructions. As shown in (88b) this prediction seem s to be borne out by the facts. The sentences in (88b) may be slightly odd, but it is far from being ju st as unacceptable as (88c). In other words, the R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 77 apparent additivity o f q-floated selv is not part o f its lexical m ake-up (as in the case o f the additive, scalar focus particle selv ‘even’), but rather contextually evoked and may be overridden by the context. Further evidence supporting the hypothesized derivational relationship between adnom inal selv and q-floated selv is provided by the exam ples in (89-91), which illustrate the possible com binations o f adnom inal selv, q-floated selv and “exclusive” adverbial selv w ithin a single sentence. (89) Statsministeren selv pudsede skoene selv. (adnom inal se/v + “exclusive” se/v) prim e m inister-the self polished shoes-the self ‘The prim e m inister him self polished the shoes w ithout help from any one.’ (90) Statsministeren pudsede selv skoene selv. (q-floated selv -I- “exclusive” se/v) prim e m inister-the polished self shoes-the self ‘The prim e m inister him self polished the shoes w ithout help from any one.’ (91) kStatsministeren selv pudsede selv skoene. (adnom inal selv + q-floated se/v) prim e m inister-the self polished self shoes-the ‘The prim e m inister him self (him self) polished the shoes.’ Com binations o f adnom inal se/v and “exclusive” adverbial se/v, and o f q-floated selv and “exclusive” adverbial se/v are both fine, see (89) and (90). However, com binations o f adnom inal selv and q-floated se/v are infelicitous (or at best highly redundant and very odd), see (91). The generalization seem s to be that one instance o f “exclusive” se/v can be com bined with one instance o f adnom inal se/v (q-floated or not), but tw o instances o f adnom inal se/v cannot be com bined in one sentence (even if one is q-floated and the other is not). This incom patibility o f adnom inal se/v and q-floated se/v confirm s the hypothesized derivational relationship between the two^'*. A further indication that sentence-m edial, q-floated se/v is derived from the adnomnal intensifier com es from stress placem ent. W hile the post-nom inal adnom inal intensifier is In this respect we differ from Siemund (2000) who claims that it is possible to construct grammatical sentences containing all three types of intensifiers, i.e. adnominal plus inclusive and exclusive adverbial intensifiers. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 78 always stressed (e.g. kongen S E L V ‘the king H IM SELF’), the additive focus particle is always unstressed (e.g. selv KONGEN ‘even the K IN G ’). As illustrated in (92), q-floated selv is necessarily stressed. (92) a. Peter er SELV en klovn. Peter is self a clown ‘Peter is H IM SELF a clow n’ b. 77PETER er selv en klovn. Peter is self a clown ‘PETER is him self a clow n’ c. llP eter er selv en KLOVN. Peter is self a clown. ‘Peter is him self a C LO W N .’ Again, this may be taken as support for the hypothesis that q-floated selv is derived from the adnom inal intensifier, w hich is also stressed. In the present state o f the language q-floated 5e/v-constructions seem to be found more often than non-q-floated adnom inal .ye/v-constructions w hich tend to be found m ost often with a sim ple unm odified DPs (e.g. Stalin, Jesus, the Pope, etc.)^\ In contrast, q-floated selv- constructions can be found with any type o f DP regardless o f their com plexity, see (93) vs. (94). (93) a. Manden (*som var skaldet) selv kom til festen. man who was bald self cam e to party-the ‘The m an who was bald him self cam e to the party’ b. Den gamle mand (’ ^med bid hukser) selv kom til festen. the old man w ith blue trousers self cam e to party-the ‘The old man with blue trousers him self cam e to the party’ (94) a. Manden som var skaldet kom selv til festen. m an-the who was bald cam e self to party-the ‘The m an who was bald him self cam e to the party’ ” Another factor to be taken into consideration is that fact that non-q-floated 5e/v-constructions tend to be found most often with a restricted set of DPs denoting high status individuals (e.g. the king, the bishop, the president, Jesus, etc.). See also the discussion of lexically defined prominence, e.g. (55), in section 2.2.3.2.2. It is also not impossible that heavy DP-effects or similar processing factors might be (partly) responsible for these differences between adnominal intensifiers in-situ and q-floated intensifiers. This might explain why q-floated selv is preferred with DPs with long post-nominal modifiers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 b. Den gamle mand m ed bid hukser kom selv til festen. the old m an w ith blue trousers cam e self to party-the ‘The old m an with blue trousers him self cam e to the party’ A s discussed in section 2.2.2.2, exam ples such as (93) seem to indicate a difference with respect to stacking options. Simple DPs, e.g. (93a,b) without the post-nom inal relative clause/PP, can be adnom inally intensified w ithout any problems. In contrast DPs m odified by post-nom inal relative clauses/PPs can only be intensified via a q-floated selv, com pare (93) vs. (94). Syntactically the intensifier selv behaves very m uch like the quantifier alle ‘all’ w hich also has both a pre-nom inal adjectival form and a post-nom inal appositional form. The post- nom inal quantifier alle ‘all’ m odifying direct and indirect object pronouns is illustrated by the sentences in (95). (95)^^ a. Lcereren roste os alle. teacher-the praised us all. ‘The teacher praised us all.’ b. Peter gav dem alle en krone. Peter gave them all a crow n (=m onetary unit in Denmark). ‘Peter gave them all a crow n.’ Just as the intensifier selv has a pre-nom inal adjectival form (i.e. selve, see section 2.2.6), the quantifier alle also has a pre-nom inal adjectival form (i.e. al, alt, alle), see (96). (96) a. A lt oksekodet er udsolgt. all beef-the is sold out. ‘All the b eef is sold o u t ’ The sentences in (95a,b) involve object shift of the personal pronouns os ‘us’, and dem ‘them’. Sentences in which the DO and 10 are lexical DPs (which cannot object shif), would have pre-nominal quantifiers in situ, e.g. (ia-h). (i) a. Laireren roste alle eleverne. teacher-the praised all pupils-the ‘The teacher praised all the pupils.’ b. Peter gav alle bornene en krone. Peter gave all children-the a crown ‘Peter gave all the children a crown’. This indicates that the quantifier alle in (95a-b), may not, in fact, be a post-nominal adjuncts, but rather pre-nominal quantifiers which have become ‘floated’ when their associates (i.e. the personal pronouns os ‘us’, and dem ‘them’) moved to a higher position due to object-shift. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 8 0 b. Jeg inviterede alle mine venner. 1 invited all my friends ‘I invited all m y friends.’ c. Du fa r al den hjcelp(common) du behover. you get all the help you need ‘Y ou’re getting all the help you need. As already illustrated by the exam ple in (76), alle m ay be q-floated ju st like selv, com pare also (97a,b) and (97c). (97) a. Eleverne bestod alle eksamen. pupils-the passed all exam -the ‘The pupils all passed the exam .’ b. Vi er alle tilfredse med resultatet. we are all satisfied w ith result-the ‘We are all satisfied w ith the result.’ c. Vi er selv tilfredse med resultatet. we are self pleased with result-the ‘We are ourselves pleased with the result.’ With the above analysis o f q-floated intensifiers in mind, let us now turn to am biguous sentences in w hich selv can be read as either an adnom inal intensifier m odifying the direct object, see (98i) and (99i), a q-floated selv, see (9911), or an “exclusive” adverbial selv (98ii) and (99iii). (98) Peter m ake muren selv. Peter painted wall-the self (i) ‘Peter painted the wall itself.’ (ii) ‘Peter painted the wall h im se lf’ (99) Peter malte den selv. Peter painted it self (i) ‘Peter painted it itself.’ (ii) ‘ Peter him self painted if (iii) ‘Peter painted it h im s e lf’ (adnom inal selv) (“exclusive” adv. selv) (adnom inal selv) (q-floated selv) (“exclusive” adv. selv) The first reading (98i), in which .sc/v is an adnominal intensifier m odifying the internal object D? muren ‘w all-the’, is illustrated in (100). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 (100) CP / \ Peteri C ’ / \ malte, IP / \ DP V 1 / t, 4 / DP 4 VP \ V ’ / \ V DP I / \ 4 DP selv I muren W hile in (99) selv may have both an “exclusive” (99iii) and a q-floated reading (99ii) in addition to the adnom inal reading, in (98) selv can only be either adnom inal (981) or “exclusive” (98ii). These differences between (98) and (99) can be shown to follow from M ainland Scandinavian (M Sc.) object shift. The sentences in (101) illustrate how the placem ent o f negation can be used to test w hether object shift has taken place or not. (101) a. Peter malte ikke muren selv. (“exclusive” 5e/v) Peter painted not wall-the self ‘Peter didn’t paint the wall him self.’ b. *Peter malte muren ikke selv. Peter painted wall-the not self c. Peter malte den ikke selv. (“exclusive” or q-floated selv) Peter painted it not self (i) ‘Peter didn’t paint it h im self (ii) ‘Peter him self didn’t paint it.’ d. *Peter malte ikke den selv. Peter painted not it self For the analysis o f MSc. object shift and sentence structure in general we follow Holm berg & Platzack (1995). In Danish, N orw egian and Swedish, certain types o f objects may be object- shifted, i.e. they may precede the negation words and sim ilar predicate adjuncts, see (101c). In M Sc., as opposed to Icelandic, object-shift is restricted to personal pronouns, see (101b) vs. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 8 2 (101c). We assum e that object shifted pronouns have m oved out o f their argum ent positions to spec-vP (cf. Platzack and H olm berg (1995:143) who take shifted object to be left-adjoined to the predicate). In other words, if the negation ikke ‘not’ is adjoined at the outer left edge o f the predicate then the fact that the pronoun den ‘it’ in (101c) can precede it w hile the DP muren ‘the w all’ cannot shows that D anish object pronouns can shift while full DPs cannot^’. Holm berg and Platzack (1995:143) observed that object shift is subject to certain restrictions. First o f all, shifted objects never crosses the main verb. This m eans that object shift only happens after overt m ovem ent o f the m ain verb out o f the predicate (i.e. vP) has taken occurred. G iven that auxiliary verbs block verb-m ovem ent, object shift is never found in clauses with auxiliary verbs. The D anish sentences in (102-103) support this analysis. (102) a. Peter havde (ikke) malt muren selv. (“exclusive” 5e/v) Peter had (not) painted w all-the self ‘Peter had not painted the wall h im se lf’ b. Peter havde (ikke) selv malt muren. (q-floated selv) Peter had (not) se lf painted wall-the ‘Peter him self had not painted the w alk’ c. *Peter havde muren (ikke) malt. Peter had wall-the (not) painted (103) a Peter havde (ikke) malt den .selv. (“exclusive” selv) Peter had (not) painted it self ‘Peter had not painted it h im se lf’ b. ?Peter havde (ikke) selv malt den. (q-floated selv) Peter had (not) self painted it ‘Peter him self had not painted i t ’ c. *Peter havde den (ikke) malt. Peter had it (not) painted Com pare (101c) w here object shift is allowed with (103e) where the auxiliary havde ‘had’ blocks object shift. The exam ples in (102-3) also illustrate the fact that auxiliaries help disam biguate betw een q-floated selv (w hich is stranded in the VP-internal subject position), see (102-3b) and the predicate-final m anner adverbial, i.e. “exclusive” selv in (102-3a). As for the movement rule responsible for object shift, H&P propose that “it is a “mixed rule”, exhibiting a particular mix o f properties of A-movement and A-bar movement”(H&P, p. 142). Since the exact nature o f this rule R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 83 Finally, the main vs. em bedded clause contrast in object shift can also be used to disam biguate betw een different uses o f selv ‘s e lf. Since in MSc. m ovem ent o f the verb out o f the predicate only takes place in m ain clauses, it follows that object shift is also is restricted to main clauses, see (104). (104) a. * . . . at Peter m ake ikke muren selv. . .. that Peter painted not wall-the self b. . . . at Peter ikke make muren selv. (“exclusive” selv) . .. that Peter not painted wall-the self ‘... that Peter didn’t paint the wall h im se lf’ c. * . . . at Peter make den ikke selv. .. . that Peter painted it not self d. .. .a t Peter ikke make den selv. (“exclusive” selv) .. . that Peter not painted it self ‘... that Peter didn’t paint it h im se lf’ Given w hat we know about object shift, the placem ent o f the negation and MSc. sentence structure in general, it is now fairly easy to explain w hy selv in (99) is am biguous between either q-floated selv, or “exclusive” adverbial selv, w hile in (98) it can only be an “exclusive” adverbial intensifier (in addition to the adnom inal reading w hich both sentences have). The structural am biguity o f (99) has been teased apart in (105-6), (105) show ing the structure underlying the q-floated reading, and (106) show ing the structure underlying the exclusive adverbial reading o f selv. is not important for the purposes of the present analysis o f selv we simply refer the reader to H&P (1995). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 84 (105) CP / \ Peter, C ’ / \ malte^ TP / \ DP T ’ I ti (106) CP / \ Peter i C ’ / \ malte, TP \ vP^® / \ den„ vP / \ DP v’ / \ / \ ti selv 4 VP 1 1 V ’ / \ V DP 4 4 / \ DP T ’ 1 / \ 4 4 vP / \ derix vP / \ DP v’ 1 / \ 4 4 VP / \ VP selv 1 V ’ / \ V DP 1 1 1 1 4 4 The structures in (105-6) is based on Platzack & Holmberg (1995). In their analysis, “Act” is a functional projecting head encoding voice which is “situated inside the predicate, taking VP as a complement and taking the external argument of the verb as its specifier” (p. 20). In currently used terminology “vP” and “TP” have replaced the now obsolete “ActP” and “IP”, but otherwise the structure in (105-6) are the same as the ones proposed by Platzack and Holmberg. In the object shift constructions “the verb has moved out of the VP, first to Act°, then to 1 ° and eventually to C°” and the “object pronoun [..] is left-adjoined to the predicate, i.e. AetP” (ibid. pp. 143, 140). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 Let us briefly sum m arize the findings o f this section. The analysis o f adnom inal selv, q- floated selv and exclusive adverbial selv proposed above enables us to correctly predict: (i) that the sentence in (98) cannot have the q-floated so-called “inclusive” reading but only tbe “exclusive” one. (ii) that the sentence in (99) can have both the “inclusive” and the “exclusive” readings, see (105) and (106). (iii) that adnom inal selv and q-floated selv cannot felicitously be com bined in the sam e sentence; since they are really both instances o f adnom inal intensification, see (91). (iv) that exclusive adverbial selv can be com bined with adnom inal selv or q- floated selv in the same sentence. M ultiply am biguous sentences like (99) illustrate one o f the main challenges faced by anyone researching the gram m ar o f intensification: the need to clearly distinguish betw een different uses o f intensifiers. The fact that in many sentences the elem ent selv is m ultiply am biguous makes it very difficult to get consistently reliable judgem ents. Further adding to this difficulty is the fact that gram m aticality jugdm ents o f sentences containing intensifiers are usually highly susceptible to discourse- and pragm atic contexts. The above analysis o f the so-called inclusive adverbial selv as a q-floated adnom inal selv illustrates one o f the goals o f this dissertation: to reduce as many seem ingly different uses o f intensifiers as possible to the sam e underlying principles. In this case, w hat was form erly believed to be an adverbial use o f the elem ent has been show n to be an instance o f adnom inal intensification falling under the contrastiveness condition (62). It also shows that DP-traces may be adnom inally intensified. 2.2.6 Fronted selv As shown in (107) and (108), the adnom inal intensifier selv may be fronted. (107) a. Jeg SEL V tor aldrig modsige ham. (adnom inal selv, in situ) 1 self dare never contradict him ‘1 m yself dare never contradict him .’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h f o w n e r. F u r th e r re p ro d u c fio n p ro h ib ite d w ith o u t p e r m is s io n . 8 6 b. K ongen SEL V har ringe magt. king-the self has slight pow er ‘The king him self has little pow er.’ (108) a. SELV tor je g aldrig modsige ham. se lf dare I never contradict him ‘I m yself dare never contradict him .’ b. SELV har kongen ringe magt. self has king-the slight pow er ‘The king him self has little pow er.’ (adnom inal selv, in situ) (fronted adnom inal selv) (fronted adnom inal selv) Such instances o f fronted selv should not be confused with the pre-nom inal scalar, additive focus adverb selv ‘even’, see the exam ples in (109)^’. (109) a. Selv JEG tor aldrig modsige ham. even 1 dare never contradict him ‘Even I dare never contradict him .’ b. Selv KONGEN har ringe magt. even king-the has slight power ‘E ven th e k in g h as little p o w e r.’ (scalar additive focus particle) (scalar additive focus particle) Both the Danish sentences and their respective English translations in (108-9) are adapted from A llan et al. (1995:169). The authors o f this descriptive gram m ar o f Danish appear to assum e fronted selv to be derived from the post-nom inal adnom inal intensifier by a productive syntactic process o f fronting. W hile we agree with this assum ption, we also realize the need for independent tests to exclude the possibility o f fronted selv being derived from the scalar, additive focus particle ‘even’. Just like in the case o f q-floated selv, discussed in section 2.2.5, different tests can be applied: (i) the so-called unique identifiability constraint on adnom inal intensification, and (ii) relative stress placem ent,, i.e. does stress fall on the elem ent selv itself or on the expression it is in construction with? Since we know that the elem ent selv is stressed when used as adnom inal intensifier, e.g. kongen SELV ‘the king h im se lf in (107b), the fact that fronted selv is stressed, see (108), could be taken as an indication that fronted selv is an instance o f the adnom inal intensifier. See section sect. 2.4. for more discussion of the scalar, additive focus adverb selv ‘even’. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 However, this argum ent is rather weak since the stress on fronted selv can also be argued to follow from general properties o f the process o f fronting, i.e. all fronted elem ents are stressed. The exam ples in (110) are intended to test w hether fronted selv (1 10a) is subject to the same unique identifiability constraint as adnom inal intensification (1 10b). (110) a. */??Selv har en konge ringe magt. se lf has a king slight power ‘A king him self has little pow er.’ b. */??En konge selv har ringe magt. a king self has slight power ‘A king him self has little pow er.’ c. Selv en konge har ringe magt. even a king has slight pow er ‘Even a king has little pow er.’ The unacceptability o f (1 10a) suggests that we are correct in assum ing that fronted selv derives from the adnom inal intensifier selv ‘h im self (110b) rather than from the additive focus particle selv ‘even’, which is ok with indefinite DPs, see (1 10c). We therefore conclude that fronted selv is sub-case o f adnom inal intensification. Interestingly, the exam ples in (108), as well as all the other exam ples o f fronted selv m entioned in Allan et al. (1995) all contain a negative element: “5'e/v in this position usually has negative connotations and therefore tends to occur with a negation {ikke ‘not’; aldrig ‘never’, etc.)” (Allen et al (1995:169). This peculiarity o f fronted selv, viz. that it appear to be possible only in negative sentences, may be related to the fact that in the Slavic languages, adnom inal intensifiers, which, incidentally, are pre-nom inal, e.g. Russian sam DP ~ (Eng.) DP himself, may be interpreted as additive focus particles w ith the m eaning ‘even’ only when occurring in negative sentences, i.e. downward entailing environments'*®. It may therefore be possible to analyze fronted selv as an interm ediary case, between the post-nom inal adnominal intensifier and pre-nom inal scalar, additive focus particle ‘even’. In section 2.4 we suggest This generalization and the potential consequences for the analysis of fronted selv in Danish were brought to my attention by R. Pancheva (p.c). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 8 8 that the differences between selv used as pre-nom inal additive scalar focus particle and selv used as post-nom inal adnom inal intensifier follow from syntactic differences (i.e. post- vs. pre- nom inal position). If this is indeed the case, then the fact that fronted instances o f the adnom inal intensifier selv should have som e properties o f the scalar, additive focus particle ‘even’ (viz. the fact that it is only found in dow nw ard entailing environm ents (just like pre- nom inal cam ‘alone, him self, even’ in Bulgarian w hen used as scalar additive focus particle ‘even’)) should com e as no surprise. In other words, it m ight be possible to use the sim ilarities betw een fronted selv in D anish and the additive scalar focus particle cam ‘alone’ in Bulgarian to bolster the argum ent that syntactic position matters, i.e. the sam e elem ent selv takes on different properties post-nom inally and pre-nom inally. 2.2.7 Adnominal intensiflcation: unification of different sub-cases A t this point tw o o f the “threefold” goals o f this chapter m entioned on page 4 have been achieved. That is, we have proposed a focus-based analysis o f adnom inal intensification, see (62) in section 2.2.3.2.2, and show n that the different uses o f selv listed in (111) are all instances o f adnom inal intensification and thus fall under the condition in (62). (Ill) A dnom inal Intensification: Exam ple: Section: i. intensified DPs (1), (14) (sect. 2.2) ii. intensified reflexives (4) (sect. 2.1, chap. 3, sect. 3.3) iii. intensified pronouns (5) (sect. 2,1, chap. 3, sect. 3.4) iv. q-floated intensifier (77) (sect. 2.2.5) V. fronted 5 e / v (108) (sect. 2.2.6) Only the third goal, viz. the unification o f all uses o f selv - both adnom inal and non-adnom inal - rem ains and will be dealt w ith in the follow ing sections as sum m arized in (112). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 89 (112) R em aining uses o f selv. Exam ple: Section: a. selv in secondary predication constructions (7), (113) (sect. 2.3) b. scalar additive focus particle ‘even’ (8), (120b) (sect. 2.4) c. 5 e /v a s a n o u n (9), (150) (sect. 2.5) d. adjectival form s o f the intensifier (10), (152) (sect. 2.6) e. selv in nom inalizations (11), (160) (sect. 2.7, ch. 7) 2.3 Selv in secondary predication constructions In this section we argue that w hat has often been referred to as the “exclusive” adverbial use o f the intensifier selv, see (113), should be analyzed as a secondary predication construction. (113) Peter reparerede bilen selv. Peter repaired car-the self ‘Peter repaired the car h im self Syntactically, this particular use o f selv appears to behave like a m anner adverbial adjoining to the VP. Sem antically, its effect is to exclude all but the referent o f the subject DP as agent o f the predicate in the sentence. Hence the term “exclusive” adverbial intensifier. In this respect, selv functions very much like the particle alene ‘alone’. Both alene ‘alone’ and selv can be m odified by the elem ent helt ‘all, com pletely’, see (1 14a,b). (114) a. Peter reparerede bilen helt selv. Peter repaired the car com pletely self ‘Peter repaired the car all by him self.’ b. Peter reparerede bilen helt alene. Peter repaired the car com pletely alone ‘Peter repaired the car all alone.’ Further support for assum ing the instances o f selv in (113-114) to be m anner adverbials comes from sentences like the ones in (1 15a-b) w hich show that neither selv in the m eaning ‘all (by) o n e se lf nor alene ‘alone’ can occur sentence-m edially but m ust occur predicate-finally, or outside the predicate, to be m ore exact'". The exam ples in (116) also illustrate that the m anner adverbial hurtigt ‘fast’ behaves in the same way. The similarities between selv and alene ‘alone’, is further bolstered by the fact that in certain languages, e.g. Russian and Bulgarian, the element used as adnominal intensifier actually means ‘alone’, e.g. Russ, sam ‘ alone, x- R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 90 (1 1 5 /^ a. * Peter make helt selv huset. Peter painted com pletely self house-the ‘Peter painted the house all by h im self, b. * Peter reparerede (helt) alene bilen. Peter painted com pletely alone house-the ‘Peter repaired the car all alone’ (116) a. * Peter make hurtigt huset. Peter painted fast house-the b. Peter m ake huset (meget) hurtigt. Peter painted house-the (very) fast ‘Peter painted the house very fast.’ Like alene ’alone’, selv usually occurs predicate-fiinally. N otice that the exclusive m anner adverbial selv ‘ all by o n eself and alene ‘alone’ share the “exclusive” m eaning w ith the focus particle kun ‘only’, see (117). (117) a. Kun Peter reparerede bilen. Only Peter repaired car-the ‘Only Peter repaired the car.’ The com m on m eaning com ponent o f exclusive selv ‘(by) h im self, the adjective alene ‘alone’, and the focus particle kun ‘only’ (in these sentences) then, consists in identifying the agent o f the predicate as unique. Finally, as already m entioned in section 2.2.5, sentence-final selv can lead to a structural am biguity betw een a reading as adnom inal intensifier and a reading as exclusive m anner adverbial, see (118a). self, even, etc.’, Bui. sam ‘alone, x-self even, etc.’. Notice that in both English and Danish, the word alone can, in certain contexts, be replaced by the PP by him self without changing the meaning, see (i-iii). (i) a. He is alone. b. He is by himself. (ii) a. He came alone. b. He came by himself. (iii) a. Taj e sam. he is alone ‘He is alone.’ b. Taj dojde sam. he came alone ‘He came alone.’ Note that sentence medial selv is not impossible, e.g. Peter havde selv malt huset ‘Peter had himself/also painted the house’, but that it is not interpreted as an exclusive manner adverbial ‘by oneself, all alone’. As shown in section 2.2.5, this sentence-medial selv should be analyzed as a q-floated adnominal intensifier. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 91 (118) a. Peter barberede kongen selv. Peter shaved king-the self (I) ‘Peter shaved the king him self.’ {selv = adnom inal intensifier, m odifying the DP kongen ‘king-the’ to which it is adjoined) (ii) ‘Peter shaved the king all by him self.’ {selv = exclusive m anner ad v erb ial m o d ify in g th e V P) b. Peter barberede kongen helt selv. {selv = excl. adv. intensifier) Peter shaved king-the com pletely self ‘Peter shaved the king all by him self.’ As show n in (1 18b) the insertion o f the adverbial m odifier helt ‘com pletely, all’ between the DP kongen ‘the king’ and the elem ent selv has the effect o f disam biguating in favor o f the exclusive adverbial reading o f selv. See section 2.2.5 for further discussion o f other ways (insertion o f negation, object shift, relative placem ent o f auxiliaries, etc.) to disam biguate betw een different uses o f selv. 2.4 Selv used as scalar additive focus particle meaning ‘even’ Typological studies have shown that in m any languages, the same elem ent w hich is used as adnom inal intensifier is also used as scalar additive focusing adverb (cf. Konig (1998)). While English uses two different words for these functions (i.e. even and himself), Danish, Germ an, French, Persian, Bulgarian and R ussian can use the same w ord in both cases, com pare the (a) and (b) exam ples in (119-124). (119) English: a. The king h im self delivered the speech, (adnom inal intensifier) b. Even the janitor delivered a speech. (scalar additive focus particle) (120) Danish: a. Kongen selv holdt talen. (adnom inal intensifier) king-the self held speech-the ‘The king him self delivered the speech.’ b. Selv visevcerten holdt en tale. (scalar additive focus particle) self janitor held a speech ‘Even the janitor delivered a speech.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 92 (121) German: a. Der Konig selbst hielt die Rede'^^. (adnom inal intensifier) the king self held the speech ‘The king him self delivered the speech.’ b. Selbst der Hausmeister hielt eine Rede, (additive focusing particle) even the janitor held a speech ‘Even the janitor gave a speech.’ (122) French: a. Le roi (lui-)meme a prononce le discours. (adnom inal intensifier) the king him self has pronounced the discourse ‘The king him -self has delivered the speech.’ b. M im e le concierge a prononce le discours. (additive focusing particle) even the janitor has pronounced the discourse ‘Even the janitor has delivered a speech.’ (123) Persian: a. Xod-e pddeshdh dar in jalaseh sherkatkard.isLdnommdWniQnsxfxQv) self king in this m eeting participated ‘The king him self participated in this m eeting.’ b^^. (i) Xod-e pddeshdh ham dar in jalaseh sherkat kard. (focusing particle) se lf king also in this m eeting participated ‘Even the king participated in this m eeting.’ (ii) l\atta pddeshdh (ham) dar in jalaseh sherkat kard. (focusing particle) even king also in this m eeting participated ‘Even the king participated in this m eeting.’ (124) Bulgarian: a. Samijat krai dojde na sdhranieto. self-the king cam e to m eeting-the The king him self cam e to the m eeting.’ b. Samijat krai ne dojde na sdhranieto. self-the king not came to m eeting-the ‘Even the king did not come to the m eeting.’ The G erm an and French exam ples in (121a-b) and (122a-h) are adapted from F.ckhardt (2000:1, ( I), (2)) As illustrated in (i), in Persian the morphologically simple element xod appears to be used both as reflexive and as adnominal intensifier. (i) Pddeshdh az xod-esh bad-esh mi-dgad. (reflexive) king from self-3SG BAD-3SG PROG-agad.(bad amadan ‘to dislike) ‘The king hates himself.’ In the literature on Persian xod has often been analyzed as a simple reflexive. Alternatively, it would, at least theoretically, be possible to analyze it as an adnominal intensifier adjoined to a zero reflexive, xod 0 . Whether such an analysis is compatible with the facts of Persian is beyond the scope of the present chapter. See chapter 5 for R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 93 As show n in (120), in Danish the scalar additive focus particle even (= Dan. selv Peter ‘even Peter’) and the adnom inal intensifier him self {= Dan. Peter selv ‘Peter h im se lf) have the same overt m orphological realization. Furtherm ore, adnom inal intensifiers and even both involve association with focus. Even also obligatorily involves evocation o f a contextually determ ined scale on w hich the value o f the expression that is the focus o f selv is ranked as the least expected/m ost rem arkable/etc. In contrast, w hether adnom inal intensifiers evoke scalar ranking o f their associates with respect to a contrast set o f alternatives depends on the context. In that respect adnom inal intensification is sim ilar to contrastive focus. A nother sem antic difference between the adnom inal intensifier and even lies in the existential presupposition o f even which is not shared by the adnom inal intensifier. That is, while (1 19-124b) presuppose that som ebody other than the janitor also delivered a speech, (119-124a) do not evoke any such presupposition, i.e. the king may or m ay not have been the only individual who delivered the speech. The sentences in (123b(i)) and (123b(ii)), illustrate the fact that, in Persian, the m eaning o f the additive focus particle hatta ‘even’ can be expressed by a com bination o f the adnom inal intensifier xod ‘s e lf and that additive particle ham ‘also, to o ’. G iven that one o f the main sem antic difference between the adnom inal intensifier and the scalar additive focus particle ‘even’ is the presence o f an existential presupposition with ‘even’, this should com e as no surprise. N otice the interesting correlation between syntactic position and the difference between the use o f intensifiers as adnominal intensifiers and additive focus particles: in G erm an, M ainland Scandinavian and French, intensifiers function as adnom inal intensifiers in post-nom inal position and as additive focus particles w hen in they occur in pre-nom inally discussion of a similar intensifier-based analysis of (what appear to be reflexive uses of) h im se lf m English as an adnominal intensifier adjoined to a 0-reflexive. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 94 position, see (120-122a) vs. (120-122b). As illustrated by the Bulgarian exam ples in (124), in the Slavic languages intensifiers are always pre-nom inal both when interpreted as adnominal intensifiers, as in (124a), and when interpreted as scalar additive focus particles, as in (124b). Interestingly, the reading o f intensifiers as scalar additive focus particles can only be obtained in dow nw ard-entailing environm ents. That is, w ithout the negations ‘not’ the elem ent sam ‘s e lf in (124b) would be interpreted as adnom inal intensifiers'*^ and not as scalar additive focus particles. In addition to the above exam ples illustrating the close synchronic relationship betw een focusing particles and intensifiers, there is also ample historical data show ing that these elem ents are intim ately related. Indeed, diachronic studies o f Germ an have established that the use o f selbst as additive focusing particle developed from the adnom inal intensifier selbst around 1700 (cf. Eckhardt (2000))^*^. Siem und (2000) notes a num ber o f sim ilarities betw een the G erm an adnom inal intensifier selbst ‘ h im self and the additive scalar focus particle selbst ‘even’ and concludes that they “are basically the sam e word, sim ply subject to slightly different (syntactic, sem antic, pragm atic) patterns o f use” (Siem und, 2000:ch.5). A sim ilar approach to the sim ilarities betw een intensifiers and additive focus particles will be adopted here. That is, we suggest that the different behavior o f the intensifier selv and the A similar connection between pre-nominal position and dependence on the presence o f a negation is also found with “fronted selv” in Danish, see section 2.2.6. Investigating this relationship between negation and the use pre- nominal intensifiers as scalar additive focus particles may potentially lead to very interesting results. However, since it falls outside the scope o f this dissertation we will simply leave it for further research. The example in (i), from S. Kierkegaard (1843) Frygt og Bceven, illustrates a transitional stage of early Modern Danish in which selv could be used post-nomlnally with the meaning ‘even’. (i) a. Hun [...] kan [....] bringe stene selv til at grcede [...] She can bring stones self to to cry ‘She can make stones themselves cry’ or ‘She can make even stones cry’ Examples like (ii) below seem to indicate that the focus particle even may be used as a post-nominal particle similar to the intensifier x-self. (ii) Mary even came to the meeting. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 focusing particle selv may potentially be accounted for by assum ing that the same m orphem e selv takes on different functions when it occurs in different syntactic positions'*’. 2.4.1 Syntax of the scalar, additive focus particle selv ‘even’ In English the adnom inal intensifier him self and the scalar additive focus particle even are tw o different words. In Danish, however, the sam e elem ent selv is used in both meanings. One way to tell the tw o uses apart is the placem ent o f selv w ith respect to the nominal expression serving as its associate. As shown in the exam ples (119-125), the adnom inal intensifier follows im m ediately after its associate, w hich has to be a nom inal expression, i.e. a DP. In contrast, when used as scalar additive focus particle, selv im m ediately precedes its associate, w hich can be either a DP as in (126a), a pronoun as in (126b), a PP as in (148c), a CP as in (126d-e). (126) a. Selv Peter var i stand til at lose problemet. even Peter was in state to to solve problem -the ‘Even Peter was capable o f solving the problem .’ b. Selv ham ville hun gerne kysse. even him w ould she w illingly kiss ‘She was w illing to kiss even him .’ c. Han hetalte selv fo r hende. he paid even for her ‘He paid even for her.’ d. Selv ndr det regner er han glad. even when it rains is he happy ‘He is happy even when it rains.’ e. Selvom Janet er englcender, taler hun flydende danslc^. even-if Janet is Englishm an speaks she fluent Danish ‘A lthough she’s English, Janet speaks fluent D anish.’ (Allan et al. (1995:472) Note, however, that the intensifier reading of even in (ii) requires a specific stress pattern. See also Eckardt (2001), and Martin (1975) for other analyses attempting unification of the same uses of selbst in German and meme in French. Danish has two conjunctions meaning ‘though, although, even though’ which can be used alternatively in most contexts: sk0nt and selvom. While skont is monomorphemic, selvom is clearly a compound consisting of selv ‘even’ + the conjunction am ‘if, whether’. While these two words can generally be used interchangeably they do differ in one respect: “Both selvom and skont can be used for factual utterances, but for hypothetical statements, selvom is generally used” (Allan et al. (1995:473)). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 96 N ote that unlike English even, Dan. selv (as well as Ger. selbst and Slavic sam) cannot have verbal associates, see (127) vs. (128). (127) Peter even BIKED around the block. (128) *Peter selv CYKLEDE rundt am blokken. Peter even BIKED around about block-the The Danish counterpart o f Eng. even w hen used with verbal associates is endog ‘even’, see (129). (129) Peter endog CYKLEDE rundt om blokken. Peter even BIKED around about block-the ‘Peter even BIKED around the block.’ A nother difference betw een the post-nom inal adnom inal intensifier selv ‘h im self and the pre- nom inal (or rather ‘pre-focaf, i.e. preceding its associate) additive focus particle selv ‘even’ is stress-placem ent. Being a focus particle, selv ‘even’ is associated w ith focus. This focus is realized phonetically as stress on the associate o f selv ‘even’, i.e. on the constituent follow ing the focus particle, see (130a). (130) a. Selv PETER loste opgaven. ‘Even Peter solved problem -the.’ b. Peter SELVloste opgaven. ‘Peter him self solved problem -the.’ In contrast, in constructions involving the adnom inal intensifier selv ‘h im self it is the elem ent selv w hich is stressed, see (130b). It could, perhaps, be argued that the difference in stress placem ent between constructions involving adnom inal intensification and constructions involving the focus particle selv ‘even’ is m erely a phonological phenom enon which follows from the fact that in Danish constituent stress fall on a prom inent constituent-final sub- constituent. If that is the case, then the difference in stress placem ent should be considered as a side-effect o f the post-nom inal vs. pre-nom inal placem ent. From this point o f view the R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 97 crucial difference between the adnom inal intensifier use o f the elem ent selv and its use as an additive focus particle is syntactic rather than prosodic. 2.4.2 Semantics of the scalar, additive focus particle selv ‘even’ D escriptively the different theories o f focus particles in the literature agree that the main sem antic contribution o f selv used as the scalar, additive focus particle selv ‘even’ consists in evoking tw o im plicatures. The so-called existential im plicature captures the additivity o f selv ‘even’ by requiring the proposition to be true for at least one o f the m em bers o f the focus generated contrast set o f alternatives to the referent o f the associate o f selv. The second im plicature is responsible for ranking the truth o f the proposition lower on a scale o f probability than the truth o f any o f the alternative propositions in the contrast set. In the follow ing we present an analysis if selv ‘even’ closely inspired by Karttunen and Peters (1979) and Eckardt (2001). First consider the sentences in (131) and (132). (131) a. Even the king came to the meeting. b. Selv kongen kom til model. even king-the cam e to m eeting-the ‘Even the king cam e to the m eeting.’ (132) The king came to the meeting. As far as truth conditions are concerned the sentence in ( O la ) , and its Danish equivalent in (131b), and the sentence in (132) are equivalent. In other words, the word selv ‘even’ plays no role in determ ining the truth conditions and, as a consequence, (131) and (132) express the same proposition. However, this does not m ean that the presence o f even in (131) contributes nothing to the m eaning o f the sentence. According to Karttunen and Peters (1979), the sem antic im port o f the particle even can best be captured in term s o f conventional implicatures. In addition to asserting that the that king cam e to the m eeting the sentence in (132) also allow the hearer to conclude that the speaker believes in the truth o f (133a-b). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 98 (133) a. Other people besides the king came to the meeting. b. O f the people under consideration, the king is the least likely to come to the meeting. The sentences (133a) and (133b) represent w hat is usually referred to as the existential and scalar im plicatures o f even, respectively. Following G rice, these im plicatures can be defined as conventional (rather than conversational) since they do not arise from the interaction o f general conversational principles (e.g. cooperation principle, etc., see G rice (1975)), and the specific properties o f the sentences them selves and their surrounding contexts. Being conventional, these im plicatures cannot be overridden by the context, see (134). (134) #Even the king came to the meeting but no one else did. The fact that (134) constitutes a contradiction thus w itnesses the conventional nature o f the existential im plicature expressed in (133a). N otice, however, that since the propositions in (133) are im plicated rather than asserted (as is (132)), their falsity is judged less detrim ental to the com m unication process than the falsity o f (132). That is, in situations w here the proposition in (132) is false uttering (131) w ould be considered as act o f lying, or m iscom m unication. In contrast, in situations w here (133a,b) are false, uttering (131) would be judged less harshly. In such cases the speaker w ould likely be considered to have uttered a sentence w hich is at least partially true. Rather than blam ing the speaker for m isrepresenting the truth the hearer m ight sim ply correct his m isconceptions by uttering either (135) or (136)'” . (135) Well yes, he did come to the meeting; but that is ju st as one should expect. (136) Well yes, he did come to the meeting; but, in fact, he was the only one who showed up. Karttunen and Peters mention yet another test to distinguish between what is asserted and what is merely conventionally implied by a given sentence: “The distinction between these two aspects of meaning in (153) can be brought out even more clearly by considering the meaning of complex sentences such as (i), which contains (153) in an embedded position”(Karttunen and Peters (1979)). (i) I ju st noticed that even the king came to the meeting. “Sentence (i) says that the speaker has just noticed that the king came to the meeting. It does not mean that he has just noticed that other people came to the meeting or just noticed that the king is the least likely person to have come. In (i), the meaning o f notice applies only to the proposition expressed by (132) - not to (133a) or (133b) or to the conjunction of (132) and (133).” (ibid. p. 13, numbers of examples changed fit present context) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 In contrast, in situations in which (132) is false, it w ould be im possible to correct the mistake o f the speaker o f(1 3 1 ) by uttering (137). (137) Yes, you wouldn ’ t expect the king to have come to the meeting; as a matter o f fact, he d id n ’ t come. The insights o f the K arttunen & Peters (1979) sum m arized above, form the basis for the analysis o f selv ‘even’ proposed here. First consider the follow ing sentences which illustrate the sim ilarities betw een focus and the additive focus particle even, see (138a,b). (138) a. [The king]p cam e to the meeting. b. Even [the king]p cam e to the m eeting. Both sentences in (138) contain a focused elem ent, i.e. [The kingjp. A ccording to B ooth’s (1992) theory o f focus, the m eaning o f any sentence containing a focused elem ent com prises an ordinary sem antic value, i.e. the proposition expressed by the sentence, see (139), and a focus sem antic value, i.e. a contrast set, see (140). As shown in (140), the contrast set generated by focus is a set o f propositions o f the type x came to the meeting in which the focused elem ent has been replaced by entities taken from the set (139) O rdinary sem antic value o f 138a,b): a. [[The kingjp cam e to the m eeting]]” = 1 iff the king came to the meeting. b. [[Even [the kingjp cam e to the m eeting]]” = 1 iff the king came to the meeting. (140) Focus sem antic value o f (138a,h): a. [[ [The kingjp cam e to the m eeting j]* ^ = {came-to-the-meeting(x) \ x e D f b. [[Even [the kingjp cam e to the meeting]] * = {came-to-the-meeting(x) \ x e D f For the form alization o f even w e w ill adopt a slig h tly m o d ified version o f the analysis proposed in Eckardt (2001) w hich uses B ooth’s ordinary and sem antic values prim itives, see (141) and (142). The set D,. is usually assumed to be constrained pragmatically to contain only those entities which are ‘relevant’ in a given context. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 100 (141) Analysis o f the additive focus particle even : even + S a. A ssertion: [[ S ]]° b. Scalar im plicature (i) '^P ^ [[ S ]]*^\{[[ S ]]°}: [[ S ]]” <c p Existential im plicature(ii) 3p g [[ S ]]^\ {[[ S ]]°}: p = 1 (142) Scale o f expectedness (Eckardt (2000)1: Let <c be an ordering o f propositions according to their probability or expectedness in a given context c. (“p <c q” reads like “p is less expected, less probable, m ore surprising than q”) A nalyzing (138b) according to (141a) we first determ ine the assertion expressed by (160b) to be its ordinary sem antic value, see (143). (143) Assertion o f (160b) = Ordinary sem antic value o f (138b): [[ [The king]p cam e to the m eeting ]]° = 1 iff the king cam e to the meeting. N ext, given the focus sem antic value o f (138b), see (144), and the form alization o f the scalar im plicature in (1 4 lb (i)) and (142), we can capture the scalar im plicature o f (138b) as in (145). (144) Focus sem antic value o f (138b): [[Even [the kingjp cam e to the m eeting]] ^ = {came-to-the-meeting(x) \ x e D^} ‘the queen came to the m eeting’, ‘the prince cam e to the m eeting’, ‘the butler cam e to the m eeting’ (145) Scalar im plicature o f (138b): Vp G { {came-to-the-meeting(x) | x g A}\{cam e-to-the-m eeting(the king)}}: came-to-the-meeting(the king) is less likely than p (145) can be paraphrased as follows: for all propositions p belonging to the set o f propositions o f the type ‘x came to the m eeting’, where x the king, the proposition ‘the king cam e to the m eeting’ is less likely to be the case than p, see also (168). (146) “For all alternatives o f the king, it would have been m ore probable for them to come to the m eeting” = “It w ould have been less surprising, had the queen come to the m eeting” “It would have been less surprising, had the prim e m inister com e to the m eeting” “It w ould have been less surprising, had the arch-bishop to the m eeting”, ...... ' The analysis of even given here is essentially the same as the one proposed in Eckardt (2001). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 101 Finally, the existential im plicature o f (138b) can be captured using the form alization in (141b(ii)), see (147). (147) “Existential” im plicature o f (138b): 3p G { {came-to-the-meeting(x) I xGZ)e}\{came-to-the-meeting(the king)}}: p = l (147) says that am ong the propositions p belonging to the set o f propositions o f the type ‘x cam e to the m eeting’, w here x the king, there is at least one true proposition, i.e. there is at least one other individual besides the king who came to the m eeting. Semantically, the difference thus lies in the tw o im plicatures o f even, see (145) and (147), neither o f w hich is shared by the adnom inal intensifier. That is, while (138b) presupposes that som ebody other than the king also cam e to the m eeting, the sentence in (138a) does not evoke any such presupposition, i.e. the king may or may not have been the only individual com ing to the m eeting. This difference between the adnom inal intensifier and the focus particle is illustrated by the clefts (which presuppose uniqueness) in (148). (148) a. *Det var selv Peter der kom fo r sent. it was even Peter who cam e for late ‘*It was even Peter who was late.’ b. Det var Peter selv der kom fo r sent. it was Peter se lf who cam e for late ‘It was Peter him self who was late.’ In section 2.2.3.2.1 the so-called unique identifiability constraint on adnom inal intensification (i.e. only definite nom inal expressions or nominal expression w hich refer to specific or uniquely identifiable entities may be adnom inally intensified) was mentioned. The sentence in (149a) shows how this constraint blocks adnom inal intensification o f indefinite DPs. In contrast, the sentence in (149b) shows that this constraint does not affect the distribution o f the scalar additive focus particle selv ‘even’. (149) a. *En student selv loste opgaven. a student self solved problem -the ‘*A student him self solved the problem .’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 102 b. Selv en student l0ste opgaven. even a student solved problem -the ‘Even a student solved the problem .’ Sum m ing up, we conclude that while both the scalar additive focus particle selv ‘even’ and the adnom inal intensifier selv ‘h im self are associated w ith focus and generate contrast sets o f alternatives, they differ in that form er carries w ith it tw o conventional implicatures, viz. (145) and (147), while the latter has none. Exactly w hat explains the differences betw een these tw o hom ophonic elem ents is still unknown. It seems likely that part o f the answ er to this question is to be found in the fact fact that the intensifier is located after its associate while the focus particle is pre-associate. W hile it is obviously too early to reach any conclusion we assum e that it m ight be possible to explain the differences between the focus particle selv ‘even’ and the adnom inal intensifier selv ‘h im self as follow ing from their syntactic differences. That is, we suggest that the different behavior o f the intensifier selv and the focusing particle selv may potentially be accounted for by assum ing that the same m orphem e selv takes on different functions when it occurs in different syntactic positions^^. Beyond these speculations we have, at this point, no satisfying account o f the exact relationship between the focus particle selv ‘even’ and the adnom inal intensifier selv. Since this topic is orthogonal to the m ain proposal o f this dissertation, we leave it for further research. 2.5 Selv used as a noun. Like its English counterpart the Danish word selv may be used as a noun by itself, see (150a- 150b). (150) a. selvets begred self-the-PO SS concept ‘the conception o f self (/the ego)’ Cf. Siemund (2000:ch. 5) who proposed the same analysis. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 103 b. Selvet er en central del o f psyken. self-the is a central part o f psyche-the ‘The self is a central part o f the psyche.’ c. Kun levende vcesner kan have et selv. only living creatures can have a self ‘Only living creatures can have a s e lf ’ In (150c) selv occurs as a noun preceded by the indefinite article. The sentences in (150a-b) show that selv behaves ju st like any other noun by taking the suffixal definite article see (150a-b), and the s-genitive, see (150a). N ot all languages allow adnom inal intensifiers to be used as nouns. As observed in Safir (1996), in the Rom ance languages intensifiers are derived from adjectival bases, e.g. French meme ‘sam e’. In these languages it is often the first person singular personal pronoun (e.g. Fr, mol ‘m e’/ ‘s e lf )^’* w hich is used as the noun m eaning ‘s e lf rather than the adnom inal intensifier, see (151c)^^ Unlike English, which does not allow this use o f the first person singular pronoun (e.g. *one’ s own m e/I vs. o n e’ s own self), Danish allows both the pronoun and the intensifier to be used nom inally with the sam e meaning, com pare (151a) and (151 b)^'’. ” ‘In Danish the definite article (or end article) singular is added to the end of the noun as an affix: -(e)n for common gender nouns and -(e)t for neuter nouns’ (Allan et al. 1995:54). Note that selv is considered a neuter noun, selv-et ‘the self. According to the Dictionary Robert, the use of the first person pronoun moi ‘me’ as noun was first attested in 1583. Note, however, that le meme ‘the self is an accepted concept in philosophical jargon (cf. Ricoeur’s Le Meme et I ’ Autre). In some instances, it appears that even the possessive intensifier egen ‘own’ may be used nominally meaning ‘one’s inner self, etc,’, see (ia) and (ib). (i) a. Hvilken forunderlig, forjcetningsfuld Tid var det ikke, hvor scelsomt ikke med 0ren at hore sin Sjcels utydelige, londomsfulde Hvisken klinge frem i Virkelighedens Luft, som vildt udfordrende Lurtoner, som Brag a f Kolleslag paa Tempelmure, .som Hvirt a f Davidsstene paa Flugt mod Goliathspander og som sejerssikker Fanfare. Det var som at hore sig selv tale i fremmede Tunger, med fremm ed Klarhed ogfremm ed Magt om det, der var Ens dybeste, inderste eget. (J.P.Jacobsen (1880) Niels Lyhne, p. 65) [...] which was one’s deepest innermost own ‘[...] which was one’s deepest, innermost se lf’ b. [...] der var Ens dybeste, inderste selv. which was one’s deepest innermost self ‘[...] which was one’s deepest, innermost se lf’ In (ia) the word egen ‘own’ is used with the meaning ‘(inner)self. (ib) shows that egen ‘own’ can be replaced with selv ‘self without changing the meaning of the sentence. This example, thus supports the assumption that selv R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 104 (151)^’ a. One should protect o n e’ s own self/*I. b. Man hor vcerne om ens eget jeg/*mig/selv. one ought protect about one’s own I/self ‘One should protect one’s own self.’ c. Chacun doit proteger sa propre *je/moi/*me/*meme/personne. everyone ought protect his own m e/sam e/person ‘Everyone ought to protect his own self/person.’ The fact that intensifiers may be used as nouns has been used as an argum ent in favor o f classifying them as nom inal expressions. H owever, rather than entering the som ew hat futile debate about what word-class it belongs to we sim ply take the elem ent selv ‘s e lf to be a m orphem e/root which can be m ade into a noun, adjective, or adverb in different contexts. As will be discussed in chapter 3, we will argue that the French intensifier -wiewe, which, unlike selv, clearly has adjectival origins, should also be analyzed as a root which can take on the characteristics o f different w ord-classes depending on its syntactic position. That is, cross- linguistically intensifiers share a num ber o f core-properties despite their different historical origins and the particular range o f m orphological realizations (adjective, noun, particle, etc.) w hich they happen to have. 2.6 Adjectival forms of selv. selve and selveste In addition to the uninflected, post-nom inal, appositional intensifier selv, M odern Danish also has an inflected, pre-nom inal, adjectival form o f the intensifier, i.e. selve, which is used ‘selP and egen/eget ‘own’ are suppletive variants o f the same word. See also the discussion of the synonymous compounds egenrisiko ‘own-risk’ and selvrisiko ‘self-risk’, in chapter 7. The fact that Danish uses the subject form of the first person personal pronoun in these cases rather than the object pronoun mig ‘me’), while French uses the emphatic object pronoun moi ‘ME’ rather than either the subject pronoun Je ‘I’ or the non-emphatic object pronoun me ‘me’, is probably due to phonological factors, i.e. stressability. The French subject pronoun je ‘1 ’ cannot be stressed on its own (In the colloquial language the stressed form of Je is moi-je ‘ME-F = ‘I m yself), while moi can. Since nouns have non-contrastive word-stress, it is thus to be expected that moi is preferred over je in nominal uses. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 105 im m ediately preceding a noun^* (in the definite form ) to em phasize the latter very strongly^’, see(152)-(154). (152) Selve huset er ikke meget vcerd. (house=core vs. rest o f property=periphery) se lf house-the is not much worth ‘The house itself isn’t w orth m uch.’ (153) Selve hiskoppen kommer paa bes0g. se lf bishop-the com es on visit ‘The bishop him self is paying a visit.’ (154) Vi bor i selve Kahenhavn. (C openhagen=core vs. suburbs=periphery) we live in self Copenhagen ‘We live in Copenhagen its e lf’ W hen used w ith nouns denoting places or inanim ate objects pre-nom inal, adjectival selve always has an core-periphery reading, with the D P m odified by selve being the core, e.g. (152) and (154). As shown by the translations o f the exam ples in (152-154), the sem antic contribution o f this adjectival form o f selv is thus very close to that o f the adnom inal intensifier*’® . Post-nom inal selv and pre-nom inal selve differ in that selve im poses a much stronger specificity constraint on its com plem ent DPs. Com pare the grammatical sentences in (156) with the ungram m atical sentence in (155). (155) *Selve en sjcellcender har fortalt mig det. self a Zealander has told me it ‘A Zealander him self told m e.’ The element selv can be combined with the adj. samme “same” in the somewhat idiomatic expressions selvsamme adj. ‘seif-same, the very same’ and selvanden adv. ‘with one other (person): (i) Den selvsamme dag, holdt han op med at ryge. The self-same day , hold he up with to smoke ‘The very same day, he stopped smoking.’ See chapter 7, section 7.3.3 for more discussion of idiomatic use o f selv ‘self in nominalizations. The examples in (152-4) are adapted from similar examples in Allan, Holmes and Lundsksr-Nieisen (1995:170ff). Indeed, in Icelandic adnominal intensifiers are pre-nominal adjectival forms which may adjoin to full lexical DPs as well as pronouns and reflexives: (Icel.) sjalva sig ~ (Dan.) sig selv. (Icel.) sjalva Peter = (Dan.) Peter selv. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 106 (156) En sjcellcender selv har fortalt mig det. a Zealander self has told me it ‘A Zealander him self told m e.’ W hile adnom inal selv can be adjoined to indefinite DPs w hich have specific reference, adjectival selve cannot under any circum stances take an indefinite DPs as its com plem ent. Furtherm ore, unlike adnom inal selv, adjectival selve cannot modify pronouns, e.g. *selve je g ‘self m e’, *selve ham ‘self ham ’ vs. je g selv ‘I m y se lf, ham selv ‘him him s e lf, etc. Finally, pre-nom inal, adjectival selve is restricted to a small class o f nouns denoting individuals o f high social status, com pare (157a) and (157b). (157) a. 11*Selve Peter kommer p a hesog. self Peter comes on visit ‘Peter him self is paying a visit.’ b. Selve hiskoppen/Dronningen/prcesidenten kommer pa hesog. self bishop-the/Q ueen-the/president-the com es on visit ‘The bishop him self is paying a visit.’ Being an adjective selve may take the so-called m orphological superlative by adding the superlative suffix -ste, i.e. selveste see (158-159)^'. This superlative form im plies an even higher degree o f intensification and is used exclusively w ith nouns referring to individuals o f high status, e.g. the Queen, the King, the prim e m inister, etc. That is, unlike the post-nom inal invariable adnom inal intensifier selv and the basic form o f the pre-nom inal adjectival intensifier selve w hich can take both anim ate and inanim ate (e.g. huset selv ‘the house itse lf « selve huset ‘the house itse lf in (152)) associates, the superlative form selveste is com patible only with [+human] nominal associates referring to high status individuals*’^. Cf. Allan et al. (1995:105): ‘Danish adjectives possess a basic (positive) form which is inflected according to number, gender and species [..], a comparative form which is uninflected and a superlative form which is inflected according to species alone. [...] The endings -ere, -est are added to the basic form.’ Basic Comparative Superlative pcen ‘nice’ pcenere ‘nicer’ pcenest ‘nicest’ Adjectives which cannot form the comparative and superlative forms by adding the endings -ere, -est use the words mere ‘more’ and mest ‘most’ for form periphrastic comparative and superlative forms: Basic Comparative Superlative intelligent ‘intelligent’ mere intelligent ‘more intelligent’ mest intelligent ‘most intelligent’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 107 (158) Vi blev inviteret til middag has selveste direktaren. we were invited to dinner at self-est boss-the ‘We were invited to dinner with the m anaging director him self.’ (159) Selveste dronningen tog imod os. self-est Q ueen-the took against us ‘The Queen herself welcom ed u s.’ Unlike selv, selve and the superlative selveste occur in pre-nom inal position, i.e. the normal position for a Danish adjective. As for the syntactic analysis o f selve and selveste, tw o solutions com e to mind; one could treat them either as adjuncts to DP (cf. Browning (1993) and Bernstein (1991,1993)) or as heads selecting DPs as their com plem ents (Sanchez (1995)). Here the latter solution will be adopted for selve/selveste. In contrast, as already discussed, the post-nom inal intensifier selv is best analyzed as a DP adjunct. Cross-linguistically intensifiers either behave as adjectives (e.g. French meme, Spanish mismo, etc.) or as DP-adjuncts (e.g. G erm an selbst, Chinese ziji, etc.). W hile in the Slavic and Rom ance languages the adnom inal intensifiers tend to display adjectival behavior, intensifiers in the G erm anic languages can behave either like adjectives (e.g. Swedish sjalva) or uninflected appositional intensifiers (e.g. Ger. selbst). From a typological perspective D anish is thus interesting in that it has both adjectival and appositional forms o f the adnom inal intensifier. The fact that adjectival form s o f selv, i.e. selve and selveste, can be shown to be subject to the principle o f unique identifiability (and thus appears to be subject to the principle o f contrastiveness as well, although this cannot be rigorously tested since selve cannot attach to reflexives in Danish), is a strong indication that selve and selveste are but different m orphological realizations o f the intensifier selv. Cross-linguistic evidence from Swedish and Icelandic, which both have adjectival intensifiers, confirm s this analysis. A lthough an “ Kote that Spanish also have what looks like a superlative form of the adjectival intensifier mismo ‘him self, viz. mismisimo, which appear to function exactly like selveste in Danish: (i) Elisa hablo con el mismisimo Papa. (= Sanchez, L. (1994:480(6))) Elisa talk with the sarae-SUPERLATIVE Pope R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 108 interesting question in itself, the different behavior o f the adjeetivai uses o f selv, i.e. selve and selveste, does bear directly on the issues discussed in this dissertation and will not be treated in any m ore detail here. 2.7 Selv ‘se lf in nominalizations As its counterparts in other Germ anic languages the elem ent selv may occur as part o f com pounds. The m ajority o f such com pounds are deverbal nom inalizations in w hich the elem ent selv appear to function as the direet object, see (160a-c). (160) Deverbal com pound nouns containing selv. C om pound: C orresponding verb: a. selvbedrag ‘self-deception’ bedrage ‘to deceive’ b. selviagttagelse ‘self-observation’ iagttage ‘to observe’ c. selvkritik ‘self-criticism ’ kritisere ‘to criticize’ It has been argued that in these cases selv should be seen as the word-internal counterpart o f reflexive pronouns. However, in ehapter 7 we will argue that it is both possible and theoretically m ore advantageous to consistently analyze such word-internal uses o f selv as adnom inal intensifiers intensifying phonetically unrealized reflexives, see (161). (161) a. [ \ 0 selv\ {bedrag^ REFL self deceit ‘self-deceit’ b. *[ [ 0 ] [bedragW REFL deceit (162) a. Peter bedrager sig selv. Peter deceives REFL self ‘Peter deceives h im self b. *Peter bedrager sig. P eter d eceiv e s R E F L Since the verb bedrage ‘deceive’ is anti-reflexive, it requires the intensified form o f the reflexive, see (162a) vs. (162b). In order to have a reflexive reading, nom inalizations based on this verb thus have to have intensified 0-reflexives, see (161a) vs. (161b). ‘Elisatalked to the Pope himself.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 109 W hile the in exam ples in (160a-c) above selv may be argued to function as a kind o f word-internal version o f the reflexive pronoun sig ‘R E FL ’, such an analysis is clearly not possible in (I63a-c) w here selv/self- seem to function as interpreted as an exclusive adverbial intensifier m eaning ‘by o n eself. (163) C om pound: C orresponding verb: a. selvbygger ‘person who builds bygge ‘to build’ (Danish) his house w ith his ow n hands.’ b. self-storage ‘place where you to store store stuff by yours- self/on your ow n’ c. self-cleanecf^ ‘a (house) cleaned to clean by o n e se lf d. self-cleaning ‘an (oven) which to clean cleans itse lf Based on the above exam ples we conclude that word-internal intensifiers can be either: (i) adnom inal intensifiers (160a-c),(161a), (ii) exclusive adverbial intensifiers (163a-c). So far we have not com e across any evidence suggesting that word-internal selv may function as either additive scalar focus particle, or any o f the other uses o f selv discussed above. See chapter 7 and appendix 1 for a m ore detailed analysis o f 5e//^nominalizations based on the account o f intensifiers proposed here. 2.8 Idiomatic uses of the element selv Finally, in the interest o f com pleteness, it should be m entioned that the elem ent selv also occurs in a num ber o f idiom atic expressions, see (164-165). (1 6 4 ) A: T akfor i aftes. (A llan et al. (1 9 9 5 :1 7 0 ) thank for in evening ‘Thank you for (the party, etc.) last night. The word self-cleaned is used (and coined?) by Barbara Ehrenreich in her (2001) book Nickel and Dimed, p. 91: “I have never employed a cleaning person or service [...] Partly this comes from having a mother who believed that a self-cleaned house was the hall-mark of womanly virtue”. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 10 B; Selv tak! self thank Thank you! (w ith stress on ‘you’.) (165) a. Det siger sig selv. it says REFL self ‘That is obvious.’ b. Doren abnede sig a f sig selv.^'^ door-the opened REFL o f REFL self ‘The door opened all by its e lf’ W hile som e o f these may potentially be accounted for in term s o f the uses o f selv described in sections 2.2-8 above, e.g. the use o f selv in (165a,b), w hich seems to be som ehow related to the exclusive adverbial intensifier, we will leave the exact analysis o f the rem aining idiom atic cases for further study 2.9 Cross-linguistic perspective on the range of uses of intensifiers. The table in (166) sum m arizes all the uses o f the elem ent selv in Danish described in sections 2.2-8'^^. It also includes a mini typological survey show ing the ranges o f uses corresponding intensifiers have in English^’, French and Chinese^**. ‘ ’‘ 'N ote that this use of the PP a f sig selv ‘by REFL self could potentially be adduced as evidence in favor of arguing for a hidden preposition in secondary predication constructions, e.g.: Peter painted the house [PP [P’ [P 0 ] himself\f\. The list examples of idiomatic uses of selv in Danish given here is far from exhaustive. Due to space limitations only a few exam ple sentences arc p rovided here. Fronted selv described in section 2.2.6 has not been included in the table, since it seems to be derived from the normal post-nominal adnominal intensifier by general syntactic processes (of fronting) and thus do not deserve to be included separately in a typological survey. Examples illustrating the different uses o f the adnominal intensifiers of the x-je/fform in English can be found in chapter 5. Examples illustrating the different uses o f the adnominal intensifier |=ici ziji ‘self-self in Chinese can be found in chapter 6. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . Ill (166) Typological survey o f different uses o f adnom inal intensifiers: Danish English French Chinse a. Adnominal int. kongen selv the king himself le roi (lui-)meme guo-wang (ta) ziji b. Intensified reflexive sig selv 0 himself soi-meme/lui-meme ta ziji/ 0 ziji c. Intensifled pronoun ham selv (him) himself (lui) lui-meme (ta) ziji d. “Exclusive adverbial” selv himself lui-meme ziji e. Q-floated adnom. inten. selv himself lui-meme f. Scalar add. focus part, selv - meme g. Nominal uses selv/Jeg self/*I *meme/moi h. Positive adjective selve very/actual mime ‘same (kind o f)’ i. Superlative adjective selveste j. Self-nominalizations selv *himself/self *meme-/auto-/sui-/etc.. *ziji/zi-/-Ji The typological survey or intensifiers in (166) shows that the elem ent selv in Danish has a w ider range o f uses than its counterparts in m ost other languages. Though such cross- linguistic data should be used w ith great caution, it still provides a rough indication o f how far attem pts o f unifying all the uses o f intensifiers like selv m ay be pushed. A lthough the elem ent selv can be used in all the ways listed in (166a) through (166j) this is not the case for its counterparts in English, French and Chinese. Only the first 4 uses, i.e. (166a-d) seem to be representative o f the range o f core uses o f intensifiers cross-linguistically. 2.10 Conclusion The contribution o f this chapter can be divided into three parts. First, in section 2.2 we have shown that the sem antic contribution o f adnom inal intensification is sim ilar to that o f focus (i.e. both involve the generation o f a contrast set o f alternatives) and is subject to the contrastiveness condition as form ulated in (62), repeated here as (167). (167) C ontrastiveness condition on adnom inal intensification: A nom inal expression (DP, pronoun, reflexive) can only be intensified adnom inally if it can be contrasted w ith other ex p ressio n s in the con text in w h ich it is found. S econd, in sectio n s 2 .2 .5 -7 w e h av e sh o w n th a t all th e d ifferen t u ses o f selv listed in (I6 8 a ,i-v ) b e lo w can b e e x p la in e d as su b ty p es o f a d n o m in al in te n sific a tio n fallin g u n d er p rin c ip le (167). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 12 (168) D ifferent uses o f selv. Examples: Section: a. adnom inal intensification i. intensified DPs (1 ),(1 4 ) (sect. 2.2) ii. intensified reflexives (4) (chap. 3, section 3.3) (1) selv in nom inalizations (160),(163d) (sect. 2.7, chap. 7) iii. intensified pronouns (5) (chap. 3, section 3.4) iv. q-floated intensifier (6), (77) (sect. 2.2.5) V. fronted intensifier (108) (sect. 2.2.6) b. selv in secondary predication constructions (7), (113) (sect. 2.3) (1) selv in nom inalizations (163a-c) (sect. 2.7, chap. 7) c. scalar additive focus particle ‘even’ (8), (120b) (sect. 2.4) d. selv as a noun (9), (150) (sect. 2.5) e. adjectival form s o f the intensifier (10), (152-4) (sect. 2.6) g- idiom atic uses o f selv (164-5) (section 2.8) In spite o f superficial differences we have shown that it is possible to unify the analysis o f selv occurring in constructions involving intensified DPs (189a,i), q-floated intensifiers (168a,iv), and fronted intensifiers (168a,v). W e furtherm ore aim to unify the analysis o f adnom inal intensification and com plex reflexives (I68a,ii), see chapter 3, section 3.3, and com plex pronouns (168a,iii), see chapter 3, section 3.4, and to show that this unification can only be achieved if binding and intensification are assum ed to belong to different m odules o f the gram m ar. W ithin this approach, binding theory deals with the interaction betw een intrinsic properties o f nom inal expressions and purely syntactic locality principles, e.g. principle A and B, while the theory o f intensification consists o f pragm atic and sem antic constraints on the distribution o f intensifiers. The argum ent is that this m odular approach com bined with the focus-based theory o f intensification proposed here stand a better chance at unifying the different uses o f adnom inal intensifiers listed in (168ai-v) than the predicate-based approaches to binding w hich assum e selv to be a reflexivizing particle (cf. Reinhart & Reuland). Third, in sections 2.3-2.8 we have illustrated the fact that the elem ent selv occurs in a large num ber o f different contexts with w hat som etim es seem s to be very different syntax and sem antics, see (168b-g). In spite o f these differences, we have argued that it is possible to unify the analysis o f all the uses o f the intensifier selv. As shown in (168), there are only five R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 13 basic uses o f the w ord selv, i.e. as adnominal intensifier (168a), in secondary predication constructions (’’exclusive” adverbial) (168b), as a scalar additive focus particle (168c), as a noun (168d), and as an adjective (168e). As argued briefly in section 2.7 and in m ore detail in chapter 7, the instances o f selv found in nom inalizations can be shown to be reducible to either intensified zero reflexives (168a,ii,(l)) or word internal secondary predication (168b,(l)). As for the idiom atic uses oiselv in (168g) discussed in section 2.8, they are presum ably ultim ately be reducible to one or the other o f the three main uses o f selv. O f the five basic uses (168a-e), the nom inal and adjectival uses (168d-e) can be accounted for as deriving from the intensifier selv by regular processes o f word formation. At first glance, the use o f selv as an exclusive m anner adverbial/secondary predication (168b) seems to be very different from the adnom inal intensifier. However, it may be possible to analyze it as an intensifier too. Rather than being adjoined to a DP as the adnom inal intensifier, it is a subject/agent oriented adverbial adjoined to the VP, i.e. it takes agentive subjects as its associates and contrasts them w ith other potential agents. Finally, as discussed in section 2.4 there is am ple historic evidence that the use o f selv as a focus particle (168c) evolved from the adnom inal intensifier. We believe that it may be possible to account for the differences betw een them by assum ing that they are basically the same m orphem e selv ‘s e lf w hich takes on different sem antic properties when occurring in different syntactic positions (cf. Siem und 2000, chap. 5). In other words, although we do not develop it in detail here we do believe it is ultim ately possible to arrive at a unified account o f all uses o f the intensifier selv ‘s e lf in (168). It is, how ever, the use o f selv ‘s e lf as an adnom inal intensifier (168a,i-v) w hich is the m ain focus o f this dissertation. In this chapter we have show n that (168i,iv,v) all fall under the sam e principle (167). C hapter 3 will be devoted to show ing that com plex reflexives (e.g. Dan. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 114 sig selv ‘REFL s e lf ) and pronouns (e.g. Dan. han selv ‘he h im se lf) are analyzable as intensified nominal expressions which also fall under the contrastiveness principle (167). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 15 Chapter 3 Binding: Reflexives and Pronouns in Modern Danish 3.1 Introduction. Binding of reflexives and pronouns in Danish. The purpose o f this chapter is two-fold. First, the distribution o f sim ple reflexives and pronouns, e.g. Dan. sig ‘R EFL’ and ham ‘him ’, will be described and an account o f these elem ents will be proposed. Second, it will be argued that com plex reflexives, e.g. Dan. sig selv ’REFL s e lf, and com plex pronouns, e.g. ham selv ‘him self/him h im self, are best accounted for by assum ing that binding and intensification are two independent m odules o f the grammar. B inding is taken to be a system o f syntactic principles w hich determ ines the distribution o f reflexives and pronouns. R ather than adopting a predicate based approach to binding (cf. Reinhart and Reuland (1993)), we propose a nominal approach in which binding facts follow from the interaction o f lexical features o f nominal expressions and syntactic dom ain constraints. In contrast, as discussed in chapter 2, the m odule o f intensification consists o f sem antic and pragm atic constraints on the distribution o f the adnom inal intensifier selv ‘s e lf. That is, it will be argued that the behavior o f com plex reflexives and com plex pronouns follow from the fact that they are adnom inally intensified form s o f their simple counterparts, e.g. sig ‘R EFL’ and ham ‘him ’, and that their specific properties can be derived com positionally from their constituent com ponents, i.e. siglham ‘R EFL/him ’ and the adnom inal intensifier selv ‘s e lf '. The rem ainder o f this section contains an introduction to binding o f reflexives and pronouns in Danish. ' For similar proposals which also advocate the independence of intensification and binding and outline compositional analyses of complex reflexives and pronouns, see McKay (1991), Baker (1995) and Kdnig and Siemund (1999). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 116 D anish distinguishes betw een sim ple and com plex reflexives, i.e. sig vs. sig selv, see (1). (1) a. Peter vasker sig/sig selv. (neutral) Peter washes REFL/REFL self ‘Peter washes (him self).’ b. Peter hader *sig/sig selv. (anti-reflexive) Peter hates REFL/REFL self ‘Peter hates h im se lf’ c. Peter hviler sig/*sigselv. (inherently reflexive) Peter rests REFL/REFL self ‘Peter is resting.’ Predicates differ w ith respect to w hether they allow both simple and com plex reflexives (la ) or w hether they allow either only com plex (lb ) or only sim ple reflexives (Ic). The com plex form sig selv ‘REFL s e lf is used w ith anti-reflexive predicates, i.e. predicates whose m eaning imply non-reflexive scenarios, see (lb ). The simple reflexive sig ‘R E FL ’ is the default form used both with predicates whose m eaning does not imply anti-reflexivity, e.g. the so-called “neutral” predicate vaske ‘w ash’ in (la ), and with inherently reflexive predicates whose m eaning does imply reflexivity, see (Ic). The sim ple/com plex distinction also interacts with the interpretation o f the referent o f the reflexive (e.g. in so-called doppelganger-effect exam ples as well as w ith respect to the choice between sloppy and strict readings in VP ellipsis constructions) and prosodic factors, both o f w hich will be discussed in greater length below. In D anish the binding dom ain for reflexives is the tensed clause, indicated with square brackets in the follow ing exam ples. Local reflexives o f the types illustrated in (1) will be discussed in sections 3.2 and 3.3. The so-called long-distance reflexives, which in Danish are lim ited to the tensed clause, see (2a) vs. (2b), will be described in section 3.3.1 w here we also discuss how intensifier-adjunction interacts with locality constraints. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 117 (2) a. [Peter, bad m ig barbere sigi], Peter ask-PA ST me shave-INF REFL ‘Peter asked me to shave him .’ b. *[Peteri sagde [at je g barberede 5/g-,]]. Peter; say-PA ST that I shave-PA ST REFL; Unlike English w here certain logophoric or contrastive uses o f what appear to be the reflexive him self can be bound by antecedents outside the tensed clause, see ( 3 a /, Danish reflexives have to be bound w ithin this domain, (2a) vs. (2b). That is, in Danish, only pronom inals, be they sim ple, e.g. ham ‘him ’, or com plex, ham selv ‘him h im self, may have antecedents outside the tensed clauses, see (3b) vs. (2b). (3) a. [Peteri said [that Mary danced with all others than himself]]. b. [Peteri sagde [at Marie dansede med alle andre end ham/ selv]]. Peter say-PA ST that M arie dance-PA ST w ith all others than him self ‘Peter said that M ary danced with all others than h im se lf’ Unlike him self m English, D anish reflexives are subject-oriented, i.e. only allow binding by antecedents functioning as subjects, see (4) vs. (5). (4) IdOi told Mary,, about herself .^ . (5)^ a. Ida,fortalte Marie^ om sigi/*^ selv. Ida told M arie about REFL self ‘Ida told M arie about h im se lf’ b. Ida.fortalte Marie^ om hende*i,^ selv. Ida told M arie about her self ‘Ida told M arie about herself (<her herself).’ If will be argued thaf subject-orienfedness is a characteristic o f ‘tru e’ reflexives'* and that all non-subject-oriented reflexives are really intensifiers or some kind o f intensified nominal ^ See chapter 5 for an analysis of locally free him self as adnominally intensified pronominals. As discussed in section 3.3.6 the simple reflexive sig cannot be stressed. The preposition om ‘about’ as used in example (5) is prosodically too light to host the clitic sig. Hence, only the complex form of the reflexive, i.e. sig selv which can be stressed, is found in such examples. So, the ungrammaticality of (1) below is not due to any anti locality of the simple reflexive sig (as is sometimes suggested, e.g. Vikner (1985)) but simply to phonological factors, see section 3.3.6 for more discussion o f the interaction o f stressability and adnominal intensification of reflexives. (i) * Peter, fortalte Marie^ om sig,. Peter told Marie about REFL self ‘Peter told Marie about himself.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 18 expressions. In fact, as will be shown in chapter 5, it is possible to claim that herself m the sentence in (4) is not a reflexive but rather a concealed intensified pronoun derived from the full form her her se lf some sort o f deletion rule. Danish distinguishes between reflexives and pronouns even for possessives, e.g. sin ‘PO SSR EFL’ (6a) and hans ‘his’ (6b), whereas English only has one possessive, nam ely the possessive pronoun his. (6) a. JohUi sagde at Peterk bad Hans,„ vaske sin*i,k m ,* z bil. John said that Peter asked Hans wash POSSREFL car ‘John said that Peter asked Hans to w ash his car.’ b. JohUi sagde at Peter^ bad Hans,„ vaske hanSi k * ,„ z bil. John said that Peter asked Hans wash his car ‘John said that Peter asked Hans to wash his car.’ One o f the greatest advantages o f assum ing binding and intensification to constitute independent m odules o f the gram m ar is that it becom es possible to unify the account o f argum ent and possessive reflexives and pronom inals in Danish. By considering intensifier- adjunction to reflexives and pronouns to be controlled by binding independent principles (i.e. the sem antic/pragm atic principles controlling adnom inal intensification) it becomes possible to defend a nom inal approach to binding in w hich the distribution o f reflexives and pronouns follow syntactic principles: the reflexives sig ‘REFL’ and sin ‘PO SSR EFL’ having to be bound by a subject (local or long-distance) inside the tensed clause (=principle A), and the pronom inals ham ‘him ’ and hans ‘h is’ having to be free from binding from a subject inside the same domain. The com plex form s sig selv ‘REFL s e lf Am egen ‘POSSREFL ow n’ and ham selv ‘him self//za«x egen ‘his ow n’ are adnom inally intensified versions o f their simple counterparts. All instances o f the adnom inal intensifier selv/egen ‘self/ow n’ - even the ones occurring in the so-called com plex reflexives and pronouns - are thus controlled by the binding-independent principles o f the m odule o f adnom inal intensification. As already shown Cf. Jakubowicz 1994:206, (1). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 19 in chapter 2, these principles are susceptible to sem antic and pragm atic factors as well as to the focus-structure o f the surrounding linguistic context, i.e. the sentence or the larger discourse. The rest o f the chapter is organized as follows. In section 3.2 the foundation for an analysis o f reflexives in D anish is outlined. Then, in section 3.3 this analysis will be used to account for the behavior o f reflexives in a num ber o f different contexts. Section 3.4 is dedicated to the treatm ent o f com plex and sim ple pronouns. Finally section 3.5 sum m arizes the results o f the previous sections and concludes the chapter. 3.2 Basic properties of Danish reflexives Before we start discussing how predicate m eaning and focus affect the distribution o f the adnom inal intensifier in com plex reflexives and the independence o f binding and intensification in general, we first need to describe the formal characteristics o f reflexives and lay out the relevant assum ptions concerning their syntactic and sem antic properties. Section 3.2.1 contains a description o f the m orphology o f reflexives in Danish. Section 3.2.2 and 3.2.3 deals w ith their sem antic and syntactic properties respectively and in section 3.2.4 an analysis o f reflexives in Danish based on a m odified version o f principle A is proposed. 3.2.1 Morphological characteristics of reflexives in Danish As illustrated in Tables 1-2 Danish has two series o f reflexives and pronouns: sim ple forms (see Table 1) and com plex form s (see Table 2). W e assum e the com plex reflexives to be form ed by adnom inal intensification o f the sim ple reflexives. That is, intensified reflexives are form ed in exactly the same way as both intensified DPs (e.g. Kongen selv kom til model ‘The king him self attended the m eeting’) and com plex subject and object pronouns pronouns (e.g. Peter sagde at Marie dansede med all andre end ham selv ‘Peter said that M ary danced R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 120 with everyone but h im self), nam ely by intensifier-adjunction to the sim ple/unintensified forms. In tables 1-2 colum n 1 gives the m eaning o f the form s in colum ns 2-4. Colum ns 2-3 illustrate the sim ple and com plex form s o f subject and object pronouns and colum n 4 gives the corresponding sim ple and com plex reflexive pronouns. N ote that only the third person form s m aintain a formal distinction betw een object pronouns and reflexives, i.e. sig ‘R EFL’ vs. ham ‘him ’. For all first and third person forms there is no formal distinction between reflexives and object pronouns, e.g. mig ‘m y se lf and mig ‘m e’. Furtherm ore while the third person object pronouns, e.g. ham/hende ‘him /her’, etc., are m arked for both case, num ber, and gender, the third person reflexive is invariably sig w hich is unm arked for all these features. In these respects the Danish pronom inal system is very sim ilar to the French and Germ an systems. Table 1. Unintensified pronouns and reflexives. N om inative Acc./Dat. Reflexive f* ,s in g . Jeg M ig mig 2’ “', sing - informal Du D ig dig 2"^ sing. - formal De Dem Dem 3'''’ , sing. - m asculine Han Ham sig sing. - fem inine Hun hende sig 3^“, sing. - com m on gender Den Den sig 3'''', sing. - neuter Det Det sig V \ plur. Vi Os os 2”", plur. - informal I Jer jer 2"“, plur. - formal De Dem Dem 3'", plur. De dem sig R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 121 Table 2. Intensified pronouns and reflexives. N om inative Acc./Dat. Reflexive r* ,s in g . je g selv M ig selv mig selv 2"'*, sing - informal du selv Dig selv dig selv 2"^', sing. - formal De selv Dem selv Dem selv 3'‘ ‘ ‘ , sing. - m asculine han selv ham selv sig selv 3^“, sing. - fem inine hun selv hende selv sigselv 3'“, sing. - com m on gender den selv den selv sigselv 3'''', sing. - neuter ?det selv ?det selv sigselv V \ plur. vi selv os selv os selv 2"“, plur. - informal I selv Jer selv je r selv 2"", plur. - formal De selv Dem selv Dem selv 3''“, plur. de selv dem selv sigselv Com plex and sim ple reflexives differ in several ways, i.e. phonologicaily, syntactically, sem antically and pragm atically. The proposal defended here is that these differences can be explained as consequences o f adnom inal intensification. But before we go into the detailed analysis we first need to spell-out the set o f assum ptions on which our analysis o f binding is based. 3.2.2 Semantics of reflexives Unlike pronouns, w hich are specified for person, num ber, and gender, Danish reflexives are only overtly specified for person^ Sem antically the reflexive sig can be thus characterized as Since Danish, unlike Chinese and other languages, does not allow subject reflexives, the only potential case marking o f reflexives would be an ACCUSATIVE vs. DATIVE/OBLIQUE distinction. However, as the examples In (i-11) show, no such distinction is overtly marked on reflexives or pronouns in Danish: R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 122 an expression that has an incom plete set o f phi-features. That is, we assum e sig to function sem antically as a variable that is dependent on its antecedent for lexical content and reference. In contrast, that pronouns (e.g. ham ‘him ’, hende ‘her’, den ‘them ’) have phi-features is obvious from the fact that they are overtly m arked for features like person, num ber and gender and that they can be used deictically, i.e. have lexical content and reference on their own. The assum ption that the simple reflexive sig ‘REFL’ is a featureless variable is supported by the fact that it is the only elem ent that can be bound by an impersonal DP. In certain sentences, a pronoun can be used in the same configurations in w hich sig can be long distance bound, see (7). Yet only the reflexive sig ‘R E FL ’, but not the pronoun ham ‘him ’, can occur if the m atrix antecedent is an im personal DP, such as arbitrary PRO or an indefinite expressions, e.g. man ‘one’ or enhver ‘everybody’, see (8)'’ (7) a. Peteri lader fo lk tale om hamj fsig,. Peter let-PRES people talk-IN F about him /REFL ‘Peter lets people talk about him .’ b. Peter, bad mig om at invitere sig/hamj.,,. Peter ask-PA ST me about to invite-INF REFL/him ‘Peter asked me to invite him .’ (i) a. Peter vasker sig. Peter washes REFL-ACC ‘Peter washes (himself).’ b. Peter kabte sig et hus. Peter bought REFL-DAT a house-ACC ‘Peter bough himself a house.’ (ii) a. Peter vastier ham. Peter washes him-ACC ‘Peter washes himself.’ c. Peter kabte him et hus. Peter bought him-DAT a house-ACC ‘Peter bough him a house.’ Since it appears to be orthogonal to the issues discussed in this dissertation, case marking of reflexives and pronouns will not be discussed in great detail. This observation, as well as the examples in (7-8a,b), is adapted from Jakubowicz (1994:133), ’ A similar situation holds in French. In certain contexts, e.g. non-contrastive PPs, the distribution of the reflexive soi ‘REFL’ and the pronominal lui/elle ‘him/her’ overlap, see (i). However, when the antecedent is indefinite then only soi ‘REEF’ can be used, see (ii). (i) Pierre, est fier de soi/lui,. Peter is proud o f REFL/him ‘Peter is proud of himself.’ (11) Personne n ’ est fier de soi,/*lui,. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 (8) a. PRO I at lade fo lk tale om sig/ham*i,^ er kedeligt. PRO to let people talk about REFL/him be-PRES boring ‘To let people talk about one/him , is boring.’ b. Man/Enhveri bad mig om at invitere sig/ham*, ^ . one/everybody ask-PA ST me about to invite-INF REFL/him ‘O ne/Everybody asked me to invite him .’ The exam ples in (9) illustrate that intensifier-adj unction does not change the ability o f the sig to be bound by an im personal DP. This should com e as no surprise since the intensifier selv ‘s e lf is also a featureless, m orphologically invariable particle. (9) a. Enhver/maUi skal forsvare sigi (selv) modpolitiet. everyone/one m ust defend REFL against police-the ‘Everyone/one m ust defend him self/oneself against the police.’ b. PROj Atprcesentere sigi (selv) fo r studenterne er kedeligt. PRO to introduce REFL for students-the is boring ‘To introduce oneself to the students is boring.’ A ccording to proposals by Burzio (1989, 1991), only so-called ‘true reflexives’ can have im personal antecedents. In this respect both sig and sig selv qualify as true reflexives*. W hile sig lacks overt m orphological m arking o f num ber, gender and case, it can still be said to be m arked for person since it differs m orphologically from f and 2"‘ ‘ person forms, see Table 1 in section 3.2.1. Burzio (1991), however, argues that the person agreem ent o f sig w ith a definite DP acting as its antecedent is m erely a case o f “pseudo-agreem ent” ’ and that, consequently, true reflexives lack even the person feature. Indeed, as observed by E. Benveniste, in m any languages the third person, is the default category, best characterized negatively as the absence o f first and second person features, rather than as the presence o f a nobody NEG is proud of REFL/him ‘Nobody is proud of himself.’ * Cf. also Jakubowicz (I994:116(la,b)). ’ Cf. Burzio (1991,p. 14): “alpha pseudo-agrees with beta if: (i) beta has no gender, no number, and no person, and, (ii) alpha is third person. Burzio further argues that “person markings have a higher relative weight than either gender or number markings since a featureless beta pseudo-agrees with and alpha o f any number and any gender, but not with one which is P* or 2"'* person” (ibid.). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 special third person feature. True reflexives can thus be defined as genderless, numberless, and personless, i.e. as lacking phi-features. We adopt B urzio’s proposal and assum e that both simple sig and com plex sig selv classify as true reflexives'” in this sense. In the rest o f the dissertation the term ‘true reflexives’ will be used to refer to nominal expressions which are: (i) featureless (i.e. lack phi-features) and, consequently, bindable by impersonal indefinites, and (ii) subject-oriented (i.e. cannot be bound by non-subject antecedents). As m entioned above, being itself an uninflected particle, the intensifier selv does not add any phi-features to the com plex reflexive sig selv. W hile sig and sig selv are sem antically the sam e in so far as they are both featureless reflexives, they clearly differ sem antically in other respects, i.e. w ith respect to contrastive focus, doppelganger-effects, sloppy vs. strict readings in VP ellipsis constructions, etc. We argue that all these differences between simple .s/gand com plex sig selv follow from sem antic properties o f the intensifier selv ‘s e lf. That is, simple sig and com plex sig selv have the sam e properties with respect to the binding theory and w hatever properties sig selv have in addition to sim ple sig derive from the adjunction o f the intensifier selv ‘s e lf. As will be shown in the follow ing section, even the different m orphological and syntactic behavior o f sig and sig selv follow from the intensifier-adj unction o f selv to sig which transform s the phonological clitic into a full DP. 3.2.3 Syntactic structure of reflexives Simple and com plex reflexives display quite different syntactic behaviors. W hile the simple reflexives, e.g. sig. display clitic-like behavior, the com plex forms o f the reflexives, e.g. .sig selv, qualify as full DP phrases. Kayne (1975) found that the French reflexive clitic se behaves differently from regular DPs with respeet to a num ber o f different syntactic processes. Applied to com plex and simple reflexives in Danish, these tests show that sig behaves ' In tills respect Danish sig/sig selv behaves like French se/soi/soi-meme and Dutch zich/zich zelf. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 125 sim ilarly to French se in a num ber o f respects while the syntactic behavior o f the com plex reflexive sig selv patterns with that o f full lexical DPs. That is, the com plex reflexive sig selv can be stressed, used as answ er to a question, coordinated, clefted and topicalized. The simple reflexive sig, on the other hand, cannot undergo any o f these processes, see (1 Oa-e). (10) se/sig DP/sig selv/ham/himself/lui-meme a. stress * OK b. answ er to questions * OK c. coordination * OK d. clefting * OK e. topicalization * OK The exam ples illustrating the properties listed in (lO a-e) are given in section 3.3.6, and have therefore not been repeated here. Though Danish sig and French se are both phi-feature-less clitics they do, nevertheless, display different behaviors in other respects. For exam ples, while se has to attach to its host verb, sig does not. The exam ple in (11) illustrates the fact that French se cannot be separated from the verb. (11) Pourquoi se rase-t-il? [C P pourquoi [C ’ S6i rasej [TP il, t/.... ]]] why REFL shave-PRES he ‘W hy does he shave (him self)?’ In the question in (11) the clitic se attaches to the verb and moves along w ith it when it is m oved to C. W hile French se necessarily attaches to its host predicate, this is not the case for D anish sig, w hich can be separated from its verb by other material. Being a verb-second (V2) language, Danish requires the verb to raise to C in m ain clauses". The exam ples in (12a,b) show that in such cases the clitic sig rem ains behind thereby becom ing separated from the verb b y the subject. " The literature on Mainland Seandinavian languages contains a variety of proposals as to the analyze V2 sentences. Since this issue does not bear directly on the topic under discussion here, it will not be discussed in any detail. The bracketed sentences in (12) are slightly adapted versions of Holmberg and Platzack’s (1989, 1995) analysis of syntax of mainland Scandinavian V2 phenomena. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 1 2 6 (12) a. Hvorfor barberer han sig? [CP hvorfor [C ’ barbererj [TP hani tj sigi ]]]? why shaves he REFL ‘W hy does he shave (him self)?’ b. Han barberer sig. [CP hani [C ’ barbererj [TP ti tj sig, ]]] he shaves RELF ‘Fie shaves (him self).’ The exam ples in (13) and (14)‘^ illustrate another difference between se and sig. The clitic se cannot be om itted in the second o f tw o coordinated verbs w ith or w ithout auxiliary verb, see (13a,b)'^. Only when the auxiliary verb is dropped too is this possible, see (13c). Danish sig shows no such constraints and can be freely om itted from all but the last o f the coordinated verbs, see (14a,b)''^. (13) a. Avant de sortir, Marie s ’ habille et *(se) peigne soigneusement. before to go out, M arie REFL dresses and (REFL) com b m eticulously ‘Before going out, M arie dresses and com bs herself m eticulously.’ b. Marie s ’ est habillee et *(s ’ )est peignee soigneusement. M arie REFL is dressed and (REFL) is com bed m eticulously ‘M arie dressed and com bed herself m eticulously.’ c. Marie s ’ est habillee et (s ’ est) peignee soigneusement. M arie REFL is dressed (REFL is) com bed meticulously ‘M arie dressed and com bed herself m eticulously.’ (14) a. Marie klceder (sig) og reder sig omhyggeligt. M arie dresses (REFL) and com bs REFL carefully ‘M arie dresses and com bs herself carefully.’ b. Peter barberede (sig), vaskede (sig) og torrede sig. Peter shaved w ashed and dried REFL ‘Peter shaved, washed and dried (him self).’ We take the differences betw een se and sig illustrated in (11-14) to m ean that while the French reflexive clitic se forms a m orphological constituent w ith its host sig does not. Sig is, however, a phonologicaily dependent form that is prosodically too light to stand on its own The examples are based on similar examples in Jakubowicz (1994:209, ex. (12-3)). The observation thatse cannot be dropped in such coordinated constructions is from Kayne (1975). Note, however, that such deletion of all but the last instance of sig is not possible with inherently reflexive verbs, see more detailed discussion such cases in chapter 4, section 4.2.8. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 127 and therefore needs to attach to a host constituent (noun or verb or any other category) with which it then forms a phonological constituent. In D an ish , b o th sim p le p ro n o u n s, e.g. ham ‘h im ’, and sim p le reflex iv e, e.g. sig ‘R E F L ’, d iffe r fro m full lex ical D P s in th a t th e y can u n d erg o o b je c t sh ift to a p o sitio n o u tsid e th e V P and attach to an a p p ro p ria te p h o n o lo g ic a ily h ea v y h o st, see (1 5 a )'^ w h ere sig an d ham p reced e th e n eg atio n ikke ‘n o t’. B u t u n lik e sig, th e sim p le p ro n o u n ham ‘h im ’ is stre ssa b le an d can th e re fo re b e left in situ if stressed , see (1 5 b ), w h ich sh o w s th a t th e stressab le sim p le p ro n o u n ham (an d th e stressed in ten sified reflex iv e sig selv) b u t n o t th e u n stressab le sig can fo llo w th e n eg atio n ikke ‘n o t’. (15) a. PeteriVaskede sigi/ *sigi selv / *H AM J h a m j *bilen ikke.(+oh]QCis\\\f() Peter washed REFL/*REFL self/him /*car-the not Peter did not wash (him self)/him /the car.’ b. Peteri vaskede ikke ’ ’ ^sigif sigi selv / H A M J "^hamj (-object shift) Peter w ashed not *REFL/REFL self/him /car-the Peter did not wash him self/him /the car.’ Sum m arizing the above, we follow H alpem (1992) and Jakubow icz’s (1994) in concluding “that Danish clitics are syntactically independent prosodically bound words, w hereas in m odern French, se as well as le are clitics that select for m orphological attachm ent. Thus sig and ham form only a prosodic constituent with their host and not a m orphological one”(Jakubow icz 1994, p. 118-9). In brief, a crucial difference betw een ham ‘him ’ and sig ‘REFL’ illustrated in (15) boils dow n to stressability; while sig can never receive stress, see the tests in (10), ham ‘him ’ can, see (15b). W hen unstressed, both sig and ham are sy n tactically in d ep en dent but p rosod ically bound w ords. S in ce the n egation ikke ‘n o t’ cannot felicitously receive stress it is not a viable host for prosodically dependent clitics. In contrast, the proper nam e Ida in (15a) can be stressed and is thus able to host clitics. The sim ple ’ Examples as well as discussion thereof are from Jakubowicz (1994:218, i-ii). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 128 pronoun ham can still be m ade acceptable in (15b) by attracting sentence stress onto itself, although thereby changing the m eaning contribution o f the sentence by adding the corresponding focus-generated im plicatures (contrast, contextually defined scales o f prom inence/rem arkability, etc.). Since the sim ple reflexive sig on its own cannot be stressed under any circum stances tbe only way to save the sentence is to enable sig to receive sentence stress by adnom inally intensifying it, see (15b). Based on the differences between sig and sig selv described above it is clear that these expressions m ust have different syntactic representations. A dopting Longobardi’s analysis o f nominal expressions (which goes back to A bney (1987)) we assum e that all nominal argum ents are projections o f a head D constituent. As argued in chapter 2, section 2.2.2.3, this approach makes it possible to arrive at an elegantly unified account o f the syntax o f nominal expressions in argum ent positions, see (16). (16) a. [ d p [ n kong ] [ d -en] ] ‘the king’ b. [dp [d ham] ] ‘him ’ c. [dp [d sig] ] ‘R EFL’ A s shown in (16) all nominal argum ents are assum ed to be projections o f a head determ iner. Pronouns and reflexives are assum ed to behave as determ iners, i.e. they are assum ed to head their own DPs, see (16b) and (16c). Besides m aking it possible to arrive at a unified analysis o f the syntax o f nom inal expressions in this approach has the additional advantage o f enabling us to unify the account o f adnom inal intensification. As discussed in chapter 2 intensified nom inals such as kongen selv ‘the king h im se lf are best analyzed as sim ple adjunction structures, see (17a). (17) a. [dp [dp [ n kong ] [d -en]] [selv]] ‘the king h im self b. [dp [dp [d ham ]] [selv ]] ‘him s e lf c. [dp [dp [o ^ 'g ]] [selv ]] ‘REFL s e lf R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 129 A s shown in (17b) and (17c) the com plex pronoun ham selv ‘him s e lf and the com plex reflexive sig selv ‘REFL s e lf can be analyzed as intensified DPs having the same syntax as intensified R-expressions. In other words, syntactically all types o f nom inal expressions behave alike w ith respect to intensification. The fact that both sim ple pronouns and sim ple reflexives can undergo object shift follow from their am biguous X 7X P status. In contrast, full lexical DPs as well as intensified reflexives and pronouns, neither o f w hich can undergo object-shift, are unam biguously XPs. The differences between sig and ham illustrated in (lOa-e), follows from stressability rather than syntax: ham can be stressed, while simple sig cannot"’. 3.2.4 Consequences of relegating binding and intensiilcation to different modules: a syntactic approach to binding of reflexives: principle A The central claim defended here is that simple sig and com plex sig selv have the same properties w ith respect to the binding theory and that w hatever properties sig selv has in addition to sim ple sig derive from the intensifier selv ‘s e lf. One often m entioned difference betw een sim ple and com plex reflexives, w hich appear to contradict this claim, is their seem ingly different behavior w ith respect to long-distance binding, see (18) which illustrate that in an out-of-the-blue context the simple reflexive sig can be long-distance bound while the com plex reflexive sig selv cannot. (18) Peter, bad J0rgeri2 barbere sig,,^ / sig*,/^ selv. Peter ask-PA ST Jorgen shave-INF REFL ‘Peter asked Jorgen to shave h im /h im self’ Since Faltz (1975) there has been a tendency to link the m orphological com plexity o f reflexive elem ents to locality. For D anish and sim ilar languages (e.g. the other m ainland Scandinavian languages, Dutch, etc.) this lead to the proposal that the sim ple reflexive sig is anti-local (i.e. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 130 specialized for LD -binding) w hile the com plex reflexive sig selv has to be locally bound (cf. V ikner 1985, am ong others). As will be shown below, both o f these assum ptions are wrong. The sim ple reflexive sig can indeed be a locally-bound, theta-role receiving argum ent anaphor (as shown by the exam ple in (18)) and, given the right circum stances, com plex reflexives may be LD-bound. N otice that this does not am ount to the claim that sig and sig selv have identical behaviors. On the contrary, it is quite obvious that sig and sig selv do behave differently in many respects. But rather than trying to account for these differences w ithin binding theory, we divide the explanatory burden betw een two m odules, i.e. binding and intensification. That is, we argue that any differences between sig and sig selv should be accounted for not by reference to binding principles but by reference to factors related to adnom inal intensification, i.e. focus, prosodic differences, doppelganger-effects, etc. For expository reasons, we will lim it the follow ing description o f binding properties o f Danish reflexives to the behavior o f simple sig, - the assum ption being that the any differences between sig and sig selv follow either directly from the binding-independent principles o f the m odule o f intensification or are phonological, syntactic or sem antic side-effects o f intensifier-adjunction. Before jum ping ahead to the analysis o f com plex reflexives in Danish, let us first consider the relevant descriptive generalizations concerning the binding o f the sim ple reflexive sig. Consider the sentences in (19) which illustrate the locality constraints restricting the distribution o f sig. (19) a. [PeterI barberede sigi * , Peter shave-PA ST REFL ‘ Peter shaved (himself). ’ ’ See also section 3.3.6 where stressability will be discussed in more detail. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 131 b'^. [Peteri badmig^ barbere sigi/*^]- Peter ask-PA ST me shave-INF REFL ‘Peter asked me to shave him .’ c. [Peteri bad Jargen^ barbere J . Peter ask-PAST Jorgen shave-INF REFL ‘Peter asked Jorgen to shave him /him self.’ d. *[Peter, sagde [at je g barberede Peter; say-PA ST that I shave-PA ST REFL; The sentence in (19a) illustrates that while sig may be locally bound'*, it cannot have a sentence-external antecedent or be used deictically. The sentences in (19b,c) illustrate that sig need not be bound by the m ost local sentence-internal subject, but may be long-distance bound by a higher subject provided that it (pseudo-)agrees in person. Finally the sentences in (19d) vs. (19c) illustrate the locality constraint on long-distance binding: long-distance binding o f sig is only ok as long as the antecedent is still located inside the m inimal tensed clause containing sig. Though languages w hich allow long-distance anaphora seem to differ widely with respect to the type o f com plem ent out o f w hich binding o f reflexives can occur, typological studies claim to have discovered the im plicational universal in (20) (adapted from H uang (2000:92-3), cf. also Burzio 1996, 1998). (20) A n im plicational universal for long-distance anaphora com plem ent tvpes: N Ps > small clauses > infm itivals > subjunctives > indicatives W hat (20) says is that if a language allows LD -binding o f reflexives into a certain type o f com plem ent then it will also allow LD -binding o f reflexive into all the types o f com plem ents lower on the hierarchy. That is, if a language allows LD -binding into indicative com plem ent ‘^Note that unlike Chinese ziji g iL ‘self-self, Danish reflexives are not subject to blocking by intervening L‘ or 2"‘ * person pronouns. This difference may be due to the fact that sig is a reflexive anaphor while ziji is an adnominal intensifier. Exploring this idea is, however, beyond the scope of this dissertation and will be left for future research. A number o f analyses of Danish and similar languages (e.g. Vikner (1985) etc.) deny that such examples show that sig can be locally bound. Instead, they claim that instances o f local sig are not true reflexive anaphors but R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 132 clauses then it will also allow binding o f reflexives into all the other types o f com plem ents. Danish, w hich allows LD -binding out o f infinitivals thus also have LD -binding out o f small clauses (and resultatives) and N Ps as discussed in section 3.3. In contrast, Italian is reported to allow LD binding out o f subjunctives and (cf. Huang (2000:93)) and Old Icelandic is claim ed to be o f the m ost inclusive type, allow ing LD -binding even out o f indicative clauses (cf. Sigurdsson 1990a, quoted in H uang (2000:93)). M any attem pts have been made to give a unified account o f the various types o f LD -binding found in the worlds languages. W hile the reduction o f such cross-linguistic variation to unified account based on a small set o f general principles is, o f course, the ultim ate goal o f linguistic theory it rem ains beyond the scope o f the present dissertation. The rest o f this chapter will therefore be lim ited to binding o f sig in Danish. That is, our aim will be to explain the follow ing descriptive generalization concerning dom ain restrictions, see (21). (21) Descriptive generalization 1: sig m ust be bound inside the m inimal tensed clause''’. N ow let us turn to another property o f D anish reflexives, viz. subject orientation. U nlike Eng. him self v/hich appears to allow binding by non-subject antecedents^”, see (22), Dan. sig can only be bound by subject antecedents, see (23). (22) Peter told Mary/ about herself. rather non-thematic grammatical markers forming part of a complex predicate and that thematic sig is specialized for long-distance binding. See chapter 4 for more discussion and criticism o f such analyses. '“ ’ The generalization in (21) is based on examples with infinitival clauses functioning as complements of verbs denoting acts of communication (e.g. saying, uttering, etc.). Theoretically it should be possible to find examples of LD-binding o t sig out of non-complement infinitival clauses. The sentence in (i) exhibits LD-binding of sig out of an adverbial clause: (1) Den asketiske munkflyttede hen til Ganges floden fo r at kunne vaske sig in helligt vand hver morgen. the ascetic monk moved over to Ganges river-the for to be-able-to wash REFL in holy water eveiy morning ‘The ascetic monk moved to the River Ganges to be able to wash himself in holy water every morning.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 133 (23) Peter ifortalte Marie^ om sigi*^ selv. Peter told M arie about REFL self ‘Peter told M arie about him self.’ This leads us to the m ore precise descriptive generalization in (24). (24) Descriptive generalization 1 (revised version): sig m ust be bound by a subject inside the m inim al tensed clause. C hom sky’s (1981, 1986) binding theory, see (25)-(27), has met with a great deal o f criticism over the years. It has been claim ed that a num ber o f cross-linguistically w ide-spread phenom ena cannot be captured properly by these principles, e.g. logophors, overlap in distribution o f reflexives and pronouns in certain contexts, e.g. possessive constructions, picture-NP, etc. In this dissertation, however, it will be argued that the original insights o f C hom sky’s syntactic approach to binding are basically sound. That is, the distribution o f anaphors and pronouns is controlled by syntactic dom ain constraints like principles A and B, see (25)-(28). (25) a. An anaphor is bound in a local domain, b. A pronominal is free in a local domain. (26) alpha binds beta if and only if (i) alpha is in an A-position, (ii) alpha c-com m ands beta, and (iii) alpha and beta are co-indexed. (27) alpha c-com m ands beta if and only if (i) alpha does not dom inate beta, (ii) beta does not dom inate alpha and (iii) the first branching node dom inating alpha also dom inates beta. See chapter 5 where it is argued that English self-fovms locally bound by non-subject antecedents (as in (22) above) are really reduced forms o f intensified pronouns, e.g. herself in (22) is assumed to derive from the adnominally intensified pronominal her herself. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 134 (28) Binding Domain^' alpha is a binding dom ain for beta if and only if alpha is the m inimal eategory (i.e. the sm allest DP or IP/S) containing beta, a case-licensor or beta, and a SUBJECT aceessible to beta. A s discussed in the following, apparent exceptions to these principles should be explained either by reference to param etric variation (e.g. with respect to what counts as binding dom ain in a given language or w ith respect to m orphological properties o f reflexives (i.e. their status as either affixes, clitics, free forms, etc.)) or by reference to binding-independent factors (e.g. sem antic/pragm atic factors influencing adnom inal intensification o f reflexives, prosodic factors (e.g. stressability), etc.) rather than by radically changing the architecture o f binding theory. As the descriptive generalization in (24) states, in Danish the local dom ain is the tensed clause. This is illustrated by the sentence in (29) w hich shows that any subject w ithin the m inimal tensed clause is a potential antecedent for sig. (29) Peteri horte Marie^ bede sygeplejersken,, vaske sig; ^ Peter heard M ary ask-lN F nurse-the w ash-lN F REFF ‘Peter heard M arie ask the nurse to wash R E FL .’ That is, m ore specifically, in (29) the antecedent o f sig can be either the closest subject sygeplejersken ‘the nurse’, the interm ediate subject Marie or the m atrix subject Peter. Thus the tw o facts about sig w hich any theory m ust be able to explain is why it m ust be bound w ithin the minimal tensed clause and why the binder m ust be a subject. In the follow ing a m odified version o f Pica’s (1984,86) LF m ovem ent approach to LD-binding o f reflexives will be The “ b in d in g dom ain” referred to in (28) corresponds w hat in o ld er versions o f the theory used to he called the Governing Category, usually defined as in (i). (i) Governing Category (GC) Alpha is a GC for beta if and only if alpha is the minimal category (i.e. the smallest DP or IP/S) containing beta, a governor of beta, and a SUBJECT accessible to beta. Cf. also the older version of Governing Category from Chomsky 1986, p. 169, given in (ii) below. (ii) Governing Category (GC) (Chomsky (1986) “A governing category of alpha is a maximal projection containing both a subject and a lexical category governing alpha (hence containing alpha). A governing category is a ‘complete functional complex’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 135 outlined. The advantages o f Pica’s LF m ovem ent analysis o f LD -binding have been argued to be that it provides a unified analysis o f three properties considered to be com m on to all LD- anaphora listed here in (30a-c). (30) a. LD -binding b. subject-orientation c. m orphological simplicity Successive cyclic m ovem ent from low er to higher INFLs/Ts via C O M P allow one to consider LD binding o f sig as obeying the locality requirem ent o f principle A. As shown in section 3.2.2, sig lacks phi-features. We therefore assum e that it has to adjoin to INFL/T in order to obtain features from the subject via spec-head agreem ent. This explains why sig is always subject-oriented (cf. Jakubow icz 1994:119). Though both Danish sig ‘R EFL’ and its French counterpart se ‘R EFL’ are clitics base-generated in argum ent position, they still differ in im portant ways as described in section 3.2.3. Unlike Rom ance clitics, e.g. French se, w hich cannot be separated from their host, D anish sig can be separated from the verb by other m aterial, see exam ples (11) and (12) in section 3.2.3. That is, French se m oves along with its host verb when it m oves overtly to adjoin to IN FL/T (cf. Pollock 1989), see (11). In contrast, D anish sig is a syntactically independent prosodically bound word which does not from a m orphological constituent with its host^^ see (12). Thus when sig m oves from its based generated position to a higher INFL/T this has to be an instance o f covert m ovem ent at LF. Long-distance binding o f sig can thus be explained as a resulting from cyclic m ovem ent o f sig from the low er INFL/T to a higher INFL/T (cf. Jakubow icz 1994:126, Pica 1987, Cole, Hermon, and Sung 1990, etc.). (CFC) in the sense that all grammatical functions compatible with its head are realized in it” (Chomsky 1986, p. 169). In contrast to Jakubowicz (1994:123), who assumes that certain Danish verbs (i.e. [+alTectedness] verbs) “can incorporate the element projected in their internal argument position”, we do not assume incorporation of sig into the verb in cases o f locally bound sig (e.g. Peter forsvarer sig ‘Peter defends himself). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 136 The claim that only m orphologically sim ple reflexives can be LD-bound stems from the observation by Faltz (1975) that, cross-linguistically, sim plex reflexives tend to allow LD- binding w hile m orphologically com plex reflexives m ust be locally bound. Since, in the LF m ovem ent analysis, only head elem ents can undergo head-to-head m ovem ent, that allegedly explains why, as it is claimed, only m orphologically sim ple reflexive elem ents can be LD- bound. W hile we agree that Pica’s LF m ovem ent analysis provides a satisfactory account o f both LD -binding and subjection orientation (30a-b), we disagree w ith its account o f the difference betw een sim ple and com plex reflexives (30c). First o f all the descriptive generalization that only sim ple reflexives can be LD -bound while com plex reflexives m ust be locally bound may not be as solid as it seems. The fact that so-called com plex reflexives can be long-distance bound has been brought up in the literature, e.g. the follow ing Germ an and Japanese exam ples o f long-distance binding o f the com plex intensifier-based reflexives sich selbst ‘REFL s e lf, and zibun-zisin ‘self-self discussed in H uang (2000:96, exam ple (2.169c,d)), see (31)-(32). (31) Willi/ dachte, dafi Hans2 mit Fritz3 iiher sich selbst 12gesprochen hat. Willi thought that Hans w ith Fritz about REFL self spoken has ‘Willi thought that Hans has spoken w ith Fritz about h im se lf’ (32) Takasii-wa Hirosi-ga zibun-zisin/-ni kasite kureta kuruma-o kowasite simatte. Takasi-TO P H irosi-SUBJ self-self lend give car-O BJ broken ended up ‘Takasi has broken the car which Hirosi lent se lf-se lf’ Based on such exam ples we conclude that the LF m ovem ent analysis has to be m odified so as to allow for the possibility o f LD -binding o f adnom inally intensified reflexives. N ote that the com plex reflexives in (31) and (32), i.e. sich selbst ‘REFL s e lf, and zibun-zisin ‘self-self, both qualify as adnom inally intensified reflexives in that they are com posed o f the simple reflexives sich ‘REFL’ and zibun ‘SELF’ to which the adnominal intensifiers selbst ‘s e lf and R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 137 zisin ‘s e lf have been adjoined. In this sense they are entirely parallel to the Danish intensified reflexive sig selv ‘REFL s e lf. Since we know that adnom inal intensifiers can be stranded (Q- floated) by m ovem ent o f their associate^^, it is only logical to assum e that the sam e thing may happen in the case o f intensified reflexives. That is, we assum e that, at LF, the X° reflexive sig may m ove covertly to a higher INFL or T ” node on its own leaving the intensifier behind. For expository reasons detailed discussion o f LD -binding o f com plex reflexives will be postponed till section 3.3.1. The rem ainder o f this section will contain a brief outline o f how the version o f the LF m ovem ent analysis defended here accounts for both local and LD -binding o f simple sig. T h e re fle x iv e is assu m e d to h av e in te rp re ta b le b u t n o n -v alu ed featu res. It th u s n eed s to m o v e to g et its featu res v alu ed , see (33). (33) sig [T in terp retab le, n o n -v alu ed ] featu res The structure in (34c) illustrates a sentence containing a locally bound reflexive at spell-out. (34) Local binding o f them atic sis with neutral predicates (em bedded clauses!: a. Array: {at, John, vasker, sig, ikke} b. Sentence: .... at John ikke vasker sig. that John not washes REFL ‘... that John doesn’t wash him self.’ ' See the analysis of q-floated adnominal intensifiers in chapter 2, section 2.2.5. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 138 c. CP / \ C ’ / \ at TP / \ John T ’ / \ r v P / \ N E G vP I / \ ikke v ’ / \ / v ’ / \ V VP / \ V DP i I vaskede sig In (34) the sim ple reflexive sig m erges w ith the neutral verb vaske ‘w ash’ from w hich it receives both case and the internal theta-role. A n em bedded clause was chosen for this exam ple in order to avoid the com plications o f V2 w ord-order and object-shift - both phenom ena which occur only in main clauses. The negation - which is assum ed to adjoin to the left edge o f the predicate (i.e. vP jdoes not play any role here and is ju st added to facilitate com parison with structures discussed in (36). Just like traditional LF-m ovem ent analyses o f LD -binding a la Pica (1984, 1986), we assum e that it is the specific featural properties o f sig which dictate that it m ust move to adjoin to T° at LF in order to get interpreted. That is, sig is assum ed to have [Tinterpretable, -valued] features. In other words, since sig has no lexical content o f its own it has to m ove to enter into a spec-head relation with the subject John in order to be interpreted. The bracketed sentence in (35c) illustrate how the LF-m ovem ent analysis accounts for LD-binding o f sig. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 139 (35) LD -binding o f them atic sis w ith neutral predicates (em bedded clauses'): a. Array: {at, John, bad, Hans, vaske, sig} b. Sentence: ...at John had Hans vaske sig. that John asked Hans w ash REFL ‘ . that John asked Hans to wash him /him self. ’ c. [cp [c’ at [tp Johni [t - [T° sigi\ [vp bad HanSk [cP t," h ^P R O , [t- [ T tj] [vp vaske ]]]]]]]]] d. [(;p [(^’ at {jp John/ [x’ [X ] [yp bad Hansi^ [cp [tp p r o , [t^ [V sig,] [vp vaske L]]]]]]]]] Since sig is a head it can undergo successive cyclic m ovem ent from low er infinite T° to a higher finite T° w here it may receive lexical content from its antecedent through spec-head agreem ent. It m ay also, as shown in (35d), get its features valued by spec-head agreem ent with the subject o f the em bedded clause. In m ain clauses, D anish has V2 word order as well as object shift o f personal pronouns and reflexives. V2 word order entails m ovem ent o f a constituent (often the subject) and the m ain verb to spec-CP and C “ respectively. And object-shift involves the m ovem ent o f pronouns and reflexives from their argum ent positions to spec-vP (cf. Platzack and Holm berg (1995), M cGinnis (1999)). Both processes are illustrated in (36) w hich shows the sentence at spell-out. In order to get interpreted the reflexive sig will still have to m ove to T° at LF as described above. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 140 (36) Local binding o f them atic sie w ith neutral predicates (main clausel: a. Array: {John, vasker, sig, ikke} b. CP / \ John C ’ / \ vasker TP / \ t T / \ t vP / \ sig v ’ / \ N EG v ’ I / \ ikke DP v ’ / \ V VP / \ V DP I I t t So far only exam ples w ith loeal and LD -binding o f the simple reflexive sig in argum ent position o f neutral verbs have been discussed. In the follow ing section the viability o f the analysis proposed here will be scrutinized by a eloser investigation o f how it deals w ith both simple and com plex reflexives in a num ber o f different syntactic environm ents as well as with different types o f predicates. The analysis o f non-them atic sig w hich occurs with certain “deponenfV inherently reflexive verbs (e.g. skamme sig ‘(h f) sham e REFL/be asham ed’, brcende sig ‘(lit.) burn REFL/(accidentally) get burned’ will be discussed in chapter 4, section 4.3. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 141 3.3 Binding of reflexives in different contexts: testing the independence of binding and intensification. In the follow ing sections, the descriptive and explanatory adequacy o f the analysis o f Danish reflexives presented in section 3.2 will be put to the test by taking a closer look at the behavior o f sim ple and com plex reflexives in different syntactic positions. That is, we discuss m orphological com plexity and locality (section 3.3.1), reflexives as direct objects o f different predicates (section 3.3.2), reflexives in resultative constructions (section 3.3.3), reflexives in ECM constructions (section 3.3.4), possessive reflexives (section 3.3.5), as well as reflexives in prepositional predicates and PPs (section 3.3.6). In all cases, it will be shown that the differences betw een sig and sig selv follow from adnom inal intensification rather than binding- related factors. 3.3.1. The independence of intensification and locality constraints. A ccording to accounts o f Danish currently found in the literature (e.g. V ikner (1985), Jakubow icz (1994), etc.), the com plex reflexive sig selv is a subject-oriented reflexive which cannot be long-distance bound, see (37). (37) McArthuri had sygeplejererik give en morfmindsprojtning. M cA rthur asked nurse-the give a m orphine injection ‘M cArthup asked the nurse, to give him self a m orphine injection.’ a. sigik REFL b. sig*ikselv REFL Jakubow icz (1994:130-131) explains the absence o f LD -binding o f the com plex reflexive sig selv as a consequence o f its syntactic structure. U nlike sig, w hich is a head, sig selv is a m axim al projection w hich cannot undergo successive cyclic m ovem ent from a lower to a higher INFL/T via CO M P w ithout violating the ECP. However, since sig selv is also R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 142 featureless it m ust m ove the closest IN FL/T to obtain features w hich explains why, like sig, it can only be bound by a subject antecedent. As m entioned above, sig selv has so far been assum ed to be a local reflexive. However, given the right context it is possible to find exam ples in w hich sig selv may (at least m arginally) felicitously be LD bound, see (38) and (39). (38) Context: M cA rthur is an extrem ely tough general who feels that the lightly w ounded soldiers ought to put up w ith the pain in order to save painkillers for the truly needing. W hen M cA rthur him self got a large piece o f shrapnel in his thigh he stubbornly refused to take any kind o f painkillers. But in the end the pain becam e too m uch for him. So far his principles had dictated him to ask the nurses to give the painkillers to the other soldiers in his ward. Men igar, sent pa natten bad McArthur j migk endeligt give_________en morfinindsprajtning. B ut yesterday, late on night-the asked M cA rthur m e finally g iv e-lN F _________a m orphine injection ‘But yesterday, late at night M cArthur, finally asked me^ to g iv e _________a m orphine injection.’ a. Isigi selv REFLj self b. Isig, REFL; In (38) the local subject mig ‘m e’ is first person and is thus ruled out as a potential antecedent for sig due to lack o f (pseudo-)agreem ent with the reflexive. The only possible antecedent for sig is therefore the m atrix subject McArthur. The elaborate context given in (38) also contributes to m aking the m atrix subject the m ost likely antecedent. Furtherm ore, the context explicitly contrasts a typical situation, in w hich M cA rthur asks the nurse to give one o f the other w ounded soldiers an injection, with the truly exceptional situation in which he asks the nurse to give him self an injection. In other words, the explicitly contrastive context triggers focusing o f the reflexive sig which, due the fact that sim ple sig cannot be stressed, has to be realized as the adnom inally intensified reflexive sig selv. The exam ple in (39) is identical to (38) except for the local subject which has been replaced by the 3"'* person DP sygeplejeren ‘the nurse’ w hich is a potential antecedent for the reflexive since it (pseudo-)agrees with sig. R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 143 (39) Context: sam e as in (38). Men igar, sent pa natten bad McArthuri endeligt sygeplejerenk g ive_________ en morfinindsprajtning. But yesterday, late on night-the asked M cA rthur finally nurse-the giv e-IN F _________a m orphine injection ‘But yesterday, late at night M cArthuri finally asked the nursek to g iv e ________ a m orphine injection.’ a. Isigi selv REFLj self b. sigi REFLj c. sig/t selv REFLk self (odd reading in this context) d. *sigk REFLk (* because give is anti-reflexive) A s show n in (39a) the presence o f this potential subject antecedent does m ake LD-binding o f sig selv m ore difficult to obtain than LD -binding o f sim ple sig (39b), but it does not exclude it. The in (39d) is due to the anti-reflexivity o f the predicate g/ve ‘give’. N ow , if it is the case that sig selv is even m arginally acceptable long-distance bound (given the right context) then Pica and Jakubow icz’s accounts o f sig selv as a local anaphor cannot be correct. We know from the discussion o f q-floated intensifiers in chapter 2, section 2.2.5, that adnom inal intensifiers may be stranded in the sam e way as floated quantifiers when the associate has to move to a higher position, e.g. V2 m ovem ent o f the subject DP, see (40a). We also know that intensifiers can be fronted while leaving their associate behind, as discussed in chapter 2, section 2.2.6, see (40b). (40) a. Lcereren havde selv skrevet en stil. [Lcereren, havde [dp f, selv] skrevet en stil] teacher-the had self written an essay ‘The teacher had him self w ritten an essay.’ b. Selv har kongen ringe magt. [Selv, har [dp kongen ti ] ringe magt] self has king-the slight pow er ‘The king him self has little pow er.’ It is therefore natural to assum e that the sig o f the intensified reflexive sig selv may leave the adjoined intensifier selv behind when it undergoes cyclic LF m ovem ent to get features from the m atrix subject in sentences like (38a) and (39a). W hile such an approach makes it possible to account for LD -binding o f sig selv, it does not by itself explain why LD-binding o f com plex R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 144 reflexives is so rare as to have been overlooked (or m arginalized) by researchers w orking on Danish reflexives for so long. W ithin the analysis proposed in this dissertation a straightforw ard explanation o f the rarity o f LD -bound sig selv naturally presents itself. As discussed in greater detail in section 3.3.2, one o f the main m otivations for intensifier- adjunction to sim ple reflexives is predicate m eaning. Anti-reflexive predicates like aflose ‘replace, relieve (som eone as guard)’ are not readily com patible with reflexive scenarios. Therefore, w hen such readings are forced, they have to be construed as relations between the referent o f the antecedent and a representation o f the referent o f the reflexive. As discussed in section 3.3.2, this doppelganger-effect triggers the intensification o f locally bound reflexives with anti-reflexive predicates, see (41a). (41) a. Vagterii qfl0ste *sigi / sigi selv. guard-the replaced REFL/REFL self ‘The guard replaced h im se lf’ b. Vagterii bad Peterafl0se sigi ^ k / sig7 ? , ii selv. G uard-the asked Peter replace REFL ‘The guard asked Peter replace h im se lf’ W hen the m eaning o f the predicate is com patible w ith reflexive scenarios (e.g. neutral and inherently reflexive predicates) then s'e/v-adjunction to locally bound sig is not m andatory for sem antic reasons. In the case o f neutral predicates, 5e/v-adjunction to sig thus has to be m otivated either by explicitly contrastive contexts or phonological factors (as in the case o f certain prepositional predicates). The sam e applies to LD -bound instances o f reflexives in argum ent position o f anti-reflexive verbs, see (41b). That is, as show n in (41b) vs. (38-9), in the absence o f an explicitly contrastive context, as for instance, the kind triggered by anti reflexive predicates (41a), intensification o f LD -bound 5/g ^e/v is not felicitous. O ther contexts w hich license intensification o f reflexives are the so-called wax m useum contexts involving the use o f reflexive pronouns to refer to statues/representations o f R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 145 the individual denoted by the antecedent. First consider the instances o f locally bound sig and sig selv in (42). (42) Context: Im agine Bill Clinton visiting the w ax museum. He notices a statue o f him self w ith an unshaven face. Since he doesn’t like the look o f the statue he takes out a razor and starts to shave it. A third party w itnessing this situation describes it in the follow ing terms: a. Bill Clintorii barberede sigi. b. ....... sigi selv. Bill Clinton shaved RELF/REFL self ‘Bill Clinton shaved 0 / 0 him self.’ The sentence w ith the sim ple reflexive in (42a) can only have the interpretation in w hich the real C linton shaves him self (i.e. the real Clinton), see (43a). It cannot have the so-called statue- reading, see (43b), in w hich the real Clinton shaves a statue o f Clinton. In contrast, the sentences w ith the com plex reflexive in (42b) can have both reading (43a) and (43b). (43) a. Clinton<real> shaves him self<real>. b. Clinton<real> shaves him self<statue>. W hat these exam ples show is that in order to get a statue-reading o f a reflexive pronoun it must be adnom inally intensified. The exam ple in (44) is designed to test w hat happens when a statue-reading o f a LD -bound reflexive is forced. (44) Context: Im agine Bill and Hillary C linton visiting the w ax m useum together. Bill notices a statue o f him self with w ith an unshaven face. He happens to have a razor in his inner pocket but since he is tem porarily in a wheel chair (due to a recent golfing accident) he hands the razor to Hillary and asks her to go over and shave the statue. A third party w itnessing this situation describes it in the follow ing terms: a. Clinton, bad H illaryharbere #sigi. b. (//sigi selv. C linton asked H illary shave REFL/REFL self ‘C linton asked H illary to shave him .’ A lth ou gh so lid ju d gm en ts are hard to co m e by in sen ten ces like (44)^“ * it seem s to be the case that sig selv is m ore felicitous than sig. O nce again, the generalization that sig selv cannot be Two factors conspire to make sentences like (44) extremely difficult to judge. First, there is a growing tendency to use pronouns instead of LD-bound reflexives. That is, most native speakers rarely use LD-bound reflexives, preferring to use pronouns instead. Second, many informants do not like statue-readings o f reflexives. The usual reaction being: “I would never say “Clinton shaved him self’ in such cases but rather “Clinton shaved the statue of R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 146 LD-bound appears to be too strong^^ That is, syntactically, sig and sig selv behave alike with respect to binding-related locality constraints. Pragm atically and sem antically they differ with respect to w hich contexts they may or may not occur in. But these differences falls under the theory o f intensification, as presented in chapter 2. So w hile it is true that there is a strong tendency for sig selv to have local antecedents this is due to the pragm atic/sem antic constraints on intensification rather than to syntactic locality constraints on the distribution o f reflexives and pronouns. 3.3.2 Locally bound reflexives in argument position: complex reflexives = a sub-type of adnominal intensiflcation. In the follow ing four sections (i.e., sections 3.3.2.1-4) an intensification-based account o f the distribution o f D anish sim ple and com plex reflexives, i.e. sig and sig selv, w ith different types o f predicates will be outlined. Based on w hat kinds o f nominal expressions they allow in object position, Danish verbal predicates can be divided into three main types^^: (i) “neutral” predicates w hich allow both sim ple and com plex reflexives (i.e. sig and sig selv) and DPs, see (45), (ii) “anti-reflexive” predicates w hich allow the com plex reflexive sig selv and DPs but not the sim ple reflexive sig, see (46), and (iii) “inherently reflexive” predicates w hich allow only the simple reflexive sig, see (47). (45) “N eutral” predicates: a. Peter vasker sig /sig selv / bilen. Peter washes REFL / REFL-SELF / car-the ‘Peter washes him self / the car.’ him self’, etc...”. However, if only given the choice between sig and and sig selv in sentences like (44), with a forced statue-reading of the reflexive, then sig selv appears more felicitous. See also example (109b) in section 3.3.5 which clearly shows that intensified possessive reflexives (e.g. sin egen ‘POSSREFL own’) may be LD-bound. This tripartite typology of predicates is directly inspired by a similar proposal by Zribi-Hertz (1995). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 147 b. Peter forsvarer sig /sig selv / Marie. Peter defends REFL / REFL-SELF / M ary ‘Peter defends him self / M ary.’ (46) “A nti-reflexive” predicates: a. Peter mistcenker *sig / sig selv / Marie. Peter suspects *REFL / REFL-SELF / M ary ‘Peter suspects him self / M ary.’ b. Peter misunder *sig / sig selv / Marie. Peter envies *REFL / REFL-SELF / M ary ‘Peter envies him self / M ary.’ (47) “Inherentlv reflexive” predicates'^: a. Peter shammer sig /*sig selv /*Maire. Peter sham es RELF / *REFL-SELF / *Mary ‘Peter is asham ed o f him self / M ary.’ b. Peter dukker sig / *sig selv / * Marie. Peter ducks REFL / *REFL-SELF / *Mary ‘Peter ducks *him self / *M ary.’ The distribution o f different types o f nom inal expression in object position w ith the different verb types is sum m arized in (48). (48) Distribution o f nominal expressions in obiect position^^: a. A nti-reflexive *sig / sig se lv / DP b. N eutral sig / sig se lv / DP c. Inherently reflexive sig / ^sigselv/ *DP We argue that the so-called “neutral” transitive verbs that allow the sim ple reflexive in direct or indirect object position retain their transitivity even when they occur with the simple reflexive sig. In contrast, other approaches (e.g. Reinhart and Reuland 1993, and others) assum e that the so-called “neutral” verbs occur tw ice in the lexicon: once, as a true transitive The exact definition “inherently reflexive” is discussed in chapter 4 where a distinction between inherently rettexive predicates with thematic reflexives vs. inherently reflexive predicates with non-thematic reflexives is introduced. The term “object position” is used here as a cover term for both direct and indirect object. As shown in (i-iii) below, di-transitive predicates can also be divided into the three main types: (i) anti-reflexive ditransitive : Peter solgte Peter sold (ii) neutral ditransitive (iii) inher. refl. ditransitive ■'sig REFL : Peter kobte sig Peter bought REFL : Peter anskaffede sig Peter acquired REFL / sig selv / REFL self sig selv / REFL self *sigselv / REFL self Mary Mary Mary Mary et hus. a house et hus. a house *Mary et hus. Mary a house ‘Peter sold/bought/acqulred himself/Mary a house.‘ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 148 verb (w hich require the com plex reflexive sig selv), and once as an inherently reflexive verb (w hich does not need to be overtly reflexive-m arked by the SELF elem ent o f the com plex reflexive). These com peting views o f “neutral” predicates will be com pared and evaluated in more detail in chapter 4, section 4.2. W ith respect to binding, reflexives, pronouns, and DPs have widely different properties. W hile DPs and pronouns, have sem antic content and are able to refer on their own, the reflexive sig does not have phi-features and behaves sem antically like a variable w hich is dependent on its binder for reference. However, from the point o f view o f intensifier- adjunction, reflexives, pronouns, and DPs are all nominal expressions w hich may be the target o f adnom inal intensification. In other words, by assum ing that the selv in sig selv is the same adnom inal intensifier as the one we find in intensified DPs, e.g. Peter selv kom til modet ‘Peter him self came to the m eeting’, we can now account for the intensified reflexive sig selv using the same analysis as the one used to analyze intensified DPs in chapter 2. Sim ply observing the distribution o f sim ple reflexives, com plex reflexives and DPs in object position, see (45-48), is not the only way to classify predicates either as “anti-reflexive”, “neutral”, or “inherently reflexive” . There are, as will be shown in the follow ing sections, binding-independent sem antic tests w hich can be used to classify predicates. 3.3.2.1 Anti-reflexives predicates: predicates which are incompatible with reflexive scenarios In this sectio n w e sh o w that it is p o ssib le to predict w hether or not a g iv en verb is an ti reflexive by testing its sem antic com patibility with reflexive scenarios. To illustrate how this sem antic testing o f predicate m eaning works let us take a closer look at the transitive predicates in (49). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 a. aflose ‘replace (the guard), succeed to ’ b. b0nfalde ‘implore, beseech, entreat, plead’ c. efterf0lge ‘succeed (e.g. Peter succeeded Kim as director)’ d. mistcenke ‘suspect (som ebody o f doing som ething)’ e. misunde ‘envy, be envious o f f. barbere ‘shave’ g- t0rre ‘dry’ h. pynte ‘adorn’ Since the verbs in (49) can all take full lexical DPs in object position, they cannot, per definition, be inherently reflexive. But how do we determ ine w hether they are “anti-reflexive” or “neutral” w ithout looking at the distribution o f sim ple and com plex reflexives? As m entioned above, we argue that testing for sem antic com patibility with reflexive scenarios makes it possible to predict w hether a transitive verb is neutral or anti-reflexive. On the basis o f this kind o f testing, we predict that the verbs in (49f-h) are neutral while those in (49a-e) are anti-reflexive. Based on w hat they mean, the predicates in (49a-e) sim ply do not m ake much sense in reflexive scenarios. For exam ple, how can one o f the guards standing guard in front o f B uckingham Palace possibly replace himself, see (49a) and (50a)7 The new w ell-rested guard can relieve/replace the old exhausted and sleep-deprived guard exactly because they are different persons. In other words, it is not possible for one person to replace him self in this sense and still rem ain the same person. The sam e goes for (49c) efterfolge ‘succeed’. Strictly speaking it m akes little sense to say that King Henry IV succeeded him self as king o f England, especially since in traditional m onarchies a new king can only be crow ned after the death o f the old, see (50b). One way to make sense o f such an utterance would be to assum e that King Henry IV only pretended to be dead and then dressed up as the crown prince in order to be crow ned in his place. In other words, he would have to appear in som eone else’s guise or under a different representation (i.e. that o f the crow n prince). That is, the generalization seems to be that anti-reflexive predicates presuppose that the linguistic expressions occupying R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 150 the positions o f external and internal argum ent do not refer to the same individual, or, as in the sentences in (50), do not refer to the same representation o f a given individuaP^. (50) a. Vagten a/l0ste *sig/ sig selv. guard-the replaced REFL / REFL self ‘The guard replaced him self (as guard). b. Henry IV efterfulgte *sig/ sig selv. Henry IV succeeded REFL / REFL self ‘Peter succeeded him self (as king o f England).’ c. Peter bonfaldte *sig / sig selv. Peter im plored REEL / REFL self ‘Peter im plored him self.’ d. Peter mistcenkte *sig/sig selv. Peter suspected REFL / REFL self ‘Peter suspected him self o f doing som ething.’ e. Peter misude *sig/ sig selv. Peter envied REFL / REFL self ‘Peter envied h im self The predicates b0ufalde ‘im plore’, mistcenke ‘suspect (som ebody o f doing som ething)’ and misunde ‘envy’ also fall under this generalization in that they only make sense when the argum ents are two different individuals, or the same individual appearing under a different representation, see (50c-e). Take the exam ple o f mistcenke ‘suspect’ in (50d). Suspecting som ebody o f m urder m eans that you have reasons to believe that this person com m itted a crim e but are still unsure w hether or not he is the culprit because you lack the evidence to prove it. Since people are norm ally assum ed to know w hether or not they have com m itted a crim e it does (under normal circum stances) not m ake sense to say that they suspect them selves o f having com m itted a crime. The uncertainty presupposed by the predicate mistcenke In the following the terms “representation”, “representational identity” and “representational non-identity” are used to refer to statue-readings, see (i), qua-sentences, see (ii), as well as any other type of sentence which involve differences in referential status between antecedent and reflexive. (i) Peter<real> barberede *sig<statue./sig<statue> selv. (statue-reading) Peter shaved REFL self ‘Peter shaved himself.’ (ii) Peter, i egenskab a f barber, barberede *sig/sig selv. (qua-sentence) Peter, in quality o f barber, shaved REFL ‘Peter, working as barber, shaved himself. The term “doppelganger-effect” is also used as a cover term for a wide range of phenomena involving differences in referential status. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151 ‘suspect’, illustrated by the contradictory flavor o f the sentence in (51), clashes with the presupposed knowledge o f one’s own acts and m akes the reflexive use o f this predicate infelicitous, unless som e kind o f presupposed representational non-identity is involved. (51) #Peter mistcenkte Mary for at have skudt Kennedy og han vidste at hun havde gjort det. # ‘Peter suspected M ary o f having shot K ennedy and he knew that she had done i t ’ That is, one way to m ake sense o f (50d) would be to assum e that Peter is schizophrenic and thus has m ultiple personalities, or to assum e that he has had m emory loss, etc^°. Let us now turn to the verbs barbere ‘shave’, torre ‘dry'’ and pynte ‘adorn’ in (49f-h). Testing them for com patibility w ith reflexive scenarios indicates that they are neutral rather than anti-reflexive. A proposition like Peter shaved/dried/adorned himself does not have any o f the non-sensical or weird flavor found w ith the sentences in (50a-e). The activities denoted by the predicates in (49f-h) are entirely com patible with reflexive scenarios. T hat is, the verbs barbere ‘shave’, torre ‘dry’ and pynte ‘adorn’ do not evoke any presupposition o f (representational) non-identity. A sentence like Peter shaved himself c& n (and usually does) m ean that Peter shaved his own face. It does not have to mean that he shaved him self under a different representation (e.g. a statue o f him self, etc.). Hence, based on com patibility with reflexive scenarios, we predict that these verbs should be neutral (i.e. be able to take both sig and sig selv in object position). The exam ples in (52) show that this prediction turns out to be true. (52) a. Peter barberede/torrede/pyntede sig / sig selv / John. P eter sh aved / dried / adorned R E F L / R E F L s e lf / John. ‘Peter shaved/dried/adom ed him self/John.’ The sentence in (i) might be construed as a counterexample to this account. (i) / suspected m yself o f having committed a fashion faux pas. Note, however, that in (i) it is not the identity of the person having committed a faux pas which is unknown, but rather the specific conditions under which wearing a specific garment might be considered a faux pas in a given social context, (p.c. J. Higginbotham) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 152 Based on the above exam ples and discussion we conclude that testing a predicate’s sem antic com patibility with reflexive scenarios can be used to determ ine w hether or not a predicate is “anti-reflexive” ^ '. If a transitive predicate is sem antically incom patible with reflexive scenarios then it is “anti-reflexive” . In other words, an anti-reflexive predicate can be defined as a predicate whose m eaning presupposes that its argum ents are either referentially different (i.e. have different sem antic values) or (if co-extensional) appear under different representations'^^ see (53). (53) A nti-reflexive predicates: A given predicate is anti-reflexive if it evokes a presupposition o f (representational) non-identity o f its argum ents that is not cancelable by context. Conversely, if a transitive predicate is sem antically com patible with reflexive scenarios then it is “neutral”. It is im portant to note that the definition o f anti-reflexivity defended in this dissertation does not crucially rely on the notion o f co-argum enthood. W hat is im portant is the relationship between reflexive and antecedent in term s o f presupposed representational identity vs. non-identity. Certain contexts trigger a presupposition that reflexive and antecedent are (representationally) non-identical. Predicate m eaning is only one am ong different source o f this presupposition. In sections 3.3.3, 3.3.4, and 3.3.5 we show that the m eaning o f certain resulative constructions, EC, constructions, and possessive constructions, e.g. (100), (104) and (111), also trigger a presupposition o f representational non-identity o f reflexive and anteeedent. The definition in (53), as well as the definitions o f neutral and inherently reflexive predicates given throughout section 3.3.2, are thus only to be considered definitions o f different predicate-types. The notions o f anti-reflexivity, neutrality and inherent Cf. E. Konig’s notion o f “other-directedness” and “other-directed” predicates. For more discussion of the notions of identity and representational identity, see section 3.3.2.7. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 153 reflexivity are more general and do not rely on the notion o f co-argum enthood but rather on presupposed identity vs. (representational) non-identity o f antecedent and reflexive. 3.3.2.2 Predicates which are semantically incompatible with non-reflexive scenarios Predicates that are sem antically incom patible w ith non-reflexive scenarios, i.e. predicates w hich require reflexive scenarios^^, can be classified as “inherently reflexive,” see (54)-(56). (54) a Peter I is out of hisi /*his, own / *DP^ ’ s mind. b. Peteri cleared hiSi / *hiSj own / *DP^ ’ s throat by saying ahem. (55) Peter solede sig / '*sig selv / *Marie. (Danish) Peter tanned REFL / REFL self / M arie ‘Peter was tanning/sunbathing.’ (56) a. Peter took the knife with him / *himself / *Mary. b. Peter tog kniven med s ig / *sigselv / *Mary. (Danish) Peter took knife-the w ith RELF / REFL self / M ary ‘Peter took the knife with him / * h im self/ *M ary.’ The m eanings o f the com plex predicate ‘be out o f one’s m ind' in (54) and the sim ple verbal predicate sole ‘tan ’ (55) both presuppose the identity o f the participants involved in the action/event. Unlike normal bathing, sunbathing can only be perform ed on one’s own body. Since the unique object cannot be contrasted with other potential sunbathees (i.e. Peter solede *Marie/*barnet/etc. ‘Peter sunbathed *M arie/*the child/etc.’) intensifier-adjunction to sig is not possible. Similarly, since it not possible to go out o f som ebody else’s mind, intensifier- adjunction o f owf'' to the possessive pronoun his in (54a) is not ok. The English and Danish exam ples in (56) also qualify as inherently reflexive. In the account proposed here the ju d gm en ts in th e ex a m p les in (5 6 ) are exp lain ed in th e sam e w ay as (5 4 )-(5 5 ). T he m ean in g o f Note the difference between the narrow definition of reflexivity (antecedent-anaphor relation between co arguments) and the more loosely defined notion of ‘reflexive scenario’ (which does not necessarily involve co arguments), e.g. while both (i) and (ii) qualify as reflexive scenarios only (i) involves reflexivity in the narrow sense. (i) Peter j washed 0 , himself. (ii) Peter j cleared [hiSi throat\. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 154 the w hole construction ‘X took a knife with Y ’ is only com patible with reflexive scenarios in which X=Y. All non-reflexive scenarios in which X-^^Y necessarily crash because they lead to non-sensical m eanings. The theory o f adnom inal intensification developed in chapter 2 can thus be used to explain why the com plex form o f the reflexive is not possible. Since the unique object cannot be contrasted w ith other individuals with whom the knife could have been brought (i.e. *Peter tog kniven m ed Marie/barnet/etc. ‘*Peter took the knife with M arie/the child/etc.’) intensifier-adjunction to sig is not possible because it would lead to a violation o f the constrastiveness condition on adnom inal intensification. Presupposed identity o f the participants in the action denoted by the predicate/construction can thus be used to determ ine w hether or not a given predicate is inherently reflexive, see (57). (57) Inherent reflexive predicates: A given predicate is inherently reflexive if it evokes a presupposition o f identity o f its argum ents that is not cancelable by context. Since the exam ples in (54), involving possessive pronouns and intensifiers, obviously do not involve antecedent-anaphor relations betw een co-argum ents o f the same predicate, they constitute strong evidence against both the lexical am biguity accounts o f local reflexives (cf. Reinhart and Reuland 1993) and reanalysis accounts (cf. V ikner 1985). The exam ples in (56) also pose a threat to lexical am biguity and reanalysis accounts since it is not obvious that the Subject DP Peter and him/sig in (56a,b) are co-argum ents o f the sam e predicate takeltage ‘take’. Unless, o f course, one analyzes take _ with as a com plex predicate. In contrast, since the nom inal approach to binding defended here is not based on co-argum enthood, it eneounters none o f these difficulties in providing a unified account o f (54-56). ' As discussed in section 3.3.5, we assume that own is the suppletive form of the adnominal intensifier himself. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 155 3.3.2.3 Neutral predicates: predicates which are semantically compatible with reflexive scenarios and which are not semantically incompatible with non- reflexive scenarios By com bining the sem antic tests proposed in the previous tw o sections, a given predicate can now be identified as “neutral” (independently o f binding-theoretic factors) if it is sem antically com patible w ith reflexive scenarios while at the sam e tim e not being sem antically incom patible witb non-reflexive scenarios. The verbs in (58a-c) m eet tbese requirem ents and tbus qualify as neutral. (58) a. barbere ‘sbave’ b. tarre ‘dry’ c. pynte ‘adorn’ A s discussed above, the verbs in (58a-c) are entirely com patible w ith reflexive scenarios, i.e. they do not evoke any presupposition o f non-identity o f the representations o f their argum ents. A sentence like Peter shaved (himself) can (and usually does) mean that Peter shaved his own face. It does not have to m ean that he shaved him self under a different representation. We thus conclude that (58a-c) are not anti-reflexive. Furtherm ore, these predicates are also com patible w ith non-reflexive scenarios. That is, sentences like Peter shaved John are perfectly m eaningful. This shows that these verbs are not inherently reflexive, i.e. do not evoke any presupposition o f identity o f their argum ents. Based on such sem antic testing for anti-reflexivity, see (53), and inherent reflexivity, see (57), we conclude that (58a-c) are neutral predicates. M ore generally, we define neutral predicates negatively as predicates that are devoid o f particular presuppositions concerning the (representational) identity o f their argum ents. If (5 8 a -c ) are neutral predicates then th ey sh ou ld a llo w sig, sig selv and D P s in object position. As (59) shows, this prediction turns out to be true. (59) Peter, barberede/torrede/pyntede sig, / sigi selv / John. Peter shaved /dried / adorned REFL / REFL self / John ‘Peter shaved (him self) / John.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 156 N otice, however, this kind o f sem antic testing o f neutral predicates is not problem -free. Even if a transitive predicate is logically com patible w ith reflexive readings this does not necessarily mean that it always allows the sim ple reflexive sig in object position (as a neutral predicate should do). There are predicates that appear to behave as “anti-reflexives” even though they are logically com patible w ith reflexive readings. The verb koge ‘boil’ is an exam ple o f such a predicate, w hich at first glance appear to behave as an anti-reflexive in that it disallows simple sig, see (60), but is nevertheless sem antically com patible with reflexive readings. (60) Peter kogte #sig / sig selv / Mary. Peter boiled REFL / REFL self / M ary ‘Peter boiled him self / M ary.’ W hile it may be strange or unexpected given w hat we know about the world, it still does not lead to nonsense to say that som eone consciously puts him self in a big water-filled pot on the stove and then turns on the gas in order to boil himself. Sem antically both barbere ‘shave’, see (59), and koge ‘boil’, see (60), classify as predicates which are com patible with non-reflexive scenarios while at the sam e tim e not being incom patible with reflexive scenarios. They should thus both behave as neutral predicates. However, as illustrated by (59) and (60) they differ w ith respect to the acceptability o f sim ple sig. How do we explain this difference? The fact that predicate m eaning is not sufficient to tell the tw o verbs apart suggests that their differences are to be explained at the level o f expectations generated by standard assum ptions about the world rather than lexical semantics. U nder normal circum stances a person is not expected to boil him- or herself. P redicates o f th is typ e, e.g. k o g e ‘boil, can ch an ge b eh avior (e.g. from “a n ti-reflex iv e” to “neutral”) as a result o f changes in the larger context in which it occurs. Com pare the way the predicate koge ‘boil’ behave in a “norm al” context, see (60) w ith how it behaves in science-fiction context, see (61). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 157 (61) Science-fiction context based on “special” assum ptions about the w orld: Viste du at marsbeboenes hud kan tale langt hojere temperaturen end vores. Faktisk er det ganske normalt fo r marsbebeoere at vaske sig i varm olie. Nagle steder paa Mars er det enddog normalt at koge sig i olie m ed regelmcessige mellemrum. ‘Did you know that the skin o f the M artians can endure m uch higher tem peratures than ours. Indeed, it is quite norm al for M artians to wash them selves {sig) in hot oil. Some places it is even norm al to boil oneself {sig) in oil with regular intervals.’ The different contexts in (60) and (61), based on different basic assum ptions about the way the world works, give rise to different sets o f expectations. Based on w hat we know about boiling and the consequences it has for the hum an body we expect the verb koge ‘boil’ to be other- directed, i.e. under norm al circum stances it denotes an activity that is directed at things or people other than oneself. Hence, it is odd (but not im possible) to say that som ebody boiled him self. This explains why sim ple sig is not fully acceptable in (60). N ote that sig is not ungram m atical in (60) but m erely unacceptable^^. As indicated in (60), under normal circum stances focusing o f the reflexive through adnom inal intensification (i.e. sig selv) is necessary to m ark the (representational) non-identity evoked by the expectations o f other- directedness associated with the predicate koge ‘boil’ in the normal, non-science-fiction context. In contrast, in the science-fiction-like context in (61) the background assum ptions about the w orld are different and no longer clashes w ith the reflexive scenario x boiled x. Hence unintensified sig can occur as the object o f koge ‘boil’ w ithout m aking the sentence unacceptable. Based on the above discussion we conclude that there are tw o sub-types o f “neutral” predicates, see (62). We use the symbol * to indicate grammaticality violations and # to indicate violation of pragmatic/discourse principles. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 158 (62) Sub-types o f neutral predicates: a. (NormaO neutral predicates: A given predicate is neutral if it lacks presuppositions concerning the (representational) identity o f its arguments. Exam ples: vaske ‘w ash’ barbere'shdMQ' t0rre ‘dry’ forsvare ‘defend’ b. “H idden” neutral predicates: A given neutral predicate is a “hidden” neutral verb if normal background assum ptions about the world trigger expectations o f (representational) non identity o f its arguments. Examples: koge ‘boil’ dolke ‘stab’ film e ‘film, shoot’ forgylde ‘gild’ Both the norm al neutrals in (62a) and the “hidden” neutrals in (62b) are defined by their lack o f presuppositions o f either (representational) non-identity or identity o f their argum ents. To the untrained eye, however, the predicates in (62b) at first appear to behave like anti-reflexives in that they disallow sim ple sig under normal background assum ptions about the world, see (60). But given the right context (i.e. a different set o f background assum ptions about the w orld) they can be shown to accept sim ple sig in object position, see (61). Since their true nature as neutral verbs (i.e. the lack o f presuppositions about the identity or non-identity o f their argum ents) is hidden under normal circum stances, we refer to these verbs as “hidden” neutral predicates. N ote that the anti-reflexive predicates discussed in section 3.3.2.1 cannot be made to allow sim ple sig no m atter how m uch the context is changed. That is, the presupposition o f non-identity o f the representations o f the argum ents evoked by anti-reflexive predicates cannot be overridden or cancelled by context. That is, even in a science-fiction scenario in which M artians are described as incurable schizophrenics who pass their days suspecting them selves o f this and that; even in such scenarios, sim ple sig is still not ok in sentences like (63). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 159 (63) Marsbeboeren mistcenkte *sig / sig selv / Peter. M ars-dw eller-the suspected REFL / REFL self / Peter ‘The M artian suspected him self / Peter.’ The present approach to reflexives can account for the change-of-behavior data in (60-61) w ithout assum ing the existence o f any specific lexical features distinguishing between “hidden neutral” and “norm al neutral” predicates. Instead, it is assum ed that the predicate koge ‘boil’ is one and the sam e in both (60) and (61) and that it is the changes in context (i.e. background assum ptions about the w orld) w hich are responsible for generating different sets o f presuppositions in (60) and (61). This contrasts w ith other approaches to reflexives (cf. Reinheart and Reuland 1993) which w ould have to assum e the existence o f two hom ophonous predicates: (i) kogel ‘boil (som ething)’ a “norm al” transitive predicate, and (ii) koge2 ‘boil (oneself)’, an inherently reflexive predicate. 3.3.2.4. Summary: Presuppositions triggered by predicate meaning In sections 3.3.2.1-3 we have shown that predicate m eaning and pragm atic factors can be used to predict w hether a given verb is anti-reflexive, inherently reflexive, or neutral. W hether a predicate is anti-reflexive or not depends on w hether or not it presupposes (representational) non-identity o f its argum ents, see the definition o f anti-reflexivity in (53), repeated here as (64). W hether a predicate is inherently reflexive or not depends on w hether it presupposes identity o f its argum ents, see (57), repeated here as (65). As shown in (66) neutral predicates are characterized by the lack o f presuppositions concerning the identity or non-identity o f their argum ents. Finally, “hidden” neutral predicates, w hich also lack such presuppositions, are distinguished from (norm al) neutral predicates in that they evoke a expectations (based on w orld-know ledge) o f the (representational) non-identity o f their argum ents, see (62b) repeated here as (67). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 160 (64) A nti-reflexive predicates: A given predicate is anti-reflexive if it evokes a presupposition o f (representational) non-identity (o f its argum ents) that is not cancelable by context. (65) Inherently reflexive predicates: A given predicate is inherently reflexive if it evokes a presupposition o f identity (o f its argum ents) that is not cancelable by context. (66) N eutral predicates: A given predicate is neutral if it lacks presuppositions concerning the identity (o f its argum ents). (67) “H idden” neutral predicates: A given neutral predicate is a “hidden” neutral verb if it normal background assum ptions about the world trigger expectations o f (representational) non-identity (o f its argum ents). It is im portant to rem em ber that the different types o f predicates defined in (64-7) are not based on featural differences. Rather the behavior o f a given predicate with respect to intensification o f reflexives depends directly on its m eaning (w hether or not it has presuppositions concerning the identity o f its argum ents) and, in the case o f hidden neutrals, on the m eaning o f the verb plus background assum ptions about the world. That is unlike, for exam ple Zribi-H ertz who bases her predicate typology on a [+/-disjoint reference feature], the present account does not rely on binding-related features. Furtherm ore, as m entioned above, although the definitions in (64)-(67) refer to argum ents o f predicates, the notion o f argum enthood is not crucial for sem antic notions o f “anti-reflexivity”, “neutrality” and “inherent reflexivity” . 3.3.2.S Complex reflexives as arguments of proto-typical neutral predicates. In the previous section, we explained why locally bound reflexives occurring in object position o f neutral predicates do not have to be intensified. However, we still haven’t addressed the questions why neutral predicates, unlike inherently reflexive predicates, can have locally R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 161 bound intensified reflexives as direct objects and why they do take have intensified reflexives in som e contexts by sim ple reflexives in other. A s shown in chapter 2, adnom inal intensification o f any kind o f nominal expression is felicitous only in contrastive contexts, i.e. in contexts in w hich the intensified DP could have some other referent, see the contrastiveness principle repeated here as (68). (68) Contrastiveness condition on adnom inal intensification: A nom inal expression can only be intensified adnom inally if it can be contrasted with other expressions in the context in w hich it is found. O f the three unsaturated propositions in (69-71b) both (69b) and (70b) give rise to felicitous sentences when expressions referring to entities other than the subject Peter are saturating the em pty argum ent position indicated by x. In other words, with both neutral and anti-reflexive predicates the referent o f the expression filling the internal argum ent position can potentially be contrasted with other entities. Hence intensification o f sig is possible w ith these kinds o f verbs. In contrast, w ith inherently reflexive predicates (71) the argum ent position can only be filled by an expression w hich corefers w ith the subject Peter, i.e. sig. Hence, the referent o f the expression filling the internal argum ent position cannot be contrasted with other entities. Consequently, (68) rules out intensification o f reflexives with inherently reflexive predicates. (69) a. Peter mistcenkte *sig / sig selv / Mary. Peter suspects * R E F L / REFL s e lf/ M ary b. Peter suspects x. (70) a. Peter vasker sig / sig selv / Mary. Peter washes REFL / REFL self/ M ary b. Peter washes x. (71) a. Peter tog en kniv med .‘ tig / *v/g- selv / *Mary. Peter took a knife w ith REFL / REFL self/ M ary b. Peter took a knife with x. A s show n in section 3.3.2.1, in the case o f anti-reflexive predicates it is the presupposition o f (representational) non-identity o f their argum ents which trigger the intensification o f sig. But R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 1 6 2 since, as argued in the previous section, neutral predicates do not presuppose the (representational) non-identity o f their argum ents, the intensification o f sig w ith proto-typical neutral predicates has to have some other m otivation. In the literature it has som etim es been claim ed that intensification o f sig is optional w ith groom ing predicates (e.g. wash, dress, clean, etc..). However, as the exam ples in (72)-(73) show, this generalization needs qualification. Intensification o f sig w ith proto-typical neutral predicates is not optional, but rather determ ined by discourse factors like focus and contrastiveness. That is, intensification o f sig has to occur w hen the referent o f sig is focused or explicitly contrasted with som e other entity, see (72), and it cannot occur in contexts w here sig is not being contrasted or focused, see (73). (72) Q: Hvem var det nu at Peter havde vasket? Havde han vasket sin hund? who was it now that Peter had w ashed had he washed PO SSREFL dog ‘W ho was it that Peter had w ashed? Had he w ashed his dog?’ A: Nej, han havde vasket SIG S E L V / *SIG. N o he had w ashed REFL SELF / REFL ‘N o, he had w ashed h im se lf’ (73) Q: Hvad var det nu at Peter havde gjort fo r han gik i seng? w hat was it now that Peter had done before he went to bed ‘W hat was it Peter had done before he w ent to bed?’ Havde han barberet sig? bad be shaved REEF ‘Had he shaved him self?’ A: Nej, han havde VASKET sig / #sig selv. no he had W ASHED REFL / REFL self ‘N o, He had W ASHED h im se lf’ The exam ple in (73) shows that when the contrast is not on the object o f vaske ‘w ash’ as in (72) but rather on the predicate itself, then sig can not felicitously be intensified by selv- adjunction. This is due to the fact that no m ore than one instance o f focus is allowed in one sentence (cf. B aker’s (1995)). It has been argued more that a sentence may contain more than on focused elem ent, e.g. Sim pson & W u (2002). Regardless o f w hether one or m ore foci is allowed per sentence, the unacceptability o f (73) follows from the lack contextual m otivation R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 163 for focusing the sim ple reflexive sig, in addition to the predicate. In other words, we do not exclude the possibility that it m ay be possible to find contexts in w hich (73) is ok. Based on the above discussion, we conclude that although neutral predicates allow both sim ple unintensified and intensified reflexives, intensifier-adjunction o f reflexives still needs to be licensed. C ontrastive focus is one way to license com plex reflexives w itb neutral predicates. In the section 3.3.2.7 additional “contextual” triggers o f intensifier adjunction o f reflexives w ith neutral predicates will be discussed. 3.3.2.6 Using inalienable possession to test whether a verb is neutral or anti-reflexive Inalienable possession may be used as a binding-independent method to test w hether a given predicate is neutral or anti-reflexive. N eutral verbs differ from anti-reflexive verbs not only with respect to their m eaning (i.e. w hether or not they presuppose (representational) non identity o f their argum ents) and, consequently, the distribution o f sim ple and com plex reflexives, but also w ith respect to the possibility o f taking inalienably possessed direct objects. The exam ples in (74-75) illustrate the differences between neutral and anti-reflexive with respect to inalienable possession. (74) N eutral predicates allow inalienable possession: Peter vaskede hcenderne. Peter w ashed hands-the. o k / ok, (i) ‘Peter w ashed his own tw o hands.’ (inalienable) (ii) ‘Peter w ashed the hands (o f the body he was dissecting).’ (alienable) (75) A nti-reflexive predicates do not allow inalienable possession: Peter stjal hcenderne. Peter stole hands-the. (i) ‘Peter stole his ow n tw o hands.’ °'‘(ii) ‘Peter stole the hands (o f the body he was dissecting).’ (inalienable) (alienable) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 164 (76) “H idden neutral” predicates tend not to allow inalienable possession: Peter kogte hcenderne. Peter boiled hands-the. *(i) ‘Peter boiled his own tw o hands.’ (inalienable) “'‘(ii) ‘Peter boiled the hands (o f the body he was dissecting).’ (alienable) For the sentence in (74) both o f the readings in (i) and (ii) are ok. For the sentence in (75) only the reading in (ii) is possible. That is, while neutral predicates allow inalienable possession, anti-reflexive predicates, e.g. stjcele ‘steal’ in (75), do not. As expected, “hidden neutral” predicates fall in between. Under norm al circum stances they are incom patible with inalienable possession, see (76), but given the right context, e.g. the science-fiction context discussed in section 3.3.2.3, it can be allowed, see (77). (77) Marsmanden kogte hcenderne i d ie forend han lavede mad. (inalienable) M artian-the boiled hands-the in oil before he m ade food ‘The M artian w ashed his hands in oil before cooking.’ The table in (78) sums up the behavior o f different types o f predicates with respect to inalienable possession. (78) Testing the availability o f inalienably possessed DP objects. Predicate tvne Allows inalienable possession a. A nti-reflexive b. N eutral c. “H idden N eutral” d. Inherently reflexive N O YES YES/NO (depending on context) (D oes not apply) A ccording to Lodrup (1997), who analyzes sig as a generalized inalienable, the ability to license inalienable possession is a com m on property o f all neutral predicates. Lodrup needs this assum ption to support his claim that locally bound sim ple sig and inalienable possessed objects are found in the sam e contexts. W hile we do not share these assum ptions we do consider inalienable possession to be a useful (though not fool-proof) m ethod to distinguish betw een anti-reflexive and neutral predicates. H owever, rather than analyzing sig as a generalized inalienable, as proposed by Lodrup (1997), we adopt V ergnaud and Zubizarreta’s (1992) analysis o f inalienable R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 165 constructions w hich is based on the assum ption that inalienable nouns take a possessor argum ent w hich is syntactically bound to the external possessor (which in Danish is usually the subject DP). Vergnaud and Zubizarreta argue tbat w hile inalienable nouns take a possessor argum ent, see (79a), alienable nouns do not, see (79b). (79) a. hcenderne(x) ‘the hands’ b. hcenderne ‘the hands’ They furtherm ore assum e the existence o f a lexical redundancy rule relating the tw o lexical entries in (79), cf. Vergnaud and Zubizarreta 1992, p. 596, 601, num bers o f exam ples bave been adapted. A dopting this approach enables us to view inalienably possessed nouns to be a kind o f anaphoric nom inal expressions whose distribution can be accounted for by the principles o f binding theory. The fact that inalienable possessed nouns seem to be found in the same environm ents as sim ple sig is therefore not surprising. The presupposed (representational) non-identity o f anti-reflexive predicates explains why the inalienable reading o f hcenderne ‘the hands’ in (75) is not available. In other words, the presupposition o f (representational) non-identity extends to inalienably possessed objects. Since tbe anti reflexive m eaning o f stjcele ‘steal’ in (75) presupposes representational non-identity o f its argum ents it also presupposes tbe representational non-identity o f the subject and the anaphoric possessor argum ent o f the inalienable object noun^®. Note that it could be argued that the analysis defended here would lead to the wrong prediction that intensification of the 0 possessor argument anaphor should be possible, see (1). (i) *Peter stjal 0 egm hcender(ne). Peter stole own hands Furthermore, it could also be argued that the analysis proposed here does not explain why (ii) is (marginally?) grammatical without the intensifier egen ‘own’. (ii) Peter stjal sine hcender. Peter stole POSSREFL hands ‘Peter stole his hands.’ One possible answer to these objections might be found in the fact that in the case o f possessive reflexives the full paradigm includes three different forms, e.g. 0 (i.e. inalienable possession), sin ‘POSSREFL’, and sin egen ‘POSSREFL own’, in addition to the possessive pronouns hans ‘his’ and hans egen ‘his own’. In a sense, the simple unintensified possessive intensive sin ‘POSSREFL’ can thus be considered the ‘intensified’ version of the 0 possessor argument in inalienable constructions. Another solution to the problems raised above, could potentially be found in the fact that unlike sig ‘REFL’ which is unstressable, the possessive reflexive sin ‘POSSREFL’ can R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 166 3.3.2.7 Linking doppelganger effects and anti-reflexivity Intensifier-adjunction to sim ple reflexives can take place in a num ber o f different contexts. In sections 3.3.2.1 and 3.3.2.3 anti-reflexive (containing presuppositions o f (representational) non-identity) and “hidden neutral” predicates (w hich together with background assum ptions about the w orld generate expectations o f (representational) non-identity)) were shown to trigger intensification o f sim ple reflexives. In section 3.3.2.5 contrastive contexts were identified as another trigger o f intensification o f simple sig with neutral predicates. In this section a num ber o f other contexts triggering intensification o f sim ple reflexives with neutral predicates will be discussed, nam ely w ax m useum contexts (statue-readings), qua-sentences, and strict readings o f reflexives in VP-ellipsis constructions. In the follow ing we will show that all these intensification triggering contexts, see (80d(i-iv)), involve some kind o f ‘doppelganger-effects’, i.e. (representational) non-identity o f the reflexive and its antecedent. (80) Com plex reflexives (e.g. sis selv): Exam ple: a. anti-reflexive predicates (50) b. hidden neutral predicates (60-1) c. contrastive contexts (w ith neutral predicates) (72) d. ‘doppelganger-effects’ (w ith neutral predicates): (i) statue-readings (81) (ii) qua-sentences (84) (iii) strict (and sloppy) readings in VP ellipsis (86) In other words, we argue that all the phenom ena in (80a-d) should be given a unified analysis in term s o f (representational) non-identity o f the reflexive and its antecedent, the only difference being the locus o f the presupposition o f representational non-identity: w ith the anti reflexive predicates (80a) the presuppositions is part o f the m eaning o f the predicate; in the case o f the different types o f doppelganger-effects listed in (80b-d), the presupposition comes from the surrounding context or larger discourse. carry stress on its own without the help of the adnominal intensifier. See section 3.3.6 for more discussion of the link between stressability and adnominal intensification. While the exploration of these issues is highly relevant to the present analysis, space considerations force us to leave this topic for future research. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 167 3.3.2.7.1 Selv as marker of statue-readings Since Jack en d o ff s (1992) paper “M adam e Tussaud m eets the binding theory” called attention to the fact that nam es can be used to refer to statues or representations o f their normal referents, it has been known that sim ple and com plex reflexives differ in this respect. First consider the instances o f locally bound sig and sig selv in (8 1). (81) Context: Im agine Bill Clinton visiting the w ax museum. He notices a statue o f him self with an unshaven face. Since he doesn’t like the look o f the statue he takes out a razor and starts to shave it. a. Bill Clintorii barberede sig/. b. Bill Clintorii barberede sigt selv. Bill Clinton shaved RELF/REFL self ‘Bill Clinton shaved (him self).’ The sentence w ith the simple reflexive sig in (81a) can only have the interpretation in which the real Clinton shaves him self (i.e. the real Clinton), see (82a). It cannot have the so-called statue- or doppelganger-reading, see (82b), in w hich the real C linton shaves a statue o f Clinton. In contrast, the sentence w ith the com plex reflexive sig selv in (81b) can bave both reading (82a) and (82b)” . (82) a. Clinton<real> shaves him self<real>. b. Clinton<real> shaves him self<statue>. We argue that this sem antic difference betw een sim ple/unintensified reflexives and com plex/intensified reflexives is a consequence o f intensifier-adj unction. As described in chapter 2, adnom inal intensification autom atically generates a set o f alternative referents for the associate o f the intensifier. In the case o f sig selv it is thus the focus-generated set o f alternative sem antic values for sig (triggered by adnom inal intensification) which gives rise to doppelganger-effects, i.e. statue-readings in w ax-m useum contexts. N ote that a sim ilar At first glance, the fact that the complex reflexive sig selv can have both readings in (82) may seem to constitute a problem for the analysis presented here. If intensification of sig licenses statue-readings, then why can we have instances o f sig selv without statue-readings (e.g. (81b) read as (82a))? We suggest that in these cases the selv- adjunction simply serves other purposes, e.g. contrastive focus, see section 3.3.2.5. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 168 proposal is found in K onig & Siem und (1999:48)^*. In its unintensified/sim ple form sig behaves as a variable w hich has to be referentially and representationally identical to its binder. Thus the present approach achieves a (sem antically and m orphologically) fully com positional analysis o f com plex reflexives. That is, w hile the sim ple reflexive (e.g. Dan. sig) has to be corefential (and co-representational) with its antecedent, the com plex reflexive (e.g. Dan. sig selv) may be either coreferential w ith its antecedent, or refer to a representation o f the antecedent or the referent o f the antecedent under a different representation^"*. 3.3.2.7.2. Qua-sentences The exam ples in (83) illustrate the m otivation for analyzing dobbelganger-effects in term s o f representational identity/non-identity rather than in term s o f referential identity. (83) a. ??*Peter, working as barber, shaved. b. Peter, working as barber, shaved himself. Discussing similar examples in Dutch, see (i) below, where the complex reflexive zich z e ^ ‘REFL self allows statue-reading but the simple reflexive zich ‘REFL’ does not, Konig & Siemund (1999:48) suggest that “What zelf adds to the meaning of the reflexive is the evoking of alternatives to the value given which is, of course Mary. Since in the context given there is no mention of other people being present, who could be such alternatives, and given that we know that there are many statues of famous people present, one of which represents Mary, there is a tendency to think of the statues as alternatives to the value given, which is therefore interpreted as Mary’s statue rather than Mary herself.’’ (i) Context: Mary is famous and walks into Madame Tussaud’s: a Ze keek in een spiegle en ze zag zich in een hoek siaan. ‘She looked into a mirror and she saw herself standing in a corner.’ b. Ze keek in een spiegle en ze zag zichzelf in een hoek staan. ‘She looked into a mirror and she saw herSELF standing in a corner.’ In Russian the simple reflexive sebja ‘REFL’ allows both coreferential and doppelganger readings. This appear to be a counter-example to generalization that simple reflexives require coreferntiality while complex reflexives also allow doppelganger readings. Note, however, that unlike, simple sig in Danish, which cannot be stressed, sebja is stressable. This difference in stressability may be responsible for the different behavior of these reflexives w.r.t. doppelganger effects. Russian also has an unstressable simple reflexive, i.e. the suffixal -sja, which behaves like Dan. sig in that it has to be both referentially and representationally identical to its antecedent. Russian sebja is thus more similar to German sich which also allow both coreferential and doppelganger readings, than Danish sig. See section 3.3.6 for more discussion of the link between stressability, intensification and ‘doppelganger’ effects. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 169 Im agine a context w here a group o f soldiers are lost in a forest for weeks. In order to m aintain a certain level o f hygiene the platoon leader picks Peter to function as cam p barber and im m ediately orders him to shave everybody in the camp including him self. In this context, (83b) is clearly preferred over (83a). Similarly, in the corresponding Danish sentence in (84b) the com plex reflexive sig selv is much better than the sim ple reflexive sig in (84a). (84) a. Peter, i egenskab a f barber, barberede ??*sig. Peter in quality o f barber shaved REFL ‘Peter, qua barber, shaved.’ b. Peter, i egenskab a f barber, barberede sig selv. Peter in quality o f barber shaved REFL self ‘Peter, qua barber, shaved h im se lf’ Now, in all the sentences in (83a-b) and (84a-b) there is clearly referential identity betw een the antecedent Peter and the reflexive pronoun. The qua-sentences thus differ from the statue- reading sentences, discussed in the previous section, w here the relation betw een reflexive and antecedent does not involve referential identity but rather a relation betw een a real person (the referent o f the antecedent) and a statue o f this person (the referent o f the reflexive). So if the dopppelganger-effect is defined as involving lack o f referential identity betw een antecedent and reflexive (as in the case o f (82b)) then the m andatory com plex reflexive in qua-sentences w ould not qualify as a doppelganger-effect. Hence, we would expect (83a) and the corresponding sentence with simple sig in (83c) to be ok. This however, is clearly not the case. Furtherm ore, at an intuitive level, the sentences in (83b) and (84b) are clearly felt to involve som e kind o f doppelganger o f Peter. That is, the individual Peter is perceived o f as occupying two roles at the sam e time. O n the one hand, he is perform ing the duty o f squad barber. On the other, he is the object o f the shaving efforts o f the squad barber. In other words, Peter is perceived as appearing in two different formal roles or representations. So rather than involving both referential and representational non-identity o f the argum ents o f the predicate (as was the case o f statue-reading in w ax m useum contexts) qua-sentences involve only R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 170 representational non-identity. The referents o f reflexive and antecedents are the same in (83b) and (84b), i.e. the individual Peter, but he appears under different representations. In this respect, the qua-sentences are very sim ilar to the sentences with anti-reflexive predicates discussed in section 3.3.2.1, e.g. (50b) repeated here as (85). (85) Henry IV efterfulgte * sig /sig selv. Henry IV succeeded REFL / REFL self ‘Peter succeeded him self (as king o f England).’ Strictly speaking the com plex reflexive sig selv and the antecedent Henry IV refer to the same individual in the world. However, as discussed in section 3.3.2.1 the only way to make sense o f such an utterance would be to assum e that reflexive and antecedent refer to different representations o f the same individual, i.e. in this case Henry IV him self vs. Henry IV appearing disguised as the crown prince after feigning his own death. A nalyzed in this way, anti-reflexivity (85), statue-readings (82b), qua-sentences (83b, and 84b), can all be said to involve representational non-identity. In the case o f anti-reflexive predicates, it is the m eaning o f the predicate w hich presupposes representational non-identity. Since both statue-readings and qua-sentences can be constructed w ith sem antically neutral predicates (e.g. barbere ‘shave’ in (82) and (83-84)) the representational non-identity cannot have its source in the predicate in these cases. As discussed above, in the case o f statue-readings, representational non-identity is m otivated by the w ax-m useum context, or sim ilar contexts involving an individual and a representation (i.e. statue, painting, picture, photo, tape-recording, video recording'"’, etc.) o f this individual. In the case o f qua-sentences, the representational non identity is introduced by the word qua itself, or one o f its equivalents, e.g. working as, etc. In brief, anti-reflexivity, statue-readings, and qua-sentences all involve representational non identity, also referred to here as the ‘doppelganger-effect’, albeit from different sources. Introducing the notion o f representational identity thus m akes it possible to arrive at a unified R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 171 analysis o f these three phenom ena w hich have hitherto been considered to be unrelated. In the follow ing section we argue that this analysis should also be extended to another phenom enon involving com plex reflexives, i.e. the choice betw een strict and sloppy reading o f reflexives in VP-ellipsis constructions. 3.3,2.7.3 Strict readings of reflexives in VP-ellipsis In this section we argue that the strict identity reading o f reflexives in VP-ellipsis constructions is due to doppelganger-effects, i.e. it involves representational non-identity. The exam ple in (86) illustrate the difference betw een sim ple and com plex reflexives w ith respect to the availability o f strict and sloppy reading in ellipsis constructions. (86) a. Peter vaskede sig og det gjorde Hans ogsa. (only sloppy) Peter w ashed REFL and it did Flans also ‘Peter w ashed (him self) and so did Peter.’ b. Peter vaskede sig selv og det gjorde Hans ogsa. (strict and sloppy) Peter w ashed REFL self and it did Hans also ‘Peter w ashed him self and so did Peter.’ N ote that the sam e paradigm is found in English, see (87). (87) a Peter washed 0 and so did Hans. (only sloppy) b. Peter washed 0 him self and so did Hans. (both sloppy and strict) As show n in (86) the sim ple reflexive sig can only be given a sloppy reading while the com plex reflexive sig selv can have both sloppy and strict readings. The sentences in (86) both involve tw o propositions the first being expressed by the sentences Peter vaskede sig in (86a) and Peter vaskede sig selv in (86b) and the second being expressed by the sam e sentence, i.e. og det gjorde Hans ogsa, in both cases. The first proposition in both (86a) and (86a) involve referential identity o f the argum ents o f the verb wash, they differ with respect to representational identity. The first conjoint in (86a), i.e. Peter vaskede sig, expresses a self w ashing activity. In contrast due to the presence o f the intensified selv ‘s e lf, the first conjoint See Rooryck and van den Wyngaerd (1999) for a number of interesting examples in Dutch. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 172 in (86b), i.e. Peter vaskede sig selv, implies that Peter washed som ebody and that this som ebody happened to be Peter. In this respect, the first conjoint in (86b) is sim ilar to a qua- sentence, i.e. it involves a case o f accidental identity o f tw o participants in an activity. That is, in both (86b) Peter w ashes/shaves som ebody who ju st happens to be himself. In other words, Peter is perform ing a shaving/w ashing action on his doppelganger, i.e. a representation o f himself. Hence, the necessary presence o f the intensifier selv. It has been claim ed that long-distance sig has both strict and sloppy readings, see (88). (88) Peteri bad sygeplejersken vaske sigt og det gjorde Hans ogsa. Peter asked nurse-the w ash REFL and it did Hans also ‘Peter asked the nurse to wash himself, and so did H ans.’ Reliable data on the availability o f strict and sloppy readings o f LD-5ig^ is difficult to com e by. A lthough (88) has been claim ed to be ok with a strict reading o f sig, it appears to me that (88) can only felicitously be given a sloppy identity reading in an out-of-the-blue context. Indeed if asked to m ake-up a context for (88) m ost inform ants assum e a scenario in w hich Peter and Hans are patients in the same ward in a hospital w ho both ask the nurse to washes them selves. So far none o f the inform ants that I have consulted provides a context in w hich Peter is the only patient and Hans is a relative who happens to be visiting Peter and who out o f concern for his welfare asks the nurse to w ash Peter. Furtherm ore, if forced to provide a sentence w ith a strict reading, then m ost speakers prefer the sentence in (89) with the pronoun ham ‘him ’ instead o f the reflexive. (89) Peter, had sygeplejersken vaske ham, og det gjorde Hans ogsa. Peter asked nurse-the wash REFL and it did Hans also ‘Peter asked the nurse to wash him self and so did H ans.’ In other words, there is no need to force long-distance reflexives to have strict readings since exactly in those contexts pronouns (which can always have strict readings) are allowed (and, for m ost speakers, preferred over reflexives). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 173 Based on the above discussion o f reflexives in VP-ellipsis constructions we conclude that the availability o f strict readings o f reflexives is an instance o f the doppelganger-effect (i.e. (representational) non-identity) triggering adnom inal intensification o f the sim ple reflexive sig. 3.3.2.S Complex reflexives and distributivity Com plex reflexives have som etim es been claim ed to disam biguate in favor o f distributive readings, see (90a) vs. (90b). (90) a. Soldaterne forsvarede sig. (collective/distributive) soldiers-the defended REFL ‘The soldiers defended them selves.’ b. Soldaterne forsvarede sig selv. (distributive) soldiers-the d efen d ed R E F L s e lf ‘T h e so ld iers d efen d ed th e m se lv e s.’ H o w ev er, in a c o n tra stiv e c o n te x t in w hich the soldiers as a group are contrasted with another group o f individuals, the collective reading is also possible with com plex reflexives, com pare (91 b) and (91 b). (91) a. (7)Soldaterne forsvarede sig, men overladte civil befolkningen til fjenden. soldiers-the defended REFL but left civil population to enem y-the ‘The soldiers defended but left the civilians to the enem y.’ b. Soldaterne forsvarede sig selv, men overladte civil befolkningen til fjenden. soldiers-the defended REFL self but left civil population to enem y-the ‘The soldiers defended them selves but left the civilians to the enem y.’ The fact that the sentence in (91a) is slightly odd w ithout the intensifier selv is due to the explicitly contrastive context w hich calls for intensification o f the reflexive. The sentence in (91b) clearly falsifies the claim that the com plex reflexive sig selv m ust have a distributive reading, thus refuting the alleged direct link betw een distributivity and intensification o f reflexives. Furtherm ore, it seem s that in m ost cases the choice between collective vs. distributive readings o f reflexives depends on factors other than intensification o f the reflexive. See, for R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 174 exam ple, the sentences in (92) w hich show that the distributive reading can be triggered by predicate meaning. (92) a. Soldaterne vaskede sig. (distributive) soldiers-the w ashed REFL ‘The soldiers washed (them selves).’ b. Soldaterne vaskede sig selv. (distributive) soldiers-the w ashed REFL self ‘The soldiers washed them selves.’ W hile (90a) the sim ple reflexive sig is readily interpreted as having a collective reading, in (92a) the default reading is the distributive reading. It seems obvious that this difference betw een (90a) and (92a) follows from the m eaning o f the predicates. Soldiers are per definition assum ed to w ork together under unified com m and to achieve a com m on goal, e.g. to defend or attack som eone or something. It is therefore not surprising that in (90a) with the predicate forsvare ‘defend’ sig is m ost readily given a collective reading. In contrast, the default choice o f a distributive reading o f sig in (92a) follows from the m eaning o f the groom ing verb vaske ‘w ash’. Under normal assum ptions, washing (by show ering or sponging) is a self-groom ing activity w hich is perform ed by each individual on his or her own body. As a consequence, the collective reading o f sig in (92a) is blocked by the relative difficulty o f im agining a plausible scenario in w hich the soldiers engage in a collective groom ing activity and as a result all get washed. Since the verbs used in (90-92) are all neutral, anti-reflexivity does not interfere with the intensification o f reflexives in these examples. The fact that the collective readings o f com plex reflexives are quite felicitous given the right context, see (91b), shows that, unlike what has been claim ed in the literature, intensification o f reflexives does not necessarily trigger distributive readings. Furtherm ore the fact that the choice betw een distributive and collective readings o f reflexives can be shown to vary from predicate to predicate, as in (90a) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 175 vs. (92a), provides additional evidence that intensification o f reflexives is not directly linked to distributivity, but depends on other factors, such as, for example, predicate meaning. 3.3.2.9 Summary In the previous sub-sections o f section 3.3.2 we have discussed the behavior o f locally bound reflexives occurring as internal argum ents o f verbal predicates. A ccording to our proposal that binding and intensification constitute separate m odules o f the gram m ar we predict the binding behavior o f reflexives sig to follow from principle A as form ulated in section 3.2, while intensification o f reflexives (i.e. the distribution o f the so-called com plex reflexive sig selv) should follow from the principles o f intensification, i.e. the principle o f contrastiveness as form ulated in chapter 2. We have show n that this approach to binding and intensification can successfully account for the distribution o f sim ple and com plex reflexives found in the contexts sum m arized in the descriptive generalizations in (93) and (94). (93) Sim ple reflexives fe.g. Dan, sis): a. N eutral predicates (section 3.3.2.3) b. Inherently reflexive predicates (section 3.3.2.2) (94) Com plex reflexives (e.g. Dan, sis selv\. a. anti-reflexive predicates (section 3.3.2.1) b. hidden neutral predicates (section 3.3.2.3) c. contrastive contexts (w ith neutral predicates) (section 3.3.2.5) d. doppelganger-effects (w ith neutral predicates): (section 3.3.2.7) (iv) statue-readings (section 3.3.2.7.1) (v) qua-sentences (section 3.3.2.7.2) (vi) strict (and sloppy) readings in VP ellipsis (section 3 3 2 .1 3 ) In the follow ing sections the distribution o f sim ple and com plex reflexives in resultatives (section 3.3.3), ECM constructions (section 3.3.4), and possessive constructions (section 3.3.5) will be argued to be am enable to the sam e type o f explanation. Section 3.3.6 presents a phonological account o f the distribution o f com plex reflexives in prepositional predicates w hich explains intensification o f reflexives w ith certain PPs a consequence o f the inability o f R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 176 simple reflexives (w hich behave phonologically like clitics) to carry stress on their own. Section 3.3.6 also contain discussion o f the role o f stressability o f reflexives with respect to adnom inal intensification in other languages (French, and German). 3.3.3 Reflexives in resultative constructions As show n in (95-98), resultatives can behave as either neutral (95), anti-reflexive (96), or inherently reflexive predicates (97-98), depending on the m eaning o f the words involved (cf. Veraart 1996). (95) N eutral resultative construction: a. Peter arbejdede sig rig. b. Peter arbejdede sine forceldre rige. c. Peter arbejdede sig selv rig. Peter w orked REFL/REFL self/his parents rich ‘Peter worked him self/his parents rich.’ (96) A nti-reflexive resultative construction: a. Peter drak #sig under bordet. b. Peter drak Hans under bordet. c. Peter drak sig selv under bordet. Peter drank REFL/REFL self/Flans under table-the ‘Peter drank him self/H ans under the table.’ (97)"“ Inherentiv reflexive resultative construction: a. Peter arbejdede sig svedig. b. Peter arbejdede #sine forceldre svedige. c. Peter arbejdede #sigselv svedig. Peter worked REFL/REFLself/his parents sweaty ‘Peter w orked him self/#his parents sweaty. Unlike its English counterpart, the Danish verb arbejde ‘work’ cannot be used as a causative with the meaning ‘make somebody work’, compare (i) and (ii). (i) Peter worked his employees pretty hard. (ii) * Peter arbjedede sine ansatte hardt. Peter worked POSSREFL employees hard The Danish translation of (i) would necessarily involve a periphrastic causative, e.g. Peter fik sine ansatte til at arbejde hardt ‘Peter got his employees to work hard.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 177 (98) Inherentiv reflexive resultative construction: a. Peter dansede sig til verdensmesterskabet. b. Peter dansede #sine forceldre til verdensmesterskabet. c. Peter dansede #sigselv til verdensmesterskabet. Peter danced REFL/REFLself/his parents to the world cham pionship ‘Peter danced him self/#his parents to the w orld cham pionship.’ N otice that the only difference betw een the sentences in (95) and (97) is the adjective denoting the end result o f Peter’s work, i.e. rig ‘rich’ vs. svedig ‘sw eaty’. In other words, simply replacing one with the other changes the w hole resulative from a neutral (95) to an inherently reflexive (97) or vice versa. Likewise, (96) could change from “anti-reflexive” to “neutral” sim ply by replacing the phrase under bordet ‘under the table’ w ith the fu ld ‘drunk’. In the analysis proposed here the fact that changing the adjective in (95) and (96) can change the behavior o f these resultative constructions can be explained as a change in the presuppositions triggered. The resultative construction drink somebody under the table in (96) is sem antically anti-reflexive (or ‘hidden’ neutral) because it triggers the presupposition that the reflexive is representationally non-identical to its antecedent Peter. By changing the phrase under the table to drunk the resultative construction no longer presupposes non-identity o f Peter and the reflexive. H ence the neutral behavior show n in (96). In other words (95) and (97) are structurally identical, but differ sem antically with respect to presupposed (representational) non-identity o f reflexive and antecedent. Unlike purely syntactic accounts o f binding, which have difficulties accounting for the different distribution o f simple and com plex reflexives in such cases, the account adnom inal intensification proposed here is sensitive to sem antic and pragm atic factors (i.e. presupposed representational non-identity) and, as a result, faces no such difficulties. U nless resultatives are analyzed as com plex predicates these exam ples constitute a serious challenge for predicate-centered theories o f binding like Reinhart & Reuland (1993). However, in the nominal account o f binding proposed here nothing hinges on w hether or not R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 178 resultatives are com plex predicates. Indeed, since binding and intensification are independent o f each other, binding theory has nothing to say about the distribution o f the intensifier selv ‘s e lf w hich falls under the contrastiveness principle discussed in chapter 2. 3.3.4 Reflexives in ECM constructions As shown in (99-100), ECM constructions can behave as either neutral, see (99), or anti reflexives predicates, see (100), depending on the m eaning o f the words involved. (99) N eutral ECM construction: a. Peter ansa sig fo r at vcere intelligent. b. Peter ansa sine forceldre fo r at vcere intelligente. c. Peter ansa sig selv fo r at vcere intelligent. Peter considered REFL/REFL self/his parents for to be intelligent ‘Peter considered him self/his parents to be intelligent.’ (100) ‘H idden’ utral ECM construction: a. Peter ansa #sig fo r at vcere dod. b. Peter ansa sine forceldre fo r at vcere dode. c. Peter ansa sig selv fo r at vcere dod. Peter considered REFL/REFL self/his parents for to be dead ‘Peter considered him self/his parents to be dead.’ C onsidering som ebody to be intelligent is a neutral activity w hich can be directed and oneself as well as others, see (99). In contrast, considering som ebody to be dead cannot be directed at oneself, at least not in the literal sense. The only way to save such a sentence would be to assum e the existence o f some kind o f doppelganger, e.g. Peter’s soul (liberated from his body at death), Peter’s ghost roam ing around after the physical death o f Peter, etc., who considers the physical Peter to be deceased. Hence (100) behaves as an anti-reflexive or ‘hidden’ neutral with respect to intensification o f reflexives. That is, the m eaning o f the ECM construction in (100) triggers the presupposition that Peter and the individual he considers to be dead are (representationally) non-identical. It is this presupposition, which is present in (100) but not in (99), w hich triggers adnom inal intensification o f the sim ple reflexive sig. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 179 O nce again, binding and intensification are independent o f each other. Binding o f sig falls under principle A w hich is satisfied in both (99) and (100) since the reflexive are bound by a subject (Peter) inside the m inimal tensed clause. The distribution o f selv ‘self on the other hand, falls under the contrastiveness principles o f the m odule o f intensification and is thus directly sensitive to the presuppositions o f identity triggered by the m eaning o f the words involved in the construction (e.g. intelligent ‘intelligent’, vs. d0d ‘dead’) and background encyclopedic knowledge'*^. Together with resultative constructions (discussed in section 3.3.3 above) and possessive construction (discussed in the following section), ECM constructions constitute potentially problem atic cases for predicate-based approaches to binding (e.g. Reinhart and Reuland (1993)). Since the notions o f anti-reflexivity, neutrality and inherent reflexivity invoked in the analysis proposed here do not rely on the notion o f co-argum enthood but rather on the presence/absence o f presuppositions o f (representational) (non-)identity o f reflexive and antecedent it does not run into these problem s. 3.3.5 Complex reflexives and pronouns in possessor position In this section we show that the distribution o f sim ple and com plex possessive reflexives, e.g. sin ‘PO SSR EFL’ and sin egen ‘PO SSR EFL’, follow the sam e general pattern as the distribution o f reflexives in argum ent position (i.e. sig and sig selv). That is, the binding “ '^Note that the same doppelganger-effect discussed in section 3.3.2 above can also be found in ECM-construction, see(i). (i) a. Peter sd sig/sig selv danse i spejlet. Peter saw REFL/REFL self dance in mirror-the ‘Peter saw himself dance in the mirror.’ b. Peter sd #sig/sig selv danse pd video-optagelsen. Peter saw REFL/REFL self dance on video-recording-the ‘Peter saw himself dance in the video-recording.’ According to Rooryck and van den Wyngaerd (1999), wherefrom these examples are adapted, the difference between (ia) which allow simple sig and (ib) which doesn’t should be explained in terms of time-slice identity. In our analysis the difference has to be due to representational non-identity. Somehow the contemporary reflection of a person in a mirror comes closer to full identity than the digital representation o f a video-recording (which thus trigger a presupposition o f representational non-identity). R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 180 behavior o f possessive reflexives (e.g. Dan. sin “PO SSR EFL’) are accounted for by principle A as form ulated in section 3.2, and the distribution o f the intensifier egen ‘ow n’, follow s the sam e principle o f contrastiveness as its suppletive variant selv ‘s e lf. O ne o f the main strengths o f the analysis proposed here is thus that it can be straightforw ardly extended to intensified possessive reflexives, which constitute a m ajor problem for m ost predieate-based accounts, e.g. Reinhart & Reuland (1993). The Danish system o f nominal expressions in possessor position is given in (101). Unlike English, Danish has a possessive reflexive, i.e. sin ‘PO SSR EFL/his/her/one’s ’, in addition to the possessive pronouns hans ‘h is’. (101)"*^ D anish nom inal expressions in possessor position: Sim ple/unintensified Com plex/intensified a. Reflexive sin ‘his/her/one’s’ sin egen ‘his/her/one’s ow n’ b. Pronoun hans ‘his’ hans egen ‘his ow n’ c. DP kongens ‘the king’s ’ kongens egen ‘the king’s ow n’ Except for the suppletive form o f the intensifier, see (103), and the different case form s (i.e. assum ing sin ‘P O SSR EFL’ to be the genitive form o f sig ‘R E FL ’) this system is exactly the sam e as the ones found in argum ent position, com pare (101) and (102). (102) D anish nom inal expressions in argum ent position: Sim ple/unintensified C om plex/intensified a. Reflexive sig ‘R E FL ’ sig selv ‘REFL self/him self b. Pronoun ham ‘him ’ ham selv ‘him him self/him self c. DP kongen ‘the king’ kongen selv ‘the king h im self ’ The full paradigm o f possessive reflexives and pronouns can be found in Appendix ii. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 181 (103)'*'* Suppletive variants o f intensifier: a. A rgum ent position: selv ‘h im self b. Possessor position: egen ‘ow n’ Like their argum ent position counterparts (i.e. sig ‘him, her, one’ and ham ‘him ’) the distribution o f sin and hans are constrained by the principles A and B o f the binding theory. Sin (like sig) is a reflexive and m ust be bound by a subject in a local dom ain and hans ‘his’ (like ham ‘him ’) is a pronoun and m ust be free in a local domain, i.e. the m inimal tensed clause, see (104a-d). (104) a. Peteri sagde at Johukvaskede sin*, h. *^/hanSi, * k z tegnebog. ‘Peter said that John washed P O SSR E FL /his w allet.’ b. Peter, sagde atJohn^ vaskede sin*, ^ * z egen / hans, * 4 % egen tegnebog. ‘Peter said that John vaskede PO SSREFL ow n /h is own w a lle t’ c. Peter, sagde at Johuk stjal sin*,. ? * k * z / hans,, * ,, z tegnebog. ‘Peter said that John stole PO SSR EFL/ his w a lle t’ d. Peter, sagde at John^ stjal sin*, k. *z egen / hans, *k *z egen tegnebog. ‘Peter said that John stole PO SSREFL own/ his own w allet.’ The sentences in (105) illustrate LD -binding o f the possessive reflexive sin follow the same pattern as LD -binding o f sig ‘R EFL’, i.e. its antecedent can be found outside infinitival clauses. (109) a. Peter, bad HanSk vaske sin,,k hund. Peter asked Hans w ash PO SSREFL dog ‘Petep asked HanSi to w ash hisi/k dog.’ The existence of synonymous words egenrisiko ‘own-risk’ and selvrisiko ‘self-risk’ can be construed as evidence for the hypothesis that they are suppletive variants. Even when used as a noun the words selv ‘self and egen ‘own’ can be found to be interchangeable, see (i) which is from J.P.Jacobsen (1880) Niels Lyhne, p. 65. (i) Hvilken forunderlig, forjcetningsfuld Tid var det ikke, hvor scelsomt ikke 0ren at hare sin Sjcels utydelige, londomsfulde Hvisken klinge frem i Virkelighedens Luft, som vildt udfordrende Lurtoner, som Brag a f Kalleslag paa Tempelmure. som Hvin a f Davidsstene paa Flugt mod Goliathspander og som sejerssikker Fanfare. Det var som at here sig selv tale i fremmede lunger, med fremmed Klarhed ogfremmed Magt om det, der var Ens dybeste, inderste eget. ‘ [....] It was like hearing oneself speak in foreign tongues, with foreign clarity and foreign power about that which was one’s deepest, innermost own.’ In the example in (i) the possessive intensifier eget ‘own’ is used in a context in which one might just as well tlnd the intensifier selv ‘self, see (ii) (ii) [...] der var Ens dybeste, inderste selv. [...] which was one’s deepest innermost self ‘[...] which was one’s deepest innermost self.’ Although, selv and egen are in overlapping distribution there are still subtle meaning differences. However, rather than taking the origin of these to be lexical, we assume them to be parasitic on the possessive relation itself. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 182 b. Peter lod forceldrene sove i sin egen seng mens han selv sov pd sofaen is stuen. Peter let parents-the sleep in PO SSREFL own bed while he self slept on sofa- the in living room -the ‘Peter let his parents sleep in his own bed while he him self slept on the sofa in the living room .’ Though LD -binding o f the simple possessive reflexive sin is m ore frequently found, LD- binding o f intensified possessive reflexives is by no m eans excluded, i.e. given an appropriate context that satisfies the contrastiveness principle o f adnom inal intensification, see (105b). That is the adjunction o f the intensifier egen ‘ow n’ does not affect the LD-potential o f the possessive reflexive sin. In other words, ju st like in the case o f sig, see section 3.3.1, intensification o f possessive reflexives does not directly affect locality constraints''^ The exam ples in (106-108) illustrate that D anish possessive reflexives sin and sin egen exhibit the sam e overall distributional pattern as the argum ent reflexives sig and sig selv with respect to “neutral”, “anti-reflexive” and “inherently reflexive” constructions/sentences'"’. The same independence of intensification and locality constraints can be found with the English possessives his and his own. Contrary to what has often been claimed, the complex possessive his own is not an anaphor which must be locally bound, but rather an intensified pronominal element which may be bound either locally or non- locally, see (i). (i) a. The Housing Associationi are encouraging people,, e, to buy their, , houses. b. The Housing Association, are encouraging people, e, to buy their own houses. c. The Housing developers were encouraging people, e, to buy their,,, own houses. ((a-b) from Quirk et al. (1985:363), (c) from Zribi-Hertz (1995:361)) The minimal pair (ia,b) has been interpreted as evidence in favor of assuming that adding own to the possessive pronoun gives rise to an anaphoric element which must be locally bound. However, as shown by Zribi-Hertz (1995), changing the matrix subject from housing association to housing developers gives rise to a sentence in which long-distance binding o f his own is pragmatically more acceptable more, see (ic). The importance o f this example in the context of this paper is that it provides further evidence in favor of the proposal that binding and intensification belong to different modules of the grammar. At first glance intensifier-adjunction may seem to have changed the possessive pronoun his to an anaphor his own. But when more attention is paid to the pragmatic context o f the sentences, it becomes clear that his own is not an anaphor but an intensified possessive pronoun which is decomposable into his and own (the suppletive form of the intensifier him self whose distribution is controlled by the principles o f the module o f intensification). The minimal pair (ib,c) also shows that the distribution o f the intensifier own is sensitive to semantic/pragmatic factors. See also chapter 5 for more discussion of the analysis o f possessives in Modern English. Examples involving L* person possessives can also be used to illustrate that intensifier adjunction is triggered by something else than a need for disambiguation, e.g.: (i) Je suis *mon/mon propre ennemi. T am my own enemy.’ (Fr.) (ii) / am *my/my own enemy. (iii) Jeg er *min/min egen fjende. T am my own enemy.’ (Dan.) (iv) Wo shi *wo / wo ziji de di-ren. T am my own enemy.’ (Ch.) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 183 (106) “N eutral” constructions: a. Peteri vasker sin /sin egen / John's tegnehog. Peter washes POSREF /PO SREF own/ John’s wallet. ‘Peter washes his /his ow n / John’s w allet.’ b. Peter hader sin /sin egen / Jo h n ’ s mor. Peter hates PO SREF /PO SREF ow n/ John’s mother. ‘Peter hates his / his own / John’s m other.’ (107) “A nti-reflexive” constructions: a. Peter er *sin /sin egen / John's jjende. Peter is *POSREF /PO SR EFL ow n/ John’s enemy. ‘Peter is his / his own / his enem y.’ O '* ’. P eterstjal *??sin / sin egen /J o h n ’ s tegnehog. Peter stole *??PO SREF/ POSSREF own / John’s wallet. ‘Peter stole his / his own / John’s w allet.’ (108) “ Inherently reflexive” constructions: a. Han var ved at gaa ud a f sit /?* sit eget/*Peters gode skind o f glcede. he was about to go out o f PO SR EF/?*PO SR EF ow n/Peter’s good skin o f happiness ‘He nearly jum ped out o f his good skin o f sheer happiness.’ h. Han hyttede sit / ??sit eget/*Peters skin. (cont. bet. a/b) he saved PO SREF/ ??PO SREF ow n/*Peter’s skin ‘He saved his ow n life.’ The im portance o f the above exam ples involving Danish reflexive possessives sin and sin egen is that they show that disam biguation is not the prim e m otivation for intensifier adjunction to possessives. In sentences like the ones in (104) w ith tw o potential antecedent for the possessive sin only the local antecedent can bind sin. These sentences (unlike the English transiations thereof) are not am biguous; sin can only have John (the local subject) as its antecedent since it is an anaphor w hich has to abide by principle A and hans, being a pronoun which has to be free in its local domain, can only have Peter as its antecedent. Since it cannot be the need to disam biguate betw een m ultiple potential antecedents w hich m otivates the presence o f egen ‘ow n’ on sin in the “anti-reflexive” constructions in (107) it has to be See chapter 4 for more discussion of P ‘ and 2 " * * person pronouns and reflexives. Note that adding the particle back saves the sentences in (ia), see (ib): (i) a. *Hej stole hisj wallet. (simple predicate: steal ) b. Hci stole hiSj wallet back. (complex predicate:^tea/_ _ back R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 184 som ething else'^*. As the contrast betw een (104a,b), (106) and (104c,d), (107) shows it is the sem antic/pragm atic m ake-up o f the sentence (i.e. the presupposed or expected (representational) non-identity o f the reflexive and its antecedent''^) w hich triggers intensifier- adjunction to sin. Based on the m eaning o f the predicates vaske ‘w ash’ and stjcele ‘steal’ and w orld knowledge, native speakers know that for a given individual x, the proposition x washes x ’ s wallet is pragm atically OK. The proposition x steals x ’ s wallet, however, is pragm atically odd; stealing is per definition an “anti-reflexive” activity. The m eaning o f the verb steal presupposes that the stealer does not originally possess the stealee. Indeed, anti-reflexivity, i.e. presupposed representational non-identity o f the stealer and the original ow ner o f the stolen goods, seems to be an integral part o f the m eaning o f the predicate stjcele ‘steal’, i.e. given normal circum stance the expectation is that people w ould not steal their own things. The exam ples in (108) show that D anish also have “inherently reflexive” possessive constructions (usually o f idiom atic nature) w hich disallow all but the simple unintensified reflexive possessive sin. We can therefore conclude that the distribution o f Danish sim ple and com plex possessives {sin and sin egen) in “neutral”, “anti-reflexive” and “inherently reflexive” predicates follow the sam e general pattern as the sim ple and com plex argum ent reflexives sig and sig selv. In this section we have shown that the sam e generalizations hold for both adnom inal intensification o f reflexives in argum ent position (e.g. sig selv ‘REFL s e lf ) and adnominal intensification o f reflexives in possessor position (e.g. sin egen ‘PO SSREFL ow n’). This constitutes a m ajor obstacle for analyses o f binding w hich are based on the notions o f co- The sentences in (I) thus indicate that some kind of analysis of resultatives as complex predicates is needed. For the example in (111b) and similar examples an structural i-within-i filter explanation has been proposed, see Vergnaud and Zubizarreta (1992). If reflexivity is defined as corefence between co-arguments, then the relation between sin and its antecedent John in the sentences in (104-108) is not strictly speaking one o f reflexivity. “Reflexivity” is here used more loosely to refer to self-directed predicates/constructions in a broader sense. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 185 argum enthood, e.g. Reinhart and Reuland (1993). It seems rather im plausible to m aintain that the difference betw een (106a) and (108b) should derive from the a lexical feature (e.g. Reinhart and R euland’s [+/-inherently reflexive], or Z ribi-H erts’s [+/- disjoint reference]) distinguishing betw een the predicate vaske ‘w ash’ and the predicate stjcele ‘steal’. The structural relationship betw een the reflexives (i.e. sin and sin egen) and their antecedent has to be assum ed to be the sam e all the sentences in (106-108). And yet, the sentences differ as to w hether they allow the sim ple possessive reflexive or not. W hile this differenee can be accounted for in term s o f focus and presuppositions as illustrated, it seem s im possible to account for it in term s o f any kind o f predicate-centered ‘reanalysis’ or ‘lexical am biguity’ account w hich only applies to co-argum ent binding^®. 3.3.6 Binding of simple and complex anaphors in PPs and prepositional predicates: phonological factors affecting intensification of reflexives In this section it will be shown that phonological factors - notably stress placem ent - interact with the distribution o f intensifier-adjunction to the sim ple reflexive sig. We argue that the focus-based analysis o f intensifier-adj unction proposed here stands a better chance o f giving a satisfying account o f the nature o f this interaction between phonology and syntax than com peting analyses o f com plex reflexives^'. The rest o f this section will be organized as follows. First, in section 3.3.6.1, the clitic-like properties o f the sim ple reflexive sig (e.g. the fact that it cannot be stressed) are illustrated u sin g a num ber o f syn tactic tests. T hen, in section 3 .3 .6 .2 , it is sh ow n that the It would potentially be interesting to test possessive reflexives for doppelganger-effects (e.g. statue-readings in wax-museum context, qua-sentences, etc.). We would expect sin vs. sin egen to display the same differences with respect to representational non-identity as sig and sig selv. Due to space limitations we leave this topic for future research. Cf. Siemund (2000), R e p r o d u c e d w ith p e r m is s io n of t h e c o p y rig h t o w n e r. F u r th e r re p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 1 8 6 idiosyncrasies o f the distribution o f sim ple and com plex reflexives in prepositional predicates follows from sig's status as a phonologically ‘light’ clitic (cf. Zwicky 1977). Finally in section 3.3.6.3 stressability o f reflexives will be discussed as a potential explanation o f cross-linguistic variation in intensification o f reflexives^^ 3.3.6.1. Stressability of reflexives: prosodic uses of selv = stress carrying element In this section a set o f syntactic tests are used to show that sim ple reflexive sig is a phonological ly ‘light’ clitic-like elem ent w hich cannot carry word-stress on its own and therefore requires intensifier-adj unction in order to be able carry stress like a full lexical DP. The different behaviors o f the sim ple and com plex reflexives in D anish are sum m arized in (109). Unlike the com plex reflexive anaphor sig selv, the simple reflexive anaphor sig is a clitic that cannot be stressed, used to answ er questions, coordinated, clefted, or topicalized. (109) C om paring sim ple sig and com plex sig selv. Tests \ expressions sig sig selv I D P ! h a m ! ham selv a. Stress N O YES (see (110)) b. A nsw er to questions N O YES (see (111)) c. Coordination N O YES (see (112)) d. C lefting N O YES (see (113)) e. Topicalization N O YES (see (114)) The sentences in (110)-( 114) illustrate the differences betw een sig and sig selv with respect to the properties in (109)” . (110) Stess: Peter had Plans om at vaske *SIG / SIG SEL V/ KONGEN. Peter asked Hans about to wash REFL/REFL SELF/ KlNG-the ‘Peter asked H ans to wash *REFL/REFL SELF/the K IN G .’ Cf. Siemund (2000). The tests in (109) and the examples in (110-4) are based on Jacubowicz (1994) who in turn was inspired by Kayne(1975). R e p r o d u c e d with p e r m is s io n of t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 187 (111) A nsw ering questions: Q: Hvem vasker Peter? A: *Sig / sig selv / Marie W ho w ashes Peter REFL/REFL self/M arry ‘W ho does Peter w ash?’ ‘*REFL / REFL SELF / M ary’ (112) Coordination: Peter vaskede barnet og *sig / sig selv / hunden. Peter w ashed child-the and REFL /REFL self /dog-the ‘Peter w ashed the child and *REFL / REFL SELF / the dog.’ (113) Clefts: D el var *sig / sig selv / hunden Peter vaskede. it was REFL / REFL self / dog-the Peter washed ‘It was *REFL / REFL SELF / the dog that Peter w ashed.’ (114) Topicalization: *Sig / sig selv / hunden vaskede Peter ikke. REFL / REFL self / dog-the w ashed Peter not ‘*REFL / REFL SELF / the dog Peter didn’t w ash.’ N otice that the verbal predicate vaske ‘w ash’ w hich is used in the above exam ples is a sem antically neutral predicate w hich allows the sim ple reflexive anaphor .sig as its internal argum ent, e.g. Peter vasker sig ‘Peter washes (him self)’. The m andatory presence o f selv in these exam ples thus cannot be due to any sem antically “anti-reflexive” nature o f the p red icate’"* . We therefore conclude that intensification o f sig in these cases m ust be prosodically m otivated, i.e. it m ust be due to a need to m ake the clitic sig heavy enough prosodically to receive stress. In this section, we have established that intensifier-adj unction to sig interacts with prosodic factors like stress. Being a phonologically ‘light’ clitic-like elem ent (cf. Zwicky 1977) sig cannot carry stress” . Therefore, intensifier-adj unction is necessary to make the The French reflexive soi ‘REFL’, which can both be stressed (i.e. SOf) and intensified (i.e soi-meme), illustrates that inability to be stressed is not a universal property of simple reflexives. German sicti and Russian sebja are also both stressable. In contrast, Russian ~sja and English 0-reflexives are both unstressable, just like Danish sig. ” There is one idiomatic expression which contains an instance of stressed, simple/unintensified sig, see (i). (i) Peter er iktee som alle de andre born. Han er nogetfor sig. Peter is not like all the other children he is something for REFL ‘Peter is not like all the other children. He is something special.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 188 reflexive behave like a full lexical DP. In the next section we propose that this susceptibility to phonological factors can be used to explain the distribution o f com plex reflexives in certain types o f prepositional predicates. 3.3.6.2. Intensifier-adjunction to reflexives in PPs and the interaction with sentential stress The m ain purpose o f this section is to show that in addition to sem antic factors (i.e. anti- reflexivity and inherent reflexivity defined in term s o f presupposition o f identity or (representational) non-identity), prosodic factors (i.e. stress-placem ent) also plays an im portant role in determ ining when reflexives m ust be adnom inally intensified. The claim is that a num ber o f exceptions to the sem antic account o f intensification o f reflexives can be explained as following tfom special prosodic properties o f certain prepositional constructions. D istributional ly, m ost prepositional predicates in Danish exclude the use o f the simple reflexive sig while allow ing intensified reflexives and DPs. As shown in (115-7) this generalization holds for prepositional verbs as well as prepositional nouns and adjectives. (115) Prepositional verbs: a. tale til ‘speak to ’ b. pege pa ‘point to /a f c. dromme om ‘dream about’ d. lytte til ‘listen to ’ e. sigte pa ‘aim a f f. Hans peger pa *sig /s ig selv / bilen. ‘Hans points at * 0 / 0 him self / the car.’ At this point we have no account o f this fact. Note, however, that some speakers prefer sig selv over stressed SIG in (i). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 189 a. et billede a f ‘a picture o f b. en beskrivelse a f ‘a description o f c. en hyldeste til ‘a tribute to, an hom age d. en bog om ‘a book about’ e. en evaluering a f ‘an evaluation o f f. . . . . en evaluering a f * sig/ sig s e lv / bilen. ‘. . . . an evaluation o f * 0 / 0 him self / the car. ’ Prenositional adiectives: a. sikker pa ‘sure o f b. interersseret i ‘interested in ’ c. optaget a f ‘occupied by ’ d. tilfreds med ‘satisfied w ith’ e. glad fo r ‘pleased with, like’ f. forskellig fra ‘different from ’ g Peter er tilfreds m ed * sig / sig selv / bilen. ‘ Peter is satisfied with * 0 / 0 him self / the car’ W hile prepositional predicates pattern w ith anti-reflexive predicates distributionally, i.e. with respect to intensifier-adjunction to the sim ple reflexive sig, they do not necessarily all have anti-reflexive m eanings w hich are logically incom patible (or even im plausible) w ith reflexive scenarios. Since D anish sig ‘R EFL’ cannot be stressed, as shown in section 3.3.6.1, we have to look for a language w ith stressable reflexives to see if the distinction betw een sem antically anti-reflexive and neutral predicates can hold for prepositional predicates. French is such a language, since it has the prosodically strong reflexive soi ‘R EFL’ w hich can host stress on its own. And as show n in (118) and (119), French reflexives in PPs may have both the intensified and the unintensified form depending on the m eaning o f the predicate, see (118) vs. (119). That is, French prepositional predicates can be classified as either anti-reflexive, neutral or inherently reflexive based on sem antic/pragm atic criteria. (118) Sem anticallv neutral prepositional predicates (French'): a. Personne n ’ est fier de soi/soi-meme/Marie. person N E is proud o f REFL/REFL-sam e/M arie ‘N obody is proud o f him self/M arie.’ b. Personne n ’ est content de soi/soi-meme/Marie. person N E is satisfied o f REFL/REFL-sam e/M arie ‘N obody is happy w ith him self/M arie.’ R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 190 (119) Sem anticallv anti-reflexive prepositional predicates (French): a. Personne n ’ est jaloux de *soi/soi-meme/Marie. person N E is jealous o f REFL/REFL-sam e/M arie ‘N obody is jealous o f him self/M arie.’ b. Personne ne bavarde avec ’ ^soi/soi-meme/Marie. person N E chats w ith REFL/REFL-sam e/M arie ‘N obody is chatting w ith him self/M arie.’ (120) Sem anticallv inherentlv reflexive prepositional predicates (French): Personne n ’ est hors de soi/*soi-meme/*Marie de fureur. person N E is out o f REFL/REFL-sam e/M arie o f fury ‘N obody is beside REFL/*REFL self /*M ary w ith rage.’ The predicates etre fier de and etre content de, in (118), are sem antically neutral in that they are com patible with both reflexive and non-reflexive scenarios. In contrast, the predicates etre jaloux de and havarder avec, in (119), are sem antically anti-reflexive in that they are com patible only w ith non-reflexive scenarios and thus trigger intensifier-adjunction to the sim ple reflexive soi ‘R E FL ’^^. Finally, the predicate in (120) presuppose identity o f its argum ents and is thus inherently reflexive. Since Danish also uses intensifier-adjunction to license reflexive scenarios with predicates w hose m eaning is logically incom patible (or pragm atically im plausible) with reflexive scenarios, as shown in section 3.3.2, one would assum e that sem antically neutral prepositional predicates corresponding to the French predicates in (118) would allow both simple and com plex reflexives. This prediction, however, does not hold. As shown in (121- 122), in D anish prepositional predicates, regardless o f meaning, need to have a com plex reflexive. (121) Sem anticallv neutral prepositional predicates: a. Peter er stolt a f \sig/sig selv/Marie. Peter is proud o f REFL/REFL self/M aire Peter is proud o f him self/M arie.’ b. Peter er tilfreds med *sig/sig selv/Marie. Peter is satisfied o f REFL/REFL self/M arie ‘Peter is happy w ith him self/M arie.’ This observation is from Zribi-Hertz (1995) who invokes a [+/-disjoint reference] feature to capture the differenee between PPs like fier de ‘proud o f vs. jaloux de ‘ jealous o f. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 191 (122) Sem anticallv anti-reflexive prepositional predicates a. Peter er misundeligpa *sig/sig selv/Marie. Peter is jealous on REFL/REFL self/M aire ‘Peter is Jealous o f him self/M arie.’ b. Peter sludrer m ed *sig/sig selv/Marie. Peter chats w ith REFL/REFL self/M arie ‘Peter is chatting w ith him self/M arie.’ M eaning-w ise one w ould expect the predicates vcere stolt a f ‘be proud o f and tilfreds med ‘satisfied w ith’ to behave like their French counterparts in (118), but in spite o f their “neutral” m eanings they still behave distributionally like anti-reflexives as shown in (I2 la ,b ). So if it is not the sem antic anti-reflexivity o f the predicates which triggers 5c/v-adjunction to sig in (I2 la ,b ), then w hat it it? We argue that it is prosodic factors. Sim ple/unintensified sig is a non-stressable elem ent w hich needs to cliticize to a prosodically stronger host to receive stress. The prepositional predicates in (115-117) all have unstressed prepositions. Since the language does not allow such unstressed prepositions to be followed by unstressable clitics, intensification o f sim ple reflexives is necessary to make the reflexive able to carry non- contrastive sentence stress w hich is required on the com plem ent o f such prepositional predicates. Hence, only sig selv and DPs are allowed w ith the prepositional predicates in (1 15- 117). A s m entioned above, m any o f the prepositional predicates in Danish are o f the type illustrated in (115-117), which, regardless o f their meaning, do not allow simple sig. There are, however, a num ber o f prepositional constructions w hich allow sim ple sig. They can be divided into those w hich only allow sim ple sig, and those which allow both sim ple and com plex reflexives as well as DPs. Let us first consider the prepositional expressions in (123- 124). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 192 (123)” Prepositional constructions allow ing only sim ple s is : a. at vcere bange a f sig to be afraid o f REFL ‘to be (naturally) tim id’ b. at have en revolver pa sig to have a gun on REFL ‘to carry a gun (with oneself)’ c. at bide smerten i sig to bite pain-the in REFL ‘to bear the pain w ithout flinching’ d. at sla fra sig to hit from REFL ‘to defend oneself, hit back, fight back’ e. at bide fra sig to bite from REFL ‘(fig) hit back, fight back, give as good as one gets’ f. at vcere om sig to be about REFL ‘be enterprising, be busy; have an eye for opportunities’ (124) Peter er bange a f sig / *sig selv / *Mary. Peter is afraid o f REFL / REFL self / M ary ‘Peter is (naturally) tim id’ The exam ple sentences in (124) illustrate how the expressions listed in (123) behave, i.e. allow ing sim ple sig, but neither intensified sig selv, nor DPs. The expressions in (129) differ from the prepositional predicates in (115-117) in several ways. First, the m eanings o f the predicates in (115) are clearly form ed eom positionally from the m eaning o f the verb plus the m eaning o f the preposition. Take for instance, tale til ‘speak to ’ which is straightforw ardly com posed o f the verb tale ‘speak’ plus the preposition til ‘to ’. As in G erm an and other related languages, m any o f the predicates o f the type illustrated in (115) have alternative forms where the preposition is fused with the verb, e.g. tiltale ‘address som eone, speak to som eone’^^ In contrast, the predicates in (123) are o f a m ore idiom atic nature. In m any cases, the m eaning o f Note that out of the 6 predicates in (123), only (123b) has a reflexive (i.e. oneself) in the English translation. In all the other cases the Danish reflexive sig ‘gets lost in translation’. The fact that sig does not need to be translated in these cases indicates that it is most likely non-thematic. See chapter 4 for more discussion of the distinction between thematic and non-thematic reflexives. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 193 the w hole expression cannot be deduced eom positionally from its com ponent parts, e.g. it is not im m ediately obvious how (123f) vcere ‘be’ + om ‘about’ comes to mean ‘be active; have an eye for opportunities’. In other cases, however, the m eaning is m ore transparent, e.g. (123b) have en reveler p a sig ‘have a gun on o n e se lf. Finally, none o f the expressions in (123) have variant form s o f the tiltale ‘to-speak, address’ vs. tale til ‘speak to, address’ type whieh, as discussed above, is frequently found w ith the expressions in (115). That is, com pounds such as *omvcere ‘about-be’, *frabide ‘from -bite’ are all impossible. In addition to prepositional constructions allow ing only sim ple sig, Danish also has a series o f prepositional expressions w hieh allow sim ple and eom plex reflexives as well as DPs, see (125). (125) Prepositional constructions w hich allow both sim ple and com plex reflexives as well as full lexical D Ps: a. Peter sd en slange bag sig / sig selv /Mary. Peter saw a snake behind REFL / REFL self /M ary ‘Peter saw a snake behind him self / M ary’ b. Peter sd en slange fo r an sig / sig selv /Mary. Peter saw a snake before REFL / REFL self /M ary ‘Peter saw a snake in front o f him self / M ary’ N otice that the prepositions found in (125) are different from both (115-7) and (123). First o f all, they are not part o f the verbal/adjectival/nom inal predicate but function as locative and directional adverbials. Second, the prepositions involved, e.g.foran ‘in front o f , hag ‘behind’, over ‘over’, etc., all assign theta-roles on their own. In contrast, in the predicates in (115-117) it is the verbal part o f the predicates w hieh assign theta-roles while the prepositions merely In theory it should be possible to find a particle verb like tiltale which behaves as (semantically and distributionally) neutral when combined but as (distributionally) anti-reflexive (while still semantically neutral) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 194 function as case-assigners. As for the expressions in (123) the them atic status o f sig is debatable^®. Even if considered to be them atic, the sim ple sig in (123) cannot plausibly be claim ed to received a theta-role form the verbal com ponent o f the expression. Hence, if the instances o f sig in (123) have a theta-role, it m ust come from the preposition itself. Finally, in addition to being m orphologically m ore com plex, the prepositions in (125) are all prosodically strong enough to carry w ord stress on their own, and, even m ore im portantly, have to be stressed. As diseussed below, this difference in stressability will turn out to be crucial. The distribution o f different types o f prepositions am ong the three distributional classes o f prepositional predicates is sum m arized in (126). (126) Prenositions constructions: Prenositions a. Tvne (115-1171: d /[-s tre ss ] t t T * sig / sig selv / DP O W [-sfress]‘ about’ /^ ^ [- s tr e s s ] O ri / r a [ - s ,r e s s ] ‘ f r o m ’ ^[-stress] ‘in’ ^ ^ ^ [ - s t r e s s ] ‘w ith’ fo r [.stress]'for\ etc. b. TvD e(123): ^ ^ T s tre s s ] sig / *sig selv / *DP n /W [+ stress] ‘about’ _ P ^ [+ stress] _ /r£ r[+ stress] ‘from ’ ^[+stress] ‘in’, etc. c. Type (125): ^ ttg [ + s tr e s s ] ‘behind’ s ig /s ig selv / DP fOTQyi [+stress] ‘ in front o f O V C r[+gt|-ess] ‘over’, etc. As shown in (126), the prepositions found in (115-117)/(126a) also appear in well-form ed constructions with the sim ple reflexive sig (123)/(126b). However, at closer inspection a very im portant difference between tbe constructions in (115-117) vs. the constructions in (123) becom es apparent. In the case o f the com plex predicates in (115-117), non-contrastive sentence stress always falls on the main verb/noun/adjective rather than the preposition, w hich when separated, due to prosodic factors. However, so far we have not been able to find an example. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 195 Is always unstressed, see (127a), w here stressed syllables are m arked in bold. In contrast, in tbe expressions in (123), it is tbe preposition w hich is unstressed, thus allow ing tbe reflexive to be unstressed, see (127b). (127) a. Peter pegede Peter pointed ‘Peter pointed b. Jeg havde en revolver I bad a gun ‘I was carrying a gun.’ p a mig. (type illustrated in (115-7)) on me at me. ’ pd mig. (type illustrated in (123)) on me Tbe reason why in (127) first person pronoun/reflexive mig ‘m e’ is used instead o f third person forms, is that mig ‘m e’ is stressable while sig ‘R EFL’ is not, thus m aking it possible to construct a paradigm as in (127) w here tbe only variable is stress placem ent. Since we know from section 3.3.6.1 that sig is not stressable and thus has to be adnom inally intensified in order to carry stress, we can now explain som e o f tbe differences betw een tbe prepositional constructions in (115-7) vs. (123) as follow ing from tbe differences in stress-patterns, see (128) and (129). Peter pegede pd (128) a. b. c. (129) a. b. c. Peter pointed on ‘Peter pointed at bim self/M arie.’ Peter havde en revolver pd Peter bad a gun on ‘Peter was carrying a gun.’ sig "'‘ sig selv "'‘ M arie REFL/REFL self/M arie "'‘ sig sig selv M arie REFL/REFL self Tbe unacceptability o f (128a) follows directly from tbe fact that sig cannot carry stress on its own. Hence intensification o f sig is necessary, as show n in (128b). Furtherm ore tbe fact that sim ple (unstressed) sig is ok in (129a) follows from tbe fact that in this case tbe preposition pd is stressed, and as a consequence, able to host tbe clitic sig. Cf. Vikner (1985) who claim that the sig in bange a f sig ‘be timid’ and similar “inherently reflexive’’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 196 W hile (128a-b) and (129a) can be explained by reference to tbe stress patterns, tbe unacceptability o f (129b) cannot. In this case, sig selv should be ok. There is no prosodic reason why sig selv should not be allow ed here. Based on w hat tbe expressions in (123b)/(129) m ean we suggest that it is their sem antically “inherently reflexive” nature (i.e. the presence o f a presupposition o f (representational) identity o f sig and its antecedent) which is responsible for the unacceptability o f (129b). That the expression in (129) is sem antically inherently reflexive is confirm ed by the unacceptability o f (129c). N otice that even with unstressed pa ‘o n ’ and stressed Marie, (129c) w ould still be unacceptable. Hence, the unacceptability o f (129b) follows from the Constrastiveness C ondition on adnom inal intensification. Based on the above exam ples we conclude that the presence/absence o f intensification o f sig in prepositional constructions is triggered by either sem antic factors, e.g. m andatory absence o f intensification in the “inherently reflexive” expressions in (123), (129b), and (131b), or prosodic factors (e.g. m andatory presence o f intensification in the distributionally, but not necessarily sem antically, “anti-reflexive” expressions in (115-7), (121), (128) and (130). (130) Sem anticallv neutral expression w ith stress-determ ined preposition (-stress'): a. Peter tog et billede * a f sig (prosodic) b. a f sig selv c. a f barnet Peter took a photo o f REFL self/child-the ‘Peter took a photo o f him self/the child.’ (131) Sem anticallv inherentlv reflexive expression: a. Peter er bange a f sig. b. *af sig selv (sem antic) c. * a f barnet (sem antic) Peter is afraid o f REFL ‘Peter is afraid o f him self/the child.’ prepositional expressions is non-thematic. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 197 In (130-131) stress is m arked in bold. B ehind each unacceptable exam ple the type o f violation is indicated in parenthesis. So far all o f the prepositional predicates involving the prepositions listed in (126a-b)“ w hich allow simple sig, i.e. (123) and (129), have been (or at least have m eanings w hich can plausibly be argued to be) sem antically inherently reflexive^' (i.e. they presuppose the (representational) identity o f the referents o f sig and its antecedent). There are how ever a small num ber (/at least one) sem antically neutral prepositional predicates involving the prepositions in (126a-b), w hich allow sim ple sig, in addition to sig selv and DPs, see (132) w here stress is m arked in bold and reasons for unacceptability indicated in parenthesis follow ing the example. (132) Sem anticallv neutral expression with stress-undeterm ined preposition^^: a. Peter tog tojet a f sig. b. *af sig (prosodic) c. * a f sig (prosodic) d. w sig (prosodic) e. *af sig selv (prosodic) f. a f sigselv g * a f sigselv (prosodic) h. *af sigselv (prosodic) i. *af barnet (prosodic) j- a f barnet k. * a f barnet (prosodic) 1 . *af barnet (prosodic) Peter took clothes o f REFL/REFL self/child-the ‘Peter undressed (him self)/the child.’ In (132) all the logically possible com binations o f stress and absence o f stress on both the preposition o f ‘o f and its com plem ent have been spelled out. Only, three com binations, i.e. (132a,f,j) are acceptable. The rest are unacceptable due to violations o f prosodic well- '’“ The prepositions listed in (126) also allow simple sig. But as already mentioned they differ from the ‘short’ prepositions in (126) in many other ways, as will therefore be treated separately below. Or, alternatively, to be deponent/unaccusative verbs which are constructed with non-thematic sig, see chapter 4, section 4.3. There appear to be very few examples of neutral constructions of this type, all involving sig as indirect object. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 198 form edness constraints^. The m ost interesting aspect o f these exam ples is the variable nature o f the preposition a / ‘of. W ith the sim ple reflexive sig, it has to be stressed, see (132a) vs. (132b-d). But with sig selv or a DP it has to be unstressed, see (132f,j) vs. (132e,g,b,i,k,l). W hether a f ‘o f is stressed or not depends on w hat type o f expression follows: a f ‘o f is stressed when follow ed by an unstresseable elem ent like sig, and stressed when follow ed by a stressed expression. Since both the com plex reflexive sig selv and DPs have w ord-stress, a f ‘o f is unstressed in front o f these expressions (132f,j). This generalization is further confirm ed by the behavior o f a / ‘o f in the expression in (132) when found with pronom inal com plem ents, see (133). (133) Sem anticallv neutral expression w ith stress-undeterm ined preposition: a. Sygeplejersken tog tojet a f ham. b. a f ham c. *af ham d. *af ham nurse-the took clothes o f him ‘The nurse undressed h im /him .’ The personal pronouns displays a dual behavior with respect to stress. On the one hand they can be unstressed (like sim ple sig). In this case, the preceding preposition a f ‘o f m ust be stressed, see (133a) vs. (133d). On the other hand, they can also carry stress, e.g. when contrastively focused. W hen that happens, the preposition a / ‘o f is stressless, see (133b) vs. (133c). Exam ples such as the above am ply illustrate the fact the absence/presence o f stress on the prepositions in (126a-b) vary depending on the following expression*’ '* . The judgments in (132) are based solely on non-contrastive sentence stress. That is, stress triggered by contrastive focus has not been taken into account. In (132g,i,k) stress on the preposition is ok if triggered by contrastive focus, e.g. (i): (i) Han tog tojet a f sig selv, ikke m ed sig selv. He took clothes-the o f REFL self not with REFL self ‘He undressed rather than bringing the clothes’ The observational that prepositions can be divided in to different groups based on the way they interact with stress is not new. See, for example, Allen et al. (1995:383-385): “As far as prepositional stress is concerned, the interaction between the preposition and its complement is of vital importance. Personal pronouns are usually unstressed when they function as prepositional complements, unless they receive contrastive stress. Generally speaking, prepositions are stressed when their complement is unstressed. Otherwise prepositions fall into two groups: R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 1 9 9 N ow let us return to the prepositions in (126c). As the exam ples in (134) illustrate these prepositions can also be found in neutral expressions, see also (125). (134)^^ N eutral predicate w ith m andatorilv stressed preposition: a. Han lagde uret bag sig. b. *bag sig c. *bag sig d. *bag sig (1) Group 1 consists of the following prepositions: ad, af, for, has, i, med, om, pa, til, ved. These are unstressed when their complement is stressed: Stressed Unstressed complement complement f r a ‘ fyrtarnet fra d e t ‘from the lighthouse’ i ‘ lommen ‘ iden ‘in the pocket’ med ‘ bornene ‘ med dem ‘with the children’ pd ‘ bordet ‘ pd del ‘on the table’ t i l ‘ dig 'til dig ‘to you’ (ii) Group II consists of the following prepositions: bag, efter, foran, forbi, for, (i)gennem, (i)mod, (i)mellem, inden, indtil, lungs, omkring, over (for), siden, uden, under. These are either stressed or unstressed when their complement is stressed: Stressed Unstressed complement complement ()efter ‘ valget ‘ efter del ‘after the election’ ( ‘ ) f o r ‘ krigen ‘ fo r den ‘before the war’ ( ) la n g s ‘ vejen ‘ lungs den ‘along the road’ ( ‘ )over ‘ doren ‘ over den ‘over the door’ ( ‘ )u d e n ‘ dem 'uden dem ‘without them’ (c) The distance of the complement from the preposition also plays a role for prepositional stress, insofar as a preposition will have at least secondary stress if it is not immediately followed by the complement. The preposition may be separated from its complement in two ways: (i) Something may intervene between the preposition and the complement, e.g. a coordinated verb phrase whose object is also the prepositional complement: Jeg ledte ‘ efter ogfandt ogsd et egetrcesbord. ‘I looked for and found and oak table.’ (ii) The prepositional complementary be fronted, leaving the preposition ‘stranded’ in final position. This is particularly common when the complement is a pronoun: ‘ Den bog har jeg ikke hort om. ‘I haven’t heard of that book.’ ‘ Ham kan vi ikke regne med. ‘We can’t count on him.’ ‘ Hvad lytter du til? ‘ What are you listening to?’ ‘ Del er hende, som jeg serpd. ‘She is the one I’m looking at.’ fetc...l” As expected, the pronoun ham ‘him’ can be both stressed or unstressed after stressed, prosodically heavy (i.e. bi- or poly-moraic) prepositions, see (i). (i) Neutral expression with mandatorilv stressed preposition: a. Hun lagde uret bag ham. b. *bag ham c. bag ham d. *bag hum she put watch-the behind him ‘She put the watch behind him/him.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 0 0 e. (*)bag sig selv f. *bag sig selv g bag sig selv h. *bag sig selv i. *bag barnet j- *bag barnet k. bag barnet 1 . *bag barnet he put watch-the behind REFL/REFL self/child-the ‘The put the watch behind him self/the child.’ The exam ples in (135) illustrate these prepositions can also be found with inherently reflexive expressions. (135)“ Sem antically inherently reflexive construction with m andatorily stressed preposition: a. Hun kunne ikke Icegge det bag sig. b. ^bag sig selv. c. ^bag barnet. she could not put it behind REFL/REFL self/child-the ‘She couldn’t leave it behind/she couldn’t put it behind her(self).’ The unacceptability o f (135b-c) m ust be due to the inherently reflexive sem antics o f the expression Icegge noget b a g ‘put som ething b eh in d ’ rather than the prosodic properties o f the preposition bag ‘behind’. U nlike the preposition a f'o ? in the neutral prepositional expression tage tojet a f ___ ‘take clothes-the o f ’ illustrated in (132), the preposition bag ‘behind’ in the neutral expression Icegge uret bag ___ ‘put watch-the behind ’ in (134) and the inherently reflexive predicate in (135) cannot be com pletely destressed, see (134f,j) vs. (132f,j). We attribute this difference betw een the prepositions in (126a-b) and (126c) to a difference in syllabic/m oraic structure. The prepositions in (126a-b) are all m ono-syllabic expressions with the syllable structure (C)(C)V, w hich m eans that they are all mono-m oraic. In contrast, the Notice that the English expression put something beh in d (with the meaning ‘forget’) is also inherently reflexive, see (i); (i) Peter; couldn’ t put it behind him/*him/*himself/*Mary. See chapter 5 for more discussion of such expressions in English, R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 0 1 prepositions in (126c) are bi- or even poly-syllabic w ith the minimal structure (C)VC(V), which m eans that they are bi- or poly-m oraic, see (136). (136) Stress-underspecified prepositions vs. a. Syllabic structure CV b. Moras 1 c. Qualify as phonological word NO d. Examples (126a-b) e. Stress - can be stressed or unstressed depending on the following expression - are always unstressed with prepositional predicates of type (137(1)) Lexically stressed prepositions (C)VC(V) 2 or more YES (126c) - always have some degree of stress regardless o f what follows The behavior o f different types o f prepositional predicates discussed above is sum m arized i m the table in (137). (137) Type 1(a) Prenositional nredicates: (i) (ii) (iii) (iv) Semantics th-role fr. th-role fr. prepos. ofpredic. v./adj/n. prep. +/-stress anti-ref. + (v) distribution o f selv *sig/sig selv/DP (vi) explanation of [+l-selv] prosody (/semantic) (vii) prepos. afpa (viii) example(s) (122) 1(b) n/hidd.n. + - - *sig/sig selv/DP prosody a f pa (121) 1(c) inh. ref. (-/+) - *sig/sig selv/*DP prosody o f (138) 1 1 (a) inh (+th) - inh (-th) - + undersp. undersp. sig/*sig selv/* DP sig/*sig selv/*DP semantic semantic afpa afpa (123b) (123a,c-t) 11(b) neut. + undersp. sig/sig selv/DP semantic af (132) 1 1 1 (a) neut. Ill (b) inh. ref. + sig/sig selv/DP semantic bag. foran (134) + sig/*sigseh/*D P semantic bag, foran (\35) As shown in (137) we divide prepositional predicates into three types depending on w hether or not the preposition is stressed. Predicates o f type I, e.g. (115-7), have prepositions which are always unstressed'^’. Sem antically these predicates can be either anti-reflexive (sub-type la), neutral (sub-type lb), or inherently reflexive (sub-type Ic). Exam ples o f anti-reflexive and neutral predicates o f this types were given in (122) and (121) respectively. Inherently reflexive prepositional constructions o f type (137(lc)) are illustrated in (138). The reason why the prepositions af, pa, etc. ‘of, on, etc.’ are always unstressed in I(a-c) but variable in Il(a-b) is still not entirely clear. At this point we simply assume that the non-contrastive stress pattern of the expressions of type I require unstressed af, pa, etc. ‘o f on, etc.’. We leave the question why this should be the case for further research. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 0 2 (138) Peter er ude a f *sig/sig selv/*Mary a f glcede. Peter is out o f R EFL/REFL self/M ary o f happiness ‘Peter is beside him self/*M ary with happiness.’ The French exam ple (due to Zribi-H ertz (1995)) in (139) (~(120)), shows that the stressable reflexive soi ‘R E FL ’ can, and m ust (due to sem antic constraints, i.e. m eaning o f predicate plus contrastiveness condition on adnom inal intensification) be unintensified in the sem antically inherently reflexive construction etre hors d e de jo ie ‘be b e sid e w ith happiness’. (139) Pierre est hors soi/*soi-meme/*Marie de joie. (see also (126)) Pierre is beside R EFL/REFL-sam e/M arie o f joy ‘Pierre is beside him self/*M arie w ith happiness.’ As indicated in (137) colum n (vi) the presence selv is m andatory w ith neutral and even with inherently reflexive predicates. These exam ples thus clearly show that prosodic factors can override sem antic constraints. In (138) the Contrastiveness Condition on adnom inal intensification is overridden by the prosodic constraint barring unstressed a f ‘o f to be follow ed by an unstressed clitic (i.e. sig). Since in these expressions the preposition (e.g. a f p d ‘of, on’) is m andatorily unstressed, the reason for the presence o f selv with sem antically neutral and inherently reflexive predicates is due purely to prosodic factors^*. Predicates o f type II have prepositions with variable stress, see (123) and (132). Sem antically these predicates are either inherently reflexive (sub-type 11a) or neutral (sub-type 11b). In contrast to predicates o f type I, the presence/absence o f selv on sig can be explained purely in term s o f sem antic factors, see (137) colum n (vi). As shown by the exam ples in (132), in these predicates the preposition is underspecified with respect to stress and varies according to the follow ing expression. If the expression follow ing the preposition is stressed then the preposition is unstressed and vice versa. Hence it is ultim ately sem antic factors (i.e. the Cf. Veraart (1996) who proposes an analysis o f Dutch reflexives which includes the assumption that prosodic factors can affect the choice between complex and simple reflexives. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 203 sem antic difference betw een neutral and inherently reflexive predicates + the Contrastiveness Condition on adnom inal intensification) w hich determ ine the distribution o f sig and sig selv. Type III predicates have prepositions w hich always have som e degree o f word stress. In these cases, it is sem antic factors (i.e. the sem antic difference betw een neutral and inherently reflexive predicates + the Contrastiveness C ondition on adnom inal intensification) which straigthforw ardly determ ine the distribution o f sig and sig selv. 3.3.6.3 Can intensification of reflexives be reduced to lack of stressability? In this section the question w hether adnom inal intensifieation o f reflexives can be reduced to stressability is discussed. First in section 3.3.6.3.1 and 3.3.6.3.2 the sim ilarities between stressed and intensified reflexives in French and G erm an are discussed. T hen in section 3.3.6.3.3 the stressability-properties o f pronouns and reflexives in Danish are contrasted. Finally in section 3.3.6.3.4 tw o different hypotheses concerning the relationship between stressability and intensification are discussed. 3.3.6.3.1 Stressable reflexives in French The exam ples in (140-142) show that the intensified reflexive soi-meme ‘R E F L -self and the stressed reflexives SO I ‘R E FL ’ behave sim ilarly in that they are both allow ed in anti-reflexive constructions like etre jaloux de ‘be jealous o f (141b,c) and excluded in inherently reflexive constructions like etre hors de “ be out o f ’ (142b,c). (140) a. Personne n ’ est fie r de °'^soi b. “ '"soi-meme c. “ ^SOI d. “ '"Marie. ‘N obody is proud o f him self/M arie.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 204 (141) a. Personne n ’ est jaloux de "soi b. "'"soi-meme c. "'"SOI d. "'"Marie. ‘N obody is jealous o f him self/M arie.’ (142) a. Personne n ’ est hors de "'"soi b. 'soi-meme c. "SOI d. "Marie de fureur. ‘N obody is beside REFL/*REFL self /*M ary with rage.’ As already discussed above, the sim ilar behavior o f soi-meme and SO I follow from the contrastiveness principle, as form ulated in chapter 2. Both intensification and focus (realized as stress, e.g. (140-2c)) require the possibility o f creating a contrast set o f alternatives. Since this is excluded by tbe sem antics o f inherently reflexive constructions both intensified and stressed reflexives are excluded in (142). Conversely, since both focus and intensification trigger the generation o f a set o f alternatives (and thus allow doppelganger-readings o f reflexives, i.e. (representational) non-identity o f the referents o f the reflexive and its antecedent) they can both license reflexives in anti-reflexive constructions like (141). Note, however, that although intensified and stressed reflexives behave sim ilarly there still is a subtle difference betw een them, viz. generally soi-meme is usually preferred over SO I in sentences like (141). 3.3.6.3.2 Stressable reflexives in German Like soi ‘R EEL’ in French, in G erm an the sim ple reflexive sick ‘R EFL’ is also stressable and, as predicted by the theory o f intensification proposed here, the behavior o f stressed reflexive (e.g. SICH) is sim ilar to that o f intensified reflexives (e.g. sick selbst), see (143-145). R e p r o d u c e d wifh p e r m is s io n o f fh e c o p y rig h f o w n e r. F u r f h e r re p ro d u c fio n p ro h ib ife d w ifh o u f p e r m is s io n . 205 (143) a. Peter ist stoltz a u f "''sich b. "'‘ sich selbst c. "'‘ SICH d. "'‘ Marie ‘Peter is proud o f him self/M arie.’ (144) a. Peter ist elfersUchtig a u f “ sich. b. "'‘ sich selbst c. "'‘ SICH d. "'‘ Marie ‘Peter is jealous o f him self/M arie.’ (145) a. Peter schamt "'‘ sich b. sich selbst c. ‘SICH d. ‘Marie ‘ Peter ashamed. ’ The exam ples in (143-5) show that the intensified reflexive sich selbst and the stressed reflexives SIC H behave sim ilarly in that they are allow ed in anti-reflexive constructions like a u f eifersiichtig sein ‘be jealous o f (144), and neutral constructions like stoltz ‘proud o f (143), and excluded in inherently reflexive constructions like sich schdmen ‘ be asham ed’ (145). N ote that although intensified and stressed reflexives behave sim ilarly there still is a subtle difference betw een them, viz. sich selbst is better than SICH, ju st like, in French, soi- meme is generally preferred over SOI. 3.3.6.3.3 Stressable pronouns in Danish A s show n in section 3.3.6.1 sim ple sig is a syntactically independent but prosodically dependent elem ent w hich cannot host stress on its own, see (146a(iii)). In this respect, it contrasts with Germ an sich and French soi, both o f w hich are stressable, see (146b,c). Notice, however, that Danish object pronouns are stressable, see (146d), ju st like G erm an and French reflexives. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 206 (146) (i) unstressed, unintensified (ii) intensified (iii) stressed a. Dan. sig vs. sig selv *SIG b. Fr. soi/se vs. soi-meme SO I c. Ger. sich vs. sich selbst ~ SICH d. Dan. ham vs. ham selv ~ H AM The sim ilarities o f stressed and intensified pronouns (e.g. Dan. H AM vs. ham selv) will be discussed further in sections 3.4.2-2. For now suffice it to say that, in certain contexts, stress and intensification appear to have a certain degree o f functional equivalence. 3.3.6.3.4 Reducing intensification of reflexives to unstressability Taken together, the functional equivalence o f stress and intensification in G erm an (e.g. SICH ~ sich selbst), French (e.g. SO I ~ soi-meme) and Danish (e.g. H AM ~ ham selv), see sections 3.3.6.3.1-3, and the unstressability o f sim ple sig, see section 3.3.6.1, may lead one to the hypothesis that intensification o f reflexives is A LW A Y S prosodically m otivated, see (147). (147) Hvpothesis I: Onlv one function o f intensification stress-carrier elem ent (required to prosodically strengthen unstressable clitics) Basically w hat hypothesis I says is that adnom inal intensification o f sim ple sig is always m otivated by stress. The reason why stress m ust fall on sig may vary, as shown in (148a-c), but in all cases intensification o f sig takes place in order to make and unstressable simple reflexive stressable (as m arked in bold in (148)). (148) Stress on si2 is triggered bv: a. Sem antics (representational non-identity => focus => stress => intensification (cf. (94a,b,d) in section 3.3.2.9)™ As mentioned in sections 3.3.6.3,1-2, stressed and intensified reflexives do differ in many respects. In certain contexts intensified forms are more felicitous than stressed forms, and in other contexts the opposite is the case. The fact that there are cases in which intensified are preferred over stressed reflexives, and vice verse, should not be ignored. It would be interesting to investigate the differences between stressed and intensified reflexives in more detail. If Hypothesis 1 is correct, then (148a) leads to the prediction that stressed forms of French soi, and German sich should be acceptable in statue-sentences in wax-museum contexts, see (i-ii): (i) Bill Clintorii est fier de SOI<statue>. Bill Clinton is proud of RELF (ii) Bill Clintorij razlert SICHi<statue>. Bill Clinton shaves REFL R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 207 b. Contrastive focus => stress => intensification (cf. (94c) in section 3.3.2.9, section 3.3.6.1)) c. Prosody (non-contrastive sentence stress) => stress => intensification (section 3.3.6.2) W hile hypothesis I may w ork for reflexives, it is not obvious that it can be extended to both pronouns and DPs. With pronouns and DPs we find both stressed and intensified forms, see (149-150). (149) a. H AN kom til modet. HE cam e to m eeting-the ‘HE cam e to the m eeting.’ b. Han selv kom til modet. He self came to m eeting-the ‘ He him self cam e to the meeting. ’ (150) a. KONGEN kom til modet. K lN G -the cam e to m eeting-the ‘THE KIN G cam e to the m eeting.’ b. Kongen selv kom til modet. K ing self cam e to m eeting-the ‘The king him self cam e to the m eeting.’ As show n in chapter 2, the sem antics o f intensification and focus are similar, i.e. both involve the generation o f a contrast set o f alternatives (which, depending on the context, m ay or may not be scalarly ordered). It is thus not surprising that in m any cases they can be found in the sam e contexts. However, the mere fact that certain languages allow for the existence o f both intensified and focused expressions is a strong indication that there m ust be som e sem antic/functional difference betw een the two. Due to space lim itations we leave the investigation o f this topic to further research. However, if differences betw een focus and intensification are found, w hich seems to be likely, then Hypothesis I m ust be revised. A possible alternative hypothesis is form ulated in (151). (151) Hypothesis 1 1: Tw o different functions o f intensification a. Contrastiveness/representational non-identity(cf.(94a-d) in section 3.3.2.9)) b. Stress-carrier elem ent (required to prosodically strengthen clitics, (121), (138)) Due to space limitations we leave this topic for further research. R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 208 Hypothesis II assum es tw o different functions o f selv. (i) a contrastiveness m arker which falls under the contrastiveness condition on adnorainal intensification (151a), (ii) a stress-carrier elem ent w hich is required solely to enable unstressable reflexives to host stress (which is required by non-contrastive sentence stress) (151b). W hile less elegant and less econom ical than H ypothesis 1 , H ypothesis II has the advantage o f not predicting focus and intensification to be sem antically and functionally identical (a prediction which is probably too strong). Furtherm ore, the existence o f exam ples like (138) clearly shows that, in certain cases, prosodic constraints can outw eigh sem antic constraints’'. Ultim ately, the choice betw een Hypothesis 1 and II is, o f course, extrem ely im portant. H ow ever, since it does not both are com patible with the analysis o f intensification and binding proposed here we leave this topic for father research. 3.3.7 Summary The im portant lesson to be draw from sections 3.3.1-6 is that binding and intensification are independent, see (152). (158) Independence o f binding and intensification: a. Independence o f locality and intensification (sect. 3.3.1) b. Predicate-m eaning and intensification (sect. 3.3.2.1-4) c. D oppelganger-effects and intensification (sect. 3.3.2.7) d. Resultative constructions (sect. 3.3.3) e. ECM -constructions (sect. 3.3.4) f. Possessive reflexives (sect. 3.3.5) g- Reflexives in prepositional predicates and stressability (sect. 3.3.6) Binding o f reflexives (i.e. the distribution o f (them atic) sig) falls under principle A as form ulated in section 3.2. T hat is, the distribution o f the reflexive anaphor sig can be explained by a Pica-style account based on LF m ovem ent o f the X° reflexive to INFL/T. As shown such an analysis explains both the LD -behavior and subject-orientation o f sig. Independently thereof the distribution o f the adnom inal intensifier selv ‘s e lf and its suppletive See also discussion of inherently retfexive verbs with intensified sig selv in chapter 4, section 4.3.3. R e p r o d u c e d with p e r m is s io n of t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 209 variant egen ‘ow n’ can be shown to be determ ined by the non-binding related sem antic, pragm atic, and prosodic factors. Furtherm ore, as show n in sections 3.3.3, 3.3.4, and 3.3.5. the nom inal approach to binding adopted here has no difficulties w ith non-coargum ent bound reflexives (e.g. in resultative, ECM and possessive constructions) while m any predicate- centered approaches flounder on such cases. In section 3.4, the consequences o f applying the same approach to binding and intensification o f sim ple and com plex pronouns are discussed. 3.4 Distribution of simple and complex pronouns In this section we propose an account o f sim ple and com plex pronouns in Danish based on the assum ption that com plex pronouns, e.g. ham selv ‘him self/him h im self, are the intensified versions o f the sim ple pronouns, e.g. ham. But before we proceed to the discussion o f com plex pronouns, we need to determ ine the principles accounting for the distribution o f sim ple pronouns, e.g. ham ‘him. This will be done in section 3.4.1 w here we argue that a syntactic approach to the distribution o f pronouns based on principle B provides the best account. Then in section 3.4.2 and 3.4.3 the special properties o f intensified pronouns (e.g. ham selv ‘him s e lf ) in object and subject position respectively will be discussed in more detail. It will be show n that the differences between simple and intensified pronouns are explained as follow ing from general properties o f adnom inal intensification. T hat is, the difference betw een ham and ham selv should not be explained w ithin binding theory but rather with the m odule o f intensification. Section 3.4.4 contains a b rief discussion o f the notion o f logophoricity which has often been adduced to account for the behavior o f intensified pronouns (as well as non-locally bound reflexives). Finally, section 3.4.5. concludes the discussion o f pronouns and sums up the chapter. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 1 0 3.4.1 Binding of pronouns: principle B As m entioned above, we argue that the binding o f sim ple pronouns should be accounted for by purely syntactic principles, see (153) and (154). (153) Binding Principle B : A pronoun is free in a local domain. (154) Binding D om ain: alpha is a binding dom ain for beta if and only if alpha is the m inimal category (i.e. the sm allest DP or IP/S) containing beta, a case-licensor or beta, and a SUBJECT accessible to beta. Assum ing that the relevant local dom ain is defined as in (154) in Danish, B principle in (153) accounts for the distribution o f pronouns, see (155). (155) a. [Peter, barberede ham*i o\. Peter shave-PA ST him ‘Peter shaved him»j/o.’ b. [Peteri bad mig^ barbere hami * z]- Peter ask-PA ST me shave-IN F him ‘Peter asked me to shave him, -z.’ c. [Peter/ bad JargeUz barbere ham, *z\ Peter ask-PA ST Jorgen shave-INF him ‘Petep asked Jorgenz to shave him,/»z.’ d. [Peter/ sagde [at je g barberede hami\\. Peter; say-PA ST that 1 shave-PA ST him ‘Peter said that I shaved him .’ Exam ple (155a) illustrates that ham ‘h im ’ cannot be locally bound. It also shows that ham ‘him ’ m ust have phi-features since it can be used deictically. In both these respects it differs from the reflexive sig (see section 3.2). Sentences (155b) and (155c) show that ham can be LD -bound by a non-local intra-sentential antecedent, a property it has in com m on w ith the simple reflexive sig ‘R EFT’. However, (155d) shows that, unlike sig, ham can be bound by an antecedent outside the tensed clause. This leads us to the descriptive generalization in (156). (156) Descriptive generalization 11: ham m ust be free inside the m inim al tensed clause R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 1 The sentences in (157) and (158), w hich contain what appear to be instances o f pronom inals bound inside their binding dom ains, indicate that we m ay have to modify the generalization in (156). (157)^^ a. Peter ifortalte Martin^om ham^selv. b. #ham^. c. *sig^ (selv). Peter told M artin about him self/him /REFL self (158) a. PeteVi gav M artinfs maleri til ham^selv. b. Uliam^. c. *sigz (selv). Peter gave M artin’s painting to him self/him /REFL self W hat (157) and (158) show is that pronouns can have an antecedent inside their binding dom ains as long as it is not a subject. This leads us to the new generalization in (159). (159)” Descriptive generalization II (revised version): ham m ust be free from binding by a subject inside the m inimal tensed clause W hile the distribution o f pronouns can be accounted for syntactically w ithin binding theory (i.e. principle B), the presence/absence o f se/v-intensification o f pronouns, e.g. (157a)-(158a) vs. (157b)-(158b), has to be explained by reference to the principles controlling adnom inal intensification. In the case o f exam ples (157-8), a sem antic account naturally presents itself. Under normal circum stances one does not give som ebody inform ation about him self, since everybody is norm ally assum ed to be well inform ed about them selves” . In the follow ing It would be interesting to test to see if stressed ///IM ‘HIM’ might fare better in these sentences. If stressed HAM and intensified ham selv are interchangeable in certain contexts, as discussed in section 3.3.6.3.3-4, then one would expect stressed HAM to be acceptable, or at least better than unstressed, unintensified ham in these examples. As already shown in section 3.3.5, example (104), his generalization also holds for possessive pronouns, e.g. hans ‘his’ and hans egen ‘his own’. The fact, that pronouns cannot be bound by a subject has been referred to as “subject-obviation”, cf. Huang (2000:24, note 5): “(some forms of) possessive pronouns in languages like Danish, Norwegian, and Icelandic seem to be subject-obviative, that is, they must be free from the closest potential subject in a specific syntactic domain.’’ (I) a. Johnjortalte Oku om sin/ / *hansi kone. b. JohHifortalte Oleu om *sinu / hansu kone. John told Ole about POSSREFL his wife R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 1 2 section we will discuss m ore evidence show ing that the so-called com plex pronouns, e.g. ham selv ‘him self/him h im se lf, are in fact nothing but adnom inally intensified version o f their simple counterparts. 3.4.2 Intensified object pronouns. If binding and intensification constitute separate m odules o f the gram m ar, as claim ed in this dissertation, then adjunction o f the adnom inal intensifier selv ‘s e lf to sim ple pronouns should follow from the principles o f intensification. The exam ples in (160-1) appear to confirm this prediction. (160)^^ a. Jon, vil giftes med en kvinde som er stolt o f hamj / '^haMi selv / ^sig, / *sigi selv^^’ ‘Jon w ants to m arry (passive) w ith a wom an who is proud o f .’ b. Kongen troede at ingen kunne tide hamj / *hami selv / *sig, / *sigi selv. king-the thought that nobody could like him. [=(2e) in section 1 above] ‘The king thought that nobody liked him .’ (161) a. JoHi plejede at hade fo lk som var anderledes end ?hami / hami selv / *sigi / *sigi selv. ‘Jon used to hate people who were different fro m .’ Sentences containing (intensified) pronouns with local non-subject antecedents are very rare, cf. Olsen (1992) quoted in Jakubowicz (1994). In the entire novel Niels Lyhne by J.P. Jacobsen (1880) we only found two examples, both of which are of the same “anti-reflexive” nature as (i) and (ii). (Page numbers from Gyldendal’s 1980 edition). (i) [...] Nielsjfors0gte [...] at gore Frithiof komisk fo r ham^ selv [...]. (1980:86) Niels tried to make Frithiof comical for him self ‘Niels tried to make Frithiofj comical to himself^.’ (ii) Hans Kjcerlighed, var stoerk som hendes, men den mangled den fine, mandige 0mhed, der vcerner den elskede Kvinde^ mod hende^ selv og vaager over hendes^ Vcerdighed. (1980:149) His love was strong as hers but it lacked the fine masculine tenderness that guards the beloved woman against her self and protects her dignity ‘His love was strong as hers but it] lacked the fine masculine tenderness thatj guards the beloved woman^ against herself^ and protects her^ dignity.’ Although we have yet to find any convincing cases o f semantically “neutral” or “inherently reflexive” constructions with non-subject-bound, local intensified pronouns, we cannot exclude that such examples might exist. If this is the case, then they may constitute a potential problem for the account o f complex pronouns presented here. The examples in (160a) and (161a,c,d) are from Safir (M.S.), p. 17-18, (35a,b,d,e). The reflexives sig and sig selv are included here only for completeness and will not be discussed below. The ungrammaticality of both sig and sig selv. with the indices indicated, in all the examples in (160) and (161) follows directly from binding theory and has nothing to do with the module o f intensification. As shown in section 3.2.4, simple sig and complex sig selv are both anaphors which must be bound inside the tensed clause. Since that condition is not met in the sentences in (160-161), both sig and sig selv are ungrammatical. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 213 b. Carl, sagde at Marie havde snakket med alle andre end ham, / ?ham, selv / *sigi / selv. ‘Carl said that M arie had talked to all others th a n __ c. Carli sagde at disse mcend hader alle inklusive/undtagen ham, / ?ham, selv / *sig, / *sig, selv. ‘Carl said that the m en hate all including/excluding__ d. Jon, sagde at han aldrig ville tillade sin datter sd meget som at overveje at gifte sig m ed en mand helt forskelligfra ?ham, / ham, selv / *sigj / *sig, selv. ‘Jon said that he never would allow his daughter so much as to consider to m arry herself to a man com pletely different fro m __ N otice that w hile the exam ples in (160) only allow the sim ple pronoun ham ‘him ’, the exam ples in (161) all allow for both simple ham ‘him and com plex ham selv ‘him self (him him self)’. Contrast seem s to he the m ost im portant characteristic o f the contexts allow LD- hound ham selv illustrated in (160). Both the so-called “sim ilarity predicates”, e.g. (161 a,d), and “exclusion predicates”, e.g. (161h,c), are inherently contrastive predicates w hich trigger adnom inal intensification. The term “sim ilarity predicate” is from Safir (1992). Safir (M .S., p. 14) defines this class o f predicates as follows; “The sem antic class o f [similarity] predicates [...] very often appears in copular constructions the sam e way as other adjectives do as in the com parisons in [(162)], yet other adjectives do not induce the sam e logophoric effects, as illustrated in [(163)]. A list o f sim ilarity predicates (from Safir (1992)) appears in [(164)]”. (162) a. Darby is fairly deferential / similar to Selena. h. Esther is quite estranged / different from Daffy. c. Angela is rather angry at / like Andrea. (163) a. Sissy insists that Darby is fairly similar / *? deferential to herself. a ’. Etta attacks people similar / *?deferential to herself. h. Theo says Esther is quite different / *? estranged from himself. h ’. Thor threatens people different / *? estranged from himself. c. Otto admits that Mary is rather like / *angry at himself. d. Abraham always abhors someone like / * angry at himself. (164) like, unlike, similar to, different from, identical to, such as R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 214 “W hile the “sim ilarity” predicates, all having to do with identity o f person or properties in relation to another, all perm it the logophoric effect, none o f the non-sim ilarity predicates do. The R & R account predicts that none o f these predicates perm it logophoricity if they apply their definition carefully. H owever even if the definition is interpreted the way they intend, that is, if they interpret it to mean that similar to is not a predicate and that it does not assign Case, then presum ably they m ust say the same thing about adjective plus preposition units like deferential to, angry at and estranged from . Either way, they fail to m ake a significant distinction because they fail to take into account the sem antic force o f the predicates involved.” (Safir (M .S.), p. 14). The term “exclusion” predicate is from Safir (1992). Safir (M .S.) describes the properties o f exclusion predicates as follows: “unlike sim ilarity predicates exclusion predicates [see (165)] are not possible in copular environm ents [see (166)], yet [like sim ilarity predicates] they also are typically tw o place relations and all arguably assign Case. (165) apart from, but, rather than, except, save, besides, other than, in addition to, including, excluding (166) *Hillary is apart from / rather than / other than Bill. By the latter criterion, the SELF-form s are syntactic argum ents in the R&R account if exclusion predicates are indeed predicates. Y et all o f these exclusion contexts perm it logophoric behavior [see (167)]. (167) Dole pointed out that the Republicans would look foolish if anyone other than / rather than / apart from / except / save him self were nominated. Safir concludes ‘that exclusion predicates are predicates and as such they are ju st as problem atic for R& R as the sim ilarity predicates are” (Safir (M .S.), P. 14-15). The sim ilarity predicates x er anderledes end y ‘x is different than y ’ and x er forskellig fra y ‘x is different from y ’ in (161a,d) both involve directly contrasting tw o R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 215 individuals or groups. Similarly, the exclusion predicates alle andre end x ‘all others than x ’, alle medregnet y ‘all including y ’, and alle undtagen y ‘all excluding y ’ in (161b,c) also involve direct contrast o f tw o groups/individuals; i.e. in (161b,c) the group o f people whom M arie spoke to is explicitly contrasted w ith the group o f individuals with whom she didn’t speak, see (168). a. A er anderledes end Y ‘X is different from Y ’ b. alle andre end Y ‘ all others than Y ’ c. alle inklusive/undtagen Y ‘all including/excluding Y ’ d. A er forskelligfra Y ‘X is different from Y ’ e. X IS CO N TRA STED W ITH Y Inherently contrastive predicates can be defined as predicates which involve explicitly contrasting two individuals or groups (X and Y) with respect to som e property or event, see (168f)’^. In exam ples with inherently contrastive predicates it is thus the m eaning o f these predicates which satisfies the contrastiveness principle, as form ulated in chapter 2, and thereby licenses intensifier adjunction to personal pronouns. N ow , let us return to the exam ple in (160a). A ccording to the Judgm ents reported in Safir (1997), the intensified pronoun ham selv is unacceptable in this exam ple. N otice however, that ham selv can be rendered felicitous, even in the sentence in (160a), in explicitly contrastive situations, see (169A). Note that with respect to the distribution of simple and complex reflexives in Danish (i.e. sig vs. sig selv), inherently contrastive predicates are “anti-reflexive” (i.e. “other-directed”), e.g.: (i) Peter er forskelligfra *sig/sig selv / John. ‘Peter is different from himself/John.’ (ii) Peter barberede all undtagen *sig/ sig selv / John ‘Peter shaved all excluding himself/John.’ R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 1 6 (169) Q: Hvem er det Jon vil have, at den kvinde han vil giftes med skal vcere stolt of? Vil Jon absolut giftes m ed en kvinde, som er stolt afhans far? who is it Jon will have that the wom an he will m arry-PA SS with shall be proud o f will Jon absolutely m arry-PA SS with a wom an w ho is proud o f his father ‘W ho is it that Jon wants the w om an he is going to m arry to adm ire? Does Jon absolutely w ant to m arry a wom an who adm ires his father?’ A; Nej, dit torskehovedei Jon vil giftes med en kvinde som er stolt o f ham selv. N o you cod-head Jon will m arry-PA SS w ith a w om an who is proud o f him self ‘N o, idiot! Jon w ant to marry a w om an w ho is proud o f (him ) h im se lf’ It thus seem s to be the case that ham selv can felicitously occur in sentences like (160a) under the condition that strong contrast is directly expressed or im plied by the larger context as in (169) (w here ham selv, referring to Jon, is directly contrasted with hans fa r ‘his father’). If this condition is not met, e.g. (160a), then the presence o f selv does indeed seem s rather infelicitous. T he exam ples discussed above seem to indicate that the presence o f selv in the com plex pronoun ham selv can be licensed in at least one o f tw o different ways: (i) by the inherent contrastiveness o f its governing predicate, see (161a-d), or (ii) by the contrastiveness evoked by the larger (linguistic or extra-linguistic context), see (169). N ow com pare that with the situation holding for sig selv. As already discussed in section 3.3.2, with “neutral” predicates the presence o f intensifier-adjunction is optional and usually only occurs in contexts w hich involve strong contrast betw een tw o individuals/groups, see (170), and cannot occur in sentences in w hich another elem ent has already been focused, see (171). See also Keenan (2001), Konig & Siemund (1999), Zribi-Hertz (1989) for discussion of “contrastive predicates” R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 217 (170) Q: Hvem var det nu at Peter havde vasket? Havde han vasket sin hund? ‘W ho was it that Peter had w ashed? Had he w ashed his dog?’ A: Nej, han havde vasket SIG SEL V / *SIG. ‘N o, he had w ashed him self.’ (171) Q; Hvad var det nu at Peter havde gjort fa r han gik i seng? ‘W hat was it Peter had done before he w ent to bed?’ A: Han havde VASKET sig /? ? sig selv. ‘He had W ASHED h im se lf’ As illustrated by the contrast betw een (160a) vs. (169) and (170) vs. (171), selv is often adjoined to the sim ple reflexive sig and the simple pronoun ham for the purpose o f indicating contrast. Indeed, the contrast betw een (160a) vs. (169) and (170) vs. (171) illustrates that optional intensifier-adj unction to sig and ham is felicitous only when contrastiveness is som ehow m otivated (or at least com patible with) by the larger context. T he fact that ham selv can occur freely with all the predicates in (161a-d), even w ithout explicit contrast present in the extra-sentential context, can now be explained simply as follow ing from the fact that all those predicates are inherently contrastive, see (168e), and thus in them selves provide the contrastive context necessary to m otivate the presence o f selv. This also explains why 5e/v-adjunction is not licensed in those constructions involving predicates like stolt a / ‘proud o f w hich are not inherently contrastive, unless the larger context contains explicit contrast betw een the argum ent o f the predicate stolt a / ‘proud o f and some other entity, see (160a) vs. (169). In the above discussion we have dem onstrated that com plex pronouns, e.g. ham selv, share some properties with intensified reflexive, e.g. reflexive sig selv discussed in section 3.3. That is both occur in contrastive contexts, i.e. contexts which satisfy the contrastiveness condition on adnom inal intensification discussed in chapter 2. There are, however, also im portant differences betw een ham selv and and sig selv which follow from sem antic and prosodic differences betw een sim ple sig and ham, some o f which are listed in (172). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 218 (172) C om oarina s is and ham: sig ham a. C an carry stress : N O YES b. C an refer on its ow n ; N O YES As also discussed in section 3.3.6.3.3-4, the fact that ham is potentially stressable while sig is not, see (173a), may be the reason why selv is optional in the exam ples (161a-d). (173) a. Carl sagde at Marie havde snakket m ed alle andre end ham selv b. Carl sagde at Marie havde snakket med alle andre end HAM. c. ? *Carl sagde at Marie havde snakket med alle andre end ham. Carl said that M arie had talked to all others than him self/H lM /?*him The exam ples in (173a,b) show that both intensifier adjunction {ham selv) and focus on H AM ‘H IM ’ can be used to satisfy the contrastiveness requirem ent o f the “inherently contrastive” predicate alle andre end ‘all others th an ’’® . In contrast, the sentence degrades considerably if the pronoun ham is neither focused nor intensified, see (173c)™. The fact that ham has phi-features and thus can have reference on its own, see (173 b), is the reason why sim ple ham gives rise to referential am biguities w hich do not occur with sim ple sig, see (174a). (174) a. Peteri sagde at orkestret ikke maatte spille med andre end hamj ^ som dirrigent. b. Peteri sagde at orkestret ikke maatte spille med andre end ham, selv som dirrigent. Peter said that orchestra-the not could play w ith others than him / him self as conductor ‘Peter said that the orchestra couldn’t play w ith anybody but him self as conductor.’ Simple ham in (174a) m ay refer either to the m atrix subject Peter or to som e 3 '^ '* person male individual not m entioned in the sentence. The com plex ham selv, on the other hand, seem s to This is reminiscent o f the situation holding in English where stressed HIM and him self are also (to a certain extent) in free variation, see Baker (1995). See also Siemund’s (2000) critique o f this analysis. Note that the sentence in (i) may potentially constitute a problem for the account of the distribution o f ham and ham selv proposed here. (i) Peter„ havde snakket med alle andre end *ham,,/ ham„ selv / *sig,,/ sig„selv. ‘Peter had talked to all others than__ According to the contrast-based account proposed in 3.3.6.3.3, we would expect a stressed form of the simple pronoun ham ‘him’ to be possible in (i). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 219 be restricted to referring back to the m atrix subject. It thus seems that intensifier-adj unction to ham has the effect o f lim iting the range o f possible antecedents. This effect is w hat Zribi-H ertz (1995) refers to as the endophoricizing effect o f intensifier-adjunction. It does not seem im plausible that, in certain instances, the use o f ham selv rather than sim ple ham could be motivated by a need/intention to disam biguate the reference o f ham. N otice, however, that intensification o f pronouns does not always disam biguate between tw o or m ore possible antecedents, see (175). (175) a. Rune^ troede at Peteri havde sagt at orkestret ikke maatte spille m ed andre end ham, 2 selv som dirrigent. Rune thought that Peter had said that orchestra-the not could play w ith others than him self as conductor ‘Rune thought that Peter had said that the orchestra couldn’t play with anybody but him self as conductor.’ b. Rune^ var ikke synderligt hange fo r at nogen anden skulle usurpere hans^ plads som dirrigent p a det kongelige teater. Peteri havde nemlig sagt at orkestret ikke maatte spille m ed andre end ham,u z selv som dirrigent. Rune was not particularly afraid for that som ebody else should usurp his place as conductor at the royal theater. Peter had actually said that orchestra-the not could play with anybody but him self as conductor. ‘Rune was not particularly afraid for that som ebody else should usurp his place as conductor at the royal theater. That was because Peter had said that the orchestra couldn’t play w ith anybody but him self as conductor.’ In (175a) the antecedent o f the intensified pronoun ham selv could be either the Peter or the m atrix subject Rune. As shown in (175b), given the right context ham selv may even skip a sentence internal subject in order to be bound by an extra-sentential antecedent. Based on these exam ples, we conclude that the apparent disam biguation function o f xe/v-adjunction is sim ply a side-effect o f adnom inal intensification. T h e sen ten ce in (1 7 6 ) illustrates that in ten sified pronouns do n ot n eed to b e bound by logophoric subjects and that the main function o f adnom inal intensification is contrastive. R e p r o d u c e d with p e r m is s io n of t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 2 0 (176) For hendej var det, som om dode, lykkelige Dage rejste sig a f deres Grav og gik igjen, saa Alting scelsom sodt forvirredes og Fortid og Nutid smelted sammen til en solvsloret, dcemrende Drommedag, hvor hunt elskede Ynglingen^, halvt som ham^ selv, halvt som en Andens Skygge, og gav ham^ helt sint halve Sjcel. For her was it, as if dead happy days rose REFL o f their grave and w ent again so everything strangely sw eet confused-PA SS and past and present m elted together to a silver-veiled dim dream -day w here she loved the swain h a lf as him self h alf as a other’s shadow and gave him w hole PO SSREFL h alf soul ‘For hen it was as if dead, happy days rose from their grave and cam e back hauntingly, so that everything was confused in a strangely sw eet way and past and present melted together to a silver-veiled, dim dream -day w here shei loved the swain*, h alf as himself*, h alf as som ebody else’s shadow, and gave him* unconditionally h a lf o f hep soul.’ (J.P.Jacobsen (1880) Niels Lyhne, p. 81, bold and italic added here) In (176) intensification o f the pronoun ham is clearly not m otivated by a need for referential disam biguation, but rather by the explicit contrast betw een Ynglingen ‘the sw ain’ and en Andens Skygge ‘som ebody else’s shadow ’. We suggest that the prim ary function o f intensification is to indicate some kind o f contrast, and that w hat appears to be a disam biguation function o f selv is no m ore than a secondary effect o f intensification. 3.4.2 Intensified subject pronouns: Intensified pronouns as subjects of embedded clauses in English and Danish In the previous section intensification o f object pronouns was discussed. This section will present an account o f sim ple and com plex subject pronouns. As w ith com plex reflexives and com plex pronouns discussed above, we argue that com plex subject pronouns, e.g. Dan. han selv ‘he h im self, are sim ple adnom inally intensified versions o f the corresponding sim ple forms, e.g. han ‘h e’. In his 1986 Linguistic Inquiry squib “He Himself: Anaphor, Pronouns, Or . . . ?” Bickerton claims that the com plex form he him self is a special kind o f anaphor having the R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 2 1 features [+anaphor, +pronom inal] in C hom sky’s (1982) system*®. This feature com bination would then explain the (alleged) fact it appears to have properties associated with both anaphors and pronouns. Bickerton (1986:347)*' arrives at this conclusion based on data such as (177)-(l 82). Sentences (177) and (180) show that like pronouns, she herself can have extra- sentential antecedents. The fact that its antecedent does not have to c-com m and it, see (181) and (182a), also confirm s its pronom inal behavior. However, it behaves m ore like anaphor in other respects. It does not allow extra-sentential antecedents in the presence o f potential intra- sentential antecedents, see (178). Furtherm ore, if it is bound inside the sentence, then it must be by a c-com m anding antecedent, see (178) and (179). W hile com plex form s like she herself share properties o f both anaphors and pronouns they also differ from both with respect to Case-m arking and reconstruction effects. Unlike reflexives and object pronouns, they receive nom inative case and appear to allow Principle C violations in LF reconstructions, see (182a) vs. (182b). (177) A: How will Maryi do in the exam? B : I d o n ’t k n o w , b u t she; h e rs e lf say s shej’ll pass. (178) A: How will Maryi do in the exam? B: I d o n ’ t know, hut Susans says that shei /t /she herself*i * will pass. latridou’s (1986) LI squib “An Anaphor Not Bound in Its Governing Category” proposes that Modern Greek has an expression very similar to Bickerton’s he himself, i.e. o idhios ‘he him self, see (i). (i) a. O Yanisi pistevi oti o idhiosi tha kerdhisi. (=latridou 1986(6)) John believes that himself will win ‘John, believes that he himself will win.’ b. O Yanis, theli [i Maria na voithisi ton idhioj. John wants Mary helps himself, ‘Johni wants Mary to help himj.’ c. O Yanis, ipe ston Costai,[oti i Maria aghapa ton idhion.t„J- John said to Costa that Mary loves himself ‘John, said to Costa,^ that Mary loves him,* Like English himself, the expression idhios is also used as adnominal intensifier, see (ii). (ii) a. O Yanis o idhios pighe sto scholia. b. Aftos o idhios pighe sto scholia. c. pro o idhios pighe sto scholia. ‘John / He / pro went to the school himself.’ Modern Greek is a pro-drop language and latridou argues that (lie) shows that pro can be adnominally intensified by the intensifier idhios ‘same, him self. *' The account o f examples (177-182) is based on a paraphrase of Bickerton (1986:347). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 2 2 (179) a. Johrii told Bilik’ s sister that hCi/k / he himselfi * k had been arrested. b. Susan i told everyone who knew Maryk that she a / she/. * k herself was pregnant. (180) Even Billi’ s genius has its limits. The problem that Mary ju st raised is one that he him self admits he ca n ’ t solve. (181) a. The essays that Mary, wrote were things that she, herself attached little importance to. b. Explaining what John, really believes is something that he, him self ca n ’ t always do. c. The ways in which the twins, behave suggest that they, themselves d o n ’ t always know what they ’ re doing. (182) a. Which pictures ofJohn, does he, him self like t? b. *Which pictures o f John, does he, like t? Unlike B ickerton (1986), M cKay (1991) argues that he him self is not an anaphor. Furtherm ore, according to M cKay the “em phatic appositive use o f reflexives”, i.e. w hat in this dissertation is called intensifiers, “require a clearly indicated referent and a relevant contrast or com parison class [..]”(M cK ay 1991:368). In other words, M cKay essentially proposes that he him self is an intensified pronoun. Syntactically, it behaves like its unintensified counterparts he, him, etc. Pragm atically, it is subject to the sam e discourse based conditions (i.e. contrastive contexts) as other intensified expressions. In this sense, M cKay (1991) is the forerunner o f Baker (1995) as well as the present dissertation. That is, we share M cK ay’s conclusion that “/le him self is grammatical w henever he is gram m atical, and the same range o f antecedent relationships is gram m atically possible, though he requires a relevant contrast or com parison for pragm atic appropriateness” (M cKay 1991:370-371). M cKay (1991) uses exam ples like (183) to show that he him self \s not a new type o f anaphoric exp ression (as argued by B ick erton ), but sim p ly an in ten sified p ronouns w h ich is still su bject to principle B. (183) A: Mary, has been concerned about her friends. Susank said that several were going to fa il the course, and Susank might be right. But Mary, should think more about her, own work. How will Mary, do on the exam? R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 223 B: 1 d o n ’ t know, but Susanj^ says that shci (sheij (shei herself) (shek herself will pass. We agree with M cKay that the exam ple in (183) falsifies B ickerton’s (1987) claim that he him self hehdiVQs, like an anaphor in that it “cannot have an antecedent outside the sentence if there is a possible antecedent inside”. B ickerton’s claim was based on exam ples like (178). M cK ay’s main contribution is to show that, given the right context, as in (183), he him self can be bound by a sentence-external antecedent in spite o f the presence o f a potential antecedent inside the sentence. That is, the pronom inal nature o f subject pronouns is not affected by adnom inal intensification. In other words, binding and adnom inal intensification are independent o f each other. As the exam ples in (184-185) show, D anish intensified subject pronouns, e.g. han selv ‘he h im se lf, behave sim ilarly to their English counterparts, i.e. they are pronom inals falling under principle B o f the binding theory. Independently thereof, w hether or not they are intensified depends on pragm atic factors (i.e. contrast). (184) A: Mary, has been concerned about her friends. Susan/, said that several were going to fa il the course, and Susan/, might be right. But Mary, should think more about her, own work. How will Mary/ do on the exam? B: Det ved je g ikke, men Susan/, siger at huUi /jhun/ /, selv vil bestd uden problemer. that know 1 not, but Susan says that she/she herself will pass w ithout problem s ‘1 don’t know, but Susan^ says that shci/k/shej/k herself will pass w ithout any problem s.’ (185) N iels Lyhne/ havde derfor ogsaa digtet ude fra en cesthetisk Personlighed i al Almindelighed, der fandt Foraaret svulmende, Havet stort, Kjcerligheden erotisk og Doden melankolsk. Han/ selv var ikke kommet videre med denne Poesi, han, bare lavede versene. N ie ls L yhne had therefore a lso com p o sed -p o etry out from an aesth etic person ality in all generality, w hich found spring-the swelling, sea-the great, love-the erotic and death-the m elancholic he self was not com e further w ith this poetry he m erely made verses-the ‘N iels Lyhnei had therefore also com posed poetry based on a general aesthetic personality who found the spring swelling, the sea great, love erotic and death m elancholic. HCj him self had not gotten any further with this poetry; he; m erely made the verses.’ (J.P.Jacobsen (1880) Niels Lyhne, p. 83, bold and italic added here) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 224 In (184) Mary is contrasted with her friends and in (185) Niels Lyhne is contrasted with en cesthetisk Personlighed i al Almindelighed ‘a general aesthetic personality’. In brief, intensified pronouns in Danish, e.g. han selv ‘he h im self, are not a special type o f anaphor but simply adnom inally intensified versions o f their sim ple counterparts w hich have to satisfy the contrastiveness condition on adnom inal intensification. 3.4.3 Logophors and logophoricity vs. intensification and intensified pronouns The term “logophor” is often used loosely in the literature to refer to locally free instances o f him selfm English*^. It has also som etim es been used to refer to the intensified pronoun ham selv w hich tends to be the form used in Danish in contexts were English w ould have locally free instances o f himself. This use o f the term logophor is, however, potentially m isleading since the term logophor is also widely used in a more narrow sense as referring to an author- denoting pronom inal elem ent w hich has to be bound by an attitude operator (cf. Clem ents (1975), C hierchia (1989), H uang& Liu (2000), Schlenker (1999), etc..)*^. This definition o f logophoricity was first developed to describe the behavior o f m orphologically specialized logophoric elem ents in W est African languages (cf. Hagege (1974), and Clem ents (1975)). The exam ples in (186-87) (from Clem ents (1975)) exem plifies the behavior o f the logophoric elem ent ye in Ewe. (186) a. Kofi be ye-dzo Kofi say LO G -leave ‘Kofi said that he (K ofi) le f t ’ b. Kofi he e-dzo K ofi sa y h e/sh e-lea v e ‘Kofi said that he/she (T^Kofi) left.’ E.g., Reinhart & Reuland (1993) who refer to all instances of him self which are not coindexed by a co-argument as “logophors”. In other words, Reinhart & Reuland’s approach implies that any SELF form which does not fall under Principle A must be a logophoric element. Sells (1987), Zribi-Hertz (1989), Reinhart & Reuland (1993), etc. for different definitions o f logophoricity. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 225 a. *Ama do nku nynnuvi hi dze ye gbz) dyi A m a set eye girl WH stay LOG side on b. Ama do nku nyzmuvi hi dze e gb=> dyi A m a set eye girl W H stay pro side on ‘A m a set eye on (rem em bered) the girl who stayed with her’ c. Ama ghl=' he ye-do nku nyrmuvi hi dze ye g b ^ dyi A m a say that LOG eye girl W H stay LOGside on ‘ Ama; said shei rem em bered the girl who stayed with her,’ The exam ple in (186a) show the standard case o f a logophoric pronoun hound hy an overtly realized attitude operator (i.e. the m atrix subject). Interestingly, the English and Danish translations o f this exam ple do not necessarily involve the so-called “logophors”, i.e. him self and ham selv, hut rather the simple subject pronouns he and han ‘h e’, see (188). (188) a. Kofi said that he / ^ '‘ ^he him self / *himself left. h. Kofi sagde at han/ ^'^han selv / *selv tog afsted. Kofi said that he / he self/ self took of-place ‘Kofi said that he/ '^ "'* h e him self/*him self left.’ The intensified pronoun han selv, m arked w ith a in (188) is perfectly acceptable when the larger context im plies contrast (see the discussion o f exam ples (160a) and (169) in section 3.4.2). In this respect, han selv seem s to behave exactly like he him self \n English, see also section 3.4.2. The Ewe exam ples in (187) illustrate the fact that logophors cannot occur inside a relative clause, unless the relative clause is itself em bedded under an attitude verb. Let us now test to see if Danish ham selv behaves the same way in relative clauses, see (1 8 9 ). (189) a. Jon vil giftes med en kvinde som er stolt a f ham / '' "fiam selv^'' / *sig / *sig selv. ‘Jon w ants to marry (passive) w ith a w om an w ho is proud o f h im se lf’ b. Jon sagde altid at han ville giftes med en kvinde som var stolt a f ham / "^ham selv / *sig/ *slg selv. ‘Jon always said that he w anted to m arry (passive) with a w om an who is proud o f .’ c. Jon vil giftes med en kvinde som er anderledes end ?ham / ham selv / *sig / *sig selv. ‘Jon wants to m arry (passive) with a wom an who is different from .’ The judgments indicated as ” °''ham selv here means that the use of the form ham selv is unfelicitous (i.e. ??) unless when the whole sentence occurs in an explicitly contrastive context (i.e. ok). R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 226 d. Jon sagde altid at han ville giftes med en kvinde som var anderledes end ?ham / ham selv / * sig/ *sig selv. ‘Jon always said that he w anted to m arry (passive) with a w om an who is different fro m . ’ As already discussed in the previous section, the sentence (189a), does seem to be infelicitous as it is, i.e. in the absence o f an explicitly contrastive larger context. The question is now w hether it will behave like its Ewe counterpart (187a,c) and im prove when em bedded under an attitude verb, see (189b). It is my im pression as a native speaker that the acceptability o f ham selv in (189a) does not im prove significantly w hen the sentence is em bedded under an attitude verb w hose subject is coreferent with ham selv, see (189b)*^ W hile (189b) may be slightly better than (189a) it is still the absence vs. presence o f an explicitly contrastive context w hich is the m ost im portant factor deciding the acceptability o f both (189a) and (189b). The exam ples in (189c,d) involve the inherently contrastive predicate anderledes end ‘different from ’ and as a consequence, the use ham selv always results in a perfectly acceptable sentence regardless o f the presence/absence o f overt attitude operators and the nature o f the context. In other words, unlike the situation holding for Ewe logophors, see (186) and (187), in Danish, it is not the presence/absence o f an attitude operator which determ ines the acceptability o f the sentences w ith ham selv in (189), but rather the contrastive/non-contrastive nature o f the predicate governing ham selv. Based on the above exam ples we conclude that Danish ham selv is not a logophor in the narrow sense o f the term but rather the intensified form o f the pronoun ham. 3.4.4 Summary W hile there are still a num ber o f unsolved problem s surrounding the question o f the distribution o f ham selv in Danish (e.g. the endophorizing/logophorizing function o f selv- R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r . F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 227 adjunction to ham), it seems clear that contrast (either provided by the larger context or the governing predicate) plays a central role. In other words, intensification (the distribution o f selv) and binding (distribution o f pronouns han/ham/hans ‘he/him /his’) are mutually independent. Com plex pronouns (e.g. han selv ‘he h im self, ham selv ‘him h im self, and hans egen ‘his ow n’) com posed o f a pronom inal part (e.g. han/ham/hans ‘he/him /his) plus an adnom inal intensifier (selv/egen ‘self/ow n’). The pronom inal part fall under principle B o f the binding theory while the intensifier is subject to the contrastiveness condition on intensification. Consequently, intensified pronouns in D anish and English are neither logophors (i.e. elem ents w hich have to be bound by an attitude operator), nor a special kind o f anaphor (cf. latridou (1986), Bickerton (1986)) as has som etim es been proposed. 3.5 Conclusion The independence of binding and intensification In this chapter we have presented an account o f sim ple and com plex reflexives and pronouns in D anish based on the proposal that binding and intensification should be clearly separated although they overlap in the case o f com plex reflexives and pronouns w hich are here analyzed as adnom inally intensified form s o f their simple counterparts. That is, we proposed that the deseriptive generalizations in (190)-(193) are best accounted for w ithin a fram ew ork w hich separates binding and intensification into the independent m odules sum m arized in (194) and (195) respectively. (190) Sim ple reflexives (e.g. si2): m ust be bound by subject in a local dom ain (=m inim al tensed clause) (191) Sim ple pronouns fe.g. ham): m ust be subject-free in a local dom ain (=m inim al tensed clause) Since intensifier adjunction is highly sensitive to various context factors (e.g. implicatures, contrast, ...) 100% consistent native speaker judgments are hard to obtain cases like these. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 228 (192) C om plex reflexives (e.g sis selv): a. anti-reflexive predicates b. hidden neutral c. contrastive contexts (w ith neutral predicates) d. doppelganger-effects (i) statue-readings (ii) qua-sentences (iii) strict (and sloppy) readings in V P ellipsis e. stress-carrier (e.g. after unstressed prepositions) (193) C om plex pronouns (e.g ham selv): a. inherently contrastive predicates b. explicitly contrastive contexts (194) Binding theorv: a. Principle A (accounts for (190)). b. Principle B (accounts for (191)). (195) Contrastiveness condition on adnom inal intensification (=(2) chapter 21 (accounts for 1192-1931: A nom inal expression (DP, pronoun, reflexive) can only be intensified adnom inally if it can be contrasted with other expressions in the context in which it is found. Some o f the consequences o f the present proposal for linguistic theory in general are listed in (196) and in (197) particular consequences for the analysis o f M odern D anish can be found. (196) Consequences for binding theorv: a. Intensification and binding are independent m odules o f the grammar. b. Binding is purely syntactic phenom enon. c. The unified account o f possessive and argum ent reflexives is a strong argum ent in favor o f nom inal approaches to binding and against current predicate-centered approached (e.g. R einhart and Reuland (1993)). d. “D oppelganger-effects” (statue-readings, qua-sentences, strict reading in VP ellipsis, etc.) and anti-reflexivity (the triggering o f adnom inal intensification o f reflexives with certain predicates) can be unified as cases o f adnom inal intensification falling under the contrastiveness condition. (197) Consequences for analvsis o f M odem D anish (and sim ilar languages): (i) Sim ple reflexives (e.g. Dan. sig ‘R E FL ’) are not anti-local (discussed in m ore detail in chapter 4). (i) C om plex reflexives (e.g. Dan. sig selv ‘REFL self) are not m andatorily local. (ii) C om plex reflexives, e.g. sig selv, are adnom inally intensified reflexives. (iii) The com plex object pronoun ham selv is not a logophor but an adnom inally intensified object pronoun. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 229 (iv) The com plex subject pronoun han selv is not a special kind o f anaphor (cf.Bickerton 1986) but an intensified subject pronoun. (v) The elem ent selv ‘s e lf is not a reflexivizing particle w hich falls under binding theory (e.g. Reinhart and Reuland 1993), but an intensifier w hich falls under the m odule o f intensification (see chapter 2) (vi) The elem ent egen ‘ow n’ is the suppletive variant o f the adnom inal intensifier selv ‘se lf. (vii) The argum ent anaphor sig ‘R E FL ’ and the possessive anaphor sin ‘PO SSR EFL’ have the same distribution and can both be accounted for by the same principle A o f the binding theory. As show n above, the contrast-based account intensification proposed in chapter 2 brings out the core com m on feature o f 5e/v-adjunction to reflexives (e.g. sig) and pronouns (e.g. ham), i.e. the contrastiveness condition on intensification. As shown in section 3.3.2 and 3.3.3 o f this paper, intensifier-adjunction to reflexives can only take place when the predicate/construction containing sig/sin^'' allows for contrast with other entities. Likewise, intensifier adjunction to ham has a strong tendency to prefer contrastive contexts. Com pared with previous analysis o f reflexives and pronouns in Danish, the focus-based analysis o f intensifier-adjunction proposed in chapter 2 has a better chance at eventually arriving at a unified account o f intensified reflexives, e.g. sig selv, intensified pronouns, e.g. ham selv, and intensified DPs, e.g. kongen selv ‘the king h im self. Indeed, it seems to be the failure to treat intensification and binding as separate subsystem s o f the gram m ar w hich has m ade it im possible for many researchers to even conceive o f a unified analysis o f the function o f selv across reflexives, pronouns and DPs. In chapter 4 a few loose ends will be tied up. That is, the them atic status o f simple reflexives with neutral and inherently reflexive verbs will be discussed in greater detail and a late-insertion analysis o f reflexives and pronouns, which also accounts for L* and 2"'* person As argued in section 3.3.5, the distribution of the intensifier egen/eget ‘own’ with possessive pronouns, e.g. hans egen ‘his own’, also follow the contrastiveness principle. Due to space limitations this topic cannot be explored in detail here. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 230 forms, is proposed. In chapters 5 and 6 sim ilar approaches to binding and intensification in English and M andarin Chinese will be put to the test. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 231 Chapter 4 The syntactic status of sig 4.1 Introduction In this chapter, the hotly debated question o f the lexical and syntactic status o f the sim ple reflexive sig ‘R EFL’ (=Nor. seg, Dutch zich, etc.) is discussed. We argue that sig can be a locally-bound theta-m arked reflexive pronoun (la ), (2a). Hence, it is not an anti-local anaphor as has som etim es been proposed. W e furtherm ore argue that the lexicon only contains one entry for the elem ent sig and that the difference between them atic and non-them atic uses o f sig depends on w here in the derivation this elem ents m erges (lb ). (1) Proposal: a. sig is not anti-local, i.e. them atic sig can have both local and LD -antecedents. b. the lexicon contains only one entry for sig. This elem ent can be either a them atic anaphor, w hen it occurs in argum ent position o f an active transitive predicate (2a), or a non-them atic gram m atical m arker o f unaccusativity, i.e. in m iddles or passive, when it occurs in the v o f the voice-projection vP (2b). (2) a. Them atic s is : VP / \ .... V ’ / \ V sig b. N on-them atic s is : vP / \ v’ / \ / \ VP sig V / \ .... V ’ / \ V .... The exam ples in (3) illustrate them atic uses o f sig ‘R E FL ’. (3) a. Peter vasker sig. Peter washes REFL ‘Peter washes (him self).’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 232 b. Peter lagde bogen bag sig. Peter put-PA ST book-the behind REFL ‘Peter put the book behind him .’ Both instances o f sig in (3) are them atic in that they receive them atic-roles from their predicates. In contrast, the exam ples in (4) illustrate non-them atic uses o f sig. (4) a. Peter skynder sig. Peter hurries REFL ‘Peter is in a hurry/Peter hurries.’ b. Peter shammer sig. Peter sham es REFL ‘Peter is asham ed.’ In section 4.3 we show that the predicates skynde sig ‘hurry’ and skamme .sig ‘be asham ed’ are not transitive predicates but rather lexically unaccusative intransitive predicates w hich only assign one theta-role to the internal argum ent Peter w hich later m oves to the subject position for case reasons. Unlike the transitive verb vaske ‘w ash’ in (3a), they do not denote events or actions involving two objects. Rather, they denote particular states o f one individual, i.e. the referent o f the theta-m arked subject. Hence, since there is no theta-role left for the reflexive, the instances o f sig in (4) are all non-them atie. Some verbs, e.g. dbne ‘open’ and lukke ‘close’, have both agentive transitive uses (5) and deagentive unaccusative uses with sig (6). The exam ples in (4) and (6) thus illustrate lexical and derived unaccusative predicates respectively. (5) a. Peter dbner daren. Peter opens door-the ‘Peter opens the door.’ b. Peter lukker daren. Peter closes door-the ‘Peter closes the door.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 233 (6) a. Doren dbnede s ig . door-the opened REFL ‘The door opened.’ c. Doren lukkede s ig . door-the closed REFL ‘The door closed.’ A s m entioned in (1), we propose to analyze the difference between them atic sig, see (3), and non-them atic sig, see (4,6), as a syntactic rather than as a lexical difference. Lexically there is only one elem ent sig w hich can take on different functions depending on w here it m erges in the derivation. Them atic sig m erges in a theta-position, e.g. internal argum ent o f verbal predicate (3a) or argum ent o f preposition (3b), while non-them atic sig m erge with the v o f the voice projection vP, see (2b), (4) and (6). It is the clitic properties o f sig, i.e. its sim ultaneous X° and X P status, w hich enable it to behave differently depending on its position in the derivation. N eedless to say, only them atic sig falls under the binding theory, while non-them atic sig falls under a different m odule o f the gram m ar (i.e. argum ent structure/voice). We argue that this approach to the analysis o f all occurrences o f sig in Danish is both m ore explanatory as well as m ore econom ical than previous accounts. A ccording to a num ber o f current accounts o f Danish (and sim ilar languages, e.g. N orw egian, Swedish, and Dutch) sim ple reflexives (e.g. Dan. sig) are anti-local. That is, it is assum ed that local sig is never a true anaphor but rather a non-them atic gram m atical m arker o f intransitivity w hich is part o f the lexical m ake-up o f deponent verbs and reanalyzed deagentive unaccusative verbs (cf. H olm berg (1984), V ikner (1985) am ong others). We argue that although this analysis does apply to som e instances o f local sig, e.g. (4) and (6), it does not extend to ALL verbs allow ing local sig. As will be shown, the large group o f verbs referred to here as “neutral” do allow local sig (even when bound by co-argum ents) w ithout show ing any signs o f reanalysis or R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 234 lexical am biguity. The fact that the neutral predicate vaske ‘w ash’ has the sam e m eaning in (3a) as in (7) indicates that no reanalysis has taken place in (3a). (7) Peter vasker d0ren. Peter washes door-the ‘Peter washes the door.’ In both (3a) and (7) the predieate vaske ‘w ash’ assigns the same two theta-roles (A G EN T and TH EM E) and thus denotes a relation o f w ashing betw een tw o entities (the only difference being that in (3a) the w asher and w ashee happen to be the same). We argue that exam ples such as (3a) and (7) falsify the anti-locality accounts o f sig which would have to claim that the sig in (3a) is non-them atic gram m atical m arker o f intransitivity. U sing a num ber o f syntactic and sem antic tests we show that there is no evidence for assum ing the verb vaske ‘w ash’ to have been reanalyzed as an unaccusative intransitive verb in (3a). In contrast, several tests show that vaske ‘w ash’ is still a transitive verb assigning tw o theta-roles. As already discussed in chapter 3, based on w hat kinds o f nominal expressions they allow in object position, Danish verbal predicates can be divided into three m ain types: (i) “neutral” predicates w hich allow both sim ple and com plex reflexives (i.e. x /g an d sigselv) and DPs, see (8), (ii) “anti-reflexive” predicates w hich allow the com plex reflexive sig selv and DPs but not the sim ple reflexive sig, see (9), and (iii) “inherently reflexive” predicates which allow only the sim ple reflexive sig, see (10). (8) “N eutral” predicates: a. Peter vasker s ig / sig selv / bilen. (them atic sig) Peter washes REFL / REEL-SELF / car-the ‘Peter washes him self / the car.’ b. Peter forsvarer sig /sig selv / Marie (them atic sig). Peter defends REFL / REFL-SELF / M ary ‘Peter defends him self / M ary.’ (9) “A nti-reflexive” predicates: a. Peter mistcenker *sig / sig selv / Marie, (them atic sig selv) Peter suspects *REFL / REFL-SELF / M ary ‘Peter suspects him self / M ary.’ R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 235 b. Peter misunder *sig / sig selv / Marie, (them atic sig selv) Peter envies *REFL / REFL-SELF / M ary ‘Peter envies him self / M ary.’ (10) “ Inherentiv reflexive” predicates: a. Peter skammer sig/*sig selv/*M aire. (non-them atic 5/g) Peter sham es RELF / *REFL-SELF / *Mary ‘Peter is asham ed o f him self / M ary.’ b. Peter dukker sig / *sig selv / * Marie. (them atic sig) Peter ducks REFL / *REFL-SELF / *Mary ‘Peter ducks *him self / *M ary.’ The distribution o f different types o f nom inal expressions in object position with the different verb types is sum m arized in (11). (11) D istribution o f nom inal expressions in obiect position: a. A nti-reflexive *sig / sig selv / DP b. N eutral sig / s ig s e lv / DP c. Inherently reflexive sig / *sigselv/ *DP The relation betw een the sem antic properties o f predicates and adnom inal intensification o f sim ple reflexives has already been discussed in great length in chapter 3 and will not be repeated here. In the rest o f this chapter the focus will be on the types o f predicates w hich allow sim ple sig, i.e. neutral (8) and inherently reflexive predicates (10). We argue that “neutral” transitive verbs retain their transitivity even when they occur w ith the simple reflexive sig, see (8) as well as (3a) vs. (7). In contrast, other approaches (e.g. Reinhart and Reuland 1993, and others) assum e that all the “neutral” verbs occur tw ice in the lexicon: once, as a true transitive verb w hich require the intensified reflexive sig selv, e.g. (7), and once as an inherently reflexive verb w hich does not need to be overtly reflexive-m arked by the SELF elem ent o f the com plex reflexive, e.g. (3a). We argue that such lexical am biguity is both unwarranted and unnecessary. That is, neutral predicates can allow sim ple sig as direct object w ithout being “inherently reflexive”. In section 4.2. a num ber o f syntactic and sem antic tests will be used to com pare and evaluate these com peting views o f “neutral” predicates and it will be argued that the approach adopted here makes it possible to arrive at a sim pler, more R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 236 econom ical, account o f binding w hich does not need to stipulate double entries for neutral verbs. T he intensified form o f the reflexive, i.e. sig selv, has been claim ed always to be them atic. See section 3.3.6.2 in chapter 3 and section 4.3.3 in chapter 4 for discussion o f instances o f non-them atic sig selv, w here the presence o f selv is triggered by prosodic rather than sem antic factors. Section 4.3 contains a discussion o f different types o f “inherently reflexive” predicates, i.e. predicates w hich can only occur w ith sig, see (10). It will be argued that the large m ajority o f inherently reflexive predicates are o f a type o f “deponent” unaccusative verbs where sig functions as a non-them atic m arker o f unaccusativity occurring in the v o f the voice projection vP, cf. (2b), (10a). However, there are also a num ber o f inherently reflexive predicates w hich fail to behave as unaccusatives w ith respect to different syntactic and sem antic tests. These predicates thus appear to be sem antically inherently reflexive but still fully transitive predicates w hich assign two theta-roles but which, due to strict sem antic selection restrictions, only allow simple reflexives as internal argum ents (2a), (10b). From a binding-theoretical point o f view, one o f the main interest o f this chapter is that it addresses the issue o f w here to draw the dividing line betw een them atic and non-them atic uses o f reflexives, e.g. while Reinhart and Reuland (1993) consider both instances o f sig in (10a,b) to be them atic, we argue that sig in (10a) in non-them atic and that the sig in (10b) is them atic. We also disagree w ith V ikner (1985) who w ould consider both instances in (10a,b) as well all the instances o f sig with neutral predicates, e.g. (3a,b) and (8a,b), to be non-them atic gram m atical m arkers’. ' In these respects .lakubowicz (1994) analysis of Danish reflexives is very similar to the one defended here. That is, she also considers local sig with neutral predicates to be thematic without stipulating double entries for all neutral predicates. R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 237 The problem posed by the fact that there is no formal distinction between reflexives and pronouns in the and 2"‘ ‘ person is taken up in section 4.4 w here we present a late- insertion analysis w hich accounts for these facts w ithout stipulating the existence o f m ultiple lexical entries. Like the adnom inal intensifier selv ‘s e lf discussed in chapter 2, the reflexive elem ent sig can be found in a num ber o f different uses with w hat seem s to be widely different syntactic and sem antic properties. In section 4.5 the full range o f uses o f reflexive elem ents (e.g. in reciprocal, m iddle, passive, etc. constructions) will be briefly discussed. This section is largely descriptive and its m ain function is to situate the uses o f sig as them atic reflexive and non-them atic m arker o f unaccusativity in the context o f the full range o f uses o f reflexive elem ents. As m entioned in chapter 1, a com plete m apping o f the full ranges o f uses o f both the intensifier selv and the reflexive sig is a necessary prerequisite to our understanding o f the behavior o f these elem ents independently as well as the instances w here they com bine to form the so-called com plex reflexives. O ur analysis o f com plex reflexives as intensified nominal expressions thus relies on carefully distinguishing betw een the contributions o f different m odules o f the gram m ar, see (12). (12) M odule: A ccounts for: a. intensification distribution o f selv b. binding theory distribution o f them atic sig c. argum ent structure, voice, etc. distribution o f non-them atic sig Finally, section 4.6 sum m arizes the results o f the previous sections and concludes the chapter. 4.2 Testing the lexical/syntactic status of sig with neutral verbs In this section a num ber o f syntactic and sem antic tests will be used to sbow that sim ple sig occurring w ith neutral predicates is best analyzed as a theta-role receiving internal argum ent o f a transitive predicate. As defined in chapter 3, neutral predicates are predicates w hich are R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 238 sem antically com patible with both reflexive and non-reflexive scenarios, and w hich therefore are neutral w ith respect to what types o f nom inal expressions can occur as internal argum ents. The groom ing activity predicates in (13), exem plify proto-typical neutral verbs. (13) G room ing verbs: a. barbere ‘shave’ b. vaske ‘w ash’ c. frisere ‘dress one’s hair’ d. bade ‘bathe’ e. rede ‘com b one’s hair’ f. tarre ‘dry’ g- pynte ‘adorn’ h. klcede__pd ‘dress’ i. klcede__a f ‘undress’ j- qfluse ‘delouse’ k. barste ‘brush’ 1 . sminke ‘m ake up’ m. frottere ‘rub (w ith a tow el)’ Verbs denoting non-translational m ovem ent and verbs denoting transform ation also constitute core groups o f neutral predicates, see (14) and (15). (14) (15) M ovem ent verbs: a. strcekke b. bevcege^ c. flytte d. dreje e. h0 je Transform ation verbs: a. forvandle b. udvikle c. cendre d. forandre ‘stretch’ ‘move (non-translationally)’ ‘move (translationally)’ ‘tu rn ’ ‘bend’ ‘transform ’ ‘develop’ ‘change, alter; m odify’ ‘change, alter’ There are also a num ber o f neutral verbs falling into neither o f these categories, see (16). ^ Only DPs denoting inalienably possessed object can occur as internal arguments of bevcege ‘move’, see (i). In c o n t r a s t , ‘move allows both alienable and inalienable objects, see (ii). (i) Peter bevcegede hdnden/foden/stolen/glasset. ‘Peteri moved his hand/hisj foot/*the chair/*the glass.’ (ii) Peter bevcegede hdnden/foden/*stolen/*glasset. ‘Peterj moved his hand/his, foot/the chair/the glass.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 239 (16) a. befri ‘liberate’ b. forsvare ‘defend’ c. redde ‘save’ d. etc. Finally, the class o f neutral predicates also include all the “hidden neutral predicates” w hich tend to pattern as “anti-reflexives” (i.e. disallow ing simple sig) although they do not presuppose (representational) non-identity o f their argum ents, see discussion o f koge ‘boil’ in chapter 3, section 3.3.2.3. By using syntactic and sem antic tests show ing that neutral verbs allow locally bound them atic sig we argue against analyses w hich are - in one way or the other - based on the alleged anti-locality o f sig (e.g. V ikner 1986, etc.), as well as analyses based on the assum ption that all neutral verbs are lexically am biguous (e.g. Reinhart & Reuland 1993). 4,2.1 Testing for agentivity Testing for agentivity is one way to determ ine w hether a given predicate is causative transitive or decausative inchoative/unaccusative. A s observed by W altereit (1999:269), the French transitive verb tuer ‘kill’ can be used with the reflexive se ‘R EFL’ in tw o different ways, i.e. (i) as a causative transitive verb (17b,c), or as (ii) a deagentive inchoative (17a). (17)^ a. Pierre s ’ est tue dans un accident. ([-agentive], unintentional death) Pierre REFL is killed in an accident ‘Pierre has died in an accident.’ b. Pierre s 'est tue. ([+agentive], intentional death) Pierre REFL is killed ‘Pierre killed h im s e lf/’Pierre com m itted suicide.’ c. Pierre a tue le juge. ([+agentive], intentional death) Pierre has killed the judge ‘Pierre has killed the ju d g e.’ In (17a) se tuer occurs in a context w hich excludes agentivity, i.e. the adverbial phrase dans un accident ‘in an accident’, and as consequence the verb tuer+se takes on the deagentive * These examples in and the following discussion thereof is closely inspired by Waltereit (1999). R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 240 inchoative interpretation ‘(accidentally, non-intentionally) die’. In the default out-of-the-blue context in (17b) the tuer+se is usually interpreted as an agentive, causative transitive verb m eaning ‘(intentionally) kill (som ebody/oneself)’ ju st like the non-reflexive scenario tuer+D? in (17c). In French, the difference betw een the true them atic reflexive (17b) and the deagentive inchoative uses o f verb non-them atic se (17a) can usually only be determ ined by the presence vs. absence o f a linguistic context which excludes agentivity, e.g. dans un accident ‘in an accident’ in (17a). The [+/-agentivity] distinction can also be used as a diagnostic for deagentive inchoative verbs in D anish (18a) w hich have causative transitive counterparts (18b,c). (18) a Peter slog sig (pd bordkanten). ([-agentive], unintentional event) Peter hit REFL on table-edge ‘Peter (accidentally) hurt him self (on the edge o f the table).’ b. Peter slog sig selv. ([+agentive], intentional event) Peter hit REFL self ‘Peter (intentionally) hit h im se lf’ c. Peter slog Hans. ([+agentive], intentional event) Peter hit Hans ‘Peter (intentionally) hit H ans.’ N otice that unlike French, w hich uses the sam e form o f the reflexive, viz. se ‘R EFL’, in both the deagentive inchoative sentence in (17a) and the causative transitive reflexive construction in (17b), D anish tend to use the simple form sig ‘R EFL’ in the deagentive inchoative sentence in (18a) and the com plex form sig selv ‘REFL s e lf in the causative transitive reflexive construction in (18b). The preference for sig selv over simple sig in causative transitive reflexive constructions involving the predicate sld ‘hit’ is due to the fact that this particular predicate is a so-called “hidden neutral” predicate, i.e. a predicate w hich - although not sem antically incom patible with reflexive scenarios - is m ost often used as an other-directed predicate and thus triggers the expectation that its argum ents denote (representationally) different entities. The verbs sld ‘hit’ in (18) and hrcende ‘burn’ in (19) thus belong to a subset R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 241 o f verbal predicates w hich display the sam e m eaning difference between verb+5/g^ (= unintentional, [-agentive]), and \erh+sig se/v/DP (=intentional, [-l-agentive]) as the French exam ples in (17a-c) discussed above. (19) a Peter brcendte sig pa ovnen. ([-agentive], unintentional event) Peter burned REFL on oven-the ‘Peter (accidentally) burned him self on the oven. b. Peter brcendte sigselv. ([+agentive], intentional event) Peter burned REFL self ‘Peter (intentionally) burned him self.’ c. Peter brcendte Hans. ([+agentive], intentional event) Peter burned Hans ‘Peter (intentionally) burned H ans.’ N otice furtherm ore that, although com plex sig selv is preferred w ith the hidden neutral predicates sld ‘hit’ and brcende ‘burn’ in (18-19b), it is possible to find agentive, transitive uses o f these predicates w ith simple sig. A dding an instrum ental (and/or locative) phrase to the sentences in (1 8 -19b) makes sim ple sig m ore acceptable even w ith the agentive reading, com pare (18b) vs. (20a) and (19b) vs. (20b). (20) a. Peter slog sig (i hovedet) (med en hammer). (agentive, intentional) Peter hit REFL (in head-the) w ith a ham m er ‘Peter (intentionally) hit him self (in the head) w ith a ham m er.’ b. Peter brcendte sig (pd armen) (med en cigaret). (agentive, intentional) Peter burned REFL (on arm -the) w ith a cigarette ‘Peter (intentionally) burned him self (on his arm) with a cigarette.’ Exam ples like (18-20) illustrate the difference betw een deagentive inchoative verbs denoting unintentional events and their causative transitive counterparts which denote activities involving tw o entities, usually an A GENT doing som ething to a PA TIEN T/TH EM E. Since the sig in (18-19a) occurs w ith the intransitive predicates sld/brcende ‘hit/burn’, it is non- them atic and should be analyzed as in (2b). In contrast, the occurrences sig selv'* (18-19b) and It has been claimed that the XP status of intensified reflexives (e.g. sig selv) assures that they are always thematic, see chapter 3 (cf, also Jakubowicz 1994). See, however, section 4.3.3 where instances o f non-thematic intensified reflexives are discussed. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 242 sig in (20a-b) are them atic and receive a theta-roles from the transitive versions o f the predicates sld/brcende ‘hit/hurn’. As m entioned ahove, Holm herg (1984) and V ikner (1985) assum e sim ple sig to he anti-local. That is, it can only he them atic when LD-hound. For them all instances o f local sig thus have to be non-them atic. Consequently, in their system, only com plex reflexive (e.g. sig selv) can he locally hound. They thus explain the difference between (18a,19a) and (18h,19h) by reference to a difference in them atic structure. In (19h,19h and (18c,19c) the subject DP Peter receives the A G EN T theta-role and the internal argum ents sig selv/Hans receive the TH EM E theta-role. In the (a) cases the verbs brcende/sld only have one theta-role to assign, i.e. the TH EM E theta-role assigned to the subject DP Peter. In other words, their analysis is based on the assum ption that local sig is “some sort o f detransitivizing elem ent that som ehow prevents the assignm ent o f the theta-role that would otherwise have been assigned to the subject. In accordance with B urzio’s generalization (cf. e.g. Burzio (1981), Haegem an (1985)) this in turn prevents the deep structure object from getting object case, and it therefore has to m ove to subject position to be case-m arked, taking along its object theta-role, in this case THEM E, parallel to the analysis o f passive in LOB: 124ff.”(Vikner (1985:50, footnote 8)). W hile we agree that an analysis along these lines may account for the difference between (18/19a) vs. (18/19b,c) we strongly disagree with their claim that it may be extended to all cases o f local sig. As shown in (20), agentive readings o f the predicates sld/brcende ‘hit/burn’ can be found even w ith locally bound sig. A dditional counter-exam ples to their theory are provided by num erous cases of^^/g^ with neutral predicates, e.g. vaske ‘w ash’ in (21). (21) a. Peter vaskede sig. ([-l-agentive], intentional event, [-contrast]) b. Peter vaskede sig selv. ([-l-agentive], intentional event, [-l-contrast]) c. Peter vaskede Hans. ([-Hagentive], intentional event) Peter w ashed REFL/REFL self/Hans ‘Peter w ashed 0/him self/H ans.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 243 The difference betw een (21a) and (21b) is not a difference o f agentivity/intentionality - as would be predicted by V ikner’s (1985) approach - but rather one o f [+/-contrastive focus]^ or [+/-doppelganger-effect]. In other words, H olm berg’s and V ikner’s reanalysis account o f sla sig and brcende sig cannot be extended to the neutral verb vaske ‘w ash’ w hich has [+intentional, +agentive] readings w ith both the sim ple reflexive sig and the com plex reflexive sig selv. The sentence in (21a) does not m ean that Peter accidentally experienced a w ashing event that som ehow ‘happened’ to him. Proof o f this is provided by the fact that while the sentences in (21a-c) are all com patible w ith adverbials like med vilje ‘deliberately’ or purpose clauses \\k& fo r at ‘in order to ’ only the (b) and (c) exam ples in (18) and (19) com bine felicitously w ith such expressions. V ikner (1985) acknowledges that such exam ples constitute a potential problem for his approach. In his analysis local sig is a non-them atic, detransitivizing elem ent w hich reduces the num ber o f theta-roles assigned by a predicate by one. W hile this analysis is justified for the exam ples in (I8 a) and (19a), there is no evidence supporting its extension to cases like (21) and (22). (22) a. .... at PeteriAG Rsatte sig, ned] b. sigi selv c. Michael^ ‘.. .that [s Peter AG R set ____ dow n]’ V ikner’s account o f (22) is as follows: “the difference in m eaning is that w here [(22a)J means straightforw ardly “Peter sat dow n”, [(22b)] m eans som ething like “Peter sat him self dow n” (pragm atically a bit odd, as if he had set him self dow n by m eans o f e.g. a crane). This difference would seem to be accounted for by assum ing the follow ing theta-roles: in [(22b)] Peter is A G EN T, sig selv is TH EM E (parallel to [(22c)]), whereas in [(22a)] Peter is AGENT, and there is no TH EM E as such. If [(22a)] is com pared to [(18/19a)], they have in com m on that sig ensures that only one theta-role is assigned, but they differ in w hich one it is. This difference is supported by my intuition that [(22a)] which has an A G EN T m ay be follow ed by a purpose clause (e.g. “in order to impress everyone in the room ”), w hich is not possible for the A G EN T-less [(18/19a)]. Sim ilarly [(21a)] ^ For discussion of adnominal intensification o f reflexives used to express contrastiveness or doppelganger-effects, see chapter 3, section 3.3.2.5 and 3.3.2.7 respectively. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 244 but not [(18/19a)] may contain an adverbial like med vilje “deliberately”.” (V ikner (1985:50-51,footnote 9) V ikner’s analysis seems rather ad hoc. W hile assum ing that the reduction in arity and the absence o f agent in (18a) and (19a) to be due to B urzio’s generalization seem s well justified, V ikner does not provide any evidence supporting the assum ption o f a different arity reduction rule responsible for the alleged absence o f the TH EM E rather than the A G EN T role in (22a). G iven w hat we now know about doppelganger-effects, it seem s quite obvious that the m eaning difference betw een (22a) and (22b) is due to the doppelganger reading triggered by the intensified reflexive sig selv, not ad hoc deletion o f the TH EM E theta-role. This analysis thus elim inates the need for stipulating any additional ad hoc arity-reduction rule responsible for elim inating the TH EM E theta-role while preserving the A GENT theta-role. Just like w ith any other transitive neutral verb, e.g. vaske ‘w ash’ in (21), the theta-grid o f scette _ tied ‘s i t __ dow n’ rem ains the sam e (i.e. AGENT, PA TIEN T/TH EM E) regardless o f w hat kind o f nominal expression occupies the position o f direct object, i.e. in all the sentences in (22a-c), as well as in (21a-c), the predicate assigns an AGENT theta-role to the subject and a TH EM E theta-role to the direct object. In other words, there is no special arity reduction rule at work in (22a). The sim ple reflexive sig sim ply receives the theta-role TH EM E while the subject gets the A G EN T theta-role. Sim ilar argum ents can be leveled against Lidz (1996) who assum es local sim ple sig to be a non-them atic, verbal reflexive. Like Vikner, Lidz runs into trouble with the neutral predicates which, as shown above, allow locally bound them atic sig. O n the basis o f such exam ples we conclude that neutral predicates (e.g. vaske ‘w ash’, etc.) rem ain agentive causative transitive predicates even when the sim ple reflexive sig ‘REEL’ occupies the position o f direct object. In other words, in spite o f w hat has been tbe accepted account for decades, sim ple sig CAN be an tbeta-role receiving argum ent even when it is locally bound by a co-argum ent. R e p r o d u c e d w ith p e r m is s io n of t h e c o p y rig h t o w n e r. F u r th e r re p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 245 4.2.2 Testing for compatibility with formation of agentive nouns in -er Form ation o f agentive nouns from verbal roots has som etim es been used to test w hether a verb is unaccusative, e.g. Evereart (1986). It has been claim ed that verbs taking the simple reflexive, e.g. D anish sig, D utch z/c/z, etc., also do not allow -~er affixation, thus indicating that they are unaccusative (cf. Lidz (1996:132) follow ing Evereart (1986)) Lidz (1996) bases his argum ent on exam ples such as (23)-(26) w hich show that unaccusative verbs disallow -e r suffixation, while transitive and unergative verbs allow it. (23) U naccusative: a. Han er faldet. he is fallen ‘He fell.’ (24) T ransitive: a. Han spiser. he eats (25) U nergative: a. H an danser he dances (26) a. Han forspiste sig (i salat). he over-ate REFL in salad ‘He overate on the salad.’ *En falder. a faller b. en (hvidlags)spiser a (garlic)eater b. en danser a dancer b. *en forspiser an overeats Lidz claim s that verbs taking sig behave like unaccusatives by disallow ing the form ation o f - er nouns. W hile this generalization turned out to be true for the particular verbs he tested, e.g. forspise sig ‘overeat’ in (26), it does not hold for the neutral verbs in (27), w hich allow form ation o f agentive nouns in -e r in spite o f the fact that they can all be found with locally bound sim ple sig, see (28). (27) a. en vasker ‘a w asher’ b. (n0gen-)bader ‘a (naked-)bather, a skinny-dipper’ c. en afluser ‘a delouser’ d. en (falck-)redder ‘(lit.) a saver, i.e. a salvage-corps m an’ e. en befrier ‘a liberator’ f. en forsvarer ‘defender, defense player (football) defense law yer’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 4 6 (28) Peter vasker/bader/afluser/etc. sig. Peter w ashes/bathes/delouses REFL ‘Peter w ashes/bathes/delouses/etc. (him self)’ N ow , if one accepts the lexical am biguity hypothesis (cf. Reinhart & Reuland (1993)), one could object that the verbs in (27) all have transitive forms and that the -e r nouns in (30) are form ed from the transitive form o f the verb rather than the inherently reflexive form o f the verb, w hich is the one found in (28). The problem with this explanation (besides the obvious problem o f having recourse to unm otivated lexical am biguity) is that it does not work for verbs like (29), w hich do allow the form ation o f -e r nouns, see (30), in spite o f their inherently reflexive nature, i.e. they allow nothing other than sig in object position, see (29). (29) Peter bosatte *0/sig/*sig selv/*Hans iAfrika. Peter settled REFL/*REFL self/*FIans in Africa. ‘Peter settled 0/*him self/*H ans in A frica.’ (30) Peter er en a f de indiske boscettere i Afrika. Peter is one o f the Indian settler-s in A frica ‘Peter is one o f the Indian settlers in A frica.’ In other words, the -e r test does N O T allow us to conclude that all verbs taking local sig behave like unaccusatives. W hile it appears to work for the verbs o f the type illustrated in (26), it does not work for (27-30)^. Furtherm ore, as discussed in section 4.3.2, the fact that a given verb does not allow form ation o f agentive nouns in -e r does not necessarily m ean that it is unaccusative. M any anti-reflexive transitive verbs do not allow form ation o f agentive nouns in — er, e.g. mistcenke ‘suspect’, *e« mistcenker ‘*a suspecter’. In other words, there is no guarantee that the reason why a given verb cannot form agentive nouns in -e r is because it is unaccusative. ^ The question of how to analyze -e r nominalizations is also discussed in chapter 7. Agentive nouns o f anti- reflexive verbs are always non-reflexive, e.g. morder ‘murderer’, unless the element selv is present to indicate a reflexive scenario (e.g. selvmorder ‘self-murderer, a person who has committed suicide’). Agentive nouns of transitive inherently reflexive nouns are rare. Those we have found are not compatible with the selv- prefix, e.g. boscetter ‘settler’ vs. *selvbosxtter ‘self-settler’. Notice that unlike -e r nouns derived from anti-reflexive verbs, e.g. morder ‘murderer’, boscetter ‘settler’ only has reflexive readings (i.e. refers to someone who settles (himself) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 24 7 4.2.3 Non-compositional meaning change and the question test Neutral verbs have the same m eaning when they are used in reflexive constructions as when they are used in non-reflexive transitive contexts, com pare (31a) and (31b). (31) a. Peter vasker/barber er Hans. (neutral) ‘Peter w ashes/shaves H ans.’ b. Peter vasker/barberer sig. (neutral, them atic sig) ‘Peter w ashes/shaves (him self).’ That is, the m eaning o f the verb+reflexive in (31b) can be derived com positionally from the m eaning o f the verb and the m eaning o f the reflexive sig in the sam e way as the m eaning o f the verb+D P in (31a) can be derived com positionally from the m eaning o f the verb and the m eaning o f the DP in the object position. In contrast, the anti-reflexive predicate bcere ‘carry’ does not have the sam e m eaning in (32a) and (32b). (32) a. Peter bar kufferten. (anti-reflexive) Peter carried suitcase-the ‘Peter carried the suitcase.’ b. Peter bar sig med star vcerdighed. (anti-refl. reanalyzed as inher. refl., Peter carried REFL w ith great dignity non-them atic sig) *(i) ‘Peter<real> carried him self<statue>.’ *(ii) ‘Peter<real> carried him self<real>.’ (M ilnchhausen story) “'‘(iii) ‘Peter behaved carried him self/behaved w ith great dignity.’ c. Peter bar sig selv. (anti-reflexive, them atic sig) Peter carried REFL self °'‘(i) ‘Peter<real> carried him self<statue>.’ (ii) ‘Peter<real> carried him self<real>.’ (M unchhausen story) *(iii) ‘Peter behaved carried him self/behaved w ith great dignity.’ In (32a) the predicate boere ‘carry’ occurs in its normal use as an anti-reflexive transitive verb m eaning to ‘carry (som ething/som ebody)’. In (32b) it is used together with the reflexive sig in the m eaning ‘carry oneself; behave’. Unlike (31b) this m eaning cannot be obtained com positionally from the basic m eaning o f the verb+reflexive pronoun. Such non- com positional m eaning differences have often been used as argum ent in favor o f assum ing that the wevh+sig com bination has been reanalyzed as an inherently reflexive verb or, alternatively, somewhere) and cannot have non-reflexive readings (e.g. someone who makes others settle somewhere). The R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h f o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 248 that bcere is lexically am biguous betw een a transitive verb with the m eaning ‘carry’ and an inherently reflexive w ith the m eaning ‘behave.’ An additional argum ent in favor o f assum ing bcere sig ‘carry oneself; behave’ to be reanalyzed as a reflexive verb is that, unlike the reflexive pronoun sig in (31b), the sig in (32b), is not them atic. In both (32a) and (31 a,b) Peter is actually w ashing an object. In (31b) this object being w ashed is identical to the w asher him self; in (31a) it is different. In (32a) the direct object DP kufferten ‘the suitcase’ serving as direct object o f bcere ‘carry’ does refer to a particular entity in the universe o f discourse, nam ely a particular suitcase being carried by Peter. In contrast, the reflexive pronoun sig in (32b) does not refer to any specific entity in the universe o f discourse; in particular it clearly does not share the reference o f its antecedent Peter. Had this been the case the sentence would have m eant that Peter carried him self (on his back), as in a M unchhausen story, see (32c), w hich is clearly not the m eaning o f (32b)^. W altereit (1999:260) proposed to use questions to test w hether sig is them atic or non- them atic in a given verb+s/g com bination. W hen the referent o f the reflexive can be questioned, as in (33), then the reflexive is them atic. Conversely, when replacem ent o f the reflexive with a question word yields unacceptable sentences then the reflexive is non- them atic, see (34). (33)* Marie se regardedans la glace. Elle regarde qui? M arie REFL looks in the m irror she looks whom ? ‘M arie looks at herself in the mirror. W hom does she look at?’ (34) Jean-Luc s 'est rendu a la gare. #11 a rendu qui? Jean-Luc REFL returned to the station He has returned whom ‘Jean-Luc w ent to the station. #He returned w hom ?’ inherently reflexive predicate boscette ‘settle’ is also discussed in section 4.3.2. ^ For more discussion of the relationship between Milnchhausen readings/doppelganger-effects and intensified reflexives, see chapter 3 section 3.3.2.7. * These examples are adapted from Waltereit (1999:260), examples (2) and (4). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 4 9 As shown by the exam ples in (35) and (36), the same question test, can be used to test the them atic status o f sig in Danish. (35) Marie vaskede sig i varmt vand. Hvem var det hurt vaskede? M arie w ashed REFL in warm water who was it she washed? ‘M arie w ashed (herself) in warm water. W ho did she w ash?’ (36) Marie bar sig m ed stor vcerdighed. #Hvem bar hurt med star vcerdighed? M arie carried REFL herself w ith great dignity who carried she w ith great dignity ‘M arie carried herself with great dignity. #W ho did she carry with great dignity?’ An inherent w eakness to this test lies in the fact that one can only ask questions about the parts o f a sentence w hich can be focused. Since inherently reflexive verbs only allow expressions that are coreferential w ith the subject to occupy the position o f direct object, it is in principle im possible to ask a m eaningful question about the referents o f the expression (which, due to the sem antic selections o f the predicate, will always be a reflexive pronoun) filling the object position o f an inherently reflexive predicate. 4.2.4 Testing for compatibility with impersonal passivization To further support the claim that verbs taking local sim ple sig behave as unaccusatives Lidz (1996) adduces exam ples such as (37-39) to show that, ju st like uaccusatives (37-38), “verbs w hich take also disallow im personal passivization [, see (39)]” (p. 133). The observation that unaccusative verbs disallow im personal passivization while unergatives allow it is due to Perlm utter (1978). L idz’s D utch exam ples have here been translated into Danish for ease o f exposition. (37) a. (38) a. Han laber. he runs ‘Fie runs.’ Han falder. he falls ‘He falls.’ b. Der hliver lahet. there becom es run ‘People are running’ *Der bliver faldet. there becom es fallen ‘People are falling.’ (unergative) (unaccusative) R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 250 (39) a. Han forspiste sig. b. *Der bliver forspist. He overate REFL there becom es overeaten ‘He overate.’ ‘ People are overeating. ’ On the basis o f such data, Lidz concludes that the occurrences o f sig in exam ples such as (39) are not anaphors but non-them atic, non-referential verbal reflexives. W hile we agree that this holds for the verbs in (39) it is certainly not the case for neutral verbs. Neutral predicates allow im personal passivization, see (40), although they also allow locally bound them atic sig, see (41). (40) Der blev vasket/badet/etc. there becam e w ashed/bathed/ ‘There was w ashed/bathed.’ (41) Peter vasker/bader/afluser/etc. sig. Peter w ashes/bathes/delouses REFL ‘Peter w ashes/bathes/delouses him self.’ These exam ples clearly show that im personal passivization cannot be used to prove that ALL verbs which allow local sim ple sig are unaccusative. The generalization that verbs taking simple sig also disallow im personal passivization could, o f course, be saved by assum ing that neutral verbs like vaske ‘w ash’ and bade ‘bathe’ have double-lexical entries, e.g. (i) transitive verbs, and (ii) unaccusative/intransitive verbs (cf. also Reinhart and Reuland (1993)). As already argued in sections 4.2.1-3, as will be argued in 4.2.5-10, there are a num ber o f facts w hich go against this assum ption and w hich are m ore readily explainable under the assum ption that neutral verbs like vaske ‘w ash’ and bade ‘bathe’ are entered only once into the lexicon, i.e. as causative transitive verbs consistently assigning the theta-role TH EM E to their internal objects (be they DPs, pronouns or the sim ple reflexive sig). The exam ple in (42), originally from Hellan (1988), w hich Lidz uses to argue that verbs w hich take sim ple sig are unaccusative, is rather curious. (42) ^Der blev vasket sig (a f manden). there becam e washed REFL (by m an-the) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 251 This exam ple appears to be flawed. If the point is to show that vaske ‘w ash’ is not a causative transitive verb w hich m ay allow sig as a true them atic anaphor in argum ent position, then one should not use a sentence like (42). Indeed if vaske ‘w ash’ is truly a causative transitive verbs in all its uses (i.e. even w ith sim ple sig) then one would expect it to be ungram m atical in im personal passive constructions W ITH the reflexive pronoun, since all causative transitive verbs behave in the sam e way, see (43). (43) a. *Der blev dolket sig (selv). (dolke ‘stab’ = anti-reflexive) there was stabbed REFL self b. *Der blev spist/vasket sig (selv). (yaske/spise ‘w ash/eat’ = neutral) there was eaten/w ashed REFL self In other words, rather than show ing that vaske ‘w ash’ behaves as an unaccusative verb, (42) shows that it behaves ju st like any other transitive verbs, see also (44), which furtherm ore underlines the parallels between vaske ‘w ash’ and the other transitive verbs. (44) a. Der blev spist/vasket. there was eaten/w ashed b. Der blev spist/vasket kartofler. there was eaten/w ashed potatoes Furtherm ore, as discussed in section 4.3.2, the fact that a given verb does not allow im personal passivization does not necessarily m ean that it is unaccusative. M any anti-reflexive transitive verbs, e.g. (45a), cannot felicitously form im personal passives, e.g. (45b) (45) a. Peter mistcenkte *sig/sigselv/M arie. Peter suspected REFL / REFL self / M arie ‘Peter suspected him self /M a rie .’ b. *Der blev mistcenkt. there was suspected In other words, there is no guarantee that the only reason a given verb cannot undergo im personal passivization is because it is unaccusative. So this test should be used with caution when applied to inherently reflexive verbs. That is, w hile it seems plausible to assum e that the R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 252 inherently reflexive verb forspise ‘overeat’ is unaccusative, the mere fact that it cannot form the im personal passive (39b) cannot be used as solid proof o f its unaccusativity. 4.2.5 Auxiliary-selection A further indication that .s/g-form verbs do not necessarily behave as unaccusatives comes from auxiliary selection in Danish. Unaccusative verbs take at vcere ‘to b e’ as auxiliary, see (46), w hile unergatives and transitive verbs take at have ‘to have’, see (47) and (48)“ ^ . (46) U naccusative predicates: Han er ankommet. he is arrived ‘He has arrived.’ Alan et al. (1995) provide a conveniently concise description of the use of the verbs have ‘have’ and vcere ‘be’ as auxiliaries in the formation of the perfect tense in Danish: “(i) Transitive verbs, and Intransitive verbs not expressing motion, use have as their auxiliary: Jeg har Icest bogen. ‘I have read the book.’ U har brugtpengene. ‘We have spent the money.’ Hun har ventetpa dig. ‘She has waited for you.’ Vasen har staet her. ‘The vase has stood here.’ Notice that have and vcere both use have as their auxiliary. Vi har haft en dejiigferie. ‘We have had a lovely holiday.’ Vi har vceret i Italien. ‘We have been to Italy.’ (ii) Intransitive verbs expressing some kind o f motion use vcere as their auxiliary (but see (iii) below): Brevet er forsvundet. ‘The letter has disappeared.’ De er kommet hjem. ‘They have come home.’ Hun er allerede rejst. ‘ She has already left. ’ Er han stukket af? ‘Has he run away?’ Notice that blive ‘be, become’ uses vcere as its auxiliary: Han er blevet professor. ‘He has become a professor.’ Det er blevet gjort. ‘It has been done.’ [...] (iii) Sometimes the same verb can combine with both auxiliaries. This is, for example, the case when a verb can be used both transitively and intransitively of motion: Jeg harflyttet bordet. (trans.) ‘1 have moved the table.’ De er flyttet til Arhus, (intrans.) ‘They have moved to Arhus.’ Han harflojet en helikopter. (trans.) ‘He has flown a helicopter.’ Han er flojet til Japan, (intrans.) ‘He has flown to Japan.’ In other cases, intransitive verbs of motion may express either an activity in the past (and use have) or a present state (and use vcere): Han er gaet 10 kilometer. ‘He has walked 10 kilometers.’ N uerhangdet. ‘Now he has gone/left.’ Hun har svommet over Kanalen. ‘She has swum the Channel.’ Hun er svommet vcek. ‘She has swum away.’ Note - The verb begynde [‘begin’] can be used with either have or vcere. Have is used when there is a direct object, e.g.: Jeg har begyndt overscettelsen. ‘I have begun the translation.’ Jeg er begyndt pd overscettelsen.'” (Alan et al. (1995:263-264)) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 253 (47) U nergative predicates: Han har talt. he has spoken ‘He has spoken.’ However, none o f the 5z^-form verbs, regardless o f w hether sig is them atic (48) or non- them atic (49), take at vcere ‘to b e’ as auxiliary; they all take at have ‘to have’. (48) C ausative transitive predicates: a. Peter har vasket sin bil. Peter has w ashed PO SSREFL car ‘Peter has w ashed his car.’ b. Peter har vasket sig. Peter has w ashed REFL ‘Peter has washed h im se lf’ (49) Inherentiv reflexive predicates: a. Peter har forspist sig i cebler. (com pare with (26b) and (39b)) Peter has over-eaten REFL in apples ‘Peter overate in apples.’ b. Peter har ofte skammet sig. Peter has often sham ed REFL ‘Peter has often been asham ed.’ As show n in (49), with respect to auxiliary selection inherently reflexive predicates, regardless o f w hether they have transitive counterparts (49a) or not (49b), all behave like unergative (47) and causative transitive predicates (48) w hich take have ‘have’, rather than like unaccusative predicates like ankomme ‘arrive’ in (46), w hich take vcere ‘be’. The auxiliary selection test for unaccusativity was originally developed for Rom ance languages, e.g. French w here unaccusative predicates, as well as all reflexive predicates, take the auxiliary etre ‘be’, while causative transitive predicates form com posite tense form s with the auxiliary avoir ‘have’. Clearly, the discrepancies between G erm anic (e.g. Danish exam ples discussed above) and Rom ance languages show that auxiliary-selection cannot be used as a cross-linguistically reliable test for unaccusativity’”. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 254 4.2.6 Ledrups tests A ccording to Lodrup (1999), there is clear evidence for the status o f sim ple anaphors as theta- role receiving direct objects with neutral predicates. In N orw egian, for instance, a referential seg ‘R E FL ’ (cognate o f Danish sig ‘R EFL’) can be m odified by the m odifier/ze/e ‘w hole’, see (50a-b), ju st like full lexical DPs (50c). (50) N orw egian: a. Hun vasker sig. She washes REFL ‘She w ashed (herself).’ b. Hun vasker hele sig. She washes w hole REFL ‘She w ashed all o f herself.’ c. Hun vasker hele bilen. She washes w hole car-the ‘She w ashed all o f the car.’ Furtherm ore, sentences with them atic seg do not allow existential constructions, in contrast to sentences with inherently reflexive verbs w hich do, see (51a) vs. (51b,c). (51) N orw egian: a. *Det vasket seg en dame i sjan. it w ashed REFL a lady in sea-the ‘There was a lady w ashing in the sea.’ b. D et lukket seg en dar. it closed REFL a door ‘A door closed.’ c. Det infannt seg en dam e pa kontoret. it appeared REFL a lady on offlce-the ‘A lady appeared in the office.’ Based on data such as (50) and (51) Lodrup concludes that local seg ‘R E FL ’ may function as a theta-role receiving internal argument. Due to m inor param etric variation the exam ples in (50)-(51) cannot be replicated in Danish. H owever, D anish and N orw egian (especially bokmal N orw egian) are very closely related languages. In alm ost all other respects, the distribution o f the sim ple reflexive seg/sig Cf. discussion o f auxiliary selection in Borer (2002) and Schein (2002) in G. Preyer and G. Peter (eds.) Logical Form and Language. Oxford University Press, Oxford. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 255 follow the sam e patterns in the tw o languages. Hence, the N orw egian data may be used as indirect evidence supporting the idea that local sig can be them atic with neutral verbs. 4.2,7 German sich is not anti-local Konig & Siem und (1999:63) also argue strongly against the idea that sim ple reflexives are anti-local: “To group these referential uses o f the sim plex anaphors together w ith the inherently reflexive verbs and analyze them as cases o f reanalysis and detransitivization is particularly absurd for a language like Germ an, where sich is practically the only reflexive m arker and is only com bined w ith the intensifier selhst in the m ost extrem e cases o f other- directed" predicates), see (52)].” (52) German: a. Paul stolperte iiber #sich/sich selbst. Paul stum bled over REFL/REFL self ‘Peter stum bled over h im se lf’ b. Maria tra f #sich/sich selbst. M aria m et REFL/REFL self ‘ M aria m et herself. ’ As illustrated in (53)-(54), verbs which are anti-reflexive or ‘hidden neutrals’ in Danish, may behave as neutral predicates in German. (53) D anish: Paul hader * sig/sigselv. (Danish, ‘hidden neutral’) Paul hates R EFL/REFL self ‘Paul hates h im se lf’ (54) G erm an: Paul hasst sich / sich selbst. (Germ an, neutral) Paul hates REFL/REFL self ‘Paul hates h im se lf’ It thus appears, as observed by Konig and Siem und (1999:64), languages may vary with respect to the kind and num ber o f predicates they classify as anti-reflexives: “ [...] There are languages like Dutch where sim plex zich is restricted to only a few verbs and there are R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 25 6 languages like Germ an, w here a reinforcing selbst is rarely needed” (K onig and Siemund, 1999:64). In other words, a certain am ount o f cross-linguistic variation w ith respect to w hich verbs are classified as anti-reflexive, neutral, and inherently reflexive will have to be assumed. In order to account for G erm an sich, Reinhart & Reuland (1993) w ould have to extend the lexical am biguity analysis arbitrarily to include alm ost all predicates. In contrast, the analysis o f anti-reflexivity in proposed in chapter 3 enables a m ore natural account o f such cross-linguistic variation. A t first glance the fact that a predicate like hate behaves differently in D anish and G erm an seems to constitute problem for the account o f intensifier-adjunction to reflexive pronouns given in chapter 3. If intensification o f reflexives is triggered by the sem antics o f the predicate (i.e. the presupposition o f non-representational identity o f internal and external argum ents) then one w ould expect a verb like hate to behave alike in different languages, and especially in the case o f closely related languages like D anish and German. At closer look it may be possible to find a natural explanation for such cross-linguistic variation. In chapter 3 a distinction was m ade between anti-reflexives and “hidden neutral predicates”, like koge ‘boil’, which appear to behave like anti-reflexives but w hich (unlike the true anti-reflexives) can be m ade acceptable w ith sim ple reflexives given the appropriate context. In the case o f true anti-reflexive predicates, e.g. stolpern iiber ‘stum ble over’ and treffen ‘m eet’ in (52), the im plicature o f non-representational identity o f the internal and external argum ents is conventionalized (i.e. triggered by the lexical m eaning o f the predicate) and cannot be overridden by context. That is, it is im possible to im agine a person m eeting him self or stum bling over him self w ithout assum ing some type o f doppelganger scenario (e.g. schizophrenia or wax m useum contexts). Hence, all languages which distinguish betw een unintensified and intensified reflexives like Danish and Germ an are predicated to require intensified reflexives w ith sem antically anti-reflexive predicates like (52)-(53) and (55). '' The original passage has “non-other-directed”. This is clearly a typo and has been emended as “other-directed”. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 257 (55) Danish: a. Peter snublede over *sig / sig selv. Peter stum bled over REFL / REFL self ‘ Peter stum bled over himself. ’ b. Peter tra f *sig/ sig selv. Peter m et REFL / REFL self ‘Peter met h im se lf’ In the case o f the “hidden neutral”, like hassen ‘hate’ in (54), the im plicature o f non- representational identity is triggered by pragm atically determ ined usage patterns (i.e. it is com m on/expected, but not logically necessary, for people to use a certain verb to express com m only other-directed actions) rather than lexical semantics. Hence it can be overridden in the right context, e.g. the science-fiction scenario discussed in chapter 3, section 3.3.2.3. The distinction between sem antically vs. pragm atically triggered presupposition o f non- representational identity makes it possible to explain why strongly anti-reflexive predicates require intensification o f reflexives in both Germ an and D anish w hile the “hidden neutral” predicates may vary cross-linguistically. This account is thus much less ad hoc than Reinhart and R euland’s (1993) account w hich do not explain why a certain predicates classified as anti reflexive rather than neutral or inherently re fle x iv e '\ A nother source o f cross-linguistic variation is due to the syntactic and prosodic properties o f the reflexive pronouns them selves. Danish sig is a syntactically free but prosodically bound form w hich cannot carry stress on its own and which has to be adnom inally intensified by se/v-adjunction in order to be heavy enough to carry stress, see Note that there are a number of problems for this account o f the differences between Danish and German, notably examples like (i) and (ii) which appear to constitute counter-examples to the generalization that semantically antl-retlexive predicates never allow simple sig, even in languages like German, which are a lot less restrictive w.r.t. to verbs which allow simple unintensified reflexives. (i) Paul mistcenker *sig / sig selv. (Danish, anti-reflexive) Paul suspects REFL/REFL self ‘Paul suspects him self’ (ii) Paul verdachtigte sich / sich selbst. (German, neutral) Paul suspects REFL/REFL self ‘Paul suspects him self’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 258 chapter 3, section 3.3.6.1. In contrast, Germ an sich ‘R E FL ’, is stressabie, see section 3.3.6.3.2. See also discussion o f stressability, focus, and intensification in chapter 3 section 3.3.6.3.4. The fact that Germ an reflexives can be focused w ithout adnom inal intensification, may be one o f the reasons why they m ay potentially obey anti-reflexivity and doppelganger requirem ents w ithout adnom inal intensification. See section 3.3.6.3.1-4 w here we discuss the hypothesis that focus (realized as stress) and adnom inal intensification are functionally equivalent, i.e. may occur in the sam e contexts w ith more or less the same m eaning contribution. O f the tw o hypotheses concerning the differences between Danish an G erm an the latter, i.e. the prosodic account also discussed in section 3.3.6.3.2-4 appear to be the most prom ising. However, due to space lim itations we leave the elaboration o f this topic for future research. 4.2.8 Constructions involving coordination of verbs'^. In this section we argue that data related to optional deletion o f (all but the last) instance o f simple sig w ith coordinated verbs can be used to support our claim that neutral verbs with simple sig are neither reanalyzed as inherently reflexive/unaccusative predicates, nor have double lexical entries. W hile coordinated anti-reflexive and neutral verbs can optionally delete all but the last instance o f sim ple sig, see (56), inherently reflexive verbs do not seem to be able to, see (57). (56) a. Peter bade vaskede (sig) og harherede *(sig) pa mindre end 5 minutter.. Peter both w ashed (REFL) and shaved REFL on less than 5 minutes We have also fond at least one native speaker who disagrees with Kdnig & Siemund’s judgment of (52a), finding the sentence perfectly acceptable even with simple unintensified sich. In other words, more works on German is needed in order to get to the bottom of these questions. This section is inspired by Jakubowicz (1994:209, (12,14)). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 5 9 ‘Peter both w ashed (him self) and shaved *(him self) in less than 5 m inutes’ b. Peter havde vasket (sig) og torret *(sig) pd mindre end 5 minutter. Peter had w ashed (REFL) and dried REFL on less than 5 minutes ‘Peter had w ashed (him self) and dried in less than 5 m inutes.’ c. Peter bad mig harbere (sig) og sminke *(sig). Peter asked me shave and m ake-up REFL ‘Peter asked me to shave and put m ake-up on him .’ (57) a. Peter soledede *(sig) og boltrede *(sig) p d stranden. Peter tanned REFL and frolicked REFL on beach-the ‘Peter tanned and frolicked on the beach.’ b. Peter havde solet *(sig) og boltret *(sig) p d stranden. Peter had tanned REFL and frolicked REFL on beach-the ‘Peter had tanned and frolicked on the beach.’ Indeed, inherently reflexive verbs with non-them atic sig are expected to behave in this way since the include sig as part o f their lexical form atives. In contrast, since the neutral verbs in (56) behave like transitives they are expected to be able to om it all but the last instance o f identical object DPs with coordinated verbs, com pare (56a) and (58a), w hich show that the simple unintensified sig behave ju st like the DP kartoflerne ‘the potatoes’ in such constructions. (58) a. Peter bade vaskede og skrcellede kartoflerne p d mindre end fern minutter. Peter both w ashed and pealed potatoes-the on less than five minutes ‘Peter both washed and pealed the potatoes in less than five m inutes.’ The fact that optional deletion o f all but the last instance o f sig is not allow ed in the exam ples in (57) indicate that these instances o f sig w ith the inherently reflexive verbs sole sig ‘tan ’ and boltre sig ‘frolic’ have a different status from the sig in the exam ples in (56). That is, the sig in (57) is not a true them atic anaphor filling the argum ent position o f tw o-place predicates, but rather a verbal reflexive left-adjoined to little v o f the vP projection''^. M ore exam ples o f the '''N ote, however, that even with inherently reflexive verbs sig is left behind when the verbs raise to C, thereby resulting in constructions where sig is separated from the verb by the intervening subject, see (ia-b). (i) a. Hvorfor boltrede Peter sig pa stranden? Why frolicked Peter REFL on beach-the ‘Why was Peter frolicking on the beach.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 26 0 im possibility o f deleting sim ple sig w ith coordinated inherently reflexive verbs are given in (59). (59) a. Peter brcekkede *(sig) og kom *(sig). Peter broke REFL and cam e REFL ‘Peter vom ited and recovered.’ b. Peter havde brcekket *(sig) og kommet *(sig). Peter have broken REFL and com e REFL ‘Peter had vom ited and recovered.’ The fact that neutral verbs taking sim ple sig can be coordinated w ithout repeating sim ple sig constitutes a serious problem for reanalysis accounts (e.g., Vikner 1985) and lexical am biguity accounts o f local sig (cf. Reinhart and Reuland 1993), w hich are both based on the assum ption that neutral verbs taking sim ple sig behave like unaccusative inherently reflexive predicates. However, as the above data clearly shows, neutral verbs continue to pattern with active causative transitive verbs even w hen they take sim ple sig. W hile optional deletion o f sig w ith coordinated predicates can be used to test w hether a verb is inherently reflexive, it should be noted that there are cases w here deletion o f sig w ith conjoined inherently reflexive predicates m ay be m arginally acceptable, or even quite felicitous, see (61a). (61) a. (??)P e/er havde tit bade skammet og cergret sig over sin mangel p d takt. Peter had often both sham ed and annoyed REFL over his lack o f tact ‘Peter had often been both asham ed and vexed over his lack o f tact.’ b. UPeter skammer og cergrer sig. Peter sham ed and chagrins REFL ‘Peter is asham ed and vexed.’ b. Hvorfor brcekkede Peter sig? Why broke Peter REFL Why did Peter vomit?’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 261 N otice however, that even in such cases, coordination o f the sam e predicates is much less acceptable in the present tense, see (61a), than in the com posite tense in (6 0 )'^ See also section 4.3.2 for m ore discussion o f the reliability o f this test. 4.2.9 The status of sig as indirect object of ditransitive verbs Di-transitive predicates can also be divided into the three m ain types, see (62)-(64). (62) A nti-reflexive ditransitive: a. Peter solgte *sig et hus. b. sigselv c. Mary ‘Peter sold _____ a house.’ (63) N eutral ditransitive: a. Peter kabte sig et hus. b. sig selv c. Mary ‘Peter bought _____ a house.’ (64) Inherentlv reflexive ditransitive: a. Peter anskaffede sig et hus. b. *sig selv c. *Mary ‘Peter a c q u ire d ____ a house.’ A ccording to the reanalysis/lexical am biguity approaches (cf. V ikner 1985, and Reinhart & Reuland 1993, Lidz 1996, etc.) one would have to argue that the verb in (63a) has som ehow been reanalyzed as an inherently reflexive ditransitive predicate, w hile it rem ains a ‘regular’ di-transitive predicate in (63b-c). N otice, however, that there is no evidence supporting such claims. That is, there is no detectable difference in the m eaning o f the verb kabe ‘buy’ in (63a) vs. (63b,c). N or are there any other indications that any ‘reanalysis/lexical am biguity’ m ight have taken place in (63a) but not in (63b,c). We therefore conclude that these exam ples constitute strong evidence against all varieties o f the ‘reanalysis/lexical am biguity’ Cf. the discussion of coordinated predicates and deletion of simple reflexives in French and Danish in chapter 3, section 3.2.3. In French the same generalization seem to apply, i.e. deletion o f se improves when the conjoined R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 262 approaches. In contrast, the proposal defended here, nam ely that the distribution o f intensified and unintensified form s o f the reflexive follow directly from a com bination o f predicate m eaning, pragm atics and w orld know ledge (see chapter 3), is strongly supported by the data in (62)-(64). That is, given w hat we know about selling transactions, nam ely that it is a proto typical anti-reflexive/other-directed activity, it is highly unexpected that anyone w ould sell anything to himself. Hence, the intensified form o f the reflexive is necessary to license (either through contrastive focus or doppelganger-effects (e.g. m em ory-loss or schizophrenia contexts)) the reflexive scenario, as show n in (62b). In contrast, buying is a neutral activity in that it is equally likely for som eone to buy som ething for him self as it is for him to buy som ething for som ebody else. Finally the im possibility o f having non coreferential DPs in object position w ith inherently reflexive di-transitive predicates like anskaffe ‘acquire’, see (64c), explains why the adnom inally intensified form o f the reflexive, i.e. sig selv, is not possible either, see (64b). 4.2.10 Non-co-argument bound local sig All the instances o f locally bound them atic sig discussed above have co-argum ent antecedents. D epending on how the local dom ain relevant for binding is defined, locally bound simple reflexives, w hich are not hound by a co-argum ent, can be argued to be found in tw o contexts in Danish: (i) locational FPs, see (65), and (ii) possessive constructions, see (66). (65) MaXi lagde bogen bag sig,. M ax put book-the behind REFL ‘M ax put the book behind him (self).’ In (65) relevant local dom ain for the anaphor is the tensed clause. The PP headed by bag ‘behind’ does not have a subject, so it does not qualify as binding dom ain. That local binding predicates are in the past participle, see example (13c) chapter 3, section 3.2.3. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 263 is possible in sentences like (65) falsifies the claim that simple reflexive pronouns in Danish are anti-local'^. (66) Peter, vasker sin, hund. Peter washes POSSREFL dog ‘Peter washes his dog.’ A ccording to the analysis o f binding proposed in chapter 3, both sig ‘R ELF’ in (65) and sin ‘PO SSR EFL’ in (66) are locally hound and thus constitute evidence supporting the claim that them atic sim ple reflexives can he locally hound. As argued in length in chapter 3, section 3.3.4 the existence o f a m orphologically specialized series o f reflexive possessives (e.g. Dan. sin ‘PO SSR EFL’) constitute a m ajor problem for predicate-based approaches to binding like R einhart and Reuland (1993). 4.2.11 Summary: Falsifying the standard analyses of as anti-local In this section we have shown that a num ber o f the tests w hich have been claim ed to show that predicates taking simple sig are unaccusative do not hold up to scrutiny. In som e cases, they turned up to he less reliable than previously assum ed, in other cases they turned out to show the opposite o f w hat they were originally intended to show. Furtherm ore, we have introduced new tests and phenom ena w hich indicates that local sig can he them atic anaphors (e.g. with neutral predicates, in certain PPs and in possessive constructions). Taken together, the data and tests discussed in sections 4.2.1-10 thus adds up to a strong refutation o f the claim that the sim ple reflexives are anti-local. 4.3 The status of sig with inherently reflexive verbs In this section we take a closer look at the so-called inherently reflexive predicates. D istrihutionally, they can only he found with the sim ple reflexive sig, see (67). ’ Cf. Lidz (1996:115-6) who uses similar arguments to argue against assuming Dutch zich ‘REFL’ to be anti-local. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 264 (67) Peter skynder sig / *sig selv / *Hans. Peter hurries REFL/REFL self/Hans ‘Peter hurries.’ The verbal predicates listed in (68-71) are all distrihutionally inherently reflexive in that they are only found in the 5/g-form'^. fdSl M alefactive verbs denoting actions w hich involve hurting oneself bv O V ERD O IN G a. something: forbygge sig ‘overbuild, build beyond one’s m eans’ b. fordriste sig ‘venture to, m ake so bold as to ’ c. forhaste sig ‘be in too great a hurry’ d. forkohe sig ‘overbuy (oneself), overpay’ e. fo rl0 fte sig ‘overstrain’ f. forregne sig ‘m iscalculate’ g- forslide sig ‘overw ork (oneself)’ h. forsluge sig ‘overeat (lit. over-sw allow )’ i. forslcehe sig ‘overstrain o n eself j- forsnakke sig ‘give oneself away (by revealing too m uch)’ k. forspise sig ‘overeat’ 1 . forsynde sig ‘offend, sin (against)’ m. fortale sig ‘make a slip o f the tongue, give oneself aw ay’ n. forvilde sig ‘lose one’s way, go astray’ 0. forvv0vle sig ‘get into a m uddle, bungled’ P- forcede sig ‘overeat’ q- overanstrenge sig ‘overexert o n eself (69) V erbs o f thinking, feeling: a. affinde sig med ‘com e to term s with, accept’ b. hekvemme sig til ‘strive to, endeavor’ c. besinde sig ‘collect oneself, regain one’s com posure’ d. bestrcehe sig *(for/pd) ‘strive to, endeavor to ’ e. betakke sig (for) ‘decline (w ith thanks)’ f. bryste sig (o f ‘(fig.) throw out one’s chest, strut, sw agger’ g- dr isle sig (til at) ‘venture to ’ h. dumme sig ‘make a fool o f oneself, make a blunder’ That is, verbs which have transitive counterparts have not been included, e.g. (ia-b). (i) a. Det can ikke betale sig at snyde. it can not pay REFL to cheat ‘Cheating doesn’t pay.’ b. Han betalte 1000 dollars fo r bilen. he paid 1000 dollars for car-the ‘Fie paid $1000 for the car.’ In many cases the existence/absence of transitive counterparts seems like an arbitrary criterion. The verb belobe sig til ‘amount REFL to’ seems to be of the same type as betale sig ‘pay, be worthwhile’ in (ia) above, in spite o f the fact that it has no transitive counterpart, e.g. *Det belober $100 ‘it amount $100’. In this respect, the majority o f all inherently reflexive verbs behave like the so-called deponent (s-form) verbs discussed in section 4.4.5. In other words, verbs like belobe sig til ‘amount to’ can be characterized as deponent j/g-fom verbs, which differ from s- form deponent verbs only in the overt realization of the reflexive element left-adjoined to the v of the vP projection. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 265 i. dy sig (for at inf) ‘behave oneself, contain o n eself j. flotte sig ‘spread oneself, do it in style’ k. forbarm e sig ‘take pity over’ 1 . forelske sig (i) ‘fall in love w ith’ m. forscette sig (at) ‘decide, resolve’ n. forgabe sig *(i) ‘fall in love w ith’ 0. opfore sig (godt/ddrligt) ‘behave w ell/m isbehave’ p. skamme sig ‘be asham ed’ (70) V erbs o f m ovem ent, location: a. alhuesigfrem ‘elbow one’s w ay’ (= resultative, see sect. 3.3.3) b. alliere sig med ‘ally oneself w ith’ c. befinde sig ‘be; feel’ d. begive sig ‘go, travel to ’ e. boltre sig ‘frolic, gam bol’ f. boscette sig ‘settle, set up house’ g. skynde sig ‘be in a hurry’ (71) “M iddles” (w ith not tr. counterpart, opp. lukke ‘close’) only [-animate] subjects: a. forgrene sig ‘branch of, ram ify’ b. ansamle sig ‘gather’ c. fortone sig ‘loom; fade out o f sight’ d. belobe sig (til) ‘am ount to ’ (comp, betale sig vs. betale (tr.)) The division o f the above inherently reflexive verbs in to four groups is im pressionistic rather than based on rigorously defined sem antic criteria. Furthermore, the lists in (68)-(71) are not exhaustive. Indeed, they w ere established sim ply by looking through all the verbs from A to F in a com prehensive Danish-English dictionary'* (+a few additional verbs discussed above). The verbs in (68-71) all take [+animate] subjects while the verbs in (71) stand apart by only allow ing [-anim ate subjects]. The verbs in (68) stand out both m orphologically and sem antically. M orphologically, they are all com pounds com posed o f a prefix (either for- (=Ger. ver-) or over- (=Ger, iiber-))'"’ + a verbal stem. Sem antically, they constitute a type o f negative benefactives or m alefactives, i.e. they denote actions/events w hich result in a negative experience on behalf o f the subject. The sem antic difference betw een (69) and (70) can be '*Vinterberg, H, and C.A. Bodelsen (1966) Dansk-Engelsk Ordbog. Gyldendalske Boghandel, Nordisk Forlag. Copenhagen, R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 266 defined, adm ittedly in rather vague/im pressionistic term s, as a difference between verbs involving cognitive activity on behalf o f the subject (69) and verbs involving physical m ovem ent/location o f the individual denoted by the subject (70). Rather than pursuing in any more depth the project o f discovering sem antically defined groups o f inherently reflexive predicates, let us instead return to the m ore im portant question o f the them atic status o f sig with inherently reflexive predicates. T hat is, is it a them atic anaphor w hich occurs in argum ent position o f an active transitive predicate (72a), or a non-them atic gram m atical m arker o f unaccusativity occurring in the v o f the voice-projection vP (72b). (72) a. Them atic s is : VP / \ .... V ’ / \ V sig b. N on-them atic s is ' . vP / \ v’ / \ / \ VP sig V / \ .... V ’ / \ V .... In the following, we argue that the vast m ajority o f inherently reflexive predicates have non- them atic sig (72b). In a sense, it w ould be m ore precise to refer to these verbs, e.g. skynde sig ‘hurry’, as deponent ^I'g-form verbs, see section 4.4.5 for discussion o f deponent 5-form verbs. In contrast, the small num ber o f inherently reflexive predicates w hich may be argued to have them atic sig (72a), e.g. hoscette sig ‘settle’, are sim ply transitive predicates w hose selection restrictions only allow reflexive anaphors to occur in object position. Which indicates that they probably are loan words from Lower German dialects which entered the language during the Hansa-era. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 26 7 4,3.1 Inherently reflexive predicates with non-thematic sig M ost o f the predicates in (68)-(71) have non-them atic simple reflexives rather than locally bound them atic sig. As show n in (73-74), in many cases, the inherently reflexive predicate (73-74b) is derived from either agentive (transitive or unergative) verbs by productive m orphological operations, e.g. in these cases by adding the prefix fo r- ‘over-’^“. (73) a. Han spiste (salat). (causative transitive) he ate salad ‘He ate salad.’ b. Han forspiste sig (i salat). (inherently reflexive) he over-ate REFL in salad ‘He overate on the salad.’ (74) a. Han talte. (unergative) he talked ‘He talked.’ b. Han fortalte sig. (inherently reflexive) he over-talked REFL ‘He m ade a slip o f the tongue/gave him self aw ay.’ Lidz adopts E veraert’s (1986) account o f these exam ples, arguing that there “is a sem antic effect corresponding to the m orphological change [between (73-74a) and (73-74b)]. The (b) cases have an interpretation in w hich “an involuntary effect ... has taken hold o f the agentive subject”(Everaert (1986:52)). This [..] suggest[s] that [5/g] may actually be a verbal reflexive [... and] that these verbs only have one sem antic argum ent. [..] In (73b), for exam ple [x/g] is not the object o f overeating. That is, the sentence does not mean that the subject ingested himself. Typically, the original verb undergoing such m orphological operations is intransitive [e.g. (74a)]. If, however, the original is transitive [,e.g. (73a)], the direct object is turned into a prepositional object, sim ilar to [..] anti-passives [, e.g. (73b) ...].”(Lidz 1996:130). O verall, we adopt this account for the predicates in (68-71). As m entioned above, and discussed at length in section 4.2.1, a num ber o f anti reflexive verbs have inherently reflexive/deagentive inchoative counterparts, e.g. Peter slar R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 268 Hans ‘Peter hits H ans’ vs. Peter slar sig ‘Peter gets hurt’. Interestingly, there are also a few inherently reflexive verbs w hich are derived from unaccusative verbs, see (75a) w hich is derived from (75b) and (76a, b). The fact that inherently reflexive verbs can be derived from unaccusative verbs is another indication o f the non-them atic status o f these occurrences o f simple sig. (75) a. Peter kom sig (inherently reflexive) Peter cam e REFL ‘Peter recovered (from a disease).’ b. Peter kom. (unaccusative) Peter came ‘Peter cam e.’ (76) a. Peter blcerer sig. (inherently reflexive) Peter blister REFL ‘Peter brags, shows o f f b. P eter’ s finger blcerer. (unaccusative) Peter’s finger blisters ‘Peter’s finger blisters.’ Unlike the inherently reflexive predicates derived by prefixation o f for- ‘over-’ in (73-74b), the m eaning o f the derived predicates in (75-76) do not appear to be related to the verbs from w hich they are derived in any system atic way. This indicates, that blcere sig ‘brag’ and komme sig ‘recover’ are lexicalized as inherently reflexive verbs. 4.3.2 “Inherently reflexive verbs” with thematic sig W hile it m ay be tem pting to conclude that all predicates labeled here as “inherently reflexive” take non-them atic sig, such a conclusion may be prem ature. First o f all, as discussed in chapter 3, the are a num ber o f constructions involving possessive reflexives w hich classify as inherently reflexive on sem antic criteria, see (77). Cf. Everaert (1986) for a similar analysis o f Dutch. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 6 9 (77) “Inherently reflexive” constructions a. Han var ved at gaa ud a f sit /? * sit eget/*Peters gode skind a f glcede. he was about to go out o f PO SREF/?*PO SREF ow n/Peter’s good skin o f happiness ‘He nearly jum ped out o f his good skin o f sheer happiness.’ b. Han hyttede sit / ??sit eget /*Peters skin. (cont. bet. a/b) he saved PO SREF/ ??PO SREF ow n/*Peter’s skin ‘H e saved his own life.’ As discussed in chapter 5, sim ilar inherently reflexive possessive constructions can be found in English, see (78). (78) a. He was out ofhis/*his own/*M ary’ s mind o f happiness. b. He said ahem several times to clear his/*his own/*M ary’ s throat. Clearly the possessive reflexive sin ‘PO SR EF’ in (77) and the possessive pronoun his in (78) cannot be analyzed as non-them atic gram m atical m arker o f unaccusativity occurring in the v o f the voice-projection vP, see (72b). Hence, as argued in chapter 3 section 3.3.5, the sem antic definition o f inherently reflexive predicates/constructions is still needed, see (80). (80) D efinition o f inherent reflexivitv: (K 6 5 ) chap. 3, sect. 3.3.2.4) A given predicate is inherently reflexive if it evokes a presupposition o f identity that is not cancelable by context. That is, for the so-called inherently reflexive possessive construction in (78)-(79), (80) is needed to explain why intensification o f the possessives sin/his is infelicitous. Due to the m eaning o f these predicates/constructions no contrast set o f alternatives can be generated. Hence, the contrastiveness condition blocks intensification in these cases. In addition to the possessive constructions discussed above, there are a few verbs w hich distrihutionally behave like inherently reflexive predicates (by only allow ing the simple unintensified reflexive sig/sin) but which, nevertheless, appear to have them atic sig, e.g. (8Ia,b). (81) a. Peter dukkede sig / *sig selv / Marie. Peter ducked REFL / REFL self / M arie ‘Peter ducked 0 / *him self / *M arie.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 7 0 b. Peter bosatte sig / *sig selv / * Marie i Afrika. Peter settled REFL / REFL self / M arie in A frica ‘Peter settled 0 / *him self / *M arie in A frica.’ The exam ples in (82)-(86) are designed to test the them atic status o f sig with the verbs dukke ‘duek’ and boscette ‘settle’ using the tests discussed in sections 4.2.1-8 above. (82) Testing for agentivitv tsection 4.2.11: a. Han dukkede sig fo r at undga at blive sldet i hovedet. he ducked REFL for to avoid to becom e hit in head-the ‘He ducked to avoid being hit in the head.’ b. Han bosatte sig i Argentina fo r at undgd retsforfolgelse i USA. he settled REFL in M exico for to avoid prosecution in the US ‘He settled in A rgentina to avoid prosecution in the U S.’ (83) Testing for com patibilitv with form ation o f agentive nouns in -e r (section 4.2.2): a. *en dukker *a ducker b. en boscetter a settler (84) Them atic vs. non-them atic j'/g (section 4.2.3): a. *Hvem dukkede Peter? W ho ducked Peter ‘W ho did Peter duck?’ b. */#Hvem bosatte Peter? W ho settled Peter ‘W ho did Peter settle?’ (85) Impersonal passivization (section 4.2.4): a. *Der blev dukket. there becam e ducked ‘There was ducked b. *Der blev hosat. there becam e settled ‘There was settled.’ (86) D eletion o f sis w ith coordinated predicates (section 4.2.8): a. (*)Peter dukkede og drejede sig rundt. Peter ducked and turned REFL around ‘ Peter ducked and turned around. ’ b. (?/ok)Peter bosatte og integrerede sig i det fremmede. Peter settled and integrated REFL in the foreign ‘Peter settled and integrated him self abroad.’ R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 2 7 1 The results o f the tests in (82-(86) are sum m arized in (88) which also includes the results o f the sam e tests run on different inherently reflexive predicates. (87) T ests: (i) A gentivity (section 4.2.1) (ii) A gentive -er) (section 4.2.2) (iii) Referentiality (section 4.2.3) (iv) Im personal passive (section 4.2.4) (v) C oordination (section 4.2.8) Inherentlv reflexive: (0 (ii) (iii) (iv) (V) a. forspise ‘overeat’ H = * * * * b. skamme ‘be asham ed’ * * (?) c. skynde ‘be in a hurry’ ok * * d. dukke^‘ ‘duck’ ok (*) (*) e. boscette ‘settle’ ok ok * * (?/ok) As show n in (88) the verbs classified as inherently reflexive (based on the fact that they allow only sig as internal argum ent) do not behave uniform ly with respect to the five tests. Com patibility w ith agentive phrases, test (i), and the agentive suffix -e r (ii) vary considerably. Only test (iii) and (iv) seem to apply uniform ly to all inherently reflexive verbs. The fact that some inherently reflexive predicates are com patible with agentive subjects (test (i)), and even allow form ation o f agentive nouns in -e r (test (ii)), may (but need not) be an indication that som e occurrences o f sim ple sig with inherently reflexive predicates are in fact true reflexive pronouns receiving an object theta-role from their predicates. In other word, we suggest that som e inherently reflexive verbs are really transitive neutral predicates whose selection restrictions exclude anything but them atic reflexive pronouns from occurring in object position. Based on these tests it may be argued that dukke ‘duck’ and boscette ‘settle’ have them atic sig and should be analyzed as in (72a) rather than as non-them atic sig in (72b). H owever, before any final conclusion is drawn let us first return to the issue o f the reliability o f these tests (already briefly discussed in sections 4.2.1-4,8). The tables in (89) and In addition to the simple reflexive sig the verb dutitie ‘duck’ can take one other DP as internal argument, viz. the inalienably possessed DP hovedet ‘the head’, see (i). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 272 (90) listed the result o f applying the sam e five tests in (87) to different anti-reflexive and neutral verbs. Anti-reflexive: (i) (ii) (iii) (iv) (V) a. hade ‘hate’ ok ok ok ok ok b. efterfalge ‘succeed’ ok ok ok * (?) e. banflade ‘im plore’ ok * ok (*) (ok) b. mistcenke ‘suspect’ * * ok * (ok) c. misunde ‘envy’ * * ok * (ok) N eutral : (i) (ii) (iii) (iv) (V) a. vaske ‘w ash’ ok ok ok ok ok b^^. klcede pd ‘ dress’ ok * ok ok c. hcenge ‘(kill by) hang(ing)’ ok ok * (*) d. forvandle ‘transform ’ ok * ok * (* ) As shown in (89-90), the neutral and anti-reflexive predicates do not form a hom ogenous group w ith respect to the five test. Some verbs, are incom patible with agentive phrases or form ation o f agentive nouns in -e r, e.g. mistcenke ‘suspect’ and misunde ‘envy’. Others cannot form im personal passives, e.g. forvandle ‘transform ’ and misunde ‘envy’. In other words, if even transitive anti-reflexive and neutral verbs do not necessarily pass these tests, then the failure to do so on the part o f a given inherently reflexive verb does not necessarily m ean that they have non-them atic sig. Though the question o f w hether sig with inherently reflexive verbs is them atic and referential or non-them atic and non-referential, is highly interesting in itself it is not crucial to the topic discussed in the present dissertation. Since our disagreem ent w ith the standard accounts (i.e. the reanalysis/lexical am biguity accounts) o f local sig can be settled sim ply by looking at the behavior o f sig w ith neutral predicates we do not need to answ er this question Note that even neutral verbs which can undergo optional i/g-deietion in some eontexts, e.g. barbere ‘shave’ in (i), cannot in other contexts, e.g. (ii). (i) Peter vaskede, barberede og torrede sig. Peter washed shaved and dried REFL ‘Peter washed, sheaved and dried himself.’ (ii) Peter barberede *(sig) og klcedte sig pa. Peter shaved and dressed REFL on ‘Peter shaved and dressed.’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 273 here and will sim ply leave it for further research. For sim plicity we will use the term “inherently reflexive” to refer to predicates/constructions which only allow the simple unintensified forms o f the reflexive, i.e. sig ‘R E FL ’ and sin “POSS R E FL ’, regardless o f the them atic status o f the reflexive. 4.3.3 Grammatical uses of inherently reflexive verbs with sig selv In the literature on binding in D anish it is often claim ed that the com plex reflexive sig selv ‘REFL s e lf is always them atic. W hile it is certainly true that sig selv is them atic in the vast m ajority o f instances, there do in fact appear to exist contexts in w hich non-them atic com plex reflexives are required. As observed by V eraart (1996), com plex reflexives, e.g. sig selv, can be used w ith unaccusative inherently reflexive verbs like skamme sig in m eta-linguistic or didactic contexts like (64). (91) a. Learner: * Peter skammer Marie. Peter sham es M arie ‘Peter sham es M arie.’ b. Teacher: Nej, Peter kan ikke .skamme andre. Han kan kun skamme sig selv. N o Peter can not sham e others he can only sham e REFL self ‘N o, Peter cannot sham e others he can only be asham ed.’ As shown in the previous section, skamme sig ‘be asham ed’ is one o f the clearer cases o f non- them atic sig. So if, skamme does in fact take non-them atic sig, then how come it is intensified in (91b)? We suggest that the answ er lies in the prosodic properties o f sig, i.e. the fact that sig is a clitic w hich cannot carry stress on its own. In (91b) sig selv seems to be licensed by the explicitly contrastive context. Since, no doppelganger-readings are available (due to the non- them atic status o f sig w ith skamme), and since the inherently reflexive m eaning o f skamme sig does not allow for the generation o f a focus-triggered contrast set, one has to argue that in (64b) it is sim ply the stress on sig that m otivates the presence o f selv. As discussed in chapter R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 7 4 3, section 3.3.6 the sim ple reflexive sig is prosodically/syntactically too w eak to receive stress on its own, consequently selv is adjoined to m ake the reflexive strong/heavy enough to carry stress. In other words, the sig selv in (91b) has all the characteristics o f being a prosodically m otivated instance o f the intensified reflexive sig selv. N otice that the existence o f com plex reflexives w ith inherently reflexive verbs cannot be explained w ithin the fram eworks o f V ikner (1985) and Reinhart and Reuland (1993), but receives a straightforw ard explanation in the present focus-based account o f adnom inal intensification. 4.3.4 Conclusion Based on the discussion o f inherently reflexive predicates in the preceding sections, we conclude that predicates which classify distrihutionally as “inherently reflexive” should be divided into tw o groups: unaccusative/deponent predicates w ith non-them atic sig in v o f vP transitive predicates with them atic sig as internal argument. M any “inherently reflexives” are lexically unaccusative, e.g. skamme sig, as opposed to derived unaccusative s/g-form predicates, e.g. abne sig ‘open’. Furtherm ore, there is probably only a few true, transitive, “inherently reflexive” predicates, e.g. boscette sig ‘settle’. In addition to the tw o types o f inherently reflexive predicates distinguished above, we also have inherently reflexive possessive constructions. The sem antic definition o f inherent reflexivity (80) thus only applies to verbs w ith them atic sig and possessive constructions w hich only allow the sim ple/unintensified possessive reflexive sin as possessor argum ent. Furtherm ore, only them atic sig and the possessive reflexive sin ‘PO SR E F’ both fall under the binding principle A discussed in chapter 3. The non-them atic sig found with m ost inherently reflexive R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 275 verbs is not a reflexive anaphors but rather a gram m atical m arker o f voice/unaccusativity w hich fall under a different m odule o f the grammar. As m entioned above, the question o f w hether sig with inherently reflexive verbs is them atic and referential or non-them atic and non-referential, is highly interesting in itself. However, it is not o f vital im portance to the topic discussed in the present dissertation, since, as shown in section 4.2, the argum ent that local sig can be them atic can be m ade on the basis o f neutral verbs alone. We therefore sim ply leave the question o f the exact status o f sig with different types o f “inherently reflexive” predicates for future research. 4.4 Late-insertion analysis of reflexives and pronouns in Danish In this section a late-insertion analysis o f Danish reflexives and pronouns is proposed (section 4.4.1), w hich has the advantage o f being able to account for and 2"‘ ‘ person pronouns and reflexives (section 4.4.2). The late-insertion analysis thus provides the m orphological com ponent o f the syntactic LF-m ovem ent (Cf. Pica 1984) analysis o f reflexives presented in chapter 3. 4.4.1 Lexical and syntactic properties of pronouns and reflexives Personal pronouns and both them atic and non-them atic reflexives all start out as the sam e unspecified determ iner, 0 ° {a person, a num ber, etc.}, see (92a-c(i)). Personal pronouns differ from reflexives in that their phi-features are specified before m erge, see (92a(ii)), w hich explains why they are able to refer on tbeir own and as well as why they are subject to principle B. Both them atic and non-them atic reflexives have their phi-features specified after merge, see (92b-c(ii)). Them atic and non-them atic reflexives get their features valued in different ways. For them atic reflexives, w hich merge into argum ent position w here they R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 7 6 receive their theta-roles (92b(iii)), it happens via binding relations, i.e. an A-chain created by m ovem ent o f reflexive from theta-position to T° (92b(v)). As argued in chapter 3, the subject- orientation o f Danish reflexives supports the assum ption that they move to T ”. N on-them atic reflexives m erge directly into (92c(iii)). W hile their features also ultim ately get valued by m ovem ent to T" they are not interpretable (92c(iv)) and are thus not bound by the subject but rather enter into a clitic-doubling chain w ith it (92c(v)). (92) Third person pronouns/reflexives: a. Pronom inal ham ‘him ’. (93a). (95c): (i) starts out as unspecified determ iner D° {a person, a num ber, etc.} (ii) phi-features are specified before merge (iii) inserted into argum ent position at merge, assigned theta-role (iv) phi-features are interpretable (v) no chain is form ed (principle B effects) (vi) morpho-Zphonological realization spelled out from valued features by phonological rule. b. Them atic sis ‘R EFL’, (93b): (i) starts out as unspecified determ iner D “ {a person, a num ber, etc.} (ii) phi-features are specified after m erge (iii) inserted into argum ent position at merge, assigned theta-role (iv) phi-features are interpretable (v) phi-features are valued via m ovem ent to T°, i.e. via A-chain. (vi) morpho-Zphonological realization spelled out from valued features by phonological rule. c. N on-them atic sis ‘R EFL’. (94a), (95a): (i) starts out as unspecified determ iner D° {a person, a num ber, etc.} (ii) phi-features are specified after merge (iii) adjoined to v at m erge, not assigned theta-role (iv) phi-features are non-interpretable but still need to be valued (v) phi-features are valued by subject (i.e. T°) via clitic-doubling chain^^ (vi) morpho-Zphonological realization spelled out from valued features by phonological rule. T h e sen ten ces in (9 3 -5 ) illustrate the pronom inal and reflex iv e elem en ts in (9 2 a -c). (93) a. Peteri vaskede sig,. (them atic sig, phi-feat specified after merge, Peter w ashed REFL A -chain, agentive, (92b)) ‘Peter w ashed him self.’ Cf. Clitic-doubling in Spanish: (i) Lo, vimos a Juan,. hinii see-lPL-PRES to Juan, 'W e see Juan.’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 7 7 (94) (95) b. Peter, vaskede HanSk. (them atic DP, phi-feat specified before Peter w ash-PA ST Hans ‘Peter washed H ans.’ m erge, no chain) c. Peter, vaskede hantk. (them atic ham, phi-feat specified before Peter w ashed him ‘ Peter washed him. ’ merge, no chain, (92a)) a. Daren, abnede sig/. (non-them atic sig (in v°), phi-features door-the opened REFL specified after merge, valued via clit- ‘The door opened.’ doubling, non-agentive, (92c)) b. Peter abnede daren. (them atic DP, phi-feat specified before door-the opened REFL m erge, no chain) ‘The door opened.’ a. HanSi slog sig,. (them atic sig (in v°), phi-feat specified after Hans hit-PA ST REFL m erge, valued via clit-doub., non-agentive. ‘Hans got hurt.’ (92b)) b. Peteri slog HanSk- (them atic DP, phi-feat specified before Peter hit-PA ST Hans ‘Peter hit H ans.’ m erge, no chain) c. Peter, slog hantk. (them atic ham, phi-feat specified before Peter hit him ‘ Peter hit him. ’ m erge, no chain, (92a)) As the exam ples in (93b-c), (94b) and (95b-c) show, the verbs vaske ‘w ash’, abne ‘open’ and sla ‘hit’ are all causative transitive verbs w hich assign theta-roles to DP and personal pronouns occurring as internal argum ents and take agentive readings when occurring in active sentences. The m ost interesting aspect o f (93-5) is the them atic status o f sig. A s already shown in 4.2.1 the verbs vaske ‘w ash’ and sla ‘hit’ in (93) and (95) are both neutral but differ in how they interact with sim ple sig. W hile the predicate vaske ‘w ash’ always takes them atic sig and thus always gives rise to agentive readings, sla ‘h it’ can take either them atic or non-them atic sig, com pare (20a) and (18a) in section 4.2.1. In the ab ove late-insertion an a ly sis the overt realization o f pronouns and r e flex iv es is determ ined by w here in the derivation (i.e. before or after m erge) the {3'^‘ * person, singular} features are specified, see (96). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 78 (96) Third person form s: O vert realization: a po person, sg., -specified before merge} => Isig/ b. D° {3'^'' person, sg., -l-specified before merge} => lham/ Features w hich are specified in the lexicon (i.e. before m erge) are per definition interpretable (e.g. pronom inals, and DPs). Features w hich are valued in syntax (via A -chain binding, or clitic-doubling chains) m ay or m ay not be interpretable depending on w hether or not they are features o f a them atic element. Them atic reflexives are always interpretable, and vice versa. Note, however, that being interpretable is not necessarily the same as being referential. All referential reflexives are interpretable, not all interpretable reflexives are referential, e.g. (97). (97) Enhver cegtemand, elsker sirii hone. every m arried-m an loves PO SS-REFL wife ‘Every m arried m an loves his w ife.’ Since sim ple sig is a clitic (i.e. sim ultaneously X ” and X P) it may be am biguous betw een them atic (in argum ent position) and non-them atic in v" o f vP. The vP is a voice/diathesis projection w hich determ ines w hether the predicate is an (active) agentive, causative transitive predicate (assigning tw o (or m ore) theta-roles, or a (passive), deagentive/non-agentive, intransitive predicate (only assigning one theta role, e.g. TH EM E to the internal object). (98)^"* In Danish v° has an EPP feature requiring the presence o f a D” elem ent. This D° elem ent may be realized in one o f the follow ing tw o ways: a. as a (them atic) D P in [Spec, vP], or b. as non-them atic sig m erging with v”. As show n in (99), (98a) gives rise to agentive, transitive predicates. (99) Peter dbner d0ren. Peter opens door-the ‘Peter opens the door.’ As shown in (100), (98b) gives rise to non-agentive, unaccusative predicates. (100) D0ven abnede sig. door-the opened REFL ‘The door opened.’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 7 9 All predicates start out as m ono-them atic (unaccusative) predicates assigning only one theta- role (TH EM E) to the internal object. W hen sig m erges in argum ent position it is assigned the internal theta-role by the predicate. In this case, the only way to satisfy the EPP feature specified in (98), is by letting a DP merge in [Spec, vP], see (98a). This DP will then get the be assigned the A G EN T theta-role o f the vP projection and the resulting predicate will be an agentive (causative) transitive predicate, see (93b). W hen sig m erges directly with v° the EPP feature (98) is already satisfied. Hence, the DP w hich merges in argum ent position o f the predicate is assigned the TH EM E theta-role and no other DP is allowed to merge in [Spec, vP]. The resulting predicate is therefore a non-agentive, unaccusative predicate, see (94a), (95a), and (100). 4.4.2 Binding theory in 1*' and 2" ^ person: the implications of person asymmetries The late-insertion account presented above is particularly good for analyzing T ’ and 2"‘ * person forms w hich otherw ise would necessitate the postulation o f form s w hich are am biguous betw een reflexives and pronouns (e.g. (102a) vs. (102b)). Languages like English w hich seem to distinguish betw een reflexives and object pronouns in the L‘ and 2"‘ ‘ person forms, see (101), are fairly rare. (101)^^ a. Ii wash myself, / 0 ,. b. Peter j washes *myself / Danish, like m ost other languages in the world, does not distinguish betw een pronom inal and reflexive form s in fhe L' and 2'"* persons^®. That is the distinction between the ham ‘him ’ and sig ‘REEL’ is only found in the 3 '^ ‘ ‘ person, see (102-3). T° features are the number and person endings found (amalgamated) with tense endings in inflectional languages). Unlike Defeatures, T“phi-features are never interpreted. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 8 0 (102) a. Peter, vaskede mig/,. (+th mig, phi-feat specified before m erge, no chain) Peter w ashed me ‘Peter w ashed m e.’ b. Jegi vaskede mig,. (+th mig, phi-feat specified after merge, A-chain) I w ashed me ‘I w ashed m e.’ (103) a. Jeg, slog migi. (-th mig (in v°), phi-feat specified after merge, I hit-PA ST me clit-doubling, non-agentive) ‘I got h urt.’ b. Peteri slog migk.{+ih mig, phi-feat spec before merge, no chain) Peter hit-PA ST me ‘Peter hit m e.’ Syntactically the first and second person forms can be divided into the same three types as the third person forms in (92). Pronom inal mig ‘m e’, w hich can refer independently and w hich is subject to principle B in the same way as ham ‘him ’, is illustrated in (102a) and (103b). The sentences in (102b) and (103a) illustrate them atic reflexive mig ‘m e’ and non-them atic mig respectively. Since the overt distinction betw een pronom inal and reflexive first and second person form s is neutralized, the relevant phonological rules do not have to be sensitive to the point at w hich features are values - w hich was necessary in the rules for the third person forms, see (104). (104) Phonological spell-out o f first and second person reflexives/pronouns: a. {r ’ person, sg.} b. {2"‘ ‘ person, sg.} c. { person, pi.} d. {2"'* person, pi.} => mig => dig => O S => Jer The differences between I wash and I wash m yself will be taken up in chapter 5 where it will be argued that English has 0-reflexives, i.e. the sentence / wash is really I wash 0 , and that I wash myself \s really I wash 0 m yself with and adnominally intensified 0-retlexive. There is a functional motivation often evoked for the absence o f morphological differentiation of P ‘ and 2 " " * person reflexives and pronouns, i.e. the absence of potential ambiguity between coreference and disjoint reference readings of T* person and 2"'* person forms. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 281 4.5 Reflexive and non-reflexive uses of reflexive elements: How many sig's do we have? The purpose o f this section is tw o-fold. First, it contains a survey o f all the so-called non reflexive uses o f reflexive elem ents in Danish. This survey will show that the sim ple reflexive sig (and its counterparts in other languages, e.g. Ger. sich, Fr. se, etc.) has a num ber o f uses outside its use as reflexive anaphor. In contrast, English x-self and Chinese ziji ‘self-self have only reflexive uses in addition to their uses as intensifiers. This difference will be used, in chapter 5-6, to bolster the argum ent that while D anish sig is a true reflexive, English x-self and Chinese ziji are always intensifiers. Second, we show that the reflexive elem ent -s in Danish, is functionally equivalent to non-them atic sig in certain contexts, e.g. deponent verbs. In m any languages w hich have reflexive elements^^ these may have m any different uses, e.g. reflexives, reciprocals, m iddles, passives, and lexical form atives o f deponent^* verbs. The French and Spanish exam ples in (105) illustrate some o f the different uses o f the reflexive elem ent se in Romance. Likewise the exam ples in (106) illustrate a sim ilarly w ide range o f the sim ple reflexive sich ‘R E FL ’ in German. (105) French tl05a.b.c.eL Spanish tl0 5 d l: a. Pierre se rase. (reflexive) Pierre REFL shaves ‘Pierre shaves (him self).’ b. Pierre et Paul se rasent (I ’ un I ’ autre), (reciprocal) Pierre and Paul REFL shave the one the other ‘Pierre and Paul shave each other.’ The term “reflexive element” is used here as a cover term referring to simple reflexive pronouns (e.g. Dan. sig, Ger. sich), reflexive clitics (e.g. Fr. se), as well as reflexive suffixes (e.g. Dan. -s, Russian ~sja, etc.). Some languages have no overtly realized reflexive elements, e.g. Old English and Frisian. Cf. also the analysis o f Modern English proposed here which is based on the assumption that Modern English only has phonologically unrealized 0-reflexives. The term “deponent verb” is used to refer to verbs which cannot occur without a reflexive element. In the deponent verbs the reflexive element is always non-thematic. In contrast, when using the term “inherently reflexive” we remain agnostic as to the status of the reflexive which may either be a thematic/referential reflexive anaphor or non-thematic reflexive element. The set of “deponent” verbs is therefore always a sub-set o f the set of “inherently reflexive” verbs. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 282 c. Ces lunettes se nettoient facilement. (m iddle) These glasses REFL clean easily ‘This pair o f eyeglasses cleans easily.’ d. Aqui se habla espagnol. (im personal passive) here REFL speaks Spanish ‘Spanish is spoken here.’ e. Pierre s ’ evanouit. (deponent/inherently reflex.) Pierre REFL faints ‘Pierre faints.’ (106) G erm an: a. Peter raziert sich. (reflexive) Peter shaves REFL ‘Peter shaves (him self).’ b. Peter und Hans razieren sich. (reciprocal/reflexive) Peter and Hans shave REFL ‘Peter and Hans shave eachother/them selves.’ c. Die Tur ojfnet sich ohne Miihe auch ohne Brechstange (m iddle) the door open REFL w ithout trouble also w ithout crowbar ‘The door opens w ithout trouble even w ithout a crow bar.’ d.^^ Die Kiste bricht sich leicht mit einer Brechstange auf. (passive) the coffer breaks REFL easily with a crow bar open ‘The coffer can easily be broken open w ith a crow bar.’ e. Peter erholt sich. (deponent/inherently reflex.) Peter relax REFL ‘ Peter relaxed G iven the fact that reflexive elem ents are frequently used in non-reflexive constructions, it is clear that in order to arrive at an account o f the binding behavior o f reflexive elem ents in a given language one m ust first determ ine w hich instances o f these elem ents count as true them atic/referential reflexive anaphors and which do not. W hile in the Rom ance languages and G erm an the same reflexive elem ent may be used in all o f the five interpretations in (105- 106a-e), in D anish the labor is divided betw een the reflexive pronoun .sig and the verbal suffix Cf. W. Abraham (1995:14) “[T]he middle refers to what is denoted by the adjectival, or statal, passive [..], not, however, the procedent, eventive, passive.” According to Abraham (106d) can be paraphrased as: (i) a. Die Kiest kann letch aufgebrochen werden. the coffer can easily open-broken become ‘The coffer can easily be broken open.’ b. Die Kiste ist leicht aufzubrechen. the coffer is easy to break open ‘The coffer is easy to break open.’ but not (ii) Die Kiste wirdleicht aufgebrochen ... the coffer becomes easily open-broken PASSIVE R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 283 -s, w hich is historically derived from encliticized versions o f the reflexives pronoun sig, i.e. O ld Icelandic sik ‘R E FL ’, in earlier stages o f the language^®, see (107a-e). (107) a reflexive sig. Peter barberer sig/sig selv. Peter shaves REFL self ‘Peter shaves (him self).’ b. reciprocal -s: (x) Peter og Marie kysses. Peter and M arie kiss-S ‘Peter and M arie are kissing (each other). (ii) Peter og M arie kysser hinanden. Peter and M arie kiss-PRES each other ‘Peter and M arie are kissing (each other). c. “m iddle” s i s /s : (i) Lceherne lasnede sig, munden hlev starre. (ii) Lceberne lasnedes, munden blev starre. lips-the loosen REFL/-S m outh-the becam e bigger ‘The lips parted, the mouth grew bigger.’ (iii) Uge i det samme skiltes Elverhoj ad right in the same split-S Elfin-Flillock apart ‘at the sam e tim e Elfm-Flillock split open.’ d. passive - s : (i) Maskinen skiltes ad fo r at blive renset. m achine-the split-S apart for to becom e cleaned ‘The m achine was taken apart to be cleaned.’ (ii) Maskinen blev skilt ad fo r at blive renset. m achine-the becam e split apart for to becom e cleaned ‘The m achine was taken apart to be cleaned.‘ (iii) Her tales spansk. (im personal passive) here speak-S Spanish ‘Spanish is spoken here.’ e. sisl-s in deponent vb : (\) Peter skammer sig. (inherently reflexive) Peter shame REFL ‘Peter is asham ed.’ (ii) Peter vcemmes blot ved tanken om mad. Peter disgust-S merely by thought-the about food ‘Peter is disgusted by the m ere thought o f food.’ ’ See Larsen (1969). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 8 4 The sentences in (107a) illustrate the by now fam iliar use o f sig and sig selv as them atic/referential reflexive pronouns. See section 4.5.1 for a brief discussion o f reflexive use o f reflexive pronouns. The sentences in (107b) illustrate the tw o ways to form reciprocals in Danish: (i) the non-productive m orphological reciprocal form ed by adding the -5 m orphem e to a small, lexically determ ined, group o f verbs, and (ii) the productive syntactic reciprocal w hich is form ed by letting the reciprocal pronoun hinanden ‘each other’ occupy the object position o f any transitive verb. In contrast, to reflexive pronouns in the Rom ance languages as well as Germ an sick, the D anish reflexive pronoun sig is not used reciprocally. Instead, as show n in (107b(i)), the reflexive/reciprocal/passive suffix -x (historically derived from a encliticized reflexive) is used. The reciprocal constructions will be discussed in m ore detail in section 4.5.2. The sentences in (107c) m ay be dubbed “m iddles” or “reflexive m iddles”^'. Actually, D anish cannot use the reflexive elem ents to form true m iddle constructions o f the Rom anee type illustrated by the French sentence in (105c). In order to render the m eaning o f (105c) in D anish one w ould have to resort to a paraphrase o f the type Disse briller er lette at vaske ‘These glasses are easy to w ash/These glasses wash easily.’ That is, unlike English, Danish does not have 0 -m iddles o f the type These glasses wash easily. In this respect, Danish is like Dutch. N ote that if one adopts the presence o f an im plicit agent as criterion for m iddle constructions (as opposed to inherently reflexive/deponent verbs) (cf. Zubizarreta, 1982) then the sentences in (107c) are clearly not true middles. In section 4.5.3 we argue that the sentences in (107c(i-iii)) should be classified as inherently reflexive verbs or, more precisely. The traditional French term for this group of verbs is “verbes pronominaux neutres” (cf. Ruwet, 1972, p. 107), Instead of using potentially misleading terms like “deponent” and “inherently reflexive verbs” which are also used in other meanings, one could decide to call these constructions “neutre”. The only problem with this term is that it may be confused with the term neutral (i.e. used to refer to neutral verbs o f the vajfe-type). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 285 inherently reflexive verbs derived from their transitive counterparts by lexical rules. N otice that in exam ples like (107c(i-ii)) sig and -s may be used w ith no difference in m eaning. In m any cases w hether sig or -5 is used is a m atter o f dialectal variation. D iachronically determ ined variation (i.e. in som e cases one form is preferred by the older generations while younger speakers prefer the other) may also play a role. See also section 4.5.5 for more discussion o f such case o f apparently free variation between - s and 5/g^-forms. The sentences in (107d) illustrate the use o f the - s suffix to form the so-called s- passive also som etim es referred to as ‘im personal passive’ due to the lack o f overtly expressed agents, see (107d(iii)). N ote that in certain cases the 5-form o f the sam e verb (e.g. skille ‘part, divide, take apart’) may be interpreted either as an inherently reflexive/m iddle in one context, see (107c(iii)), or as an 5-passive in another, see (94d(i)). The sentence in (107d(ii)) illustrates the periphrastic passive form ed by the auxiliary blive ‘be, becom e’ + the past participle o f the main verb. N ote that the m eaning o f w hat is called passive constructions in (107d) overlaps considerable with the m eaning o f the Rom ance middle, i.e. “5e m oyen”, constructions in (105d). Both the 5-passive and the periphrastic passive are very productive constructions in M odern Danish, see section 4.5.4 for m ore detailed discussion o f the 5-passive. Finally, the sentences in (107e) illustrate the use o f both sig and - 5 as part o f the lexical form atives o f deponent verbs. The verbs o f this group form a closed set. As in the case o f the exam ples in (107c), there appear to be no m eaning difference between the - 5-form s and the 5/g-forms. In fact, as will be argued in section 4.5.5, one o f the main difference between, on the o n e hand verbs o f the typ e illustrated by losne ‘lo o s e n ’ (1 0 7 c (i-ii)) and verbs o f the type R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 8 6 skamme sig ‘be ashamedVvcewOTe^ ‘be disgusted’ (107e) on the other, is that the form er have transitive counterparts while the latter does not^^ The rem ainder o f this chapter will be devoted to a com parison o f the semantic, syntactic and m orphological characteristics o f reflexive constructions, reciprocal constructions, m iddle constructions, passive constructions and deponent verbs. A nticipating som ew hat table 1 sum m arizes som e o f the result o f this comparison. Table 1. Reflexive R eciprocal M iddle Passive D epon/” Inher.Refl.' (a) non-thematic sig: sig - sig - sig (b) thematic sig: sig - - - (sigf^ {c )-s - -s -s -s -s (d) sig selv sig selv - - - (sig selv/'' (e) DP object DP (DPyhinanden (DP) DP - (t) examples (2a) (2b) (2c) (2d) (2e) (g) Romance se se se se se (h) German sich sich sich sich sich Row (a) and (c) in table 1 indicates the type(s) o f m orphological realization the different ‘diathetic’ ca te g o rie s'^ listed in the five colum ns, may take. Row (b) shows that only reflexive constructions may have them atic sig, i.e. sig as a true theta-role receiving anaphor occupying the direct object position o f the a neutral transitive verb. As discussed in chapter 3, Note that the inherently reflexive/deponent verb skamme sig ‘be ashamed’ (which does not have an alternative -s form) is an - s deponent verb in certain forms of Norwegian (a Mainland Scandinavian language closely related to Danish), e.g. skjemmes ‘be ashamed’, see also section 4.5.5. The existence of inherently reflexive verbs taking thematic sig is discussed in section 4.3.2. The existence o f non-thematic intensified reflexives is discussed in section 4.3.3. Traditionally diathesis is assumed to include two or three categories: active voice, middle voice and passive voice (cf. traditional grammars of Sanskrit and Classical Greek). Recently, it has been suggested that grammatical dimension of diathesis might be more fine-grained than previously assumed and that reflexive constructions/reflexive forms of the verb may be considered a ‘fourth voice’ situated between the active and the R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 8 7 the intensified reflexive sig selv can only be them atic, see row (d) in table 1 above. The close affinity betw een deponent verbs and the co-called m iddles is illustrated by the fact that they have the sam e values in all but row (e). The row in (e) in table 1 above shows that, except for the deponent/inherently reflexive verbs, the verbs w hich may enter the other ‘diathetic’ categories in table 1 also have transitive uses in with full lexical DPs occurring in object position. Row (g) refers to exam ples illustrating the different constructions and rows (g-h) show that in G erm an and the sim ple reflexives (se/sich) can be used in all the five colum ns in table 1. 4.5.1 Use of reflexive elements in reflexive constructions From a sem antic point o f view, the event denoted by a verb is said to be reflexive if it is the case the individual referred to by the subject D P engages in an activity that is directed at him- /herself, see (108b). (108) a. Peter hit James. ( A -> B) b. Peter, hit 0 , himself. (A -> A) The sentence in (108a) denotes a non-reflexive event in which the direction o f the hitting action goes from Peter (as the A G EN T) to James (as the PATIENT). The sentence in (108b) above denotes a reflexive event in that the action o f hitting both originates from Peter (as the AG EN T) and ends with Peter/him self (g& the PATIENT). In English the intensified 0-reflexive, 0 him self is always them atic. That is, it is always assigned a theta-role from its governing predicate. Likewise the Danish intensified middle voices. It is in this loose sense we use the term “diathetic categories” to refer to the different columns in table 1. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 88 reflexive sig selv ‘REFL s e lf is also always thematic^'’. In contrast, their sim ple counterparts (i.e. Eng. 0 and Dan. sig) have both them atic, see (108b), and non-them atic uses, see (109a). (109) a. Peter slog sig. (deagentive inchoative) Peter hit REFL ‘Peter got hurt.’ b. Peter[ slog sig, selv (med en hammer), (agentive causative transitive) Peter hit REFL self w ith a ham m er ‘Peter hit him self (w ith a ham m er).’ Sem antically, the sentence in (109b) denotes a reflexive event involving a hitter (A G EN T), i.e. Peter, and a hittee (PATIENT). In contrast, the sentence in (109a) describes a non-reflexive scenario only involving one participant, viz. Peter (THEM E). Hence, we characterize the use o f the sim ple reflexive sig ‘R EFL’ in (109a) as an “non-reflexive”/non-them atic use o f sig. The exam ples in (110) show that simple sig ‘R E FL ’ may also be a true them atic reflexive. (110) a. PeteriVasker Jamesi. ( A -> B) Peter w ashes James ‘Petep washes Jam es;.’ b. Peteri vasker sig/. (A -> A) Peter w ashes REFL ‘Petep washes 0 ;.’ In Danish, both the sim ple/unintensified reflexive sig ‘R E FL ’ and the com plex/intensified reflexive sig selv ‘REFL s e lf may be used as true them atic, reflexive anaphors^’. W hile historically derived from an encliticized reflexive pronoun, the verbal suffix -s can nevertheless not be used as a them atic reflexive in Danish. That is, the 5-forms o f verbs all represent non-reflexive uses o f this reflexive element. 4.5.2 Use of reflexive elements in reciprocal constructions In som e languages the reflexive elem ents can be used in reciprocal constructions. Unlike its French and G erm an counterparts, the Danish reflexive pronoun sig cannot be used with Exceptions to this generalization appear to be prosodically motivated, see discussion of in intensified non- thematic sig selv with the inherently reflexive verb skamme 'shame ’ in section 4.3.3. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 2 8 9 reciprocal meaning. Only the 5-form, w hich is historically derived from a phonologically reduced encliticized reflexive, o f a small lexically defined set o f verbs can receive a reciprocal interpretation, see (111). (Ill) Reciprocal s-■forms: Transitive form: a. modes ‘m eet’ mode ‘m eet’ b. trceffes ‘m eet’ trceffe ‘meet, encounter’ c. ses ‘see each other, m eet’ se ‘see’ d. slas ‘fight’ sld ‘hit, strike’ e. skcendes ‘quarrel’ (skcende p a J ‘scold’ f. modes ‘m eet’ mode ‘m eet’ g- trcettes ‘quarrel’ (trcette ‘tire, exhaust’) h. brydes ‘clash, w restle’ hryder ‘break’ i. kysses ‘kiss’ kysse ‘kiss’ j- skiftes (til at ../ta k e turns ..-in g .’ skifte (jud) ‘exchange;change sby’s diapers’ k. folges (ad) ‘accom pany (each other)’ folge ‘follow ’ 1 . hjcelpes (ad) ‘help (each other)’ hjcelpe ‘help’ m. tales ved ‘talk’ tale ‘talk’ n. snakkes ved ‘talk, chat’ snakke ‘talk, chat’ A n small num ber o f the 5-reciprocals do not have transitive counterparts. The verbs in (112) exem plify this group o f deponent reciprocal verbs. Transitive form: (112) Reciprocal s-form s: a. enes ‘agree, get on b. /or//g-e5‘becom e reconciled’ c. kappes ‘com pete’ d. hives ‘bicker’ ^ene *forlige- *kappe *kive W hile the reciprocal 5-forms may have been productive in an earlier stage o f the language (probably w hen the s- was still form ally identical to the reflexive pronoun), it certainly isn’t productive anym ore. Only in a few eases do the reciprocal 5-form and the ordinary transitive use o f the verb have the same m eaning, see (11 la,b) and (1 13a,b). (113) a. Peter m0der ofte Hans pa gaden. (m eet accid en tally ‘run into’) Peter m eet-PRES often Hans on street-the ‘Peter often m eets Hans in the street.’ b. Peter og Hans modes ofte pa gaden. ((i) ‘run into’, (ii) ‘meet Peter and Hans m eet-S often on street-the w ith’, (iii) passive ‘are ‘Peter and Hans often m eet in the street.’ encountered’) See also discussion of semantic and syntactic properties of sig in chapter 3. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 290 In m ost cases, the m eanings o f the reciprocal 5-forms are different, to a greater or lesser extent, from m eanings o f the transitive counterparts, see (114) and (115). (114) a. Peter slar Hans. Peter hits Hans. ‘Peter is hitting H ans.’ b. Peter og Hans slas. Peter and Hans hit-S ‘Peter and Hans are fighting.’ (115) a. Peter shifter luftfilter (pa sin nye bit). Peter exchanges air filter on PO SS-REFL new car Peter changes the air filter in his new car. b. Peter og Marie skiftes til at vaske op. Peter and M arie exchange-S to to wasb up ‘Peter and M arie take turns doing the dishes.’ Such non-com positional m eaning differences betw een the ordinary transitive uses, e.g. (a) exam ples in (114)-(115), and the derived reciprocal form s, e.g. (b) exam ples, is an strong indication that the reciprocal 5-forms are lexicalized as such. Further evidence o f lexicalization com es from the fact that certain verbs are particle verbs in their reciprocal uses but not in their transitive forms, see (1 llk ,l), or vice versa, see (111 e). Form ally the 5-reciprocals appear to be identical to the 5-passive forms, see (1 16a) and (116b). (116)^* a Peter og Marie modes ofte p a gaden. (reciprocal) Peter and M arie m eet-S’ ^ ® '”'’™ '’ ® ' often on street-tbe ‘Peter and M arie often m eet each other in the street.’ b Peter og Marie modes ofte p a gaden. (passive) Peter and M arie m eet-S’ ’® * ® " '® often on street-the ‘Peter and M arie are often m et (by others) in the street.’ However, in at least one case the tw o form s differ in pronunciation'^^, see (1 17a) vs. ( 1 17b). The superscripts “reciprocal” and “passive” qualifying the different occurrences of the suffix - s in the examples does not mean that we assume the existence of two different -s morphemes. At this point we leave open the question o f whether the lexicon contains one entry - s which assumes different meanings when used in different contexts or whether it contains several different suffixes with the same overt realization, e.g. a reciprocal s , a passive -s, etc. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. m (117) a. Bornene slas i skolen. (sld [sla’* * ] + reciprocal -s = [sl^s]) children-the in school-the ‘The children are fighting at school.’ b. Bornene slas i skolen. (sld [sla’^'s] + passive -5 = [sla’^s]) children-the hit-S’ ’'* ® * " ''’ in school-the ‘The children are being beaten (by the teachers) at school.’ The difference in pronunciation betw een the reciprocal 5-form slds [sl^s] and the passive s- form slds [sla’® s] thus indicates a difference in m orphological structure. The long vowel in the infinitival form sld [sla”* ] is preserved in the 5-passive form slds [sla’^s] due to the presence o f a strong inflectional m orphem e border betw een the verb stem and the suffix w hich prevents the form ation o f a closed syllable and the accom panying shortening o f the vowel. N otice that bid also changes its pronunciation from [bla’* * ] to [bN] in other derivational processes such as the form ation o f com pound nouns'*®. All this indicates that the 5-reciprocals are lexicalized in the sam e way as com pound nouns, i.e. the phonological rules operating on the reciprocal 5- forms are the same as the ones which operates on the output o f lexical form ation processes such as compounding'*'. The fact that some 5-reciprocals are derived from intransitive verbs, see (11 Im ) above, and that yet others have no transitive counterparts, see (112a-d), further illustrates the Cf. Hansen (1967). The difference in vowel length and quality between the infinitival form sla [sla’" * ], pronounced with long vowel and stod (i.e. pitch accent realized as ereaky voice), and the reciprocal form slds [s^s], which is pronounced with a slightly higher (=more closed) short version of the long vowel [aa], is also found elsewhere in the language, see the examples in (i) and (ii). The examples in (i)-(ii) illustrate the fact that the vowel [^] is the regular shortened form of the long vowel [aa] which is generally assumed to occur only in open syllables. (i) (ii) The above argumentation is based on the assumption that the deponent j-form verh omgds [''mga’“s] ‘handle, mix with’ should be analyzed as an ^-passive rather than as an s-reciprocal form. a. en bid bil (bid [bla’“l, common gender form) ‘a blue car’ b. et bldt hus (bid [bla’“] + - t (neuter) = [b rt]) ‘a blue house’ a. en bid musling (bid [bla’“], common gender form) a blue mussel b. en bldmusling (bid- [bH], compound noun) a common mussel (Mytilus edulis) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 9 2 idiosyncratic nature o f ^-reciprocals and can thus be added to the list o f argum ents in favor o f assum ing 5-reciprocal to be lexicalized as such. Some 5-reciprocals, e.g. kysses ‘kiss (each other)’ in (llli) have already becom e obsolete am ong younger speakers who now only use the pronom inal reciprocal with this verb, e.g. de kysser hinanden ‘they kiss each other’. This kind o f idiosyncratic diachronic variation further underscores the lexicalization hypothesis. Some o f the 5-reciprocals have alternative form s with the reciprocal pronoun hinanden ‘each other’, see (118). (118) a. Peter og Hans moder ofte hinanden p d gaden. (com pare w. (116a)) Peter and Hans m eet-PRES often each-other on street-the ‘Peter and Hans often m eet each other in the street.’ b. Bornene slar hinanden i skolen. (com pare w. (117a)) children hit-PRES each-other in school-the ‘The children often hit each other at school.’ W hile in the case o f (1 16a) and (1 18a) the 5-reciprocal and the pronominal reciprocal have the same meaning, this is by no means always the case. Indeed, m ost o f the 5-reciprocals w hich have alternative pronom inal reciprocals exhibit some m eaning differences between the tw o forms. As the exam ples in (1 17a) and (1 18b) illustrates, when there is a m eaning difference, then the verb form ing the reciprocal w ith hinanden ‘each other’ always have the same m eaning as it has in its transitive use. W hile the .v-reciprocal .slds, see (1 17a), refers to any kind o f physical fighting activity (i.e. w restling could qualify) involving more than one person, the transitive and reciprocal uses in (1 14a) and (118b) necessarily involves a hitting activity (i.e. boxing or fighting with clubs/sticks could qualify but wrestling could not). The exam ples in (119) exem plify the kind o f m eaning differences often found between transitive forms o f a verb and its reciprocal and inherently reflexive forms. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 293 (119) M eaning differences between lexicalized and productive form s: Lexicalized M eaning Transitive productive use: a. R eciprocal: De slds. ^ De star hinanden. they hit-S they hit each other ‘They fight.’ ‘They hit each other.’ b. R eflexive: Han star sig. Han star sig ^g/v/hunden he hits REFL he hits REFL self/dog-the ‘Fie gets hurt.’ ‘He hits him self/the dog.’ On the basis o f the above discussion o f reciprocals we conclude that, unlike French and German, D anish cannot use the sim ple reflexive pronoun, i.e. sig, to form reciprocal constructions''^. We furtherm ore conclude that the only presently productive way to form the reciprocal in M odern Danish is by using the reciprocal pronoun hinanden ‘each other’. That is, in M odern Danish reciprocals form ed w ith the reflexive elem ent -s are limited to a small (closed) set o f lexically reciprocal verbs. 4.5.3 The use of reflexive elements in middle constructions English can productively form 0-m iddles, see (120a), from m ost transitive verbs. This kind 0-m iddles are not possible in Danish, see (121a). Even the finite form s o f the 5-passive cannot felicitously be used in this way, see (121b). The m ost productive way to form the m iddle is the is/are easy to verb-INFINITE construction exem plified by the sentence in (121c). (120) a. The LAPD officers bribed easily. b. Peter easily bribed the LAPD officers. (121) a. *LAPD betjente bestikker let. L A P D o ffic e rs b rib e -P R E S -A C T e a sily ‘LAPD officers bribe easily.’ b. */??LAPD betjente bestikkes let. LA PD officers bribe-PRES-S'’‘ ‘® * "'’ easily ‘LAPD officers bribe easily.’ Interestingly, Russian is like Danish in that reciprocals are formed using the suffix s j a , but cannot he formed using the free reflexive sebja, cf. Israeli (1997:74-78). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 9 4 c. LAPD betjente er lette at hestikke. LAPD officers be-PRES easy to bribe ‘LAPD officers are easy to bribe.’ d. LAPD betjente kan let bestikkes. LAPD officers can easily bribe- ‘LAPD officers can easily be bribed.’ The modal verb+.s-passive construction o f the type exem plified by the sentence in (121d) is another frequently used, and fully productive, way to translate the m eaning o f English 0 - m iddles into Danish. The use o f the 5-passive in (121d) is in certain respects sim ilar to the so- called 5c-moyen in French, see (122). (122) a. C evestonselavefacilem ent. this w est REFL washes easily ‘This west washes easily.’ b. Ces livres se vendent rapidement. These books REFL sell fast ‘These books sell fast.’ Notice, however, that in this kind o f construction the 5-form has to be an infinite form follow ing a modal verb. This is not the case in French where the verb usually occurs in the finite 5c-moyen form, com pare (122) and (123). (123) a. ??Ce veston peut se laver facilement. this w est can REFL wash easily ‘This w est washes easily.’ b. llC e s livres pouvait se vendre rapidement. these books can REFL sell rapidly ‘These books sell rapidly.’ Based on the above exam ples we therefore conclude that the sentences o f the type illustrated in (121d) has m ore in com m on with passive 5-forms than w ith the French m iddle constructions in ( 122). W hile D anish does not use the 5/g-forms to form middle constructions o f the English and French types illustrates in (120a) and (122a,b), it does use 5/g-forms to form deagentive inchoatives, see dbne sig ‘(unintentionally) open’ (124b) w hich is derived from the original Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 295 causative form o f the verb abne ‘(intentionally) open som ething’ in (124a). The sentences in (125)-(126) illustrate sim ilar causative-inchoative pairs in English and French. (124) a. Peter abner d0ven. (agentive causative transitive) Peter opens door-the ‘Peter opens the door.’ b. D0ven abner sig. (deagentive inchoative / “m iddle” or pseudo-m iddle) door-the opens REFL ‘The door opens.’ (125) a. Peter opens the door. b. The door opens. (126) a. Pierre ouvre la porte. Peter opens the door ‘Peter opens the door.’ b. La porte s ’ ouvre. the door REFL opens ‘The door REFL opens.’ Such deagentive inchoatives involving 5/g-forms (5e-forms in French) which we have referred to here as “m iddles” or pseudo-m iddles, will be shown to have a great deal in com m on with the inherently reflexive verbs discussed in section 4.3. 4.5.4 Use of reflexive elements in passive constructions. The sentences in (127-8) illustrate the use o f the - s suffix to form the so-called 5-passive, also som etim es referred to as the impersonal passive due to the lack o f an overtly expressed agent. (127) a. Maskinen skiltes ad fo r at blive renset. (=(107d(i))). m achine-the split-S apart for to becom e cleaned ‘The m achine was taken apart to be cleaned.’ b. Maskinen blev skilt ad fo r at blive renset. (=( 107d(ii))) m achine-the becam e split apart for to be com e cleaned ‘T h e m a c h in e w a s ta k e n a p a rt to b e c le a n e d .‘ (128) a. Se habla espahol. (=(105d)) REFL speaks Spanish ‘Spanish is spoken here.’ b. Pier tales spansk. here speak-S Spanish ‘Spanish is spoken here.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 96 (129) a. Lige i det samme skiltes Elverhaj ad right in the same split-S Elfin-H iilock apart ‘At the same tim e Elfm -H illock split open.’ (=(107c(iii))) In certain cases the 5- form o f the same verb (e.g. skille ‘part, divide, take apart’) may be interpreted as a “m iddle”/pseudo-m iddles in one context, see (129a), and as an ^-passive in another, see (127a). The sentence in (127b) illustrates the periphrastic passive form ed by the auxiliary blive ‘be, becom e’ -I- the past participle o f the main verb. N ote tbat the m eaning o f w hat is called passive constructions in (128b) overlaps considerably with the m eaning o f the Rom ance m iddle constructions in (128a). Both the 5-passive and the periphrastic passive are productive in M odern Danish. The m orphology o f the so-called 5-form/5-passive is illustrated in (130). (130) The form ation o f the the ^-formA-passive**^: Infinitive Present Past Past Part M eaning a. Act. mode mode-r mod-te mod-t ‘m eet’ Pass. m o d es m o d es m od-tes m od-tes b. Act. Icegge Icegger Icegde lagt ‘lay, place’ Pass. IcEgges Icegges Icegdes la g d es c. Act. sld sld-r slog sldet ‘h it’ P ass. sla-s sla-s slo g -es slo g -es ‘fig h t’ d. Dep. vcemmes vcemmes vcemmedes vcemmedes ‘feel disgusted’ Sim ply put, the s-form is form ed by adding - 5 to the active form, w ith an epenthetic schwa, spelled with the letter ‘e ’, inserted as required by phonological well-form edness constraints. The three main uses o f the 5-form o f verbs are: (a) the passive ~s form, (b) deponent verbs, and (c ) re c ip ro c a l v erb s. T h e re c ip ro c a l h a v e a lre a d y b een d is c u sse d In th e p re c io u s s e c tio n an d the so-called deponent verbs will be discussed in section 4.5.5. 'C f. Allan, eta/. (1995:310-311). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 97 A clear distinction betw een a passive reading and an intransitive “m iddle’Vpseudo- m iddie reading can be found with the verb skilles ad ‘be taken apart; burst, go to pieces’, see (131a) vs. (131b). (131) a. maskinen skiltes ad fo r at blive renset (passive) m achine-the separate-S apart for to becom e cleaned ‘the m achine was taken apart in order to clean it.’ b. lige i det samme skiltes ElverhoJ ad (m iddle, intransitive) right in the sam e separated-S Elfm -H illock apart ‘at the sam e m om ent Elfm -H illock (a fam ous hill/m ound) split open.’ The .s-passive is productive and can thus be form ed on the basis o f any transitive verb. N eedless to say, when occurring with a full lexical object DP, such verbs never have a passive m eaning. N otice also that the 5-passive can even be form ed from intransitive verbs, e.g. (132). (132) a. Der dansedes til langt udpd natten there dance-S to long out on night-the ‘D ancing was going on until late at n ig h f b. Skulle der does sd dode man uden at klage (Cf. Hansen 1966) should there die-S then died one w ithout to complain ‘If one had to die, then one died w ithout com plaining’). 4.5.5 Deponent verbs. Allan, et al. (1995:311) define deponent verbs as follows: ‘Deponent verbs are verbs that have a passive form (-s form ) but active m eaning. They m ay be transitive or intransitive’, see the exam ples in (133). (133) a. Der findes mange dyr ear ter. there fm d-S m any anim al-species ‘There exist m any species o f anim als.’ b. Hun mindedes sin ungdom. sh e re c a ll-S P O S S -R E F L y o u th (h o w sh o u ld th e d e p o n e n ts b e g lo s se d ? ) ‘She recalled her past.’ c. Jeg synes, at det er en god ide. 1 think-S that it is a good idea ‘1 think that it’s a good idea.’ The deponent verbs constitutes a small closed list o f forms, m ost o f w hich are given in (134). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 98 (134) Exam ples o f deponent verbs: a. dages ‘daw n’ b. fades ‘lack’ c. findes ‘be, exist’ d. fcerdes ‘move, travel’ e. gronnes ‘becom e green’ (inchoativey f. kendes ved ‘acknow ledge’ g- Icenges ‘long’ h. lykkes ‘succeed’'* ^ i. mindes ‘rem em ber’ j- mis lykkes ‘fail’ k. omgds ‘handle, m ix w ith’'* * ’ 1 . rygtes ‘be rum ored’ m. synes ‘think, seem ’ n. trives ‘thrive’ (=Ei 0. vcemmes ‘feel disgusted’ P- celdes ‘becom e older’ (inchoative) 44 Per definition deponent verbs do not possess a form w ithout -s. W here a verb with such a form exists e.g. fmde/findes, minde/mindes, see exam ples in (135), there will usually be a m arked difference in m eaning from the deponent verb, and they m ust be regarded as tw o separate verbs, see (135-6). (135) a. Han m indede m ig om model. he rem inded me about m eeting-the ‘He rem inded me o f the m eeting.’ b. Jeg mindes ikke hans tale. 1 rem ind-S not his speech I can’t rem em ber his speech. Inchoative verbs can be subsumed under passives by assuming tbat they represent a passivized forms of originally causative transitive predicates. In contrast to its English equivalent ‘succeed’, lykkes cannot have an inanimate subject in Danish (cf. Allan, et al. (1995:311): (i) a. Fors0get er lykkedes. b. Det lykkedes ham at besta eksamen. c. Det er lykkedes hende at fa et job. ‘The attempt was successful.’ ‘He succeeded in passing the exam.’ ‘She has succeeded in getting a job.’ Phonology: Even though the deponent verbs are most likely lexicalized no vowel shortening takes place, in this respect they differ from the s'-reciprocals which do trigger vowel shortening, compare (i) and (ii) (i) a. b. c. (ii) a. Active sla [sla’“] ‘hit’ ro [ro’“] ‘row’ DNA Active tr. sld [sla’“] ‘hit’ S-passive i/fl-s [sla’“s] ‘he hit’ ro-s [ro’“s]/ *[r^s] ‘rowing is going on’ omgds [sla’“s] ‘handle, mix with’ Active reciprocal sld s [sDs] ‘[they] fight’ (p r o d u c tiv e 5 -p a s s .) (p r o d u c tiv e 5 -p a s s .) ( le x ic a liz e d d e p o n . v e r b ) ( le x ic a liz e d s -r e c ip r o c a l) R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 2 9 9 (136) a. Jeg kan ikke fm d e mine briller. I can not find my glasses ‘I can not find my glasses.’ b. Der kan ikke findes dyreliv pd Mars. there ean not find-S anim al life on M ars ‘There eannot exist anim al live on M ars.’ (137) Exam ples o f deponent verbs w ith transitive counterparts: a. fattes ‘lack’ fa d e ‘understand, com prehend’ b. findes ‘be, exist’ fm d e ‘find’ c. kendes ved ‘acknow ledge’ vedkende sig ‘acknow ledge’ d. mindes ‘rem em ber’ minder om _ ‘rem ind sby about sth.’ e. synes ‘think, seem ’ syne ‘inspect’ The verbs in (138) are interesting in that in addition to the regular .s-forms they also have sig- forms. Furtherm ore the .s-form and the s/g-form have the same m eaning and can be used interchangeably. These alternate s/g-form s thus bear w itness to the fact that the -5 in the s- forms originated from an encliticized reflexive, e.g. (O ld N orse) -sik > (Iceland.) -s t > (Dan.) -s. (138) .s-form: s'/g-form: transitive form : & . kendes ved ‘acknow ledge’ wedkende sig ‘acknow ledge’ ‘know ,recognize’ h.detforstds ‘understand'*’’ detforstdr sig ‘understand’ forstd ‘understand’ e. det hcends at ‘it happens that’ dethcender sig ‘it happens that’J e t hcender ‘it happens th at’'* * In substandard varieties o f D anish certain verbs (which are not found in the ~s form in the standard language) can occur in the -s form w ith the intransitive/m iddle m eaning, see (139a,b). In som e cases a 5/g-form is also found with the same m eaning as both the active form and the (deponent) passive 5-form, com pare (139a,b and c) and (140). (139)'*® a. Det hcender at, .... (Standard Danish) (active) it happens(active) th a t.. ‘It happens th a t....’ Notice that the same idiomatic use o f reflexives with verbs of understanding is found in French: Dan. Det forstas/Det forstar sig ~ Fr. (7a se comprends. Notice that the same idiomatic use of reflexives with verbs of ‘happening’ is found in French: Dan. Det hcends at/ det hcender sig at ~ Fr. il se passe ‘it happens....’. These examples are from Hansen, 1967. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 300 b. Dethcend(e)sat,... (Substandard Danish) (5-passive) it happened-S that ‘It happens-S (5-passive) t h a t ... c. Det hcendte sig at, ... (Substandard D anish) (5/g--form) it happened REFL th a t... ‘It happened t h a t . . . ’ (140) a. D etforstds / Det forstdr sig. (Danish) it understand-S it understand REFL ‘It goes w ithout saying.’ b. Qa se comprend. (French) It REFL understands ‘It goes w ithout saying.’ N otice also that the inherently reflexive verb par excellence (i.e. the exam ple m ost often m entioned in the literature, e.g. Reinhart & Reuland (1993)), viz. skamme sig ‘be asham ed’ has tw o variants in N orw egian dialects, see (141). (141) a. skamme seg (Bokmal N orw egian) sham e REFL ‘be asham ed’ b. skjemme-s (N y-norsk Norwegian)^” sham e-S ‘be asham ed’ c. skamme sig (M odern Standard Danish) sham e REFL ‘be asham ed’ As discussed in section 4.3, m any inherently reflexive verbs are in fact simply deponent verbs with sig form rather than 5-form. That is, the sig in skamme sig -b e asham ed’ (1 4 1c) fulfils the sam e gram m atical function as the - 5 in vcemmes ‘feel disgusted’ (134o), both merge in v with unaccusative verbs. Superficially, there are three m orphological classes o f deponent verbs: (i) those w hich take only - 5 , see (134), (ii) those w hich are found with both - 5 and sig (138-141), and those which take only sig, e.g. skynde sig ‘hurry’, (68-71). However, sem antically/gram m atically there is only one class o f deponents (i.e. - 5 and sig have the same function in v), see (2b). Cf. Jan Terje Faarlund, Svein Lie, and Kjell Ivar Vannebo (1999) Norsk Referanse-Grammatikk. Universitetsforlaget. Oslo, p. 5 11. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 301 4.6 Conclusion The different reflexive and non-reflexive uses o f reflexive elem ents in Danish and English have been sum m arized in the table in (142), to w hich has been added a sim ilar survey o f the uses o f reflexive elem ents in French, Germ an and Chinese. (142) Typological survey o f reflexive and non-reflexive (and non-them atic) uses o f reflexive e l e m e n t s ! Danish English French G erm an C hinese a. refl. sig him self se/soi sich (ta)^iji b. recp. -s - se sich c. p-mid sig - se sich d. mid - - se sich d. pass. -s - se sich e. dep -s/sig - se sich f. int. - him self - - (ta) ziji Interestingly, neither English nor Chinese have any non-reflexive^' uses o f their so-called reflexives, viz. him self and ziji respectively, see (142b-e). W hile French se, Danish sig and Germ an sich all classify as true reflexives, him self does not since it is not featureless. Chinese ziji, however, is (like the D anish intensifier selv) unm arked for person, num ber, and gender, etc. and could thus be a featureless “true” reflexive. N otice, however, that it behaves ju st like English him self in that it cannot be used in any o f the non-reflexive constructions. Furtherm ore, both English him self and Chinese ziji behave like intensifiers, som ething the reflexives in Danish, French, Germ an and Russian are unable to do, see (142f). So except for the reflexive uses, English him self \s basically in com plem entary distribution with the Danish reflexive sig (as well as with Ger. sich, Fr. se, Russian -sja). O n the other hand, as discussed in chapter 2, English him self and Chinese ziji display m ost o f the uses o f the Danish intensifier selv, see (143). Needless to say, the logophoric uses of him self in English (also referred to as “locally free reflexives” or LFRs, cf. Baker (1995)) are also in a sense non-reflexive. However, there are still in argument position, i.e. thematic. As will be discussed in more detail in chapter 5 such non-Iocally bound instances o f him self are best analyzed as reduced form of underlying intensified pronouns him self < him him self (Cf. Siemund 2002:81). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 302 (143) Typological survey o f different uses o f adnom inal intensifiers: Danish English French Chinse a. Adnom. inten. kongen selv the king him self le roi (lui-)meme guo-wang (ta) zij b. Complex refl. sig selv [0 himself] soi-meme/lui-meme [ta ziji/ 0 ziji] c. Prosodic inten. sig selv [HIM SELF/0 himself] [801/ soi-meme| [0 ziji] d. Doppelgrenger-mark sig selv [0 himself] [soi-meme] [0 ziji] e. Inten. pron./logoplior ham selv [(him) himself] [(lui) lui-meme] [(ta) ziji/ta ziji] f. Exclus.adv. int. selv him self lui-meme ziji g. Indus, adv. int. selv himself jlui-meme] ye h. Add. focus part. selv even meme lian__ye j. Nominal uses selv/jeg self/*l *meme/moi - k. Deverbal compound selv *himself/self *meme-/auto-/sui-/etc.. *ziji/zi-/-ji 1 . Positive adjective selve very/actual[meme] ‘same (kind of)’ - m. Superlative adjective selveste - - - We take the com plem entary distribution o f him self and sig, see (142), as well as the sim ilar distribution o f him self and selv, see (143), as evidence that him self \s always an intensifier. The apparent reflexive use o f him self m (142a) can be explained if we assum e the English have 0-reflexives which can be intensified by the adnom inal intensifier him self These assum ptions m akes it possible to arrive at the follow ing revised version o f the typological survey in (142), see (144). (144) Typological survey o f reflexive and non-reflexive (and non-them atic) uses o f reflexive elem ents: Danish English French G erm an Chinese a. refl. sig 0 se/soi sich -/0 b. recp. -s 0 se sich - c. p-mid sig 0 se sich - d. mid - 0 se sich - d. pass. -s (0) se sich - e. dep -s/sig 0 se sich - Com bined, (143) and (144) draw a picture o f M odem English in w hich him self \s consistently analyzed as an intensifier w hich may adnom inally intensify all types o f nominal expressions, i.e. 0-reflexives, pronouns, and DPs. In other words, the absence o f non-reflexive/non- them atic uses o f him self plus its consistent behavior as an intensifier provide strong support for R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 303 the analysis o f intensified and unintensified nom inal expressions in Danish and English proposed in chapter 2, see (145). Sim ple/unintensified Com plex/intensified a. Reflexives sig 0 sig selv 0 h im self b. Pronouns ham him ham selv him self (<him himself) c. DPs kongen the king kongen selv the king h im self In this chapter the hotly debated question o f the lexical/syntactic status the sim ple reflexive sig (=Nor. seg, Dutch zich, etc.) has also been discussed. The myth that sig is an anti-local anaphor was dispelled. We have also show n that local sig can be a true theta-m arked reflexive pronoun even when bound by a co-argum ent since. That is, the large group o f verbs defined here as neutral do allow local sig (even when bound by co-argum ents) w ithout show ing any signs o f reanalysis/lexical ambiguity. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. THE INDEPENDENCE OF BINDING AND INTENSIFICATION VOLUME II by Uffe Bergeton A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (LINGUISTICS) May 2004 Copyright 2004 Uffe Bergeton Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 304 Chapter 5 Independence of intensincation and binding in English: synchronic and diachronic perspectives 5.1 Introduction In this chapter we argue that the analysis o f D anish based on the independence o f intensification and binding presented in chapters 1-4 can be extended to M odem and Old English as well as to the evolution o f reflexives and intensifiers betw een these two stages o f the language. The analysis o f M odern English w hich will be presented in section 5.2 is sum m arized in (1). (1) a. English x-5e//form s (e.g. himself, herself etc.) are not reflexive anaphors but rather adnom inal intensifiers (Eng. x-self ~ Dan. selv ‘self’), see (2a). b. M odern English has 0-reflexives (Eng. 0 ~ Dan. sig ‘REEL’), see (2b-e). c. W hat looks like locally bound reflexives is really locally bound adnom inally intensified 0-reflexives, (Eng. 0 x-self ~ Dan. sig selv ‘REEL s e lf), see (2b,c,e). d. The distribution o f unintensified and intensified 0-reflexives is controlled by the same sem antic/pragm atic factors w hich control the distribution o f sig and sig selv in Danish, e.g.: (i) Predicate m eaning: 1. Anti-reflexive predicates require 0 him self (2c). 2. N eutral predicates can take either 0 or 0 him self (2b). 3. Inherently reflexive predicate require 0 , see (2d). (ii) Doppel ganger-effects: Statue-readings require intensified 0 him self see (2e). e. W hat looks like locally free reflexives (or “logophors”)' are really intensified object pronouns whose pronom inal part is not realized phonetically (due to a morpho-Zphonological deletion rule) (Eng. [him] ^ him self ~ Dan. ham selv ‘him s e lf), see {If). f. Com plex subject pronouns (e.g. he him self I m yself etc.) are intensified subject pronouns, not as special kind o f anaphor (cf. M cKay 1991) (Eng. he him self - Dan. han selv ‘he s e lf), see (2g). ' The term “logophor” is used here informally to refer to the intensified object pronoun ham selv in Danish and the corresponding English cases o f non-locally bound [him] himself. Strictly speaking Danish ham selv and English [him] him self are not a logophors but intensified pronouns, see chapter 3, section 3.4.3 for a more detailed discussion of ham selv in Danish and the notion of logophoricity. ^ Square brackets are used to indicate the phonetically unrealized pronominal. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 305 (2) a. The kingi h im se lf came to the meeting. b. The king, washed 0i.l 0 , himself / DP^. c. The kingi suspected *0,./ 0 , himself / DP^. d. The king, rested 0,.! * 0 , himself / *DP^. e. Bill Clintoni shaved *0,. sm tu e -1 0 i .n a iv e himself. f. The kingi said that the orchestra could not play with anyone other than [h im j himself as conductor. g. He himself passed the exam. The assum ptions in (1) makes it possible to unify the analysis o f intensifiers and reflexives in English and Danish, see exam ples in (3) which show that the distribution o f Danish selv, sig, sig selv, ham selv, and han selv m irrors that o f English himself, 0, 0 himself, [him] him self and he himself. (3) a. Kongen selv kom til model. (~(2a)) king-the self cam e to m eeting-the ‘The king him self cam e to the m eeting.’ b. KongeHi vaskede sigi / .sig, selv / DP^. (~(2b)) king-the w ashed REFL / REFL self / DP ‘ The kingi washed 0 , / 0 , him self! DP^. c. Kongen, mistcenkte *sig, / sig, selv / DP^. (K 2c)) king-the suspected REFL self ‘The king suspected *0, / 0 , him self / DP^. d. Kongeni hvilte sigi / *sig, selv / *DP^. (~(2d)) king-the rested REFL self ‘The king rested 0 , / * 0 , him self / *DP^. e. Bill Clinton, barherede *sigi .s,a,ue ! sig, .n a m e selv. (K 2e)) Bill C linton shaved REFL / REFL self ‘ 'Bill Clintoni shaved *0i .n a m e ! 0 , sia m e - h im self’ f. Kongeni sagde at orkestret ikke matte spille med andre end hami selv som dirrigent. (K 2 f)) king-the said that orchestra-the not m ust play with others than him self as conductor ‘The king said that the orchestra could not play with anyone other than him self as conductor.’ g. Han selv hestod eksaminen. (==(2g)) he self passed exam -the ‘He herself passed the exam .’ The English x-5e^form s are consistently analyzed as adnom inal intensifiers whose distribution are determ ined by contrastiveness condition o f the m odule o f intensification ju st like the D anish intensifier selv ‘s e lf , see chapters 2-3. Likewise, the behavior o f the English 0 - ' Bold italics are used here only for purposes of highlighting and does not necessarily indicate stress. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 306 reflexive m irrors that o f sig ‘REFL’ in Danish, see chapters 3-4. Finally, [him] him self and he him se\f are analyzed as intensified pronouns whose pronom inal parts fall under principle B o f the binding theory w hile the intensifier part, i.e. him self fall under the contrastiveness condition on intensification. These parallels betw een the Danish and English system s o f reflexives, pronouns, and intensifiers is sum m arized in (4)-(5). (4) N om inal expressions in D anish: Simple/Unintensified'* Complex/Intensified a. Reflexive sig ‘him self/herself/oneself’ sigselv ‘him self/herself/oneself’ b. Pronoun ham ‘him ’ ham selv ‘him him self/him self’ c. DP kongen'the king’ kongen selv ‘the king him self’ d. Intensifier selv ‘self’(uninflected particle) (5) N om inal expressions in E nglish: Simple/Unintensified Complex/intensified a. Reflexive him self 0 him self b. Pronoun him [him] him self {him him selff c. DP the kin g ’ the king him self d. Intensifier him self {\nf\QC\Qd for gender, num ber and person)® A s show n in (6), form a superficial point o f view the English system o f pronouns and intensifiers appears to have a m uch higher degree o f m orphological opacity that the Danish system. '* As mentioned above the terms “simple” and “complex” are used Interchangeably with the terms “unlntenslfied” and “Intensified”. Although from the point of view of this paper the terms “unlntenslfied” and “Intensified” are more correct, the terms “simple” and “complex” occasionally serves as more convenient theory-neutral terms. ^ While English has an Intensified form of subject pronouns which can be analyzed as a combination of pronoun + adnomlnal Intensifier, e.g. he himself, the corresponding form for the object pronoun, which would be him himself. Is extremely rare and Is generally not accepted by native speakers. Note, however, that some native speakers do accept Intensified object pronouns. More discussion of the form him himself can be found in Blckerton (1987), latrldou (1987), and McKay (1991). The following example passage from Boyce (1979/2001:115) Zoroastrians. Their Religious Beliefs and Practices contains the only attested Instance of the full form him him self that 1 have been able to locate, see (1). (1) In his inscriptions Kirder has little to say o f doctrinal matters, being concerned rather with observances, church discipline, conversions, and the discouragement o f infidels. He proclaims, however, the existence o f heaven and hell, and the latter part o f the inscription at Sar-Mashad (1.57 ff.) is taken up with an account, only partly legible, o f a vision o f the hereafter seen by him himself. (Boyce 1979/2001:115) The English Intensifier has 8 frequently found forms: myself, yourself, ourselves, yourselves, him self herself, itself, themselves. In addition, two less frequently found forms exist: oneself, ourself. The latter Is the, now somewhat outdated, Intensifier corresponding to the royal we (cf. Slemund, 1999:8). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 307 (6) M orphological opacity o f English vs. transparency o f D anish: Eng. him self w Dan. selv, sig, sig selv, ham selv One o f the main advantaged o f the analysis outlined in (1) is that it enables us to analyze the system o f intensified and unintensified nominal expressions in English as having the same degree o f m orphological transparency as the D anish system in (5), see the tables in (7) and (8) which spell-out the com plete English pronoun system s as analyzed here. (7) Sim ple (i.e. unintensified) pronouns and ref exives in M odem English. N om inative Acc./Dat. Reflexive E ', sing. / me 0 2"‘ *, sing. you you 0 sing. masc. he him 0 3"^, sing. fem. she her 0 3"’, sing. neut. it it 0 V \ plur. we us 0 2"“, plur. you you 0 3"‘ * , plur. they them 0 (8) Com plex (i.e. intensified) pronouns and ref exives in M odem Englis 1 . N om inative Acc./Dat. Reflexive E ', sing. / m yself [me] m yself 0 m yself 2"“ , sing. you yourself [youjyourself 0 yourself 3"^^ sing. masc. he him self [him]himself 0 him self 3''^ sing. fem. she herself [her]herself 0 herself 3"^ sing. neut. it itself [it] itself 0 itself E ‘, plur. we ourselves [we] ourselves 0 ourselves 2"^ plur. you yourselves [you]yourselves 0 yourselves 3’'', plur. they themselves [them] themselves 0 themselves Section 5.3. deals w ith on the historical developm ent o f both the elem ent -self, w hich is the adnom inal intensifier in m ost Germ anic languages, and the M odern English x-^e//form s (e.g. him self herself, etc.). It will be shown that the assum ptions that M odern English has 0 - refiexives and that the x-self forms are ALWAYS intensifiers provide the key to a num ber o f ^ Although the iorms you yourself, it itself, and you yourselves are perfectly acceptable, they seem to appear only in subject position and never in object position. * The form it itself does seem a bit odd. The same seems to be true for the intensified form of the third person neuter pronoun del selv ‘it itself in Danish. Note that intensifier adjunction to [-animate] DPs is perfectly acceptable. It thus cannot be a constraint prohibiting intensifier-adjunction to [-animate] DPs which is at play. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 308 hitherto unsolved problem s related to the evolution o f M odem English com plex intensifier x- self(e.g. himself, herself, etc.) from the sim ple intensifier seolf'm Old English. Finally section 5.4 concludes the chapter by providing a sum m ary o f the m ain results obtained and their consequence for linguistic theory. W hile the analysis o f English outlined above has the advantage o f enabling an elegant unification o f the account o f intensified and unintensified pronouns and reflexives in Danish and English, it also raises a host o f questions, a num ber w hich will be addressed in the follow ing sections. One o f the m ost recalcitrant problem s is the existence o f unstressed him self w ith anti-reflexive verbs. If him self is assum ed always to be an intensifier and if intensifiers are always stressed (as has been claim ed in the literature (cf. Siem und (2000) am ong others) then the existence o f sentences like (9) with unstressed him self constitutes a potential problems. (9) Peter suspects him self As discussed in section 5.2.4, this and other facts, lead us to the conclusion that the 0 - reflexive hypothesis m ay be better suited to account for earlier stages o f M odern English than the language o f present day native speakers. 5.2 Intensifiers, 0-reflexiyes and intensified pronouns in Modern English and the independence of intensification and binding This section is divided into three parts. First, in section 5.2.1 the analysis o f sim ple and com plex reflexives (e.g. 0 vs. 0 himself) in English outlined above is discussed in m ore detail. Section 5.2.2 parallels section 4.5 in chapter 4 by surveying the different reflexive and non reflexive use o f 0-reflexives in English (e.g. reciprocals, middles, etc.). Section 5.2.3 takes a closer look at the analysis o f sim ple and com plex pronouns (e.g. he/him vs. he himself/[himjhimself). Finally, in section 5.2.4 the results o f the present analysis are sum m arized. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 309 5.2.1 Binding of 0-reflexives in different contexts: testing the independence of binding and intensification in Modern English In this section, the descriptive and explanatory adequacy o f the analysis o f English 0 - reflexives presented above will be put to the test by taking a closer look at the behavior o f simple and com plex reflexives (i.e. 0 , and 0 himself) in different syntactic positions. T hat is, we discuss 0-reflexives as direct objects (section 5.2.1.1), 0-reflexives in resultative constructions (section 5.2.1.2), 0-reflexives in ECM constructions (section 5.2.1.3), inalienable possession and the absence o f 0-reflexives in possessive constructions (section 5.2.1.4), 0-reflexives in prepositional predicates and PPs (section 5.2.1.5), and, finally, the potential problem posed by unstressed jc-se/f form s in object positions (section 5.2.1.6). 5.2.1.1 Locally bound 0-reflexives in argument position: complex reflexives = a sub-type of adnominal intensification A ccording to the analysis o f English proposed above, the distinction between the English simple and com plex reflexives 0 and 0 x-self is assum ed to be o f the sam e kind as the distinction between the sim ple (unintensified) reflexive sig in Danish and its com plex (intensified) counterpart sig selv. One w ould therefore expect to find distribution o f simple and com plex reflexives in object position o f “neutral”, “anti-reflexive” and “inherently reflexive” predicates’ in English to correspond to the distribution o f sim ple and com plex reflexives in Danish. The exam ples in (10-12) support this prediction. (10) A nti-reflexive verbs: (i) pron.: (ii) unint. refl. fiii) int. refl. a. He; suspects *himi / *0, / 0 , him self b. Harij mistcenker *hami / *sigi / sigi selv. he suspects him REFL REFL self ' See chapter 3, sections 3,3.2.1-3 for definitions o f the terms “anti-reflexive”, “neutral” and “inherently reflexive” R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 310 (11) N eutral verbs: a. He, washes *himi / b Han, vasker *hami / he washes him (12) Inherently reflexive verbs: a. He rests *him / h. Han hviler *ham / HE rests him 0 i sigi REFL 0, sig, REFL 0 , himself, sigi selv. REFL self *0, himself. *sigi selv. REFL self Based on exam ples such as (10-12) we conclude that predicate m eaning influences the choice o f 0 vs. 0 him self m the same way as it influences the choice o f sig vs. sig selv in Danish. The predicates disallow ing unintensified 0-reflexives, exem plified in (10), are anti-reflexive in that they presuppose (representational) non-identity o f their argum ents. The predicates in (11) are neutral since they are com patible w ith both reflexive and non-reflexive scenarios. Finally, the predicates in (12) are inherently reflexive in that they require the argum ent DPs to he c o r e f e r e n tia l. Intensifier-adjunction is thus clearly sensitive to sem antic/pragm atic factors; (i) it is only allow ed when contrast w ith other entities is possible (cf. Baker 1995), see (1 0 -lc,d ) vs. (12c,d), (ii) intensifiers adjoin to 0-reflexives w ith anti-reflexive predicates since these presuppose the (representational) non-identity o f their arguments. In follow ing seven sections (5.2.1.1-7) we address the question o f how far the equations Eng. 0 « Dan. sig and Eng. 0 Dan. x-self ~ sig s e lf can he pushed. That is, do all sem antically anti-reflexive predicates in English require the com plex form o f the reflexive 0 himselfl D o all sem antically inherently reflexives predicates in English require the simple form o f the reflexive 0 ? And do all neutral verbs allow both the sim ple and com plex form o f the 0-reflexive, i.e. both 0 and 0 himself! The thematic status of simple reflexives with inherently reflexive predicates is sometimes difficult to determine. See chapter 4 for discussion of tests used to determine whether a given use of a simple reflexive is thematic or non- thematic. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 31 5.2.I.I.I. Anti-reflexive predicates The theory proposed here predicts that sem antically anti-reflexive predicates in Danish, which allow only com plex reflexives (e.g. sig selv) and DPs in object position and do not allow simple unintensified reflexives (e.g. sig), should have English counterparts w hich likewise allow only com plex reflexives (e.g. 0 himself) and DPs and disallow the sim ple unintensified reflexive (i.e. 0 ). The list o f Danish anti-reflexive predicates and their closest English equivalents in (13) was arrived at by flipping through a large D anish-English dictionary". Based on the verbs listed in (13) it seems possible to conclude that Danish anti-reflexive predicates (i.e. * sig /sig selv / DP) always have anti-reflexive English counterparts (i.e. * 0 / 0 him self / DP). N o exceptions to this generalizations were found. That is, none o f the English predicates in (13) allow 0-reflexives; e.g. *Peter insulted 0 cannot m ean ‘Peter insulted him self, etc. (13) A nti-reflexive predicates in Danish and their English equivalents: a. ahsolvere ‘absolve, give absolution’ b. adlyde ‘obey (fx one’s superior’s) c. adoptere ‘adopt’ d. afbryde ‘interrupt’ e. affcerdige ‘dism iss, brush aside’ f. afh0re ‘exam ine (witness), interrogate’ g- aflaste ‘relieve the pressure on; releive (for from )’ h. aflevere ‘deliver’ i. afl0se ‘relieve (the guard), replace, succeed to ’ j- h0nfalde ‘implore, beseech, entreat’ k. efterfolge ‘follow (upon), succeed, follow ’ 1 . eje ‘ow n’ m. eskortere ‘escort’ n. fo rf0lge ‘persecute’ 0. forfere ‘seduce’ P- fr y g te ‘fear’ q- invitere ‘invite’ r. jage ‘hunt, stalk’ s. kvcele ‘choke, suffocate, strangle, asphyxiate’ t. kvcerke ‘throttle (=kill)’ u. lagre ‘store; season m ature (wine, cheese)’ V . lede X i fristelse ‘lead X into tem ptation’ " Cf. Vinterberg& Bodelsen (1966, 1986). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 1 2 w. m istcenke X fo r at X. misunde y. mode z. strangulere ‘suspect som ebody o f doing som ething’ ‘envy, grudge; be envious o f’ ‘m eet’ ‘strangle’ 5.2.1.1.2 Inherently reflexive predicates Clear cases o f English inherently reflexive predicates seem very hard to find. Duck 0 and its Danish counterpart dukke sig ‘duck’ is one o f the few plausible pair o f cognate inherently reflexive predicates we have been able to find, see (14). (14) D anish: a. Peter dukkede sig b. *Feter dukkede sig selv c. *Feter dukkede DF E nglish: Peter ducked 0 *Feter ducked 0 him self *Feter ducked DF And yet, how can we be sure that duck is really a dyadic “inherently reflexive” predicate rather than sim ply a com m on intransitive (unaccusative) verb. It is difficult to find objective critieria for deciding w hether the English verbs frolic and gambol, w hich translate the Danish “inherently reflexive” predicate boltre sig, true “inherently reflexive” predicates rather than sim ple intransitive predicates, see (15). (15) boltre sig - s i g sig selv *~D F ‘frolic gam bol’ frolic 0 / gambol 0 "^frolic 0 oneself / *gambol 0 oneself *frolic DF / *gambol DF O ther potentially true inherently reflexive predicates in English include overeat, see (16) and acquire, see (17c). (16) fo r cede sig ‘overeat’ - s i g * - sig selv * -D F overeat 0 *overeat 0 oneself * overeat FF [TEST] (17) anskaffe sig noget - sig noget sig selv noget DF noget ‘acq u ire som ething’ acquire 0 something * acquire 0 oneself something * acquire DF something R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth er reproduction prohibited w ithout perm ission. 313 There are basically tw o types o f m ism atches betw een D anish inherently reflexive predicates and their potential counterparts in English. The first is illustrated by the verbs in (18) and the second by the verbs in (19). (18) opf0re sig ‘behave (oneself)’ a. opf0re sig behave 0 b. * opfare sig selv behave 0 oneself c. * opfore DP ^behave DP (19) se sig gal pa nogen ‘see oneself mad at som ebody’ a. se sig gal pa nogen *see 0 mad at somebody b. *se sig selv gal p a nogen see 0 oneself mad at somebody c. *se DP gal pd nogen * see DP mad at somebody The D anish predicates in (18) and (19) both behave as “inherently reflexives” in that they allow only sig in object position. Their English counterparts, however, ‘m isbehave’ in that they do not allow DPs, see (18c) and (19c) in spite o f the fact that they allow the intensified reflexive 0 oneself This is a violation o f the contrastiveness condition on adnominal intensification, see (62) in chapter 2, repeated here as (20). (20) C ontrastiveness condition on adnom inal intensification: A nom inal expression (DP, pronoun, reflexive) can only be intensified adnom inally if it can be contrasted with other expressions in the context in w hich it is found. (= (62), chapter 2) If we assum e English him self to always behave as an intensifier, then the English exam ples in (18-9b) constitute blatant counterexam ples to the condition in (20). As show n in chapter 3, section 3.3.6 counterexam ples to (20) in D anish can be explained as being due to prosodic factors, i.e. the lack o f stressability o f sig w hich leads to prosodic reinforcem ent by adnom inal intensification when prosodic rules dictate that this elem ent m ust have non-contrastive word- stress. See section 5.2.1.5 below for a sim ilar account o f the exceptions to (20) in English. A s m entioned above, it is very hard to establish w ith certainty that English has true inherently reflexive predicates at all. Even if we assum e that English has som e true inherently reflexive predicates, it still seem s safe to say that they are rarer in English than in Danish. In R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 314 (21) is listed a few predicates w hich are inherently reflexives in Danish but w hich do not seem to have inherently reflexive English counterparts. (21) a. b. c. d. forelske sig i skamme sig skynde sig etc... ‘fall in love w ith ’ ‘be asham ed’ ‘hurry, be in a hurry’ N ote furtherm ore that in Danish sim ple sig also has a num ber o f non-reflexive uses, see (22). (22) a. D 0ren lukkede sig. door-the closed REFL ‘The door closed.’ b. D 0ren dbnede sig. door-the opened REFL ‘The door opened.’ c. En lille pol dannede sig. a small puddle form ed REFL ‘A small puddle form ed.’ In section 5.2.2.2 we discuss the possibility o f analyzing English m iddles such as close and open in (22a) as involving non-reflexive use o f 0-reflexives. 5.2.1.1.3 Neutral predicates The list o f Danish and English neutral predicates in (23) was established by flipping through a large Danish-English dictionary'^. (23) N eutral predicates in English and D anish denoting groom ing activities: a. bade X bathe b. barbere shave c. borste brush d. forklcede dress up (as), disguise e. klcede pd dress f. klcede _ a f x; undress g- klcede ud som dress up like h. klo _ (pd ryggen) sctratch _ (on the back) i. torre dry j- vaske wash ■ Cf. Vinterberg& Bodelsen (1966, 1986). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 315 a. bekymre worry b. cengste ~ worry N eutral predicates in D anish and Enslish: nontranslational'^ movem ent: a. hoje bend b. strcekke stretch c. dreje _ (rundt) turn _ (around) d. flytte move e. vende _ (om) turn _ (round/over) N eutral predicates in D anish and English: verbs denoting internal change: a. cendre change b. forandre change c. forbedre a improve d. forberede prepare e. forvandle _ til noget = ; turn _ into something W hile all the verbs listed in (23-6) behave as neutral predicates in both languages (i.e. allow ing 0/sig, 0 himself/sig selv, and D P in argum ent position), there is also a num ber o f m ism atches betw een the two languages. That is there is a num ber o f “neutral” predicates in Danish which have no straightforw ard English equivalents taking intensified reflexive pronouns. As discussed below, these cases do not necessarily have to be considered as counterexam ples to the proposed parallelism betw een Danish and English. It is indeed quite norm al for tw o different languages to lexicalize the same or sim ilar concepts in different ways, see (27). (27) N eutral predicates in D anish with no direct English equivalent: frisere ‘dress som ebody’s h air’ ~ sig dress/do o n e’ s hair ~ sig selv dress/do o n e’ s own hair ~ DP dress/do DP s hair The vQvh frisere in (27) has no transitive counterpart in English and has to be paraphrased as ‘do ’s hair’. The term “non-translational movement” is from Kemmer’s (1995, 1993) classification of “middle verbs” into d iffe r e n t s e m a n tic c la s s e s . R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 316 Peter fe d 0 on onions Peter fe d 0 him self on/with onions Peter fe d DP with onions There is, however, a different set o f m ism atching predicates w hich are potentially m ore problem atic, i.e. D anish neutral predicates with English equivalent w hich do not allow 0-reflexives, see (28). (28) English equivalents o f D anish neutral predicates w hich do not allow simple 0 -reflex iv es: a. beskytte ‘protect’ ~ sig (mod) *protect 0 (against) ~ sig selv (mod) protect 0 oneself (against) ~ DP (mod) protect DP (against) b. erncere ‘support, live on/by, feed’ ~ sig ^support 0 ~ sig selv support 0 oneself ~ DP support DP c. etahlere ‘establish’ ~ sig (som tandlcege) *establish 0 as a dentist /set 0 up as a dentist ’ establish 0 oneself as a dentist/set 0 oneself up as a den establish DP as a dentist /set DP up as a dentist 'am use, enjoy’ *amuse 0 / *enjoy 0 amuse 0 oneself/ enjoy 0 oneself amuse DP / enjoy DP ‘provide, procure, get’ '^■provide 0 (with vegetables) / get 0 (vegetables) ■ sig selv (med) provide 0 oneself (with vegetables) / get 0 oneself (veg.) ■ DP (med) provide DP (with vegetables) / get DP (veg.) (com pare w. lynche ‘lynch’ w hich is “anti-reflexive) *hang 0 hang onself hang DP ‘isolate, w ithdraw ’ *isolate 0 isolate oneself isolate DP withdraw 0 (die: he withdraws from other people) withdraw 0 oneself withdraw DP ~ sig selv (som tandl.) ~ DP (som tandl.) A.fornoje sig - s i g - sig selv - D P Q.forsyne - sig (med) f. hcenge ‘hang’ g. isolere -sig ■ sig selv DP ■sig ■ sig selv -DP more -sig - sig selv DP ‘amuse, enjoy - s i g - sig selv - D P i. m u reX inde - sig inde - sig selv inde - DP inde *amuse 0 amuse 0 oneself amuse DP / *enjoy 0 / enjoy 0 oneself / enjoy DP (different meaning) im m ure *immure 0 immure 0 oneself immure DP R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 317 All the D anish predicates in (28) are neutral, but their closest equivalents in English are behave as anti-reflexives. Should they therefore be considered counterexam ples to the proposed theory o f English reflexive? Let us first consider the case Danish predicate hcenge ‘hang’ in (28f) and the cognate English predicate hang. The Danish verb is neutral and its English counterpart behaves as an anti-reflexive predicate. Rather, than a serious counterexam ple we take this to be a idiosyncratic fact o f the two languages. As discussed in chapter 3, verbs that, under normal circum stances, allow only DPs and com plex reflexive (but not sim ple reflexives) ean be divided into true anti-reflexives, and ‘hidden neutrals’, see (29). (29) a. A nti-reflexive predicates: sem antically presupposed representational non identity, non-cancelable by context. Exam ples: afl0se ‘relieve (the guard), replace, succeed to ’ efterf0lge ‘succeed’ mistcenke ‘suspect (sby. o f doing sth.)’ misude ‘envy, be envious o f’ b. ‘H idden neutrals’: pragm atically im plied/presupposed representational non-identity, cancelable by context. Examples: koge ‘boil’ dolke ‘stab’ film e ‘film, sh o o f forgylde ‘gild ’ The group o f predicates in (29a) consists o f predicates w hich can never change status from anti-reflexive to neutral no m atter w hat context is imagined. The second group consists o f predicates w hich may change status depending on the context, e.g. koge ‘boil’. Cross- linguistically we predict that in any language w hich has a distinction between anti-reflexive” and neutral predicates, the predicates belonging to group (29a) will invariably classify as anti reflexives. The predicates in (29b), on the other hand, m ay behave as neutral by default in one language, while “anti-reflexive” by default in another. The D anish verb lynche ‘lynch’, for example, while close to hcenge ‘hang’ in m eaning behaves as an anti-reflexive predicate, just like its English counterpart lynch, see (30). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 318 (30) lynche ‘lynch’ Hynche sig Hynch 0 lynche sig selv lynch 0 oneself lynche DP lynch DP Furtherm ore, due to slight m eaning ditTerences and difTerences in usage it may in many cases be difficult to determ ine w hich out o f several candidates is the closest English counterpart o f a Danish predicate, e.g. (28c) erncere w hich translates into English as either ‘support’, ‘live o n ’ or ‘feed’. D epending which one o f these translations is chosen as TH E English counterpart o f this Danish predicate, we end up with either a m ism atch or a match. Exam ples such as these illustrate the relative uselessness o f the whole project o f finding English ‘counterparts’ o f Danish predicates. A better approach w ould be to sim ply consider eaeh language separately in order to see if any language internal counterexam ples turn up. Under this second approach, none o f the ‘m ism atches’ in (28) above w ould count as counterexam ples. All they would allow us to conclude is that D anish seems to have a greater num ber o f “neutral” predicates than English. 5.2.1.1.4 Non-optionality of intensifier-adjunction to 0-reflexives with proto typical neutral predicates: focus and doppelganger-effects The sim ilarities betw een the English 0-reflexive and the D anish reflexive sig are not limited to the sensibility to predicate m eaning illustrated in (10-12). Like sig, w hich has to be intensified every tim e it is focused, the English 0-reflexive also cannot be focused w ithout being adnom inally intensified. That is, the choice o f unintensified 0-reflexive vs. intensified 0 him self m (3 1 a ,b ) is not op tion al, as often claim ed , sin ce it depends on fo cu s, e.g . (3 1 b -,c,d ) can answ er the question Who did Peter wash?, but (31a) cannot. (31) Peter, w ashes . a. 0 , h .0 i him self c.him*,,^ A. Mary^ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 1 9 As discussed in chapter 3, section 3.3.2.5, the focus-sensitivity o f 0 vs. 0 him self is also found in Dan. sig vs. sig selv, e.g. (32b,c,d) but not (32a) can answ er the question Who did Peter wash? (32) PeteriVasker___ . sigi h. sigi selv c. ham*i. 2 d. Mary^ 5.2.1.1.5 Ditransitives The difference betw een (33), (34) and (35) shows that even for ditransitive predicates it is the sem antics o f the predicate w hich determ ines w hether it behaves as an anti-reflexive, neutral predicate or inherently reflexive predicate. (33) N eutral di-transitive predicates: a. b. ‘build’ bygge sig et hus hygge sig selv et hus bygge DP et hus kobe ‘buy’ kohe sig et hus kobe sig selv et hus kobe DP et hus (34) A nti-reflexive ditransitive predicates: scelge ‘sell’ *scelge sig et hus scelge sig selv et hus scelge DP et hus build 0 a house build 0 oneself a house build DP a house buy 0 a house buy 0 oneself a house buy DP a house *sell 0 a house ’ sell 0 oneself a house ’ sell DP a house ’ (35) Inherently-reflexive ditransitive predicates: anskaffe sig noget ‘acq u ire som ething’ ~ sig noget sig selv noget D P noget acquire 0 something * acquire 0 oneself something * acquire DP something R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 320 5.2.1.1.6 Linking Doppelganger-effects and anti-reflexivity In this section we show that intensifier-adjunction to sim ple 0-reflexives in English take place in the sam e contexts w here intensifier-adjunction to simple sig takes place in Danish, i.e. the contexts in (36). (36) C om plex reflexives (e.g. 0 himself)- . Exam ple: a. anti-reflexive predicates (10,13) b. hidden neutral predicates (29b) c. contrastive contexts (with neutral predicates) (31 -2) d. ‘doppelganger-effects’ (with neutral predicates): (i) statue-readings (37-8) (ii) qua-sentences (40-1) (iii) strict (and sloppy) readings in VP ellipsis (42-3) In sections 5.2.1.1.1 and 5.2.1.1.3 above anti-reflexive (containing presuppositions o f (representational) non-identity) and hidden neutral predicates (which together with background assum ptions about the world generate expectations o f (representational) non-identity)) were shown to trigger intensification o f sim ple reflexives. In section 5.2.1.1.4 contrastive contexts were identified as another trigger o f intensification o f sim ple unintensified 0-reflexives with neutral predicates. In this section a num ber o f other contexts triggering intensification o f simple reflexives with neutral predicates will be discussed, nam ely wax m useum contexts (statue-readings), qua-sentences, and strict readings o f reflexives in VP-ellipsis constructions. As was the case for Danish, discussed in chapter 3 section 3.3.2.7, we argue that all these intensification triggering contexts, see (36d(i-iii)), involve som e kind o f ‘doppelganger-effects’, i.e. (representational) non-identity o f the reflexive and its antecedent. 5.2.1.1.6.1 X-self as marker of statue-readings The adnom inal intensifiers him self and selv are used as m arker o f statue-readings in the same way in Danish and English. The English and D anish intensified reflexives ( 0 himselflsig selv) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 321 in (37-8b) can have both readings in (39a,b) while their unintensified counterparts {0!sig) in (37-8a) only have the reading in (39a). (37) Peteri w ashes___ . a. 0 , b. 01 him self (38) Peter, vasker . a. sig, b. sig, selv (39) a. ‘Peter<reai> washes Peter<reai>.’ b. ‘ Peter<reai> washes Peter<statue>- (statue-reading) 5.2.1.1.6.2 Qua-sentences As argued in chapter 3 section 3 3 .2 J.2 , the qua-sentences in (40-1) illustrate the m otivation for analyzing dobbelganger-effects in term s o f representational identity/non-identity rather than in term s o f referential identity. A gain the distribution o f intensified and unintensified reflexives is the sam e in English and Danish. (40) a. ??*Peter, working as barber, shaved. b. Peter, working as barber, shaved himself. (41) a. ??*Peter, i egenskab a f barber, barberede sig. Peter in quality o f barber shaved REFL ‘Peter, qua barber, shaved.’ b. Peter, i egenskab a f barber, barberede sig selv. Peter in quality o f barber shaved REFL self ‘Peter, qua barber, shaved him self.’ In order to facilitate the processing o f the above qua-sentence they should be read in a context w here a group o f sold iers are lost in a forest for w eek s. In order to m aintain a certain lev el o f hygiene the platoon leader picks Peter to function as cam p barber and im m ediately orders him to shave everybody in the cam p including himself. In this context, (4 0 -lb ) is clearly preferred over (4 0 -la). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 2 2 5.2.1.1.6.3 VP-ellipsis and the sloppy vs. strict reading of reflexives In chapter 3, section 3.3.2.7.3, we argued that the strict identity reading o f reflexives in VP- ellipsis constructions is due to doppelganger-effects, i.e. it involves representational non identity. The exam ples in (42-3) illustrate the difference between sim ple and com plex reflexives w ith respect to the availability o f strict and sloppy reading in ellipsis constructions. Again, the distribution o f sim ple and com plex reflexives is the same in English (42) and D anish (43). (42) a Peter washed 0 and so did Hans. (only sloppy) b. Peter washed 0 him self and so did Hans. (both sloppy and strict) (43) a. Peter vaskede sig og det gjorde Hans ogsa. (only sloppy) Peter w ashed REFL and it did Hans also ‘Peter w ashed (him self) and so did Peter.’ b. Peter vaskede sig selv og det gjorde Hans ogsa. (strict and sloppy) Peter w ashed REFL self and it did Hans also ‘Peter w ashed him self and so did Peter.’ 5.2.1.1.7 Summary In the previous sub-sections o f section 5.2.1 we have discussed the behavior o f locally bound 0-reflexives occurring as internal argum ents o f verbal predicates. A ccording to our assum ption that binding and intensification constitute separate m odules o f the gram m ar we predict the binding behavior o f 0-reflexives to follow from principle A as form ulated in chapter 3, while intensification o f reflexives (i.e. the distribution o f com plex reflexives, e.g. Dan. sig selv and Eng. 0 x-self) should follow from the principles o f intensification, i.e. the principle o f contrastiveness as form ulated in chapter 2. We argue that this approach to binding and intensification can successfully account for the distribution o f sim ple and com plex reflexives found in the contexts sum m arized in the descriptive generalizations in (44-5). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 2 3 (44) Sim ple reflexives fe.g. Eng. 0 ~ Dan, s i q . ) : a. N eutral predicates (section 5.2.1. 1.3) b. Inherently reflexive predicates (section 5.2.1.1.2) (45) C om plex reflexives (e.g. Eng. 0 him self ~ Dan, s is selv): a. anti-reflexive predicates (section 5.2.1. 1.1) b. hidden neutral predicates (section 5.2.1.1.3) c. contrastive contexts (w ith neutral predicates) (section 5 .2 .1. 1.4) d. doppelganger-effects (w ith neutral predicates): (section 5.2.1.1.6) (i) statue-readings (section 5.2.1.1.6.1) (ii) qua-sentences (section 5.2.1.1.6.2) (iii) strict (and sloppy) readings in VP ellipsis (section 5.2.1. 1.6.3) In the follow ing sections the distribution o f sim ple and com plex reflexives in resultatives (section 5 .2 .1.2), ECM constructions (section 5 .2 .1.3), and possessive constructions (section 5.2.1.4) will be discussed. Section 5.2.1.5 presents a phonological account o f the distribution o f com plex reflexives in prepositional predicates which explains intensification o f reflexives w ith certain PPs a consequence o f the inability o f simple reflexives (which behave phonologically like clitics) to carry stress on their own. W hile there are som e apparent “exceptions” to the correspondence betw een sig and 0 these are either due to lexical differences (e.g. lack o f exact sem antic correspondence betw een predicates in tw o languages, cf. the discussion o f neutral predicates in section 5.2.1.1.2), or can be explained by reference to other aspects o f gram m ar or language use (e.g. phonetic factors: 0-reflexives behave differently from overly realized clitics, etc., see sect. 5.2.1.5 below). The rem aining set o f more recalcitrant counterexam ples is fairly restricted and does, in our opinion, not in itself w arrant abandoning the viability o f the general approach binding and intensification proposed here. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 324 5.2.1.2 Resultatives and the 0-reflexive analysis of English If the 0-reflexive is the English equivalent o f simple unintensified sig in Danish, then one would expect to find it in same type resultative constructions as the ones found in Danish, see (46-8). (46) a. b. c. Peter arbejdede sig rig. Peter arbejdede sine forceldre rige. Peter arbejdede sig selv rig. Peter w orked REFL/REFL self/his parents rich ‘Peter worked him self/his parents rich.’ (47) a. b. c. Peter drak Peter drak Peter drak Peter drank M g Hans sigselv REFL/REFL self/Hans under bordet. under bordet. under bordet. under table-the ‘Peter drank him self/H ans under the table.’ (48) a. b. c. Peter dansede sig til verdensmesterskabet. Peter dansede M n e forceldre til verdensmesterskabet. Peter dansede M g selv til verdensmesterskabet. Peter danced REFL/REFLself/his parents to the w orld cham pionship ‘Peter danced him self/#his parents to the w orld cham pionship.’ The English counterparts o f (46-8) are given in (49-51). A s shown in (49a), (50a) and (51a), contrary to expectations, English resultatives o f the neutral and inherently reflexive types do not allow the 0-reflexive. This constitutes a potential problem for the analysis proposed here, w hich is based on the assum ption that the English 0-reflexive corresponds to the D anish sim ple/unintensified reflexive sig ‘REFL’. (49) a. Peter worked *0 rich. b. Peter worked his parents rich. c. Peter worked 0 him self rich. a. Peter worked *0 across the ocean. b. Peter worked Uhis parents across the ocean. c. Peter worked 0 him self across the ocean. a. Peter dansed * 0 to the word championship. b. Peter dansed #his parents to the word championship. c. Peter dansed 0 him self to the word championship. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 325 A t this point, we have no fully-fledged solution to this problem. A possible solution may come from com parison w ith R om ance languages, w hich lack (or have only very few) resultative constructions'^* o f the type exem plified in (46-51), see the French exam ple in (52). (52) *Pierre se travaille riche. Peter REFL works rich The gram m atical constraints barring reflexives in resultatives in French, see (52), may also be responsible o f the absence o f unintensified 0-reflexives in English resultative constructions. As discussed in more detail in section 5.2.1.5, phonological constraints on the distribution o f certain types o f clitics and zero-elem ents may also be at play. In other words, the absence o f unintensified 0-reflexives in resultative construetions does not necessarily constitute a fatal problem for the present analysis since it be explainable as being due to other m odules o f the gram m ar (i.e. phonology, etc.). 5.2.1.3 ECM constructions the 0-reflexive analysis of English If the 0-reflexive is the English equivalent o f simple unintensified sig in Danish, then one would expect to find it in same type ECM constructions as the ones found in Danish, see (53- 54). (53) a. Peter ansa sig fo r at vcere intelligent. b. Peter ansa sine forceldre fo r at vcere intelligente. c. Peter ansa sigselv fo r at vcere intelligent. Peter considered REFL/REFL self/his parents for to be intelligent (54) a. Peter ansa #sig fo r at vcere dod. b. Peter ansa sine forceldre fo r at vcere dode. c. Peter ansa sig selv fo r at vcere dod. Peter considered REFL/REFL self/his parents for to be dead The English counterparts o f (53-4) are given in (55-6). As shown in (55a) and (56a) such ECM constructions do not allow the 0-reflexive. This constitutes a potential problem for the Cf. Lidz & Williams (2002). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 326 analysis proposed here, w hich is based on the assum ption that the English 0-reflexive corresponds to the D anish sim ple/unintensified reflexive sig ‘REFL’. (55) a. Peter considered * 0 (to be) intelligent. b. Peter considered his parents (to be) intelligent. c. Peter considered 0 him self (to be) intelligent. (56) a. Peter considered * 0 (to be) dead. b. Peter considered his parents (to be) dead. c. Peter considered 0 him self (to be) dead. As in the case o f the resultative constructions discussed in the previous section, the gram m atical constraints responsible for the absence o f unintensified 0-reflexives in English resultative constructions may be unrelated to both binding and intensification. As discussed in m ore detail in section 5.2.1.5, interference o f phonological constraints on the distribution o f certain types o f clitics and zero-elem ents may be to blame. In other words, the absence o f unintensified 0-reflexives in resultative constructions does not necessarily constitute a fatal problem for the present analysis since it be explainable as being due to other m odules o f the gram m ar (i.e. phonology, etc.). 5.2.1.4 Adnominal intensifiers in possessive constructions If, as assum ed here, reflexives and pronouns are in com plem entary distribution then the local dom ain m ust be defined differently in D anish and English. In English the possessive pronoun his can be locally bound, see (58c). This is not possible in Standard M odern Danish, where the locality difference betw een pronouns and reflexives is m aintained in possessive constructions, see (57a) vs. (57b). (57) a. PeteVi vaskede sirii/t^pung. b . Peteri vaskede Hans * , ^ pung. c. Peteri vaskede M arie’ s^ pung. Peter w ashed PO SSR EFL/his/M ary’s w allet Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 327 (58) a. Peteri washed *0, wallet. b. Peteri washed *0, own wallet. c. Peteri washed his/^ wallet. d. Peter washed hisi own wallet. If, as argued above, 0-reflexives are the equivalent o f unintensified sig in Danish, then it would he natural to expect English 0-reflexives to he able to occur in possessor position just like the Danish possessive reflexive‘s sin ‘POSSREFL’ in (57a,b). As show n in (58a,h), this is not the case. We do not, however, consider this m ism atch in the distribution o f possessive reflexives and pronouns in D anish and English to constitute a serious problem for the present approach since it can he explained quite sim ply a follow ing from a m orphological differences betw een the tw o languages. Languages differ w ith respect to how m any and which gram m atical categories are captured overtly by the morphology. W hile the m orphology o f D anish and Latin differentiate overtly betw een possessive reflexives and possessive pronouns, languages such as English and French do not. W hatever the linguistic principles are which are at plays in determ ining the overt m orphology o f a language, they do not directly affect the present approach to binding and intensification proposed here. The difference betw een the binding dom ains for hans in Danish and his in English is thus a problem which any binding account has to address and thus does not necessarily constitute a specific argum ent against the proposal defended here. To add to the confusion, there are Danish dialects in which the locality difference between reflexives and pronouns is neutralized in possessive position, see (ia-b), but maintained in argument position, see (iia-b). (i) Danish Dialect (overlap o f sin and hans): a. Peteri vaskede sin, pung. b. Peter, vaskede ham,/^ pung. Peter washed POSSREFL/his wallet (ii) Danish Dialect (complementarity o f sin and hans): a. Peter, vaskede sig, * ^ . b. Peter, vaskede harn*,,^. Peter washed REFL/him (ii) Micro-parametric variation: DPs containing a subject constitute a local domain for Binding o f anaphors: a. English YES b. Standard Danish NO c. Dial. Danish YES/NO ??? Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 2 8 5.2.1.4.1 Absence of inalienable possession in English In this section we argue that the differences between English and Danish w ith respect to inalienable possession, i.e. the fact that D anish allows inalienable possession o f tbe type illustrated in (59) while English does not (60), do not constitute a serious problem for the present approach to binding and intensification. (59) Inalienable possession in D anish: Peter vaskede hcenderne. Peter w ashed hands-the “'"(a) ‘Peter w ashed his own tw o hands.’ (inalienable) °'‘(b) ‘Peter w ashed tbe hands (o f the body he was dissecting).’ (alienable) (60) A bsence o f inalienable possession in E nglish: Peter w ashed the hands. *(a) ‘Peter w ashed his own tw o hands.’ (inalienable) °'‘(b) ‘Peter w ashed the hands (o f the body he was dissecting).’ (alienable) As discussed in chapter 3, section 3.3.2.6, we follow V ergnaud & Zubizarreta in assum ing that the inalienable possessed DPs in (59a), contains an em pty possessor argum ent. That is, the readings in (59a) vs. (59b) are obtained by the representations in (61a) and (61b) respectively"’. (61) a. PeteVi vaskede hcenderne(Xi) Peter, w ashed hands-the(x,) b. Peteri vaskede hcenderne Petep w ashed hands-the I f inalienable possession is analyzed as binding o f phonetically unrealized possessor argum ents (i.e. 0-reflexives), and if English is assum ed to have 0-reflexives, then why does English not allow inalienable possession as shown in (60a)? W hile this question may, at first, seem to con stitu te a serous prob lem for the present approach, it m o st lik ely is not. L an guages sim ply differ (param etrically) with respect to w hether (and where) they allow inalienable possession. French, w hich is sim ilar to English in that it does not have possessive reflexives Vergnaud and Zubizarreta (1992) suggest that hcenderne(x) is derived from hcenderne by a lexical redundancy rule. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 329 (e.g. in the sentence Pierrej mange soUj/^ pain ‘Peteri eats hisi/^ bread’ son/sa ‘his, h er’ is am biguous in the sam e way as English his), patterns with Danish in allow ing inalienable possession, see (62). (62) Inalienable possession in French: Pierre lave les bras. ‘ ’'‘(a) ‘Peter washed his own tw o hands.’ (inalienable) ‘ ’'‘(b) ‘Peter washed the hands (o f the body he was dissecting).’ (alienable) As this cross-linguistic data indicates, the presence/absence o f inalienable possession in a given language is independent o f the presence/absence o f 0-reflexives in argum ent position. French does not have 0-reflexives (having se ‘REFL’ in the contexts where Danish has sig and English 0 ) , but it still allows inalienable possession. Furtherm ore, in spite o f the fact that Danish has a possessive reflexive (which is different from the possessive pronoun) it still allows inalienable possession. 5.2.1.5 0-reflexives in PPs and prepositional predicates in English In chapter 3 section 3.3.6 we argued at length that phonological factors (i.e. non-contrastive sentences stress, etc.) interfere w ith the distribution o f simple and com plex reflexives in certain prepositional predicates. In this section we argue that the absence o f simple unintensified 0-reflexives in prepositional predicates can be explained along sim ilar lines. The 0-reflexive analysis o f M odem English proposed here predicts that the 0-reflexive should behave like the sim ple unintensified reflexive sig ‘R E FL ’in Danish. As shown in (63-5) this predication is supported by the behavior prepositional predicates in English. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 330 a. tale til ‘speak to ’ b. pege pd ‘point to/at’ c. dramme om ‘dream about’ d. lytte til ‘listen to ’ e. sigte pd ‘aim a t’ f. Hans peger pd *sig / sig selv / bilen. Hans is pointing at * 0 / 0 him self / the car Prenositional nouns in Enalish and Danish: a. et billede a f ‘a picture o f’ b. en beskrivelse a f ‘a description o f’ c. en hyldeste til ‘a tribute to, an hom age to ’ d. en bog om ‘a book about’ e. en evaluering a f ‘an evaluation o f’ f. . . . . en evaluering a f *sig / sigselv / bilen. .... an evaluation o f * 0 I 0 him self / the car Prepositional adiectives in E nalish and Danish: a. sikker pd ‘sure o f b. interersseret i ‘interested in ’ c. optaget a f ‘occupied b y ’ d. tilfreds med ‘satisfied w ith’ e. glad fo r ‘pleased with, like’ f. forskellig fra ‘different from ’ g Peter er tilfreds med *sig / sig selv / bilen. Peter is satisfied with *0 / 0 him self / the car In both English and Danish, m ost prepositional constructions allow com plex reflexives but exclude sim ple unintensified reflexives. There is, however, a num ber o f mism atches between the languages. As m entioned in chapter 3 section 3.3.6, a small group o f prosodically heavy prepositions (‘snake’-prepositions) do allow simple sig, e.g. bag ‘behind’ in (66a). (66) a. Han plantede flaget hag sig / sig selv / Mary b. He planted the fla g behind * 0 / 0 himself/ Mary The Danish verb+PP construction plante X bag Y ‘plant X behind Y ’ in (66a) behaves as a neutral construction. A ssum ing parallelism betw een Danish sig and English 0-reflexives, we w ould expect the English verb+PP construction plan X behind Y to behave similarly. As the ungram m aticality o f the 0-reflexive in the English exam ple in (66b) shows this prediction does not hold. While in (63-5) the absence o f sim ple unintensified 0-reflexives in can be R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 331 argued to be due to the same prosodic factors excluding sim ple unintensified sig in Danish, the m ism atch between English and Danish illustrated in (66) points to a difference betw een English 0-reflexives and D anish sig. That prosodic explanation o f the absence o f 0-reflexives with certain prepositions is supported by the difference betw een unstressable se ‘R EEL’ and stressable soi ‘REEL’ in Erench; only the the latter is found in prepositional predicates, see (67) (67) a. Pierre est fier de soi. b. *Fierre est fie r de se. Peter is proud o f REEL ‘Peter is proud o f h im se lf’ Like its English counterpart, the sim ple 0-reflexive, the Erench reflexive clitic se ‘REEL’ also never occurs with any prepositions, see (68). (68) Pierre aplante le drapeau derriere *se /soi-(meme). Peter has planted the flag behind SE / SE-(sam e) ‘Peter planted the flag behind h im se lf’ As discussed in chapter 3, section 3.2.4, the differences between D anish sig and Erench se follow from the fact that the form er is syntactically independent prosodically bound word, which does not form a m orphological constituent w ith its host verb, w hile the latter fuses m orphologically w ith the verb w ith w hich it does form a constituent. The fact that English 0 - refiexives behave like Erench se in never occurring w ith prepositions, indicate that they may be o f the same nature, i.e. clitics w hich need to fuse m orphologically with a verbal host with w hich they then form a constituent. The above data show that in m any cases w here the sim ilarities betw een Dan. sig and Eng. 0 break down 0 actually patterns w ith the Erench reflexive clitic se ‘REEL’. The leads us to the conclusion that 0-reflexives have m ore in com m on with Rom ance reflexive clitics'^ '^As shown here, it unintensified 0-reflexives can only be found with verbal predicates. Notice that the same generalization holds for 0-reciprocals, see (i-ii) (i) a. Peter and Mary kissed 0 / each other. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 332 than w ith M ainland Scandinavian sig/seg and D utch zich. For the cases where Fr. se and Eng. 0 display different behaviors (e.g. ECM construction (where French se is ok while English 0 - reflexives are excluded)), we tentatively suggest that these are due to the m orpho-syntactic peculiarities o f phonologically unrealized elements. 5.2.1.6 Potential problem: stressed and unstressed forms himself It has often been claim ed that the uses o f x-^e//’forms as reflexives and intensifiers can be distinguished based on stress. X -se lf form s used as adnom inal intensifiers are stressed, as opposed to x-self forms used as reflexive anaphors w hich are unstressed. Siem und (2000:82-3, (3.173)) proposed the generalization in (69). (69) English x-self \s am biguous betw een: a. a reflexive anaphor, w hich is unstressed {x-self) and b. an intensifier [...], which is stressed (x-SELF). A ccording to Siem und (2000), the assum ption that x-self forms are am biguous as indicated in (69) m akes it possible to reduce binding conditions to (70). (70) a. An anaphor (i.e. unstressed x -s e lf is bound in co-argum ent position, b. A pronom inal is free in a co-argum ent position. Siemund, follow ing Baker ( 1 9 9 5 ) suggests that locally free x-SELF form s should be analyzed as intensified 0-pronouns, e.g. (71). (71) Johrii believes that letter was sent to everyone hut [h im j himSELF. This analysis is based on the assum ption that him self \s stressed in sentences like (71) but not in sentences like (72). (72) Peteri envies 0 , him self b. Peter and Mary fought 0 / each other. (ii) a. Peter and Mary pointed at *0 / each other. b. Peter and Mary are proud o f *0 / each other. Baker (1995) in turn is inspired by the analysis of certain instances o f locally free him self as deriving from underlying him him self proposed in Ross (1970). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 333 W hile there appears to be a tendency for x -se ^ fo rm s to be stressed in (3) but not in (4), we will argue that this does not w arrant the jum p to the conclusion (69) that adnom inal intensifiers are always stressed w hile anaphors are not. Siemund him self acknow ledges this and cites a num ber o f counterexam ples, see (73) (73) a. John noticed a picture o f him self in the post office. b. Bill likes stories about him self In spite o f the fact that the x-self forms in (73) are predicted by Siem und to be intensifiers intensifying 0-pronouns by Siemund, they are nevertheless perfectly ok with out stress. Rather than abandoning the analysis o f the x-self form s in (73) as intensified 0-pronouns, we suggest to save the analysis by questioning the very generalization that adnom inal intensifiers ALWAYS have to be stressed. Rather than being an intrinsic property o f intensifiers, stress placem ent is determ ined by a num ber o f factors, e.g. focus structure o f the sentence, non- contrastive sentence stress, stressability o f lexical elem ents (e.g. Dan. sig, Fr. se and English 0-reflexives, are all unable to carry stress and thus have to be lexically reinforced to host stress, see section 3.3.6). That this account stands a better chance o f unraveling the m ysteries o f the interactions between stress placem ent and intensification is suggested by the fact that, as observed in Siem und (2000:83), in the G erm an translations o f (73) the reflexive sich can occur w ithout the intensifier selhst. In contrast, the D anish translations o f (73) would require sig to be intensified by adjunction o f selv. As argued in chapter 3, section 3.3.6'®, this difference betw een D anish and Germ an follow from the fact that Danish sig is unstressable while G erm an sich is stressable. As argued in 5 .2 .1.5, English 0-reflexives have to be intensified (but not necessarily contrastively stressed) in sentences like (73) for prosodic reasons, i.e. just like Danish sig, they cannot carry non-contrastive sentence stress on their own and thus have to be adnom inally intensified to do so, see also chapter 3, section 3.3.6. The question o f ’ And as suggested in Siemund (2000). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 334 w hether this approach necessitates the assum ption o f tw o different kinds o f intensifiers: (i) intensifiers w hose presence is sem antically m otivated, and (ii) intensifiers w hose presence is prosodically m otivated, is discussed in section 3.3.6. For now, suffice it to say that the relationship betw een stress and intensification is far from clear and certainly not clear enough to w arrant the assum ption that all adnom inal intensifiers m ust be contrastively stressed. Awaiting further, more conclusive, research on the m atter we tentatively assum e that the generalization that adnom inal intensifiers M U ST have stress (cf. lb ) is too strong. A s argued here, the interrelations between stress and the distribution o f adnom inal intensifiers is far to com plex to be captured by the sim ple generalization that intensifiers m ust be stressed. Indeed, to proceed from the assum ption that intensifiers m ust be stressed as a god-given fact, am ounts to taking the solution to the problem under investigation as one’s point o f departure. The w hole point o f the present dissertation is to show that the dividing line betw een intensifiers and reflexives is far from clear and that a great deal o f m isunderstandings have been generated by m istaking traditional assum ptions about w hat is a reflexive and w hat is an intensifier for solid facts. If, as proposed here, all instances o f x-self forms in English are adnom inal intensifiers, then the assum ption that intensifiers are all m andatorily stressed m ust be w rong and should be revised. Logically, there is no way to decide which assum ption (that all intensifiers m ust be stressed, or that all x-self forms are intensifiers) is correct. It is, o f course, an empirical question w hich m ust be decided by looking at the facts. At this point, w hile the final answ er is still up for grasp, we hope to have presented enough data to show that the assum ption that all x-s'e/f forms are intensifiers is at least as plausible as, if not m ore plausible, than the assum ption that x-self form s are am biguous between unstressed reflexives and m andatorily stressed intensifiers. Furtherm ore, we hope to have called attention to the need to distinguish betw een non-contrastive stress^® contrastive stress. As discussed in chapter 3, section 3.3.6, all bi-moraic words are assumed to required (by phonological rules R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 335 5.2.2 Non-reflexive uses of 0-reflexives in English Pushing the equivalence o f English 0-reflexives and Danish sig to the extrem e, we should expect to find a large degree o f degree sim ilarities in their non-reflexive uses. In this section we argue that 0-reflexives function the sam e way as sim ple sig in reciprocal constructions (section 5.2.2.1), middles (section 5.2.2.2), and (deponent) inherently reflexive verbs section 5.2.2.3). 5.2.2.1 Reciprocals The observation that in many languages (e.g. Rom ance languages and Germ an) the simple unintensified reflexive (se/si/sich) can be used as both reciprocal and reflexive can be used to bolster both our analysis o f 0-reflexives in M odern English, see (74). (74) a. [Peter and Maryji kissed 0 , / each otheri. b. [Peter and Mary]i fought 0 , / each otheri. See chapter 4, section 4.5.2 for discussion o f the use o f the reflexive suffix -s , w hich is derived from an encliticized reflexive ~sik, to form reciprocal forms o f certain verbs in Danish. The evolution o f 0-reciprocals in M odem English can also be used bolster our analysis o f the historical developm ent o f 5e/f-intensifiers, see also section 5.3 below. In the same way Old English (O E) locally bound pronouns were replaced by 0-reflexives in Early M odern English (EM E), see (75a), so w ere also locally bound pronouns w ith reciprocal readings replaced with 0-reciprocals, see (75b). governing the prosodic structure of words and sentences) to have some form of stress. In contrast, mono-moraic clitics (e.g. Danish sig, French se, etc.) must be intensified (i.e. phonologically fortified by addition o f phonological material, e.g. by adjunction o f selv in Danish, and by segmental fortification of mono-moraic se to bi-moraic soi in French). As argued in chapter 3, section 3.3.6, if, for some reason, the prosodic rules governing non-contrastive stress require that non-contrastive stress fall on a clitic, then the clitic must be phonologically fortified to be able to carry stress. Consequently, this form o f intensification (e.g. Dan. sig selv) does not necessarily require the intensifier to be contrastively stressed. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 336 (7 5 ) O ld Enslish: » fEarlvl M odern English: a. Locally bound sing. pron. » 0-reflexive 7/e, washed hirrii. ‘He w ashed him self’ » He, washed 0 , (himself). b. Locally bound plur. pron » 0 - reciprocal Theyi kissed thenii. ‘They kissed 0 /each other.’ » They, kissed 0,. In other words, locally bound expressions in Old English (be they reflexive or reciprocal) were all realized as pronouns. The change from OE to EM E can thus be capture saying that locally bound pronom inals turned 0-anaphors in EME, see (76). (76) a. (O E) pronom inal reciprocals » (EM E) 0-reciprocals b. (OE) locally bound pronouns » (EM E) 0-reflexives This kind o f evidence illustrates the close affinity betw een reflexives and reciprocals and provides additional support for the analysis o f the assum ption that M odern English has 0 - anaphors (which are functionally equivalent to Dan. sig, Fr. se and Ger. sich) as well as for the historical evolution o f 0-reflexives/reciprocals proposed in section 5.3. S.2.2.2. Middles Danish and French use simple unintensified reflexives (i.e. sig and se) to form deagentive inchoatives, see abne sig /s’ ouvrir ‘(unintentionally) open’ (77b) and (79b) which is derived from the original causative transitive form o f the verb abne/ouvrir ‘(intentionally) open som ething’ in (77a) and (79a). The sentences in (78) illustrate that sim ilar causative- inchoative pairs an be found in English, w here the simple unintensified 0-reflexive functions as Fr. se and Dan. sig. (77) a. Peter abner d0ren. (agentive causative transitive) Peter opens door-the ‘Peter opens the door.’ b. Doren dbner sig. (deagentive inchoative) door-the opens REFL ‘The door opens.’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 337 (78) a. Peter opens the door. b. The door opens 0 . (79) a. Pierre ouvre la porte. Peter opens the door ‘Peter opens the door.’ b. La porte s ’ ouvre. the door REFL opens ‘The door REFL opens.’ 5.2.2.3 Inherently reflexive/deponent verbs The sentences in (80-1) illustrate typical deponent inherently reflexive verbs in Danish and English, i.e. verbs w hich only allow sim ply unintensified reflexives in object position. (80) a. Peter anskaffede sig et hus. b. *Peter anskaffede sig selv et hus. c. *Peter anskaffede Marie et hus. Peter acquired 0 / 0 self / M ary a house ‘Peter acquired (*him self/*M ary) a house.’ (81) a. Peter acquired 0 a house. b. *Peter acquired 0 him self a house. c. *Peter acquired Mary a house. W hile in Danish the presence o f the sim ple reflexive is beyond doubt since it is phonetically realized, the existence o f an unintensified 0-reflexive in (81a) is harder to prove. Keyser and Roeper (1992) claim to have found a binding-independent way to decide w hether a given predicate has zero argum ents. A ccording to them, predicates which can take the prefix re-, can have a zero argument, e.g. re-acquire. In contrast, predicates which cannot take the prefix re do not have zero argum ents, e.g. *re-come, *re-arrive. 5.5.2.4 Summary The fact that English 0-reflexives pattern w ith overt unintensified reflexives in German, Danish, and French supports the argum ent that they are functionally equivalent, see (82). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 338 (82) Typological survey o f reflexive and non-reflexive uses o f reflexive elem ents: English Danish French German a. reflexive 0 sig se/soi sich b. reciprocal 0 -s se sich c. middle 0 sig se sich e. inh.relf/depon. 0 -s/sig se sich As shown in chapter 2, English x -se/f forms have m ore in com m on w ith adnom inal intensifiers (e.g. Dan. selv, Ger. selbst, Fr. (lui-)meme), see (83). (83) Typological survey o f different uses o f adnom inal intensifiers: English Danish French German a. Adnom. inten. the king himself kongen selv le roi (lui-)meme der Konig selbst b. Complex refl. 0 himself sig selv soi-meme sich selhst d. Doppelganger-mark 0 himself sig selv soi-meme sich selbst e. Inten. pron./logophor [him] himself ham selv [lui] lui-meme ihn selbst f. Exclus.adv. int. himself selv lui-meme selbst g. Indus, adv. int. himself selv lui-meme selbst Com bined, the evidence sum m arized in (82-3) provide strong support for the analysis based on the assum ptions that English has 0-reflexives and that English x-self form?, are ALWAYS intensifiers. 5.2.3 Intensified and unintensified pronouns in English In this section we propose an account o f sim ple and com plex pronouns in English, e.g. he him self and him himself/[him] him self based on the assum ption that com plex pronouns are form ed by adjunction o f the adnom inal intensifier him self, in the same way as intensified DPs, e.g. the king him self Both subject and object pronouns fall under the sam e principle B o f the binding theory, i.e. they have to be locally free. Furtherm ore, sim ple and com plex pronouns behave alike with respect to the binding theory. Any differences betw een them are assum ed to follow from the m odule o f intensification rather than from binding related factors. As m entioned in the introduction, the present analysis thus builds on and extends sim ilar ideas expressed in earlier w ork by Ross (1970), Zribi-H ertz (1989,1995), M cKay (1991), Baker (1995) and Konig & Siem und (1999) and others. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 339 5.2.3.1 Intensified subject pronouns O ur analysis o f intensified subject pronouns, e.g. he him self and Dan. han selv ‘he s e lf , as form ed by a general, binding independent process o f adnom inal intensification has already been presented in chapter 3, section 3.4.2 and will therefore not be repeated here. S.2.3.2. Intensified object pronouns In this section we argue that the analysis o f intensified object pronouns in Danish, e.g. ham selv ‘him s e lf , see (85), can be extended to English, see (84). (84) Peter, said that Mary danced with everyone except [him jhim self. (85) Peter I sagde at Marie dansede med alle andre end ham selv. Peter said that M ary danced w ith all others than him self ‘Peter said that M ary danced with everyone except h im se lf’ In both languages the intensified object pronouns is com posed o f a pronom inal + the adnom inal intensifier. The only difference is that the pronom inal part o f the English intensified pronoun is made inaudible by some process o f haplological reduction, while it is phonetically realized in Danish. In Danish, the com plex pronoun ham selv ‘h im self, e.g. (3d), can be straightforw ardly analyzed as an intensified form o f the object pronoun ham ‘him ’, see (3e). W hile English has an intensified form o f subject pronouns which can be analyzed as a com bination o f pronoun + adnom inal intensifier, e.g. he himself, the corresponding form for the object pronoun, which would be him him self is extrem ely rare (though not unattested) and is generally not accepted by native speakers. Instead o f him him self the reduced form him self is usually used^' See chapter 3, section 3.4.2 for discussion o f the sem antic/pragm atic principles governing the distribution o f intensifier-adjunction to object pronouns. Note that English Dialects vary with respect to their use of locally free himself. According to Baker (1995), American English speakers use stressed HIM in many cases where British English speakers would use himself. While this has later been shown by Siemund (2000) to be to strong a claim, it is nevertheless true that dialects differ substantially with respect to their use of locally free s.-self R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 340 S.2.3.3 Summary: binding of intensified and unintensified pronouns W hile only a rough sketch, the analysis o f intensified pronouns in English outlined above, suggest that it may be possible to extend the analysis o f Danish pronom inals presented in chapter 3, section 3.4 to English, and thus supports the viability o f the particular separation o f binding and intensification proposed here. 5.2.4 Conclusion In the above sub-sections o f section 5.2 we have discussed the cross-linguistic viability o f the analysis binding and intensification in D anish proposed in chapter 2-4, by show ing to w hat extent it can be extended to account for sim ilar phenom ena in M odem English. W hile a num ber o f differences betw een the tw o languages were noted, e.g. the absence o f simple unintensified 0-reflexives in ECM constructions and resultatives, we argued that sim ilarities outw eighed the differences. We also argued that m ost differences can be accounted for by reference to factors not related to either binding or intensification, e.g. prosodic differences It is to the best of my knowledge still an open question why English does not allow intensified object pronouns, i.e. why do forms like *IV.me myself, *nihim him self tend to be shunned by native speakers. Haplology might be a plausible account of the absence of the forms him himself, her herself them themselves but seems to run into trouble with forms like me m yself and us ourselves. Baker (1995) suggests that the absence of intensified object pronouns might be a consequence o f his prominence condition (see (100) in section 3.5.1). If intensifier-adj unction is sensitive to syntactic prominence and if subjects are assumed to be syntactically more prominent than objects then it follows that only nominal expressions in subject position can be intensified. Such an account, however, is unable to explain why Danish allow intensified object pronouns and reflexives and why, even in English, sentence like Mary wrote a letter to the King him self are ok in the right context. Notice that Chinese and French display a similar tendency to avoid overly redundant forms. Both French and Chinese have both simple and complex forms of the adnominal intensifier, i.e. Fr. meme and lui-meme, see (i) and Ch. ziji and ta ziji, see (ii). (i) Le roi meme / Le roi lui-meme (French) the king self the king him-same ‘The king him self ‘The king him self (ii) Huang-di ziji / Huang-di ta ziji (Chinese) emperor self emperor him self ‘The emperor him self ‘The emperor him self’ While both the complex and the simple form of the intensifier can be used to intensify DPs, only the simple form can be used to intensify pronouns, see (iii) and (iv). (iii) Lui- meme / #Lui lui-meme (French) him same him him-same (iv) ta ziji / *ta ta ziji (Chinese) him self him him self R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 341 between overt (sig) and non-overt reflexive elem ents (0-reflexives), inalienable possession, etc. All in all, in spite o f the fact that the analysis o f English outlined here may have raised m ore questions that it answers, we hope to have show n that this approach is at least a viable alternative w hich deserves to be explored further. In the following section, we show that the 0-reflexive analysis o f English presented here allows for interesting new solutions to hitherto unansw ered questions concerning the historical developm ent o f reflexives and intensifiers from O ld English (OE) to Early M odern English (EM E) and eventually M odern English. 5.3 Historical development of English intensifiers The G erm anic languages have had intensifiers o f the self-type for as long back as historical records go. For this reason very little is known for certain about the early historical developm ent o f self-type intensifiers. Section 5.3.1 provides a b rief sum m ary o f current theories o f the historical developm ent o f G erm anic intensifiers. Similarly, section 5.3.2 contains an overview o f previous accounts o f the historical developm ent o f M odern English x- s e lf forms. In section 5.3.3 it is argued that the assum ptions that English has 0-reflexives and that the com pound him self is always an intensifier provide the key to a coherent, unified account o f a num ber o f the hitherto unanswered questions concerning the developm ent o f the M odem English x-self form intensifiers. Finally, section 5.3.4 sum m arizes the results and concludes section 5.3. 5.3.1 The historical development of the Germanic/Modern Danish intensifier selv. In (86a) are listed various cognates o f selv ‘self’ in other Germ anic languages and in (86b) the reconstructed Germ anic form can be found. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 342 (86)^^ (a) Cognates o f selv in the G erm anic languages^"*: Danish selv Swedish sjdlv Old D anish sia lf O ld Norse sjalfr Faroese sjdlvur N orw egian (bokm al) selv N orw egian (nynorsk) Sj0l English -self O ld English se(o)lf s y lf se lf Germ an selbst O ld High Germ an selp Dutch ze lf Afrikaans s e lf Gothic silba (b) Reconstruction form o f Com m on Germ anic origin: Germ. *selba- , *selhan In m any languages the m orphological m aterial o f which the intensifiers (as well as reflexive elem ents) are m ade can be show n to derive from expressions for ‘body parts’, see (87). (87) ^ ^ a. Georgian tviton ‘body’ b. Okinaw an du:na ‘body’ c. Arabic nafs ‘soul’ d. H ausa ni daikana ‘I w ith my head’ e. H ungarian maga ‘seed’ f. Japanese ji-sin ‘ow n body’ g. H aitian Creole tet-Emile (lit.)’head E m ile’ h. Classical Chinese ^ shen ‘body’ A ccording to E. Konig (1996:10), the cross-linguistic evidence in (87) provides support for G rim m ’s suggestion that G erm an selb-st originally derived from the possessive construction si-lib ‘his body’, see (88). (88) 5e/v(D an.) « < *selba-(Germ.) < « *si ‘h is’ *lib-, *liba ‘life, body’ The examples in (86a) as well as the reconstructions in (86b) are partly from Niels Age Nielsen (1989) Dansk Etymologisk Ordbog. Ordenes Historic. Gyldendalske Boghandel, Nordisk Forlag A/S, p. 363, partly from Siemund (1999:22). Cf. Siemund (1999:22) ‘A notable exception is Yiddish, which has lost SELF and uses ateyn (cf. E. alone) as the intensifier.’ The loss o f self-Xype intensifiers in Yiddish could, perhaps, by due to influence from neighboring Slavic languages also have adnominal intensifiers meaning ‘alone’, e.g. Russian sam ‘alone, him self. Cf. Moravcsik (1972), Schladt (1995). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 343 Based on this kind o f data Konig furtherm ore suggests that intensifiers play a prom inent role in the historical developm ents o f reflexive pronouns, see the path o f gram m aticalization illustrated in (89). (89)^® body parts > intensifiers > reflexive anaphors Konig thus takes intensifiers to be the (m issing) link between words for body parts and reflexive anaphors. The fact that intensifiers and reflexives are closely related historically in m any languages does not contradict our basic assum ption that binding and intensification belong to separate m odules o f the grammar. Instead, as discussed in the follow ing section, it helps us understand why their synchronic functions have a strong tendency to overlap. 5.3.2 Previous account of the evolution of the Modern English se/^forms M odern English reflexives and intensifiers constitute a typical exam ple o f apparent convergence o f the formal realization o f intensifiers and reflexives and can thus be used to illustrate the urgent need to relegate binding and intensification to separate m odules o f the grammar. Indeed, we argue that it is the failure to correctly separate binding and intensification that has prevented earlier accounts from arriving at a unified analysis o f the evolution o f M odern English x-5e//” forms. To the best o f our knowledge, all the existing studies o f the historical developm ent o f the com pound .seZ/'-forms are based on the assum ption that these elem ents function as reflexive pronouns in M odern English. As argued in the previous sections, there is plenty o f data indicating that this assum ption may be wrong. That is, as argued in chapter 5, M odern English has clitic-like 0-reflexives which may be intensified by adjunction o f the appropriate x-self adnom inal intensifier, e.g. 0 himself, 0 ourselves, etc. In other words, the x-.se//form s are never true reflexives but rather consistently function as adnom inal intensifiers in all contexts. ’ Cf. K onig(1996:10). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 344 B efore we develop our analysis o f the evolution o f 0-reflexives and self-^orm in M odem English let us first briefly sum m arize the status quo o f the history o f English himself. Basically m ost (if not all) existing analyses agree that the desire to disam biguate betw een coreference readings and disjoint reference readings o f local pronouns was (one of) the main m otivation for the developm ent o f the M E ^eZ/'-forms. In this respect, Siem und’s (1999:25-30) account is quite representative^’. O ld English did not have reflexive pronouns and thus used personal pronouns instead. This yielded a situation in w hich third person personal pronouns were am biguous between co reference and disjoint reference interpretations, see (90). (90)^^ a. hine he heweradh m id wcepnum [TEGram 96.11 ] ‘he defended himself with w eapons’ b. dha^‘ ^ behydde Adam hine & his w if eac swa dyde [Gen 3.9] ‘and Adam hid himself and his wife did the sam e’ A ccording to m ost traditional accounts the intensifier ‘self’ was added to the am biguous local pronouns for the sake o f disam biguation, see (91-2). (91)^° a. se Hcelende sealde hine sylfne fo r us^' [A iL et4 1129] ‘The Saviour gave himself for u s.’ b. He [Moses] sceawode hine selfne, & pinsode, dha dha him dhuhte dhcet he hit doon ne meahte, ... [CP 7.51.14, translation provided] ‘ He contem plated himself, and thought that he could not do it, ... ’ (92) a. he ... seldh G odehis (ehta, & hine seifne diobule [CP 44.327.23,] ‘He gives his possessions to God and him self to the devil.’ There is a large number o f very interesting and valuable studies of the history of English reflexives all which deserves to be mentioned in this context, e.g. van Gelderen (2000), Keenan (1994, 2003), Penning (1875), Farr (1905), Ogura (1989), and many others. Unfortunately, spaces considerations prevents us from discussion all these works here. 28 This example is from Siemund 1999, (2.44). The letter sequences ‘dh’ and ‘th’ are used here to transcribe the Old English letters for the voiced and voiceless (inter-)dental fricatives respectively. These examples are from Siemund 1999 (2.45-6). Bold face is added to Siemund’s examples to help the reader locate the relevant intensified expressions. The numbers of Siemund’s examples have also been modified to fit into the present context. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 345 b. Hannibal ... hine selfne mid atre acwealde [Or 4 11.110.2] ‘Hannibal killed himself w ith poison.’ W hile this account seems to be very intuitive, it cannot be the whole story, since, as Siemund observes, not all intensified object pronouns were locally bound, see (93). This indicates that disam biguation o f local pronouns was not the only function o f intensifiers. As argued in chapter 2 and 3 above, we argue that intensifier adjunction to locally bound elem ents is due to other sem antic factors (i.e. presupposed representational non-identity) rather than directly linked to binding and that such an account allows for the potential unification o f all used o f intensifiers. (93)^^ Be dham cwcedh se cedhela lareow sanctus Paulus: Ic wille dhcet ge sien wise to gode & bilwite to yfele. Ond eft be dhcem cwcedh Dryhten dhurh hine selfne to his gecorenum: Beo ge swa ware sua sua ncedran & sua bilwite sua culfran. [CP 35.237.18, translation provided] ‘Therefore the noble teacher St. Paul said: “I w ish ye to be w ise for good and simple for evil. A nd again, the Lord spoke about the same thing through himself to his elect: “Be cunning as adders and simple as pigeons.’” W hile the traditional disam biguation account seems very persuasive, we thus argue that it is false since it is based on w rong assum ptions. Siem und him self admits that such disam biguation accounts run into a num ber o f problem s. Notably, they do not explain why the fusion o f s e lf and pronom inal elem ents also took place in the first and second persons where there could be no am biguity betw een coreference and disjoint reference. As discussed in Siem und (1999:25-30), “analogy” has often been adduced as answ er to the question o f why 5e//^adjunction happened in th eT ' and 2"'' (cf. Penning (1875:13)). In contrast, the analysis proposed here provides a straightforw ard explanation o f why the selfH om ^ em erged in the L ‘ and 2"‘ * persons as well as in the 3"‘' person. In a nutshell, our proposal is that 5c//^adjunction to pronouns in OE was not prim arily m otivated by a need to disam biguate between coreference and disjoint reference but rather by a need to adnom inally intensify locally bound sim ple ■ This example is from Siemund 1999 (2.47). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 346 pronom inals occurring in object position o f anti-reflexive or ‘hidden neutral’ verbs, e.g. (91-2). That is we propose that .se//^adjunction to locally bound pronom inal in O E follows the same pattern as se/v-adjunction to sim ple reflexives w ith anti-reflexive predicates in M odern Danish (e.g. Jeg mistcenker *mig / mig selv ‘1 suspect *me / me s e lf), see discussion o f Danish reflexives in chap. 3. A nother problem encountered by previous accounts is the lack o f explanations for why the com pound self-^ovms cam e to be used as adnom inal intensifiers (e.g. ME the King himself) and why the OE sim ple intensifier self'was abandoned. The exam ples in (9) below show that Old English used the sim ple form se lf as adnom inal intensifier. If that is the case, then why did OE abandon the use o f s e lf as adnom inal intensifier in favor o f the self-^om s, w hich according to standard account developed as a com plex reflexive pronoun? (94)^^ a. se cyning s y lf a [GD 14.131.3] ‘the king him self b. Swa swa Crist s y lf cwcedh [43Let3 173] ‘as C hrist himself said’ A s discussed below, this problem becom es m uch easier to understand once we realize that the self-forms did N O T develop as com plex reflexive pronouns but rather were adnom inal intensifiers right from the first m om ent the pronom inal and the adjoined se lf fusod into one word. Finally, let us take a closer look at an often overlooked aspect o f Old English, which we, follow ing suggestions by Siem und (2000), believe to be crucial to the developm ent o f 0 - reflexives in M odern English. Old English had obligatory pronouns in m any contexts where M odern Danish, G erm an and French have non-them atic reflexives, i.e. with deponent inherently reflexive verbs, see Old English exam ples in (95) and G erm an and D anish exam ples in (96), see also the discussion o f non-reflexive uses o f 0-reflexives in section 5.2.2 above. ' This example is from Siemund 1999 (2.48). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 347 (95) 34 (96) 35 a. .. .dhcet dhu dhin scamige, Sidon " that you be ascham ed, Sidon.’ b. hit is cyn dhcet we ure scomigen ‘It is appropriate that we be asham ed’ c. he gereste hine on dhone seofothan dceg ‘He rested on the seventh day.’ a. Nero wandte sich /*sich selbst an Paul. N ero turned REFL/REFL self to Paul ‘N ero turned to Paul.’ b. Er schdmt sich/*sich selhst. He sham ed REFL/REFL self ‘ He was ashamed. ’ c. Peter skammede sig/*sig selv. Peter sham ed REFL/REFL self ‘Peter was asham ed.’ d. Peter hvilte sig/*sig selv. Peter rested REFL/REFL self ‘Peter rested.’ [CP 52.409.33] [CP 52.407.15] [Gen 2.2] (Germ an) (G erm an) (D anish) (D anish) Siem und (1999:25-30) draws attention to an often overlooked aspect o f the evolution o f M odern English, nam ely the loss o f locally bound pronouns in object position o f verbs like the ones exem plified in (95). He also suggests that we need to take the question “w hy English did not start anew to develop pronouns w ith these functions” into consideration. A s will be shown below, the single assum ption that at a certain point in history a recent ancestor o f M odern English developed a 0-reflexive, provides the key to a coherent, unified account o f all the four questions raised in the passages Ifom Siem und (1999) quoted above and sum m arized in (97). (97) Q uestions concerning the developm ent o f the \-s e lf forms in M odem E nglish: (a) W hy was the fusion o f pronoun and se lf not confined to the third person, but was extended to the first and second persons? This rem ains a m ystery for all the disam biguation accounts since no am biguity arises in these cases. (b) W hy did the newly developed .se//^compound forms also com e to be used as intensifiers? (c ) W hy did E n glish abandon its original in ten sifier seip. S in ce Old E n glish s e lf was, used in m uch the sam e way as its m odem equivalent and there seem s to have been no obvious reason to replace it w ith a new intensifier. These OE examples as well as the German examples in (96) are from Siemund (2000). The absence of intensified forms of the reflexive in these examples is, o f course, due to the fact that they involve inherently reflexive predicates. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 348 (d) W hy were the pronouns in the usages in (95) abandoned and, once they were lost, why English did not start anew to develop pronouns with these functions? 5.3.3 A new account of the evolution of the Modern English veZ/'-forms In the follow ing we will outline an alternative picture o f the evolution o f the English system o f reflexives and intensifiers w hich provides new answers to the questions listed in (97). As will be shown, the problem w ith m ost o f these questions is that they are based on the w rong basic assum ptions and consequently m ake us ask the w rong questions. O versim plifying the im m ense com plexity o f the evolution o f M odern English som ew hat we have decided to cut the period from spanning from Old English to M odem English into three stages: (i) Old English, (ii) Early M odern English, and (iii) M odern English. For ease o f presentation a few other sim plifications have been m ade as well, see (98). (98) Sim plifications adopted in the follow ing sections: (a) In order to facilitate com parison betw een different stages, M odern English spelling and m orphology has been used even in the Old English examples. (b) Case com plications have been ignored. That is, the fact that first and second person form s (i.e. myself, yourself, ourselves, yourselves) are com posed o f a pronoun in the genitive plus s e lf whereas the third person forms (i.e. himself, herself itself themselves) are com posed o f pronouns in accusative or dative case plus se lf has been ignored. 5.3.3.1 Stage one: intensification and binding in Old English Old English was a language w ithout m orphologically specialized reflexive pronouns. This m eant that pronouns could do double duty as either reflexives or pronouns. See the exam ple in (9 0 ), repeated in (99). (99) hinei/^ he; heweradh m id wcepnum [.diGram 96.11 ] ‘hei defended himself;/himj with w eapons’ (=Siem und 1999, (2.44)) A s shown in (99), locally bound pronouns were potentially am biguous between coreference and disjoint reference readings. A ccording to m ost existing accounts o f the evolution o f \-s e lf Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 349 forms (see section 5.3.2), se lf was adjoined to locally bound pronouns to disam biguate in favor o f coreference readings, see (92) repeated in (100). (100) Hannibal, ... hine, * ^ selfne m id atre acwealde [Or 4 11.110.2] ‘Hannibal, killed himselfyt^ w ith poison.’ (=Siem und 1999 (2.46b)) Unlike, (99) w here the pronoun is am biguous, (100) tend to have only the coreference reading o f the locally bound object pronoun hine ‘him ’. In contrast to the standard accounts, according to which disam biguation was the m ain m otivation for 5e//^adjunction to local pronouns, we argue that .se//'-adj unction was m otivated by a need to m ark representational non-identity o f the antecedent and the locally bound pronoun with anti-reflexive (or “hidden” neutral predicates, e.g. (100)) in exactly the same was as ^e/v-adjunction to simple reflexives in M odern Danish, see chapter 3. The distribution o f Danish pronouns and reflexives (e.g. ham ‘him ’, sig ‘REFL 3 person’, m ig ‘m e’) with these different verb types is illustrated in (I01)-(102). (101) M odern D anish third person singular pronoun ham ‘him ’: a. Peter, mistcenkte ham*,,. (anti-reflexive predicate) ‘ Peter] suspected him*]/z.’ b. Peter, vaskede ham*,;,. (neutral predicate) ‘Peter] washed him.j/^.’ c. Peter, hvilte *ham* ,/* ,. (inherently reflexive predicate) ‘*Peter rested him .’ (102) M odern D anish third person singular reflexive s is ‘REFL’: a. Peter mistcenkte *sig/sig,,*, selv. (anti-reflexive predicate) ‘ Peter] suspected himselfj/.z. ’ b. Peter vaskede sig, *fsig, * , .selv. (neutral predicate) ‘ Peter] washed himself]/*z. ’ c. Peter, hvilte sig,,*f*sig selv. (inherently reflexive predicate) ‘Peter rested *??him self.’ Unlike O ld English, Danish has a distinction between reflexives (e.g. sig ‘REFL’) and pronouns (e.g. ham ‘him ’). As discussed in chapter 3, the distribution o f sig and ham fall under principle A and principle B respectively: sig ‘REFL’ is an anaphor w hich m ust be bound in a local dom ain (viz. the tensed clause), and ham ‘him ’ is a pronoun w hich m ust be free in the same local domain. As (101c) shows, inherently reflexive verbs do not allow pronouns in R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 350 object position. See section 3.3.2.2 in chapter 3 for discussion o f the sem antics o f inherently reflexive predicates and o f the reasons why unintensified reflexives are required in such cases, see (102c). In contrast, Old English allow ed pronouns in such positions, see (95). N ote furtherm ore, that in spite o f the fact that sig ‘REFL’ in (102a) is not am biguous betw een a co-referential and a disjoint reference reading, the presence o f the intensifier selv ‘self’ is still m andatory as a m arker o f presupposed representational non-identity with anti reflexive predicates. N ote also that se/v-intensification o f sig is optional with neutral predicates in (102b) and im possible w ith inherently reflexive predicates in (102c). This distribution o f selv clearly shows that se/v-intensification o f reflexive is not triggered by a need for disam biguation in M odern Danish, but rather by the sem antics o f the predicate: anti reflexive predicates require intensification o f reflexives to mark the reflexive reading (in contexts w hich presuppose representational non-identity, see section 3.3.2.1 in chapter 3) and 5e/v-intensification is excluded with inherently reflexive predicates since in those contexts generation o f a contrast set o f alternatives is blocked by the m eaning o f the predicate w hich presupposes the identity o f its arguments. O ur analysis o f the evolution o f him self m M odern English is based on the assum ption that in Old English ^e/f-intensification functioned like selv- intensification does in M odern Danish, i.e. it was triggered more by a need to license representational non-identity than by a need to disam biguate. The first person exam ples in (103) further corroborate this account o f M odern Danish. (103) M odern Danish first person pronoun/reflexive m is ‘m e’: a. Jeg mistcenker *mig/migi,*^ selv. (anti-reflexive predicate) ‘Ii su sp ect myselfi/*^.’ b. Jeg vasker migi *fmigi selv. (neutral predicate) ‘Ij w ash 0/m yself, c. Jeg hviler mig, *f*mig selv. (inherently reflexive predicate) ‘1 rest (*m yself).’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 351 Danish does not distinguish betw een pronouns and reflexives in the first and second persons. In spite o f this fact, no am biguity is - for obvious pragm atic reasons - possible. But even so, the presence o f selv is still m andatory w ith anti-reflexive verbs, optional w ith neutral verbs, and im possible w ith inherently reflexive verbs, see (103a) and (103c) above. As the exam ples in (103) illustrate the fact that the creation x-5e//form s took place in D’ and persons as well as in the 3 '^ ‘ ‘ person is no m ystery in the account proposed here which, consequently, does not need to recur to vague notions o f ‘analogy’. N ote that the sem antic/pragm atic approach o f intensification proposed here stands a better chance o f explaining w hy 5e//^adjunction to locally bound pronouns did not always result in unam biguously coreferential readings, see (93) repeated in (104). (104) Be dham cwcedh se cedhela lareow sanctus Paulus: Ic wille dhcet ge sien wise to gode & bilwite to yfele. Ond eft he dhcem cwcedh Dryhten dhurh hine selfne to his gecorenum: Beo ge swa ware sua sua ncedran & sua bilwite sua culfran. [CP 35.237.18, translation provided] ‘Therefore the noble teacher St. Paul said: “1 wish ye to be wise for good and simple for evil. A nd again, the Lord spoke about the same thing through himself to his elect: “ Be cunning as adders and simple as pigeons.’” (=Siem und 1999 (2.47)) The exam ples (105)-(106) illustrate the predictions this approach m akes for the interpretation o f locally bound object pronouns with the three m ain types o f predicates, i.e. anti-reflexive predicates (e.g. suspect), see (105a) and (106a), neutral predicates (e.g. wash), see (105b) and (106b), and inherently reflexive predicates (e.g. rest), see (105c) and (106c). N ote that the some o f the O E data in (105) are based on extrapolations rather than attested exam ples. O f the sentences in (105), we only have attested exam ples corroborating the judgm ents in (105c). T h e ju d g m en ts in (1 0 5 a ,b ) are based on the beh avior o f other typical a n ti-reflex iv e/’h id d en ’ neutral predicates (e.g. give and kill in 91-2) and other typical neutral predicates (e.g. defend in (90)). N eedless to say, m ore careful and exhaustive studies o f the behavior o f different verb types in OE is still needed. A t this point the analysis o f OE presented here is thus m ore an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 352 outline o f a new approach than a fully-fledged analysis whose em pirical coverage has been carefully researched. (105) Stage one: unintensified locally bound 3'^ person pronouns in Old English: a. PeteVi suspected h i m , (anti-reflexive)(com pare with (100)) b. Peter, washed him,,^. (neutral) (see (99)) c. PeteVj rested himi (inher. reflex.) (= (9 5 c)) (106) Stage one: unintensified locally bound 1 person pronouns in Old E nglish: a. 7, suspected m e„*i,*,. ( anti-reflexiye)(com pare with (100)) b. P washed me, (neutral) (see also (99)) c. /, rested mei,*^. (inher. reflex.) (see (95c)) As shown in (105a), anti-reflexive predicates are predicted to require ^e/Aadjunction to the simple pronouns in order to m ark the presupposed representational non-identity. This also applies to the “hidden” neutral^*’ predicate acwealde ‘kill’ in (100). Neutral predicates, which evoke no presuppositions concerning the representational identity o f their argum ents, may or may not display self-adjunction, see (105b) and (99). Finally, the since they presuppose identity o f their argum ents, inherently reflexive predicates do not allow non-coreferential DPs in object positions. Therefore, since no focus-based contrast set o f alternatives can possibly be generated, non-coreferential readings o f locally bound object pronouns is not allowed, see (105c). The exam ples in (107)-(108) illustrate how the intensifier-adj unction to simple pronouns function to license representational non-identity, viz. in the cases o f the anti reflexive predicates in (107-8a,d). N ote also that, since intensification requires that the generation o f a contrast o f alternatives to the referent o f the adnom inally intensified nominal expression, intensification o f sim ple pronouns in object position o f inherently reflexive verbs ’ See section 6.3. for a definition of “hidden” neutral predicates. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 353 is under no circum stances allowed, see (107-8c,f). Finally, intensification o f sim ple pronouns with neutral verbs is optional, see (107-8b,e)^’. (107) Stage one: intensified locally bound person pronouns in Old E nglish: Intensified pronouns: Unintensified pronouns: a. Peteri suspected him, ?^ self. d. Peter i suspected him?,.,. b. Peteriwashed him,/?, self. e. Peter, washed him,.,. c. Peter, rested him,,*, *self. f. Peter, rested him, * ,. (108) Stage one: intensified locally bound f* person pronouns in Old E nglish: Intensified pronouns: U nintensified pronouns: a. 7, suspected me,,*, self. d. I, suspected me, * ,. b. I, washed me,.*, self. e. I, washed me, * ,. c. I, rested me, *, *self. f. I, rested me, * ,. (109) Stage one: intensified and unintensified DPs and subject pronouns in Old E nglish: Intensified: Unintensified: a. the king s e lf etc. c. the king b. he s e lf I se lf etc. d. he, I, etc. A t this stage the prim ary function o f 5e/f-adjunction was not to disam biguate third person pronouns but to m ark representational non- identity with in anti-reflexive contexts, cf. the M odern Danish exam ples in (102a) and (103a). N ote also the existence o f exam ples like (8), w hich bear witness to the fact that even intensification o f third person pronouns was not always triggered by a desire to disam biguate. In other words, we propose that the so-called disam biguating function is derivative o f the prim ary intensifying function illustrated by the contrast between (107b) and (107e). N ote that a this stage we assume that there still has been no fusion o f him+self. The fact that se/f-intensification was m otivated by a need to license co reference scenarios which would otherwise be infelicitous w ith anti-reflexive predicates, e.g. (107a,d) and (108a,d), and not by a need to disam biguate third person pronouns, neatly explains why the intensification took place with first and second person pronouns as well as with third person pronouns. That is, question (97a) is no longer a mystery. Indeed, it is now clear that (97a) was the w rong question to ask in the first place. The system o f intensified and As discussed in chapter 3, section 3.3.2.5, intensifier-adj unction to simple reflexives with neutral predicates is never truly optional, but rather dependent on focus and other discourse factors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 354 unintensified nominal expression in Old English (stages one and tw o) is sum m arized in the table in (110). (110) Stage one: system o f intensification o f nom inal expressions in O ld E nglish: U nintensified Intensified a. L og. bound pr. him him s e lf b. Subj.pronoun he he s e lf c. O bj.pronoun him him s e lf d. D P the king the king s e lf A s shown in table 1, Old English had a m orphologically transparent and fully com positional system o f intensification in w hich all kinds o f nom inal expressions (DPs and pronom inals) were intensified through the same process o f 5e//^adjunction. In this respect Old English is sim ilar to M odern Danish, see the table in (111). (111) System o f intensification o f nominal expressions in M odern D anish: U nintensified Intensified a. Reflexive sig ‘him self/herself/etc.’ sig selv ‘him self/herself b. Subj. pronoun han ‘he’ han selv ‘he h im se lf c. Obj. pronoun ham ‘him ’ ham selv ‘him him self/him self’ d. D P kongen ‘the king’ kongen selv ‘the king him self S.3.3.2 Stage two: intensification and binding in Early Modern English The schem as in (112)-(114) sum m arize the changes w hich took place between O ld English (OE) and Early M odern English (EM E). In EM E the locally bound pronouns o f O E were replaced by 0-reflexives w bile locally free pronouns rem ained, see ((112-3), and the OE intensifier s e lf has, been replaced by him self as intensifier across the board, see (114). (112) The evolution o f adnom inal intensifiers: a. (OE) s e lf » b. (E M E ) him self (113) The evolution o f locally bound pronouns: a. (OE) him b. (EM E) 0-reflexive » R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 355 (114) The evolution o f locally free pronouns: a. (OE) him » b. (EM E) him The three changes in (112-4) is ail that is needed to account for how the Early M odern English system o f binding and intensification, see (115), evolved out o f Old English. N otice that the 0 - reflexive analysis proposed here allow us to account for EM E as having the sam e kind o f m orphologically com positional system o f intensification as the ones found in Old English and M odem Danish, see (110-1), repeated as (116-7). (115) Stage two: Svstem o f intensification o f nom inal expressions in Earlv M odern E nglish: U nintensified Intensified a. Reflexive 0 0 himself b. Subj.pronoun he he himself c. Obj .pronoun him him himself d. DP the kin g ’ the king himself (116) Stage one: svstem o f intensification o f nom inal expressions in O ld English: U nintensified Intensified a. Loc. bound pr. him him self b. Subj.pronoun he he self c. O bj.pronoun him him self d. DP the king the king self (117) Svstem o f intensification o f nominal expressions in M odern D anish: U nintensified Intensified a. Reflexive sig ‘him self/herself/etc. ’ sig selv ‘him self/herself’ b. Subj. pronoun han ‘h e’ han selv ‘he him self’ c. Obj. pronoun ham ‘him ’ ham selv ‘him him self/him self’ d. DP kongen ‘the king’ kongen selv ‘the king him self In other words, the 0-reflexive approach allows us to account o f the evolution o f the EM E system in (115) from the O E system in (116) w ithout stipulating any shift from one type o f system to another. Basically the m orphological com positionality o f the system o f intensification rem ains sam e m odulo a change in the form o f the adnom inal intensifier from the sim ple x-self form to the com plex x-self forms^*. Furtherm ore, a new reflexive pronoun. Notice that this change from simple to complex form of the intensifier closely mirrors the (still ongoing) change of the older simple intensifier -m eme into the complex intensifier lui-meme in Modem French, e.g. le roi meme ‘the king self » le roi lui-meme ‘the king him self. Notice also the close parallels in behavior between the clitic- R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 356 the 0 -reflexive was introduced to replace the locally bound pronouns in OE. We argue that these tw o changes are intim ately connected. That is, we claim that the analysis proposed here can explain the link betw een the loss o f locally bound pronouns and the formal change o f the adnom inal intensifier from s e lf to himself, see questions (b-d) in (118). First let us illustrate the changes described in (118-121) with concrete examples'^’. (118) The evolution o f unintensified locally bound pronouns: O ld English Early M odern English a. ??*Peteri suspected himi. » *Peteri suspected 0j. b. Peter, washed him,. » Peter, washed 0,. c. Peter, rested him,. » Peter, rested 0,. d.. Peter, took a knife with him,. » *Peter, took a knife with 0,. (119) The evolution o f intensified locally bound pronouns: O ld English Early M odern English a. Peter, suspected him, self. » Peter, suspected 0 , himself. b. Peter, washed him, self. » Peter, washed 0 , himself. c. *Peter, rested him, self. » *Peter, rested 0 , himself. d.. *Peter, took a knife with him, self. » * Peter, took a knife with 0 , himself. (120) The evolution o f unintensified locally free pronouns: Old English Early M odern English a. Peter, suspected him^. » Peter, suspected him,,. b. Peter, washed him„. » Peter, washed him^. c. *Peter, rested him^. » *Peter, rested him^. d.. *Peter, took a knife with him^. » *Peter, took a knife with him^. (121) The evolution o f unintensified locally free pronouns: O ld English Early M odem English a. ...Peter, [..] except him„ self. » ...Peter, [..] except (him)^ himself. The exam ples in (118-9) illustrate the consequences o f the shift from locally bound pronouns to 0-reflexives. As shown, the distribution o f unintensified and intensified forms is the same in both stages o f the languages, i.e. anti-reflexives (due to the presupposition o f like 0-refiexive in English and the French reflexive clitic se ‘RELF’. Similarly the simple and complex pronouns lui T im ’ and lui-meme ‘him self also closely mirror the behavior of their English counterparts him and locally free/logophoric himself. Interestingly, the shift from locally bound pronouns to 0-reflexives occurred in a time when the French exerted a strong influence on the English language. One could therefore speculate that this shift from locally bound pronouns to 0-reflexives was due to influence from French. As mentioned above, these OE “examples” are based in large part on extrapolations from similar data. More work is needed to find attested examples for all these judgments. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 357 representational non-identity) require intensification, see (118a) and (119a), neutrals are com patible with both intensified and unintensified forms, see (118b) and (119b), and inherently reflexives are only com patible w ith unintensified forms, see (1 18c-d) and (1 19c-d). The only unexpected judgm ent is ungram m aticality o f (118d) w ith sim ple unintensified 0 - reflexive. However, this “exception” can be explained as a consequence o f the clitic-like behavior o f 0-reflexives, i.e. clitics can only occur in direct argum ent-positions (cf. Fr. Pierre estfier de *se/soi ‘Pierre is proud o f REFL’). Since the 0 -clitic cannot is barred in (188d) for m orphological reasons, and since the inherently reflexive sem antics o f the expression take a knife with excludes intensifier-adjunction, the personal pronoun him is the only overt realization offered by the m orphology o f the language. N ote that Danish has sim ple unintensified sig in such expressions, see chapter 3, section 3.3.6. The exam ples in (120-1) illustrate the fate o f locally free pronouns in OE. Unlike locally bound pronouns, locally free pronouns rem ained unchanged in EM E. Likewise, adnom inal intensification o f pronouns took place in the sam e environm ents in OE and EME, i.e. with in contrastive contexts, e.g. Peter [said that M ary danced with everybody] except himself, see (121b). N ote that in EM E locally free intensified pronouns could take the form him himself, her herself us ourselves, etc. w hich is banned for phonological reasons (i.e. possibly haplology) in m ost versions o f M odem English. These forms thus clearly show that the form o f the intensifier had changed from se lf to him self N ote also that the total absence o f exam ples like (122) in the history o f English, is a witness to the fact that the change locally bound him » 0-reflexive took place sim ultaneously with (or ju st prior to but not after) the change s e lf» him self (122) *Peteri washes/suspects himi him self R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 358 In other words, the evolution o f OE into EM E m ust have taken place as follows. First the simple intensifier began to be adjoined to locally bound pronouns in OE to m ark representational non-identity in anti-reflexive contexts. Since the m ajority o f verbs are either anti-reflexive or “hidden” neutral predicates this happened in a large num ber o f contexts. Hence the com bination him +self came to be associated with reflexive readings and, as a consequence, the unintensified pronouns, e.g. him, etc., came to be associated with non reflexive, disjoint reference readings. A s a certain point the association between unintensified pronouns and disjoint reference readings becam e so strong that the deletion o f locally bound pronouns was enforced, e.g. pronouns could no longer be used with inherently reflexive predicates. Thus the 0-reflexives were created. And at the same tim e the com bination him +self vtas reinterpreted as an adnom inal intensifier, see (123). As m entioned above, the em ergence o f him self as intensifier is corroborated by the sim ultaneous em ergence o f intensified (locally free) pronouns o f the form him himself, and intensified DPs o f the form the king him self (123) The creation o f com pound ‘reflexives Vintensifiers: him + self m e+ self etc. => him self m yself^, etc. (=’reflexivesVintensifier w hich could only be used in contrastive contexts) (123) illustrates the fusion o f him + self me+self, etc. into the so-called com pound intensifiers him self myself, etc., which took place fusion betw een OE and EM E. The com pound x-self forms w hich arose through this fusion process have a very peculiar and som ew hat surprising property: they are never found in inherently reflexive contexts (which is exactly the kind o f context w here one would expect to find reflexives rather than pronouns, see Danish exam ples in (125)), see (124). (124) Peter took a knife with him/*himself As mentioned above, we do not take the case complications into account here. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 359 (125) Peter tog en kniv med sig/*sig selv. Peter took a knife with R EFL/REFL self ‘Peter took a knife w ith him /*him self.’ The fact that the x-self forms are found m ainly in contrastive situations confirm s our analysis o f them as adnom inal intensifiers. That is, they have the sam e m eaning contribution and (to a large extent) the same distribution as the sim ple intensifier selv ‘s e lf in Danish and the simple intensifier xe//'in O ld English, see (125-7). The assum ption that English developed 0-reflexives is also necessary to explain why sentences like (126a) are bad, while its OE and M odem Danish/Dutch counterparts in (126b,c) are ok. (126) a. I shave *me. b. Jeg barberer mig. (Danish) ‘I shave m e.’ c. Ich scheere mich. (D utch) T shave m e.’ The exam ples in (127-132) illustrate that correspondence betw een English 0-reflexives and simple unintensified reflexives in Danish, colum n (hi) in (127-132). N ote that while both D anish and English distinguish betw een 3 '^'’ person pronouns and reflexives, com pare colum ns (i) and (hi) in (127-129), only English distinguish form ally between f ' (and 2"‘ * ) person pronouns and reflexives, i.e. me vs. 0-reflexive, see colum n (i) and (hi) in (130-132) . As shown in colum ns (i) and (hi) in (130-132) D anish f ‘ and (2"‘ * ) person pronouns and reflexives share the sam e formal expression, e.g. mig ‘m e’ in colum ns (i) and (hi) in (130-132). (127) A nti-reflexive verbs: (i) pron.: (ii) 0 -p ro n . (iii) unint. refl. (iv) int. refl. c. Ij suspect *me, / *0^ / *0, / 0 , myself. d. Jeg, mistoenker (mig,) / *0^ / *mig, / mig, selv. (128) N eutral verbs: a. 1 , wash *me, / 0^ / 0 , / 0 , myself. b Jeg, vaster (mig) / 0^ / mig, / mig, selv. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 360 (129) Inherently reflexive verbs: a. b. I, rest Jeg, hviler *me, / (mig,) / 0, mig, *0 , myself. *mig, selv. (130) A nti-reflexive verbs: (i) pron.: e. He, suspects *him, / f. Han, mistcenker *ham,/ (131) N eutral verbs: a. He, washes *himi / b Han, vasker *ham, / (132) Inherently reflexive verbs: a. He, rests *him, / b. Han, hviler *ham, / (ii) 0 -p ro n . (iii) unint. refl. (ivl int. refl. 0 z 0 . * 0 z *0, *sig, 0, sigi 0, sig, 0 , himself, sig, selv. 0 , himself, sig, selv. *0, himself. *sig, selv. A ssum ing the English system o f intensified nom inal expressions to follow the Danish pattern illustrated in (127-132), yields the picture o f Early M odern English given in (133)-(135). Just like the simple Danish reflexive sig is intensified by intensifier adjunction (e.g. sig selv), so the simple English reflexive 0 is intensified by adjunction o f the appropriate form o f the com plex intensifier him self (e.g. 0 himself). (133) Stage tw o (third personl: Intensified reflexives: a. Peter, suspects 0 , * ^ himself. b. Peter, washes 0,,*^ himself. c. Peter, rests 0,/*^ *himself. (134) Stage tw o (first personl: Intensified reflexives: a. b. c. I, suspect 0,/*2 myself. I, washe 0 , myself. I, rest 0 , *myself. (135) Stage two (intensified DPs, subiect pronouns'): Intensified expressions: a. the king him self c. b. he him self I m yself etc. d. U nintensified reflexives: d. Peter, suspects 0 ., e. Peter, washes 0 , f. Peter, rests 0 , Unintensified reflexives: d. /, suspect 0 ., e. I, wash 0 , f. I, rest 0 , * ^ . U nintensified expressions: the king he, I, etc. N ote that the neat com positional character o f the system o f intensified and unintensified nom inal expressions o f EM E illustrated in (115) is perfectly paralleled in the m odern Danish R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 361 system in (117). That is, in both system s the diflference betw een intensified and unintensified forms is signaled by the presence/absence o f the adnom inal intensifier selv/himself. EM E differs from M odern English in that it allows for intensified object pronouns, e.g. him him self {= Dan. ham selv ‘him him self’). That these intensified pronouns are pronominal expressions is dem onstrated by the fact that they obey principle B, i.e. they cannot be locally bound, see (136)-(139), in this sense they behave like their Danish counterparts in (138)-(139). (136) Stage tw o (third person pronouns)'” Intensified pronouns: Unintensified pronouns: a. PeteVj suspects him*i,z himself. d. Peteri suspects him*, z. b. Peter, washes him*,z him self e. Peter, washes him*,z- c. Peter, rests *him,/*z*himself f. Peter, rests *him*, * z . (137) Stage tw o (first person pronouns) Intensified pronouns: Unintensified pronouns: a. He, suspects me*,/z m yself d. He, suspects me*,z- b. He, washes me*,z m yself e. He, washes me*,z- c. He, rests *me*, * z *myself f. Peter, rests *me*, * z . N ote that, unlike Old English, EME does not allow sim ple unintensified pronouns in object position o f inherently reflexive verbs, see (136c,f) and (137c,0- We now see that question (97c) (i.e. ‘W hy were the pronouns in the usages in (95) abandoned and, once they were lost, why English did not start anew to develop pronouns w ith these functions?’) asks the w rong question. English did develop a new kind o f pronom inal elem ent to replace the simple pronouns in object position o f inherently reflexive verbs, viz. the simple unintensified 0 - reflexive w hich thus can be seen as the English counterpart o f the sim ple unintensified reflexive sig in Danish, see tables 3-4. The exam ples in (138)-(139) illustrate the distribution o f intensified and unintensified form s o f the 1 ® ' and 3 '^ '* person pronouns in M odern Danish. A s described above, close com parison o f the English stage four pronom inal system in and the D anish system in (127)-(132) reveals one interesting difference. W hile English chose The judgments in (56-59) are based on the assumption that these sentences are placed in the appropriate ‘iogophoric’ contexts where the intensified pronouns are bound by a non-iocai antecedent referring to a subject of consciousness (cf. Zribi-Hertz (1989), Kuno (1987), Huang & Liu (2001), etc.). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 362 to replace all (i.e. T ‘, and 2"“ * person form s as well as 3 " * * persons form s) locally bound sim ple pronouns with the zero reflexive, D anish chose (at som e point in history) to live w ith a system in w hich the reflexive and pronom inal form s in the T ' and 2 ' " ' persons are form ally identical, com pare the ungram m atical sentence / wa^/z me in (128a) versus the gram m atical sentence je g vasker m ig ‘(lit-) I wash m e’ in (128b). N otice also that the same form mig ‘m e’ o f the personal pronoun is used in the non-reflexive sentence in (139e). Furtherm ore notice that M odern Danish does not have zero reflexives, com pare (128a) and (128b). This fact strongly supports the hypothesis that English zero reflexives correspond to Danish sim ple unintensified reflexives (i.e. Eng. 0 reflexive = Dan. sig ‘REFL’, mig dig ‘you'^® *’'^ ’ ''™’). See also chapter 4, section 4.4.2 for a late-insertion analysis o f T ' and 2"‘ ’ person form s in Danish. (138) M odem D anish third person pronouns: Intensified pronouns: Unintensified pronouns: a. Peteri mistcenker ham ^ selv. d. Peter, mistcenker ham*, ,. ‘Petep suspects him*i/^ h im se lf’ ‘Petep suspects him*i/-^.' b. Peteri vasker ham*i,zSelv. e. Peter, vasker ham*, ^. ‘Petep washes him-i/^ him self.’ ‘Peter, washes him*n^.' c. Peter hviler *ham selv. f. Peter hviler *ham. ‘Petep rests /n'm j/.z*him self’ ‘Petep rests 0i/*z-’ M odern Danish first person pronouns: Intensified pronouns: U nintensified pronouns: a. Ptan, mistcenker mign^ selv. d. Han, mistcenker mig*,,. ‘Hc; suspects we*i/z m yself.’ ‘HCi suspects me*,/^.’ b. Hani vasker mig*,^ selv. e. Han, vasker mig*,.^. ‘He, washes me»,/z m y se lf’ ‘ Hci washes mety^,. ’ c. Hanj hviler *mig selv. f. Han, hviler *mig. ‘Hei rests *w7c.i/.z*myself’ ‘Petep rests ^me*y*^.'’ N ote that the change to 0-reflexives is necessary to explain the fact that the sentence I shave me (cf. Ger. Ich scheere mich, Dan. Jeg barberer mig, in (126)) was ok in OE but has b eco m e im possible in M odern English. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 6 3 As illustrated by (136)-(137) and (115), stage tw o (EM E) is represented by O lder M odern English (e.g. Jane A usten’s English''^) w hich does allow intensified object pronouns (e.g. him himself). Later, in stage five, a m orphological redundancy rule elim inated such forms. 5.3.3.3 Stage three: intensification and binding in Modern English Stage five is characterized by the introduction o f a m orpho-(phono-)logical rule'*^ banning repetition o f person, num ber and gender features in intensified object pronouns; him him self => [him] him self me m yself=> [me] myself, etc., see the table in (140). (140) System o f intensification o f nom inal expressions in stage five (=M odem English). Unintensified Intensified a. Reflexive 0 0 h im self b. Subj. pronoun he he h im self c. Obj. pronoun him [him] h im self (<him him selff^ d. D P the kin g ’ the king h im self 5.3.3.4 Additional evidence from the evolution of pronominal reciprocals into 0 - reciprocals The observation that in many languages (e.g. Rom ance languages and G erm an) the simple unintensified reflexive (se/si/sich) can be used as both reciprocal and reflexive is highly relevant to our analysis o f the historical developm ent o f 5e//^intensifiers and 0-reflexives in M odern English. In the same way O E locally bound pronouns w ere replaced by 0-reflexives in EM E so were also locally bound pronouns with reciprocal readings, see (141a), replaced with 0-reciprocals, see (14Ib). See Baker (1995) for a detailed description of Jane Austen’s English. Cf. Siemund’s (1999:81) rule: “If two expressions El and E2 form a complex expression Ec, and if the semantic features of El are a subset o f those of E2, then E l is superfluous and can be dispensed with.” The fact that the form him him self \s extremely rare (but not unattested) in present-day Modern English indicates that the change from Early Modern English to Modern English took place fairly recently and that for some speakers (esp. of conservative versions of British English) this changes has yet to be fully implemented. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 364 (141) a. {OE) They, kissed them,. » b. {EME) They, kissed 0,. ‘They kissed 0 /each other.’ ‘They kissed (each other).’ (142) (OE) pronom inal reciprocals » (EM E) 0-reciprocals This kind o f evidence illustrates the close affinity betw een reflexives and reciprocals and provides additional support for the analysis o f the historical evolution o f 0 - reflexives/reciprocals proposed above. 5.3.4 Summary The above discussion o f the developm ent o f M odern English (M E) him self has, shown that the assum ptions that: (i) M E has 0-reflexives w hich are functionally equivalent o f Dan. .sig, Fr. se, Ger. sich, etc. (ii) ME x-self forms o f the type him self are always intensifiers. (iii) predicate m eaning (presupposed representational non-identity) affects the distribution o f intensified and unintensified reflexives (e.g. Dan sig vs. sig selv, and Eng. 0 vs. 0 x-self). The above approach to the evolution o f x-self form s and 0-reflexives offer interesting new answers to all o f the hitherto unansw ered questions listed in (97) and repeated in (143), concerning the evolution o f the com pound intensifiers/reflexives him self m M odem English. (143) Q uestions concerning the developm ent o f the x -se/f intensifier in M odern E nglish: (a) W hy was the fusion o f pronoun and s e lf not confined to the third person, but was extended to the first and second persons? This rem arkable in so far as no am biguity arises in these cases. (b) W hy did the new ly developed com pound form also come to be used as an intensifier? (c) W hy did English abandon its original intensifier self! Old English se lf was used in much the sam e way as its m odern equivalent and there seems to have been no obvious reason to replace it w ith a new intensifier. (d) W hy were the pronouns in the usages in (95) abandoned and, once they were lost, why English did not start anew to develop pronouns with these functions? As discussed above, question (143a) is asked in the w rong way to begin with. Since the adjunction o f s e lf to pronouns in O ld English was not prim arily m otivated by a need for disam biguation (but rather by a need to m ark representational non-identity in anti-reflexive, R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 365 see (91-2), or inherently contrastive contexts, see (93), it is not at all surprising but, indeed, expected that the fusion o f pronoun and se lf should be extended to the first and second persons. Indeed, as shown be the D anish exam ples discussed above, even in the absence o f am biguity (between coreference and disjoint reference readings) se/v-adjunction to pronouns and reflexives is still necessary to m ark representational non-identity with anti-reflexive and ‘hidden neutral’ predicates. There is therefore nothing rem arkable in the fact that the fusion o f se lf and pronoun was took place across the board regardless disam biguation issues. As for questions (143b) and (143c) the answ er lies in the 0-reflexive approach. As argued in section 5.3.3.2, the fact that the com bination him +self cam e to be associated with reflexive readings lead to the unintensified pronouns being associated with non-reflexive, disjoint reference readings. A t a certain point the association between unintensified pronouns and disjoint reference readings becam e so strong that the deletion o f locally bound pronouns was enforced, e.g. pronouns could no longer be used w ith inherently reflexive predicates. Thus the 0-reflexive were created. And at the same tim e, the com bination him + self was reinterpreted as an adnom inal intensifier. The tables in (144-147), neatly sum m arizes the account o f historical developm ent o f reflexives and intensifiers in M odem English and its parallels to the system o f intensification o f nominal expression in M odern Danish. (144) Stage one: system o f intensification o f nom inal expressions in Old English. U nintensified Intensified a. Loc. bound pr. him him self b. Subj.pronoun he he self c. O bj.pronoun him him s e lf d. DP the king the king self R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 366 (145) Stage two: system o f intensification o f nom inal expressions in Early M odern English. U nintensified Intensified a. Reflexive 0 0 himself b. Subj.pronoun he he himself c. O bj.pronoun him him himself d. DP the kin g ’ the king himself (146) Stage three: system o f intensification o f nom inal expressions in M odem English. U nintensified Intensified a. Reflexive 0 0 himself b. Subj.pronoun he he himself c. Obj .pronoun him (him) himself (<him himself) d. DP the kin g ’ the king himself (147) System o f intensification o f nom inal expressions in M odern Danish. Unintensified Intensified a. R eflexive sig ‘him self/herself/etc.’ sig selv ‘him self/herself b. Subj. pronoun han ‘he’ han selv ‘he him self’ c. Obj. pronoun ham ‘him ’ ham selv ‘him him self/him self’ d. DP kongen ‘the king’ kongen selv ‘the king him self Finally, the question in (143d) receives a straightforw ard explanation if we assum e that the pronouns in object position o f inherently reflexive verbs were replaced by 0-reflexives. Further evidence for the existence o f 0-reflexives is provided by the 0 /h im self alternations found w ith neutral predicates, see (68). (148) N eutral verbs: a. I wash 0 / 0 myself. b. Jeg vasker m ig/m ig selv. Rather than being optional (as is often claim ed in the literature), the alternation betw een sim ple unintensified reflexive and com plex intensified reflexive is m otivated by the same sem antic/pragm atic factors (i.e. contrastive focus, doppelganger-effects, anti-refiexivity, etc.) in both English and D anish (i.e. Eng. 0 vs. 0 m yself = Dan. mig ‘m e’ vs. mig .selv ‘me se lf’). Only the assum ption that English have 0 reflexives allows for the unification o f the analysis reflexives and intensifiers in English and Danish. Indeed, this assum ption is, as we have argued above, the very key which unlocks that m ysteries surrounding the evolution o f the com pound form s o f the intensifier him self \n M odern English. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 367 5.4. Conclusion W hile potentially raising more questions than it answers, we hope that the analysis o f English intensifiers and reflexives outlined above has convincingly shown that the analysis o f Danish based on the independence o f intensification and binding presented in chapters 1-4 can be extended to both M odern and O ld English. The articulation o f binding and intensification proposed here has far reaching consequences for our understanding o f both modules. First, in contrast to both traditional gram m ars and m odern linguistic accounts o f binding, English x-self forms are not am biguous betw een intensifiers (e.g. Peter himself) and reflexives {Peter succeeded himself), but rather ALWAYS function as intensifiers. In this respect, Eng. \-s e lf form s have m ore in com m on w ith the D anish intensifier selv than w ith the sim ple reflexive sig, see (149). (149) Typological survey o f different uses o f adnom inal intensifiers: English Danish French German a. A dnom . inten. the king himself kongen selv le roi (lui-)meme der Konig selbst b. C om plex refl. 0 himself sig selv soi-meme sich selbst d. D oppelganger-m ark 0 himself sig selv soi-meme sich selbst e. inten. pron./logophor [him] himself ham selv [lui] lui-meme ihn selbst f. Exclus.adv. int. himself selv lui-meme selbst g. In d u s, adv. int. himself selv lui-meme selbst Furtherm ore, the existence o f 0-reflexives in M odem English finds support in the fact that they have largely the sam e distribution as sim ple unintensified reflexives in Danish, see (150). (150) Typological survey o f reflexive and non-reflexive uses o f reflexive elem ents: English Danish French German a. reflexive 0 sig se/soi sich b. reciprocal 0 -s se sich c. m iddle 0 sig se sich e. inh.relf/depon. 0 -s/sig se sich One o f the advantages o f the 0-reflexive analysis o f English is that locally free reflexives no longer constitute potential exceptions to principle A. That is, the distribution o f anaphors (0 - reflexives and reciprocals, both 0-reciprocals and each other) is now lim ited to the local dom ain (principle A), w hile pronom inals have to be free in the sam e domain. Finally, as R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 368 discussion in section 5.3, the 0-reflexive analysis o f M odem English allows for a m ore straightforw ard analysis o f the evolution o f reflexives and intensifiers. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 369 Chapter 6 Independence of binding and intensification in Mandarin Chinese 6.1 Introduction In this chapter we discuss to w hat extent the analysis based on the independence o f intensification and binding presented in chapters 2-5 can be applied to reflexives and intensifiers in M andarin Chinese. Applied to Chinese this approach gives rise to the hypotheses in (1). (1) a. Chinese ziji [ = } 3 ‘se lf-se lf' is always an intensifier, never a reflexive. b. Chinese has 0-reflexives w hich can be adnom inally intensified by adjunction o f ziji. W hat appears to be reflexive uses o f ziji are really adnom inally intensified 0-reflexives, e.g. Zhangsan hen 0 ziji ‘Zhangsan hates 0 h im self. A s will be show n below, the tw o hypotheses in (1) yields a radically new approach to the analysis o f reflexives and intensifiers in M andarin Chinese. 6.2 Setting the stage: evidence supporting the proposals that ziji is always an intensifier and that Chinese has 0-refiexives In this section the main properties o f the elem ent ziji will be introduced in order to set the stage for the follow ing sections. The m ain focus will be on showing that there is a large body o f data supporting the proposals in (1). The Chinese intensifier elem ent ziji has a great deal in com m on w ith the intensifier elem ent him self m M odem English and it will be shown that these sim ilarities can be used to argue that the 0-reflexive analysis o f M odern English proposed in chapter 5 can be extended to Chinese. First, like the English adnominal intensifier him self the Chinese adnom inal intensifier ziji also appears to be able to function as a reflexive anaphor, see (2)-(3). ' In the linguistic literature on Chinese g H ziji is usually glossed as ‘self. In section 6.3.1 we argue that ‘self- self is a better gloss since it is a bi-morphemic word composed of the morphemes § zi and t3 j i which both mean ‘self. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 370 (2) [ii S iR T ° {ziji = adnom inal intensifier) Huang-di ziji lai-le. Em peror self-self com e-PERF ‘The em peror him self cam e.’ {himself = adnom inal intensifier) (3) (zi}7 = reflexive) Huang-di hen ziji. Em peror hate self-self ‘The em peror hates him self.' {himself = reflexive) From the beginning o f m odern form al syntax in the sixties and seventies till today, binding-theoretical accounts o f h im se lf and ziji have tended (W ith a few exceptions, e.g. M oyne (1974), am ong others) to treat these elem ents purely as reflexive anaphora (3) w ithout taking into consideration the fact that they can also be used as intensifiers as illustrated in (2). W e argue that this approach is too narrow in scope and that any viable theoretical account o f these elem ents has to take their use as intensifiers into account. Indeed, in m any cases the behavior o f ziji and h im se lf m w hat seem s to be reflexive uses, e.g. (3), follow from the fact that they are first and forem ost adnom inal intensifiers. O ne o f the peculiarities o f intensifier-based reflexives like C hinese ziji and English h im se lf \s that they do not have any o f the non-reflexive uses^ (e.g. reciprocal, m iddle/unaccusative verbs, m edio-passives, deponent verbs, etc.) displayed by sim ple reflexive elem ents^ in m any other languages (e.g. D anish sig, and G erm an sich, French se, etc.), see (4). ^ Needless to say, the Iogophoric uses o f ziji and him self (also referred to as “locally free reflexives” or LFRs, cf. Baker (1995)) are also in a strict sense non-reflexive. However, they are still in argument position, i.e. thematic. As was argued in Chapter 5 such non-locally bound instances of him self are best analyzed as reduced forms of underlying intensified pronouns him self < him him self (cf. Siemund 2002:81). In section 6.2.2 a similar approach of Iogophoric uses of ziji and to ziji will be discussed. ^ The term “reflexive element” is used here to refer to different forms of reflexive morphology ranging from affixes (e.g, Dan. -s, Rus. -sja), clitics (e.g. Fr. se), syntactically free but phonologically dependent clitics (e.g. Dan. sig), syntactically and phonologically free (i.e. independently stressable) forms (e.g. Fr. soi, Ger. sich, Rus. sebja). See also chapter 4, section 4.5 for more discussion of the different uses of reflexive elements in Danish. The passive suffix -s is derived from an encliticized reflexive pronoun in older stages of the language, see chapter 4 sections 4.5.2, 4.5.4 and 4.5.5 for more discussion of this suffix. R e p ro d u c e d with p e rm ission of th e copyright ow ner. F u rth er reproduction prohibited w ithout perm ission. 371 (4)^ Typological survey o f reflexive and non-reflexive uses o f reflexive elements: (i) Chinese (ii) English (iii) French (iv) G erm an (v) Danish a. reflexive ziji him self se/soi sich sig b. reciprocal - - se sich -s c. middle - - se sich - d. passive - - se sich -s e. deponent verbs - - se sich -s/sig f. intensifier ziji him self (lui-)meme selbst selv N otice that neither English nor Chinese have any non-reflexive uses o f their so-called reflexives, viz. him self m d ziji, see (4b-e). W hile French se, Danish sig and Germ an sich all classify as true reflexives'", him self does not since it is not featureless. Chinese ziji, however, is (like the Danish intensifier selv) unm arked for person, num ber, and gender, and could thus be a featureless “true” reflexive. In spite o f this, it still behaves ju st like English him self In that it cannot be used in any o f the non-reflexive constructions in (4). Furtherm ore, both English him self and Chinese ziji can function as intensifiers, som ething the reflexives in Danish, French, and G erm an are unable to do, see (4f). So except for the reflexive uses, Chinese ziji ‘self-self and English him self are basically in com plem entary distribution w ith the Danish reflexive sig (as well as with Ger. sich, Fr. se, etc.). O n the other hand, as discussed in Chapter 2, Chinese ziji and English /i/w se/f display m ost o f the m ain uses o f the Danish intensifier selv, see (5). ^ The table in (4) is from chapter 4, section 4.5 where example sentences illustrating the different use of reflexive elements in non-reflexive constructions (e.g. reciprocal, middle, passive, etc.) can be found. The term “true reflexive’’ is from Burzio (1991), see also chapter 3. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 372 (5)^ Typological survey o f different uses o f adnom inal intensifiers: C hinese English French D anish a. A dnom inal intensifier guo-wang ziji the king himself le roi (lui-)meme kongen selv b. C om plex reflexive 0 ziji 0 himself soi-meme/lui-meme sig selv c. D oppelganger-m arker 0 ziji 0 himself soi-meme sig selv d. Intensified pronoun [ta] ziji / ta ziji [him] himself [lui] lui-meme ham selv e. E xclusive adv. inten. ziji himself lui-meme selv f. Deverbal com pound *zi]i/zi-/-ji * himself/self *meme-/auto-/sui-/etc.. selv We take the com plem entary distribution o f ziji and him self vs. sig in (4-5) as strong support o f the assum ption that both ziji and him self are intensifiers rather than reflexive anaphors. The apparent reflexive uses o f ziji and him self in (3) and (4a) can thus be explained if we assum e that C hinese and English both have 0-reflexives* w hich can be intensified by the adnominal intensifiers ziji and him self respectively, e.g. 0 him self and 0 ziji in (5b) and (6a). Com bined, (4) and (5) thus draw a picture o f Chinese and M odem English in w hich ziji and him self axe consistently analyzed as intensifiers w hich may adnom inally intensify different types o f nom inal expressions, i.e. 0-reflexives, pronouns, and DPs. In other words, the absence o f non-them atic uses o f ziji and him self {Ah-o) plus their consistent behavior as intensifiers (5a-e) provide strong support for the analysis o f intensified and unintensified nom inal expressions in Chinese outlined in (6) w here ziji is consistently analyzed as an intensifier (like selv in Danish and him self m English) while 0-reflexives correspond to the simple reflexive sig in Danish. (6) System s o f intensified and unintensified nom inal expressions in Chinese, Danish and English. Sim ple/unintensified Com plex/intensified a. Reflexives (i) Chinese 0 0 ziji (ii) D anish sig sig selv (iii) English 0 0 himself b. Obj. Pron. (i) Chinese ta (ta) ziji (ii) Danish ham ham selv 9 (iii) English him [him ] h im self ^ The table in (4) is repeated from chapter 2 where example sentences illustrating the different uses o f intensifiers can be found. * See chapter 5 for more discussion of 0-reflexives in English. ® The material enclosed in square brackets is assumed to have been rendered phonologically zero by a morphological rule, see discussion of similar rule in English in chapter 5. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 373 c. Subj. Pron. (i) Chinese ta (II) D anish han (III) English he d. DPs (I) Chinese huang-di (II) Danish kejseren (III) English the emperor ta ziji han selv he himself huang-di ziji kejseren selv the emperor himself M ost Plca-style LF m ovem ent approaches to LD -anaphora tend assum e that Chinese ziji and D anish sig share som e properties since they can both be LD-bound. M ost Im portantly they are both assum ed to be heads since only heads are able to undergo the successive cyclic m ovem ent w hich is necessary in order to be LD-blndable. Usually the com parison o f properties o f sig and ziji ends there. It Is, however, Im portant not to overlook that, as shown above, sig and ziji are different In alm ost all other respects and that ziji has far more properties in com m on with the English intensifier him self (and D anish selv ‘s e lf) than with the sim ple reflexive sig in Danish. H im self and ziji shares at least five properties w hich D anish sig does not have, see (7). (7) Com m on properties o f ziji and himself. B oth ziji and him self can: (I) be focused (e.g. In answers to questions, as associates o f focus particles, etc.) (II) receive both sloppy and strict Interpretations In sentences w ith verb ellipsis (III) receive both near-reflexive vs. pure-reflexive readings (Iv) be used as adnom inal intensifier o f DPs (v) be used as exclusive adverbial intensifier. N ote that the sim ple reflexive sig In D anish has none o f the properties In (7). In contrast, the Intensified reflexive sig selv ‘REEL s e lf has (7i-lil) and the adnominal selv ‘s e lf on its own has the properties (7iv-v). In other words, both ziji and him self behave m ore like intensified reflexives and Intensifiers than as unintensified reflexives. As the sentences In (8-12) show, assum ing the existence o f 0-reflexlves (which may or may not be intensified) m akes it possible to consistently analyze him self and ziji as adnom inal intensifiers on a par with Danish selv ‘s e lf ’ The sentences in (8-12) Illustrate the properties In (71-v). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 374 (8) A bsence o f focused unintensified 0-reflexives in Chinese (7i'): Q : Zhangsan tui-jian le ni-de di-di ma Zhangsan recom m end PERF your brother ? ‘Has Zhangsan recom m ended your brother?’ A: Mei-you, ta jzh i tui-jian-le 0 izijH *0i not-have he only recom m ended PERF self-self ‘N o, hej only recom m ended 0 ; h im self/* 0 i (9)'° Ziji can have both sloppy and strict readings in elliptical constructions (7 iit: Zhangsan bi Lisi wei ziji bianhu de hao. Zhangsan than Lisi for self-self defend DE good (i) ‘Zhangsan defended him self better than Lisi defended him self.’ (=sloppy) (ii) ‘Zhangsan defended him self better than Lisi defended him .’ (=strict) (10)” Ziji have both <statue> and <real> ti.e.both pure- and near-reflexivet readings (7iiit: Mao Ze-dongj ba zijij qiang-hi le. M ao Ze-dong BA self-self shoot PERF (i) ‘M ao Ze-dong<real> shot him self<statue>.’ (=M ao vandalized his statue) (ii) ‘M ao Ze-dong<real> shot him self<real>.’ (=M ao com m itted suicide) (11) Ziji can function as adnom inal intensifier (7iv'): 7 :? ^ g ° {ziji = adnominal intensifier) Huang-di ziji lai-le. yellow -em peror self-self com e-PERF ‘The em peror him self cam e.’ {himself = adnom inal intensifier) (12) Ziji as exclusive adverbial intensifier t7 v l: Zhangsan bu ziji zhuo gong-ke Zhangsan not self-self do hom ework ‘Zhangsan doesn’t do the hom ew ork h im s e lf’ (i.e. som ebody else helps him) T h e D an ish ex a m p le s in (1 3 -1 7 ) illu strate th a t in all th ese resp ects ziji h as m o re in co m m o n w ith th e D an ish in ten sified re fle x iv e sig selv (see (1 3 -1 5 )) an d th e ad n o m in al This example is adapted from Lidz (2001a:239, (25)). '' This example is adapted from Lidz (2001a:239, (24)). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 375 an d a d v e rb ia l in te n sifie rs selv ‘s e l f , see (1 6 ) an d (1 7 ), th an w ith th e u n in ten sified re fle x iv e sig. (13) S is selv can be focused, sis cannot (IW. Q: D id Peter w ash your brother? A: Nej, han vaskede *sig / sig selv. N o, he washed REFL / REFL self ‘No, he w ashed h im se lf’ (1 4 ) S is selv has both sloppy and strict readings, sis only has sloppy readings (7ii): a. Peter, forsvarede sig. Og det gjorde Hans,, ogsa. Peter defended REFL and it did Hans too ‘Peter defended him self and so did H ans.’ (i) HanSk defended h im self. (sloppy) *(ii) HanSk defended himi. (strict) b. Peter forsvarede sig selv. Og det g a rd e Hans ogsa. Peter defended REFL self and it did Hans too ‘Peter defended him self and so did H ans.’ (i) HanSk defended himselfk. (sloppy) (ii) HanSk defended himi. (strict) (15) S is selv has both near- and pure-reflexive readings, sis only has pure-reflexive readings (7iii): a. Peter barherede sig. Peter shaved REFL ‘Peter shaved 0 . ’ (i) ‘Peter<real> shaved 0 < real> .’ *(ii) ‘Peter<real> shaved 0 < statu e> .’ b. Peter barherede sig selv. Peter shaved REFL self ‘Peter shaved h im se lf’ (i) ‘Peter<real> shaved him self<real>.’ (ii) ‘Peter<real> shaved him self<statue>.’ (16) Selv ‘s e lf can be used as adnom inal intensifier. s is ‘REFL’ cannot (7 iv t: Peter selv/ *sig kom til modet. Peter self REFL cam e to m eeting-the ‘Peter him self attended the m eeting.’ (17) Selv ‘s e lf can be used as exclusive adverbial intensifier. s is ‘R EFL’ cannot (7v): Peter loste opgaven selv / *sig. Peter solved task-the self ‘Peter solved the problem h im se lf’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 376 We conclude that ziji should be treated as an intensifier (or intensified 0-reflexive) since it has five properties in com m on w ith the intensifier selv and the intensified reflexive sig selv ‘REFL s e lf but only one property in com m on w ith sig (i.e. LD-binding). The analysis o f Chinese outlined in (6), w hich yields a m orphologically transparent system o f adnom inal intensification on a par with those proposed for Danish and English in chapters 3 and 5 respectively, is based on the hypotheses listed in (18). (18) a. Chinese ziji is not a reflexive anaphor but rather an adnom inal intensifier (e.g. huang-di ziji ‘the em peror him self). b. Chinese has 0-reflexives, e.g. Peteri xi-le 0,- ‘Peter w ashed 0 j , see (19)'^. c. W hat looks like locally bound reflexives, e.g. Peter, hen ziji, ‘Peter hates h im se lf, is really locally bound adnom inally intensified 0-reflexives, e.g. Peter, hen 0 , ziji ‘Peter hates 0 i h im se lf, see (20). d. W hat looks like locally free reflexives (also called “logophors”) is really adnom inally intensified pronom inals, e.g.: Peter M ary ° Peter, shuo Mary gen chu-le [ ta j ziji yi-wai suo-you de ren tiao-wu P say M with except he self-self INST-outside that-have DE people dance ‘ Peteri said that M ary danced w ith everyone except [him;] h im self.’ The sentences in (19-24) illustrate the elem ents predicted to exist by the analysis sketched in (6) and (18): (19) testifies to the existence o f 0-reflexives (or at least phonetically null argum ents w hich m ay receive a reflexive interpretation), (20) illustrates w hat is m eant by intensified 0-reflexives, (21)-(22) illustrate unintensified and intensified object pronouns, and (23)-(24) illustrate unintensified and intensified subject pronouns. Adnom inally intensified DPs have already been illustrated in (2). '^Note that the absence o f overtly realized objects in sentences o f this type does not have to be construed as evidence that Chinese has 0-reflexives. Mandarin Chinese is known for being able to leave out arguments which can be recovered from information in the surrounding discourse or pragmatic context. What the sentence in (19) does show, is that phonologically unrealized objects are possible in Chinese and that it is at least possible to analyze them as 0-reflexives. Needless to say, more work is needed to determine the exact nature of such empty categories in Chinese. One possibility is that the 0-reflexive is identical to little pro, which has to be identified by the closest c-commanding antecedent (or bound by a null operator). More work is needed before any definite conclusion can be made. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 377 (19) U nintensified i0-reflexives in C hinese: Q : N i you-mei-you xi-zao? you have not-have have wash-bath ‘Have you bathed’ A: W , ■ > You, wOj yi-jing xi 0,-/e. have 1 already wash PERF ‘1 have, 1 have already w ashed 0 . ’ (20)'^ A dnom inailv intensified 0-reflexives in Chinese Q: Ni you-mei-you tui-jian ni-de tong-xue? you have-not-have recom m ended your classm ates ‘Have you recom m ended your classm ates?’ A: Mei-you, w o/zhi tui-jian-le 0iZiJi. not-have 1 only recom m ended PERF self-self ‘No, 1 only recom m ended 0 m yself.’ (21) U nintensified/sim ple object pronouns in Chinese N i qu huang-gong de shi-hou you-mei-you kan-dao huang-dii? you go em peror palace DE tim e have-not-have see-reach em peror ‘Did you see the em peror when you w ent to the palace.’ A: W , You, wo you kan-dao tU i. have 1 have see-reach him ‘ 1 have, 1 saw him (there). ’ (22)*'^ A dnom inailv intensified/com plex object pronouns in Chinese While ziji is acceptable in (20), the complex form wo ziji ‘me self would in most contexts be more felicitous, see (i). The existence of locally bound complex forms, e.g. wo ziji ‘me self, will be discussed in more detail below, (i) Wo/yi-jing tui-jian le wOiZiji- 1 already recommend PERF 1 self-self ‘1 already recommended m yself’ The answer in (22) sounds a bit awkward with the intensified pronoun to ziji. The sentence in (i) would be more appropriate. (i) Wo you kan dao huang-di ta ziji. I have see reach emperor he self-self ‘I saw the emperor himself.’ Alternatively, using the adnominal intensifier ben-ren ‘(lit.) root-person, in person’ instead of ziji would also yield a more felicitous sentence: (ii) “ Wo you kan dao ta ben-ren. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 378 Q: N i qu huang-gong de shi-hou you kan-dao huang-dii Hai-shi zhi you kan-dao ta-de shi-wei? you go em peror palace DE tim e have see-reach em peror or is only have see reach he DE serve guard ‘Did you see the em peror when you w ent to the palace? O r did you m erely see his guards?’ Wo you kan-dao ta zijii- I have see-reach him self-self ‘1 saw him himself (there).’ (23) U nintensified/sim ple subiect pronouns in Chinese < > Zhangsanj chang-chang shuo ta, bu xi-huan kan dian-ying. Zhangsan often-often say he not like see m ovies ‘Zhangsan often said that he didn’t like m ovies.’ “The Lord o f the R ings” ° Ke-shi zuo-tian tUj qu kan “The Lord o f the Rings but yesterday he go see “The Lord o f the R ings” ‘But yesterday he w ent to see “The Lord o f the Rings’” 1 have see reach he root-person ‘I saw him himself.’ Needless to say an exhaustive account of adnominal intensification in Chinese should include discussion o f other adnominal intensifiers than ziji, e.g. ben-ren, A J I' ben-shen, etc. However, since such a study would largely exceed the scope o f this dissertation, we leave that for further research. The Chinese informants consulted for these sentences were speakers of Standard Taiwanese Mandarin Chinese. This explains why the verb you A ‘have’ is used to for the perfect. Speakers of Mainland Mandarin Chinese would use the particle le T instead, see (i) below. Needless to say, the formation of the perfect does not directly bear on the behavior of reflexives and intensifiers. (0 “ Wo kan-dao le ta zijii- 1 see-reach PERF him self-self ‘I saw him him self (there).’ R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. (24)''’ A dnom inailv intensified/com plex subiect pronouns in Chinese 379 Zhangsan, chang-chang shuo ta^ bu xi-huan kan dian-ying. Zhangsan oflen-often say he self-self not like see movies ‘Zhangsan often said that he didn’t like m ovies.’ “The Lord o f the R ings” ° Ke-shi zuo-tian tUi ziji qu kan “The Lord o f the Rings but yesterday he self-self go see “The Lord o f the Rings” ‘But yesterday he himself w ent to see “The Lord o f the Rings’” In the follow ing sections the hypotheses listed in (18) will be discussed in more detail as the present analysis o f Chinese is com pared to previous analyses. W hile it may still be too early to reach a definitive verdict, argue that the present analysis has the advantage o f m aking it possible to explain the intensifier-behavior o f ziji and ta ziji and thereby unify the analysis o f adnom inally intensified DPs (e.g. huang-di ziji ‘the em peror h im se lf) and uses o f ziji which have hitherto been analyzed as reflexive anaphors and logophors. O ther analyses o f ziji appear to be incapable o f achieving a unified account o f all uses/properties o f ziji. Furtherm ore, m ost existing accounts o f Chinese (e.g. H uang and Liu 2001) suffer from the draw back that they need to assum e'^ the existence o f two, three, or even more different zijVs: (i) zijii = adnom inal intensifier (e.g. huang-di ziji ‘the em peror h im self), (ii) zyC = locally bound reflexive anaphor subject to a syntactic principle A, and (iii) zijis = locally free reflexive/logophor'*. '^T o maintain the analysis proposed here one would have to assume the existence of 0-subject pronouns (which may or may be adnominally intensified by adjunction of ziji) as in (i). Such examples will be discussed more in later sections. (0 ^ Zhangsan, chang-chang shuo 0 , ziji bu xi-huan kan dian-ying. Zhangsan often say self-self not like see movie ‘Zhangsan often says that he himself does not like to watch movies.’ Huang and Liu (2001) adduce a long list of solid facts illustrating the different behavior of local (syntactically bound) ziji and LD (pragmatically governed) ziji. While we do not challenge the validity of these facts, we do dispute the need to assume two different ziji'% to account for them. We argue that the difference between local ziji and LD ziji should not be captured as a lexical difference between to homophonous words, but rather as a difference between adnominally intensified (locally boound) 0-reflexives ( 0 ziji) and adnominally intensified locally free phonetically unrealized pronominals ([to] ziji). In other words, there is only one ziji which may intensity different nominal expressions, see sections 6.4-6.6 below. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 380 In this section we have discussed data w hich support analyzing intensifiers and reflexives in M andarin Chinese based on the assum ptions in (1) and (18). The follow ing sections will be dedicated to m ore detailed discussion o f different aspects o f this analysis. 6.3 Internal strncture of adnominal intensifiers in Chinese In this section the internal structure o f adnom inal intensifiers in Chinese will be discussed. In section 6.3.1, w hich focuses on the m orphology o f ziji, it will be argued that this elem ent is bi- m orphem ic rather than m ono-m orphem ic as is often claim ed in the literature. Section 6.3.2. contains a discussion o f w hether Chinese has com plex intensifiers o f the English type, e.g. ta zji ‘he/him self-self w hich is com posed o f a pronom inal elem ent (ta ‘he/him ’) plus an intensifier elem ent {ziji ‘self-self). 6.3.1 Is ziji really mono-morphemic? In the literature on binding, ziji has often been claim ed to be m ono-m orphem ic. The distinction betw een sim ple/m ono-m orphem ic reflexives and com plex reflexives dates back to Faltz (1977) who observed that in many languages m ono-m orphem ic reflexives allow LD- binding (i.e. clause-external binding) while com plex/m ulti-m orphem ic reflexives have to be locally bound. Since ziji allows clause-external antecedents it is naturally to assum e that it is m ono-m orphem ic. W hile it is certainly true that ziji is m orphologically sim pler than ta ziji, this does not m ean that it is m ono-m orphem ic. Except for a few loan words Chinese lexical To this list o f different ziji's most analyses would have to add one more, namely the so-called exclusive adverbial intensifier, see (i). (i) Zhangsan bu ziji zhuo gong-ke Zhangsan not self-self do homework ‘Zhangsan doesn’t do the homework himself.’ See chapter 2 for more discussion of the exclusive adverbial use o f intensifiers. As discussed in chapter 2, the approach to intensification defended in this dissertation makes the unification o f all uses of intensifiers seem less utopian. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 381 m orphem es are all m onosyllabic. The fact that ziji is bi-syllabic is thus strong indication that it consists o f two m orphem es. The bi-m orphem ic structure o f ziji is further confirm ed by the fact that both the |i} zi- and the 3 -ji parts o f ziji can be used independently, see (25-6). (25) a. zi-qi self-cheat ‘self-deceit’ b . ga zi-ai self-love ‘have regard for oneself; self-respect(ing)’ (26) a. ^P t3 zhi ji know self (i) ‘bosom /intim ate friend’, (ii) ‘be intim ate/close’, (iii) ‘know o n e se lf, b. g J B i g li j i zh u yi benefit self m aster thought ‘egoism (lit. ‘self-benefitism ’)’ Since both the g zi- and the 3 -ji parts o f ziji can be used independently, g 3 ziji is obviously a bi-m orphem ic com pound, com posed o f tw o different m orphem es w hich both m ean ‘s e lf'" ’. N ow the question is w hether its bi-m orphem ic status affects its binding properties. In Classical Chinese g zi and H j i also both mean ‘self but they had different syntactic functions. While g zi was used as pre-verbal reflexive clitic indicating coreference between two arguments of a predicate, see (i), H j i was more often used as a logophor (i.e. a reflexive ‘referring not to the subject of its own verb but to that of a clause in which its clause is embedded’ (Pulleyblank 1995:83)), see (ii). ( i) s a s wang zi sha king self kill ‘The king killed himself.’ (ii) bu huan ren zhi bu j i zhi not worry people DE not self know ‘[1] do not worry that people do not know me.’ Motice, that the typical word order of the Classical language (i.e. [ij zi being pre-verbal while cl j i is post-verbal) still tends to be preserved in Modern Chinese compounds. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 382 A ccording to the Pica-style LF m ovem ent approach to LD -anaphora only reflexives which can undergo successive cyclic head-m ovem ent are able to be LD-bound. This is how the alleged difference between sim ple and com plex reflexives in Danish, i.e. sig and sig selv, has often been accounted for (cf. Jakubow icz (1994) discussed in Chapter 3). Since sig is a m ono-m orphem ic phonological clitic it autom atically qualifies as a head. The com plex reflexive sig selv, on the other hand, claim ed to be unable to undergo head m ovem ent since it behaves syntactically as an XP. Sim ilarly the difference between ziji (w hich can be both locally and LD bound) and ta ziji (w hich allegedly^® has to be locally bound) has been explained as follow ing from the m orphological properties o f these elem ents, i.e. the simple reflexive ziji being able to function as a head while the com plex reflexive ta ziji is blocked from head m ovem ent due to its status as XP (cf. Cole, H erm on and Sung 1991, and others). N otice that the question w hether ziji is mono- or bi-m orphem ic does not have to bear on its ability to function syntactically as a head. Syntactically words can still be heads even though they are not m ono-m orphem ic. In other words, even though ziji is clearly bi- m orphem ic and is thus not a clitic it may still function syntactically as a head. In other words, the bi-m orphem ic status o f ziji does not by itself suffice to exclude a Pica-style analysis o f this elem ent as a head reflexive able to undergo successive cyclic head movement. However, as shown in section 6.2, there is am ble evidence elsewhere in the gram m ar show ing that ziji behaves m ore like him self (i.e. com plex reflexive/adnom inal intensifier) than like the sim ple reflexive sig in Danish. There is plenty of evidence suggesting that this generalization is wrong. Sentences with LD-bound ta ziji will be discussed below. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 383 6.3.2 Ta ziji' complex intensifier, complex reflexive 0 ta ziji, or intensified pronoun ta ziji? It has been claim ed that Chinese has tw o form s o f the adnominal intensifier, viz. (i) ziji ‘self- s e lf (27a) and (ii) ta ziji ‘he/him s e lf (27b). (27) a. ” Huang-di ziji lai le. (ziji = adnom inal intensifier) Em peror self-self com e PERF ‘The Em peror him self cam e.’ b. Huang-di ta ziji lai le. (ta ziji = adnom inal intensifier?) Em peror he self-self com e PERF ‘The Em peror him self cam e.’ (28) a. = Huang-di hen ziji. (0 z iji = intensified reflexive) em peror hate self-self ‘The em peror hates him self.’ b. » Huang-di hen ta ziji. ( 0 ta ziji = intensified reflexive) em peror hate he self-self ‘The em peror hates h im se lf’ W hile the existence o f com plex intensifiers o f the ta ziji type in M andarin Chinese, see (27b), is still debated^', the existence o f languages w hich have both sim ple and com plex form s o f intensifiers is beyond dispute. Some registers o f French allow two different forms o f the adnom inal intensifier, the sim ple form meme ‘same, s e lf, see (29a), and the com plex form lui- meme ‘h im self, see (29b), which is com posed o f the pronoun lui ‘him ’ -l- meme in the same way as M odern English him self was form ed by adding the Old English sim ple adnom inal intensifier self'sQ lf to the object pronoun him. (29) a. Le roi meme est venu. the king sam e is come ‘The king him self cam e.’ People who do not recognize the existence of to ziji as a complex adnominal intensifier in Mandarin Chinese may chose to analyze the sentences in (27b) as consisting of Huang-di ‘emperor’ as topic which is resumed by a subject pronoun to intensified by ziji ‘self-self. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 384 b. Le roi lui-m em e est venu. the king him -sam e is come ‘The king him self cam e.’ It is therefore not unrealistic to imagine that, during the transition period from Old English to M odern English, English also may also have had both a simple and a com plex form o f the intensifier. However, unlike French w hich appears to still be in such a transition period, M odern English now only allow the com plex form o f the intensifier, see (30a) vs. (30b). (30) a. *The king s e lf came. b. The king h im self came. One w ould expect both the sim ple and the com plex form s o f the adnom inal intensifier in languages like French and Chinese to be able to intensify any kind o f nom inal expression. This is, however, not the case. There seem s to be a strong tendency to avoid using the com plex form o f the intensifier to intensify object pronouns, see the Chinese exam ples in (31a,b). (31) a. Peter H j^ M a iy M ^ T i m B l i i m m m X m n ' > Peter, skua Mary gen chu-le tui ziji yi-wai suo-you de ren tiao-wu P say M with except he self-self INST-outside that-have DE people dance ‘Peteri said that M ary danced w ith everyone except [hinij] himself.’ b. *Peter M ary fffefffiS “ Peter, skua Mary gen chu-le ta tUi ziji yi-wai suo-you de ren tiao-wu c. Peter M ary g!M | 577 g ° Peter, shuo M ary gen chu-le zijii yi-wai suo-you de ren tiao-wu P say M with except self-self INST-outside that-have DE people dance ‘Peter; said that M ary danced w ith everyone except him self.’ English displays a sim ilar tendency to avoid overly redundant forms. That is repetitive forms like him him self arc usually sim plified to himself, see (32a-b). (32) a. Peter, said that Mary danced with everyone except himself, b. */??him, himself. c. *self R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 385 It is to the best o f my know ledge still an open question why English does not allow intensified object pronouns, i.e. why do forms like */??we myself, * nihim him self tend to be shunned by native speakers. H aplology m ight be a plausible account o f the absence o f the form s him him self her herself them themselves but such an explanation seem s to run into trouble with forms like me m yself and us ourselves^^. In any case, some rule m ust be at play reducing him him self to him self W hatever the rule is, the same phenom enon appears in Fr., com pare (32a-c) and (33a-c). (33) a. Pierre bavardait avec lui-meme b. *lui lui-meme c. *meme Peter chatted w ith him -sam e/him him -sam e/sam e ‘Peter chatted with h im se lf’ In the case o f Chinese, one can im agine several different accounts o f why *ta ta ziji does not occur. One way w ould be to assum e the existence o f a m orpho-phonological rule sim plifying ta ta ziji to [ta] ta ziji in the same way as him him self reduced to [him] him self m English. A nother approach would be to assum e that only the sim ple adnom inal intensifier zi]i is allowed to intensify the object pronoun ta. English and French illustrate that the absence o f object pronouns intensified by com plex intensifiers (e.g. *him him self and *lui lui-meme) is not in itself enough to falsify the hypothesis that a language has com plex intensifiers. That is, the absence o f form s such as ta ta ziji does not by itself suffice to falsify the hypothesis that Chinese has com plex intensifiers o f the ta ziji type. We thus conclude that the existence o f com plex intensifiers {Huang-di ta ziji ‘the Em peror h im se lf) in M odem Chinese is at least a theoretical possibility. Baker (1995) suggests that the absence of intensified object pronouns might be a consequence of his prominence condition. If intensifier-adjunction is sensitive to syntactic prominence and if subjects are assumed to be syntactically more prominent than objects then it follows that only nominal expressions in subject position can be intensified. Such an account, however, is unable to explain why Danish allow intensified object pronouns and reflexives and why, even in English, sentence like Mary wrote a letter to the King him self are ok in the right context. See the critique o f Baker’s prominence condition in chapter 2, section, 2.2.3.2.2, (56). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 386 Critics o f this analysis m ight claim that the sequence Huang-di ta ziji in (27b) should be analyzed as the DP Huang-di in topic position resum ed by the intensified pronoun ta ziji in subject position. W hile such an analysis o f (27b) is possible it is worth rem em bering that the topic position is characterized by being separated from the rest o f the sentence by a b rief intonational pause. Since (27b) can be pronounced w ithout intonational pause we conclude that Huang-di ta ziji can be analyzed as a com plex DP. Independent evidence for this com es from other com plex post-nom inal m odifiers, e.g. yi-ge-ren ‘alone’^^, w hich are known to be able to form com plex DPs, com pare (34) and (35) (34) A dnom inally intensified DP: DP / \ D P ta ziji huang-di (35) D P m odified by j^/-ge-re«‘alone’: DP / \ DP yi-ge-ren huang-di In section 6.5.1 below we argue that locally bound ta ziji is in fact a 0-reflexive intensified by the com plex intensifier ta ziji, i.e. 0 ta ziji, see (36). (36) Intensified DP: DP / \ DP ta ziji 0 Cf. Siemund (2000) who also discusses the close similarities between adnominal intensifiers and other post- nominal modifiers, e.g. Ger. alein ‘alone’ In Der Konlg alein ‘the King alone’ vs. der Konig selbst ‘the king him self. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 387 6.4 Analyzing what appears to be uses of d ji as reflexive anaphors as adnominal intensifiers In this section we outline analyses o f w hat appears to be (and is generally assum ed to be) uses o f ziji as locally (section 6.4.1) or LD -bound (section 6.4.2) sim ple reflexive anaphors as com plex intensified 0-reflexives (i.e. 0 ziji). Section 6.4.3 concludes the section by sum m arizing the results. 6.4.1 Locally bound 0-reflexives in argument position: complex reflexives = a sub-type of adnominal intensification In this section we investigate w hether the distribution o f locally bound unintensified 0 - reflexives and intensified 0-reflexives in M andarin Chinese can be accounted for w ithin the fram ew ork assum ed here. Is the distribution o f locally bound unintensified 0 -reflexives and intensified 0 -reflexives (i.e. 0 ziji) determ ined by predicate m eaning in the sam e way as the distribution o f sig vs. sig selv in D anish and 0 vs. 0 him self m English? D o we have evidence o f a distinction betw een anti-reflexive, neutral, and inherently reflexive predicates in Chinese? 6.4.1.1 Anti-reflexive predicates A ccording to the analysis proposed here, all sem antically anti-reflexive (or “hidden” neutral) predicates should potentially require the intensified reflexive 0 ziji in object position^'*. As far as we can tell at this point (not having done an exhaustive survey), this prediction is hom e out by facts. That is, the anti-reflexive verbs in (37) do seem to require the intensified form o f the 0-reflexive, see (38a,b), (39a,b) and (40). Anti-reflexive predicates require adnominal intensification of reflexive pronouns because they presuppose representational non-identity. This presupposition creates an inherently contrastive environment which triggers adnominal intensification. See chapter 3 for more details of this analysis of the relation between anti-reflexivity and intensification. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 388 (37) a. ® hen ‘hate’ b. qi-pian ‘cheat’ c. ti-dai ‘replace, substitute fo r’ d. zhui ‘follow, court, pursue’ e. etc. (38) a. Zhangsan hen 0,. Zhangsan hate 0 ‘*Zhangsan hates 0 i.’ b. Zhangsan hen 0 , ziji. Zhangsan hate 0 self-self ‘Zhangsan hates 0 i h im se lf’ (39) a. Zhangsan qi-pian 0j. Zhangsan cheat 0 ‘*Zhangsan cheats 0 j .’ b. o Zhangsan qi-pian ziji. Zhangsan cheat 0 self-self ‘Zhangsan cheats 0 j him self.’ (40)^' a. » Xiaom ingj yi-wei lao-shi2 you yao ze-guai 0 le. X iaom ing think teacher again will blam e CSR ‘X iaom ing] thinks that the teacher2 will blam e (m e/you/him ]/3/her/him self/herself/us/you/them .. . ) again.’ b. » Xiaoming I yi-wei lao-shi you yao ze-guai 0-ziji le. X iaom ing think teacher again will blam e self-self CSR ‘X iaom ing] thinks that the teacher2 w ill blam e (him ]/him self2) again.’ The exam ples in (38) and (39) show that local unintensified 0-reflexives are ungram m atical with anti-reflexive predicates. The exam ple in (40a) (which is from Huang (2000:86) furtherm ore illustrates that null objects in Chinese can receive a num ber o f different While finding it marginally acceptable, none of my informants liked the sentence in (40a), which Huang (2000:86) finds to be ok. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 89 interpretations. However, in spite o f the range o f possible interpretations o f the null object in (40a), it cannot (or only very m arginally) be interpreted as a 0-reflexive bound by the local subject. A dnom inally intensification is necessary to (i.e. 0 ziji in (40b)) save the sentence. Sa-constructions or co-verb + verb sequences also behave as anti-reflexive constructions in that they do not allow unintensified 0-reflexives, see (41) and (42). (41) a. f E b a ___qiangbi ‘s h o o t ’ b. JG b a ___diu-chu-qu ‘th ro w o u f (42) a. w e i bianhu . ‘d efen d ’ b. KM g e n __shuohua ‘speak w ith_ _ ’ In this respect they behave like Danish prepositional predicates w hich also do not allow the simple unintensified reflexive. A nd sim ilar to the case o f D anish prepositional predicates, the absence o f unintensified 0 -reflexives w ith Zia-constructions or co-verb + verb sequences may follow from the m orpho-phonological properties o f 0 -reflexives (i.e. lack o f stressability), rather than being due to the sem antic properties o f the predicate. 6.4.1.2 Neutral predicates As described in chapter 3-5 many o f the neutral predicates in English and Danish are either groom ing verbs, or verbs denoting movement^^, see (43-44). (43) N eutral groom ing verbs: a. Hei washes DP^ / 0 , / 0 , himself. b HaUi vasker DP^ / sig, / sig, selv. (44) N eutral verbs denoting non-translational m ovem ent: a. Hei bends DP^ / 0 , / 0 , himself. b Hani bajer DP^ / sigi / sig; selv. Some verbs denoting emotional states may also qualify as neutral in Danish, e.g. bekymre ‘worry’, etc. See chapter 4 for more discussion o f different types o f neutral predicates. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 90 O ne w ould therefore expect verbs o f these sem antic types to qualify as neutral predicates in Chinese as well. This, how ever as shown in (45)-(46), appears not always to be the case. (45) Predicates denoting groom ing activities: a. *)5feSD P xi-zao xi-zao ziji xi-zao DP w ash-bath w ash-bath self-self w ash-bath DP 'wash (oneself), take a bath, bathe (oneself)' b. I l s l n i s i DP shu-tou shu-tou ziji shu-tou DP com b-head com b-head com b-head DP 'com b (oneself) c. * S iJ f lB D P gua hu-zi gua hu-zi ziji gua hu-zi DP cut beard-N O M cut beard-N O M cut beard-NOM ‘shave (oneself).’ (46) Predicates denoting non-translational m ovem ent: *'W T K §B *)^TI3^dp wan xia yao wan xia yao ziji wan xia yao DP bend down waist bend down w aist self bend down waist DP 'bend down, duck' b. i& M B f i S l S B * f £ s lD P di tou di tou ziji di tou DP low head low head self-self low head DP 'duck' c. # M # f i * # M # f l g B * # M # f l D P shen zhan shen ti shen zhan shen ti ziji shen zhan shen ti extend spread body body 'stretch oneself Unlike the typical groom ing verbs in English and D anish allow reflexives in object position, the Chinese counterparts are often com plex predicates com posed o f a verbal elem ent denoting the action + either a nom inal elem ent denoting an inalienable body-part, e.g. tou ‘head’ in shu- tou ‘(lit. com b head) com b’ as in (45b) and (46a,b,c), or a cognate object as in (45a). Since these cognate or inalienably possessed objects saturate the position o f internal argum ent there is no room for any m ore direct objects. This accounts for the judgm ents in (45-46); the starred sentences are ungram m atical because they contain tw o internal argum ents w here there is only room for one them atically R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 391 One way to contrast the activity o f com bing oneself with com bing som ebody else is to add an overt possessor to the otherwise inalienably possessed object, see (47). (47) a. b. c. o Zhangsan shu 0 (de) tou. com b head ‘Zhangsan combs 0 . ’ SB m S i ° Zhangsan shu ziji de tou. com b-head self-self DE head ‘Zhangsan combs 0 h im se lf’ (?)5g=^j^ lie §I “ Zhangsan shu wo de tou. com b I DE head ‘Zhangsan com bs m e.’ Notice, however, that the sentences in (47b,c) are only m arginally acceptable. The idiom atically correct way to say “Zhangsan com bs m e’ w ould be by using a circum locution with the verb ^ bang ‘help’ as in (48). (48) a. W S Zhangsan shu hang wo shu-tou. Zhangsan help me com b-head ‘Zhangsan com bs m e.’ In the case o f the verb+cognate object xi-zao ‘w ash’, the only way to make this verb transitive is to use the ^a«g-construction illustrated in (48), see (49)^’. (49) a. a f t » Zhangsan xi wo de zao Zhangsan w ash I DE bath b. a Zhangsan .shu hang wo xi-zao. Zhangsan help ‘Zhangsan bathes m e.’ me w ash-bath Likewise m ost verbs denoting non-translational bodily m ovem ent also cannot be found (or are highly infelicitous) in the genitive construction illustrated in (47), see (50)“ . ^ Notice, however, that strongly contrastive contexts can, in many cases, improve judgments. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 92 (50) a. ° Zhangsan di wo de tou. Zhangsan lower I DE head b. S T a W Zhangsan wan xia wo de yao. Zhangsan bend down I DE waist ‘Zhangsan bends m y waist dow n.’ = ‘Zhangsan bends me dow n.’ N otice, however, that the ftang^-constructions are all anti-reflexive^‘ ’, see (51). (51) n *0/(??)0 Zhangsan shu hang 0 / 0 ziji / Lisi xi-zao. Zhangsan help 0 / 0 self-self / Lisi wash-bath ‘Zhangsan bathes him self/Lisi.’ A nother way to aehieve a reflexive reading o f an em pty object position is by om itting the cognate object, see (52A). (52) U nintensified 0-reflexives in C hinese: Q : N i you-mei-you xi-zao? you have not-have have wash-bath ‘Have you bathed’ A: ° You, wOjyi-jingxi 0ile. have I already w ash PERF ‘1 have, I have already w ashed 0 . ’ Finally, notice that the ability to have inalienably possessed objects is one way to test w hether a verb is neutral or anti-reflexive. Just like in Danish, Chinese neutral verbs are com patible with inalienably possessed object while anti-reflexive and ‘hidden” neutral verbs are not^°, see (53). Notice that form some reason certain verbs denoting non-translational movement fare better in the genitive construction than certain other neutral constructions involving inalienable possession, compare (50a) and (50b). As anti-reflexive to/jg-constructions should in principle be able to take 0 ziji. However, due to their meaning reflexive scenarios are highly unlikely (since helping is typically an other-directed activity) and thus tend to be only marginally acceptable. Furthermore, as discussed in section 6.4.1.1 all ^-constructions are distributionally anti reflexive because of the phonological properties of 0-reflexives. 0-reflexives cannot carry stress on their own and are thus unable to occur in PPs. In this respect, 0-reflexives behave like the simple reflexive sig in Danish, see chapter 3 section 3.3.6 and chapter 5 section 5.2.1.5. See chapter 3, section 3.3.2.6 for discussion of inalienable possession in Danish. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 393 (53) a. W im M ° wo shu tou. I com b head (lit.) ‘I com b my head’ = ‘I am com bing (m yself).’ b. wo hen tou. 1 hate head *(i) ‘I hate m y head’ °'‘(ii) ‘I hate heads’ / i hate the head’ (i.e. som ebody else’s head) We conclude that the absence o f neutral predicates o f the English/D anish type (e.g. Peter washes 0/shaves 0/ducks 0/etc.) in Chinese (except for forced exam ples as (52A), w hich probably involve deleted objects rather than 0-reflexives) does not constitute a serious problem for the present proposal since it is the formal properties o f predicate-form ation in Chinese w hich are responsible for the absence o f m ono-syllabic neutral transitive verbal predicates^'. That is, it appears to be the case that neutral predicates (i.e. groom ing verbs, m ovem ent verbs, etc.) in M andarin Chinese are characterized by the m andatory presence o f either cognate objects (e.g. xi-zao ‘w ash-bath’) or inalienably possessed objects (e.g. .shu-tou ‘com b-head’). H ence the absence o f m ono-syllabic neutral verbs o f the type xi 0 ‘to wash o n eself can not be used as proof that Chinese does not have neutral predicates w hich take 0 - reflexives. All we need to do is to classify predicates as anti-reflexive, neutral or inherently reflexive depending on w hether they allow DPs, intensified 0-reflexives, and unintensified 0 - reflexives in possessor position, see (54). (54) Predicate tv p e: D istribution: Exam ples: a. Anti-reflexive: DP / 0 ziji / * 0 (55) b. N eutral: DP /0 z iji / 0 (47) c. Inherently reflexive: *D P /* 0 ziji / 0 (58) The sentence in (55) illustrates the type o f “anti-reflexive possessive predicate” referred to in (54). It would thus be interesting to test whether bi-morphemic neutral verbs of the tui-jian ‘recommend’ type can take 0-refiexives. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 394 (55) a. ° Zhangsan tou-le Lisi de qian-bao. Zhangsan steal-PERF Lisi DE m oney-w rap ‘Zhangsan stole Lisi’s wallet. b. Zhangsan tou-le ziji de qian-bao. Zhangsan steal-PERF self-self DE m oney-wrap ‘Zhangsan stole his ow n wallet. c. Zhangsan tou-le 0 qian-bao. Zhangsan steal-PERF m oney-w rap ‘Zhangsan stole wallet. 6.4.1.3 Inherently reflexive predicates In chapters 3, 4 and 5 we discussed inherently reflexive expressions in D anish and English, see (56-7). (56) a. Peter skynder sig / *sig selv / *Hans. Peter hurries REFL / REEL self / Hans ‘Peter is in a hurry.’ b. Peter tog en kniv med sig / *sig selv / * Plans. Peter took a knife w ith REFL / REFL self / Hans ‘Peter took a knife w ith him .’ c. Peter dukker sig / *sig selv / *Hans. Peter dueks REFL / *REFL self / *Hans. ‘Peter ducks 0 / * 0 him self / *H ans.’ (57) a. Peter is out o f his / *his own / *Hans’ s mind. b. Peter took a knife with him / *himself / *Hans. c. Peter ducks 0 / * 0 him self / *Hans. N otiee that the intensified form s o f the reflexives (i.e. sig selv/sin egen and 0 himself/his own) are never allow ed w ith inherently reflexive predicates. This (surprising) behavior o f him self can be explained by assum ing that it is an intensifier. Thus the absence o f him self in those construction can be seen as follow ing from the contrastiveness condition on adnom inal intensification (see (62) in chapter 2, section 2.2.3.2.2) w hich states that intensifiers can only occur in positions w here contrast w ith other entities is possible. O ur analysis therefore predicts that Chinese should not have any predicates/constructions which are sem antically R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 395 inherently reflexive w hile still allow ing ziji. To the best o f our know ledge this prediction seems to be true: Chinese does not (appear to) have any inherently reflexive expressions which allow ziji. The predicate { £ § 1 di-tou ‘low er head/duck’, discussed above (see (50)), may be a potential candidate for an inherently reflexive predicate, see (58). (58) a. i& 0 , S i ° Zhangsan/ di 0 / tou. Zhangsan low head ‘Zhangsan lowers his head/ducks.’ b. f £ 0 / S B gli o Zhangsan/ di 0 / self-self de ton. Zhangsan lower self-self DE head c. S i ° Zhangsan, di wo^ de ton. Zhangsan low er I DE head 6.4.1.4 Predicate types: summary O n the basis o f the discussion in this section we conclude that Chinese data relating to the behavior o f em pty object positions and locally bound ziji do not em pirically falsify the proposals in (1), and (18) on which the present fram ew ork is based. W hile there is a great deal o f data supporting the hypothesis that ziji always has the same properties as intensifiers in other languages (e.g. Eng. himself, and Dan. selv) and thus should be consistently analyzed as an intensifier, conclusive evidence that Chinese has 0-reflexives is harder to com e by. Some evidence o f a contrast between anti-reflexive, neutral and inherently reflexive predicates was found in the “possessive predicates” discussed in the previous sections, but, needless to say, we still need to specify m ore precisely w hat exactly Chinese 0-reflexives are. This and other related questions will be taken up in the follow ing sections. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 96 6.4.2 LD-bound ziji'. complex pronoun or complex reflexive? As described in H uang and Liu (2001) (am ong others) local ziji and LD ziji differ from each other in im portant ways. In their term inology, local ziji is a reflexive anaphor whose distribution is determ ined by the syntactic principles o f the binding theory. LD ziji, on the other hand, is a logophoric elem ent w hich has to be bound by som e kind o f attitude operator and w hose distribution thus fall under both syntactic AND pragm atic principles. W hile we agree with their account o f the facts, we disagree w ith their analysis which is based on the assum ption that the lexicon contains two entries for ziji: (i) ziji, = a reflexive anaphor subject to principle A o f the binding theory, and (ii) zZ/'L = a logophoric elem ent bound by attitude operator and susceptible to pragm atic factors. Instead o f proposing tw o lexical entries for ziji, we propose that ziji is always an intensifier and that the observable differences between local and LD ziji are due to a difference betw een the phonetically unrealized elem ents that ziji is intensifying in the tw o cases. In the case o f “local ziji", the elem ent ziji is intensifying a 0 - reflexive (e.g. Zhangsan hen 0 ziji ‘Zhangsan hates 0 h im self), which falls under principle A o f the binding theory. In contrast, we suggest that in the case o f “LD ziji", ziji is intensifying a phonetically unrealized pronom inal (e.g. [ta] ziji ‘[him] h im se lf(= Dan. ham selv)) w hich falls under principle B o f the binding theory. See section 3.4 in chapter 3, and section 5.2.3 for discussion o f intensified object pronouns in Danish and English. Intensified pronouns in M andarin Chinese will also be taken up in section 6.6 below. 6.5 Intensifier analysis of local and LD la ziji In this section, the consequences o f extending the intensifier-based analysis proposed here to local (section 6.5.1) and LD ta ziji (section 6.5.2) will be discussed. We follow the convention, introduced in chapter 5, to indicate phonetically unrealized material with square brackets. Instances of locally free him self in Modern English, are thus represented as [him] himself. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 397 6.5.1 Locally bound ta ziji: complex pronoun or complex reflexive? A t first sight the fact that Chinese allow the com plex form ta ziji to be locally bound may seem like a serious counterexam ple to the analysis proposed here, see (59b). (59) a. *Zhangsarii ai ta/. b. tQi ziji. c. 0 i ziji. Z hangsan love him /him self-self/self-self ‘Zhangsan loves h im se lf’ The simple unintensified pronoun ta ‘he/him ’ in (59a) behaves as expected, i.e. it cannot be locally bound w ithout violating principle As show n in (60a,b), intensification o f pronouns in D anish does not affect their binding properties, i.e. they rem ain pronouns subject to principle B w hether they are intensified or not. (60) a. Peteri vasker *hami. b. *hami selv. Peter washes him /him self The sam e applies to pronouns in M odern English, see (61a-b). (61) a. Peteri washes *himi. b. */i/m, himself. c. himself. If ta ziji were an intensified pronoun o f the sam e type as Danish ham selv ‘him s e lf then (59b) should be ju st as bad as (60b) and (61b). The fact that this is not the case indicates that ta ziji in (59b) is not an intensified pronoun. But if ta ziji in (59b) is not an adnom inally intensified pronoun, then what exactly is it? Traditionally it has been analyzed as a com plex reflexive which due to its status as an XP has to be locally bound, cf. Cole, Herm on & Sung (199?) follow ing Pica (1987). Since this solution im plicitly assum es that adding z iji to the pronom inal ta changes the binding properties o f the latter it cannot be adopted in the present fram ew ork which is based on the independence o f binding and intensification. One option ” Genitive constructions which appear to violate this generalization (that simple ta ‘he/him’ cannot be locally bound) are discussed in more detail in section 6.6.1.1.1 below. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 398 consists in analyzing the occurrence o f ta ziji in (59b) as a 0-reflexive adnom inally intensified by the com plex intensifier ta ziji, see (62a) and (63 a). (62) a. 0-reflexive + com plex intensifier; DP (e.g. (63a)) b. 0-reflexive + simple intensifier: (63) a. b. Zhangsan, ai 0 , ta Zhangsan loves REFL he ‘Zhangsan loves h im se lf’ Zhangsan, ai 0 , ziji. Zhangsan loves R EFL s e lf self ‘Zhangsan loves him self / DP I 0 / DP 0 ziji. s e lf self \ ta ziji DP (e.g. (63b)) Zljl (see (62a)) (see (62b)) In other words we need to assum e that M odern Chinese has both sim ple and com plex intensifiers w hich can intensify 0-reflexives, com pare (62a) and (62b). As discussed in section 6.3.2 there is independent evidence supporting the existence o f both sim ple and com plex adnom inal intensifiers in M andarin Chinese. 6.5.2 LD-bound ta ziji: complex pronoun or complex reflexive? In spite o f w hat has often been claim ed in the literature, the com plex form ta ziji, w hich is com posed o f the third person pronoun ta ‘he/sh e/if + the adnominal intensifier ziji ‘s e lf, can be long-distance bound, see the follow ing exam ple from Huang (2000:96(2.169a))^'^, see (64) (64) X iaom ingj shuo lei-sheng ha ta zijiy xiao le y i tiao. X iaom ing say thunder-sound BA 3SG self-self frighten PERF one jum p ‘X iaom ing said that the loud crash o f thunder had given him a fright.’ ' See also Pan (1998) for more such examples. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 3 9 9 Such exam ples has som etim es been construed as evidence that com plex reflexives can be LD- bound, contrary to Faltz’s (1985) generalization that LD reflexives are m orphologically simplex. N otice, however, that one does not have to assum e ta ziji ‘3SG self-self to be a com plex reflexive in the first place. In the fram ework adopted here, it could also be considered an intensified pronoun on a par with Danish intensified pronouns like ham selv ‘him s e lf, etc. In others words, the apparent LD-binding o f ta ziji follows from the pronom inal character o f ta ‘3SG ’. See (65) w hich illustrates that the pronom inal ta can be bound by the m atrix mh]Qct Xiaoming. (65) Xiaomingi shuo leisheng ha tai xiao ie y i tiao. X iaom ing say thunder-sound BA 3SG frighten PERF one jum p ‘X iaom ing said that the loud crash o f thunder had given him a fright.’ In sum m ary, LD ta ziji is not a com plex reflexive but rather a pronom inal ta ‘he/him ’ w hich has been intensified by adjunction o f the adnom inal intensifier zi/7 ‘se lf-se lf, see (66b). (66) a. ta -I- com plex intensifier: D P (e.g. (64)) / \ DP ta ziji [ta] b. ta -f- sim ple intensifier: D P (e.g. (64)) / \ D P ziji I ta Just like with the pronom inals in D anish (ham ‘him) and English (discussed in chapter 3 and chapter 5) adnom inal intensification (e.g. ham seiv ‘him s e lf) does not alter their binding properties, i.e. they still fall under principle B requiring them to be locally free^^ Needless to say, the question why occurrences locally free ta ziji are so rare in Chinese compared with Danish ham selv ‘him self needs to be addressed. Due to space limitations, we leave this question for future research. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 0 0 6.6 Intensified and unintensified pronouns in Mandarin Chinese In this section we investigate the behavior o f intensified and unintensified pronouns in M andarin Chinese. A ccording to the assum ption that binding and intensification are independent o f each other, we predict that intensification o f pronom inals should not interfere with the binding properties o f the latter. As shown in chapters 3 and 5, this prediction seems to be borne out by the facts in both Danish and English. In the follow ing we test to w hat extent this analysis can be extended to Chinese. 6.6.1 Simple/unintensified pronouns in Mandarin Chinese: Principle B The sentences in (67-8) dem onstrate the existence o f principle B effects with third person personal pronouns in Chinese. (67) a. Zhangsariihen b. zijij*Q. Zhangsan hate him /self-self ‘Zhangsan hates him /him self.’ (68) a. Zhangs an i jiao Lisik da ta,;*t„- b. zijij Zhangsan ask Lisi hit him /self-self ‘Zhangsan asks Lisi to hit h im /h im self’ The sentences in (67-8a) illustrate the fact that the pronom inal ta ‘he/him ’ m ust be locally free. N ow the question is w hether intensification o f pronom inals affect their binding properties. The sentences in (69) seem to suggest that it does. (69) a. ZhangsaUi hen tai,*„ziji. ‘Zhangsan hates h im /h im self’ b. Zhangsan, jia o Lisik da ta,.k * „ ziji. ‘Zhangsan asks Lisi hit h im /h im self’ The com plex form ta ziji ‘him self-self seem to follow the pattern o f ziji in (6 0 -6 lb). Traditionally this has been taken as evidence that ta ziji is a com plex reflexive w hich m ust be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 401 locally bound. However, as discussed in sections 6.3 and 6.5.2 above several different analyses o f ta ziji ‘him self-self are possible, see (70). (70) a. Intensified pronoun: DP / \ DP ziji I ta b. Intensified 0-reflexive: D P (=(69a),(69bi)) / \ DP ta ziji I 0 c. Intensified pronoun: D P (=(69b,)) / \ DP ta ziji I [ta] N otice furtherm ore, that the sentence in (70b,) illustrate LD -binding o f ta ziji, see also discussion o f (64-66). 6.6.1.1 Overlapping distribution of simple ta and ziji/ta ziji The distribution o f sim ple ta vs. ziji/ta ziji overlaps in tw o contexts: (!) possessive construction (section 6.6.1.1.1), and (ii) em bedded subject position (section 6.6.1.1.2). 6.6.1.1.1 Possessive constructions The com plem entarity o f sim ple pronom inals (e.g. ta fiij’he/him ’) and ziji/ta ziji breaks down in p o sse ssiv e con struction s, se e (7 1 -2 ). (71) a. Zhangsan, ai tai ,, de gou. b. zijii/*,, c. ta zijii * ,, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 0 2 (72) a. Zhangsarii shuo Lisik ai tai/y„ de gou. b. zijii/w*„ c. ta zijij/k/*o The existence o f tw o different possessives in Chinese, i.e. ta and ziji, has som etim es led to Chinese being com pared to Danish w hich distinguishes between pronom inal and reflexive possessives, see hans ‘h is’ vs. sin ‘PO SSR EFL’ (73). (73) a. Peter, siger at Hansk hader hans, *k/o (egen) hund. his (own) b. sinnk*o (egen) PO SSREFL (own) ‘ Peter says that Hans hates his/PO SSREFL (own) dog. ’ N otice, how ever that, unlike Chinese, the com plem entarity betw een pronom inals and reflexives is m aintained in Danish regardless o f w hether the intensifier egen ‘ow n’ is present or not. We therefore conclude that the Chinese system is closer to the English (or French) w here both his and his own (Fr. son vs. son propre) can be locally or LD-bound, ju st like Chinese ta and [ta] ziji/ta ziji. 6.6.1.1.2 Embedded subject pronominals In M andarin Chinese the subject o f em bedded clauses can be realized in three different ways: (i) a a sim ple unintensified pronoun ta ‘he’ (74a), as an intensified pronoun [taj ziji (74b), or as an a pronoun intensified by the com plex intensifier [taj ta ziji (74c). (74) a. Zhangsan; shuo ta;/o kanjian-le Lisi. b. zijU*o c. ta zijii/f,, Zhangsan say he/self-self/he self-self saw -Perf Lisi ‘Z hangsan says that h e/h e h im s e lf saw L is i.’ In this section we explore to w hat extent the analysis o f intensified Danish and English pronouns proposed in chapter 3 and 5 can be extended to em bedded subject pronom inals in M andarin Chinese. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 403 As show n in (75-76) sim ple unintensified pronouns behave alike (i.e. obeying principle B) in both Chinese and English. (75) Joarii believes that shCi^ will win. (76) Joan g ° JoaHi xiang-xin tai,^ huiying. Joani believes she^z will win. ‘Joanj believes that shei/z will w in.’ Since the sim ple pronoun does not display any surprising behavior we will focus our attention on the two types o f intensified subject pronouns in Chinese, i.e. [ta] ziji and [ta] ta ziji. The exam ples in (77-8) illustrate both [taj ziji and [taj ta ziji behave like their English counterpart she herself in out-of-the-blue contexts. (77) JoaHi believes that she, *^herself will win. (78) a. ° JoaUi xiang-xin [ taj ta zijii hui ying. b. *[taj tazijh c. [taj zijii d. [taj *ziji^ Joani believes she herself/herself will win. ‘Joani believes that she herself will w in.’ In his (1991) article, M cKay showed that, given the right context, intensified pronouns can be bound by an extra-sentential antecedent, see (79) (79) Anni wants to interview the winner. Joank believes that sheii herself will win. As show n in (80) the judgm ents o f the direct translations o f (79) differ from those o f their English counterparts. (80) a. *Ann, xiangyao fang-wen de-jiang-zhe. JoaHk xiang-xin ta ziji, huiying. b. Ann, xiangyao fang-wen de-jiang-zhe. Joan,, xiang-xin ta zijit hui ying. c. *Ann,xiangyao fang-wen de-jiang-zhe. JoaHk xiang-xin zijii huiying. d. AnHi xiangyao fang-wen de-jiang-zhe. Joank xiang-xin zijik hui ying. A nn think w ant interview receive-prize-N O M . Joan believe self-self will win ‘Anni wants to interview the winner. Joan believes that she*i/k herself will w in.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 0 4 It thus appears to be the case that [ta] ziji and [ta] ta ziji are m ore likely to be bound the closest subject than intensified pronouns in English. However, we do not w ant to rule out the possibility that Chinese intensified pronouns m ay be bound extra-sententially. As show n by (81-2) and (83-4) the intensified pronoun [taj ta ziji m ay indeed have an extra-sentential antecedent. (81) How will she herself do on the exam ? (82) a. T aziji kao-shi hui kao-de zen-meyang? b. (??/*)Ziji kao-shi hui kao-de zen-me yang? (he) self-self exam will exam DE how type ‘How will she herself do on the exam .” (83) M ary I has been worrying about the grades o f her friends. How will shcj herself do on the exam? (84) Margi zui-jin hen dan-xin ta de peng-you de cheng-ji. M ary recently very w orry she DE friend(s) D E grade(s) ‘ M ary has been w orrying about the grades o f her friends. ’ a. Ke-shi ta, ziji hui kao-de zhen-me yang? b. *zijii but she self-self/self-self will exam DE how type ‘But how will she herself do on the exam ?’ The asym m etry betw een the (a) and (b) exam ples in (81-4) show that subject [taj ziji is m ore restricted in its choice o f antecedent than [taj ta ziji. A t this point we have no explanation o f this difference. The contrast betw een the (85-6) and (87-8) show that, in both English and Chinese, the choice o f context does m ake affect the ability o f intensified pronouns to skip a closer non subject antecedent in favor o f a m ore distant m atrix subject antecedent. (8 5 ) Susan i told everyone who knew Maryk that she, h erself was pregnant. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 405 (86) a. Susarii gao-su suo-you ren-shi Maryk de ren ta, ziji huai-yun le. b. *takziji c. d. tO k e. ziji, f. * zijik Susan tell all know M ary D E people pregnant LE ‘Susan told everyone who knew M ary that was pregnant.’ (87) Maryk always lectured at the pregnant girls. Susan/ did something to get even. Susan/ told everyone who knew Maryk that shek herself was pregnant. (88) Maryk chang-changjiao-xun huai-yun de nu-hai-zi. Susan/ xiangyao hao-fu. Mary Maryk always lectured pregnant DE girls. Susani think want get-even ‘Maryk always lectured at the pregnant girls. Susan, did som ething to get even.’ a. Susan/ gao-su suo-you ren-shi Maryk de ren ta/ ziji huai-yun le. b. tOkZiji c. ta/ d. tO k e. ziji, f. * zijik Susan; told all know Maryk DE people that ____ was pregnant. ‘Susan; told everyone who knew Maryk that ____ was p re g n a n t’ The contrast betw een (86b) and (88b) is entirely due to the richer context o f (88). This kind o f data (w hich shows that intensification o f subject pronouns (e.g. [ta] ta ziji) is subject to sem antic/pragm atic factors) is evidence that intensification and binding o f subject pronouns are independent in sim ilar ways in M andarin as in Danish and English. N ote how ever, that ziji (which we analyze as an intensified pronoun [ta] ziji) behaves in an unexpected way. As m entioned above, at this point we have nothing to say about sentences such as (86f) and (88f) and sim ply leave the m atter for further research. 6.7 C onclusion In this chapter we have outlined an analysis o f M andarin Chinese based on the proposal that binding and intensification are m utually independent m odules o f the gram m ar. W hile a num ber o f problem s rem ains we hope to have showed that this approach is at least a viable Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 0 6 option w hich deserves to be explored in greater detail. A m ong the evidence adduced in its favor w ere the fact that ziji has m ore in com m on w ith adnom inal intensifiers in other languages than w ith true reflexives, see (4-5) repeated here as (89-90). (89) Typological survey o f reflexive and non-reflexive uses o f reflexive elem ents: (i) Chinese (ii) English (iii) French (iv) G erm an (v) Danish a. reflexive ziji him self se/soi sich sig b. reciprocal - - se sich -s c. middle - - se sich - d. passive - - se sich -s e. deponent verbs - - se sich -s/sig f, intensifier ziji him self (lui-)meme selbst selv (90) Typological survey o f different uses o f adnom inal intensifiers: Chinese English French Danish a. Adnominal intensifier guo-wang ziji the king himself le rot (lui-)meme kongen selv b. Complex reflexive 0 ziji 0 himself soi-meme/lui-meme sig selv c. Doppelgftnger-marker 0 ziji 0 himself soi-meme sig selv d. Intensified pronoun [ta] (ta) zJji [him] himself [lui] lui-meme ham selv e. Exclusive adv. inten. ziji himself lui-meme selv The m ain advantage o f the present proposal is that it enables a unified account all uses o f the intensifier elem ent ziji ‘self-self. In contrast, m ost existing analyses o f Chinese are forced to assum e ziji to be lexically am biguous betw een (i) a sim ple reflexive anaphor, (ii) an adnominal intensifier, and (iii) a locally free logophor. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 407 Chapter 7 The element selv ‘self in nominalizations 7.1 Introduction In this chapter the word-internal uses o f the elem ent selv ‘s e lf will be discussed. It has been argued that selv ‘s e lf functions as a reflexive in nom inalizations such as selvrespekt ‘self- re sp e c f. That selv ‘s e lf should function as a reflexive in nom inalizations (rather than the simple reflexive sig ‘R EFL’, e.g. *sig-respekt ‘R E F L -respecf, has been taken as evidence in favor o f analyzing selv ‘s e lf as a reflexivizing elem ent (cf. Reinhart and Reuland (1993)) and could thus be construed as an argum ent against the analysis defended here, nam ely that binding o f reflexives, e.g. sig, and the distribution o f intensifiers, e.g. selv ‘s e lf, are controlled by separate m odules o f the grammar. In this chapter these issues will be discussed and an alternative analysis will be proposed in w hich the elem ent selv ‘s e lf is consistently assum ed to be an intensifier even in its word-internal uses. In the rem ainder o f this introductory section the main facts concerning nom inalizations containing the elem ent selv ‘s e lf will be introduced. Section 7.2 will contain m ore detailed description o f the different functions o f the elem ent selv ‘s e lf in nom inalizations. Section 7.3 offers a cross-linguistic perspective on the use o f reflexives and intensifiers in nom inalizations and also takes a brief look at the diachronic aspects o f word form ation processes and lexicalization. Finally, in section 7.4, the exact status o f word- internal uses o f selv ‘s e lf will be discussed (i.e. is it an intensifier or a reflexive, or som ething else?) and a possible analysis o f selv ‘selt^ in nom inalizations will be outlined. For expository reasons an alphabetically ordered list o f all the m orphologically com plex words containing the elem ent selv ‘s e lf occurring in one o f the m ost com prehensive Danish-English dictionaries has been placed in A ppendix 1 . R e p r o d u c e d with p e r m is s io n o f t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 40 8 The elem ent selv ‘s e lf occurs in a num ber o f derived nouns, see (1). ( 1 ) a. selvrespekt ‘self-respecf b. selvanalyse ‘self-analysis’ c. selviagttagelse ‘self-observation’ d. selvbedrag ‘self-deception’ e. selvforsvar ‘self-defense’ The nouns listed in (1), are all derived from verbal stems, see (2). (2) a. respektere ‘to respect’ => respekt ‘respect’ b. analysere ‘to analyze’ => analyse ‘analysis’ c. iagttage ‘to observe’ => iagttagelse ‘observation’ d. bedrage ‘to deceive’ => bedrag ‘deceit’ e. forsvare ‘to defend’ => forsvar ‘defense’ The verbal m eanings o f the words in (2) are preserved in their respective nom inalizations in (1). The elem ent selv ‘s e lf occurring at the first part o f these nom inalizations could thus potentially be interpreted as receiving the theta-role associated with the direct object o f the verbs in (2). In other words, the m eanings o f the nom inalizations in (1) correspond roughly to the m eanings o f the corresponding reflexive clauses in (3). As also shown in (3), the external theta-role o f the verbs form ing the base o f the nom inalizations in (1) can be expressed by an optional possessor phrase. (3) a. Peter bedrager sig selv. « Peters selvbedrag b. Peter respekterer sig selv. ~ Peters selvrespekt c. Peter iagttager sig selv. ~ Peters selviagttagelse d. Peter analyserer sig selv. ~ Peters selvanalyse e. Peter forsvarer sig selv. ~ Peters selvforsvar ‘Peter respects/analyzes/etc. h im se lf’ ‘Peter’s self-respect/etc..’ A t first glance, the close correspondence betw een clauses and nom inalizations illustrated by the exam ples in (3) may appear to indicate that the elem ent selv ‘s e lf functions as a reflexive elem ent inside nom inalizations in the sam e way that the reflexive pronoun sig (selv) functions as a reflexive elem ent in full clauses. H ow ever, as w ill be proposed in section 7.4, one may also analyze the selv elem ent in nom inalizations to be an intensifier adjoining to a 0 reflexive. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 40 9 7.2 Different uses of the morpheme selv ‘se lf in derived nouns and adjectives In this section, different uses o f the m orphem e selv ‘s e lf in derived nouns and adjectives will be described. Two m ain uses can be distinguished: (i) w hat appears to be “reflexives” uses, w hich we will argue are really adnom inal intensifiers, discussed in section 7.2.1, and (ii) “exclusive” adverbial uses, discussed in section 7.2.2. Finally Section 7.2.3 contains a brief discussion o f idiom atic expressions. 7.2.1 Word-internal selv ‘se lf appears to behave as a “reflexive”, yet it really is an adnominal intensifler As m entioned in section 7.1, the close correspondence between clauses and nom inalizations illustrated by the exam ples in (3) may at first glance appear to suggest that the elem ent selv ‘s e lf functions as a reflexive elem ent inside nom inalizations in the sam e way that the reflexive pronoun sig (selv) functions as a reflexive elem ent in full clauses. A lm ost all o f the nom inalizations in w hich the elem ent selv ‘s e lf appears to function as a reflexive are derived from anti-reflexive verbs, see (4)-(6). (4) (5) (6) a. b. c. a. b. c. a. b. c. selvbedrag selviagttagelse selvkritik bedrage iagttage kritisere ‘self-deception’ ‘self-observation’ ‘self-criticism ’ ‘to deceive’ ‘to observe’ ‘to criticize’ Peter bedrager *sig/sig selv/Marie. Peter deceives REFL/REFL self/M arie ‘Peter deceives him self/M arie.’ P eter iagttager ksig/sig selv/M arie. Peter observes REFL/REFL self/M arie ‘Peter observes him self/M arie.’ Peter kritiserer #sig/sig selv/Marie. Peter criticizes REFL/REFL self/M arie ‘Peter criticizes him self/M arie.’ (anti-reflexive) (hidden neutral) (hidden neutral) (anti-reflexive) (h id d en neutral) (hidden neutral) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 1 0 The nom inalizations in (4) are based on the verbs in (5) w hose anti-reflexive or hidden neutral nature is illustrated by the exam ples in (6). The anti-reflexive m eaning o f these predicates generates a presupposition o f representational non-identity o f their argum ents. Therefore adnom inal intensification o f the sim ple reflexive is necessary to m ark the reflexive as representationally non-identical to its antecedent'. In other words, though the instances o f selv ‘s e lf in (4) appear to function as reflexives, they are really adnom inal intensifiers intensifying a word-internal 0-reflexive (see the analysis proposed in section 7.4 below). There are, how ever, also a few o f exam ples o f such nom inalizations derived from neutral verbs, see (7)-(9). (7) a. selvforsvar ‘self-defense’ b. selvforsorgende ‘self-supporting, independent’ (8) a. forsvare ‘defend’ (neutral, transitive verb) b. forsorge ‘support’ (neutral, transitive verb) (9) Peter forsvarer/forsorger sig/sig selv/Marie. Peter defends/supports REFL/REFL self/M arie ‘Peter defends/supports him self/M arie.’ The nom inalization in (7a) is based on the neutral verbs in (8) whose neutral nature is illustrated by the exam ples in (9). As expected there are no ^e/f-nom inalizations based on “groom ing” verbs, e.g. *selwaskning ‘self-w ashing’^. See analysis of intensified reflexives in chapter 3. ^ The only apparent exception is adjectives like selvrensende ‘self-cleaning/automatically cleaning’. In this derived adjective, however, the element selv seem to derive from the adverbial adjunct a f sig selv ‘by itself/automatically’, see (i), rather than from a reflexive pronoun, see (ii). (i) Ovnen kan rense sig a f sig selv. oven-the can clean-INF REFL of REFL self ‘The oven can clean itself automatically/by itself.’ (ii) Ovnen rensede sig (selv). oven-the clean-PAST REFL self ‘The oven cleaned itself.’ R e p r o d u c e d with p e r m is s io n of t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 41 There are, however, som ew hat unexpectedly, a couple o f nom inalizations based on inherently reflexive verbs, in w hich the elem ent selv ‘s e lf seems to play the role o f reflexive, see(10)-(12). (10) a. selvbesindelse ‘collectedness, self-com m union; self-reflection’ b. selvudvikling ‘self-developm ent; spontaneous evolution (medical term inology)’ (11) a. besinde sig ‘collect o n e se lf (inherently reflexive) b. udvikle sig ‘develop’ (inherently reflexive/transitive) (12) Peter besinder sig/*sig selv/*Marie. Peter collects REFL/REFL self/M arie ‘Peter collects him self/M arie.’ The nom inalizations in (10) are based on the inherently reflexive verbs in (11) whose inherently reflexive nature is illustrated by the exam ples in (12). The fact that an inherently reflexive verb like besinde sig ‘collect o n e se lf is able to form nom inalizations w itb selv- is rather surprising. As discussed in chapter 4, there is am ple evidence in favor o f assum ing the sig occurring with (m ost) inherently reflexive verbs to be a non-them atic gram m atical m arker rather than a theta-role receiving nom inal argum ent. That is, besinde sig ‘collect o n e se lf is assum ed to have been reanalyzed as an intransitive verb unable to assign m ore than one theta- role. In contrast, the verbs in (5) and (8) are transitive verbs assigning two theta-roles and there is evidence that in nom inalizations based on these verbs, the elem ent selv-^ receives the internal theta-role, see (13) and (14). (13) a. Peters kritik a f sig selv. ‘Peter’s criticism o fR E F L s e lf b. Peters kritik a f Marie. ‘P eter’s criticism o f M a rie.’ (14) a. * Peters selvkritik a f sig selv. ‘Peter’s self-criticism o fR E F L s e lf b. * Peters selvkritik a f Marie. ‘Peter’s self-criticism o f M arie.’ Alternatively, as proposed in section 7.4, it is a 0-reflexive to which the intensifier selv is adjoined which receives the theta-role from the verbal stem. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 1 2 c. Peters selvkritik Peter’s self-criticism In (13a,b) both theta-roles o f the deverhal noun kritik ‘criticism ’ are assigned: one to the possessor phrase Peters ‘Peter’s ’ and the other to the nom inal expression in the prepositional com plem ent a f ‘o f Likewise, in (14c) both theta-roles o f the deverhal noun kritik ‘criticism ’ are assigned: one the possessor phrase Peters ‘Peter’s ’ and the other to the elem ent selv ‘s e lf. The exam ples in (14a,h) can he construed as evidence that the internal theta-role o f the verbal stem kritik- in the nom inalization selvkritik ‘self-criticism ’ has been assigned to the elem ent selv-. That is, the ungram m aticality o f (14a,h) can he explained as a consequence o f the DPs in the prepositional com plem ent a f ‘o f ’ in (14a,b) receiving no theta-role, since the tw o theta-roles o f the deverhal noun kritik ‘criticism ’ have already been assigned to other elements. However, if it is assum ed that the elem ent selv- in nom inalizations always receives a theta-role from the verbal root it is prefixed to, then how do we explain the presence o f selv in nom inalizations based on inherently reflexive verbs (see (10a,b))? As discussed below, there are even cases o f ^e/v-com pounds based on intransitive verbs. But in those cases, e.g. selvdad ‘self-dead (used to describe anim als which have died o f disease or other natural causes rather than having been killed)’, the elem ent selv is interpreted as an adverbial phrase m eaning ‘by itself, autom atically, spontaneously, etc’ rather than as an argument. N ote that it is possible to adopt a sim ilar adverbial analysis o f selv in the com pound selvbesindelse ‘collectedness, self- com m union; self-reflection’ (10a). By consistently analyzing self-nom inalizations based on inherently reflexive verbs (o f w hich there is no more than a handful out o f the total o f 235 com pounds listed in A ppendix I) as adverbial .se//^nominalizations we arrive at the generalization that reflexive self-nom inalizations only can be form ed on the basis o f anti reflexive and neutral verbal stems, see list o f self-nom inalization in A ppendix 1 . The term Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 413 ‘reflexive self-nom inalizations’ refers to deverbal nom inalizations in w hich selv appear to function as the internal argum ent (hut, as we argue, really is an adnom inal intensifier). In contrast, the term ‘adverbial self-nom inalizations’ refers to deverhal nom inalizations in which selv functions as an adverbial. 7.2.2 Word-internal selv ‘se lf behaving as an adverbial Some transitive verbs are not, or only m arginally, com patible with reflexive scenarios. In nom inalizations o f such verbs the selv elem ent tends to take on an adverbial m eaning ‘on one’s own, by oneself, w ithout help, all alone, etc.’, see (15). In other words, the occurrence o f the elem ent selv in the com plex noun in (15) should not be interpreted as the direct object o f the verbal elem ent bygge ‘build’ but rather as the m anner adverbial selv ‘by him self, etc.” occurring sentence-finally in the exam ple in (16). (15) selvhygger self-build-er ‘person who builds his house with his own hands.’ (16) Peter byggede huset (hell) selv. ( ‘exclusive’ adverbial use o f selv ‘s e lf) Peter built house-the all self ‘Peter built the house (all by) h im s e lf’ O ther exam ples o f adverbially used word-internal instances o f the elem ent selv are listed in (17) w hich contain adjectivizations/nom inalizations based on the intransitive verbs listed in (18). (17) a. selvdod self-dead (A djective used to describe anim als w hich have died o f disease, old age or other natural causes rather than having been killed) b. selvklatrende self-clim bing ‘(bot.) clim bing, creeping, trailing (plant)’ (18) a. d0 intr.‘d ie’ b. klatre intr. climb, clam ber R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 1 4 The adjectivizations/nom inalizations in (19) are based on the transitive verbs listed in (20). But unlike (15), the elem ent selv here m eans ‘by itself, autom atically’ rather than ‘all by itself, w ithout help.’ (19) a. selvspillende self-playing (adj. used to describe a piano w hich can play by itself/autom atically) b. selvbinder self-bind-er (adj. used to describe bay-raking m achine w hich binds the hay into bundles on its own/autom atically) c. selvhcevende self-rising ‘self-rising (fx flour)’ d. selvlukkende self-closing ‘self-closing (fx door)’ (20) a. spille (et instrument) ‘play (an instrum ent’ b. binde (Tip) ‘bind (hay)’ e. hceve (i) tr. ‘raise (fx one’s arms; o n e’s glass); lift, lift up’ (ii) intr. ‘swell, tumefy; (dough) rise ’ d. lukke ‘close (a door/gate/w indow /etc.)’ The m ain difference between (16), w hich allows exclusive adverbial selv, and (21a), which doesn’t, lies in the nature o f the predicate. Bygge ‘build’ in (16) is a causative transitive predicate. In this case the exclusive selv can associate w ith the agentive subject Peter. (21) a. *Dejen hcevede (belt) selv. (Com pare (21 a) and (16)) dough-the rose all self ‘The dough rose all by its e lf’ b. Dejen hcevede (helt) a f sig selv. dough-the rose all o fR E F L self ‘The dough rose all by itse lf Since the verb hceve ‘rise’ in (21) is unaccusative, it does not assign any A G EN T theta-role. Hence exclusive adverbial selv can not be used and the phrase a f sig selv ‘o f REFL s e lf is found instead. 7.2.3 P ro b lem atic cases It is unclear how to characterize the exact function/m eaning o f selv ‘s e lf in the more or less idiom atic adjectival/nom inal com pounds in (22)-(26). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 415 (22) a. selvklog adj. ‘self-opinionated; conceited’ b. klog adj. w ise (fx a w ise person); 2 prudent; 3 shrewd; 4 sensible; 5 intelligent, clever (23) a. selvretfcerdig adj. ‘self-righteous’ b. selvretfcerdighed n. ‘self-righteousness’ c. retfcerdig adj. just; righteous (24) a. selvmdl n. ‘(sports) own goal’ b. mat n. goal (fx in football) (25) a. selvsamme pron./ adj. ‘self-sam e; the very sam e’ b. samme adj. same (26) a. selvtredje adv. ‘w ith the tw o others’ b. tredie num. ‘third’ M ost o f these ‘problem atic’ cases are usually N O T deverbal nom inalizations but rather com pounds based on non-verbal nominal or adjectival roots w hich do not take argum ents, or express actions (hence cannot take either adnom inal, or exclusive adverbial selv). The adjectives klog ‘wise, intelligent’ and retfcerdig ‘just, righteous’ in (22-23) do not allow internal argum ents. It is thus im possible to suggest that in the .se//^eompounds in (22a) and (23a,b) the elem ent selv functions as an internal argum ent. The ^e/f-compounds in (24-26) seem even m ore idiosyncratic. In at least one case the adnom inal intensifier selv and its suppletive form in possessive constructions, egen ow n’, can occur in the sam e eom pound w ithout changing the m eaning, see (27). (27) egenrisiko = selvrisiko ow n-risk self-risk ‘own risk’ ‘own risk’ (in insurance term inology) The words in (27) are interesting since they m ay potentially be used as evidence that selv ’s e lf and egen ‘ow n’ are suppletive variants o f the sam e elem ent, an assum ption whieh is crucial to the analysis defended in this dissertation, see section 3.3.5 in chapter 3 for more discussion o f the possessive intensifier egen ‘ow n’. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 416 A nother set o f idiom atic uses o f selv is given in (28), w here it may be possible to argue that the selv in the com pound in (28a) corresponds to the com plex reflexive in the idiom atic phrase in (28c) though the exact analysis o f both (28a) and (28c) rem ains to be worked out. (28) a. selvsagt self-said adj. ‘self-evident, obvious, plain’ b. sige vt. say c. det siger sig selv it says REFL self ‘It is obvious/it goes w ithout saying/etc.’ 7.3 Cross-linguistic perspective on use of intensifiers in derived words In this section it will be argued that while word-external ly (syntactically) used adnom inal intensifiers have the same basic properties in m ost languages (i.e. as m arkers o f representational non-identity, markers o f contrastive focus, etc., see (29), w ord-internal ly used intensifiers seem to display a great deal m ore cross-linguistic variation w hich are m ainly due to the effects o f lexicalization and blocking. (29) Typological survey o f different uses o f adnominal intensifiers'*: Danish English French Chinse a. Adnom, inten. kongen selv the king him self le roi (lui-)meme guo-wang (ta) ziji b. Complex refl. sig selv [0 himselfl soi-meme/lui-meme [ta ziji/ 0 ziji] c. Prosodic inten. sig selv [HIM SELF/0 himself] [SOI/soi-memej [0 ziji] d. Doppelganger-mark sig selv [0 himself] [soi-meme] [0 ziji] e. Inten. pron./logophor ham selv [(him) himself] [(lui) lui-meme] [(ta) ziji] f Exclus.adv. int. selv himself lui-meme ziji g. Indus, adv. int. selv himself [lui-meme] ye h. Add. focus part. selv even meme lian__ye j. Nominal uses selv/jeg selt7*I *meme/moi . k. Deverbal compounds selv *himself/self *meme-/auto-/sui-/etc.. 1 . Positive adjective selve very/actual [ meme] ‘same (kind ot)’ - m. Superlative adjective selveste - - - 7.3.1. French The nom inal intensifier used to m ark non-guise-identity in French is -mem e, see (30). ' See chapter 2 for examples illustrating the different uses of intensifiers referred to in this table. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 1 7 (30) Use oi~m em e to m ark non-guise-identity. a. II travaille contre soi*(-meme). (anti-reflexive predicate) b. II travaille pour soi(-meme). (neutral predicate) Even though the adnom inal intensifier -m em e is used to m ark non-guise-identity in the same w ay is the adnom inal intensifier selv in D anish the Fr. m orphem e is not used as the first part o f any derived nouns and adjectives. It apparently only occurs in one com pound, i.e. the adverb memement ‘likewise; especially’. This clearly shows that the m orphem e used as adnom inal intensifier in a given language does not necessarily behave the same way in derived nouns and adjectives. In French the prefix auto- (derived from the G reek reflexive/intensifier autos ‘s e lf) is more productive than it is in the Germ anic languages and thus is able to take over some o f the functions o f-m em e in derived nouns and adjectives, see (31). (31) N om inalizations in auto- Danish: French: a. b. c. d. selvanklager self-accuser selvforsvar self-defense selvportrcEt self-portrait selvplageri self-torture auto-accusateur *meme-accusateur auto-accuser auto-defense '^meme-defencse auto-defense auto-portrait *meme-portrait auto-portrait auto-punition *meme-punition auto-punishm ent The exam ples in (32) and (33) illustrate that G erm anic self-nom inalizations often correspond to French nom inalizations in which the internal argum ent is relegated to a prepositional complement. (32) Paraphrases w ith PP + soi-meme Danish: French: a. selvbebrejdelse b. selvbegrcensning c. selvkritik g rief contre soi-meme limitation de soi-meme critique de soi-meme!autocritique R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 418 (33) Paraphrases with PP + soi(-meme) D anish: French: a. selvbehag contentement de soi, ??complaisance de soi auto-complaisance b. selvtilfreds content de soi [-meme] c. selvforsvar defense de soi-meme/autodefense The word-internal distribution o f soi and soi-meme in nom inalizations o f the types illustrated here seem to neatly follow the w ord-external distribution o f soi and soi-meme and o f lui and lui-meme w ith non-contrastive prepositional predicates. That is, anti-reflexive predicates require the com plex form s lui-meme/ soi-meme, see (30a) and (32), w hile neutral predicates allow both sim ple soi/lui and com plex lui-meme/ soi-meme, see (30b) and (33). Finally, notice that in neither French nor Danish is it possible to form com pounds with the reflexives se and sig, e.g. *se-punition, *sig-plageri. Indeed, it seems to be im possible to find ‘tru e’ reflexives used in com pounds in this way in any language. The absence o f nom inalizations with incorporated reflexive pronouns may be due to a constraint on w hat kind o f elem ents may undergo incorporation. That is, being DPs, reflexive pronouns cannot incorporate, while the m orphem e selv being an N is allow ed to incorporate. 7.3.2 Chinese In M odern M andarin Chinese the reflexive pronoun and the intensifier appear to have the sam e form, i.e. ziji |i |t 3 ‘s e lf , see (34). (34) a. ° Zhangsan hen ziji. Zhangsan hates self ‘Zhangsan hates him self.’ b. “ Zhangsan ziji hen Lisi. Zhangsan self hates Lisi ‘Zhangsan him self hates Lisi.’ R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 1 9 Like French -m em e, ziji g B ‘s e lf does not seem to be used in derived nouns and adjectives. Instead, the m orphem es zi i l ‘s e lf an d j i B ‘s e lf, w hich both have the basic m eaning ‘s e lf, are used independently^ see (35) and (37). In Classical Chinese both zi j = j ‘s e lf and j i B ‘s e lf could be used as independent words. The word zi g ‘s e lf was mostly used as a verbal prefix, and was therefore usually only used as local reflexive. Ji B, on the other hand, could be used both as local and as long-distance reflexive. In Classical C hinese the noun shen # ‘body’ was also used as em phatic reflexive/adnom inal intensifier'’. All three words now occur in the M odem Chinese counterparts o f English/D anish words containing the elem ent selv ‘s e lf, see (35)-(37). (35) zi g ‘s e lf + verb’ * a. zi-bei ‘self-abasem enf b. zi-qi SSf ‘self-deception’ c. zi-ai ilg ‘self-love’ d. zi-wei ‘self-defense’ (see (49a-b)) verb + shen # ‘self, body’ a. hu-shen ‘self-defense’ b. fang-shen fiJjM' ‘self-defense’ verb + j i B ‘s e lf a. zhi-ji ^PB ‘know o n eself b. zhi-ji ^ P B n. ‘bosom /intim ate friend, be intim ate/close’ 'N o te incidentally, that ziji [||c l ‘self-self is often wrongly assumed to be mono-morphemic, see chapter 6 for more discussion of the intensifier ziji ll| B ‘self-self. ^ As in wu shen (lit.) ‘I body’ = ‘1 m yself. ’ zi III‘self + verb is a frequently found type of xe/f-compound in Modern Chinese. The prefix zi llj‘self is only found in a few, idiomatic, constructions which are not based on deverbal nouns, e.g. zi-wo (lit. ‘self-me’) ‘self; oneself; ego’. *Note that verbal compounds containing z/j=i ‘self abound in Chinese (e.g. zi-sh a ^jjk ‘(lit.) self-kill, commit suicide’ but are almost non-existing in Danish, as well as the other Germanic languages. There is only one verbal compound containing selv ‘self in the entire list in Appendix 1 . R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 2 0 The form ation o f m any o f the nom inalizations now found in m odem Chinese, e.g. (35-37), occurred long tim e ago at a tim e w hen ziji g 3 as reflexive/intensifier had not yet been formed. It could therefore be argued that the existence o f such nom inalizations in the lexicon is now blocking the productive form ation o f com pounds containing ziji l i | 3 , e.g. *ziji-ai *[e) 3 S ‘self-love’, com pare with (35c). 7.3.3 Summary The above m ini typological survey o f the behavior o f the w ord-internal uses o f intensifiers shows that the elem ents occurring (w ord-externally) as reflexives/intensifiers in clausal constructions need not be identical to the ones occurring in same functions w ord-internally. Furtherm ore, even if the same elem ents are used there is no guaranty that they will behave the same way when used w ord-intem ally as when used in clausal constructions. The fact, that the elem ent selv ‘s e lf appear to behave as a reflexive word-internally does therefore not constitute any strong evidence in favor selv being a reflexivizing element. 7.4 Outline of analysis of selv ‘se lf in derived nouns and adjectives In this section it will be proposed that word-internal selv and word-external selv should be analyzed in the sam e way, i.e. either as an adnom inal intensifiers adjoining to a simple reflexive or as an adverbial phrase m odifying a predicate. The distribution o f the w ord-extem al adnom inal intensifier w ith difference verb types is given in (38). (38) a. N eutral predicates: sig / sig selv / DP b. Anti-reflexive: *sig / sig selv / DP c. Inherently reflexive: sig / *sigselv / DP R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 421 (39) N eutral predicates: a. Peter fo r svarer sig selv. b. Peter forsvarer sig. ‘Peter defends REFL / REFL self.’ A s stated in (38a) and illustrated in (39a,b) .se/v-adjunction to the sim ple reflexive sig is optional with neutral predicates like forsvare ‘defend’. Likewise in ‘reflexive’ nom inalizations based on neutral predicates the presences o f the elem ent selv is optional, see (40a,b). (40) N eutral predicates: a. [ [ 0 selvWforsvar]} ‘self-defense’ b- [ [0 ] [forsvar'W ‘defense (o f oneself unless otherwise specified)’ As show n in (40), we assum e nom inalizations o f neutral verbs like forsvar ‘ defense’ and selvforsvar ‘self-defense’ to contain a 0 reflexive w hich may or may not be intensified by the adnom inal intensifier selv. Even when the intensifier selv is not present the nom inalization still has a reflexive reading (see (41(i)) below) as its default reading. O nly when a non reflexive argum ent is overtly present (e.g. His defense o f Mary) is a non-reflexive reading possible. (41) The u s e football team had a problem with their defense. (i) their defense (o f them selves) N ow let’s turn to the anti-reflexive predicates. As stated in (38b) and illustrated in (42a,b) 5e/v-adjunction to the sim ple reflexive sig is m andatory with anti-reflexive predieates like kritisere ‘criticize’. (42) A nti-reflexive predicates: a. Peter kritiserer sig selv. b. Peter kritiserer *sig. ‘Peter criticizes REFL / REFL s e lf ’ Likewise in ‘reflexive’ nom inalizations based on anti-reflexive predicates the presence o f the elem ent selv is m andatory, see (43a,b). R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 2 2 (43) Anti-reflexive: a. [[0 selv][kritik]\ ‘self-criticism ’ b. *[ [0 ] [kritik]] ‘criticism (o f som ebody other than oneself unless otherwise specified)’ As show n in (43), in nom inalizations based on anti-reflexive verbs the 0 reflexive has to be intensified to get the reflexive reading. W hen the intensifier selv is not present the nom inalization cannot have a reflexive reading, see (44(i)). (44) a. The USC football team had a problem with their criticism. *(i) their criticism o f them selves “'‘(ii) their criticism o f som eone else/other team s b. The USC football team had a problem with their selfcriticism. “'‘(i) their criticism o f them selves *(ii) their criticism o f som eone else/other team s That is, the default reading o f nom inalizations o f anti-reflexive verbs is a non-reflexive reading, see (43b), and to obtain the reflexive reading the intensifier m ust be adjoined to the w ord-internal 0 reflexive. The fact that a few (only one or tw o) nom inalizations o f inherently reflexive verbs allow the optional presence o f selv, see (45a,b), constitutes a potential problem for this analysis. (45) Inherently reflexive: a. [ [0 selv] [besindelse]] self collectedness b. [ [0 ] [besindelse^ collectedness c. besindelse a f sig s e lf » selvbesindelse collectedness o f REFL self » ‘collectedness’ As indicated in (38c) inherently predicates do not allow the com plex reflexive. Consequently on e w o u ld ex p ect the n om in alization s o f su ch verbs to e x clu d e the p resen ce o f the adnom inal intensifier selv. A possible way out m ight be to assum e the selv in (45a) to be an adverbial m odifier rather than the adnom inal intensifier adjoined to a 0 reflexive, see (45c). R e p r o d u c e d with p e r m is s io n o t t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 2 3 A s already m entioned in section 7.3.1 above, in neither French nor Danish is it possible to form com pounds w ith the reflexives se and sig, e.g. *se-punition, *sig-plageri. Indeed, it seem s to be im possible to find ‘tru e’ reflexives used in com pounds in this way in any language, see (46). (46) G eneralization: True reflexives (e.g. Dan. sig, Fr. se, Ger. sich, etc.) do not occur in nom inalizations, e.g. Fr. *se-punition, Dan. *sig-plageri S ince re fle x iv e s c a n n o t be used in c o m p o u n d s, it h as been su g g ested th a t th e e le m e n t selv fu n ctio n s as a re fle x iv e w o rd -in tern ally . In th is sectio n , w e h av e o u tlin ed an altern ativ e a n a ly sis b ased on th e assu m p tio n th a t w o rd -in tern al re fle x iv e s are p h o n o lo g ic a lly u n realized e le m e n ts th a t m a y o r m ay n o t be in ten sified b y th e a d n o m in al in te n sifie r selv. A s su g g e ste d ab o v e, th is an aly sis m ak es it p o ssib le to unify th e ac c o u n t o f b o th w o rd -e x te rn a l an d w o rd -in te rn a l o c c u rre n ce s o f th e ad n o m in al in te n sifie r selv. In ad d itio n to its use as an ad n o m in al in ten sifier, th e e le m e n t selv can also be u sed as an ad v erb ial m o d ifier. A g ain , th e w o rd -in te rn a l uses m irro r th e w o rd -e x te rn a l uses, see (3 8 ) vs. (39), an d (4 1 c) vs. (4 3 b ). R e p r o d u c e d with p e r m is s io n o t t h e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 2 4 Chapter 8 Conclusion In this dissertation we have presented an account o f simple and com plex reflexives and pronouns based on the proposal that binding and intensification should be clearly separated. The tw o m odules overlap in the case o f com plex reflexives and pronouns which are here analyzed as adnom inally intensified forms o f their sim ple counterparts. That is, we proposed that the descriptive generalizations in (l)-(4 ) are best accounted for within a fram ew ork which separates binding and intensification in to the independent m odules sum m arized in (5) and (6) respectively. (1) Sim ple reflexives (e.g. Dan, sis): are bound by subject in local dom ain (=m inim al tensed clause) (chapter 3, section 3.2.4) (2) Sim ple pronouns (e.g. Dan, ham): are subject-free in local dom ain (=m inim al tensed clause) (chapter 3, section 3.4.1) (3) C om plex reflexives (e.g. s is selv) are found in the following contexts: a. anti-reflexive predicates (chapter 3, section 3.3.2.1) b. hidden neutral predicates (chapter 3, section 3.3.2.3) c. contrastive contexts (with neutral predicates) (chapter 3, section 3.3.2.5) d. doppelganger-effects (chapter 3, section 3.3.2.7) (i) statue-readings (section 3.3.2.7.1) (ii) qua-sentences (section 3.3.2.7.2) R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 425 (iii) strict (and sloppy) readings in V P ellipsis (section 3.3.2.7.3) e. stress-carrier (i) after unstressed prepositions (ex. (144), section 3.3.6.1) (ii) m eta-linguistic contexts (w ith inherently reflexive) (chapter 4, section 4.3.3, ex. (91)) (4) C om plex pronouns (e.g ham selv) are found in the following contexts: a. inherently contrastive predicates (chapter 3, section 3.4.2, ex. (167-8)) b. contrastive contexts (w ith neutral predicates) (section, 3.4.2, ex. (169) (5) B inding theorv (see chapter 31: a. Principle A, see (chap. 3, sect. 3.2.4, (25-28)) b. Principle B, (chap. 3, sect. 3.2.4, (25-28)) (6) Theorv o f intensification (see chapter 2 ): C ontrastiveness condition on adnom inal intensification: A nom inal expression (DP, pronoun, reflexive, etc.) can only be intensified adnom inally if it can be contrasted with other expressions in the context in w hich it is found, (chapter 2, ex. (2/62)) The syntactic binding principles in (5a) and (5b) account for the distribution o f reflexives and pronouns in (1,3) and (2,4) respectively. The pragm atic/sem antic principle o f contrastiveness in (6a) is the basis o f the account o f the distribution o f the intensifier selv in the contexts listed in (3) and (4). R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 2 6 The approach to binding and intensification proposed here has a num ber o f im portant consequences for linguistic theory in general, see (7), and for the analysis o f Danish in particular, see (8). (7) Consequences for binding theorv: a. intensification and binding are independent m odules o f the gram m ar b. binding is a purely syntactic phenom enon c. unified account o f possessive and argum ent reflexives d. doppelganger-effects (statue-readings, qua-sentences, stict reading in VP ellipsis, etc.) and anti-reflexivity (the triggering o f adnom inal intensification o f reflexives w ith certain predicates as well as in other sem antically/pragm atically anti-reflexive contexts (e.g. resultatives, ECM , and possessive constructions)) can be unified. (8) C onsequences for analvsis o f M odern D anish (and sim ilar languages): a. sig not anti-local b. sig selv is not m andatorily local (cf. his own). c. com plex reflexives, e.g. sig selv, are intensified reflexives. d. ham selv is not a logophor but an intensified object pronoun. e. han selv is not a special kind o f anaphor (cf. Bickerton 1986) but an in ten sified subject pronoun. f. selv ‘s e lf is not a reflexivizing particle w hich falls under binding theory (e.g. Reinhart and Reuland 1993), but an intensifier which falls under the module o f intensification. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 2 7 g. egen ‘ow n’ is the suppletive variant o f selv ‘s e lf. h. the possessive reflexive sin ‘PO SSR EFL’ is a m orphological variant o f sig ‘R E FL ’ and falls under the sam e principle A o f the binding theory (in contrast, Reinhart and Reuland (1993) have to assum e sin to be fundam entally different from sig, since in their analysis reflexives have to be argum ents o f predicates), In chapters 5 and 6 sim ilar approaches to English and M andarin Chinese were put to the test, see (9)-(l 1). (9) Consequences for analvsis o f M odern English: a. English has 0-reflexives (= Dan. sig) b. English x-self forms are not reflexives but A LW AYS intensifiers c. English have both sim ple and com plex reflexives, i.e. 0 vs. 0 him self d. The distribution o f sim ple (i.e. 0 ) and com plex reflexives (i.e. 0 himself) in M odern English m irrors that o f sim ple (i.e. sig) and com plex reflexives (i.e. sig selv) in Danish, see (3) d. Locally free him self \s not a logophor but an intensified object pronoun e. he him self is not a special kind o f anaphor (cf. Bickerton 1986) but an intensified subject pronoun (cf. M cKay 1991) f. -self is not a reflexivizing particle w hich falls under binding theory (e.g. R einhart and R euland 1993), but part o f co m p lex in ten sifiers, i.e. the x -self forms, which falls under the m odule o f intensification g. own is the suppletive variant o f the adnom inal intensifier him self R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 428 (10) C onsequences for the analvsis o f the evolution o f English reflexives and intensifiers: a. O ld English (OE) locally bound pronouns (e.g. he, washes him,) were replaced by 0-reflexives in Early M odern English (EM E) (e.g. he, washes 0 ) . b. The fact that the loss o f locally bound pronouns took place across the board explains the loss o f the pronoun in OE he rested him w hich turned into he rested 0 in EM E and M E. T hat is, Eng. 0-reflexive occurs w here Dan. has sim ple sig, e.g. han hvilte sig ‘he rested’. c. OE s e lf fused w ith its associate him and was reanalyzed as an adnom inal intensifier adjoined to a 0-reflexive, e.g. he washed 0 himself. d. A fter em erging as the new intensifier EM E x-self replaced the sim ple OE intensifier seolf'm all contexts (e.g. 0 him self Peter him self etc.) e. Since intensifier-adj unction to sim ple reflexives is m otivated by sem antic/pragm atic factors (e.g. predicate meaning, presuppositional context) rather than a need to disam biguate betw een disjoint reference and coreference, the fact that it also happens in the T ' and 2"‘ ‘ persons is no longer a mystery. (11) Conseq uences for analvsis o f M andarin C hinese: a. Chinese has 0-reftexives Dan. sig) b. [ij S ziji is not a reflexive but A LW A Y S an intensifier c. Chinese have both sim ple and com plex reflexives, i.e. 0 vs. 0 ziji d. the distribution o f sim ple (i.e. 0 ) and com plex reflexives (i.e. 0 ziji) in Chinese m irrors that o f sim ple (i.e. sig) and com plex reflexives (i.e. sig selv) in Danish, see (3) e. com plex reflexives, e.g. 0 ziji, are intensified reflexives R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 2 9 f. Locally free ziji is not a logophor but an intensified object pronoun g- fttHij B ta ziji in subject position is not a special kind o f anaphor but an intensified subject pronoun. h. Locally bound ta ziji is an intensified 0-reflexive, 0 ta ziji. i. B B ziji is not a reflexivizing particle w hich falls under binding theory (e.g. R einhart and Reuland 1993), but rather an intensifier w hich falls under the m odule o f intensification. One o f the m ain lessons to be learned from the analysis defended here is that the conventional w isdom o f traditional gram m ars cannot always be taken for granted. Just because Eng. x-self forms and Ch. ziji have traditionally been considered to be first and forem ost reflexive anaphors this does not have to be taken for a good-given truth. Traditional gram m arians were inclined to look for m atches to categories found in Latin grammar. Since Latin has overt reflexive pronouns (w hich differ m orphologically from object pronouns), and since M odern English x-self form s and Chinese ziji do occur in m any contexts w here Latin reflexives are found, it is not surprising that x -se/f forms were taken to be reflexive pronouns. However, as we have seen in chapter 5 and 6, there is a large body o f data which suggest that 0-reflexives are the true Eng. and Ch. counterparts o f Rom ance (and Latin) reflexives and that x-5e//"forms and ziji are really intensifiers. Finally, the present dissertation stresses the need for com prehensive accounts w hich take the interaction between several different m odules into account. The behavior o f elem ents like Dan. sig, selv, and sig selv, is extrem ely com plex exactly because they are found at the intersection o f m any different m odules, e.g. intensification, binding, focus, phonology, argum ent structure o f predicates, etc. A central them e o f this dissertation has been to show R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 43 0 that m any current accounts o f binding fail to recognize the im portance o f taking intensification into account when explaining the behavior o f com plex reflexives (e.g. Dan. sig selv) and pronouns (e.g. Dan, ham selv), see chapter 2-3. We have also shown that correct understanding o f the them atic status o f reflexive elem ents is a necessary prerequisite o f any binding theory, see chapter 4. Finally, we have shown that the role played by stressability and non-contrastive stress is crucial to prosodically m otivated instances o f com plex reflexives. N arrow accounts o f binding phenom ena that are lim ited to only to strictly binding related principles are thus likely to have correspondingly lim ited descriptive and explanatory adequacy. The best way to im prove our understanding o f both binding and intensification (as well as the other m odules m entioned here, i.e. focus, argum ent structure, prosody, etc.) lies in the developm ent o f m ulti-m odular m odels that are able to capture com plex m odular interaction. It is our hope that by developing an alternative com prehensive approach to binding and intensification, w hich differ radically from current m ainstream theories o f binding, the present dissertation constitutes a small step in this direction. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 3 1 References Abraham , W erner (1995) Diathesis: The M iddle, Particularly in W est-Germ anic. W hat Does Reflexivization have to do w ith V alency Reduction? In Abraham , W, T. Givon, and Sandra A. T hom pson (eds.) Discourse Grammar and Typology. Papers in Honor o f John W.M. Verhaar. Studies in Language C om panion Series Vol. 27, pp. 3-47. John Benjam ins Publishing Company. Am sterdam /Philadelphia. Abraham , W erner (2000) The structural and lexical space between reflexive binding and logophorics: Sundry paradigm s o f reflexives and anaphora. In Reflexives: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) V olum e 40, pp. 104-124. Am sterdam : John Benjam ins. Abush, Dorit (1989) “Reflexives, Reference Shifters, and A ttitudes”, in E.J. Fee and K. Hunt (eds.), W CCFL 8, Stanford Linguistics A ssociation, Stanford, pp. 1-13. Aissen, Judith (1982) “Valence and coreference.” In Paul H opper and Sandra A. Thom pson (eds.). Studies in Transitivity, 7-35 [Syntax and Semantics 15.] N ew York: Academ ic Press. Allan, R. & Holmes, P. & T. Lundskaer-Nielsen (1995) Danish: A Comprehensive Grammar. Routledge. London and N ew York. A nagnostopoulou, Elena, and M artin Everaert (1999) Tow ard a m ore com plete typology o f anaphoric expressions. Linguistic Inquiry 30:97-118. Anderson, S.R. (1972) “HOW TO G ET even”. Language 48.4:893-906. Anderson, S.R. (1986) “The typology o f anaphoric dependencies: Icelandic (and other) reflexives, in L. Hellan and K.K. Christensen (eds.). Topics in Scandinavian Syntax. (Studies in natural language and linguistic theory.) D ordrecht: D. Reidel, pp. 65-88. A nscom bre, J.C. (1973) M eme le roi de France est sage. U n essai de description sem antique. Communications 1973, t. 20, pp. 40-82. Aoun, Joseph (1985) A Grammar o f Anaphora. Linguistic Inquiry M onographs. The M IT Press, Cam bridge, M assachusetts. Ariel, M ira (1988) “Referring and accessibility”. Journal o f Linguistics 24:65-87. Ariel, M ira {\990) Accessingnoun-phrase antecedents. London: Croom Helm. Ariel, M ira (1994) “Interpreting anaphoric expressions: a cognitive versus a pragm atic approach”. J. Linguistics. 30:3-42. Baker, C.-L. (1995). Contrast, discourse prom inence, and intensification, with special reference to locally-free reflexives in British English. Language 71.1. 63-101. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 3 2 Barker, Chris (1995) Possessive descriptions. D issertations in Lunguistics. CSLI Publications: Stanford, California. Barker, Stephen (1991) "''Even, still and counterfactuals” . Linguistics and Philosophy 14:1-38 Barss, A ndrew and H ow ard Lasnik (1986) “A N ote on A naphora and Double O bjects”, Linguistic Inquiry 17:347-354. Battistella, Edwin and Yonghui X u (1990) Rem arks on the reflexive in Chinese. Linguistics 28:205-240. Beas, O m ar (2001) “The D erivation o f M eaning and Binding Properties o f SA M E-elem ents in Spanish: A naphors and N on-A naphors”. MS. USC. Beas, O m ar (2002a) “On the Role o f Clitics in Spanish Reflexive D oubling”. MS. USC. Beas, O m ar (2002b) “On the Role o f Clitics in Spanish Reflexive D oubling” . H and-out for Student W orkshop, April 29, 2002. USC. Belletti, A driana and Luigi Rizzi (1988) “Psych-Verbs and theta-Theory”, N L L T 6:291-352. Benveniste, Em ile (1966) Problemes de linguistique generale. Paris: Gallim ard. Bickerton, Derek (1987) He him self; Anaphor, pronoun, or ...? Linguistic Inquiry 18, 345-48. Bolinger, D w ight L. (1961) “Contrastive A ccent and Contrastive Stress”. Language 37.1:83- 96. Borkovec, V era Zanda (1973) Grammatical and Stylistic Uses o f Certain Reflexive Forms o f Verbs in Czech and in Russian. G eorgetow n University Ph.D. Brinton (1995) “N on-anaphoric reflexives in free indirect style” . In Dieter Stein and Susan W right (eds.). Subjectivity and Subjectivization in Language. Cam bridge: Cam bridge University Press, 193-194 Brown, Karen Leigh (1996) “Reflexivity and B elief De Se”. In Jerry Seligm ann and Dag W esterstahl (eds.) Logic, Language and Computation, pp. 109-120. Stanford University. Biihring, Daniel (1997) The Meaning o f Topic and Focus. The 59"' Street Bridge Accent. Routledge Studies in Germ an Linguistics. Routledge: London and N ew York. Burzio, Luigi (1991) The m orphological basis o f anaphora. Journal o f Linguistics 27:81-105. Burzio, Luigi (19??) “On the non-existence o f disjoint reference principles”. 3-27. Burzio (1996a) “The Role o f the antecedent in anaphoric relations”. In Robert Freidin (ed.). Current Issues in Comparative Grammar. D ordrecht: Kluwer, 1-45. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 3 3 Burzio (1996b) “A naphora and soft constraints”. Proceedings o f the Workshop on Optimality in Syntax “ Is the Best Good Enough? ”, M IT W orking Papers in Linguistics. Canpado, M aria (1999) “Exceptional binding w ith psych verbs?”. Z /(1999:133-143). Cantrall, W illiam R. (1973) “W hy W ould I R elate Own, Em phatic Reflexives, and Intensive Pronouns, M y Own S e lf’. In Papers form the ninth regional m eeting, Chicago Linguistic Society, April 13-15, 1973. Chicago Linguistic Society. Cantrall, W illiam R. (1974) Viewpoint, Reflexives, and the Nature o f Noun Phrases. The Hague: M outon. Carroll, S. (1986a) “O n non-anaphor reflexives”. Revue quehecoise de linguistque 15:135-66. Carroll, S. (1986b) “Reflexives and the dependence relation ‘R ’”. The Canadian Journal o f Linguistics 1:1-43. Chafe, W allace (1976) “Givenness, Contrastiveness, D efiniteness, Subjects and Topics”. In Charles N. Li (ed.) Subject and Topic. N ew York. A cadem ic Press. Chierchia, (1989) A naphora and A ttitudes De Se. In R. Bartsch, J. van Benthem, and P. van Em de Boas, eds.. Semantics and Contextual Expression: 1-31. Dordrecht: Foris. Chom sky, N. (1980) On binding. Linguisticlnquiry 11:1-46. Chom sky, N. (1981) Lectures on government and binding. Dordrecht: Foris. Chom sky, N. (1986) Knowledge o f language: Its nature, origin, and use. N ew York: Praeger. Chom sky, N. (1995) The Minimalist Program. Com bridge, Mass.: M IT Press. Chomsky, N. (2000) M inim alist inquiries: The fram ework. In Step by step: Essays on minimalist syntax in honor o f Howard Lasnik, ed. Roger M artin, D avid M ichaels, and Juan Uriagereka, 89-155. Cam bridge, M ass.: M IT Press [Also available as M IT Occasional Papers in Linguistics 15, M ITW PL, D epartm ent o f Linguistics and Philosophy, MIT, Cam bridge, Mass.] Cinque, G uilelm o (1993) A null theory o f phrase and com pound stress. Linguistic Inquiry 24(2):237-97. C lem en ts, G eorge, N . (1 9 7 5 ) T h e lo g o p h o ric p m ou n in E w e: Its role in d iscou rse. Journal o f West African Languages 10, 141-77. Cole, Peter, G abriella Hermon, and Li-M ay Sung (1990) “Principles and param eters o f long distance reflexives”. Linguistic Inquiry 21:1-22. Cole, Peter and Li-M ay Sung (1994) “H ead M ovem ent and Long-D istance Reflexives”. Linguistic Inquiry 25:355-387. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 3 4 Cole, Peter, Chengchi W ang (1996) “A ntecedents and Blockers o f Long-D istance Reflexives: The Case o f Chinese ZijT. Linguistic Inquiry 27.3:357-390. Dacygier, Barbara (1997) Reflexive m arkers in Polish: Parcipiants, M etaphors, and Constructions. In Verspoor, M arjolijn, Kee Dong Lee and Eve Sw eetser (eds) Lexical and syntactical constructions and the construction o f meaning. Prooceedings o f the bi-annual ICLA m eeting in A lbuquerque, July 1995. A m sterdam studies in the theory and history o f linguistic science. V olum e 150,pp. 311-327. Dirven, R. (1973) “Em phatic and Reflexive in English and D utch”. Leuvense Bijdragen, Tijdschrift voor G erm aanse Filologie 63:285-299. Eekardt, Regine (1999) ‘Focus with N om inal Q uantifiers’, in P. Bosch and R. van der Sandt (eds), Focus & Natural Language Processing, vols. 1-3, Heidelberg: IBM, 281-290. Eekardt, Regine (2000). Reanalysing “ selbst”, Handout, Sinn und B edeutung V, Am sterdam , December, 2000. Eekardt, Regine (2001). Reanalysing “ selbst”, MS. U niversity o f Konstanz.. Edmondson, J.A. & Plank, F. (1978). G reat expectations: an intensive self analysis. Linguistics and Philosophy 2. 373-413. Erteschik-Shir, N. (1973) O n the nature o f island constraints. Doctoral dissertation, MIT. Erteschik-Shir, N. (1981) “On extraction from noun phrases (picture noun phrases). Theory o f Markedness in Generative Grammar. Proceedings o f the GLOW conference. Pisa: Seuola N orm ale Superiore, pp. 147-69. Evans, Gareth (1980) Pronouns. Linguistic Inquiry 11:337-362. Everaert, M artin (1986) The syntax o f reflexivization. Dordrecht: Foris. Everaert, M artin (1991) “Contextual determ ination o f the anaphor/pronom inal distinction”. In J. K oster & E. Reuland (eds.) Long-distance Anaphora, Cam bridge: Cam bridge University Press. Everaert, M artin (2000) Types o f anaphoric expressions: Reflexives and reciprocals. . In Reciprocals: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl. Typological Studies in L anguage (T S L ) V o lu m e 4 1 , 6 3 -8 3 . A m sterdam : John B enjam in s. Faarlund, Jan Terje, Svein Lie, and Kjell Ivar V annebo (1997) Norsk Referansegrammatik. U niversitetsforlaget. Oslo. Faltz, Leonard M. (1977) Reflexivization: A study in universal syntax. Doctoral dissertation. U niversity o f C alifornia at Berkeley. [D istributed by University M icrofilm s International, Ann Arbor, M ich., and London.] R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 435 Faltz, Leonard M. (1985) Reflexivization: A study in universal syntax. N ew York: Garland. Fauconnier, Gilles (1975) “Pragm atic Scales and Logical Structure”. Linguistic Inquiry Vol.VI, N um ber 3:353-375. Ferro, Lisa (1993) “O n ‘s e lf as a focus m arker”. ESCOL ’92: Proceedings o f the N inth Eastern States C onference on Linguistics. Ithaca, NY: Cornell University, 68-79. Fiengo, Robert, and Jam es Higginbotham (1981) “O pacity in N P ”. Linguistic Analysis 7.4:395421. Fiengo, Robert, and R obert M ay (1994) Indices and identity. Cam bridge, M ass.: M IT Press. Fowler, G eorge (1993) “A Syntactic A ccount o f Derivational -s ja in R ussian.” American Contributions to the Eleventh Congress o f Slavists. Eds. Robert A. M aguire and Alan Tim berlake. Colum bus, Ohio: Slavica. 270-284. Frajsyngier, Zygm unt (2000) Dom ains o f point o f view and coreferentiality: System interaction approach to the study o f reflexives. In Reflexives: Forms and functions, ed. Frajzyngier, Zygm unt and Traci Curl. Typological Studies in Language (TSL) Volum e 40, 104-124. Am sterdam : John Benjam ins. Frajzyngier, Zygm unt and Traci Curl (ed.) (2000) Reflexives: Forms and functions. Typological Studies in Language (TSL) Volume 40. Am sterdam : John Benjamins. Frajzyngier, Zygm unt and Traci Curl (ed.) (2000) Reciprocals: Forms and functions. Typological Studies in Language (TSL) Volum e 41. Am sterdam : John Benjam ins. Fraser, Bruce (1971) “An Analysis o f “even” in English” . In Fillmore, Charles J. and D. Terence Langendoen (eds) Studies in Linguistic Semantics. Frazier, Lyn (1999) On sentence interpretation. Dordrecht: Kluwer. Ganjavi, Shadi (2002) “Anaphors in Persian” . M.S. USC. Gelderen, Elly van (2000) Bound pronouns and non-local anaphors: The case o f Earlier English. In Reflexives: Forms and functions. Typological Studies in Language (TSL) Volume 40. Am sterdam : John Benjamins. G eniusiene, Em m a (1987) The Typology o f Reflexives. Berlin: M outon de Gruyter. Giorgi, A lessandra (1984) “Tow ard a theory o f long distance anaphors. A GB approach.” The Linguistic Review 3:307-361. Grice, H. P. (1975) Logic and Conversation. M organ and Cole (eds) Syntax and Semantics, Vol. 3, Speech Acts. A cadem ic Press, 1975. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 3 6 G rim shaw, Jane (1982) “On the lexical representation o f Rom ance reflexive clitics”. In Joan Bresnan (ed.), The Mental Representation o f Grammatical Relations. Cam bridge, Mass.; The M IT Press, 87-148. Grim shaw, Jane {\990) Argument Structure. M IT Press, Cam bridge, MA. Grodzinsky, Yosef, and Tanya Reinhart (1993) The innateness o f binding and coreference. Linguistic Inquiry 24:69-101. Gussenhoven, Carlos (1984) On the Grammar and Semantics o f Sentence Accents, Dordrecht: Foris. Hagege, Claude (1974) Les Pronom s Logophoriques, Bulletin de la Societe de Linguistique de Paris 69, 287-310. Haiman, John (1995) Gram m atical Signs o f the D ivided Self. A Study o f Language and Culture. In A braham , W, T. Givon, and Sandra A. Thom pson (eds.) Discourse Grammar and Typology. Papers in Honor o f John W.M. Verhaar. Studies in Language C om panion Series Vol. 27, pp. 213-234. John Benjam ins Publishing Com pany. Am sterdam /Philadelphia. Hale, K enneth & Samuel Jay K eyser (1993) “On Argum ent Structure and the Lexical Expression o f Syntactic R elations”, in K. Hale & S. J.K eyser (eds.) The View from Building 20, Cam bridge: M IT Press. Halle, M orris (1973) “Prolegom ena to a Theory o f W ord Form ation”. Linguistic Inquiry 4 .1:3- 16. Hansen, Aage (1967) Moderne Dansk. / - / / / . G rafisk Forlag: Kobenhavn. Heim, Irene (19??) “O n the Projection Problem for Presuppositions”. Proceedings o f the Second West Coast Conference on Formal Linguistics. Heim, Irene (1993) A naphora and sem antic interpretation: A reinterpretation o f R einhart’s approach. SfS report 07-93. U niversity o f Tubingen. [Reprinted in M IT w orking papers in linguistics 25: The interpretive tract, 205-246. M ITW PL, Departm ent o f Linguistics and Philosophy, M IT, Cam bridge, M ass.] Heim, Irene (1993) “Puzzling reflexive pronouns in de se reports.” Hand-out, Bielefeld conference M arch 1994. H eim , Irene, and A n g elik a K ratzer (1 9 9 8 ) Semantics in Generative Grammar. B la ck w ell P ublishers. Helke, M ichael (1971) The grammar o f English reflexives. Unpublished dissertation. MIT, Cam bridge Mass. Helke, M ichael (1979) The Grammar o f English Reflexives. N ew York: G arland Publications. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 3 7 Hellan, Lars (1986) On anaphora and predication in N orw egian. In L. Hellan and K. Koch Christensen (eds.), Topics in Scandinavian Syntax, 103-124. Hellan, Lars (1988). Anaphora in Norwegian and the theory o f grammar. Dordrecht: Foris. Hellan, Lars (1991) C ontainm ent and connectedness anaphors. In Long-distance anaphora, eg. Jan K oster and Eric Reuland, 27-49. Cam bridge: Cam bridge U niversity Press. Hellan, Lars, and K. Koch Christensen (eds.) (1986) Topics in Scandinavian Syntax, Studies in N atural Language and Linguistic Theory. D.Reidel Publishing Company. Dordrecht: Kluwer. Hestvik, Arild (1991) “Subjectless B inding D om ains”, Natural Language and Linguistic Theory 9:455-496. Hestvik, Arild, and W illiam Philip (1997) Reflexivity, anti-subject orientation and language acquisition. In NETS 27, 171-185. GLSA, U niversity o f M assachusetts, Amherst. Higginbotham , J. (1985) On sem antics. Linguistic Inquiry 16.4. 547-593. Higginbotham , Jam es (1980) “Reciprocal interpretation”. Journal o f Linguistic Research 1:97- 117. Hopper, Paul J. and Sandra Thom pson (1980) “Transitivity in gram m ar and discourse.” Language 56:251-99. Horn, L. R. (1969) A presuppositional analysis o f only and even. Papers from the fifth regional meeting Chicago Linguistic Society. 98-107. Horn, Laurence R. (1998) “Presupposition and Im plicature”. In ???? (eds.) Handbook o f Semantics. Blackwell. Huang, C.T. Jam es (????) Distributivity and Reflexivity. In ????? . Huang, C.T. Jam es & C.S. Luther Liu (2001) Logophoricity, attitudes and ziji at the interface. In Syntax and semantics. Volum e 33. Huang, Yan (2000) Anaphora. A Cross-linguistic Study. Oxford Studies in Typology and Linguistic Theory. O xford U niversity Press. Hyman, Larry M. and Bernard Com rie (1981) “Logophoric Refence in G okana.” Journal o f African Languages and Linguistics 3:19-37. latridou, Sabine (1986) A n A naphor N ot B ound in Its G overning Category. Linguistic Inquiry 17, pp. 766-772. latridou, Sabine (1988) “Clitics, anaphors and noun phrase interpretation”. Linguistic Inquiry 19: 698-703. R e p r o d u c e d with p e r m is s io n o f th e c o p y rig h t o w n e r. F u r th e r r e p ro d u c tio n p ro h ib ite d w ith o u t p e r m is s io n . 4 3 8 Israeli, A lina (1997) Semantics and Pragmatics o f the “ Reflexive” Verbs in Russian. Verlag Otto Sagner. Miinchen. Jackendoff, R.S. (1972) Semantic Interpretation in Generative Grammar, Cam bridge: Cam bridge University Press. Jackendoff, R.S. (1987) “The Status o f Them atic Relations in Linguistic Theory”, Linguistic Inquiry 18:369-411. Jackendoff, R.S. (1992) M me Tussaud m eets the binding theory. Natural Language & Linguistic Theory 10: 1-33. Jaeggli, O svaldo A. (1986) “Passive”. L I 17.4:587-622. Jakubow icz, C. (1992) "Rig en D anois:syntaxe et acquisition”. In H .G .O benauer and A. Zribi- Hertz (eds.) Theories du Liage et structue de la phrase. Presses Universitaires de Vincennes. Jakubow icz, C. (1994a) “O n the m orphological specification o f reflexives: im plications for acquisition.” In VETS'24:205-219, M erce G onzalez (eds), Univ. o f M assachusetts Amherst. Jakubow icz, C. (1994b) “Reflexives in French and Danish: M orphology, Syntax, and A cquisition”. Janda, Laura A. (1993) “Cognitive Linguistics as a Continuation o f the Jakobsonian Tradition: the Sem antics o f R ussian and Czech Reflexives.” American Contributions to the Eleventh Congress o f Slavists. Eds Rober A. M aguire and A lan Tim berlake. Colum bus, Ohio: Slavica. 310-319. Jayaseelan, K.A. (1997) Anaphors and pronouns. Studia Linguistica 51:186-234. Joseph, Brian D. (1979) “On the agreem ent o f reflexive form s in English.” Linguistics 17:519- 523. Kardela, Henryk (19xx) “Target: Emphatics. A note on governm ent, binding and case assignm ent in Polish. Societas Linguistica Europaea. M outon Publishers, The Hague. K arttunen, L. & Peters S. (1979) Conventional Im plicature Syntax and Semantics 11, 1-56. Keenan, Edward (1971) Nam es, quantifiers and a solution to the sloppy identity problem. Papers in Linguistics 4.2. Keenan, Edward (1976) “Tow ards a Universal D efinition o f “Subject”” . In Charles N . Li (ed) Subject and Topic (1976). A cadem ic Press, N.Y. Keenan, Edw ard (1988) O n sem antics and the binding theory. In Explaining language universals, ed. John Hawkins, 104-155. Oxford: Blackwell. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 3 9 Keenan, E. (1994) Creating Anaphors: An Historical Study o f the English R eflexive Pronouns. MS. UCLA. Keenan, E. (1998) The Historical Creation o f Anaphors in English. MS. UCLA. Keenan, (2000) A historical explanation o f English binding theory. Paper presented at SHELL 1, UCLA, M ay 27. Keenan, (2000) An Historical Explanation o f Some Binding Theoretic Facts in English. MS. UCLA, N ovem ber 12, 2000. Keenan, E, (2001) Explaining the Creation o f Reflexive Pronouns in English. Handout. UCLA. Keenan, E, (2001) Explaining the Creation o f Reflexive Pronouns in English. MS. UCLA, M ay 2001. Kemmer, Suzanne (1993) The Middle Voice. A m sterdam and Philadelphia: John Benjamins. Kemmer, Suzanne (1994) M iddle Voice, Transitivity and the Elaboration o f Events. In Fox, Barbara, and Paul J. H opper (eds.) (1994) Voice: Form and Function. John Benjamins. Am sterdam Philadelphia, pp. 179-230. Kemmer, Suzanne (1995) “Em phatic and reflexive -self: expectations, view point and subjectivity”. In Dieter Stein and Susan W right (eds.), Subjectivity and Subjectivization in Language. Cam bridge: Cam bridge U niversity Press, 55-82. Kibrik, Aleksandr. A., and Ekaterina B ogdanova (1995) “Sam kak operator korrektsii ozidanij adresata”. [Russian sam as operator o f correction o f hearer’s expectations]. V oprosy jazykoznanija 3:4-47. Klaim an, M.H. (1992) “M iddle verbs, reflexive m iddle constructions, and m iddle voice”. Studies in Language 16:1, 35-62. Klenin, Emily (1975) “The Pronoun sebja. Particle sebe, and A ffix -s ja .” Slavic and East European Journal 19.2:188-199. Klenin, Em ily (1980) “Sentential and D iscourse Prom inence: Tbe case o f the Em phatic Pronoun .” Russian Linguistics 4:269-280. Koenig, Jean-Pierre (1998) “French body-parts and the sem antics o f binding”. Natural Language and Linguistic Theory ????, pp. 219-265. Konig, Ekkehard (1989). O n the Historical D evelopm ent o f Focus Particles. In Weydt, H. (ed) Sprechen mit Partikeln. W later de Gruyter. Berlin. N ew York. Konig, Ekkehard (1991). The m eaning o f focus particles: a com parative approach. The Linguistic Review 3. 307-361. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 0 Konig, Ekkehard (1991). The meaning o f focus particles: a comparative perspective. Routledge. London and N ew York.. Konig, Ekkehard (1993) Distribution und Bedeutung von R eflexivpronom ina im Englishen. V ersuch einer historischen Erklarung. Lecture, Free U niversity o f Berlin. Konig, Ekkehard (1996a) Tow ards a Typology o f Intensifiers (Em phatic Reflexives), MS. FU Berlin. Konig, Ekkehard (1997a) “From expressions for body parts to reflexive anaphors: Sem antic change in the developm ent o f intensifiers”. In Gerald F. Carr et al. (eds.). Interdigitations: Essays fo r Irmgard Frankfurt: Lang. Konig, Ekkehard (1997b) “Tow ards a typology o f intensifiers”. XVIth International Congress o f Linguists (Paris, July 1997), Proceedings. Konig, Ekkehard (1998) Intensifiers and reflexive pronouns. MS. Presented at USC. Konig, Ekkehard (1998) Intensifiers and reflexives: a typological perspective. MS. Presented at USC. Konig, Ekkehard & Peter Siem und (1998) O n the developm ent o f reflexive pronouns in English: A case study in gram m aticalization. ms. Institut fur Englishe Philologie, Freie U niversitat Berlin. Konig, Ekkehard & Peter Siem und (1998) Tow ards a Typology o f Intensifiers (Em phatic Reflexives). Paper given at the X V Ith International Congress o f Linguistics, Paris July 1997. Konig, Ekkehard & Peter Siem und (1999) Intensifiers and reflexives: a typological perspective. In Reflexives: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volum e 40, 41-74. Am sterdam : John Benjamins. Koopm an, Hilda, and A nna Szabolcsi (2000) Verbal complexes. Cam bridge, M ass.: MIT Press. Koster, Jan (1987) Domains and Dynasties. Dordrecht: Foris. Koster, Jan (1991) “Tow ard a N ew Theory o f A naphoric Binding”. In Koster, Jan and Eric Reuland (eds.) (1991) Long-distance anaphora. Cam bridge: Cam bridge U niversity Press.Dordrecht: Foris. Koster, Jan and Eric Reuland (eds.) (1991) Long-distance anaphora. Cam bridge: Cam bridge University Press. Kuno, Susum o (1972) Pronom inalization, Reflexivization, and Diriect Discourse. Linguistic Inquiry 3(2): 161-195. Kuno, Susumo (19??) “The Syntax o f Com parative C lauses”. l l T l l l R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 1 Kuno, Susumo (1987) Functional Syntax: Anaphora, discourse and empathy. Chicago: The University o f Chicago Press. Kuno, Susumo, and E. Kaburaki (1977) “Empathy and syntax” . Linguistic Inquiry 8:627-672. Lakoff, G eorge (1971) ‘Presupposition and Relative W ell-form edness’, in D. Steinberg and L. Jacobovits (eds) Semantics - An Interdisciplinary Reader, Cam bridge: Cam bridge U niversity Press. 329-340. Lam brecht, Knud (\994) Information Structure, Cam bridge: Cam bridge University Press. Larsen, Erling G eorg (1969) Norron Grammatik. U niversitetsforlaget. Oslo. Lebeaux, D avid (1983) “A distributional difference between reciprocals and reflexives”. Linguistic Inquiry 14: 723-30. van der Leek, Frederike (1991) “Iconicity and tw o-form reflexive sysm tem s”. Chicago Linguistic Society 27: 445-463. Levin, Beth, and M alka Rappaport Hovav (1995) Unaccusativity at the syntax-lexical semantics interface. Linguistic Inquiry M onographs 26. The M IT Press, Cam bridge, M assachusetts. London, England. Levinson, S. (1991) Pragm atic reduction o f the binding conditions revisited. Journal o f Linguistics 21. 107-61. Levinson, S. (2000) Presumptive Meanings. M IT Press. Cam bridge, M assachusetts. Lichtenberk, Frantisek (1994) “Reflexives and reciprocals.” In R.E. A sher and J.M .Y. Simpson (eds.). The Encyclopedia o f Language and Linguistics, vol. 7. Oxford: Pergam on Press, 3504-9. Lidz, Jeffrey (1995) M orphological reflexive m arking: Evidence from Kannada. Linguistic Inquiry 26: 705-710. Lidz, Jeffrey (1996) Dimensions o f reflexivity. Doctural Dissertation, U niversity o f Delaware, Newark. Lidz, Jeffrey (2001) Condition R. Linguistic Inquiry 32, 123-140. Lobel, Elisabeth (2000) “Copular verbs and argum ent structure: participant vs. non-participant roles”. Theoretical Linguistics 26:229-258. Lodrup, H. (1999). Inalienables in N orw egian and binding theory. Lyutikova, Ekaterina A. (2000) Reflexives and em phasis in Tsaxur (N akh-D agestanian). In Reflexives: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl, 227-255. Am sterdam : John Benjamins. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 2 M aling, Joan (1984) N on-clause bounded reflexives in Icelandic. Linguistics and Philosophy 7:211-41. M aldonado, Ricardo (2000) Conceptual distance and transitivity increase in Spanish reflexives. In Reflexives: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volum e 40, 153-185. Am sterdam : John Benjamins. M artin, R. (1975) Sur I’unite du m ot meme. Travaux de linguistique et de litterature. Centre de philologie et de litteratures romanes de I ’ universite de Strasbourg. X lll, 1. p. 227-243. M cKay, Thom as J. (1991) He him self: U ndiscovering an anaphor. Linguistic Inquiry 22, 368- 73. M egerdoom ian, Karine (2002) Beyond words and phrases: A unified theory o f predicate composition. Ph.D. dissertation, USC. M ilner, Jean-Claude (1989) Introduction d une .science du langage. Editions du Seuil. Paris. M oravcsik, Edith A. (1972) Some C rosslinguistic G eneralizations about Intensifier Constructions. Chicago Linguistic Society 8: 271-77. M oyne, J.A. (1971) Reflexive and Em phatic. Lg. Al. 141-63. M uysken, Pieter (1993) Reflexes o f Ibero-Rom ance reflexive clitic + verb com binations in Papiam entu: Them atic grids and gram m atical relations. In Focus and grammatical relations, ed. F. Byrne and D. W inford. Am sterdam : John Benjam ins. M uysken, Pieter, and N orval Smith (1994) Reflexives in the creole languages: An interim report. In Creolization and language change, ed. D. A done and I. Plag. Tiibingen: Max N iem eyer Verlag. N ielsen, N iels Age (1989) Dansk Etymologisk Ordhog. Ordenes Historie. G yldendalske Boghandel, N ordisk Forlag A/S. Pan, Haihua (1997) Constraints on reflexivization in Mandarin Chinese. G arland Publishing: N ew Y ork & London. Pesetsky, D avid (2000) Phrasal movement and its kin. Cam bridge, M ass.: M IT Press. Pica, Pierre (1987) O n the nature o f the reflexivization cycle. In Proceedings o f N ELS 17, 483-499. GLSA, University o f M assachusetts, Amherst. Pica, Pierre (1991) On the interaction betw een antecedent-governm ent and binding: The case o f long-distance reflexivization. In Long-distance anaphora, eg. Jan K oster and Eric Reuland, 27-49. Cam bridge: Cam bridge U niversity Press. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 3 Plank, F. (1979) “Exclusivierung, Reflexivierung, Indentifizierung, relationale A uszeichnung: V ariationen zu einem sem antisch-pragm atischen Them a”, in 1 . Rosengren (ed.), Sprache und Pragmatik. Lund: C W K G leerup, pp. 330-54. Pollard, Carl, and Ivan Sag (1992) A naphors in English and the scope o f the binding theory. Linguistic Inquiry 23:261-305. Primus, Beatrice (1992) “Selbst-variants o f a scalar adverb in G erm an”. In Joachim Jacobs (ed.), Inform ationsstruktur und Gram m atik, Opladen: W estdeutscher Verlag, 54-88. Quirk, R. G reenbaum , S., Leech, G. & Svartik, J. (1985). A comprehensive grammar o f the English language. London: Longman. Reinhart, Tanya {\9S3) Anaphora and semantic interpretation. London: Croom Helm. Reinhart, Tanya (2000) Strategies o f anaphora resolution. In Interface strategies, ed. Hans Bennis, M artin Everaert, and Eric Reuland, 295-325. Amsterdam: Royal Academ y o f Arts and Sciences. Reinhart, Tanya (To appear) The theta system: Syntactic realization o f verbal concepts. Cam bridge, M ass.: M IT Press. Reinhart, T. & Reuland, E. (1991) Anaphors and logophors: An argum ent structure perspective. In Long-distance anaphora, eg. Jan K oster and Eric Reuland, 27-49. Cam bridge: C am bridge U niversity Press. Reinhart, T. & Reuland, E. (1993). Reflexivity. Linguistic Inquiry 24. 657-720. Reuland, Eric (1996) Pronouns and features. In NELS 26, 319-333. GLSA, University o f M assachusetts, Am herst. Reuland, Eric (1998) Structural conditions on chains and binding. In N ELS 28, 341-356. GLSA, University o f M assachusetts, Am herst. Reuland, Eric (2000a) Anaphors, logophors and binding. In Long-distance reflexives, ed. Peter Cole, G abriella Hermon, and C.-T. Jam es Huang, 343-370. Syntax and Sem antics 33. San Diego, C a lif: A cadem ic Press. Reuland, Eric (2000b) The fine structure o f grammar: Anaphoric relations. In Reflexives: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl, 227-255. Am sterdam : John Benjamins. Reuland, Eric, and Tanya Reinhart (1995) Pronouns, anaphors and Case. In Studies in comparative Germanic syntax, ed. H ubert Haider, Susan Olsen, and Sten Vikner, 241-269. Dordrecht: Kluwer. R e p ro d u c e d with p erm issio n of th e copyright ow ner. F u rth e r reproduction prohibited w ithout perm ission. 4 4 4 Reuland, Eric, and SigriSur Sigurjonsdottir (1997) Long distance “binding” in Icelandic: Syntax or discourse? \n Atomism in binding, ed. Hans Bennis, Pierre Pica, and Johan Rooryck, 323-340. D ordrecht: Foris. Rizzi, Luigi (1990) Relativized M inim ality, Cam bridge, M ass.: M IT Press. Rochem ont, M ichael (1986) Focus in Generative Grammar, A m sterdam /Philadelphia: John Benjamins. Rooryck, J. & G. Vanden W yngaerd (1998) ‘The S elf as Other, A M inim alist approach to zich and zichzelf in D utch’ NELS IS, 359-373. Rooth, M ats {\ 9S5) Association with Focus, PhD dissertation. University o f M assachusetts at Amherst. Rooth, M ats (1992). A theory o f focus interpretation. Natural Language Semantics 1. Ross, John R. (1970) “On declarative sentences”. In R.A.Jacobs and P.S. Rosenbaum (eds). Readings in English Transformational Grammar, W altham, Mass.: Ginn & Co. Rothstein, Robert A. (1970) “Reflexive “R eflexive V erbs” in Polish”, Slavic and East European Journal [SEEJ], Vol. XIV, No. 2, pp. 194-197. Rullm ann, H otze (1988) Binding, Stress and the Distribution o f Pronouns in Dutch, ms. G roningen University, the Netherlands. Ruwet, N icolas (1972) “Les constructions pronom inales en Franfais. Restriction de selection, transform ations et regies de redondances.” Le Frangais Moderne 40:102-25. Safir, K. (1992) Im plied non-coreference and the pattern o f anaphora. Linguistics and Philosophy 15: 1-52. Safir, K. (1996). Sem antic atom s o f anaphora. Natural Language and Linguistic Theory 14. 545-589. Safir (1997) Symmetry and Unity in the Theory o f Anaphora. M.S. Saxon, Leslie (1991) “O n one’s own: The sem antics and pragm atics o f reflexives”. In Carol G eorgopoulos and Roberta Ishihara (eds.). Interdisciplinary Approaches to Language: Essays in honor o f S.-Y.Kuroda. Dordrecht: Kluwer, 501-17. Schachter, Paul (1976) “A N ontransform ational A ccount o f G erundive N om inals in English”, Linguistic Inquiry 1'.205-241. Schladt, M athias (2000) The typology and gram m aticalization o f reflexives. In Reflexives: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) V olum e 40, 104-124. Am sterdam : John Benjamins. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 445 Schlenker, P. (1999) Propositional A ttitudes and Indexicality: A Cross-Categorical Approach. Ph.D. D issertation M IT. Sells, Peter (1987) A spects o f logophoricity. Linguistic Inquiry 18.445-79. Sells, Peter, A nnie Zaenen, and Draga Zee (1987) “Reflexivization variation; Relations betw een syntax, sem antics and lexical structure” . In Working papers in grammatical theory and discourse structure, ed. M asayo lida, Stephen W echsler, and D raga Zee, 169-238. Stanford, Calif.: CSLI Publications. [D istributed by Cam bridge University Press.’ Siemund, Peter (2000) Intensifiers in English and German. A Comparision. Routledge. Sola, Jaum e (1994) A uniform analysis for SELF elements. U npublished manuscript. U niversity o f G roningen. [Have not been able to procure copy of this paper] Thrainsson H oskuldur (1991) Long-distance reflexives and the typology o f NPs. In Long distance anaphora, eg. Jan K oster and Eric Reuland, 27-49. Cam bridge: Cam bridge U niversity Press. Tim berlake, A lan (1979) “Reflexivization and the Cycle in Russian.” Linguistic Inquiry 10.1:109-141. Townsend, Charles E. (1967) “Voice and V erbs in -s ja .” Slavic and East European Journal 11.2: 196-203. Valfells, Sigridur (1970) “M iddle voice in Icelandic.” In Hreinn B enediktsson (ed.). The Nordic languages and modern linguistics 1, 551-572. Reykjavik: V isindafelag Islendinga. van Gelderen, Elly (1996) “S elf in the history o f English”. Germanic Linguistics Annual Conference 2. A rizona State University, Handout. Vat, Jan (1980) “Zich en Z ich zelf’, in Linguistics in the Netherlands 1980:127-138. Veraart, Fleur (1996) On the Distribution o f Dutch Reflexives. MA thesis, MIT. Vergnaud, J.-R. & Zubizarreta, M .-L. (1992). The definite determ iner and the inalienable constructions in French and in English. Linguistic Inquiry 23. 595-652. V interberg, H, and C.A. Bodelsen (1966) Dansk-Engelsk Ordhog. G yldendalske Boghandel, N ordisk Forlag. Copenhagen. W altereit, Richard (2000) W hat it m eans to deceive yourself; The sem antic relation o f French reflexive verbs and their corresponding transitive verbs. In In Reflexives: Forms and functions, ed. Zygm unt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) V olum e 40, 104-124. Am sterdam : John Benjam ins. W ehrli, Eric (1986) “On som e properties o f French clitic se”. In Hagit Borer (ed.), The Syntax o f Pronominal Clitics [Syntax and Semantics 19]. N ew York: Academ ic Press, 263-83. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 6 W illiam s, K em p (1988) “Exceptional B ehaviour o f A naphors in A lbanian”. Linguistic Inquiry 19:161-8. W illiams, Edwin (1985) “PRO and Subject o f N P ”, Natural Language and Linguistic Theory 3, 297-315. Zarechnak, M ichael (1971) “-s ja Verbs in Russian.” Slavic and East European Journal 15.2:199-209. Zribi-Hertz, Anne (1980) “Coreferences et pronom s reflechis: notes sur le contraste lui/lui- meme en franfais”, Linguisticae Investigationes IV: 1. Zribi-Hertz, Anne (19??) “Econom isons-nous: A propos d ’une classe de formes reflexives m etonym iques en frangais.” In ????? vol. ??, pp. 104-128. Zribi-Hertz, A nne (1987) Le reflexivite ergative en lfan9ais moderne. In Le Frangais Moderne vol. 55, pp. 23-54. Zribi-H ertz, Anne (1989) A napbor binding and narrative point o f view: English reflexive pronouns in sentence and discourse. Lg. 65.695-727. Zribi-Hertz, A. (1995) ‘Em phatic or reflexive? O n the endophoric character o f French lui- m em e and sim ilar com plex pronouns’, J. Linguistics 31 (1995), 333-374. Zribi-H ertz (1996) “Some w ondering rem arks on the developm ent o f syntactic thories: The case o f long-distance reflexives”. Recherches Linguistiques de Vincennes 25. Zubizarreta, M aria-Luisa (1998) Prosody, Focus, and W ord Order. Linguistic Inquiry. M onograph Thirty-Three. M IT Press. Cam bridge, M assachusetts. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 7 Appendix I Danish se(^-compounds I .l A bbreviations T h e fo llo w in g list o f w o rd s c o n ta in in g th e e le m e n t selv is b ased o n V in terb erg , H , an d C .A . B o d e lse n (1 9 6 6 ) D ansk-E ngelsk O rdhog. G y ld en d alsk e B o g h an d el, N o rd isk F o rlag . C o p en h ag en . A bbreviations: A N T /a a n ti-re fle x iv e (tran sitiv e) N E U /n e u n eu tral (tran sitiv e) IN H /i in h eren tly re fle x iv e (tran sitiv e) D O /do selv ~ d ire c t o b je c t o f v erb A D V /ad v selv ~ ad v erb ial m o d ifie r o f v erb 1 0 /io selv ~ in d ire c t o b je c t o f v erb DITEUditr = d itra n sitiv e verb IN T /in tr. = in tran sitiv e vi = in tran sitiv e verb v t = tra n sitiv e verb n = n o u n adj = ad jectiv e T h e a b b re v ia tio n s A N T /a , N E U /n e u an d IN H /in h in d icate th e ty p e o f th e v erb on w h ich th e .ye//'-com pound is b ased . IN T /in tr and D IT R /d itr fu rth e r d e sc rib e s th e v erb ty p e b y in d icatin g w h e th e r th e v erb is in tra n sitiv e o r d itran sitiv e. It is assu m e d th a t all v erb s d e sc rib e d as e ith e r A N T /a, N E U /n e u , o r IN Fl/inh are e ith e r tra n sitiv e or d itran sitiv e. T h e a b b re v ia tio n s D O /d o , lO /io , an d A D V /a d v in d icate th e sy n tactic ro le p lay ed b y th e e le m e n t selv ‘s e l f in sid e th e 5c//'-com pound. T h at is, is it th e d ire c t o b ject o f th e v erb (e.g. P eter respects him self, P e te r ’ s self-respect), an ad v erb ial e le m e n t (e.g. P eter sto red the goods h im se lf self-starage), o r p e rh a p s th e in d irect R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 48 o bject. B e lo w each i'c ^ -c o m p o u n d all th e ad d itio n al c o m p o u n d s b ased on th e sam e v erb al b ase are listed, e.g.: A N T N E U IN H IN T D O A D V 1 0 D IT R selvm ord n. ‘suicide’ selvm order n. ‘(person committing) suicide’ selvm orderisk adj. ‘suicidal’ selvm ordsm ani n. ‘suicidal mania’ selvm ordspilot n. ‘suicide pilot’ m yrde v t.‘kill, m urder’ m ord n. ‘m u rd er’ a do B e lo w th e list o f related .s'c^-eom pounds th e b ase w o rd (s) fro m w h ich th e self com pound is d e riv e d are listed , see en try fo r selvm o rd ‘su ic id e ’ ab o v e. W hen th e b ase w o rd (s) h a s/h a v e m o re th an o n e m ean in g , th e re le v a n t m ean in g has b een h ig h lig h te d in b o ld fonts. F in ally , o n th e last line o f th e en try th e ty p e o f th e v erb al b ase is d escrib ed an d th e fu n ctio n o f th e e le m e n t selv is in d icated . In th e case o f selvm o rd ‘su ic id e ’ ab o v e, th e o rig in al verb myrde ‘kill, m urder’ is a transitive, anti-reflexive(a) verb. And selv can be interpreted as direct object (do). In the statistical surveys, xeZ/^compounds like selvm o rd ‘su ic id e ’ are c o u n te d o n ly once. 1.2 T he m orph ology o f derived nouns in D anish In D an ish m o st o f th e c o m p o u n d s c o n ta in in g selv ‘s e l f are e ith e r n o u n s o r a d jectiv es d eriv ed form a v erb al stem . T h e fo llo w in g lists o f d eriv atio n al su ffix es an d ex a m p le s are in te n d e d to illu strate th e w o rd fo rm atio n p ro c e sse s in v o lv ed in th e fo rm atio n o f th e x e/v -co m p o u n d s fo u n d in th e list in A p p e n d ix I. T h a t is w o rd fo rm atio n p ro c e sse s no t e n c o u n te re d in th e w o rd s in th is list are n o t in clu d ed in th e fo llo w in g d e sc rip tiv e su rv ey o f d eriv atio n al m o rp h o lo g y . R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 4 9 (1) N O U N fo rm a tio n d eriv atio n al su ffix es a d d ed to v erb al stem s: (4) a. -else b. -(n)ing c. -eri d. -(a)tion e. -er f. -(e)lig-hed g- -som -hed h. - 0 (v erb + -lig => ad jectiv e, see (8 b )) (v erb + -som => ad jectiv e, see (8 d )) (2) E x a m p le s o f d ev erb al n o m in a liz a tio n s c o n ta in in g th e d eriv atio n al su ffix es liste d in (1): V erb E n d in g N o m in a liz a tio n a. bedomme ‘ judge’ -else bedommelse ‘judgment’ b. begrcense ‘limit’ -ning begrcensning ‘limitation’ c. bygge ‘build’ -eri byggeri ‘the activity of building’ d. am putere ‘amputate’ -ation amputation ‘amputation’ e. bygge ‘build’ -er hygger ‘builder’ f. virke ‘work, function’ -som -hed virksom hed ‘business’ g- behage ‘please’ -lig-hed behagelighed ‘pleasantness’ h. bedrage ‘deceive’ - 0 bedrag ‘deceit’ (3) N o m in a liz a tio n s o f th e ty p e s illu strated in (1) c o n ta in in g th e e le m e n t selv. a. selvbedom m else ‘se lf-ju d g m e n t’ b. selvbegrcensning ‘se lf-k n o w le d g e ’ c. selvbyggeri ‘b u ild in g o n e ’s h o u se w ith o n e ’s o w n h a n d s ’ d. selvam put ation ‘sp o n ta n e o u s a m p u ta tio n ’ e. selvbygger ‘p e rso n w h o b u ild s h is h o u se w ith h is o w n h a n d s ’ f. se lw irkso m h e d ‘se lf-a c tiv ity ’ g- selvbehagelighed ‘(se lf-)e o m p la ce n c y , self-sa tisfa c tio n , sm u g n e ss’ h. selvbedrag ‘self-d elu sio n , self-d eceit, se lf-d e c e p tio n ’ N O U N fo rm atio n d eriv atio n al su ffix es a d d ed to ad jectiv al stem s: a. -hed (adj. + -h e d ^ > n o u n ) (5) E x am p les o f ‘d e -a d je c tiv a f n o m in a liz a tio n s c o n ta in in g th e d eriv atio n al su ffix es listed in (4): V erb E n d in g N o m in a liz a tio n a. virkesom ‘working, functional’ -hed virksom hed ‘business’ b. behage/ig ‘pleasant’ -hed behagelighed ‘pleasantness' (6) N o m in a liz a tio n s o f th e ty p e s illu strated in (5) e o n ta in in g th e e le m e n t selv. a. selw irk so m h e d ‘se lf-a c tiv ity ’ b. selvbehagelighed ‘(se lf-)c o m p la ce n c y , se lf-sa tisfa c tio n , sm u g n e ss’ ((6 a,b ) = ((3 f,g )) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 5 0 (7) A D JE C T IV E fo rm a tio n d eriv atio n al su ffix es ad d ed to v erb al stem s: a. -ende (g e ru n d /p re se n t p articip le) b. -(e)lig c. - (e)t/d/en (p ast p a rtic ip le ) d. - som e. -(a)tiv (8) E x am p les o f d ev erb al ‘a d je c tiv iz a tio n s’ c o n ta in in g th e d eriv a tio n a l su ffix es listed in (7): V erb E n d in g A d je c tiv e a. beskue ‘look at, observe’ -ende beskuende ‘observing’ b. behage ‘please’ -lig behagelig ‘pleasant, pleasing’ c. erhverve ‘acquire -t ejhven’et ‘acquired’ d. virke ‘work, function’ -som virksom ‘working, functional’ e. forme ‘form’ ~(a)tiv formativ ‘formative’ (9) ‘A d je c tiv iz a tio n s’ o f th e ty p es illu strated in (8) c o n ta in in g th e e le m e n t selv. a selvbeskuende ‘in tro sp ectiv e; n a rc issistic ’ b. selvbehagelig ‘(se lf-)c o m p la ce n t, self-satisfied , sm u g ’ c. se lv e rh v e rv e t ‘self-acq u ired , se lf-ta u g h t (fx k n o w le d g e )’ d. selw irkso m ‘a c tiv e ’ (10) A D JE C T IV E fo rm a tio n d eriv atio n al su ffix es ad d ed to n o m in al stem s: a. - isk (11) E x am p les o f a d je c tiv iz a tio n s c o n ta in in g th e d eriv atio n al su ffix es listed in (10): N o u n E n d in g A d je c tiv e a. morder^ ‘murderer’ -isk m orderisk ‘murderer-like’ (12) A d je c tiv iz a tio n s o f th e ty p e s illu strated in (11) c o n ta in in g th e e le m e n t selv. a. selvm orderisk ‘su ic id a l’ 1.3 L ist o f com pounds containing the m orphem e selv ‘s e lf A N T N E U IN H IN T D O A D V 1 0 D IT R selvafhryder n . ‘automatic cutout’ afbryde 1 interrupt, break (fx the silence); 2 interrupt (fx conversation); cease to do, interrupt, break off (fx work), 3 (electric current) interrupt, cut off, switch off; 4 (teleph) cut off; disconnect [afbryde sig selv switch-off REFL self ‘switch-off itself] a do [afryde strammen a f sig selv switch-off current-the o f REFL self ‘switch off the current by itself] a adv ' The agentive noun morder ‘murderer’, in (11a) above, is derived from the verb myrde ‘to kill’ through both (apophonic) vowel change (y > o) and adding the agentive suffix -er. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 451 selvqfladning n. ‘self-discharge’ aflade 1 (naut) ship; unload; 2 discharge (fx a battery) [aflade sig selv discharge REFL self a do ‘discharge itself] selvqflceggende selvafleegger adj. ‘self-dump(fx hayrake)’ n. ‘self-rake’ (note: aflcegger = ‘(agricult.) side-delivery reaper’ 1 put down; 2 take off, discard; 3 drop, leave off, discard; 4 (gardening) layer, lay [aflcegge h0et a f sig selv drop hay-the o f REFL self ‘drop the hay by itself] adv selvagtelse n. ‘self-respect, self-esteem’ agte 1 esteem , respect, reverence, revere, venerate; 2 intend; mean; 3 regard as, consider [agle sig selv esteem REFL self do ‘esteem oneself] selvamputation n. ‘spontaneous amputation’ amputere 1 am putate [amputere sig selv amputate REFL self a do [amputeredes a f sig selv amputated-PASSI VE o f REFL self a adv ‘spontaneously’ selvanalyse n. ‘self-analysis’ analysere 1 analyze [analysere sig selv analyze REFL self a do ‘amputate itself] ‘was amputated by itself’] ‘analyze oneself” selvanden adv. ‘with one other (person)’ 1 (num..) second, (adj.) other {?? anden ‘other’ is not a verb} selvangive v. ‘file one’s (income) tax return’ selvangivelse n. ‘confession; tax return; tax form’ angive 1 state; note; 2 indicate; 3 allege; profess; 4 inform against, denounce [angive skatten selv state tax-the self a adv [angive sig selv denounce REFL self a do ‘state the tax oneself] ‘denounce/inform against oneself selvanklage n. ‘self-accusation’ selvanklagende adj. ‘self-accusing, self-accusatory’ selvanklager n. ‘self-accuser’ anklage 1 accuse. [anklage sig selv accuse REFL self a do ‘accuse oneself” selvansvarlig adj. ‘responsible’ ansvarlig 1 adj. responsible [ansvarlig fo r sig selv responsible for REFL self {?? ansvarlig ‘responsible’ is not a verb} ‘responsible for oneself”] selvantcendelig adj. ‘spontaneously combustible’ selvantcendelse n. ‘spontaneous combustion, self-ignition’ selvantcending n. ‘spontaneous combustion, self-ignition’ selvantcendt adj. ‘ignited spontaneously’ antcende 1 set on fire; set fire to, kindle, ignite (fx ignite the petrol in a motor) [antcende sig selv selvarbejde n. ‘self-activity’ arbejde 1 w ork intr. ignite REFL self (do) (adv) ‘spontaneously’ adv (??) ‘ignite itself] R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 452 selvavl n. ‘spontaneous generation, autogenesis’ selvavlet adj. ‘generated spontaneously’ avle 1 grow, raise; 2 breed, 3 beget, proereate, engender, generate a (do) (adv) ‘spontaneously’ sehbebrejdeise adj. ‘self-reproach’ 1 reproach a do selvbedrag n. ‘self-delusion, self-deceit, self-deception’ bedrage 1 deceive, cheat a do selvbedammelse n. ‘self-judgment’ bedamme 1 judge; estim ate; m ark a do selvbefrugtende adj. ‘self-fertilizing, selfing’ selvbefrugtning n. ‘self-fertilization’ befrugte 1 (biol.) fertilize, feeundate, (artificially) inseminate, a do selvbegrcensning n, ‘self-knowledge’ (??) begrcense 1 limit; 2 reduce, restrict begrcense sig 1 keep within reasonable bounds, a do selvbehag n. ‘(self-)complaceney, self-satisfaction, smugness’ selvbehagelig adj. ‘(self-)complacent, self-satisfied, smug’ selvbehagelighed n. ‘(self-)complacency, self-satisfaction, smugness’ behage 1 please, appeal to; 2 deign (to do...) a (do) (although behagelig is an adj. it is derived from a verb behage ‘please’ selvbeherskelse n. ‘self-control, self-command, self-restraint’ selvbehersket adj. ‘self-possessed, self-controlled’ beherske 1 rule over, govern; 2 control, dominate, be master of; 3 master (fx a language) beherske sig 1 control oneself, restrain oneself, keep one’s temper (a) (i) do selvbehotd n. ‘retention’ beholde 1 keep, retain, a do selvbekendelse n. ‘(voluntary) confession’ bekende 1 confess meronymic a (do) (adv)(??) selvbekrceftelse n. ‘self-affirmation’ bekrcefte 1 certify; 2 affirm , confirm a do selvbesindelse n. ‘self-reflection; self-communion, collectedness’ besinde sig 1 colleet oneself, collect on e’s thoughts; 2 regain one’s composure; 3 think twice, pause, deliberate i (do) (adv) selvbeskatning n. ‘self-taxation’ beskatte 1 tax; rate; assess a do selvbeskikket adj. ‘self-appointed, self-constituted’ beskikke 1 appoint, 2 allot a do R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 453 selvbeskuelse n. 'introspection, self-examination, self-observation’ selvbeskuende adj. ‘introspective; narcissistic’ beskue 1 contem plate, behold, view , gaze at. a do selvbeskceftigelse n. ‘self-activity; self-communion’ beskceftige 1 occupy, engage, 2 employ beskceftige sig (med) 1 be occupied with, occupy oneself with, devote one’s time to. (a) (n) (i)(??) do selvbesmittelse n. ‘self-abuse, self-pollution, masturbation, onanism’ besmitte 1 defile, pollute; sully (fx one’s name), contaminate (fx c. sby’s morals) a do selvbespejling n. ‘narcissism’ hespejle NO VERB spejle 1 reflect (fx the lake reflects the trees); mirror, 2 fry (fx an egg) spejle sig I be reflected, be mirrored (fx in the water) (?? there is no verb *bespejle} selvbeslaltet adj. ‘self-appointed, self-constituted’ bestaltet adj. duly appointed NO VERB (a) (do) selvbestemmelse n. ‘self-determination’ selvbestemmelsesrel n. ‘(right of) self-determination’ bestemme 1 decide, m ake up o n e’s m ind, determ ine, resolve; 2 determine, fix (fx the price) a adv selvbestpvende sehbestpver selvbestovning bestpve a adj. ‘autogamous, self-compatible’ n. ‘autogamous plant’ n. ‘autogamy, self-pollination’ 1 pollinate (do) (adv) sehbesdning besa selvbetjenende selvbeljening n. ‘self-seeding’ 1 sow (fx a field) (do) adj. ‘automatic, self-acting’ n. ‘self-service’ sehbetjeningsvaskeri n. ‘launderette’ (adv) betjene 1 serve; attend; w ait on, 3 operate, work (fx machine) do selvbelraglning n. ‘introspection, self-communion, self-observation’ betragte 1 look at, contem plate; view a do selvbeundring beundre a n. ‘self-admiration’ 1 adm ire do selvbevidst selvbevidsthed bevidsi {?? there is no verb} adj. ‘self-conceited, self-opnionated, self-important, arrogant’ n. ‘self-conceit, self-importance, arrogance’ 1 conscious; 2 self-confident; selvbevcegelig adj. ‘self-propelling (fx guns)’ selvbevcEgende adj. ‘self-propelling (fx guns)’ bevcege 1 move; agitate; 2 move (fx the scene moved him); affect, touch bevcege sig 1 move, stir (a)?? (n)?? (i)?? (do) (adv) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 5 4 sehhindende adj- mejemaskine) see selvbinder'’ selvbinder n. ‘reaper-binder, harvester, self-binder’ selvbindersejl n. ‘canvas’ binde 1 bind, tie; tie up (fx a dog); 2 tie (fx a knot); 3 bind (fx this promise binds me) a adv {the machine is not binding itself, but biding hay by itselfj selvbiograf sehbiograjisk biograf n n. ‘autobiographer’ adj. ‘autobiographic(al)’ 1 n. biographer (do) selvblegende adj. ‘(gart) ~ bladselleri self-blanching celery’ blege 1 bleach, blanch, w hiten a do selvbuden adj. ‘~g®st intruder; gate-crasher’ byde/b0d/hudt(buden) 1 ask, invite; 2 order, command indbyde 1 invite a do selvbygger n. ‘[person who builds his house with his own hands’ selvbyggeri n. ‘[building one’s house with one’s own hands’ selvbygget adj. ‘[built with one’s own hands]’ bygge 1 build, construct (fx a house) a adv selvbcerende adj. ‘standing on its own base’ bcere 1 carry (compare w. bcere s ig ‘behave ’) a do selvcentrerende centrere a adj. ‘self-centering’ 1 center do selvdannelse 'a .\h m \) = selvavl' selvdannelsesevne n. ‘power of spontaneous generation’ selvdannet adj. ‘self-created’ danne 1 form, make (up), frame, shape, mould; 2 form, found (fx a society); 3 make, constitute; 4 form; produce; generate (fx heat) danne sig 1 form (itself) (fx a red spot formed on his hand) (a) (i) (do) (adv) ‘spontaneously’ selvdisciplin n. ‘self-disciplin’ disciptiner e 1 vt. disciplin a do selvdreven adj. ‘self-propelled’ drive 1 drive, move, drift (fx cattle); 4 drive, operate, work, (~frem fx car, ship, airplane) propel a do selvdyrkelse n. ‘self-worship’ dyrke 1 w orship (fx a god), idolize (fx a hero) a do selvdad adj. ‘dead from accident or disease (about animals)’ d0 1 die, pass away INTRANS int adv selvdamme n. ‘[the right of the plaintiff to decide the penalty to be paid by the defendant]’ adv damme 1 judge, decide, a (do) selve adj. ‘the very’ {?? selve = adjectival form o f selv, no verbal base) R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 455 seivegen egen {?? no-verbal base) selveje selvejendom selvejende selvejer selvejerbil selvejerbonde selvejergdrd selvejervogn eje adj, ‘original, particular, peculiar, specific’ 1 adj. own n. ‘private property; freehold’ n. ‘private property; freehold” adj. ‘-institution private foundation, ~ landm and owner-farmer ’ n. ‘owner; (af land) freeholder, owner-occupier’ n. ‘private ear’ n. ‘freeholder, yeoman (farmer), peasant proprietor’ n. ‘freehold farm’ n. ‘private carriage’ 1 ow n, possess adv [selveje does not mean that one owns oneself but rather that owns something by oneself, cf. selvbygger, selvbinder, self storage, etc. Seems to be a productive way to form self- nominalizations! selverfaring erfare n. ‘personal experience’ 1 experience adv selverhverv n. ‘independent employment’ selverhvervende adj. ‘self-supporting’ selverhvervet adj. ‘self-acquired; self-taught (fx knowledge)’ erhverve NO TR. VERB erhverve sig 1 acquire; gain, get, obtain; aehieve (fx status) erhverve sit udkomme 1 earn one’s living. erhverv 1 n. occupation, trade, pursuit, calling a adv ditr selverkendelse n. ‘self-knowledge’ erkende 1 admit (fx admit that you are right), acknowledge (tx one’s mistakes); recognize 2 understand, realize; (phil) apprehend, know, arrive at a cognition of a do selveste adj. the very, the highly esteemed, etc. {?? superlative o f selve, no verbal base} selvforagt n. ‘self-contempt’ foragte 1 despise, disdain, scorn, hold in contempt; a do selvforblindelse n. ‘delusion’ forblinde 1 (fig) blind, dazzle, infatuate, a do selvforhrcending n. ‘{see selvantcendelsef forbrcende 1 vt/vi. burn (fx wood burns easily); 2 bum (fx one’s hands); 3 scorch, parch, intrans. a (do) (adv) selvfordybelse n. ‘self-communion’ fordybe 1 deepen, make deeper; fo rd y b e sig i I becom e deeply absorbed in, becom e im m ersed in, bury on eself in ?? (the nominalization is derived from fo rd yb e sig i rather than fordybe) a (i) (io)?? selvfordojelse n. ‘(fysiol) autolysis’ fordaje 1 vt. digest a (adv) {DO does not make sense here ( selvfordom m else n. ‘self-condemnation’ fordom m e 1 vt condemn (tx c. shy. for his conduct); denounce (fx he denounced his conduct publicly) a do R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 5 6 selvforgiftning forgifte a n. ‘autointoxication’ 1 vt poison. do selvforglemmelse n. ‘forgetting oneself sehforglemmende adj. ‘self-forgetting, self-denying (fx love)’ forglem me 1 forget; ikke at ~ not forgetting. forglem me sig 1 forget oneself (ix he forgot himself as far as to strike her) (a) (n) do selvforgudelse n. ‘self-worship’ forgude 1 vt. idolize, dote on. forgudelse 1 n. idolization a do selvforherligelse n. ‘self-exaltation, self-glorification’ forherlige 1 vt. glorify, exalt, extol. forherligelse 1 n. glorification, a do selvforklarende forklare sehform dt • a {11} form al selvfornedrelse fornedre adj. ‘self-explanatory’ 1 vt explain; 2 expound (fx the bible); interpret; 3 explain; account for; (fx one’s conduct) 4 account for (fx that accounts for his conduct), do (adv)?? n. ‘end in itself 1 n. aim, end, object, purpose, objective. n. ‘self-abasemenf 1 vt degrade, debase, abase; hvo sig selv ophojer skal ~s, og hvo sig selv ~r skat ophojes (bib!) whosoever shall exalt himself shall he abased; and he that shall humble himself shall be exalted, do selvfornceglelse n. ‘self-denial’ selvforncegtende adj. ‘self-denying’ forncegte 1 vt disown (fx one’s son), disclaim, disavow, renounce (fx the authority of the law); repudiate; 2 vt deny (fx deny God, one’s faith, one’s country) hart ~de sig ikke T he ran true to form; ~ sig selv (bibl) deny oneself (fx let him deny himself and take up his cross and follow me): ban ~r ikke sig selv that is just what he would do, that is him all over. (a) (n) do selvforplejning n. ‘[supplying oneself with provisions]’ forpleje 1 vt board, feed; cater for. a do selvforringelse n. ‘self-disparagement’ forringe 1 vt reduce (fx the possibilities of..); impair (fx our influence abroad); 2 depreciate, detract from the value of; 3 disparage (=forklejne) a do selvforsikrer selvforsikret selvforsikring forsikre a selvforskytdl forskyldt {11 no verbal base) selvforstcerkende forstcerke n. ‘self-insurer’ adj. ‘self-insured’ n. ‘self-insurance’ 1 vt insure; 2 assure (fx he assured me that I was wrong) (do) (adv)?? adj. ‘self-inflicted’ 1 adj.yjj Ion som ~ get one’s deserts adv adj. ‘self-increasing’ 1 vt strengthen (ix the foundations of a building); make stronger, fortify; 2 vt augment, reinforce; 3 intensify, heighten, enhance (fx the effect); 4 (radio) amplify do R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 457 selvforstaelse forsta n. ‘self-knowledge’ 1 vt & vi understand, com prehend, perceive, grasp; realize, see; 2 know (fx he knows how to do selvforsvar n. ‘self-defence’ forsvare 1 vt defend (fx one’s country, one’s interests, a fortress) (forsvare sig) 1 defend oneself, put up a defense n do selvforsynende adj. ‘self-sufficient’ selvforsyning n, ‘self-sufficiency’ selvforsyningspolitik n. ‘policy of self-sufficiency, autarkic policy” forsyne 1 vt supply shy with sth, provide sby with sth,; furnish forsyne sig (med) 1 help oneself (to) forsyne sig med noget provide oneself with sth (fx p. oneself with money for the journey) forsyne sig (has) procure one’s supplies from sby n io ditr selvforsorgende adj. ‘self-supporting, independent’ forsorge 1 vt support, keep, provide for, maintain, ~ en fam ilie support a family forsorge sig selv support oneself, be self-supporting (a) (n) (do) (adv) selvforyngelse n. ‘(forst) natural reproduction, self-sowing’ forynge 1 vt rejuvenate; (forst) regenerate foryngelse 1 n. rejuvenation, rejuvenescence; (forst) regeneration. kunstig/naturlig foryngehe (forst) artificial/natural regeneration a (do) (adv) selvfremkaldt adj. fremkalde ~sdr self-inflicted wound’ 1 cause, bring about, give rise to, produce, call forth, provoke, evoke; engender; 2 (photo) develop; 3 give a call (fx actor after the end of play) adv selvfrugtbar adj. ‘autogamous, self-fertile’ selvfrugtbarhed n. ‘autogamous, self-fertilization’ frugtbar adj fertile, rich (soil); 2 fruitful, productive (fx co-operation) adj {??} selvfyldende selvfylder fylde a selvfolelse selvfolende foie folelse sehfolge selvfolgelig selvfolgelighed folge a selvgjort a selvglad adj {??} adv adj. ‘self-filling (fx fountain-pen)’ n. ‘self-filling fountain-pen, self-filler’ a 1 vt fill (fx fill the glass, fill the pen), fill up; 2 refill, replenish; 3 inflate, blow up, fill (do) (adv) n. ‘self-esteem; egotism’ adj. ‘conceited, self-satisfied, self-important; proud’ 1 vt feel; sense (fx I sensed a eertain hostility in his manner); perceive, be sensible of n. feeling; sentiment, 2 feeling, sensation (fx sensation o f heat); 3 sense of touch; etc.. (do)?? n. ‘matter of course’ adj. ‘natural, inevitable; obvious’ n. ‘matter of course; truism; naturalness’ 1 vt follow (adv)?? adj. ‘of one’s own making, o f one’s own creation’ 1 vt do, make adv {cf. selvbygger, etc.) adj. ‘pleased with oneself, self-satisfied; complacent; smug’ adj. happy, (io) ?? R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 458 selvglcede n i m glcede selvgod sehgodhed adj {??} selvgroet god gro n. ‘self-satisfaction, complacency’ n. happiness ( 10) ?? adj. ‘self-righteous, priggish, pharisaric(al), smug’ n. ‘self-righteousness, pharisaism, smugness’ adj. good adj. ‘self-sown, self-set; in its natural shape (abt. wood); (fig.) peculiarly one’s own (Gruntvig’s mode of thought was peculiarly his own and came from within him self’ 1 vi grow; 2 vt grow intr adv selvgcering n. ‘anaerobic respiration’ giere vi ferm ent; (fig) ferment, work (fx the suggestion was beginning to work in his mind) intransitive intr adv selvgdende adj. ‘(mining) ~ malm self-fluxing ore’ gd vi walk, go, intr adv selvhelbredelse n. ‘spontaneous healing’ helbrede 1 vt cure, restore (to health); (bibl & poet about miracle cures) heal helbredelse n. cure, restoration to health (fx to cure sby); 2 recovery (fx sby recovered) n (do) (adv) ‘spontaneously’ selvhengivelse n. ‘self-devotion’ hengive 1 vt sacrifice, give up; hengive sig til abandon oneself to, give oneself up to; indulge in (fx emotion, pleasures, etc.) i{??} (io)?^ selvhersker n. ‘autocrat’ selvherskerdomme n. ‘autocracy’ selvkerskermagt n. ‘autocracy’ hersker vi reign; 2 rule; 3 be (fx there was great confusion in the house); be prevalent; prevail a (do) (adv) selvhjulpen adj. ‘resourcefiil; (=selvfors0rgende), (abt. one who has succeeded) self-made’ selvhjcelp n. ‘self-help’ selvhjcelpssalg n. ‘[sale by an unpaid vendor in order to ascertain the amount of his damages]’ hjcelpe/hjalp/hjulpet(-en) vt help sby; do selvhceftende hcefte (a) selvhardende hcerde adj. ’ ’-vildvin (Parthenocissus tricuspidata) Japanese ivy, Boston ivy’ vt fix, fasten, attach, stich; 2 stick, paste; 3 stitch, sew; 4 vi stick (the paper will not stick); cling, (intr) (do) (adv) adj. ‘~std/self-hardening steel ’ vt. harden, intr adv selvhcevdehe selvhcEvdende hcevde n. ‘self-assertion’ adj. ‘self-asserting, assertive’ assert, claim. do selvhcevende adj. ‘self-raising (fx flour)’ hceve raise (about flour, dough, etc.) intr adv selviagttagelse n. ‘self-observation, introspection; personal observation ’ iagttage vt. observe, a do R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 5 9 selvilcegger n. ‘(agricult.) self-feeder’ (itegge ?????/ laegge i) {no verbal base) {no verbal base} adv selvimprcegnering n. ‘(forst) self-impregnation’ imprcegnere vt. im pregnate, proof (fx ~ wood) a (do) (adv) selvim pulserende adj. ‘self-pulsing (fx oscillator)’ (impulsere) {no such verb in dictionary, pulsere vi. beat, throb, pulsate (intr) adv selvindlysende adj. ‘self-evident (fx a s.-e. truth), obvious’ indlysende adj {no verbal base} adv {??} {cf. sehfolge, selvform dl, etc.} selvindstillelig adj. ‘self-adjusting’ indstillelig adj. ajustable indstille vt. adjust (fx apparatus, an engine, a brake) a do selvindtrcede v. ‘to contract on on e’s own b e h a lf ENESTE verbnm l! selvindlrcedelse n. ‘contracting on one’s own behalf sehiyidlrceden n. ‘contracting on one’s own behalf seh in d tm d e r n. ‘commission agent contracting on his own behalf selvindtrddt n. ‘contracting on one’s own behalf indtrcede vi. 1 enter; 2 commence, set in (fx a crisis set in); take place intr adv selvinduktion n. ‘self-induction’ sehinduktiv adj. ‘self-inducive’ induktion n. (physics) indnction inducere vt. (physics) induce a do (ad v )?? sehinfektion adj. ‘auto-infection, self-infection’ inficere a (do) (adv) ‘spontaneously” selvinteresse n. ‘self-interest (tx enlightened self-interest)’ interesse n. interest interessere vt. interest a adv {??} {=se/y/or/nd/. Se/vtnteresse is derived from the noun interesse] selvironi n. ‘self-irony’ seh iron isk adj. ‘self-ironic’ ironi n. irony n {?? no verbal base} adv ?? (= ‘irony on one’s own behalf) selvisk adj. ‘selfish’ seh iskh ed n. ‘selfishness’ adj. {W) selvjustits n. ‘(=se/vtegt)(jur.) self-help’ ju stits n. justice, adm inistration o f justice n {??} selvklatrende adj. ‘(hot.) climbing, creeping, trailing’ klatre vi climb, clamber, intr adv selvklog adj. ‘self-opinionated; conceited’ klog adj. smart, intelligent, adj {?? no verbal base} R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 4 6 0 setvktcebende ‘adj. adhesive (a latex-gummed envelope)’ klcebe vi. stick, cleave, adhere (til, ved to) vt. stick, glue, paste; ~ sammen ‘glue together’ klcebe sig til cling to, stick to. (a) (intr) (do) (adv) selvkontrahent n. (=selvintrceder) ‘contracting on one’s own behalf selvkontrahering n. (=selvintrceden) ‘contracting on one’s own behalf kontrahere vi. contract, make a contract a adv {??) ‘on one’s own behalf selvkontrol n. (=setvbeherskelse) ‘self-control. kontrollere vt. control a do selvkritik n. ‘self-criticism’ selvkritisk adj. ‘self-critical’ kritisere vt. criticize a do selvkrojende adj. "-melle mill automatically keeping in the wind’ kroje (spellingvariantofkroge {??}) kr0ge vi. bend kroge sig bend, stoop intr adv selvkorende adj. ‘self-propelling, self-propelled (fx gun)’ k0re V I. drive, go, (abt car) run, go, travel intr adv selvkdren adj. (=selvbestaltet) adj. ‘self-appointed, self-constituted’ kdre vt. choose, elect, a do selvlade- adj- prefix-‘automatic (pistol (lit.) self-loading)’ sehladende adj. ‘automatic’ selvladepistol n. ‘automatic pistol’ selvladevdben n. ‘automatic firearm’ lade vt. load (a gun), etc. a (do) (adv) {does the gun load itself, or does it load (middle) by itself?} selvlavet adj. ‘home-made, of one’s own making’ lave vt. make, a adv selvlede n. ‘self-loathing’ lede n. disgust, loathing; distaste for. {no verbal base} (do)?? selvlemlcestelse n. ‘(jur.) self-mutilation’ lemlceste vt. mutilate a do selvlossende adj. ‘self-discharging’ selvlossepram n. ‘self-discharging lighter, hopper barge’ losse vt. (naut.) unload, discharge, unship a (do) (adv) selvlukke n. ‘med ~ self-closing’ selvlukkende adj. ‘self-closing (fx door)’ {proof that lukke is not a true inherently reflexive but a middle?} selvlukker n. ‘self-closing device; (for dor) automatic door closer’ lukke vt. close (fx a door) lukke sig vi. close (fx a door closes) lukke sig a f sig seiv (e.g. the door closes by itself) (a) (n) (do) (adv)?? R eproduced with perm ission of the copyright owner. Further reproduction prohibited without perm ission. 461 selvlyd n. ‘vowel; sonant’ selvlydsrim n. ‘assonance’ selvlydssam m enstod n. ‘hiatus’ 1yd n. sound {no verbal base} adv?? ‘by itself selvtysende lyse selvlcegende seM censende Icense adj. ‘luminous; luminescent’ vi. shine, give out light mtr adj. ‘(med.)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The independence of binding and intensification

PDF

The evolving vocabulary of otherness in pre-imperial China: From 'belligerent others' to 'cultural others'

PDF

On the identification of null arguments

PDF

Linguistic attitudes in the Basque Country: The social acceptance of a new variety

PDF

Two types of dependency

PDF

Overt focus movement and minimal information marking

PDF

The grammaticalization of present and past in Basque

PDF

The effect of stress and foot structure on consonantal processes

PDF

The effect of non-native speaker status on the use of linguistic accommodation by native speakers of Japanese: Implications for communication accommodation theory

PDF

The modal preterite phenomenon (MPP) in colloquial American English: A diachronic and synchronic analysis

PDF

Second language acquisition of Spanish morpho -syntax by Quechua -speaking children

PDF

Discourse functional units: A re-examination of discourse markers with particular reference to Spanish

PDF

Grammaticalization and the development of functional categories in Chinese

PDF

Form and meaning: Negation and question in Chinese

PDF

Representational richness in phonological development

PDF

At wit's end: The rhetoric of humor and the ends of talk

PDF

Syntactically triggered VP information focus

PDF

The face of others, the taste of things: Photogenie and sensation in silent cinema

PDF

Children's scope of indefinite objects

PDF

Issues in the syntax of resumption: Restrictive relatives in Lebanese Arabic

Asset Metadata

Creator Bergeton, Uffe (author)

Core Title The independence of binding and intensification

Degree Doctor of Philosophy

Degree Program Linguistics

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Language, General,language, linguistics,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-503170

Unique identifier UC11335800

Identifier 3140435.pdf (filename),usctheses-c16-503170 (legacy record id)

Legacy Identifier 3140435.pdf

Dmrecord 503170

Document Type Dissertation

Rights Bergeton, Uffe

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA