University of Southern California Dissertations and Theses

The independence of binding and intensification

(USC Thesis Other)

The independence of binding and intensification

PDF

Download

Open document

Flip pages

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 U ffe Bergeton UNIVERSTITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNNERSITY PARK LOS ANGELES, CALIFORNIA 90089-1695 This dissertation, written by Uffo Bergeton under the direction ofhis dissertation committee, and approved by all its members, has been presented and accepted by the Director of Graduate and Profossional Programs, in partial folfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Date May 14, 2004 Co-Chair Co-Chair II Dedication Til mine foneldre Henning A. V. Larsen og Inger J. Larsen iii Acknowledgements This dissertation is built on a large body of accumulated insight into the workings of human language in general and reflexives and intensifiers in particular found in the very rich literature on these topics. While I have tried to give credit where credit is due I fear that, in many cases, I may have failed to explicitly state in what respect and to what extent I depend on the research of others. My hope is that these omissions may be forgiven. In addition to the existing linguistic literature I am also profoundly indebted to a large number of people who have provided me with invaluable support and assistance, both personal and professional, throughout the process of writing this dissertation. Since this acknowledgement section is too brief to adequately express my gratitude to everybody, I hope that those not mentioned here will be able to forgive me and trust that the lack of mention should not be translated into lack of gratitude. Needless to say, while all of the people mentioned here have contributed to the dissertation in one way or another, directly or indirectly, and share the credit of the better parts, only I am responsible for any inconsistencies, errors and other shortcomings it may still contain. First of all I'd like to thank my faculty advisors at USC, Roumyana Pancheva, Philippe Schlenker, and Jean-Roger Vergnaud. Beginning in chronological order, Jean-Roger Vergnaud has been an invaluable source of both academic and administrative support ever since I came to USC. Since he guided me as faculty advisor during the initial stages of my research the impact of his input cannot not be overestimated. Philippe Schlenker, who was my advisor from 1999 to 2001, and co-chair of my Ph. D. committee from 2002 to 2004, has been pivotal to the development and refinement of my ideas. IV Without his sharp reasoning most of the ideas in the dissertation would still be entangled in conceptual confusion and unclarity. Roumyana Pancheva, who accepted the task of being my committee chair after Philippe left USC in 2001, has been no less important in helping me turn vague intuitions into coherent and explicit analyses. The late-insertion analysis of reflexives in chapter 4 and the analysis of the evolution of English reflexives in chapter 7 were developed in close collaboration with her. Many other parts ofthe analysis of both reflexives and intensifiers were worked out during our many meetings over the last few years which have, without a doubt, contributed the most to making the final product into a readable and fairly coherent whole. In addition to my advisors, a number of other faculty members at USC have also had a large impact on my work. Among them are, of course, my Ph.D. Committee Members, Joseph Aoun, James Higginbotham, Audrey Li, and Edward Slingerland. I thank them for their helpful comments and encouragement and not the least for having had the patience to read through earlier drafts of the dissertation. Special thanks also go to Jack Hawkins, Barry Schein, and Maria Luisa Zubizarreta for being in my screening paper committee and for their classes at USC which have contributed to forming my thinking as a linguist. Finally, I also want to thank Jean Lowenstamm, my faculty advisor at the University of Paris VII for his instruction. Without his suggestion that I apply for graduate school in the Unites States I would never have come to USC in the first place. And without his inspiring approach to the syntagmatic study of the sound structure of human language (i.e. Government Phonology) I would never have decided to pursue a career in formal linguistics. The student community in the Department of Linguistics and my circle of friends (including Shadi Ganjavi, Aaron Sonnenschein, Junichi Hayashishta, Barbara Lohse, Karine v Megerdoomian, Lina Choueiri, and many others) have provided the social setting without which I would not have thrive in Los Angeles. Last but not least I am deeply indebted to my girlfriend Iris Yim ( t±}{IR:) for having had the patience to live with a graduate student working on his dissertation. Without her unfailing support and encouragement I could never have completed this work. Table of Contents Dedication Acknowledgements Abstract Volume I Chapter 1 Introduction 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10 1.11 1.12 1.13 1.14 1.15 The failure of most current binding theories to correctly separate binding and intensification Independence of binding and intensification in Danish Revisiting the analysis of simple and complex reflexives in Danish Adnominal intensification Predicate meaning and intensification of reflexives Intensification and the meaning difference between of simple and complex reflexives Logophoricity and intensified pronouns Possessive reflexives, pronouns and intensifiers Refuting anti-locality Self-nominalizations Unification of all uses of adnominal expressions Extending the analysis to Modern English Historical evolution of English reflexives and intensifiers Extending the analysis to Mandarin Chinese Conclusion VI II iii xvi 2 5 9 12 15 18 20 22 26 30 30 31 34 35 39 VII Chapter 2 Intensification 2.1 Introduction 4I 2.2 Adnominal intensification 45 2.2.I The morphology of adnominal intensification 45 2.2.2 The syntax of adnominal intensification 48 2.2.2.1 Possible syntactic positions of adnominally intensified 48 nominal expressions 2.2.2.2 The syntactic relation between intensifier and its focus: 50 adjunction 2.2.2.3 Unified analysis of adnominally intensified expressions 54 2.2.3 Semantic and pragmatic aspects of adnominal intensifiers 55 2.2.3.1 The semantic contribution the adnominal intensifier selv 56 'himself 2.2.3.2 Semantic and pragmatic constraints on adnominal 60 intensification 2.2.3.2.1 Unique identifiability 60 2.2.3.2.2 Prominence, centrality or contrastiveness 62 2.2.4 Is adnominal intensification of PRO possible? 67 2.2.5 Selv used as q-tloated adnominal intensifier 72 2.2.6 Fronted selv 85 2.2.7 Adnominal intensification: unification of different sub- 88 cases 2.3 Selv in secondary predication constructions 89 2.4 Selv used as scalar additive focus particle meaning 'even' 91 2.4.1 Syntax of the scalar, additive focus particle selv 'even' 95 2.4.2 Semantics ofthe scalar, additive focus particle selv 'even' 97 VIII 2.5 Selv used as a noun 102 2.6 Adjectival forms of selv: selve and selveste 104 2.7 Selv 'self in nominalizations 108 2.8 Idiomatic uses of the element selv 109 2.9 Cross-linguistic perspective on the range of uses of II 0 intensifiers 2.10 Conclusion Ill Chapter 3 Binding: Reflexives and Pronouns in Modern Danish 3.1 Introduction. Binding of reflexives and pronouns in 115 Danish 3.2 Basic properties of Danish reflexives 119 3.2.1 Morphological characteristics of reflexives in Danish 119 3.2.2 Semantics of reflexives 121 3.2.3 Syntactic structure of reflexives 124 3.2.4 Consequences of relegating binding and intensification to 129 different modules: a syntactic approach to binding of reflexives: principle A 3.3 Binding of reflexives in different contexts: testing the 141 independence of binding and intensification 3.3.1 The independence of intensification and locality 141 constraints 3.3.2 Locally bound reflexives in argument position: complex 146 reflexives= a sub-type ofadnominal intensification 3.3.2.1 Anti-reflexives predicates: predicates which are 148 incompatible with reflexive scenarios 3.3.2.2 Predicates which are semantically incompatible with non- 153 reflexive scenarios 3.3.2.3 Neutral predicates: predicates which are semantically 155 compatible with reflexive scenarios and which are not semantically incompatible with non-reflexive scenarios IX 3.3.2.4 Summary: Presuppositions triggered by predicate 159 meaning 3.3.2.5 Complex reflexives as arguments of proto-typical neutral 160 predicates 3.3.2.6 Using inalienable possession to test whether 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 Selv as marker of statue-readings 167 3.3.2.7.2 Qua-sentences 168 3.3.2.7.3 Strict readings of reflexives in VP-ellipsis 171 3.3.2.8 Complex reflexives and distributivity 173 3.3.2.9 Summary 175 3.3.3 Reflexives in resultative constructions 176 3.3.4 Reflexives in ECM constructions 179 3.3.5 Complex reflexives and pronouns in possessor position 180 3.3.6 Binding of simple and complex anaphors in PPs and 185 prepositional predicates: phonological factors affecting intensification of reflexives 3.3.6.1 Stressability of reflexives: prosodic uses of selv = stress 186 carrying element 3.3.6.2 Intensifier-adjunction to reflexives in PPs and the 188 interaction with sentential stress 3.3.6.3 Can intensification of reflexives be reduced to lack of 203 stressabi I ity? 3.3.6.3.1 Stressable reflexives in French 203 3.3.6.3.2 Stressable reflexives in German 204 3.3.6.3.3 Stressable pronouns in Danish 205 3.3.6.3.4 Reducing intensification of reflexives to unstressability 206 X 3.3.7 Summary 208 3.4 Distribution of simple and complex pronouns 209 3.4.1 Binding of pronouns: principle B 210 3.4.2 Intensified object pronouns 212 3.4.2 Intensified subject pronouns: Intensified pronouns as 220 subjects of embedded clauses in English and Danish 3.4.3 Logophors and logophoricity vs. intensification and 224 intensified pronouns 3.4.4 Summary 226 3.5 Conclusion: The independence of binding and 227 intensification Chapter 4 The syntactic status of sig 4.1 Introduction 231 4.2 Testing the lexical/syntactic status of sig with neutral 237 verbs 4.2.1 Testing for agentivity 239 4.2.2 Testing for compatibility with formation of agentive 245 nouns in -er 4.2.3 Non-compositional meaning change and the question test 247 4.2.4 Testing for compatibility with impersonal passivization 249 4.2.5 Auxiliary-selection 252 4.2.6 L0drups tests 254 4.2.7 German sich is not anti-local 255 4.2.8 Constructions involving coordination of verbs 258 4.2.9 The status of sig as indirect object of ditransitive verbs 261 4.2.10 Non-co-argument bound local sig 262 4.2.11 Summary: Falsifying the standard analyses of sig as anti- local 4.3 The status of sig with inherently reflexive verbs 4.3.1 Inherently reflexive predicates with non-thematic sig 4.3.2 "Inherently reflexive verbs" with thematic sig 4.3.3 Grammatical uses of inherently reflexive verbs with sig selv 4.3.4 Conclusion 4.4 Late-insertion analysis of reflexives and pronouns in Danish 4.4.1 Lexical and syntactic properties of pronouns and reflexives 4.4.2 Binding theory in 1st and 2nd person: the implications of person asymmetries 4.5 Reflexive and non-reflexive uses of reflexive elements: How many sig's do we have? 4.5.1 Use of reflexive elements in reflexive constructions 4.5.2 Use of reflexive elements in reciprocal constructions 4.5.3 The use of reflexive elements in middle constructions 4.5.4 Use of reflexive elements in passive constructions 4.5.5 Deponent verbs 4.6 Conclusion Volume II Chapter 5 Independence of intensification and binding in English: synchronic and diachronic perspectives 5.1 Introduction XI 263 263 267 268 273 274 275 275 279 281 287 288 293 295 297 301 304 Xll 5.2 Intensifiers, 0-reflexives and intensified pronouns in 308 Modern English and the independence of intensification and binding 5.2.1 Binding of 0-reflexives in different contexts: testing the 309 independence of binding and intensification in Modern English 5.2.1.1 Locally bound 0-reflexives in argument position: 309 complex reflexives= a sub-type of adnominal intensification 5.2.1.1. 1 Anti-reflexive predicates 311 5.2.1.1.2 Inherently reflexive predicates 312 5.2.1.1.3 Neutral predicates 314 5.2.1.1.4 Non-optionality of intensifier-adjunction to 0-reflexives 318 with proto-typical neutral predicates: focus and doppelganger-effects 5.2.1.1.5 Ditransitives 319 5.2.1.1.6 Linking Doppelganger-effects and anti-reflexivity 320 5.2.1.1.6.1 X-self as marker of statue-readings 320 5.2.1.1.6.2 Qua-sentences 321 5.2.1.1.6.3 VP-ellipsis and the sloppy vs. strict reading of reflexives 321 5.2.1.1.7 Summary 322 5.2.1.2 Resultatives and the 0-reflexive analysis of English 324 5.2.1.3 ECM constructions the 0-reflexive analysis of English 325 5.2.1.4 Adnominal intensifiers in possessive constructions 326 5.2.1.4.1 Absence of inalienable possession in English 328 5.2.1.5 0-reflexives in PPs and prepositional predicates in 329 English 5.2.1.6 Potential problem: stressed and unstressed forms of 332 himself xiii 5.2.2 Non-reflexive uses of 0-reflexives in English 335 5.2.2.1 Reciprocals 335 5.2.2.2 Middles 336 5.5.2.3 Inherently reflexive/deponent verbs 337 5.2.2.4 Summary 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 of intensified and unintensified 340 pronouns 5.2.4 Conclusion 340 5.3 Historical development of Danish and English intensifiers 341 5.3.1 The historical development of the Germanic/Modern 341 Danish intensifier selv 5.3.2 Previous account of the evolution of the Modem English 343 self-forms 5.3.3 A new account of the evolution of the Modern English 348 self-forms 5.3.3.1 Stage one: intensification and binding in Old English 348 5.3.3.2 Stage two: intensification and binding in Early Modem 354 English 5.3.3.3 Stage three: intensification and binding in Modern 363 English 5.3.3.4 Additional evidence from the evolution of pronominal 363 reciprocals into 0-reciprocals 5.3.4 Summary 364 5.4 Conclusion 367 xiv Chapter 6 Independence of Binding and Intensification in Mandarin Chinese 6.1 Introduction 369 6.2 Setting the stage: evidence supporting the assumptions 369 that ziji is always an intensifier and Chinese has 0- reflexives 6.3 Internal structure of adnominal intensifiers in Chinese 380 6.3.1 Is ziji really mono-morphemic? 380 6.3.2 Ta ziji: complex intensifier, complex reflexive 0 ta ziji, or 383 intensified pronoun ta ziji? 6.4 Analyzing what appears to be uses of ziji as reflexive 387 anaphors as adnominal intensifiers 6.4.1 Locally bound 0-reflexives in argument position: 387 complex reflexives = a sub-type of adnominal intensification 6.4.1.1 Anti-reflexive predicates 387 6.4.1.2 Neutral predicates 389 6.4.1.3 Inherently reflexive predicates 394 6.4.1.4 Predicate types: summary 395 6.4.2 LD-bound ziji: complex pronoun or complex reflexive? 396 6.5 Intensifier analysis of local and LD ta ziji 396 6.5.1 Locally bound ta ziji: complex pronoun or complex 397 reflexive? 6.5.2 LD-bound ta ziji: complex pronoun or complex reflexive? 398 6.6 Intensified and unintensified pronouns in Mandarin 400 Chinese 6.6.1 Simple/unintensified pronouns in Mandarin Chinese: 400 Principle B 6.6.1.1 Overlapping distribution of simple ta and zij1lta ziji 401 6.6.1.1.1 Possessive constructions 401 XV 6.6.1.1.2 Embedded subject pronominals 402 6.7 Conclusion 405 Chapter 7 The element selv 'selr in nominalizations 7.1 Introduction 407 7.2 Different uses of the morpheme selv 'self in derived 409 nouns and adjectives 7.2.1 Word-internal selv 'self appearing to behave as a 409 "reflexive" while it really is an adnominal intensifier 7.2.2 Word-internal selv 'self behaving as an adverbial 413 7.2.3 Problematic cases 414 7.3 Cross-linguistic perspective on use of intensifiers in 416 derived words 7.3.1 French 416 7.3.2 Chinese 418 7.3.3 Summary 420 7.4 Outline of analysis of selv 'self in derived nouns and 420 adjectives Chapter 8 Conclusion 424 References 431 Appendix I Danish se/v-compounds 447 Appendix II Intensified and unintensified pronouns and reflexives in Danish 469 XVI Abstract A new approach to the interaction of binding and intensification is advanced. While most current approaches to (e.g. Reinhart&Reuland 1993, Huang&Liu 2001, among others) take the selv element in so-called SELF-anaphors (e.g., Dan. sig selv) to be itself a reflexive, we argue that it is in fact an adnominal intensifier (cf. Baker 1995, and Konig&Siemund 2000). Thus we achieve a unified account of all types of intensified nominal expressions, be they intensified DPs (e.g., Dan. Peter selv 'Peter himself), intensified reflexives (e.g., Dan. Peter hader sig selv 'Peter hates REFL self), or intensified pronouns (e.g., Dan. Peter sagde at Mary dansede med alle andre end ham selv "Peter said that Mary danced with everyone except himself.') - something most 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 which binding is defined as the interaction of intrinsic properties of nominal expressions and syntactic locality constraints. We also argue that the Danish reflexive sig (= Dutch zich, Norwegian seg, etc.) is not an anti-local anaphor specialized for long-distance binding, thus falsifying those binding theories which crucially rely on this assumption. Finally, concerning intensification, we argue that the semantic contribution of adnominal intensifiers (e.g. Danish selv, Eng. himself) is not similar to that of scalar focus particle ( cf. Eckardt 200 I) or reducible to centrality-effects ( cf. Konig (1997) and Siemund (2000)), but rather very similar to that of contrastive focus, which - depending on the context - may or may not involve scalar ordering of focus-generated alternatives. The analysis is extended to English and Chinese where himself and ziji 'self-self are argued always to be intensifiers (:::::: Dan. selv) modifying 0-reflexives (:::::: Dan. sig), e.g. Peter, shaved 0, himself. We show that this analysis provides answers to hitherto unanswered questions related to the evolution of Modern English XVII reflexives and intensifiers from Old English which had a monomorphemic selfintensifier and no reflexive pronouns. Chapter 1 Introduction This dissertation presents an approach to adnominal intensifiers and simple and complex reflexives and pronouns based on the idea that intensification and binding constitute separate modules of the grammar. While the separation of intensification and binding into different modules is hardly controversial, the question of where exactly to draw the dividing I ine is far from trivial, especially in the case of complex reflexives and complex pronouns. It will be argued that most current accounts of binding (e.g. Reinhart and Reuland (1993), Huang and Liu (200 I), among others) are misguided in that they often take what are really adnominal intensifiers for reflexive elements. In addition to achieving a descriptively and explanatorily more adequate account of intensified reflexives and pronouns, the approach defended here also has the advantage of being able to achieve a unified account of all types of intensified nominal expressions, be they intensified DPs (e.g. Dan. Peter selv 'Peter himself), intensified reflexives (e.g. Dan. Peter hader sig selv 'Peter hates REFL self), or intensified pronouns (e.g. Dan. Peter sagde at Mary dansede med alle andre end ham selv "Peter said that Mary danced with everyone except him self.') - something current approaches to binding and intensification are unable to do. Though most of 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 of binding and intensification the present proposal also has consequences for each of these modules individually. In contrast to predicate-based approaches to binding (e.g. Reinhart & Reuland 1993), we call for a nominal approach to binding in which binding is defined as the interaction of intrinsic properties of nominal expressions and syntactic locality constraints. We also argue that the Danish reflexive sig (= Dutch zich, Norwegian seg, etc.) is not an anti-local anaphor specialized for 2 long-distance binding, thus falsifying those binding theories which crucially rely on this assumption. Finally, concerning intensification, we argue that the semantic contribution of adnominal intensifiers (e.g. Danish selv, Eng. himself) is not similar to that of scalar focus particle (cf. Eckardt 2001) or reducible to centrality-effects (cf. by Konig (1997) and Siemund (2000)), but rather very similar to that of contrastive focus which - depending on the context- may or may not involve scalar ordering of focus-generated alternatives. 1.1 The failure of most current binding theories to correctly separate binding and intensification The main problem with most existing analyses of binding is that they do not take into account the fact that in many languages the elements which serve as reflexives are either identical to the elements serving as adnominal intensifiers, e.g. English himself(1 ), Chinese b3 c ziji 'self (2), and French lui-meme 'him-same' (3), or partially overlaps with adnominal intensifiers, e.g. the Danish complex reflexive sig selv 'REFL self (4a) which is composed of the simple reflexive sig 'REFL' (4b) plus the adnominal intensifier selv 'self (4c). In the examples in (1- 4) bold has been used for expository reasons to highlight the elements functioning as ad nominal intensifiers/reflexives 1 • (1) (2) a. b. Peter hated himself Peter himself attended the conference. Mandarin Chinese: a. IJR~tNS co Zhangsan hen ziji. Zhangsan hate self 'Zhangsan hates himself.' (himself= reflexive) (himself= intensifier) (ziji =reflexive) 1 See chapters 5-6 for analyses of English and Chinese based on the assumption that himself and ziji § c ·self-self always function as intensifiers. b. (3) French: iJ:R ~ ~ C3$1JO~§~ 0 Zhangsan ziji canjia huiyi. Zhangsan self attend conference. Zhangsan himself attended the conference. (ziji =intensifier) a. Pierre estjaloux de lui-meme. (lui-meme =reflexive) Peter is jealous of him-same 'Peter is jealous of himself.' b. Pierre lui-meme a assiste ala conference. (lui-meme =reflexive) Peter him-same has participated to the conference (4) Danish: a. 'Peter himself participated in the meeting.' Peter vaskede sig selv. Peter washed REFL self 'Peter washed himself.' (sig selv =complex reflexive) b. Peter vaskede sig. (sig= simple reflexive) Peter washed REFL 'Peter washed (himself).' c. Peter selv deltog i m0det. (selv =intensifier) Peter self participated in meeting-the 'Peter himself participated in the meeting.' 3 From the beginning of modem formal linguistics till the present day, most binding theoretical accounts of reflexives have tended to ignore the formal identity of intensifiers (l-3b) and reflexives (l-3a), or in the case of Danish, the identity of intensifiers (4c) and a sub-part of complex reflexives (4a). If mentioned at all, intensifiers have usually been assumed 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 many cases either been overlooked or tacitly assumed to be a historical accident not worth investigating. In the mid-nineties a number of researchers began to (re-)discover 2 the close links between intensifiers and reflexives and the important interactions of intensification and 2 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 of 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. Sec also the early typological study by Moravcsik ( 1972 ). 4 binding evident in the peculiar behavior of complex reflexives. Among these, McKay ( 1991 ), Baker (1995), Zribi-Hertz ( 1995), Veraart (\ 996), Konig and Siemund (1999), and Siemund (2000) have provided the inspiration for many of the insights on which the analysis presented here is based. Baker ( 1995) proposes to analyze what looks like locally free reflexives in English as intensified object pronouns. In a similar vein, McKay (1991) had suggested to analyze complex pronominal forms like he himself, she herself, etc. as intensified subject pronouns rather than as a special kind of anaphoric expression, as proposed by Bickerton (1987) and Iatridou (1986). We agree that adopting these proposals makes it possible to correctly separate intensification and binding into independent modules ofthe grammar. That is, a syntactic principle B accounts for the distribution of pronominals (e.g. he, him) while the distribution of intensifiers (e.g. himself) is subject to the semantic and pragmatic principles of intensification. At the end of his paper Baker ( 1995) suggests that this analysis may be extended to the analysis of complex reflexives in German. Stated in the terminology of this dissertation, his idea is basically that the complex reflexive sich selbst 'REFL self' is formed from its simple counterpart sich 'REFL' by the same process of adnominal intensification which is responsible for the formation of intensified DPs, e.g. der Konig selbst the king self 'the king himself' (see Konig and Siemund (1999) for an analysis of German along those lines). One of the goals of the present dissertation is to show that by adopting this approach it becomes possible to unifY the account of all adnominally intensified expressions, be they reflexives, pronouns or DPs - something previous accounts of binding and intensification are unable to do. The present dissertation is thus in part an attempt to flesh out the research project outlined at the end of Baker (1995): "The aim of such work would be to gain a much clearer picture than now exists of the essential nature of locality, discourse prominence, and contrast, considered as individual subsystems of grammar, and also a better understanding of 5 the ways in which they interact to produce the extremely complex phenomena that we encounter in actual language use." (Baker 1995:99). That is, we argue that the so-called complex reflexives (e.g. Danish sig selv 'REFL self are best analyzed as adnominally intensified counterparts of the simple reflexive (e.g. Dan. sig 'REFL'). By relegating reflexives anaphors and intensifiers to different modules it becomes possible to adopt a nominal approach to binding theory in which the distribution of anaphors (e.g. sig) and pronouns (Dan. ham 'him') follow from the interaction of lexical properties of nominal expressions and syntactic principles (e.g. LGB-style principles A and B). In contrast, the distribution of the adnominal intensifier selv 'self is accounted for by the semantic/pragmatic principles of intensification. That is, adnominal intensifiers (e.g. Dan. selv 'self as used in Peter selv 'Peter himself), will be given a Roothian focus-based analysis (cf. Rooth 1992). In brief, the main idea consists in analyzing ALL the elements highlighted in bold in the examples in (1-4) as adnominal intensi tiers. 1.2 Independence of binding and intensification in Danish In this section, we show that due to its morphologically transparent articulation of intensification and binding, Danish is particularly well suited to test the proposals defended here. As iilustrated in (1-3), in many languages, including English, the mutual independence of binding and intensification is obscured by what appears to be a total morphological overlap between the elements uses as intensifiers and reflexives, see (5). 6 ( 5) System of intensification of nominal expressions in English. Simple!Unintensified Complex/Intensified a. Reflexive (himself) himself b. Subj.pronoun he he himself c. Obj.pronoun him himself (<him himse(N d. DP the king the king himself Other languages, e.g. the Mainland Scandinavian languages, Dutch, German etc., have morphologically distinct realizations of intensifiers and reflexives, e.g. the Danish intensifier selv 'self and the simple reflexive sig 'REFL', see (6). ( 6) System of intensification of nominal expressions in Danish. Simple/Unintensified Complex/Intensified a. Reflexive sig 'himself/herself sig se/v 'himself/herself b. Subj. pronoun han 'he' han selv 'he himself c. Obj. pronoun ham 'him' ham selv 'him himself/himself d. DP kongen 'the king' kongen selv 'the king himself As shown in (5-6) the English system of nominal expressions is more opaque that the Danish system. The English form himselftranslates into four different expressions in Danish, viz. the simple/unintensified reflexive sig 'REFL', the complex/intensified reflexive sig selv 'REFL self, the intensified pronoun ham selv '(him) himself, and the adnominal intensifier selv 'himself. Figure I below further illustrates the mutual independence of the reflexive sig 'REFL' and the intensifier selv 'self in Danish as well as their 'overlap' in the so-called complex, viz. intensified, reflexive sig selv 'REFL self. The examples in figure I illustrate most of the different uses of sig 'REFL' and selv 'self in figure I and (6a-d). 3 The system in (5) represents the standard analyses of nominal expressions in English in which himself in (Sa) is assumed to be an anaphor (e.g. Peter; washed himself,) and himself in (5c) to be a logophor/Locally Free Retlexive(LFR) (e.g. Peter, said that Mary danced with everybody except himself,). In contrast, as indicated in (5c), we follow Baker 1995 in assuming the latter use of himself to derive from an underlying intensified pronoun him himself See chapter 5 for more detailed discussion of intensified pronominals in English. We also analyze himself in (Sa) and (I a) as an intensified 0 reflexive, Peter, hated 0, himself See chapter 5 for more detailed discussion of intensified 0 reflexives. See also section I. I 2 where this alternative analysis English is summarized in the table in (53). 7 Figure I. The intersection of the intensifier selv 'self and the reflexive sig 'REFL' and the independence of intensification and binding. selv sig (ii) (b) focus part. "middles" (iii) excl.adv (iv) incl.adv (v) (d) logophor inherently reflexives Examples illustrating the different uses of selv and sig in Figure 1: (i) Peter selv loste 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 laste opgaven selv. Peter solved problem-the self 'Peter solved the problem himself.' (iv) Peter havde selv last opgaven. Peter had self solved problem-the 'Peter had himself solved the problem.' (v) Peter sagde at Marie dansede med aile andre end ham selv. Peter said that Marie danced with all other than him self (c) deagentive inchoatives 'Peter said that Marie danced with everybody except (him) himself.' ((a)+(i)) Peter vaskede sig selv. Peter washed REFL self 'Peter washed himself.' (a) Peter vaskede sig. Peter washed REFLE 'Peter washed.' (b) Lceberne l0snede sig, munden blev st0rre. lips-the loosen REFL mouth-the became bigger 'The lips parted, the mouth grew bigger.' (c) Peter slog sig. Peter hit REFL 'Peter got hurt (accidentally).' (d) Peter skammer sig. Peter shames REFL 'Peter is ashamed.' 8 After setting up the independence of intensification and binding as the main hypothesis to be tested, we walk trough all the cases of overlap between these modules (e.g. the different uses of intensified reflexives sig selv, intensified pronouns ham selv 'him self, and intensified R- expressions Peter selv 'Peter himself) as well as all the different uses of selv and sig on their own. The core function of the element selv 'self is as a marker of adnominal intensification, e.g. sentence (i) in figure 1. The core function of the element sig 'REFL' is as a reflexive pronoun, e.g. sentence (a) in figure 1. The behavior of selv 'self and sig 'REFL' is thus determined by different parts of the grammar, viz. the modules of intensification and binding respectively. As figure 1 indicates, even when sig and selv come together to form the so- called complex reflexive sig selv, see ((a)+(i)) above, these two elements still fall under two different modules of the grammar. Besides their core functions, both selv 'self and sig 'REFL' have a number of other uses. The sentences in (ii)-(iv) illustrate the use of selv 'self as an additive scalar focus particle similar in meaning to English even (ii), as a manner adverbial meaning 'by oneself, without help' (iii), and as a quantifier-floated adnominal intensifier which is usually interpreted similarly to an inclusive particle meaning 'also, too' (iv). As will be shown in chapter 2, one of the advantages of the focus-based analysis presented here is that it allows for the potential unification of both core- and non-core uses of the element selv 'self. 9 The sentences in (b)-(d) illustrate the non-reflexive uses of the reflexive sig 'REFL' in middle constructions, deagentive unaccusative predicates and inherently reflexive predicates. The main distinction between reflexive and non-reflexives uses lies in their thematic status: reflexive sig is a theta-marked argument merging in argument position of a transitive predicate, see (a) in Figure 1, while non-reflexive sig (b-d) is non-thematic since it merges with the small v of unaccusative (forms of) predicates. Reflexive uses of simple reflexives, e.g. Dan. sig 'REFL', will be treated in chapter 3 where we also argue that sentences like (v) support analyzing so-called "logophors" (e.g. Dan. ham selv 'him himself') as adnominally intensified pronouns (cf. McKay 1991 ). Non-thematic uses of reflexive elements will be discussed in more detail in chapter 4. 1.3 Revisiting the analysis of simple and complex reflexives in Danish Since Faltz's ( 1977, 1985) observation that complex reflexives have a tendency to be locally bound while simple reflexives allow for long-distance binding, most accounts of reflexives have aimed to explain the correlation between the morphological form of reflexives and locality constraint within the binding theory. Faltz's difference between simple and complex reflexives has been claimed to be found in Danish, see (7). (7)4 a. Peter, ajloste * . szg, I sig, selv. Peter replaced REFL I REFL self 'Peter, replaced himself,' b. Peter, bad Hansk ajlose sig; *k I sig., k selv. Peter asked Hans replace REFL I REFL self 'Peteri asked Hans to replace himj.' In the binding literature, examples like (7a,b) have been adduced as evidence that the simple reflexive sig is anti-local and specialized for LD-binding while the complex reflexive sig selv requires a local antecedent. That is, Danish has been used to support the theory that the 10 morphological complexity of anaphors interacts directly with locality. Indeed, most current theories of binding assume that the correlation between morphological complexity and locality should be encoded in the binding principles themselves. Vikner (1985) proposes that simple sig is an anti-local anaphor while sig selv is a complex anaphor in which presence of selv has the direct effect of restricting the binding domain of the anaphor. According to the proposal defended here, Vikner's view is misguided in several ways. First, his approach does not take the uses of selv 'self as an intensifier, e.g. (4c) into account. Second, there is strong evidence in favor of assuming theta-marked simple sig to be a locally bound theta-role receiving reflexive -thus falsifying claims that sig is an anti-local anaphor specialized for long-distance binding, see (8a,b). (8) a. b. Peter, vaskede sig, I Peter washed REFL I 'Peteri washed (himself;).' Peter, bad Hansk vaske sig, k I Peter asked Hans wash REFL I sig, selv. REFL self sig., k selv. REFL self 'Peteri asked Hansk to wash him/himselfk.' In Reinhart & Reuland's (1993) terminology, the simple vs. complex distinction is referred to as a distinction between (simple) SE-anaphors and (complex) SELF-anaphors. They propose a predicate-centered binding theory in which SELF is a reflexivizing element whose function is to reflexive-mark predicates that are not lexically reflexive. That is, they argue that it is selv 'self' rather than sig 'REFL' which is the true reflexive element. In order to maintain this view, they are forced to assume that simple sig can be locally bound only when occurring as argument of inherently reflexive predicates, i.e. predicates which are lexically marked for reflexivity. As a consequence, they have to stipulate the existence of double entries for all the predicates which allow both locally bound simple reflexives and non-coreferential DPs as internal arguments, e.g. vaske 'wash' in (8a,b). In other words, Reinhart & Reuland's (1993) 4 We adopt the convention of using the star symbol'"*'" to indicate the an example is ungrammatical. For examples 11 analysis relies on an unwarranted stipulation of lexical ambiguity. To account for the fact that both the simple reflexive sig and the complex reflexive sig selv can be found in sentences like (8a) they need to stipulate the existence of two synonymous verbs meaning 'wash' 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 adnominally intensified version of sig. To account for the LD-binding of sig we adopt a Pica-style analysis based on successive cyclic movement of sig, first to the most local TP and then, potentially, to higher TPs at LF (cf. Pica 1984, 1986, and Jakubowicz 1994). However, in contrast to Pica who argues that it is the morpho-syntactic properties of sig selv which makes LD-binding impossible we suggest that the apparent locality constraint on sig selv is an epiphenomenal result of adnominally intensifying sig. That is, we follow Konig and Siemund ( 1999:63-4) in assuming that sig selv is an intensified reflexive whose semantic and syntactic properties can be explained as deriving compositionally from the properties of its constituent parts, i.e. the simple reflexive sig 'REFL' and the adnominal intensifier selv 'self 5 • The sentence in ((a)+(i)) in figure I above thus illustrated the overlap of the modules of binding and intensification which gives rise to the so-called complex reflexive sig selv 'REFL self, himself. Unlike Reinhart & Reuland's (1993) predicate-centered approach to binding we propose a nominal approach in which binding is defined as the interaction of intrinsic properties of nominal expressions and syntactic locality constraints ( cf. Chomsky 1981 ), see (9-10). (9) a. b. An anaphor is bound in a local domain. A pronominal is free in a local domain. which are merely pragmatically or semantically deviant/infelicitous the sharp symbol '"#" is used. 5 "[l]n many European languages (i.e. Norwegian seg selv. Spanish sf mismo, Russian sam 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). 12 (10) Binding Domain: Alpha is a binding domain for beta if and only if alpha is the minimal category (i.e. the smallest DP or IP/S) containing beta, a case-licensor of beta, and a SUBJECT accessible to beta. The binding theory outlined here will be developed in more detail in chapter 3. 1.4 Adnominal intensification Before we move on to showing how predicate meaning influences intensifier-adjunction to simple reflexives (see section 1.5), we need a theory of adnominal intensification. We argue that the semantic contribution of adnominal intensifiers (e.g. Danish selv, Eng. himse(f) neither always involves scalar ordering of focus alternatives ( cf Eckardt 2001) nor is reducible to centrality-effects (cf. Konig (1997) and Siemund (2000), but is rather very similar to that of contrastive focus which -depending on the context- may or may not involve scalar ordering of focus-generated alternatives. Unlike the focus particle selv 'even', the adnominal intensifier selv 'himself does not presuppose that the proposition is true for at least one other element in the focus-generated set of alternatives, see (II a) vs. ( 11 b). This presupposition is sometimes referred to as the existential presupposition/implicature. At this point no attempt has been made to distinguish between presupposition and implicature. The two terms are used interchangeably without difference in meaning in the rest of this dissertation. ( 11) a. b. *Del var selv biskoppen der holdt gudstjenesten. (selv ::::o even) It was self bishop-the who held service-the '*It was even the bishop who held the service.' Det var biskoppen selv der holdt gudstjenesten. (selv ::::o himself) It was bishop-the self who held service-the 'It was the bishop himself who held the service.' The example in (11a) shows that the focus particle selv 'even' cannot occur in clefts, which imply uniqueness, because that would lead to a contradiction of the so-called existential 13 implicature. Since the adnominal intensifier selv 'himself can occur in clefts, see (II b), we therefore conclude that it does not evoke this existential implicature which is mandatory with the focus particle even. The examples in (12)-(13), which contain predicates presupposing uniqueness, further confirm this difference between even and the adnominal intensifier. (12) a. b. (13) a. b. #Selv Dronningen gav nytarstalen. (selv ::::o even) even queen-the gave new-years-speech-the '#Even the Queen delivered the New Year's Speech.' #Even the President delivered the inaugural speech. Dronningen selv gav nytarstalen. (selv ::::o himself) queen-the self gave new-years-speech-the 'The Queen herself delivered the New Year's Speech.' The President himself delivered the inaugural speech. According to standard analysis of focus particles, the semantic contribution of even involves both the existential implicature illustrated (11) and a scalar implicature which orders the focus alternatives on a scale of expectedness. That is, in a sentence like Even the King came to the meeting the scalar implicature would rank the King lower on a scale of likelihood than any of its focus-generated alternatives, e.g. the Queen, the prince, the bishop, etc. Eckardt (2000) suggests that the semantic contribution of adnominal intensifiers is identical to that of the focus particle even except for the absence of existential implicature. We argue that this analysis of the semantic contribution of adnominal intensifiers in on the wrong track. While it is true that DPs adnominally intensified by selv often refer to entities which are remarkable/unexpected in a given context, this is by no means always the case, see ( 14-15). ( 14) A very powerful earthquake struck the center of Rome and the Vatican. a. The Pope himself 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. 14 Given that an earth quake does not differentiation between people of high and low status or prominence 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 remarkable location, but simply because it is contrasted with another location, viz. the suburbs. Notice that in both cases the adnominal intensifier is perfectly acceptable while the focus particle even, which always carries with it a scalar implicature, see (14b) and (15b), is not. We take this as evidence that the scalar implicature of even is NOT an integral part of the semantic contribution of adnominal intensifiers 6 • We therefore conclude that the semantic contribution ofthe adnominal intensifier involves neither existential nor scalar implicatures and therefore is similar to plain focus, see ( 16-17). The formula highlighted in bold indicates a proposition. ( 16) ( 17) a. b. a. b. [The king himselj]F came to the meeting. [The king], came to the meeting. Ordinary semantic value of(l6a): [[ [The king himself} came to the meeting]t = 1 iff the king came to the meeting. Focus semantic value of ( 16a): [[ [The king himself]F came to the meeting ]] r = {came-to-the-meeting(x) I XE De} 'the queen came to the meeting', 'the prince came to the meeting', etc. Like plain focus, the basic semantic contribution of adnominal intensification consists m contrasting the referent of the focused/intensified expression with a contextually determined set of alternatives, see ( 17b ). The contrastiveness condition on intensification in ( 18) captures the fact that intensification can only occur in contexts in where the generation of contrast sets of focus-alternatives is possible. 6 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 of 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.). 15 ( 18) Contrastiveness condition on adnominal intensification: A nominal expression (DP, pronoun, reflexive) can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. As shown in section 1.5, unlike the conditions on intensification proposed by Baker ( 1995:79- 80), Konig (1997), Siemund (2000: 154), the existence of ( 18) is supported by evidence from the distribution of complex reflexives. 1.5 Predicate meaning and intensification of reflexives In this section, we outline how the analysis of intensification proposed above can account for the interaction between predicate meaning and adnominal intensification of reflexives. One of the important contributions of the present dissertation consists in proposing a new view of the interaction between predicate meaning and the choice between simple/unintensified and complex/intensified reflexives. Early versions of the binding theory (e.g. Chomsky 1981) usually had very little to say about the interaction of predicate meaning and reflexives. Binding theories which do take predicate meaning into account usually encode the different behavior of predicates with binary lexical features (e.g. Reinhart & Reuland 1993, and Zribi- Hertz 1995). Zribi-Hertz ( 1995) showed that non-contrastive prepositional predicates in French differ with respect to whether the element used as adnominal intensifier (lui)-meme '(him-)same' is or is not adjoined to the locally bound pronoun in sentences I ike ( 19-21 ). (19) (20) Pierre estfier de lui/ lui-meme/Marie. Pierre is proud of him/him-same/Marie 'Peter is proud ofhimself/Marie.' Pierre est jaloux de *lui! lui-meme/Marie. Pierre is jealous of him/him-same/Marie 'Peter is jealous of himself/Marie.' (neutral predicate) (anti-reflexive predicate) This presupposition clashes with the expectation that. under normal circumstances. most people live in only one place. 16 (21) Pierre est hors de lui/*lui~meme/*Marie. (inherently reflexive predicate) Pierre is outside him/him-same/Marie 'Peter is besides himself/Marie (of joy/anger/etc.).' Zribi-Hertz (1995) suggests that the different behavior of predicates is encoded lexically with the feature [+/-disjoint reference]. R&R ( 1993) also resorts to lexical marking to account for the different behavior of predicates, i.e. [+/-inherently reflexive]. In contrast, the proposal defended here assumes no binding-specific lexical features. Instead, adnominal intensification of reflexives is assumed to be determined by pragmatic and semantic features of the governing predicate plus those ofthe discourse context in which it occurs. That is, predicate meaning is just one among many contextual factors which may influence adnominal intensification of reflexives. As mentioned in section 1.4, we propose a focus-based analysis of adnominal intensification. The intensifier selv 'self is associated with focus and evokes a contextually defined set of alternatives to its associate, i.e. the host DP to which it is adjoined. Unlike the focus particle selv 'even', the adnominal intensifier selv 'self does not obligatorily involve scalar ordering of alternatives. That is, it is the surrounding context which detennines whether or not the alternatives are ordered (with the associate DP being defined as the end-point of a contextually defined scale) or whether the associate is simply singled out as 'prominent'. When adjoined to the reflexive sig, the intensifier, because of its meaning, will be compatible with some predicates (22), obligatory with others (23), and impossible with still others (24). Descriptively, predicates can thus be divided into three types determined by the distribution of different kinds of nominal expressions in argument position: (i) those which can take as direct object the unintensified reflexive sig, the intensified reflexive sig selv and DPs, (ii) those which can take sig selv and DPs but not unintensified sig, and (iii) those which can only take unintensified sig, see (22-24). The tenns "neutral", "anti-reflexive" and "inherently reflexive" 17 does not imply that we assume the existence oflexically determined predicate-classes. That is, these three terms 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. Konig & Siemund (1999) also distinguish between different predicates classes. What he calls "other-directed" verbs corresponds more or less to what is referred to as "anti-reflexive" verbs here. See also Larsen (1997) who uses similar terms to refer to different verb-types. (22) "Neutral" Qredicates: a. Peter vasker sig I sig selv I bilen. Peter washes REFL I REFL-SELF I car-the 'Peter washes himself I the car.' b. Peter torrer sig I sig selv I Marie. Peter dries REFL I REFL-SELF I Mary 'Peter dries himself I Mary.' (23) "Anti-reflexive" Qredicates: a. Peter misunde *sig I sig selv I Marie. Peter envies *REFL I REFL-SELF I Mary 'Peter envies himself I Mary.' b. Peter mistcenker *sig I sig selv I Marie. Peter suspects *REFL I REFL-SELF I Mary 'Peter suspects himself I Mary.' (24) "Inherently reflexive" Qredicates: a. Peter dukkede sig I *sig selv I *Marie. Peter ducked RELF I *REFL-SELF I *Mary 'Peter ducked (*himself) I *Mary.' b. Peter tog en kniv med sig I *sig selv I *Marie. Peter took a knife with REFL I *REFL self I *Mary 'Peter took a knife with himl*himselfi*Mary.' We argue that it is the meaning of the predicates in (22-24) - in combination with pragmatic factors and world-knowledge - which determines whether sig can occur alone in object position or whether intensification by selv is necessary. The predicates in (24) are semantically "inherently reflexive" in that they are only compatible with reflexive scenarios, 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 arguments refer to (representationally) different entities, i.e. under normal circumstances it is highly unexpected for anyone to suspect himself of a crime or to envy himself. Finally, the predicates in (22) are "neutral" in the sense that they evoke no such presuppositions and consequently allow all types of direct objects. Thus, these examples illustrate that selv-intensification of reflexives follows the pattern of selv-intensified DPs by occurring only in contexts which allow for the generation of contrast sets of alternatives, thus satisfying the contrastiveness condition on adnominal intensification (18). The sentences in (24) involve predicates whose semantics excludes anything but the simple reflexive sig as internal object. Hence, since they violate ( 18), adnominal intensification is impossible. This analysis of complex reflexives as intensified nominal expression allows for the unification ofthe analyses ofadnominally intensified DPs, e.g. Peter selv 'Peter himself ( 4c ), and intensified reflexives like sig selv in (22-23a,b ). 1.6 Intensification and the meaning difference between simple and complex reflexives Since Jackendoffs ( 1992) paper "Madame Tussaud meets the binding theory" called attention to the fact that nominal expressions can be used to refer to statues or other types of representations of their normal referents, it has often been noted that simple and complex reflexives differ with respect to whether or not they can be used to refer to representations of the individual denoted by the antecedent. First consider the instances of locally bound sig and sig selv in (25). 7 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 predicate 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. 19 (25) Context: Imagine Bill Clinton visiting the wax museum. He notices a statue of himself with an unshaven face. Since he doesn't like the look of the statue he takes out a razor and starts to shave it. a. Bill Clinton, barberede sig,. b. Bill Clinton, barberede sig; selv. Bill Clinton shaved RELF/REFL self 'Bill Clinton shaved (himself).' The sentence with the simple reflexive sig in (25a) can only have the interpretation in which the real Clinton shaves himself (i.e. the real Clinton), see (26a). It cannot have the so-called statue-reading, see (26b), in which the real Clinton shaves a statue of Clinton. In contrast, the sentence with the complex reflexive sig selv in (25b) can have both reading (26a) and (26b ). (26) a. b. Clinton<real> shaves himself<real>. Clinton<real> shaves himself<statue>. We argue that the semantic difference between simple/unintensified and complex/intensified reflexives is a consequence of intensifier-adjunction. As described in section 1.4, adnominal intensification automatically generates a set of alternative referents for the associate of the intensifier. In the case of the complex reflexive sig selv it is the focus-generated set of alternative semantic values for sig (triggered by adnominal intensification) that licenses statue- readings in wax-museum contexts, e.g. (25b) (see also chapter 3, section 3.3.2.7. I for more discussion of the link between adnominal intensification of reflexives and the availability of statue-readings). In its unintensified/simple form sig behaves as a variable which 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 of its antecedent, see (25b ). Thus the present approach achieves a (semantically and morphologically) fully compositional analysis of complex reflexives. Note that a similar proposal is found in Konig & Siemund ( 1999:48). Furthermore since the anti-reflexive predicates, which require complex reflexives, see (23), all presuppose the non-identity (or at least the non-representational identity) of their arguments, we are now able to unify the account of anti-reflexivity (23) and so-called doppelganger- 20 effects as illustrated by the availability of statue-readings of sig selv in (25). This unification of the analyses of anti-reflexivity and doppelganger-effects in terms of adnominal intensification is one of the main contributions of the present dissertation. 1. 7 Logophoricity and intensified pronouns In this section we outline how the compositional analysis of reflexives based on the independence of intensification and binding (see sections 1.3 to 1.6) may be extended to the analysis of simple and complex subject and object pronouns. Let's begin by looking at simple/unintensified and complex/intensified subject pronouns. McKay (1991) argues that he himself is not a new kind of anaphor (pace Bickerton 1987). Using our terminology, McKay essentially proposes that he himself is an adnominally intensified pronoun. Syntactically, it behaves like its unintensified counterparts he and him in that it is subject to Principle B of the binding theory. Pragmatically, it is subject to the same semantic/pragmatic condition, i.e. (18), as other intensified nominal expressions. In this sense, McKay (1991) is the forerunner of Baker (1995) as well as the present dissertation. We thus share McKay's conclusion that "he himselfis grammatical whenever he is grammatical, and the same range of antecedent relationships is grammatically possible, though he himself requires a relevant contrast or comparison for pragmatic appropriateness" (McKay 1991:370- 371). McKay uses the example in (27) to show that he himself is not a special kind of anaphoric expression, but simply an intensified pronoun which is subject to principle B. (27) A: B: Mary, has been concerned about herfriends. Susank said that several were going to fail the course, and Susank might be right. But Mary, should think more about her, own work. How will Mary, do on the exam? I don't know. but Susank says that she/she/she; herself!shek herself will pass. More precisely, McKay's example in (27) falsifies Bickerton's (1987) claim that he himself behaves like an anaphor in that it "cannot have an antecedent outside the sentence if there is a 21 possible antecedent inside". McKay's main contribution is thus to show that, given the right context, he himself can be bound by a sentence-external antecedent in spite ofthe presence of a potential antecedent inside the sentence. That is, the pronominal nature of subject pronouns is not affected by adnominal intensification. In other words, binding and adnominal intensification are independent of each other. So far we have seen that adnominal intensification of reflexives and subject pronouns is independent of the binding properties of these elements. Taking the independence of 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 complex pronoun ham selv 'him self/him himself in the Danish sentences (28-29) should be analyzed as intensified versions of their simple counterpart, i.e. ham 'him'. (28) a. b. (29) a. b. Jon, vil giftes med en kvinde sam er stolt af ham, I #ham, selv. Jon wants marry-PASS with a woman who is proud of him I him self 'Jon wants to get married to a woman who is proud of him.' Kongen troede at ingen kunne !ide ham, I #ham, selv. king-the thought that nobody could like him I him self 'The king thought that nobody liked him.' Jon, plejede at hade folk sam var anderledes end ham, I ham, selv. Jon used to hate people who were different than him I him self 'Jon used to hate people who were different from himself.' Carl, sagde at Marie havde snakket med alle andre end ham, I ham, selv. Carl said that Marie had talked to all others than him I him self 'Carl said that Marie had talked to all others than himself.' While the distribution of the pronominal ham 'him' is determined by the syntactic principle B of the binding theory, i.e. pronouns have to be free from binding by a subject within the tensed clause, see (9b), the distribution of the element selv is determined by the semantic-pragmatic principle of the module of intensification, i.e. ( 18). In (29) both the unintensified pronoun ham 'him' and the intensified pronouns ham selv 'him self are allowed. In contrast, the sentences in (28) do not seem to license 22 intensifier-adjunction of the pronominal ham 'him' in an out-of-the-blue context. Adopting similar proposals by Zribi-Hertz ( 1989), Safir ( 1992), and Konig and Siemund (1999) we suggest that this difference is determined by the properties of the governing expressions. Since the expressions anderledes end 'different than' and andre end 'other than' in (29) are inherently contrastive they are able to license intensifier-adjunction to the pronoun. Lacking this 'inherently contrastive' quality the expressions stolt (l('proud of and fide 'like' in (28) are unable to do so 8 • Once again we show that the distribution of different kinds of nominal expressions (e.g. reflexive anaphors sig 'REFF', pronouns ham 'him', and r-expressions Peter) follows from syntactic principles of the binding theory, in this case principle B, while the distribution of the element selv is determined by the semantic-pragmatic principles of the module of intensification. 1.8 Possessive reflexives, pronouns and intensifiers One of the main strengths of the analysis proposed here is that it can be straightforwardly extended to intensified possessive reflexives and pronouns, which constitute a major problem for predicate-based approaches to binding, e.g. Reinhart & Reuland ( 1993). The Danish system of nominal expressions in possessor position is given m (30). Notice that except for the suppletive form of the intensifier, see (31 ), and the different case forms (assuming sin 'POSSREFL' to be the genitive form of the accusative/dative form sig 'REFL') this system is exactly the same as the one found in argument position, compare (30) with the system of argument DPs given in (6) in section 1.2. 8 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. himselj) 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. 23 (30) Danish nominal expressions in possessor position: a. b. c. Reflexive Pronoun DP Simple/unintensified Complex/intensified sin 'his/her/one's' hans 'his' kangens 'the king's' sin egen hans egen kangens egen 'his/her/one's own' 'his own' 'the king's own' (31) Suppletive variants of the adnominal intensifier in Danish: a. Argument position: selv 'himself b. Possessor position: egen 'own' Unlike English, Danish has a possessive reflexive, i.e. sin 'his/her/one's', in addition to the possessive pronouns hans 'his'. The distribution of simple and complex possessive reflexives and possessive pronouns follow the same general pattern as the distribution of simple and complex reflexives and pronouns in argument position. That is, like their argument position counterparts (i.e. sig 'him, her, one' and ham 'him') the distribution of simple forms, e.g. sin "POSSRELF' and hans 'his', is constrained by the principles A and B of the binding theory. Sin (like sig) is a reflexive and must be bound by a subject in a local domain (38a) and hans 'his' (like ham 'him') is a pronoun which must be free in its local domain, i.e. the minimal tensed clause, here indicated with square brackets (32b). (32) a. b. Peter, sagde [at Jahnk vaskede Peter said that John washed sin•; k *z tegnebag}. POSSREFL wallet 'Peteri said that Johnk washed his*i/ki*z wallet.' Peter, sagde [at Jahnk vaskede hans; *k z Peter said that John washed his 'Peteri said that Johnk washed hisii*klz wallet.' tegnebag}. wallet The sentences in (33a,b) illustrate the fact that both simple/unintensified and complex/intensified possessive reflexives may be LD-bound out of infinitival clauses. (33) a. b. Peter, bad Hansk vaske sin, k hund. Peter asked Hans wash POSSREFL dog 'Peteri asked Hansi to wash hi silk dog.' Peter, lad farceldrenez save i sin, *z egen seng mens han, selv sav pa safaen i stuen. Peter let parents-the sleep in POSSREFL own bed while he self slept on sofa the in living room-the 'Peteri let hisi parents sleep in his own bed while hei himself slept on the sofa in the living room.' 24 Though LD-binding of the simple possessive reflexive sin is more common, LD-binding of intensified possessive reflexives is not excluded, see (33b). That is, the adjunction of the intensifier egen 'own' does not affect the LD-potential of the possessive reflexive sin. Once again, adnominal intensification of reflexives does not affect locality constraints. The examples in (34-36) illustrate that Danish possessive reflexives sin and sin egen exhibit the same overall distributional pattern as the argument reflexives sig and sig selv with respect to "neutral", "anti-reflexive" and "inherently reflexive" constructions/sentences. (34) "Neutral" constructions: a. Peter vasker Peter washes sin /sin egen I John's tegnebog. POSREF /POSREF own/ John's wallet. 'Peter washes his /his own I John's wallet.' b. Peter hader Peter hates sin /sin egen I John's mar. POSREF /POSREF own/ John's mother. 'Peter hates his I his own I John's mother.' (35) "Anti-reflexive" constructions: a. Peter er *sin /sin egen I John's jjende. Peter is *POSREF /POSREFL own/ John's enemy. 'Peter is his I his own I his enemy.' b. Peter stjal *??sin I sin egen I John's tegnebog. Peter stole *??POSREF/ POSSREF own I John's wallet. 'Peter stole his I his own I John's wallet.' (36) "Inherently reflexive" constructions: a. Han varved at gaa ud afsit /?*sit eget/* Peters gode skind af glee de. he was about to go out ofPOSREF/?*POSREF own/Peter's good skin of happiness 'He nearly jumped out of his good skin of sheer happiness.' b. Han hyttede sit I ??sit eget /*Peters skin. (cont. bet. alb) he savede POSREF/ ??POSREF own/*Peter's skin 'He saved his own life.' The contrast between (34) and (35) shows that it is the semantic/pragmatic make-up of the sentence which triggers intensifier-adjunction to sin. Based on the meanings of the predicates vaske 'wash' and stjcele 'steal' and world knowledge, native speakers know that for a given individual x, the proposition x washes x 's wallet is pragmatically OK. The proposition x steals 25 x 's wallet, however, is pragmatically 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 ofthe meaning of the predicate stjcele 'steal', i.e. given normal circumstances the expectation is that people would not steal their own things. Hence, since anti-reflexive sentences of the type DPi stole POSSREFL, 's DP presuppose the (representational) non-identity of the referents of the possessive reflexive and its antecedent, intensification of the possessive reflexive sin is necessary to mark the (representational) non identity. The examples in (36) show that Danish also have "inherently reflexive" possessive constructions (usually of idiomatic nature) which- due to their semantics- disallow all but the simple unintensified reflexive possessive sin, see ( 18). We therefore conclude that the distribution of Danish simple and complex possessives (sin and sin egen) in "neutral", "anti-reflexive" and "inherently reflexive" constructions follow the same general pattern as the simple and complex argument reflexives sig and sig selv. This constitutes a major obstacle for analyses of binding which are based on the notions of co argumenthood, e.g. Reinhart and Reuland (1993). It seems rather implausible to maintain that the difference between (34a) and (35b) should derive from the a lexical feature, e.g. Reinhart and Reuland's [+/-inherently reflexive], or Zribi-Herts's [+/- Disjoint reference], distinguishing between the predicates vaske 'wash' and stjcele 'steal'. The structural relationship between the reflexives (i.e. sin and sin egen) and their antecedent has to be assumed to be the same all the sentences in (34-36). And yet, the sentences differ as to whether they allow the simple possessive reflexive or not. While this difference can be accounted for in terms of focus and presuppositions as illustrated above, it seems impossible to account for it in terms of any kind of predicate-centered 'reanalysis' or 'lexical ambiguity' account which only applies to co-argument binding. 26 1.9 Refuting anti-locality The account of the interaction between predicate/sentence meaning and intensification of reflexives outlined in section 1.5 and 1.8 above leads us to refute the standard account of sig in Danish (Norwegian, Dutch, etc.) which is based on the assumption that thematic sig is an anti- local anaphor which can never be locally bound (cf. Holmberg, (1984), Vikner (1985), etc.). That is, the binding theory proposed here does allow simple reflexives to be locally bound. Rather than being due to any inherent anti-locality of the simple reflexive sig, it is semantic or pragmatic factors which conspire to prevent local binding of unintensified reflexives in certain cases, e.g. anti-reflexive predicates (23) and (35). As shown above, intensifier-adjunction to simple reflexives is determined not by the binding theory but by the module of intensification and follows directly from semantic and pragmatic properties of the predicates and the larger context. In other words, anti-locality as an intrinsic property of simple reflexives does not exist; or, at least not in Dutch, Danish and the other Mainland Scandinavian languages. Although we do reject the idea that sig is anti-local, we still acknowledge the existence m Danish of two different uses of unintensified sig: (i) sig used as a thematic reflexive pronoun as in (37), (ii) sig used as a non-thematic grammatical marker of unaccusativity as in (38-40). (37) a. b. c. Peter vaskede sig. Peter vaskede sig selv. Peter vaskede Hans. Peter vaskede REFL I REFL self I Hans 'Peter washed (himself)/Hans.' (38) Deponent sig-verbs: a. Peter opforte sig I *sig selv I *Hans godt. Peter behaved REFLIREFL self/Hans well 'Peter behaved well.' b. Peter skammer sig I *sig selv I *Hans. Peter shames REFLIREFL self/Hans 'Peter is ashamed.' (39) Deagentive unaccusatives (inanimate subject) with transitive counter-parts: a. Doren abnede sig I *sig selv I *vinduet. (compare w. ( 41 a)) door-the opened REFLIREFL self/window-the 'The door opened.' b. Doren lukkede sig l*sig selv I *vinduet. (compare w. ( 41 b)) door-the closed REFLIREFL self/window-the 'The door closed.' ( 40) Deagentive unaccusatives (animate subject) with transitive counterparts: a. Peter slog sig (pa bordkanten). Peter hit REFL on table-edge-the 'Peter got hurt (on the edge ofthe table).' b. Peter brcendte sig (pa ovnen). Peter burned REFL on oven-the 'Peter got burned on the oven.' 27 The sig in (37a) is thematic. That is, it merges in object position of the transitive verb vaske 'wash' from which it receives the THEME 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 'shame' and skynde 'hurry' in (38), or unaccusative predicates with transitive counterparts, e.g. abne 'open, lukke close', sla 'hit', and brcende in (39-40). The transitive counterparts ofthese verbs are given in (41). (41) a. b. c. Peter abnede doren. Peter opened door-the 'Peter opened the door.' Peter lukkede doren. Peter closed door-the 'Peter closed the door.' Peter slog siglsig selviHans med en kolle. Peter hit REFLIREFL self/Hans with a club Peter hit himself/Hans with a club.' (cf. (39a)) (cf. (39b)) (cf. (40a)) d. Peter brcendte siglsig selviHans med en cigaret. (cf. (40b)) Peter burned REFLIREFL self/Hans with a cigarette 'Peter burned himself/Hans with a cigarette.' Rather than assuming the existence of two different sig's we attribute the difference between (37) vs. (38-40) to differences with respect to where in the derivation sig initially merges. As mentioned above, the sig in (37) merges in object position of the verb where it receives the 28 internal theta-role. This option is not available for the unaccusative predicates in (38-40) where it is the subject which initially merges in object position receiving the unique theta-role of the predicate before it moves up to the subject position. We propose that in these cases sig does not receive any theta-role since it merges directly with the small v of the vP projection responsible for the voice properties of the predicate. While we do acknowledge that two different uses of sig must be recognized we strongly disagree with the claim that thematic sig is anti-local ( cf. Vikner ( 1985), Lidz (200 1 ), etc.) or that it may only be locally bound be predicates which are lexically marked as [+inherently reflexive] (cf. Reinhart & Reuland (1993)). That is, we argue that thematic sig can be locally bound with certain predicates, e.g. neutral predicates like vaske 'wash' in (37). Both Vikner's and Reinhart & Reuland's approaches run into problems with (37a). In order to account for the absence of overt reflexive marking (by the element selv 'self) in (37a) R&R have to assume that vaske 'wash' is lexically [+inherently reflexive] in (37a) but not in (37b,c). Since there is no independent evidence for the existence of double entries for the verb vaske 'wash' such a stipulation is best avoided. Indeed, as shown in ( 42) the only difference in meaning between (37a) and (37b) is due to the adnominal intensification of sig, see (42a) vs. (42b). (42) a. b. c. d. Peter vaskede sig. ([+agentive], intentional event, [-contrast]) Peter vaskede sig selv. ([+agentive], intentional event, [+contrast]) Peter vaskede Hans. ([+agentive], intentional event, [-contrast]) Peter vaskede Hans selv. ([+agentive], intentional event, [+contrast]) Peter washed REFLIREFL self/Hans/Hans self 'Peter washed (himselt)/Hans (himself).' Since in our analysis the difference between ( 42a) vs. ( 42b) and ( 42c) vs. ( 42d) follows from the theory of adnominal intensification, there is no need for the ad hoc stipulation that Danish has double lexical entries for neutral predicates like vaske 'wash'. Vikner's anti-locality approach does not fare much better. Basically, his proposal consists in assuming that local sig 29 is "some sort of detransitivising element that somehow prevents the assignment of the theta- role that would otherwise have been assigned to the subject" (Vikner 1985:50, footnote 8). This analysis correctly predicts the non-agentive nature of the sentences with simple sig in (38-40), i.e. these sentences are not compatible with 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 assumed to remain the same transitive verb in all of (37a-c) and assign both an AGENT theta-role (to the external argument Peter) and a THEME theta-role (to the internal object sig). In chapter 4 a number of additional syntactic and semantic tests will be adduced to support our proposal that thematic sig is neither anti-local nor needs to be reflexive-marked by selv 'self' to be locally bound. Finally, notice that in addition to avoiding unnecessary proliferation of lexical entries for neutral predicates it also makes it possible to assume the existence of only one sig which may merge in different positions in the derivation: (i) in object position in the case of thematic sig, see ( 43a), and (ii) in the v of the voice projection vP in the case of non- thematic sig, see ( 43b ). (43) a. Thematic sig: VP I \ V' I \ v sig b. Non-thematic sig: vP I \ v' I \ I \ VP sig v I \ .... V' I \ v 30 1.10 Self-nominalizations In chapter 7 the use of the element selv 'self in nominalization will be discussed. It has been argued that selv 'self functions as a reflexive in nominalizations such as selvrespekt 'self- respect'. That selv 'self should function as a reflexive in nominalizations (rather than the simple reflexive sig 'REFL', e.g. *sig-respekt 'REFL-respect') has been taken as evidence in favor of analyzing selv 'self as a reflexivizing element ( cf. Reinhart and Reuland (1993)) and could thus potentially be construed as an argument against the analysis defended here, namely that binding of reflexives (i.e. sig) and the distribution of intensifiers (i.e. selv) are controlled by two separate modules of the grammar. In chapter 7 these issues will be discussed and an alternative analysis will be proposed in which the element selv 'self is consistently assumed to be an intensifier even in its word-internal uses. 1.11 Unification of all uses of ad nominal expressions Before moving on to exploring the consequences of extending the present analysis to English and Mandarin Chinese, let us first summarize some its main advantages. First of all, unlike current versions of binding theory (e.g. Reinhart & Reuland (1993), Lidz (2001), Huang & Liu (200 1 ), etc.) the present approach provides a morphologically fully transparent articulation of interaction between intensification and binding: nominal expressions (be they DPs, reflexives, or pronouns) to which selv 'self has been adjoined, see (44) are all intensified in the same way and have to obey the contrastiveness condition on intensification (18). (44) System of intensification of nominal expressions in Danish. Unintensified Intensified a. Reflexive sig 'himself/herself/' sig selv 'himself/herself b. Subj. pronoun han 'he' han selv 'he himself c. Obj. pronoun ham 'him' ham selv 'him himself/himself d. DP kongen 'the king' kongen selv 'the king himself 31 Furthermore, the present analysis avoids unwarranted proliferation of lexical entries. Most other approaches to binding and intensification all rely on the assumption of either (i) double lexical entries for neutral predicates (e.g. Reinhart & Reuland (1993)), or (ii) double/multiple lexical entries for intensifiers (e.g. Vikner (1985), Reinhart & Reuland (1993), or both. In the present approach selv is ALWAYS an intensifier and thus falls under the binding-independent module of intensification. The different subparts of complex reflexives (e.g. sig selv 'REFL self) and pronouns (han selv 'he himself) thus fall under different modules, see (45). (45) Independence of binding and intensification of intensified nominal expressions: Binding: Intensification: a. Intensified reflexives: sig (principle A (9a)) selv ( contrastiveness condition (18)) b. Intensified pronouns: ham (principle B (9b)) selv ( contrasti veness condition ( 18) c. Intensified DPs: Peter (principle C) selv ( contrastiveness condition (18)) Finally, unlike Reinhart & Reuland's (1993) predicate-centered approach to binding we propose a nominal approach in which binding is defined as the interaction of intrinsic properties of nominal 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 of the analysis defended here is demonstrated by showing how it can be extended to reflexives, pronouns and intensifiers in Modem English. We argue that the proposal in ( 46) make it possible to analyze the system of intensified and unintensified nominal expressions as having the same degree of morphological transparency as the Danish system in (44), see (47). ( 46) a. b. c. English x-se(f( e.g. himself, herself, etc.) are not reflexive anaphors but ALWAYS adnominal intensifiers (e.g. the king himself). Modem English has 0-reflexives, e.g. Peter, washes 0,. What looks like locally bound reflexives, e.g. Peter replaces himself, are really locally bound adnominally intensified 0-reflexives, e.g. Peter, replaces 0, himself(cf. Siemund 2000 for an outline of a similar analysis) 32 d What looks like locally free reflexives (also called "logophors") are really intensified pronominals whose pronominal 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). ( 4 7) System of intensification of nominal expressions in Modern English. Unintensified Intensified a. Reflexive 0 0 himself b. Subj. pronoun he he himself c. Obj. pronoun him [him] himself (<him himse(f) d. DP Peter Peter himself We propose to consistently analyze himself as an intensifier everywhere it occurs. Such an analysis becomes possible if we assume the existence of a 0 reflexive in English. By assuming that the internal argument position of the verb wash in (48) is filled by a 0 reflexive in (48a(ii)) and an intensified 0 reflexive, i.e. 0 himse(f, in (48a(iii)), we are able to unify the account of intensified and unintensified reflexives in Danish and English: in both languages the intensified forms of the reflexive pronouns are formed by the same process of intensifier adjunction which is responsible for intensification of DPs like Peter selv 'Peter himself and which is subject to the contrastiveness condition (18). (48) Neutral verbs: (i) Qron.: (ii) unint. refl. a. He, washes *him, I 0, b Han, vasker *ham; I sig, (49) Anti-reflexive verbs: a. He, suspects *him, I *0, b. Han, mistcenker *ham, I *sig; (50) Inherently reflexive verbs: a. He, rests *him, I 0, b. Han, hviler *ham, I sig, I I I I I I (iii) int. refl. 0, himself sig, selv. 0, himself sig; selv. *0, himself *sig; selv. We 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 washed'. The English x-self form reflexives can thus be viewed as intensified versions of the unintensified 0-reflexives, e.g. Peter washed 0, himself:::: (Dan.) Peter vaskede sig selv 'Peter washed 33 REFL self. Furthermore, the fact that, to a large extent, the distribution of himself follow that of the adnominal intensifier selv in Danish also supports our claim that English x-self, e.g. himself, herself, ourselves, etc., are always intensifiers- never reflexives; just like selv 'self in Modern Danish, and selfin Old English. In a similar vein, the neutralization of the morphological distinction between intensified pronouns and intensified reflexives in English (both appearing to be realized as himself), can be explained by assuming the existence of a deletion rule which erases redundant pronominal elements. Since the English intensifier himself is morphologically specified for the same features as the pronoun him, the intensified pronoun resulting from the adjoining the former to the latter would end up being overtly marked twice for the same features 9 . With such a rule it now becomes possible to analyze locally free instances of himself (also called "logophors") as intensified pronouns, e.g. [him}himselj; on a par with their Danish counterparts, e.g. ham selv 'him himself, see (51) and (52). The pronominal enclosed in square brackets e.g. [him]himself, indicates the morpheme deleted by the morphological reduction rule. This notation will be used hereafter to refer to the morphologically reduced intensified object pronouns. (51) Peter, said that Mary danced with everyone except {himJ himself (52) Peter, sagde at Marie dansede med aile andre end ham;selv. Peter said that Mary danced with all others than him self 'Peter said that Mary danced with everyone except himself.' 9 The fact that reduction of intensified pronouns applies to accusative/dative forms (e.g. him himself > [him] himselj) but not to nominative forms (e.g. he himselj) 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 taz 'he hates him'). does not allow subject pronouns to be intensified by complex intensifiers (e.g. *Ta taziji 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. 34 In both languages the intensified object pronouns are composed of a pronominal + the adnominal intensifier. The only difference is that the pronominal part of the English intensified pronoun is rendered inaudible by the morphological reduction rule (51), while it is phonetically realized in Danish (52). See chapter 5 for further discussion of these and other issues related to the analysis of Modem English. 1.13 Historical evolution of English reflexives and intensifiers In the second half of chapter 5 we argue that the account of Modem English proposed above makes it possible to understand hitherto unexplained facts in the evolution of intensifiers and reflexives from Old English into Modem English, see (53). (53) a. b. b. As shown in The Old English intensifier self was fused with pronouns to form the compound x-selfintensifiers, e.g. him+self>> himself Old English lost locally bound pronouns and developed 0-reflexives to replace them, e.g. Peter, washed him,>> Peter, washed 0,. The formation of x-self forms also took place in the 1st and 2nd persons (this is unexpected in most traditional accounts which assume that the initial motivation for adnominal intensification of locally bound pronouns in Old English was to disambiguate between disjoint reference and coreference readings). (54) Old English had a morphologically simple intensifier self but lacked morphologically specialized reflexives, using locally bound pronominals instead (54a). (54) Stage one: system of intensification ofnominal expressions in Old English 10 • Unintensified Intensified a. Locally bound pron. him him self b. Subject pronoun he he self c. Locally free pronoun him him self d. DP the king the king self 1 ° 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 of Old English morphology and spelling. The Old English intensifier self was overtly marked for case agreement. Though it has been argued that ca>e 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 of intensifiers and reflexives in English. 35 (55) System of intensification of nominal expressions in Modern English (=(47)). Unintensified Intensified a. Reflexive 0 0 himself b. Subject pronoun he he himself c. Object pronoun him [him} himsel/(<him himself) d. DP Peter Peter himself The sentences in (56) illustrate local binding of pronominals in Old English. They also illustrate how predicate meaning affects intensifier-adjunction (killing is an other- directed/anti-reflexive activity while defending is neutral). (56) 11 a. b. Hine, he, beweradh mid wcepnum. him he defended with weapons 'He; defended himsel~ with weapons.' Hannibal, ... hine, selfne mid atre acwealde. Hannibal him selfwith poison killed 'Hannibal killed himself with poison.' (Old English) (Old English) In chapter 5 we argue that assuming the system of intensified and unintensified nominal expressions in Modern English to be as outlined in (55), allows us to explain the evolution of the Old English (54) into Modern English in a rather straightforward way as a consequence of replacement of locally bound pronouns with 0-reflexives at a certain point in history, compare (54a) and (55a). As a result ofthe substitution of locally bound pronouns with 0-reflexives, in Old English the form him self was reanalyzed as an intensifier. 1.14 Extending the analysis to Mandarin Chinese In chapter 6 we argue that the analysis based on the independence of intensification and binding can be extended to reflexives and intensifiers in Mandarin Chinese. The Chinese intensifier element z!ji El c 'self-self has a great deal in common with Modern English himse(f and it will be shown that these similarities can be used to argue that the 0-reflexive analysis of Modern English proposed in chapter 5 may be extended to Chinese. First, like the 36 English adnominal intensifier himself (57) the Chinese adnominal intensifier § c ziji also appears to be able to tunction as a reflexive anaphor, see (58). (57) (58) ~wE:lc*To Huang-di ziji lai-le. emperor self-self come-PERF 'The emperor himself came.' ~1&t!H3c 0 Huang-di hen ziji. emperor hate self-self 'The emperor hates himself.' (ziji = adnominal intensifier) (himself= adnominal intensifier) (ziji =reflexive) (himself= reflexive) However, as discussed above, the peculiar properties of ziji and himself when allegedly used as reflexives, compared with reflexive anaphors in other languages, e.g. Danish sig, Dutch zich, French se, follow from the fact that they are ALWAYS adnominal intensifiers, which have long been mistakenly assumed to be reflexive anaphors. The differences between, elements like Chinese ziji and English himself (which are both ALWAYS intensifiers) and true reflexives like Danish sig and Dutch zich are illustrated by the fact is that ziji and himself do not have any of the non-reflexive uses (e.g. reciprocal, middles, medio-passives, unaccusatives, deponent verb, etc.) displayed by simple reflexive elements many other languages, e.g. Dan. sig, Ger. sich, see (59b-e). 11 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. (59) Typological survey of reflexive and non-reflexive uses of reflexive elements 12 : a. reflexive b. reciprocal d. middle Chinese English French German Danish ziji himself selsoi sich sig se sich -s se sich d. medio-pass. - se sich -s e. deponent se sich -s/sig f. intensifier ziji himself 37 Furthermore, as shown m (59t), both English himself and Chinese ziji can function as intensifiers, something the reflexives in Danish, French, German and Russian are unable to do, see (59t). So except for the reflexive uses, Chinese ziji and English himself are basically in complementary 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 himself display most ofthe uses of the Danish intensifier selv, see (60). (60) Typological survey of different uses of adnominal intensifiers: Chinese English French a. Adnominal intensifier guo-wang (ta) ziji the king himself le roi (lui-)meme b. Complex reflexive 0 ziji 0 himself soi-meme/lui-meme c. Doppelganger-marker 0 ziji 0 himself soi-meme d. Intensified pron,/logop. [ta} ziji Ita ziji [him] himself [lui}lui-meme e. Exclusive adverb. inten. ziji himself lui-meme Danish kongen selv sig selv sig selv ham selv selv We take the complementary distribution of ziji and himse(f vs. sig, see (59), as well as the similar distribution of ziji and himself vs. selv, see (60), as strong support of the assumption that both ziji and himself always are intensifiers rather than reflexive anaphors. The apparent reflexive uses of ziji and himself in (58)/(59a) can be explained if we assume that Chinese and English both have 0-reflexives which can be intensified by the adnominal intensifiers ziji and himself respectively, see 0 himself and 0 ziji in (60a). In other words, the absence of non- thematic uses of ziji and himself plus their consistent behavior as intensifiers provide strong 12 Danish has two reflexive elements: (i) the ·free form' reflexive sig 'REFL ·• and (ii) the bound morpheme s which is a medio-passive/reciprocal suftix derived historically from an cncliticized 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'. 38 support for the analysis of intensified and unintensified nominal expressions m Chinese, outlined in (61). (61) Systems of intensified and unintensified nominal expressions in Chinse, Danish and English: SimQle/unintensified ComQlex/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/ 3 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 of Chinese outlined in (61 ), which yields a morphological transparent system of adnominal intensification, is based on the hypotheses listed in (62). (62) a. b. c. d. Chinese ziji is not a reflexive anaphor but rather an adnominal intensifier (e.g. huangdi ziji 'the emperor himself). Chinese has 0-reflexives, e.g. Peter, shu lJ; [de] tou 'Peter combed 0/ 14 • What looks like locally bound reflexives, e.g. Peter; hen ziji, 'Peter hates himself, is really locally bound adnominally intensified 0-reflexives, e.g. Peter; hen lJ; ziji 'Peter hates 0; himself. What looks like locally free reflexives (also sometimes called "logophors") is really adnominally intensified pronominals, e.g.: Peted)t Mary ~N~~7 ({ill) E327tJ:9i-?JTif8"})\_~Jt~ o Peter, shuo Mary gen chu-le (taJ ziji yi-wai suo-you de ren tiao-wu P say M with except he self-selfiNST-outside that-have DE people dance 'Peteri said that Mary danced with everyone except [him;] himself.' 13 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. 14 As discussed in chapter 6, the clearest evidence for the existence of 0-retlexive in Mandarin Chinese are found in inalienable constructions in which the 0-retlexive occurs in possessor position: (i) a. 5ff=~ f)TE!J,FP<'fEI~ilJ! , Zhangsan, shu xiaa haiziz de tau. Zhangsan comb small child DE head b. '*=::-tm§ c'fs"JilJ!" Zhangsan, shu 0;zi)i de tau. Zhangsan comb self-self DE head c. 5N =::-tmBJ! o Zhangsan; shu 0, Zhangsan comb 0 [de] tau. head 'Zhangsan combs the child's/his (own) head.' 39 In addition to exploring the pros and cons of adopting the analysis outlined above for Mandarin Chinese, chapter 6 will also contain discussion of whether Chinese allows for adnominal intensification of PRO or if sentences like (62d) should be analyzed as having intensified 0 subject pronouns. 1.15 Conclusion The main proposals of the approach to binding and intensification advocated in this dissertation are summarized in (63). (63) Main proposals: a. Binding and intensification belong to separate modules of the grammar. b. Adnominal intensification is subject to pragmatic/semantic factors, should be given a focus-based account, see ( 18). c. A nominal rather than a predicate-based approach to binding is necessary. The distribution of nominal expressions is determined by the interaction of the feature properties of nominal expressions and syntactic locality constraints (principles A and B), see (9)-(1 0). d. Complex reflexives are intensified nominal expressions on a par with intensified DPs. That is, we propose a unified analysis of all uses of adnominal intensifiers (e.g. Dan. kongen selv 'the king himself, and sig selv 'REFL self ::::; Eng. the king himself, lJ himself, etc.). e. Intensification of simple reflexives is determined by a combination of: (i) pragmatics, i.e. background knowledge of participants in the utterance situation, and (ii) predicate meaning. f. (I e) leads to a tripartite typology of predicates: (i) anti-reflexive predicates, which presuppose (representational) non-identity of their arguments, (ii) inherently reflexive predicates, which presuppose identity of their arguments, and (iii) neutral predicates, which evoke no such presuppositions. Additional results ofthe analysis proposed here are listed in (64). (64) a. b. c. Contrary to what is often proposed in the literature, the simple reflexive sig 'REFL' in Danish(:::::: Dutch zich, Nor. seg, etc.) is not-anti-local. Local sig can be a true theta-role receiving anaphor. 40 Unification of reflexive and non-reflexive uses of sig: the same element is used in both cases but it merges in different places in the derivation: reflexive, theta-marked sig merges as internal argument of the predicate, non-reflexive sig merges with the head of the vP projection responsible for the diathesis of the sentence. In terms of its feature composition it is the same sig in both cases. As mentioned throughout this introductory chapter many of the ideas presented in this dissertation have been proposed before. However, they have rarely been elaborated in any detail and they have never before been put together to form a coherent and comprehensive framework. Furthermore, the cross-linguistic validity of the analysis of intensification and binding presented here has been illustrated by outlining how it can be extended to both Modern English and Mandarin Chinese. Finally, the 0-reflexive analysis of modern English proposed here receives further support from the fact that it makes it possible to explain the historical evolution of intensifiers and reflexives. 2.1 Introduction Chapter 2 Intensification 41 This chapter is dedicated to the theory of intensification. Basing the discussion mainly on Danish and English data, we propose a focus-based analysis of adnominal intensifiers, e.g. himself and selv 'self in (1( (1) Intensified DPs: a. Peter selv deltog i modet. Peter self partook in meeting-the 'Peter himself attended the meeting. b. Peter himself attended the meeting. The semantic contribution of adnominal intensifiers has been previously thought to involve scalar ordering of focus alternatives (as argued by Eckardt 2001 ), or to be reducible to centrality-effects (as argued by Konig ( 1997) and Siemund (2000)). Instead, we propose that it is similar to that of focus, which- depending on the context - may or may not involve scalar ordering of focus-generated alternatives. In certain respects, the present analysis is close to Baker (1995), whose analysis of adnominal intensification is based on prominence and contrastiveness. But unlike Baker (1995) whose analysis includes both a prominence and a contrastiveness condition on intensification, we propose to reduce the analysis of intensification to a single condition, viz. the contrastiveness condition in (2). (2) Contrastiveness condition on adnominal intensification: A nominal expression (DP, pronoun, reflexive) can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. 1 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, leibhafiig, 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. 42 The analysis of the morphology, syntax and semantics of adnominal intensifiers is presented in section 2.2. The advantage of the analysis of adnominal intensification proposed here is that it can be extended to uses of selv 'self which have usually been considered to be different from adnominal intensifiers, see (3a, ii-iv). (3) a. b. c. d. e. f. Different uses of selv: adnominal intensification i. intensified DPs 11. intensified reflexives 111. intensified pronouns iv. q-floated intensifier selv in secondary predication constructions scalar additive focus particle 'even' selv as a noun adjectival forms ofthe intensifier selv in nominalizations Examples: (I) (4) (5) (6) (7) (8) (9) (1 0) ( 11) Section: (sect. 2.2) (chap. 3, section 3.3) (chap. 3, section 3.4) (sect. 2.2.5) (sect. 2.3) (sect. 2.4) (sect. 2.5) (sect. 2.6) (sect. 2. 7, chap. 7) As indicated in (3a), we propose to unify as cases of adnominal intensification a number of uses of selv 'self hitherto analyzed as different phenomena. Most importantly, in chapter 3 we argue that the so-called complex reflexives (3a,ii), e.g. Dan sig selv 'REFL self, see (4), and complex pronouns (3a,iii), e.g. Dan. ham selv '(him) himself, see (5), should be analyzed as intensified nominal expressions on a par with intensified DPs (3a,i), e.g. Dan. Peter selv 'Peter himself, see (1 ). (4) Intensified reflexives: Peter mistcenkte sig selv. Peter suspected REFL self 'Peter suspected himself.' (5) Intensified pronouns: [..] pa onsdag flyver Jakobsen til Malaysia for at forhandlei hvad der ifolge ham, selv kan blive til kontrakter pa 300 km kystsikring. (JP July 20, 2003) on Wednesday flies Jakobsen to Malaysia for to negotiate what there according-to him self can become to contracts on 300 km coast securing 'Wednesday Jacobsen flies to Malaysia to negotiate what according to (him) himself could tum into contracts for 300 km coast securing.' Since intensified reflexives and pronouns are dealt with in detail in chapter 3, they will only be mentioned cursorily in the remainder of this chapter, which focuses more on the other uses of 43 intensifiers. For example, in section 2.2.5 we argue that what has been referred to as an "inclusive adverbial intensifier" (cf. Konig (1997)) meaning 'also, too', compare (6a-b) and (6c), should be analyzed as a quantifier-floated adnominal intensifier (3a,iv). (6) Quantifier-floated adnominal intensifiers: a. Peter; var [t, selv] blevet nomineret til en Oscar. Peter was self become nominated to an Oscar 'Peter had himself been nominated for an Oscar.' b. Peter, had [t, himself} been nominated for an Oscar. c. Peter had also been nominated for an Oscar. In addition to showing that the phenomena in (3a,i-iv) are all instances of adnominal intensification falling under the contrastiveness condition in (2), we propose that the analysis of intensification presented here has the potential to be extended to all the non-adnominal uses of selv 'self, see (3b-t). Cross-linguistically the elements used as adnominal intensifiers are also frequently found in a number of other uses, e.g. as so-called "exclusive adverbial intensifiers" meaning 'by oneself, without help' (3b), see Eng. himself and Dan. selv in (7), and as additive scalar focus particles meaning 'even' (3c), see Fr. meme 'same' and Dan. selv in (8). (7) Intensifiers in secondary predication constructions ("exclusive adverbial selv"): a. Peter skrev stilen selv. Peter wrote essay-the self 'Peter wrote the essay himself.' 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. same Peter is come 'Even Peter came.' In section 2.3 we propose that what has sometimes been referred to as an exclusive adverbial intensifier meaning '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 44 selv 'himself and the focus particle selv 'even' (Sa-b) as the same element which assumes different properties when occurring in different positions. In sections 2.5 and 2.6, the nominal and adjectival uses of selv (3d-e), illustrated in (9)-(1 0), will be argued to be morphological variants of the same intensifier element selv 'self. (9) Nominal uses of intensifiers: Selvet er en central del af psyken. self-the is a central part of psyche-the 'The self is a central part of the psyche.' (10) Adjectival uses of intensifiers: Selve huset er ikke meget vcerd. self house-the is not much worth 'The house itself isn't worth much.' Detailed treatment of the use of the element selv in nominalizations (3t), briefly described in section 2. 7, will be postponed till chapter 7, where we argue that word-internal uses of selv follow the same pattern as the adnominal intensifier (3a), see (II a), or as selv used in secondary predication constructions (3b ), see (II b). (II) Intensifiers in nominalizations: a. selvrespekt self-respect 'self-respect' b. selvbygger self-builder 'person building a house by himself/herself Section 2.8, completes the survey by listing and briefly discussing a few idiomatic uses of the element selv 'self. Finally, in section 2.9 a mini typological survey of the ranges of uses of intensifier elements in Danish, English, French, and Chinese, is given. In brief, as summarized in section 2.1 0, the contribution of this chapter is threefold. First, in section 2.2, we propose a focus-based analysis of adnominal intensification based on the contrastiveness principle in (2). Second, in sections 2.2.5-6, we claim that many 45 phenomena which have hitherto been considered unrelated, i.e. (3a,i-iv), can be accounted for within this analysis of adnominal intensification. Third, in sections 2.3-8, we suggest that the lexicon contains only one element selv 'self which can take on different syntactic functions (and yield different semantic contributions) when found in different positions in the derivation, i.e. (3b-c). This proposal makes it possible to unify the analysis of both adnominal (3a,i-iv) and the so-called non-adnominal uses (3b-t) of the intensifier selv 'self in Danish. 2.2 Adnominal intensification In this section the main focus will be on adnominal intensification of full lexical DPs. After going through the morphological, syntactic and semantic characteristics of adnominal intensifiers, a focus-based analysis of adnominal intensification will be proposed. The rest of this section is organized as follows. First in section 2.2.1 and 2.2.2 the morphology and syntax of adnominal intensification will be discussed. Then, in section 2.2.3, a semantic analysis of adnominal intensifiers will be outlined. 2.2.1 The morphology of adnominal intensification This section provides a brief description of the basic morphological properties of adnominal intensifiers in Danish and English. English intensifiers are in association with a DP with which they display agreement in the phi-features, i.e. person, number and gender, see ( 12). ( 12) English Danish a. !myself jeg selv 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. 46 In Danish, the intensifier selv 'self is morphologically invariable and it does not decline to match the features of the nominal element that it is intensifying, as shown in (12). Ofthe Scandinavian languages, Norwegian (bokmal) behaves like Danish in that it has uninflected intensifiers of the German type. In contrast, Swedish and Icelandic intlect their intensifiers. See the Swedish examples in (13) in which the adnominal intensifier sjalv 'self displays agreement in gender and number. ( 13) 2 a. Han iir godtheten sjiilv. she is goodness-the self-SING-COMMONGENDER 'She is kindness itself' b. Barnet sjiilv-t sa ingenting. child-the self-SING-NEUTR said nothing 'The child himself said nothing.' c. Barnen sjiilv-a bestamde. children-the self-PLUR decided 'The children themselves decided.' According to typological surveys ( cf. Siemund 2000: I 0-11 ), intensifiers with overtly marked agreement (e.g. English himself, French lui-meme, Russian sam-, etc.) are more frequent than morphologically invariable intensifiers (e.g. Danish selv, French -meme, German selbst, etc}. The above discussion of the morphology of intensifiers naturally leads us to the question what class of linguistic expressions adnominal intensifiers belong to. While in the Slavic and Romance languages adnominal intensifiers tend to display adjectival behavior, intensifiers in the Germanic languages primarily 4 behave as adnominal adjuncts (e.g. Ger. selbst, Dan. selv). These differences are witnessed by the fact that, Germanic intensifiers tend 2 These Swedish examples arc adapted from Siemund, 2000:8 and pp. 17-22. 3 In Old English the intensifier seoif'selr 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. 4 Note. 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 47 to allow nominal uses 5 , see the Danish examples in (14), while Romance intensifiers tend to allow adjectival uses but disallow nominal uses, see the Spanish examples in (15). ( 14) (15) a. b. c. a. Selvets begred self-the-POSS concept 'the conception of self (/the ego)' Selvet er en central del af psyken. self-the is a central part of psyche-the 'The self is a central part of the psyche.' Kun levende vcesner kan have et selv. only living creatures can have a self 'Only living creatures can have a self.' el mismo cache the same car 'the same car' b. la misma casa the same house 'the same house' c. *el mismo/*la misma the self In ( 14c) the Danish intensifier selv occurs as a noun preceded by the indefinite article et 'a (neut., sing.)'. The sentences in (14a-b) further illustrate that selv behaves just 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 sometimes been adduced as evidence supporting the classification of selv as a nominal expression. In contrast, the Spanish examples in ( 15a-b) have been taken to show that Spanish intensifiers are a special kind of adjectives which cannot be used as nouns meaning 'the self, see (15c). However, rather than entering the somewhat futile debate about what word-class the element selv 'self belongs to, we simply take the 5 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 (i). See section 2.5 for more discussion intensifiers used as nouns. (i) The *oneseljl*himseljlselfis a central part of the psyche. 48 element selv 'self' to be a morpheme/root which can assume nominal, adjectival, or adverbial properties depending on where in the derivation it is inserted 6 • 2.2.2 The syntax of ad nominal intensification In this section the syntax of the use of the adnominal intensifier selv 'himself' with full lexical DPs will be described in some detail. The discussion of adnominal intensification will be structured around the questions in ( 16). ( 16) The syntax of adnominal intensification: a. Are there any constraints as to what syntactic positions intensified nominal expressions may occur in? b. What is the exact nature of the syntactic relation (i.e. complementation, adjunction, etc.) between the adnominal intensifier selv 'himself' and its associate, i.e. the DP which it is intensifying? 2.2.2.1 Possible syntactic positions of adnominally intensified nominal expressions Let us now tum to the question of whether there are any constraints as to what syntactic positions intensified nominal expressions may occur in. The answer seems to be no: apparently nominal expressions intensified by adnominal selv can occur in the same range of syntactic positions as their unintensified counterparts, see the Danish examples in (17)-(23): (17) Intensified nominal expressions in subject 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; jeg kom kun til at tale med hans kane. I came not to to interview chancellor-the self; I came only to to talk to his wife 'I did not get to interview the chancellor himself; I only got to talk to his wife.' 6 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. ( 19) Intensified nominal expressions in indirect object position: Jeg gav ikke Dronningen selv blomsterne; jeg gav dem til hendes kammerpige. I gave not Queen-the selffiowers-the; I gave them to her chamber-maid 'I didn't give the Queen herselfthe fiowers; I gave them to her maid.' (20) Intensified nominal expressions as objects of prepositions: Jeg gav blomsterne til Dronningen selv. I gave fiowers-the to Queen-the self 'I gave the fiowers to the Queen herself.' (21) Intensified nominal expressions in clefts: Det var Dronningen selv, som modtog os. it was Queen-the self who received us 'It was the Queen herself who received us.' (22) Intensified nominal expressions in topicalizations: Dronningen selv sa vi ikke. Vi sa kun prinsene. Queen-the self saw we not we saw only princes-the 'The Queen herself we didn't see. We only saw the princes. 49 For DPs in possessor position the intensifier selv 'himself assumes the suppletive form egen 'own', see (23). (23) a. Form ofintensifier: b. Example Argument DPs selv 'self, himself kongen selv king-the self 'the king himself Possessor DPs: egen 'own' kongens egen hund king-the-POSS own dog 'the king's own dog.' Adnominal intensification in possessive constructions will be discussed in more detail in chapter 3, section 3.3.5. Though there are subtle differences between egen 'own' and selv 'self, we will argue that they are not due to lexical differences between these two forms, but rather are parasitic on the specific syntactic and semantic properties of possessive constructions. The only arguments inside nominalizations which are intensified are 0- elements, see discussion in chapter 7 where we argue that nominalizations such as selvrespekt 'self-respect' in (11 a) above contain a 0-refiexive intensified by the adnominal intensifier selv 'self. 50 As the above descriptive survey shows, selv may intensify different types of nominal expressions 7 occurring in a wide range of syntactic positions, e.g. subject, direct object, indirect object, prepositional complement, possessor position, etc. This leads us to conclude that there are no significant syntactic constraints as what positions intensified nominal expressions may occur in (cf. Siemund (2000)) 8 • 2.2.2.2 The syntactic relation between intensifier and its focus: adjunction Adnominal selv forms a constituent with the DP it modifies. This can be shown by fronting the whole constituent in a cleft construction as in (24). (24) Det var Kongen selv, der laflede klaveret. It was king-the self who lifted the piano 'It was the King himself who lifted the piano.' Secondly, Danish is a V2 language. This means that in root sentences, only one constituent can precede the verb. Dronningen 'the Queen' and selv in (19a) in section 2.2.2.1 must therefore belong to the same constituent, the subject DP. This kind of evidence leads us to conclude that selv must be some kind of adnominal modifier forming a syntactic constituent with its associate. But it still does not tell us whether it is an adjectival modifier 9 , an adjunct or something else. Unlike regular adjectives, which in Danish are always pre-nominal, see (25), selv must occur after the full DP which it is modifying, see (26). 7 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). 8 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. 9 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 adnomina1 or an adverbial interpretation.'' (25) a. b. c. den gamle mand the old man 'the old man' *den mand gamle the man old *manden gamle man-the old (26) a. b. c. *den selv mand the man self (??)den mand selv 10 that man self 'that man himself manden selv man-the self 'the man himself 51 Furthermore, notice that selv, unlike adjectives, can modify 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 PPs are placed after the DP they modifY, see (27a-b), as are DP complements, see (27c). (27) a. skuespilleren sam havde modtaget prisen actor-the who had received award-the 'the actor who had received the award' b. dronningen med den gyldne krone queen-the with the golden crown 'the queen with the golden crown.' c. dronningen af Sverige queen-the of Sweden 'the queen of Sweden' As illustrated in (28), the different post-nominal modifiers in (27) do not all display the same behavior with respect to adnominal selv. (28) ??/*skuespilleren sam havde modtaget prisen selv actor-the who had received award-the self 'the actor who had received the award himself 10 In Danish the definite article can be realized in two ways: (I) as a suffixal determiner -enl-et 'the', or (II) as a free standing determiner den!det 'the'. The determiner denldet 'the' is always used when the noun is modified by an adjective. see (25a). Unmodified head nouns and head nouns moditied 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 denldet. the latter is usually interpreted as a demonstrative 'this/that' 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 intensitiers usually only modify DPs whose referent has been introduced earlier in the same situation/discourse, the sentence in (26b) sounds rather odd. 11 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'. b. ??l*dronningen med den gyldne krone selv queen-the with the golden crown self 'the queen with the golden crown herself c 12 • dronningen afSverige selv queen-the of Sweden self 'the queen of Sweden herself 52 If a DP is already modified by a restrictive relative clause or a PP of the with-type then it cannot felicitously be intensified by selv, see (28a-b ). However, if a DP takes a complement of the oftype then it can be intensified by selv, see (28c). The examples in (27) and (28) indicate that selv is in complementary distribution with post-nominal prepositional adjuncts of the med-type and restrictive relative clauses 13 • On the basis of such facts it seems plausible to suggest that selv also has the syntactic status of a post-nominal adjunct modifying DPs 14 • Given that selv forms a constituent with the DP, it must be the highest element in the DP, see (29). (29) 15 The structure ofDPs intensified by selv: I DP I \ DP \ selv self N D 'himself kong -en king the 'the king' 12 This 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 af 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 of 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 of the issue itself (continued for several hours) (from Siemund (2000: 157) example (6.94a)). 13 Cf. Everaert ( 1986:37 -38) who observed the same pattern in Dutch: "it appears that a post-nominal quantifier like zelfoccupies the same position as met/zonder-PP-complements and restrictive relatives". 14 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. 15 Some researchers now believe that natural languages never use right-adjunction of the type illustrated in (29). The question whether right-adjunction is among the structures available to natural languages seems to be orthogonal 53 While stacking of adjuncts yields unacceptable structures in (28a-b) where the intensifier selv is the last in a series of adjuncts, it appears to be possible to construct acceptable examples involving adnominal intensifiers and adjunct-stacking, provided that the intensifier precedes the other adjunct, see (30). (30) Dronningen selv sammen med prinserne /g.ftede klaveret. Queen-the self together with princes-the liftede piano-the 'The Queen herself, together with the princes, lifted the piano.' Notice, however, that in this respect the PP sammen med 'together with' behave differently from the other constructions involving PPs and relative clauses in (27), see (31 ). (31) a. *skuespilleren selv sam havde modtaget prisen actor-the self who had received award-the 'the actor himself who had received the award' b. */ok 16 dronningen selv med den gyldne krone queen-the self with the golden crown 'the queen herself with the golden crown.' c. *dronningen selv a(Sverige queen-the self of Sweden 'the queen herself of Sweden' The contrast between (30) and (31 ), which are all unacceptable, indicates that the PP sammen med 'together with' is different from the adjuncts in (31a-b) and the prepositional complement in (31c). This difference is further confirmed by the examples in (32) and (33). (32) Dronningen selv /g.ftede klaveret sammen med prinserne. (33) Queen-the self liftede piano-the together with princes-the 'The Queen herself lifted the piano together with the princes.' a. b. *skuespilleren selv kom til modet sam havde modtaget prisen actor-the self came to meeting-the who had received award-the 'the actor himself came to the meeting who had received the award' *dronningen selv satte sig pa tronen med den gyldne krone queen-the self sat REFL on throne-the with the golden crown 'the queen herself sat down on the throne with the golden crown.' to the issues discussed here. The reader is therefore free to translate the structure in (29) to left-adjunction plus movement, if he/she is so inclined. 16 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. c. *dronningen selv satte sig pa tronen af Sverige queen-the self sat REFL on throne-the of Sweden 'the queen herself sat down on the throne of Sweden' 54 Based on the above examples we conclude that the PP sammen med 'together with' in (30) and (32) is used as a VP-adverbial or secondary predicate (which can either precede or follow the VP) rather than as an adnominal adjunct. Hence, it does not constitute a counter-example to the generalization that PPs adjunct and relative clauses are in complementary distribution with the adnominal intensifier selv 'self. 2.2.2.3 Unified analysis of ad nominally intensified expressions Adopting a modified version of Longobardi's (1994) analysis of nominal expressions we assume that all nominal arguments are projections of a head D constituent. This approach makes it possible to arrive at a unified account of the syntax of nominal expressions, see (34) a. b. c. d. [op [op [op [oP kong ] konger ] [o [o [o [o -en ] 0 ] ham] sig ] 'the king' 'kings' 'him' 'REFL' The structures in (34a) and (34b) illustrate the syntax of DPs with overt and null determiner respectively. Pronouns and reflexives are assumed to behave as determiners, i.e. they are assumed to head their own DPs, see (34c) and (34d) 18 • In addition to making it possible to 17 As mentioned above. Danish has both post-nominal and pre-nominal articles, see (i-ii). (i) Hus-et House-the (ii) Det store hus the big house The question of which order is the basic order of 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. 18 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))). 55 arrive at a unified analysis of the syntax of nominal expressions, this approach has the advantage of enabling us to unify the account of adnominal intensification. As discussed above, intensified nominals such as kongen selv 'the king himself' are best analyzed as simple adjunction structures, see (35a). (35) a. b. c. d. [or [or [N [or [or [N [or [or [o [or [or [o kong ] [ 0 -en ] konger ] [ 0 0 ham] ] sig] ] [selv ]] [selv]] [selv ]] [selv]] 'the king himself' (??)'kings themselves' 19 'him self' 'REFL self' As shown in (35c) and (35d) the complex pronoun ham selv 'him self' and the complex reflexive sig selv 'REFL self' can be analyzed as intensified DPs having the same syntax as intensified R-expressions. In other words, syntactically all types of nominal expressions, e.g. full lexical DPs, subject pronouns, object pronouns, possessive pronouns, and reflexive pronouns, behave alike with respect to adnominal intensification 20 . 2.2.3 Semantic and pragmatic aspects of ad nominal intensifiers The literature on intensifiers contains a number of different proposals of how to best capture the semantic and pragmatic properties of intensifiers some of which are listed in (36). (i) [IP [DP [D We] [N psychologists]] [I' [I don't] [VP [V trust] [DP [D you] [N linguists]J]]] (ii) [IP [DP [D We]] [I' [I don't] [VP [V trust] [DP [D you]]]]] 19 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. 20 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 of 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. 56 (36) a. Dirven (1973) b. c. d. emphasis expectedness/remarkabi I ity centrality Edmondson & Plank ( 1978), Eckardt (200 1) Konig (1991) (discourse) prominence obviation Baker ( 1995) Baker (1995) e. f. g. focus contrasti veness Konig (1991 ), Ferro (1993), Sanchez (1994) McKay (1991 ), Baker ( 1995) In section 2.2.3.1 a focus-based, non-scalar account of the semantic contribution of adnominal intensifiers will be proposed. In section 2.2.3.2 semantic constraints on intensified DPs will be discussed. 2.2.3.1 The semantic contribution the adnominal intensifier selv 'himselr In Danish, as well as in German and French, the element used as adnominal intensifier, i.e. Dan. selv, Ger. selbst, and Fr. -meme, can also be used as an additive, scalar focus particle similar to the particle even in English. In all three languages the adnominal intensifier is post- nominal while the focus particle 'even' is pre-nominal, compare the (a) ahd (b) examples in (37)-(39). (37) Danish: a. b. Kongen selv holdt en tale. king-the self held a speech 'The king himself delivered a speech.' Selv visew:erten holdt en tale. self janitor held a speech 'Even the janitor delivered a speech.' (38) German: a. Der Konig selbst hielt eine Redi 1 • the king self held a speech 'The king himself 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) 21 The German and French examples in (38a-b) and (39a-b) are adapted from Eckhardt (2000: I (I), (2)). (39) French: a. Le roi (lui-)meme 22 a prononce un discours. the king him self has pronounced a discourse 'The king him-self has delivered a speech.' b. Meme le concierge a prononce un discours. even the janitor has pronounced a discourse 'Even the janitor has delivered a speech.' 57 (intensifier) (focus particle) Inspired by the overt similarity between the two uses of the intensifier elements illustrated in (37-39) a number of researchers have proposed to analyze the adnominal intensifier as a scalar focus particle similar to 'even'. In the following, we argue that such analyses are misguided and that adnominal intensifiers should be given a non-scalar, focus-based analysis. Unlike the focus particle selv 'even', the adnominal intensifier selv 'himself does not presuppose that the proposition is true for at least one other element in the focus-generated set of alternatives, see ( 40a) vs. ( 40b ). (40) a. b. *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.' Det var biskoppen selv der holdt gudstjenesten. (intensifier) It was bishop-the self who held service-the 'It was the bishop himself who held the service.' The example in ( 40a) shows that the focus particle selv 'even' cannot occur in clefts which presuppose uniqueness because that would lead to a contradiction of the so-called existential presupposition it always evokes, see ( 44b ). Since the adnominal intensifier selv 'himself can occur in clefts, see ( 40b ), we therefore conclude that it does not evoke this existential presupposition which is mandatory with the focus particle even. The examples in ( 41 )-( 42), which contain predicates presupposing uniqueness, further confirm this difference between even and the adnominal intensifier. 22 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) tor more discussion of lui-meme and -meme as intensifiers. (41) a. b. (42) a. b. #Selv Dronningen gav nytarstalen. even queen-the gave new-years-speech-the '#Even the Queen delivered the New Year's Speech.' Even the President delivered the inaugural speech. Dronningen selv gav nytarstalen. queen-the self gave new-years-speech-the 'The Queen herself delivered the New Year's Speech.' The President himself delivered the inaugural speech. 58 According to Eckardt (2000) the semantic contribution of 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 of the existential presupposition/implicature, compare (43) with the analysis of even in (44). As mentioned in chapter I, at this point no attempt has been made to distinguish between presupposition and implicature. The two terms are used interchangeably without difference in meaning in the rest of this dissertation (43) Analysis of the adnominal intensifier himself himself+ S a. Assertion: [[ S W b. Scalar implicature: (i) \fp E {[[ s Jt\ [[ s nT [[ s ]]" <c p (44) Analysis ofthe additive focus particle even: a. b. even+ S Assertion: Scalar implicature (i) Existential implicature(ii) [[ s ]]" VpE {[[S]t\[[SJn:[[S]t <cp ::Jp E {[[S]]f\[[S]] 0 }:p= J Following Rooth ( 1992), [[ S Jt and [[ S ]{ are used to distinguish between the ordinary semantic value of a sentence, i.e. its truth value, see (48a), and the so-called focus semantic value of a sentence. As illustrated in ( 48b ), the focus semantic value of a sentence is a set of propositions obtained by replacing the associate of the intensifier with elements from a contextually defined set of relevant alternatives. In ( 43-44b) the scalar ordering of alternative propositions on a scale of likelihood is expressed by the comparative operator <c, that is 'p <c q' should be read as 'the proposition pis more likely to be true than the proposition q'. 59 We argue that the analysis of adnominal intensifiers in ( 43) is on the wrong track. While it is true that DPs adnominally intensified by selv often refer to entities occurring in propositions which are unexpected or remarkable in a given context, this is by no means always the case, see ( 45-46). (45) A very powerful earthquake struck the center of Rome and the Vatican. a. The Pope himself perished in the rubble. b. #Even 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 of high and low status or prominence 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 remarkable location, but simply because it is contrasted with another location, viz. the suburbs. Notice that in both cases the adnominal intensifier is perfectly acceptable while the focus particle even, which always carries with it a scalar implicature, see (44b(i)), is not. We take this as evidence that the scalar implicature in (43b) is NOT an integral part ofthe semantic contribution ofadnominal intensifiers. Based on the examples discussed above, we conclude that the semantic contribution of the adnominal intensifier involves neither a scalar implicature (44b(i)), nor an existential implicature, (44b(ii)), and is thus different from the focus particle even, which has both these implicatures. The semantics ofadnominal intensification is similar to that of focus, see (47a)- (48). (47) a. b. (48) a. [The king himselj] 1 . came to the meeting. The king came to the meeting. Ordinary semantic value of(47a): [[ [The king himself]F came to the meeting]r = 1 iff the king came to the meeting. b. Focus semantic value of(47a): [[ [The king himself]F came to the meeting ]] f = { came-to-the-meeting(x) I x E De} 'the queen came to the meeting', 'the prince came to the meeting', 60 Like focus, the basic semantic contribution of adnominal intensification consists in contrasting the referent of the focused/intensified expression with a contextually determined set of 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 of alternatives who also did (or might have) come to the meeting. The generation of a set of alternatives gives rise to contrastiveness condition on adnominal intensification which 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 adnominal intensification were discussed. In this section, semantic/pragmatic constraints on adnominal 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 Moravscik ( 1972) and Edmondson and Plank (1978) it has been generally known that a DP modified by an adnominal intensifier must have certain referential properties. However, the exact formulation of this constraint still seems to elude the community of researchers working on intensification. Examples like (49) and (50) seem to indicate the existence of a constraint requiring adnominally intensified DPs to be definite. (49) (50) a. b. a. b. *The LA philharmonic is lookingfor a truly gifted violinist himself *A truly gifted violinist himself would never miss note. The Queen herself showed up to the conference. I meet the Queen herself at the conference. 61 However, as Edmondson and Plank (1978:382) observed, things are not that simple. In certain cases formally definite DPs, i.e. DPs preceded by the definite article the, cannot be felicitously intensified either, see (51 c). (5li 3 a. We wanted to call the doctor. b. We wanted to call the doctor himself c. *We wanted to call the doctor himself but we didn't know any. In (51 a) the DP the doctor is ambiguous between a specific reading (referring to a contextually identifiable individual) and a non-specific reading (referring to any doctor). By adding the adnominal intensifier, as in (51 b), the reading of the doctor as non-specific is excluded. Hence, in (5lc) adnominal intensification is not felicitous. These examples clearly show that formal definiteness alone is not enough to qualify a DP for adnominal intensification. In the preceding we have showed that certain definite DPs cannot be intensified. As the example in (52) shows, in certain cases, even indefinite DPs can be intensified. (52) A: All Cretans lie. B: Where did you hear that? A: a. A Cretan himself told me. b. Cretans themselves told me. Based on this and similar examples Edmondson and Plank (1978) concluded that indefinite DPs may allow adnominal intensification if they are interpreted as having a specific reference. Does this mean that specificity is the correct generalization? While a large number of examples can be adduced in support of this hypothesis, e.g. (53a), the existence of examples like (53b ), from Siemund (2000: 162(6.1 05)) seems to refute it. 23 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)). (53) a. b. 62 *Unicorns themselves exist. Man himself is a product of his environment. It may be possible to explain the difference between (53a) and (53b) as following from a semantic difference between the DP unicorns, which is an indefinite plural DP, and the DP man, which can be argued to refer to a specific species (i.e. homo sapiens) which can be contrasted with other species (e.g. cat, dogs, spiders, etc.) 24 . Siemund (2000), who discusses these problems in great detail, concludes that the relevant constraint on adnominal intensification should be formulated as in (54). (54) Unique Identifiability Condition: In order to be successfully intensified by an adnominal intensifier a DP "must denote a uniquely identifiable referent where referent can be understood in the broadest sense of the word." (Siemund (2000: 154, 170(6.137)) As it stands this constraint is still formulated in rather vague terms. It is thus clear that more works needs to be done to arrive a more precise understanding of the linguistic principles involved 25 • At the end of the following section we argue that the condition in (54) may 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 adnominal intensification were discussed. In this section, semantic and pragmatic prominence conditions on intensification will be investigated in some detail and it will be shown that different constraints formulated in terms of prominence, centrality and contrastiveness, as well as the Unique Indentifiability Condition discussed above, see (54), can be subsumed under one contrastiveness requirement. 24 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 of our imagination. 63 It has long been noticed that there is a strong tendency for adnominal intensifiers to modify DPs which denote prominent individuals of high status 26 , see (55). (55) a. b. c. d. dronningen selv queen-the self 'the Queen herself' biskoppen selv bishop-the self 'the bishop himself' statsm inisteren selv prime minister-the self 'the prime minister himself' prcesidenten selv president-the self 'the president himself' Data such as (55) has lead to the formulation of different prominence or centrality requirements on adnominal intensification. Baker (1995) proposes the following condition on the use of intensive NPs. (56) Baker's ( 1995) Condition of Relative Discourse Prominence: Intensive NPs can only be used to mark a character in a sentence or discourse who is relatively more prominent or central than others. The notion of prominence is intended to cover both cases of lexical prominence, e.g. (55), and case of contextually defined prominence, e.g. (57). According to Baker (1995:79), the sentences in (57a) and (57b) both involve contrasting the custodians with other employees. Inspite of these similarities, intensification of the DP the custodians is only acceptable in (57a), but not in (57b). (57) a. All of the employees of 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. 25 For more detailed discussion of these issues. see Edmondson & Plank ( 1978) and Siemund (2000). 26 Cf. Keenan (1994) who observes this tendency for the adnominal intensifier se(o)/f in Old English. Cf also Moravcsik ( 1972). 64 b. All of the employees of 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 examples in (55) exemplify cases of expressions which, by their meaning alone, tend to refer to high-status individuals (e.g. kings, presidents, etc.) who, under normal circumstances, are prominent in most contexts. In contrast, Baker's example in (57) contains an intensified DP, i.e. the custodians, denoting low status individuals. Nevertheless, the example in (57a) is ok since, in the given context, the DP the custodians can be construed as referring to a contextually prominent group of individuals. According to Baker (1995:80), "the much greater naturalness of the intensive in [(57a)] as compared to that in [(57b)] follows directly from our perception that custodians are much more likely to play a central role in an episode of illegal trash disposal than in an episode of check-kiting". Baker therefore claims that these examples strongly support the existence of his prominence condition on intensification. Konig's (1998) notion of centrality, see (58), is similar to Baker's ( 1995) prominence condition. ( 58) 27 Konig's ( 1998) Conditions for the use of adnominal intensifiers: Adnominal intensifiers relate a center X (=referent of the focus) to a periphery of alternative values, such that: a. X has higher rank than Y in a real-world hierarchy. b. X is more important that Y in a specific situation. c. Y is identified relative to X (kinship terms, part-whole, etc .. ) d. X is the subject of consciousness, center of observation, etc .. The examples in (59a-d) illustrate the different kinds of centrality defined in (59a-d). (59) a. b. c. d. The Pope himself is against this view. Nobody cared about the fans when the fire broke out, but the rock star himself was quickly whisked away. Lucy's sister is more intelligent then Lucy herself 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. 27 The centrality condition in (58) as well as the examples in (59) are from Konig (1997). 65 Siemund (2000:154) who adopts a version of Konig's centrality requirement, admits that the existence of this constraint is very difficult, if not impossible, to test since "given a sufficient amount of adequate context or some imagination on the part of the decoder, almost any referent can be thought of as being central and hence associated with another referent which forms its periphery" (Siemund (2000: 154)). The elusive nature of the prominence and centrality requirements in (56), (58), as well as the virtual impossibility of constructing any example which clearly violate these constraints indicate that we would be well advised to look for a better formulation of the constraint or perhaps a different constraint altogether. McKay (1991) offers a possible solution, see (60). (60) McKay's (1991 :368) constraints on adnominal intensification: In order to be successfully intensified by an adnominal intensifier a DP must have a. "a clearly indicated referent", and b. "a relevant contrast or comparison class" The constraint in (60a) corresponds to the unique identifiability requirement discussed in the previous section. The contrastiveness requirement in (60b) offers an interesting alternative to the prominence and centrality requirements proposed by Baker (1995) and Konig (1997). Baker (1995) suggests that we need both the prominence requirement in (56) and a contrastiveness requirement which he formulates as in ( 61 ). (61) Bakers' (1995:77) contrastiveness condition: "Intensive NPs are appropriate only in contexts m which emphasis or contrast is desired" In the following, we argue that only McKay's contrastiveness condition (60b) is needed. Furthermore, using evidence from inherently reflexive constructions we show that, unlike the prominence and centrality requirements, solid evidence for the existence of a contrastiveness requirement, as formulated in (62), can be found. (62) Contrastiveness condition on adnominal intensification: A nominal expression (DP, pronoun, reflexive) can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. 66 Zribi-Hertz (1995:349-350) uses the incompatibility of intensification with inherently reflexive constructions such as (63-64) as evidence supporting a scalar analysis of intensifiers. In contrast, we propose to account for the absence of intensified DPs in inherently reflexive constructions as a consequence of the contrastiveness condition (62). (63) a. Bill, is out of his, *z mind. b. *Bill is out of John's mind. c. *Bill is out of your mind. d. *Bill is out of his own mind. (64) a. Bill; took a knife with him; *z· b. *Bill took a knife with John. c. *Bill took a knife with you. d. *Bill took a knife with himself Due to the inherently reflexive meaning of the constructions m (63-64), only expressions which are coreferential with 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 impossible to construct solid, unambiguous violations of Konig's centrality condition and Baker's prominence condition, we conclude that they are not needed. On the other hand, the existence of data like (63-4) clearly supports the assumption of a contrastiveness constraint as (62). Indeed, it seems possible to derive prominence and centrality phenomena from the contrastiveness condition in (62). Intensification of a DP is thus very similar to focus. Just like focus it triggers the generation of a set of alternatives which are contrasted with the intensified expression. Depending on the context, the relationship between the associate of the intensifier and the focus-generated contrast set may or may not be ordered in terms of prominence or centrality. In other words, centrality and 67 prominence are not essential parts of the process of intensification itself, but simply different interpretations which intensification may get in different contexts. At this point we have identified two conditions on adnominal intensification: the unique identifiability condition (54) and the contrastiveness condition (62). It seems very likely that these two conditions are connected 28 and that they may potentially be further reduced to even more general underlying principles. Detailed comparison of focus and adnominal intensification reveals a link between unique identifiability and contrastiveness. Intensification always involves the generation of a contrast set of alternatives to the referent of the associate of the intensifier. In order to generate such a contrast set one must be able to (uniquely) identify the referent of the associate. However, while is possible to subsume the unique identifiability condition (54) under the contrastiveness condition (62) in this way, for ease of exposition we will still continue to refer to the condition in (54) when testing different uses of intensifiers. 2.2.4 Is ad nominal intensification of PRO possible? As shown in (65) PRO resists adnominal intensification 29 • In this section we discuss three possible explanations ofthis fact. 28 Cf. Siemund (2000: 170) who also speculates that unique identifiability and centrality may somehow be ··intimately connected." 29 At tirst glance. Danish sentences like (i) may appear to falsify the claim that PRO cannot be adnominally intensified: (i) a. b. c. Peter lovede os selv at skrive en artikel om skolesystemet. Peter, 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." 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.' 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 himself.' 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: (I) a q-floated adnominal intensitier intensifying the matrix subject (ia). (II) an adnominal intensifier intensifying the pronoun os 'us' (ib). or. 68 (65) 30 a. Bill hopes PRO to win the election. b. *Bill hopes [PRO himse(f] to win the election If, as argued in 2.2.2.3, all nominal expressions (DP, pronouns, reflexives, traces, etc.) can be intensified in the same way by adjunction of the adnominal intensifier, one has to answer the question why PRO disallows intensification. That is, why is (65b) ungrammatical? Based on our analysis of intensification in (66) we have to argue that, with respect to the syntax of intensifier-adjunction, intensification of PRO is ok. That is, by itself, (67) should be syntactically well-formed. (67) [or [or [o PRO] ] [selv ]] 'PRO self Furthermore, since our analysis of English reflexives proposed in chapter 5 is based on the assumption that phonologically zero elements may be adnominally intensified, we cannot explain (65b) simply by assuming the existence of a contraint banning adnominal intensification of phonologically unrealized elements 31 • Since it not the syntax of intensfier-adjunction which is responsible for the ungrammaticality of (65), it has to be due to something else. In the following, three different hypotheses will be discussed: (i) an explanation based on the contrastiveness condition on adnominal intensification, (ii) an explanation based on the assumptions about the lexical properties of PRO, and (iii) an explanation based on the PRO-theorem. It seems very tempting to explain the absence of intensified PRO in terms of the contrastiveness condition on adnominal intensification (62). As the data in (69-71) show, the (III) a pre-verbal exclusive adverbial intensifier meaning "without help. all by oneself which is adjoined to the VP (ic). 30 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 himself 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 of intensitiers as ·exclusive' manner adverbials. 69 PRO constructions in (69) appear to violate the contrastiveness condition in the same way as inherently reflexive constructions (70-71) by not allowing for the creation of 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. d32. *Bill hopes [PRO himself] to win the election. (70) a. Bill, is out of his, z mind. b. #Bill is out of John "s mind. c. #Bill is out of your mind. d. #Bill is out of his own mind. (71) a. Bill, took a knife with him, z· b. #Bill took a knife with John. c. #Bill took a knife with you. d. #Bill took a knife with himself While the unacceptability of (69) and (70-71) appear to be due to violations of the contrasitiveness condition in (62), the ultimate reasons why these sentences fail to generate contrast sets are probably different. In the case of the inherently reflexive expressions in (70- 71) the failure to generate contrast sets, and thus meet the requirements of (62), is due to the semantics of these expressions. Syntactically there is nothing wrong with (70b-d) and (71 b-d). It is the meaning of these sentences which exclude the possibility of commutation of his and him with any other expressions. Since it is semantic rather than syntactic factors which are responsible for the unacceptability of (70-71 c-d), these sentences have been marked with the symbol "#", indicating semantic or pragmatic anomaly, rather than "*" which is reserved for syntactic anomaly. In the case of the PRO sentences in (69), however, the failure to generate contrast sets, and thus meet the requirements of ( 62), may be due to syntactic contraints. Overt DPs have to 31 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. Sec also discussion of fronted selv in section 2.2.6. 32 Note that this sentence may be acceptable to certain speakers when himself is analyzed as a q-tloated adnominal intensifier modifying the matrix subject DP Bill. see section 2.2.5 for more discussion of q-tloated intensifiers. 70 have case, while the PRO-theorem requires PRO to be case less (or have null-case, depending on analyses). As shown in (72-73), the ungrammatical sentences in (69b-d) can be saved by providing case-assigners for the subjects of the embedded clauses. (72) a. b. c. d. (73) a. b. c. d. Bill hopes that he will win the election. Bill hopes that John will win the election. Bill hopes that you will win the election. Bill hopes that he himself will win the election. Bill hopes for him to win the election. Bill hopes for John to win the election. Bill hopes for you to win the election. Bill hopes for himself to win the election. In (72) nominative case is assigned to the embedded subject by the finite T projection. In (73) accusative case is provided by the preposisitonfor. Thus the impossibility of generating a set of alternatives to the referent of PRO in (65b/69d) is due to the fact that case-constraints require overt DPs and PRO to be in complementary distribution. However, regardless of the different reasons for failure to satisfy (62), the ungrammaticality of (69d) and (70-71d) all appear to be reduced to violations of this contrastiveness principle. On closer inspection, however, this account may not be as succesful as it seems. Since the contrastiveness condition is of semantic nature, there is no reason why generation of a constrast in (69) should not possible. After all, the sentences in (72-73), clearly show that semantically the contrast sets can be created once the syntax allows it. Let us now tum to another, perhaps more promising, account of the absence of intensified PRO, which is based on the assumption that PRO is incompatible with intensification. As noted by Chierchia (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 semantic paraphrases in (74). 71 (74) a. John hopes that A.x[x wins the election] (DE SE, PRO, (69-70)) b. Johni hopes that hei wins the election (DE RE, non-PRO, (72-73)) The DE SE reading in (74a) establishes a relation between John and a property, in this case the property of winning the election. In contrast, since the paraphrase in (74b) does not imply self-ascription of any property, it is ambiguous between DE SE and DE RE readings. Assuming PRO to be lexically specified to require a bound-variable interpretation may then explain the incompatibility of PRO and adnominal 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]i hopes PROi to be elected) it is necessary to assume that PRO must be bound by the A.-operator in the closest Comp. Finally, an even simpler way to explain the apparent absence of adnominally intensified PRO, consists in assuming that adnominal intensification of phonological zero elements turns these elements into overt DPs which must have non-null case. Since PRO is per definition in a case-less (or null-case) position, intensification of PRO via intensifier- adjunction would necessarily lead to violations of case-requirements. See also discussion of the possibility of adnominally intensifYing PRO in English and Chinese in chapters 5 and 6 respectively. Notice that this account would not run into any problems with 0-reflexives in English and Chinese, since these all occur in case-positions. Furthermore, this account needs no extra stipulations, but follows directly from the PRO-theorem. If this last, case-based account of the absence of intensified PRO is correct then it constitutes a potentially powerful argument against current analyses of PRO as receiving zero Case. At this point the third explanation seems to be the most straightforward 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 of the present dissertation. 72 2.2.5 Selv used as q-floated adnominal intensifier In this section we demonstrate how the analysis of intensification proposed above can be used to unify the account of two uses of selv which have hitherto been assumed to be of very different nature, viz. the adnominal intensifier and what has been called the "inclusive adverbial intensifier". The terms "inclusive" and "exclusive" adverbial uses of selv were introduced by Ekkehart Konig to refer to what he takes to be different adverbial uses of intensifiers. The use of the intensifier himse(f in the sentence Peter had himself written an essay is "inclusive" in the sense that it presupposes the inclusion of Peter into the set of individuals who have written essays. The use of himself in the sentence Peter had written the essay himself is exclusive in that it excludes all agents but Peter, i.e. Peter wrote the essay without help/all alone. When the intensifier selv is placed sentence-medially, 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 selfwritten an essay 'The teacher had himself written an essay.' Although selv in this usage has the appearance of an adverbial element we propose to analyze it as being derived from an adnominally intensified DP. That is, we propose that the selv in (75) is 'floated' in the same sense that the quantifier aile 'all' in (76) is stranded by quantifier-floating (cf Sportiche (1988)), see (77) which gives the syntactic tree for the sentences in (75-76). (76) Eleverne havde aile skrevet en stil. Pupils-the had all written an essay 'The pupils had all written an essay.' 73 (77) CP I \ Lcereren, C' \ IP I \ DP I' I I \ l; fz VP I \ DP V' I \ I \ DP selv v DP I I I f; skrevet en stil As shown in (77) q-floated selv is syntactically derived from adnominal selv. The existence of this derivational relationship is supported by the fact that the interpretation of selv in this usage is closely related to the meaning of adnominal selv. The adjunction of selv to a DP, e.g. the teacher himself had written an essay, may serve to emphasize the teacher's involvement 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)1(77) appears to have the same semantic contribution. As mentioned in section 2.2.3.2.1, the distribution of adnominal selv is limited the unique identifiability constraint on the DP which is being intensified, see (54) 33 • The examples in (78- 83) illustrate that quantifier-floated selv also displays the same identifiability effects and thus has more in common with adnominal selv than with selv used as an "exclusive" manner adverbial. (78) (79) *lngen selv pudsede skoene. nobody self polished shoes-the '*Nobody himself polished the shoes.' Statsministeren selv pudsede skoene. prime minister-the self polished shoes-the 'The prime minister himself polished the shoes.' (adnominal selv) (adnominal selv) 31 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 docs not exist. If we continue to refer to the unique identifiability condition, it is therefore only for ease of exposition. (80) Ingen pudsede skoene selv. ("exclusive" selv) nobody polished shoes-the self 'Nobody polished the shoes all by himself' (81) Statsministeren pudsede skoene selv. ("exclusive" selv) prime minister-the polished shoes-the self 'The prime minister polished the shoes all by himself.' 74 While adnominal 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) which are both fine. The question is now whether or not quantifier-floated selv is subject to the same unique identifiability constraint. On the basis of the proposed analysis of quantifier-floated selv, see (77), we predict that sentences with q- floated selv should display some sort of identifiability effects. The examples in (82) and (83) offer evidence in favor of this hypothesis. (82) a. ??1*/ngen pudsede selv skoene. (q-floated selv) nobody polished self shoes-the '*Nobody polished himselfthe shoes.' b. Statsministeren pudsede selv skoene. (q-floated selv) prime minister-the polished self shoes-the 'The prime minister himself polished the shoes.' (83) a. *Ingen er selv en klovn. (q-floated selv) nobody is self a clown '*Nobody is himself a clown.' b. Statsministeren er selv en klovn. (selv q-floated) prime minister-the is self a clown 'The prime minister is himself a clown.' 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 adnominal intensifier which is stranded when its associate moves to a higher position, see (77). That (82a) seems slightly more acceptable to some speakers than (85a), is due to the nature of the predicate. In (82) the predicate pudse sko 'polish shoes' is compatible with agent-oriented adverbs, while in (83) the predicate were en klovn 'be a clown' is not. This 75 means that in (82) selv can be forced to be read as an exclusive adverbial, which is compatible with indefinite subjects, while such a reading is excluded by the non-agentive predicate in (83). The impossibility of combining exclusive adverbial selv with non-agentive predicates is iII ustrated by the sentence in (84 ). (84) ??Statsministeren er en klovn selv. prime minister-the is a clown self (Compare with (81 )) As already mentioned exclusive adverbial selv is a manner adverbial which usually occurs predicate-finally. While such a predicate-final exclusive adverbial selv is ok with the agentive predicate pudse sko 'polish shoes', as shown by (81 ), it is quite infelicitous in (84 ). Unlike the exclusive manner adverbial selv 'all by oneself, the q-tloated selv 'too, also' does not imply uniqueness of the agent, see the example in (83b) which contains a predicate which is incompatible with manner adverbials. The meaning of (83b) is not incompatible with scenarios in which people other than the prime minister are clowns. In fact, (83b) seems to imply the existence of a set of propositions of the form 'xis a clown', i.e. the focus generated contrast set, in which at least one proposition other than 'The prime minister is a clown' (i.e. the foreign minister is a clown, the secretary is a clown, etc.) is also true. In other words, (83b) implies the inclusion of the proposition 'the prime minister is a clown' into the non-empty set oftrue propositions ofthe form 'xis a clown'. In this respect, the meaning of the inclusive selv 'also, too' seems to be similar to the meaning of the additive particle ogsa 'also', compare (83b) and (85). (85) Statsministeren er ogsaa en klovn. prime minister-the is also a clown 'Peter is also a clown.' Like q-tloated selv 'too, also', the additive focus particle ogsa 'also' is also usually placed immediately after conjugated verb or auxiliary (unlike the exclusive manner adverbial selv 'all by oneself which is VP final), see (86). (86) *Statsministeren er en klavn agsaa. prime minister-the is a clown too 76 (compare with (85) and (87)) Note, however, that inclusive selv and the additive focus particle agsa 'also' can co-occur in the same sentence, see (87). (87) Statsministeren er agsaa selv en klavn. prime minister-the is also self a clown 'The Prime minister is also a clown himself.' The fact that inclusive selv and the additive focus particle agsa 'also' can co-occur in the same sentence is a strong indication that they do not fulfill exactly the same semantic function(s). Otherwise, such co-occurrence would be impossible. The 'additive' aspect of the meaning of inclusive selv seems to indicate that it might somehow be related to the pre-nominal additive focus particle selv 'even', discussed in the previous section, which also evokes additivity. However, unlike the additive focus particle selv 'even', q-floated selv does not necessarily evoke a scale of expectedness. Nor does it seem to involve the same kind of additivity. The presence of additivity can be tested using clefts, see (88). (88) a. b. c. Det var dranningen selv, sam var blevet namineret til en Oscar. it was queen-the self who was become nominated to an Oscar "It was the Queen herself who had been nominated for an Oscar." (?)Det var dronningen, sam selv var blevet namineret til en Oscar. it was queen-the self who was become nominated to an Oscar It was the Queen who had herself been nominated for an Oscar. *Det var selv dranningen, sam var blevet namineret til en Oscar. it was queen-the self who was become nominated to an Oscar *It was even the Queen who had been nominated for and Oscar. Cleft-constructions presuppose uniqueness and are thus incompatible with additive focus particles like selv 'even' in (88c). In contrast, since the adnominal intensifier selv does not presuppose additivity, it is fine in cleft-constructions like (88a). If q-floated selv is basically a stranded adnominal intensifier, then it should also be compatible with cleft-constructions. As shown in (88b) this prediction seems to be borne out by the facts. The sentences in (88b) may be slightly odd, but it is far from being just as unacceptable as (88c). In other words, the 77 apparent additivity of q-floated selv is not part of its lexical make-up (as in the case of 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 adnominal selv and q-floated selv is provided by the examples in (89-91), which illustrate the possible combinations of adnominal selv, q-floated selv and "exclusive" adverbial selv within a single sentence. (89) Statsministeren selv pudsede skoene selv. (adnominal selv + "exclusive" selv) prime minister-the self polished shoes-the self 'The prime minister himself polished the shoes without help from any one.' (90) Statsministeren pudsede selv skoene selv. (q-floated selv +"exclusive" selv) prime minister-the polished self shoes-the self 'The prime minister himself polished the shoes without help from any one.' (91) #Statsministeren selv pudsede selv skoene. (adnominal selv + q-floated selv) prime minister-the self polished self shoes-the 'The prime minister himself (himself) polished the shoes.' Combinations of adnominal selv and "exclusive" adverbial selv, and of q-floated selv and "exclusive" adverbial selv are both fine, see (89) and (90). However, combinations of adnominal selv and q-floated selv are infelicitous (or at best highly redundant and very odd), see (91). The generalization seems to be that one instance of "exclusive" selv can be combined with one instance of adnominal selv ( q-floated or not), but two instances of adnominal selv cannot be combined in one sentence (even if one is q-floated and the other is not). This incompatibility of adnominal selv and q-floated selv confirms the hypothesized derivational relationship between the two 34 • A further indication that sentence-medial, q-floated selv is derived from the adnomnal intensifier comes from stress placement. While the post-nominal adnominal intensifier is 34 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. 78 always stressed (e.g. kangen SELV 'the king HIMSELF'), the additive focus particle is always unstressed (e.g. selv KONGEN 'even the KING'). As illustrated in (92), q-tloated selv is necessarily stressed. (92) a. Peter er SELV en klavn. Peter is self a clown 'Peter is HIMSELF a clown' b. ?? PETER er selv en klavn. Peter is self a clown 'PETER is himself a clown' c. ??Peter er selv en KLOVN. Peter is self a clown. 'Peter is himself a CLOWN.' Again, this may be taken as support for the hypothesis that q-tloated selv is derived from the adnominal intensifier, which is also stressed. In the present state of the language q-tloated selv-constructions seem to be found more often than non-q-tloated adnominal selv-constructions which tend to be found most often with a simple unmodified DPs (e.g. Stalin, Jesus, the Pope, etc.) 35 • In contrast, q-tloated selv- constructions can be found with any type of DP regardless of their complexity, see (93) vs. (94). (93) a. b. (94) a. Manden (*sam var skaldet) selv kam til festen. man who was bald self came to party-the 'The man who was bald himself came to the party' Den gamle mand (*med bla buher) selv kam tilfesten. the old man with blue trousers self came to party-the 'The old man with blue trousers himself came to the party' Manden sam var skaldet kam selv til fest en. man-the who was bald came self to party-the 'The man who was bald himself came to the party' 35 Another factor to be taken into consideration is that fact that non-q-floated selv-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 tor these differences between adnominal intensifiers in-situ and q-tloated intensifiers. This might explain why q-tloated selv is preferred with DPs with long post-nominal modifiers. b. Den gamle mand med bla bukser kom selv til fest en. the old man with blue trousers came self to party-the 'The old man with blue trousers himself came to the party' 79 As discussed in section 2.2.2.2, examples such as (93) seem to indicate a difference with respect to stacking options. Simple DPs, e.g. (93a,b) without the post-nominal relative clause/PP, can be adnominally intensified without any problems. In contrast DPs modified by post-nominal relative clauses/PPs can only be intensified via a q-tloated selv, compare (93) vs. (94). Syntactically the intensifier selv behaves very much like the quantifier aile 'all' which also has both a pre-nominal adjectival form and a post-nominal appositional form. The post- nominal quantifier aile 'all' modifying direct and indirect object pronouns is illustrated by the sentences in (95). (95)36 a. Lcereren roste OS aile. teacher-the praised us all. 'The teacher praised us all.' b. Peter gav dem aile en krone. Peter gave them all a crown (=monetary unit in Denmark). 'Peter gave them all a crown.' Just as the intensifier selv has a pre-nominal adjectival form (i.e. selve, see section 2.2.6), the quantifier aile also has a pre-nominal adjectival form (i.e. al, alt. aile), see (96). (96) a. er Alt all IS 'All the beef is sold out.' oksek0det beef-the udsolgt. sold out. 36 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-b). (i) a. b. La'reren roste aile teacher-the praised all 'The teacher praised all the pupils.' Peter gav aile bernene eleverne. pupils-the Peter gave all children-the 'Peter gave all the children a crown'. en krone. a crown This indicates that the quantifier aile 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. 80 b. Jeg inviterede aile mine venner. I invited all my friends 'I invited all my friends.' c. Du far at den hjcelp(common) du behcJVer. you get all the help you need 'You're getting all the help you need. As already illustrated by the example in (76), aile may be q-floated just like selv, compare 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 with result-the 'We are all satisfied with 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 of q-floated intensifiers in mind, let us now turn to ambiguous sentences in which selv can be read as either an adnominal intensifier modifying the direct object, see (98i) and (99i), a q-floated selv, see (99ii), or an "exclusive" adverbial selv (98ii) and (99iii). (98) (99) Peter Peter Peter Peter malte muren selv. painted wall-the self (i) 'Peter painted the wall itself.' (ii) 'Peter painted the wall himself.' malte den selv. painted it self (i) 'Peter painted it itself.' (ii) 'Peter himself painted it' (iii) 'Peter painted it himself.' (adnominal selv) ("exclusive" adv. selv) (adnominal selv) (q-floated selv) ("exclusive" adv. selv) The first reading (98i), in which selv is an adnominal intensifier modifying the internal object DP muren 'wall-the', is illustrated in (I 00). 81 (100) CP I \ Peter, C' I \ maltez IP I \ DP I' I I \ t, tz VP I \ DP V' I I \ t, v DP I I \ tz DP selv I muren While in (99) selv may have both an "exclusive" (99iii) and a q-floated reading (99ii) in addition to the adnominal reading, in (98) selv can only be either adnominal (98i) or "exclusive" (98ii). These differences between (98) and (99) can be shown to follow from Mainland Scandinavian (MSc.) object shift. The sentences in (101) illustrate how the placement of negation can be used to test whether object shift has taken place or not. (101) a. b. c. Peter malte ikke muren selv. Peter painted not wall-the self 'Peter didn't paint the wall himself.' *Peter malte muren ikke selv. Peter painted wall-the not self Peter malte den ikke selv. Peter painted it not self (i) 'Peter didn't paint it himself (ii) 'Peter himself didn't paint it.' d. *Peter malte ikke den selv. Peter painted not it self ("exclusive" selv) ("exclusive" or q-floated selv) For the analysis of MSc. object shift and sentence structure in general we follow Holmberg & Platzack ( 1995). In Danish, Norwegian and Swedish, certain types of objects may be object- shifted, i.e. they may precede the negation words and similar predicate adjuncts, see (!Ole). In MSc., as opposed to Icelandic, object-shift is restricted to personal pronouns, see (lOlb) vs. 82 ( 101 c). We assume that object shifted pronouns have moved out of their argument positions to spec-vP (cf. Platzack and Holmberg (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 of the predicate then the fact that the pronoun den 'it' in (101c) can precede it while the DP muren 'the wall' cannot shows that Danish object pronouns can shift while full DPs cannoe 7 • Holmberg and Platzack (1995:143) observed that object shift is subject to certain restrictions. First of all, shifted objects never crosses the main verb. This means that object shift only happens after overt movement of the main verb out of the predicate (i.e. vP) has taken occurred. Given that auxiliary verbs block verb-movement, object shift is never found in clauses with auxiliary verbs. The Danish sentences in (102-103) support this analysis. (102) a. b. c. (103) a b. c. Peter havde (ikke) malt muren selv. Peter had (not) painted wall-the self 'Peter had not painted the wall himself.' Peter havde (ikke) selv malt muren. Peter had (not) self painted wall-the 'Peter himselfhad not painted the wall.' *Peter havde muren (ikke) malt. Peter had wall-the (not) painted Peter havde (ikke) malt den selv. Peter had (not) painted it self 'Peter had not painted it himself.' ? Peter havde (ikke) selv malt den. Peter had (not) self painted it 'Peter himselfhad not painted it.' *Peter havde den (ikke) malt. Peter had it (not) painted ("exclusive" selv) (q-floated selv) ("exclusive" selv) (q-floated selv) Compare (101c) where object shift is allowed with (103c) where the auxiliary havde 'had' blocks object shift. The examples in (102-3) also illustrate the fact that auxiliaries help disambiguate between q-floated selv (which is stranded in the YP-internal subject position), see (102-3b) and the predicate-final manner adverbial, i.e. "exclusive" selv in (102-3a). 37 As for the movement rule responsible for object shift, H&P propose that "it is a ""mixed rule'', exhibiting a particular mix of properties of A-movement and A-bar movement"(H&P, p.l42). Since the exact nature of this rule 83 Finally, the main vs. embedded clause contrast in object shift can also be used to disambiguate between different uses of selv 'self. Since in MSc. movement of the verb out of the predicate only takes place in main clauses, it follows that object shift is also is restricted to main clauses, see (1 04). (1 04) a. b. c. d. * ... at Peter malte ikke muren selv. . .. that Peter painted not wall-the self ... at Peter ikke malte muren selv. ... that Peter not painted wall-the self ' ... that Peter didn't paint the wall himself.' * ... at Peter malte den ikke selv. . .. that Peter painted it not self ... at Peter ikke malte den selv. ... that Peter not painted it self ' ... that Peter didn't paint it himself.' ("exclusive" selv) ("exclusive" selv) Given what we know about object shift, the placement of the negation and MSc. sentence structure in general, it is now fairly easy to explain why selv in (99) is ambiguous between either q-floated selv, or "exclusive" adverbial selv, while in (98) it can only be an "exclusive" adverbial intensifier (in addition to the adnominal reading which both sentences have). The structural ambiguity of (99) has been teased apart in (1 05-6), (1 05) showing the structure underlying the q-floated reading, and (1 06) showing the structure underlying the exclusive adverbial reading of selv. is not important for the purposes of the present analysis of selv we simply refer the reader to H&P ( 1995). 84 (105) CP I \ Peter, C' I \ maltez TP I \ DP T' I I \ t, fz vP38 I \ den, vP I \ DP v' I \ I \ t, selv tz VP I V' I \ (I 06) CP v DP I \ I I Peter; C' fz t, I \ maltez TP I \ DP T' I I \ t, fz vP I \ den, vP I \ DP v' I I \ t, fz VP I \ VP selv I V' I \ v DP I I t= t, 38 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 (I 05-6) are the same as the ones proposed by Platzack and Holmberg. In the object shift constructions "the verb ha<> moved out of the VP, first to Act 0 , then to I 0 and eventually to C 0 " and the "object pronoun[ .. ] is left-adjoined to the predicate. i.e. ActP'' (ibid. pp. 143, 140). 85 Let us briefly summarize the findings of this section. The analysis of adnominal 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 the "exclusive" one. (ii) that the sentence in (99) can have both the "inclusive" and the "exclusive" readings, see (I 05) and (I 06). (iii) that adnominal selv and q-floated selv cannot felicitously be combined in the same sentence; since they are really both instances of adnominal intensification, see (91 ). (iv) that exclusive adverbial selv can be combined with adnominal selv or q floated selv in the same sentence. Multiply ambiguous sentences like (99) illustrate one of the main challenges faced by anyone researching the grammar of intensification: the need to clearly distinguish between different uses of intensifiers. The fact that in many sentences the element selv is multiply ambiguous makes it very difficult to get consistently reliable judgements. Further adding to this difficulty is the fact that grammaticality jugdments of sentences containing intensifiers are usually highly susceptible to discourse- and pragmatic contexts. The above analysis of the so-called inclusive adverbial selv as a q-floated adnominal selv illustrates one of the goals ofthis dissertation: to reduce as many seemingly different uses of intensifiers as possible to the same underlying principles. In this case, what was formerly believed to be an adverbial use of the element has been shown to be an instance of adnominal intensification falling under the contrastiveness condition (62). It also shows that DP-traces may be adnominally intensified. 2.2.6 Fronted selv As shown in (I 07) and (I 08), the adnominal intensifier selv may be fronted. (107) a. Jeg SELV t0r aldrig modsige ham. I self dare never contradict him 'I myself dare never contradict him.' (adnominal selv, in situ) b. (108) a. b. Kongen SELV har ringe magt. king-the selfhas slight power 'The king himself has little power.' SELV tor jeg aldrig modsige ham. self dare I never contradict him 'I myself dare never contradict him.' SELV har kongen ringe magt. self has king-the slight power 'The king himselfhas little power.' 86 (adnominal selv, in situ) (fronted adnominal selv) (fronted adnominal selv) Such instances of fronted selv should not be confused with the pre-nominal scalar, additive focus adverb selv 'even', see the examples in (I 09) 39 • (109) a. b. Selv JEG tor aldrig modsige ham. even I dare never contradict him 'Even I dare never contradict him.' Selv KONGEN har ringe magt. even king-the has slight power 'Even the king has little power.' (scalar additive focus particle) (scalar additive focus particle) Both the Danish sentences and their respective English translations in (I 08-9) are adapted from Allan et a!. ( 1995: 169). The authors of this descriptive grammar of Danish appear to assume fronted selv to be derived from the post-nominal adnominal intensifier by a productive syntactic process of fronting. While we agree with this assumption, we also realize the need for independent tests to exclude the possibility of fronted selv being derived from the scalar, additive focus particle 'even'. Just like in the case of q-tloated selv, discussed in section 2.2.5, different tests can be applied: (i) the so-called unique identifiability constraint on adnominal intensification, and (ii) relative stress placement,. i.e. does stress fall on the element selv itself or on the expression it is in construction with? Since we know that the element selv is stressed when used as adnominal intensifier, e.g. kongen SELV 'the king himself in (I 07b ), the fact that fronted selv is stressed, see (I 08), could be taken as an indication that fronted selv is an instance of the adnominal intensifier. 39 See section sect. 2.4. for more discussion of the scalar. additive focus adverb se/v ·even". 87 However, this argument is rather weak since the stress on fronted selv can also be argued to follow from general properties ofthe process of fronting, i.e. all fronted elements are stressed. The examples in (II 0) are intended to test whether fronted selv (II Oa) is subject to the same unique identifiability constraint as adnominal intensification (II Ob ). (110) a. b. c. */??Selv har en konge ringe magt. self has a king slight power 'A king himself has little power.' */??En konge selv har ringe magt. a king self has slight power 'A king himself has little power.' Selv en konge har ringe magt. even a king has slight power 'Even a king has little power.' The unacceptability of (II Oa) suggests that we are correct in assuming that fronted selv derives from the adnominal intensifier selv 'himself (I lOb) rather than from the additive focus particle selv 'even', which is ok with indefinite DPs, see (IIOc). We therefore conclude that fronted selv is sub-case of adnominal intensification. Interestingly, the examples in (I 08), as well as all the other examples of fronted selv mentioned in Allan et al. (1995) all contain a negative element: "Selv 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 of fronted selv, viz. that it appear to be possible only in negative sentences, may be related to the fact that in the Slavic languages, adnominal intensifiers, which, incidentally, are pre-nominal, e.g. Russian sam DP:::: (Eng.) DP himself, may be interpreted as additive focus particles with the meaning 'even' only when occurring in negative sentences, i.e. downward entailing environments 40 • It may therefore be possible to analyze fronted selv as an intermediary case, between the post-nominal adnominal intensifier and pre-nominal scalar, additive focus particle 'even'. In section 2.4 we suggest 40 This generalization and the potential consequences for the analysis of fl·onted selv in Danish were brought to my attention by R. Pancheva (p.c). 88 that the differences between selv used as pre-nominal additive scalar focus particle and selv used as post-nominal adnominal intensifier follow from syntactic differences (i.e. post- vs. pre- nominal position). If this is indeed the case, then the fact that fronted instances of the adnominal intensifier selv should have some properties of the scalar, additive focus particle 'even' (viz. the fact that it is only found in downward entailing environments Gust like pre- nominal cam 'alone, himself, even' in Bulgarian when used as scalar additive focus particle 'even')) should come as no surprise. In other words, it might be possible to use the similarities between fronted selv in Danish and the additive scalar focus particle cam 'alone' in Bulgarian to bolster the argument that syntactic position matters, i.e. the same element selv takes on different properties post-nominally and pre-nominally. 2.2. 7 Adnominal intensification: unification of different sub-cases At this point two of the "threefold" goals of this chapter mentioned on page 4 have been achieved. That is, we have proposed a focus-based analysis of adnominal intensification, see (62) in section 2.2.3.2.2, and shown that the different uses of selv listed in (Ill) are all instances of adnominal intensification and thus fall under the condition in (62). (111) Adnominal Intensification: Example: Section: I. intensified DPs ( 1 ), ( 14) (sect. 2.2) II. intensified reflexives (4) (sect. 2.1, chap. 3, sect. 3.3) Ill. intensified pronouns (5) (sect. 2,1, chap. 3, sect. 3.4) IV. q-fioated intensifier (77) (sect. 2.2.5) v. fronted selv (108) (sect. 2.2.6) Only the third goal, viz. the unification of all uses of selv- both adnominal and non-adnominal -remains and will be dealt with in the following sections as summarized in (112). (112) Remaining uses ofselv: a. selv in secondary predication constructions b. scalar additive focus particle 'even' c. selv as a noun d. adjectival forms ofthe intensifier e. selv in nominalizations 2.3 Selv in secondary predication constructions Example: (7),(113) (8), (120b) (9), (ISO) (10), (152) (II), (160) Section: (sect. 2.3) (sect. 2.4) (sect. 2.5) (sect. 2.6) (sect. 2.7, ch. 7) 89 In this section we argue that what has often been referred to as the "exclusive" adverbial use of 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 himself Syntactically, this particular use of selv appears to behave like a manner adverbial adjoining to the VP. Semantically, its effect is to exclude all but the referent of the subject DP as agent of 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 modified by the element heft 'all, completely', see ( 114a,b ). (114) a. b. Peter reparerede bilen heft selv. Peter repaired the car completely self 'Peter repaired the car all by himself.' Peter reparerede bilen heft alene. Peter repaired the car completely alone 'Peter repaired the car all alone.' Further support for assuming the instances of selv in ( 113-114) to be manner adverbials comes from sentences like the ones in (liSa-b) which show that neither selv in the meaning 'all (by) oneself nor alene 'alone' can occur sentence-medially but must occur predicate-finally, or outside the predicate, to be more exact 41 • The examples in (116) also illustrate that the manner adverbial hurtigt 'fast' behaves in the same way. 41 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- ( 115) 42 a. b. (116) a. b. *Peter malte helt selv huset. Peter painted completely self house-the 'Peter painted the house all by himself. *Peter reparerede (heft) alene bilen. Peter painted completely alone house-the 'Peter repaired the car all alone' *Peter malte hurtigt huset. Peter painted fast house-the Peter malte huset (meget) hurtigt. Peter painted house-the (very) fast 'Peter painted the house very fast.' 90 Like alene 'alone', selv usually occurs predicate-finally. Notice that the exclusive manner adverbial selv 'all by oneself' and alene 'alone' share the "exclusive" meaning with the focus particle kun 'only', see (117). ( 117) a. Kun Peter reparerede bilen. Only Peter repaired car-the 'Only Peter repaired the car.' The common meaning component of exclusive selv '(by) himself, the adjective alene 'alone', and the focus particle kun 'only' (in these sentences) then, consists in identifying the agent of the predicate as unique. Finally, as already mentioned in section 2.2.5, sentence-final selv can lead to a structural ambiguity between a reading as adnominal intensifier and a reading as exclusive manner adverbial, see ( 118a). self, even, etc. •. Bul. sam 'alone, x-selt~ even, etc. •. Notice that in both English and Danish, the word alone can, in certain contexts. be replaced by the PP by himself 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.' 42 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. (118) a. Peter barberede kongen selv. Peter shaved king-the self 91 (i) 'Peter shaved the king himself.' (selv = adnominal intensifier, modifying the DP kongen 'king-the' to which it is adjoined) (ii) 'Peter shaved the king all by himself.' (selv =exclusive manner b. Peter barberede kongen heft selv. Peter shaved king-the completely self 'Peter shaved the king all by himself.' adverbial modifying the VP) (selv = excl. adv. intensifier) As shown in (118b) the insertion of the adverbial modifier heft 'completely, all' between the DP kongen 'the king' and the element selv has the effect of disambiguating in favor of the exclusive adverbial reading of selv. See section 2.2.5 for further discussion of other ways (insertion of negation, object shift, relative placement of auxiliaries, etc.) to disambiguate between different uses of selv. 2.4 Se/v used as scalar additive focus particle meaning 'even' Typological studies have shown that in many languages, the same element which is used as adnominal 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, German, French, Persian, Bulgarian and Russian can use the same word in both cases, compare the (a) and (b) examples in ( 119-124 ). ( 119) English: a. The king himself delivered the speech. ( adnominal intensifier) b. Even the janitor delivered a speech. (scalar additive focus particle) (120) Danish: a. Kongen selv holdt talen. king-the self held speech-the 'The king himself delivered the speech.' b. Selv viseva:rten holdt en tale. self janitor held a speech 'Even the janitor delivered a speech.' (adnominal intensifier) (scalar additive focus particle) (121) German: a. Der Konig selhst hielt die Rede 43 • (adnominal intensifier) the king self held the speech 'The king himself delivered the speech.' b. Selhst 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. (adnominal intensifier) the king him self has pronounced the discourse 'The king him-selfhas delivered the speech.' b. Meme 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 sherkat kard. (adnominal intensifier) self king in this meeting participated 'The king himself participated in this meeting.' bN. (i) Xod-e pddeshdh ham dar injalaseh sherkat kard. (focusing particle) self king also in this meeting participated 'Even the king participated in this meeting.' (ii) Halla pddeshdh (ham) dar injalaseh sherkat kard. (focusing particle) even king also in this meeting participated 'Even the king participated in this meeting.' (124) Bulgarian: a. Samijat kral dojde na sdbranieto. self-the king came to meeting-the The king himself came to the meeting.' b. Samijat kral ne dojde na sdbranieto. self-the king not came to meeting-the 'Even the king did not come to the meeting.' 43 The German and French examples in ( 121 a-b) and (122a-b) are adapted from Eckhardt (2000: I, (I). (2)) 92 44 As illustrated in (i). in Persian the morphologically simple clement 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-iigad.(bad iimadan '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 (7). Whether such an analysis is compatible with the facts of Persian is beyond the scope of the present chapter. See chapter 5 for 93 As shown in (120), in Danish the scalar additive focus particle even(= Dan. selv Peter 'even Peter') and the adnominal intensifier himself(= Dan. Peter selv 'Peter himself) have the same overt morphological realization. Furthermore, adnominal intensifiers and even both involve association with focus. Even also obligatorily involves evocation of a contextually determined scale on which the value of the expression that is the focus of selv is ranked as the least expected/most remarkable/etc. In contrast, whether adnominal intensifiers evoke scalar ranking of their associates with respect to a contrast set of alternatives depends on the context. In that respect adnominal intensification is similar to contrastive focus. Another semantic difference between the adnominal intensifier and even lies in the existential presupposition of even which is not shared by the adnominal intensifier. That is, while (119-124b) presuppose that somebody other than the janitor also delivered a speech, ( 119-124a) do not evoke any such presupposition, i.e. the king may or may not have been the only individual who delivered the speech. The sentences m (123b(i)) and (123b(ii)), illustrate the fact that, in Persian, the meaning of the additive focus particle hatta 'even' can be expressed by a combination of the adnominal intensifier xod 'self and that additive particle ham 'also, too'. Given that one of the main semantic difference between the adnominal intensifier and the scalar additive focus particle 'even' is the presence of an existential presupposition with 'even', this should come as no surprise. Notice the interesting correlation between syntactic position and the difference between the use of intensifiers as adnominal intensifiers and additive focus particles: in German, Mainland Scandinavian and French, intensifiers function as adnominal intensifiers in post-nominal position and as additive focus particles when in they occur in pre-nominally discussion of a similar intensifier-based analysis of (what appear to be reflexive uses of) himself in English as an adnominal intensifier adjoined to a 0-reflexive. 94 position, see (120-122a) vs. (120-122b). As illustrated by the Bulgarian examples in (124), in the Slavic languages intensifiers are always pre-nominal both when interpreted as adnominal intensifiers, as in ( 124a), and when interpreted as scalar additive focus particles, as in (124b ). Interestingly, the reading of intensifiers as scalar additive focus particles can only be obtained in downward-entailing environments. That is, without the negations 'not' the element sam 'self in (124b) would be interpreted as adnominal intensifiers 45 and not as scalar additive focus particles. In addition to the above examples illustrating the close synchronic relationship between focusing particles and intensifiers, there is also ample historical data showing that these elements are intimately related. Indeed, diachronic studies of German have established that the use of selbst as additive focusing particle developed from the adnominal intensifier selbst around 1700 ( cf. Eckhardt (2000)) 46 • Siemund (2000) notes a number of similarities between the German adnominal intensifier selbst 'himself' and the additive scalar focus particle selbst 'even' and concludes that they "are basically the same word, simply subject to slightly different (syntactic, semantic, pragmatic) patterns of use" (Siemund, 2000:ch.5). A similar approach to the similarities between intensifiers and additive focus particles will be adopted here. That is, we suggest that the different behavior of the intensifier selv and the 45 A similar connection between pre-nominal position and dependence on the presence of 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 of this dissertation we will simply leave it for further research. 46 The example in (i). from S. Kierkegaard (1843) Frygt og Bt:even, illustrates a transitional stage of early Modem Danish in which selv could be used post-nominally with the meaning 'even'. (i) a. Hun[. .. ] kan [. .. .] bringe stene selv til at gra'de [. .. ] 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. 95 focusing particle selv may potentially be accounted for by assuming that the same morpheme selv takes on different functions when it occurs in different syntactic positions 47 . 2.4.1 Syntax of the scalar, additive focus particle selv 'even' In English the adnominal intensifier himself and the scalar additive focus particle even are two different words. In Danish, however, the same element selv is used in both meanings. One way to tell the two uses apart is the placement of selv with respect to the nominal expression serving as its associate. As shown in the examples ( 119-125), the adnominal intensifier follows immediately after its associate, which has to be a nominal expression, i.e. a DP. In contrast, when used as scalar additive focus particle, selv immediately precedes its associate, which 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. b. c. d. e. Selv Peter var i stand til at lose problemet. even Peter was in state to to solve problem-the 'Even Peter was capable of solving the problem.' Selv ham ville hun gerne kysse. even him would she willingly kiss 'She was willing to kiss even him.' Han betalte selv for hende. he paid even for her 'He paid even for her.' Selv nar det regner er han glad. even when it rains is he happy 'He is happy even when it rains.' Selvom Janet er englcender, taler hunjlydende dansk 48 • even-if Janet is Englishman speaks she fluent Danish 'Although she's English, Janet speaks fluent Danish.' (Allan et al. (1995:472) Note, however. that the intensifier reading of even in (ii) requires a specific stress pattern. 47 See also Eckardt (200 I). and Martin ( 1975) for other analyses attempting unification of the same uses of selbst in German and me me in French. 48 Danish has two conjunctions meaning 'though, although, even though' which can be used alternatively in most contexts: skent and se/vom. While skont is monomorphemic. selvom is clearly a compound consisting of selv 'even' + the conjunction om 'if, whether'. While these two words can generally be used interchangeably they do differ in one respect: "Both se/vom and skent can be used for factual utterances, but for hypothetical statements, selvom is generally used'' (Allan et al. ( 1995:473)). 96 Note 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 om blokken. Peter even BIKED around about block-the The Danish counterpart of Eng. even when 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.' Another difference between the post-nominal adnominal intensifier selv 'himself and the pre- nominal (or rather 'pre-focal', i.e. preceding its associate) additive focus particle selv 'even' is stress-placement. Being a focus particle, selv 'even' is associated with focus. This focus is realized phonetically as stress on the associate of selv 'even', i.e. on the constituent following the focus particle, see ( 130a). (130) a. b. Selv PETER leste opgaven. 'Even Peter solved problem-the.' Peter SELV leste opgaven. 'Peter himself solved problem-the.' In contrast, in constructions involving the adnominal intensifier selv 'himself it is the element selv which is stressed, see (130b). It could, perhaps, be argued that the difference in stress placement between constructions involving adnominal intensification and constructions involving the focus particle selv 'even' is merely a phonological phenomenon which follows from the fact that in Danish constituent stress fall on a prominent constituent-final sub- constituent. If that is the case, then the difference in stress placement should be considered as a side-effect of the post-nominal vs. pre-nominal placement. From this point of view the 97 crucial difference between the adnominal intensifier use of the element 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' Descriptively the different theories of focus particles in the literature agree that the main semantic contribution of selv used as the scalar, additive focus particle selv 'even' consists in evoking two implicatures. The so-called existential implicature captures the additivity of selv 'even' by requiring the proposition to be true for at least one of the members of the focus- generated contrast set of alternatives to the referent of the associate of selv. The second implicature is responsible for ranking the truth of the proposition lower on a scale of probability than the truth of any of the alternative propositions in the contrast set. In the following we present an analysis if selv 'even' closely inspired by Karttunen and Peters (1979) and Eckardt (200 I). First consider the sentences in (131) and (132). ( 131) a. b. Even the king came to the meeting. Selv kongen kom til m0det. even king-the came to meeting-the 'Even the king came to the meeting.' (132) The king came to the meeting. As far as truth conditions are concerned the sentence in ( 131 a), and its Danish equivalent in (131 b), and the sentence in (132) are equivalent. In other words, the word selv 'even' plays no role in determining the truth conditions and, as a consequence, (131) and (132) express the same proposition. However, this does not mean that the presence of even in ( 131) contributes nothing to the meaning of the sentence. According to Karttunen and Peters (1979), the semantic import of the particle even can best be captured in terms of conventional implicatures. In addition to asserting that the that king came to the meeting the sentence in ( 132) also allow the hearer to conclude that the speaker believes in the truth of ( 133a-b ). 98 (133) a. Other people besides the king came to the meeting. b. Of the people under consideration. the king is the least likely to come to the meeting. The sentences (133a) and (133b) represent what is usually referred to as the existential and scalar implicatures of even, respectively. Following Grice, these implicatures can be defined as conventional (rather than conversational) since they do not arise from the interaction of general conversational principles (e.g. cooperation principle, etc., see Grice ( 1975)), and the specific properties of the sentences themselves and their surrounding contexts. Being conventional, these implicatures 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 witnesses the conventional nature of the existential implicature expressed in ( 133a). Notice, however, that since the propositions in (133) are implicated rather than asserted (as is (132)), their falsity is judged less detrimental to the communication process than the falsity of ( 132). That is, in situations where the proposition in (132) is false uttering ( 131) would be considered as act of lying, or miscommunication. In contrast, in situations where (133a,b) are false, uttering (131) would be judged less harshly. In such cases the speaker would likely be considered to have uttered a sentence which is at least partially true. Rather than blaming the speaker for misrepresenting the truth the hearer might simply correct his misconceptions by uttering either ( 135) or ( 136) 49 . ( 13 5) Well yes, he did come to the meeting; but that is just as one should expect. ( 136) Well yes, he did come to the meeting; but, in fact, he was the only one who showed up. 49 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 just 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 of 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) 99 In contrast, in situations in which (132) is false, it would be impossible to correct the mistake of the speaker of ( 13 I ) by uttering ( 13 7). (137) Yes, you wouldn't expect the king to have come to the meeting: as a matter of fact, he didn't come. The insights of the Karttunen & Peters (1979) summarized above, form the basis for the analysis of selv 'even' proposed here. First consider the following sentences which illustrate the similarities between focus and the additive focus particle even, see (138a,b). (138) a. b. [The king]F came to the meeting. Even [the king]F came to the meeting. Both sentences in (138) contain a focused element, i.e. [The king]F. According to Roath's ( 1992) theory of focus, the meaning of any sentence containing a focused element comprises an ordinary semantic value, i.e. the proposition expressed by the sentence, see (139), and a focus semantic value, i.e. a contrast set, see (140). As shown in (140), the contrast set generated by focus is a set of propositions of the type x came to the meeting in which the focused element has been replaced by entities taken from the set De 50 (139) Ordinary semantic value of 138a,b): a. [[The king]F came to the meetingJr = I iff the king came to the meeting. b. [[Even [the king]F came to the meeting]r = I iff the king came to the meeting. (140) Focus semantic value of(l38a,b): a. [[ [The king]F came to the meeting Jt = { came-to-the-meeting(x) I x E De} b. [[Even [the king]F came to the meeting]] r = { came-to-the-meeting(x) I x E De} For the formalization of even we will adopt a slightly modified version of the analysis proposed in Eckardt (2001) which uses Roath's ordinary and semantic values primitives, see (141) and (142). 50 The set De is usually assumed to be constrained pragmatically to contain only those entities which are 'relevant' in a given context. (141) Analysis ofthe additive focus particle even 51 : a. b. even+ S Assertion: Scalar implicature (i) Existential implicature(ii) ( 142) Scale of expectedness (Eckardt (2000)): [[ s Jt \ip E [[ S ]]f\ {[[ S ]]o}: [[ S ]]o <c p :lpE [[S]{\{[[S]] 0 }:p=l 100 Let <c be an ordering of propositions according to their probability or expectedness in a given context c. ("p <c q" reads like "p is less expected, less probable, more surprising than q") Analyzing ( 138b) according to ( 141 a) we first determine the assertion expressed by ( 160b) to be its ordinary semantic value, see (143). (143) Assertion of(l60b) =Ordinary semantic value of(l38b): [[ [The king]F came to the meeting Jt = I iff the king came to the meeting. Next, given the focus semantic value of(l38b), see (144), and the formalization ofthe scalar implicature in ( 141 b(i)) and (142), we can capture the scalar implicature of ( 138b) as in (145). (144) Focus semantic value of(l38b): [[Even [the king]F came to the meeting]] r = {came-to-the-meeting(x) I XE De} 'the queen came to the meeting', 'the prince came to the meeting', 'the butler came to the meeting' (145) Scalar implicature of(l38b): \ip E { {came-to-the-meeting(x) I xE De}\{came-to-the-meeting(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 of propositions of the type 'x came to the meeting', where x =t the king, the proposition 'the king came to the meeting' is less likely to be the case than p, see also ( 168). (146) "For all alternatives ofthe king, it would have been more probable for them to come to the meeting" ="It would have been less surprising, had the queen come to the meeting" "It would have been less surprising, had the prime minister come to the meeting" "It would have been less surprising, had the arch-bishop to the meeting", ..... . 51 The analysis of even given here is essentially the same as the one proposed in Eckardt ( 200 I). I 01 Finally, the existential implicature of (138b) can be captured usmg the formalization in (141 b(ii)), see ( 14 7). (147) "Existential" implicature of(138b): ::Jp E { { came-to-the-meeting(x) I x E De}\{ came-to-the-meeting( the king)}}: p=l (147) says that among the propositions p belonging to the set of propositions of the type 'x came to the meeting', where x -=1:- 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 meeting. Semantically, the difference thus lies in the two implicatures of even, see (145) and (147), neither of which is shared by the adnominal intensifier. That is, while (138b) presupposes that somebody other than the king also came to the meeting, the sentence in ( 138a) does not evoke any such presupposition, i.e. the king may or may not have been the only individual coming to the meeting. This difference between the adnominal intensifier and the focus particle is illustrated by the clefts (which presuppose uniqueness) in (148). (148) a. *Det var selv Peter der komfor sent. it was even Peter who came for late '*It was even Peter who was late.' b. Det var Peter selv der komfor sent. it was Peter self who came for late 'It was Peter himself who was late.' In section 2.2.3.2.1 the so-called unique identifiability constraint on adnominal intensification (i.e. only definite nominal expressions or nominal expression which refer to specific or uniquely identifiable entities may be adnominally intensified) was mentioned. The sentence in ( 149a) shows how this constraint blocks adnominal intensification of indefinite DPs. In contrast, the sentence in ( 149b) shows that this constraint does not affect the distribution of the scalar additive focus particle selv 'even'. (149) a. *En student selv laste opgaven. a student self solved problem-the '*A student himself solved the problem.' b. Selv en student lraste opgaven. even a student solved problem-the 'Even a student solved the problem.' 102 Summing up, we conclude that while both the scalar additive focus particle selv 'even' and the adnominal intensifier selv 'himself are associated with focus and generate contrast sets of alternatives, they differ in that former carries with it two conventional implicatures, viz. ( 145) and (147), while the latter has none. Exactly what explains the differences between these two homophonic elements is still unknown. It seems likely that part of the answer 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. While it is obviously too early to reach any conclusion we assume that it might be possible to explain the differences between the focus particle selv 'even' and the adnominal intensifier selv 'himself as following from their syntactic differences. That is, we suggest that the different behavior of the intensifier selv and the focusing particle selv may potentially be accounted for by assuming that the same morpheme selv takes on different functions when it occurs in different syntactic positions 52 • Beyond these speculations we have, at this point, no satisfying account of the exact relationship between the focus particle selv 'even' and the adnominal intensifier selv. Since this topic is orthogonal to the main proposal of this dissertation, we leave it for further research. 2.5 Selv used as a noun. Like its Eng! ish counterpart the Danish word selv may be used as a noun by itself, see ( 150a- !SOb). (150) a. selvets begred self-the-POSS concept 'the conception of self (/the ego)' 52 Cf. Siemund (2000:ch. 5) who proposed the same analysis. b. Selvet er en central del af psyken. self-the is a central part of psyche-the 'The self is a central part of the psyche.' c. Kun levende w:esner kan have et selv. only living creatures can have a self 'Only living creatures can have a self.' 103 In (150c) selv occurs as a noun preceded by the indefinite article. The sentences in (150a-b) show that selv behaves just like any other noun by taking the suffixal definite article ~et 53 , see (150a-b), and the s-genitive, see (150a). Not all languages allow adnominal intensifiers to be used as nouns. As observed in Safir (1996), in the Romance languages intensifiers are derived from adjectival bases, e.g. French meme 'same'. In these languages it is often the first person singular personal pronoun (e.g. Fr, moi 'me' /'self) 5 ~ which is used as the noun meaning 'self rather than the adnominal intensifier, see (I51c) 55 • Unlike English, which does not allow this use of the first person singular pronoun (e.g. *one's own mel! vs. one's own self), Danish allows both the pronoun and the intensifier to be used nominally with the same meamng, compare (ISla) and (151 b) 56 • 53 '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. 54 According to the Dictionary Robert. the use of the first person pronoun moi ·me' as noun was first attested in 1583. 55 Note. however, that le me me 'the self is an accepted concept in philosophical jargon ( cf. Ricoeur's Le Me me et l 'Autre). 56 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. Hvilkenforunderlig. forj(J!fningsfuld Tid var det ikke, hvor S(J!/somt ikke med 0ren at here sin Sj(J!/s utydelige, londomsfulde Hvisken klinge frem i Virkelighedens Luft, som vildt udfordrende Lurtoner, som Brag af Kelleslag paa Tempelmure, som Hvin af Davidsstene paa Flugt mod Goliathspander og som sejerssikker Fanfare. Det var som at here sig selv tale ifremmede Tunger, medfremmed Klarhed ogfremmed 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 self.' b. [. . .} der var Ens dybeste, inderste selv. which was one's deepest innermost self '[ ... ]which was one's deepest. innermost self.' 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 (151) 57 a. b. c. One should protect one's own seljl*I. Man b0r vcerne om ens eget jeg/*mig/selv. one ought protect about one's own 1/self 'One should protect one's own self.' Chacun doit proteger sa propre *je/moi/*mel*memelpersonne. everyone ought protect his own me/same/person 'Everyone ought to protect his own self/person.' 104 The fact that intensifiers may be used as nouns has been used as an argument in favor of classifYing them as nominal expressions. However, rather than entering the somewhat futile debate about what word-class it belongs to we simply take the element selv 'self to be a morpheme/root which can be made into a noun, adjective, or adverb in different contexts. As will be discussed in chapter 3, we will argue that the French intensifier -meme, which, unlike selv, clearly has adjectival origins, should also be analyzed as a root which can take on the characteristics of different word-classes depending on its syntactic position. That is, cross- linguistically intensifiers share a number of core-properties despite their different historical origins and the particular range of morphological realizations (adjective, noun, particle, etc.) which they happen to have. 2.6 Adjectival forms of selv: selve and selveste In addition to the uninflected, post-nominal, appositional intensifier selv, Modern Danish also has an inflected, pre-nominal, adjectival form of the intensifier, i.e. selve, which is used "self and egenleget 'own' are suppletive variants of the same word. See also the discussion of the synonymous compounds egenrisiko ·own-risk' and selvrisiko ·self-risk", in chapter 7. 57 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 T or the non-emphatic object pronoun me ·me'. is probably due to phonological factors, i.e. stressability. The French subject pronoun je T cannot be stressed on its own (In the colloquial language the stressed form ofje is moi-je "ME-l':::: "I myself). 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. 105 immediately preceding a noun 58 (in the definite form) to emphasize the latter very strongli 9 , see ( 152)-(154 ). ( 152) Selve huset er ikke me get veer d. (house=core vs. rest of property=periphery) self house-the is not much worth 'The house itself isn't worth much.' (153) Selve biskoppen kommer paa besog. self bishop-the comes on visit 'The bishop himself is paying a visit.' (154) Vi bar i selve Kobenhavn. (Copenhagen=core vs. suburbs=periphery) we live in self Copenhagen 'We live in Copenhagen itself.' When used with nouns denoting places or inanimate objects pre-nominal, adjectival selve always has an core-periphery reading, with the DP modified by selve being the core, e.g. (152) and ( 154 ). As shown by the translations of the examples in ( 152-154 ), the semantic contribution of this adjectival form of selv is thus very close to that of the adnominal intensifier 60 • Post-nominal selv and pre-nominal selve differ in that selve imposes a much stronger specificity constraint on its complement DPs. Compare the grammatical sentences in ( 156) with the ungrammatical sentence in (155). (155) *Selve en sjcellcender har fortalt migdet. self a Zealander has told me it 'A Zealander himselftold me.' 58 The element selv can be combined with the adj. samme '"same'' in the somewhat idiomatic expressions selvsamme adj. ·self-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 ofselv ·self in nominalizations. 59 The examples in (152-4) are adapted from similar examples in Allan, Holmes and Lundskaer-Nielsen ( 1995: 170ft). 60 Indeed, in Icelandic adnominal intensifiers are pre-nominal adjectival forms which may adjoin to full lexical DPs as well as pronouns and reflexives: (Ice!.) sjalva sig::: (Dan.) sig selv. (Ice!.) sjalva Peter= (Dan.) Peter selv. (156) En sja:llamder selv har fortalt mig det. a Zealander self has told me it 'A Zealander himself told me.' 106 While adnominal selv can be adjoined to indefinite DPs which have specific reference, adjectival selve cannot under any circumstances take an indefinite DPs as its complement. Furthermore, unlike adnominal selv, adjectival selve cannot modify pronouns, e.g. *selve jeg 'self me', *selve ham 'self ham' vs.jeg selv 'I myself, ham selv 'him him self, etc. Finally, pre-nominal, adjectival selve is restricted to a small class of nouns denoting individuals of high social status, compare (157a) and (157b). (157) a. b. ??*Selve Peter kommer pa besog. self Peter comes on visit 'Peter himself is paying a visit.' Selve biskoppen!Dronningen/pra:sidenten kommer pa besog. selfbishop-the/Queen-the/president-the comes on visit 'The bishop himself is paying a visit.' Being an adjective selve may take the so-called morphological superlative by adding the superlative suffix -ste, i.e. selveste see (158-159) 61 • This superlative form implies an even higher degree of intensification and is used exclusively with nouns referring to individuals of high status, e.g. the Queen, the King, the prime minister, etc. That is, unlike the post-nominal invariable adnominal intensifier selv and the basic form of the pre-nominal adjectival intensifier selve which can take both animate and inanimate (e.g. huset selv 'the house itself ;::; selve huset 'the house itself in ( 152)) associates, the superlative form selveste is compatible only with [+human] nominal associates referring to high status individuals 62 • 61 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 pam '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' ( 158) Vi blev inviteret til middag has selveste direktoren. we were invited to dinner at self-est boss-the 'We were invited to dinner with the managing director himself.' ( 159) Selveste dronningen tog imod os. self-est Queen-the took against us 'The Queen herself welcomed us.' 107 Unlike selv, selve and the superlative selveste occur in pre-nominal position, i.e. the normal position for a Danish adjective. As for the syntactic analysis of selve and selveste, two solutions come 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 complements (Sanchez (1995)). Here the latter solution will be adopted for selve/selveste. In contrast, as already discussed, the post-nominal 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. German selbst, Chinese ziji, etc.). While in the Slavic and Romance languages the adnominal intensifiers tend to display adjectival behavior, intensifiers in the Germanic languages can behave either like adjectives (e.g. Swedish sjdlva) or uninflected appositional intensifiers (e.g. Ger. selbst). From a typological perspective Danish is thus interesting in that it has both adjectival and appositional forms ofthe adnominal intensifier. The fact that adjectival forms of selv, i.e. selve and selveste, can be shown to be subject to the principle of unique identifiability (and thus appears to be subject to the principle of 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 morphological realizations of the intensifier selv. Cross-linguistic evidence from Swedish and Icelandic, which both have adjectival intensifiers, confirms this analysis. Although an 62 Note that Spanish also have what looks like a superlative form of the adjectival intensifier mismo 'himself, viz. mismisimo. which appear to function exactly like selveste in Danish: (i) Elisa hab/6 cone/ mismisimo Papa. (=Sanchez. L. ( 1994:480(6))) Elisa talk with the same-SUPERLATIVE Pope 108 interesting question in itself, the different behavior of the adjectival uses of selv, i.e. selve and selveste, does bear directly on the issues discussed in this dissertation and will not be treated in any more detail here. 2. 7 Selv 'selr in nominalizations As its counterparts in other Germanic languages the element selv may occur as part of compounds. The majority of such compounds are deverbal nominalizations in which the element selv appear to function as the direct object, see (I 60a-c ). (160) Deverbal compound nouns containing selv. Compound: Corresponding 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 of reflexive pronouns. However, in chapter 7 we will argue that it is both possible and theoretically more advantageous to consistently analyze such word-internal uses of selv as adnominal intensifiers intensifying phonetically unrealized reflexives, see (161). (16 I) a. [0 selv] [bedrag]] REFL self deceit 'self-deceit' b. *[ [0 ] [ bedrag]] REFL deceit (162) a. Peter bedrager sig selv. Peter deceives REFL self 'Peter deceives himself b. *Peter bedrager sig. Peter deceives REFL Since the verb bedrage 'deceive' is anti-reflexive, it requires the intensified form of the reflexive, see (162a) vs. (162b). In order to have a reflexive reading, nominalizations based on this verb thus have to have intensified 0-reflexives, see ( 161 a) vs. (161 b). "Elisa talked to the Pope himself.' 109 While the in examples in ( 160a-c) above selv may be argued to function as a kind of word-internal version of the reflexive pronoun sig 'REFL', such an analysis is clearly not possible in (163a-c) where selv/se(f- seem to function as interpreted as an exclusive adverbial intensifier meaning 'by oneself. (163) ComQound: CorresQonding verb: a. selvbygger 'person who builds bygge 'to build' (Danish) his house with his own hands.' b. self-storage 'place where you to store store stuff by yours- self/on your own' c. self-cleanecl' 3 'a (house) cleaned to clean by oneself d. self-cleaning 'an (oven) which to clean cleans itself Based on the above examples we conclude that word-internal intensifiers can be either: (i) adnominal intensifiers (160a-c),(161a), (ii) exclusive adverbial intensifiers (163a-c). So far we have not come across any evidence suggesting that word-internal selv may function as either additive scalar focus particle, or any of the other uses of selv discussed above. See chapter 7 and appendix I for a more detailed analysis of se(f-nominalizations based on the account of intensifiers proposed here. 2.8 Idiomatic uses of the element selv Finally, in the interest of completeness, it should be mentioned that the element selv also occurs in a number of idiomatic expressions, see ( 164-165). (164) A: Takfor i aftes. (Allan et al. (1995:170) thank for in evening 'Thank you for (the party, etc.) last night. 63 The word selfcleaned is used (and coined?) by Barbara Ehrenreich in her (200 I) 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''. B: Selv tak.' self thank (165) a. b. Thank you! (with stress on 'you'.) Det siger sig selv. it says REFL self 'That is obvious.' D0ren abnede sig af sig selv. 64 door-the opened REFL of REFL self 'The door opened all by itself.' 110 While some of these may potentially be accounted for in terms of the uses of selv described in sections 2.2-8 above, e.g. the use of selv in (165a,b), which seems to be somehow related to the exclusive adverbial intensifier, we will leave the exact analysis of the remaining idiomatic cases for further studl 5 • 2.9 Cross-linguistic perspective on the range of uses of intensifiers. The table in ( 166) summarizes all the uses of the element selv in Danish described in sections 2.2-8 66 • It also includes a mini typological survey showing the ranges of uses corresponding intensifiers have in English 67 , French and Chinese 68 • 64 Note that this use of the PP af 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]]]. 65 The list examples of idiomatic uses of selv in Danish given here is far from exhaustive. Due to space limitations only a few example sentences are provided here. 66 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. 67 Examples illustrating the different uses of the adnominal intensitlers of the x-seifform in English can be found in chapter 5. 68 Examples illustrating the different uses of the adnominal intensifier § l3 ziji ·self-self in Chinese can be found in chapter 6. I I I (166) Typological survey of different uses of adnominal 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 zijil 0 ziji c. Intensified 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 seljl*l *memelmoi h. Positive adjective selve very/actual meme 'same (kind of)' i. Superlative adjective selveste j. Self-nominalizations selv *himseljlse(f *meme-/auto-/sui-letc.. *=ijil=i-1-ji The typological survey or intensifiers in (166) shows that the element selv in Danish has a wider range of uses than its counterparts in most other languages. Though such cross- linguistic data should be used with great caution, it still provides a rough indication of how far attempts of unifying all the uses of intensifiers like selv may be pushed. Although the element 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 of the range of core uses of intensifiers cross-linguistically. 2.10 Conclusion The contribution of this chapter can be divided into three parts. First, in section 2.2 we have shown that the semantic contribution of adnominal intensification is similar to that of focus (i.e. both involve the generation of a contrast set of alternatives) and is subject to the contrastiveness condition as formulated in (62), repeated here as ( 167). ( 167) Contrastiveness condition on adnominal intensification: A nominal expression (DP, pronoun, reflexive) can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. Second, in sections 2.2.5-7 we have shown that all the different uses of selv listed in (168a,i-v) below can be explained as subtypes of adnominal intensification falling under princi pie ( 167). 112 ( 168) Different uses of selv: ExamQles: Section: a. adnominal intensification i. intensified DPs (I), ( 14) (sect. 2.2) ii. intensified reflexives (4) (chap. 3, section 3.3) (1) selv in nominalizations (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 nominalizations (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 forms ofthe intensifier (10), (152-4) (sect. 2.6) g. idiomatic uses of selv (164-5) (section 2.8) In spite of superficial differences we have shown that it is possible to unify the analysis of selv occurring in constructions involving intensified DPs (189a,i), q-floated intensifiers (168a,iv), and fronted intensifiers (168a,v). We furthermore aim to unify the analysis of adnominal intensification and complex reflexives (168a,ii), see chapter 3, section 3.3, and complex pronouns (168a,iii), see chapter 3, section 3.4, and to show that this unification can only be achieved if binding and intensification are assumed to belong to different modules of the grammar. Within this approach, binding theory deals with the interaction between intrinsic properties of nominal expressions and purely syntactic locality principles, e.g. principle A and 8, while the theory of intensification consists of pragmatic and semantic constraints on the distribution of intensifiers. The argument is that this modular approach combined with the focus-based theory of intensification proposed here stand a better chance at unifying the different uses of adnominal intensifiers listed in (168ai-v) than the predicate-based approaches to binding which assume selv to be a reflexivizing particle (cf. Reinhart & Reuland). Third, in sections 2.3-2.8 we have illustrated the fact that the element selv occurs in a large number of different contexts with what sometimes seems to be very different syntax and semantics, see ( 168b-g). In spite of these differences, we have argued that it is possible to unify the analysis of all the uses of the intensifier selv. As shown in ( 168), there are only five 113 basic uses of the word 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 more detail in chapter 7, the instances of selv found in nominalizations can be shown to be reducible to either intensified zero reflexives (168a,ii,(1)) or word internal secondary predication (168b,(1)). As for the idiomatic uses of selv in ( 168g) discussed in section 2.8, they are presumably ultimately be reducible to one or the other of the three main uses of selv. Of the five basic uses (168a-e), the nominal and adjectival uses (168d-e) can be accounted for as deriving from the intensifier selv by regular processes of word formation. At first glance, the use of selv as an exclusive manner adverbial/secondary predication (168b) seems to be very different from the adnominal intensifier. However, it may be possible to analyze it as an intensifier too. Rather than being adjoined to a DP as the adnominal intensifier, it is a subject/agent oriented adverbial adjoined to the VP, i.e. it takes agentive subjects as its associates and contrasts them with other potential agents. Finally, as discussed in section 2.4 there is ample historic evidence that the use of selv as a focus particle (168c) evolved from the adnominal intensifier. We believe that it may be possible to account for the differences between them by assuming that they are basically the same morpheme selv 'self which takes on different semantic properties when occurring in different syntactic positions (cf. Siemund 2000, chap. 5). In other words, although we do not develop it in detail here we do believe it is ultimately possible to arrive at a unified account of all uses of the intensifier selv 'self in ( 168). It is, however, the use of selv 'self as an adnominal intensifier (168a,i-v) which is the main focus ofthis dissertation. In this chapter we have shown that (168i,iv,v) all fall under the same principle (167). Chapter 3 will be devoted to showing that complex reflexives (e.g. Dan. 114 sig selv 'REFL self) and pronouns (e.g. Dan. han selv 'he himself) are analyzable as intensified nominal expressions which also fall under the contrastiveness principle ( 167). Chapter 3 Binding: Reflexives and Pronouns in Modern Danish 3.1 Introduction. Binding of reflexives and pronouns in Danish. 115 The purpose of this chapter is two-fold. First, the distribution of simple reflexives and pronouns, e.g. Dan. sig 'REFL' and ham 'him', will be described and an account of these elements will be proposed. Second, it will be argued that complex reflexives, e.g. Dan. sig selv 'REFL self, and complex pronouns, e.g. ham selv 'him self/him himself, are best accounted for by assuming that binding and intensification are two independent modules of the grammar. Binding is taken to be a system of syntactic principles which determines the distribution of reflexives and pronouns. Rather 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 of lexical features of nominal expressions and syntactic domain constraints. In contrast, as discussed in chapter 2, the module of intensification consists of semantic and pragmatic constraints on the distribution of the adnominal intensifier selv 'self. That is, it will be argued that the behavior of complex reflexives and complex pronouns follow from the fact that they are adnominally intensified forms of their simple counterparts, e.g. sig 'REFL' and ham 'him', and that their specific properties can be derived compositionally from their constituent components, i.e. siglham 'REFL/him' and the adnominal intensifier selv 'self 1 • The remainder of this section contains an introduction to binding of reflexives and pronouns in Danish. 1 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 Konig and Siemund ( 1999). (I). (I) 116 Danish distinguishes between simple and complex reflexives, i.e. sig vs. sig selv, see a. b. c. Peter vasker siglsig selv. Peter washes REFLIREFL self 'Peter washes (himself).' Peter hader *sig/sig selv. Peter hates REFLIREFL self 'Peter hates himself.' Peter hviler sigl*sig selv. Peter rests REFLIREFL self 'Peter is resting.' (neutral) (anti-reflexive) (inherently reflexive) Predicates differ with respect to whether they allow both simple and complex reflexives (Ia) or whether they allow either only complex (I b) or only simple reflexives (I c). The complex form sig selv 'REFL self is used with anti-reflexive predicates, i.e. predicates whose meaning imply non-reflexive scenarios, see (I b). The simple reflexive sig 'REFL' is the default form used both with predicates whose meaning does not imply anti-reflexivity, e.g. the so-called "neutral" predicate vaske 'wash' in (I a), and with inherently reflexive predicates whose meaning does imply reflexivity, see (I c). The simple/complex distinction also interacts with the interpretation of the referent of the reflexive (e.g. in so-called doppelganger-effect examples as well as with respect to the choice between sloppy and strict readings in VP ellipsis constru~tions) and prosodic factors, both of which will be discussed in greater length below. In Danish the binding domain for reflexives is the tensed clause, indicated with square brackets in the following examples. Local reflexives of the types illustrated in (I) will be discussed in sections 3.2 and 3.3. The so-called long-distance reflexives, which in Danish are limited to the tensed clause, see (2a) vs. (2b), will be described in section 3.3.1 where we also discuss how intensifier-adjunction interacts with locality constraints. (2) a. [Peter, bad mig barbere sigJ. Peter ask-PAST me shave-INF REFL 'Peter asked me to shave him.' b. *[Peter, sagde [at jeg barberede sigJ]. Peter; say-PAST that I shave-PAST REFL; 117 Unlike English where certain logophoric or contrastive uses of what appear to be the reflexive himself can be bound by antecedents outside the tensed clause, see (3ai, Danish reflexives have to be bound within this domain, (2a) vs. (2b). That is, in Danish, only pronominals, be they simple, e.g. ham 'him', or complex, ham selv 'him himself, may have antecedents outside the tensed clauses, see (3b) vs. (2b ). (3) a. b. [Peter, said [that Mary danced with all others than himself]]. [Peter, sagde [at Marie dansede med alle andre end ham; selv]]. Peter say-PAST that Marie dance-PAST with all others than him self 'Peter said that Mary danced with all others than himself.' Unlike himself in English, Danish reflexives are subject-oriented, i.e. only allow binding by antecedents functioning as subjects, see (4) vs. (5). ( 4) Ida, told Maryz about herself/Z. a. IdaJortalte Mariez om sig, *z selv. Ida told Marie about REFL self 'Ida told Marie about himself.' b. IdaJortalte Mariez om hende., z selv. Ida told Marie about her self 'Ida told Marie about herself (<her herself).' It will be argued that subject-orientedness is a characteristic of 'true' reflexives 4 and that all non-subject-oriented reflexives are really intensifiers or some kind of intensified nominal 2 See chapter 5 for an analysis of locally free himself as adnominally intensified pronominals. 3 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 (i) below is not due to any anti locality of the simple reflexive sig (a~ is sometimes suggested, e.g. Vikner ( 1985)) but simply to phonological factors. see section 3.3.6 for more discussion of the interaction of stressability and adnominal intensification of reflexives. (i) * Peter,forta/te Marie= om sig,. Peter told Marie about REFL self ·Peter told Marie about himself.· 118 expressions. In fact, as will be shown in chapter 5, it is possible to claim that herself in the sentence in ( 4) is not a reflexive but rather a concealed intensified pronoun derived from the full form her herself by some sort of deletion rule. Danish distinguishes between reflexives and pronouns even for possessives, e.g. sin 'POSSREFL' (6a) and hans 'his' (6b), whereas English only has one possessive, namely the possessive pronoun his. (6) a. John, sagde at Peterk bad Hans"' vaske sin., k"' *= bil. John said that Peter asked Hans wash POSSREFL car 'John said that Peter asked Hans to wash his car.' b. John, sagde at Peterk bad Hans 111 vaske hans, k *mz bil. John said that Peter asked Hans wash his car 'John said that Peter asked Hans to wash his car.' One of the greatest advantages of assuming binding and intensification to constitute independent modules of the grammar is that it becomes possible to unify the account of argument and possessive reflexives and pronominals in Danish. By considering intensifier- adjunction to reflexives and pronouns to be controlled by binding independent principles (i.e. the semantic/pragmatic principles controlling adnominal intensification) it becomes possible to defend a nominal approach to binding in which the distribution of reflexives and pronouns follow syntactic principles: the reflexives sig 'REFL' and sin 'POSSREFL' having to be bound by a subject (local or long-distance) inside the tensed clause (=principle A), and the pronominals ham 'him' and hans 'his' having to be free from binding from a subject inside the same domain. The complex forms sig selv 'REFL self/sin egen 'POSSREFL own' and ham selv 'him self /hans egen 'his own' are adnominally intensified versions of their simple counterparts. All instances of the adnominal intensifier selvlegen 'self/own' - even the ones occurnng m the so-called complex reflexives and pronouns - are thus controlled by the binding-independent principles of the module of adnominal intensification. As already shown 4 Cf. Jakubowicz 1994:206. (I). 119 in chapter 2, these principles are susceptible to semantic and pragmatic factors as well as to the focus-structure of the surrounding linguistic context, i.e. the sentence or the larger discourse. The rest of the chapter is organized as follows. In section 3.2 the foundation for an analysis of reflexives in Danish is outlined. Then, in section 3.3 this analysis will be used to account for the behavior of reflexives in a number of different contexts. Section 3.4 is dedicated to the treatment of complex and simple pronouns. Finally section 3.5 summarizes the results ofthe previous sections and concludes the chapter. 3.2 Basic properties of Danish reflexives Before we start discussing how predicate meaning and focus affect the distribution of the adnominal intensifier in complex reflexives and the independence of binding and intensification in general, we first need to describe the formal characteristics of reflexives and lay out the relevant assumptions concerning their syntactic and semantic properties. Section 3.2.1 contains a description of the morphology of reflexives in Danish. Section 3.2.2 and 3.2.3 deals with their semantic and syntactic properties respectively and in section 3.2.4 an analysis of reflexives in Danish based on a modified version ofprinciple A is proposed. 3.2.1 Morphological characteristics of reflexives in Danish As illustrated in Tables 1-2 Danish has two series of reflexives and pronouns: simple forms (see Table 1) and complex forms (see Table 2). We assume the complex reflexives to be formed by adnominal intensification of the simple reflexives. That is, intensified reflexives are formed in exactly the same way as both intensified DPs (e.g. Kongen selv kom til modet 'The king himself attended the meeting') and complex subject and object pronouns pronouns (e.g. Peter sagde at Marie dansede med all andre end ham selv 'Peter said that Mary danced 120 with everyone but himself), namely by intensifier-adjunction to the simple/unintensified forms. In tables 1-2 column I gives the meaning of the forms in columns 2-4. Columns 2-3 illustrate the simple and complex forms of subject and object pronouns and column 4 gives the corresponding simple and complex reflexive pronouns. Note that only the third person forms maintain a formal distinction between object pronouns and reflexives, i.e. sig 'REFL' vs. ham 'him'. For all first and third person forms there is no formal distinction between reflexives and object pronouns, e.g. mig 'myself and mig 'me'. Furthermore while the third person object pronouns, e.g. ham/hende 'him/her', etc., are marked for both case, number, and gender, the third person reflexive is invariably sig which is unmarked for all these features. In these respects the Danish pronominal system is very similar to the French and German systems. Table I. Unintensified pronouns and reflexives. Nominative Acc./Dat. Reflexive 1st' sing. Jeg Mig mig 2"ct, sing Du Dig dig -informal 2"d . , smg. De Dem Dem -formal 3rct, sing. Han Ham sig -masculine 3rct, sing. Hun hende sig -feminine 3rd, sing. Den Den sig - common gender 3rd, sing. Det Det sig -neuter Is\ plur. Vi Os OS 2"d, plur. I Jer jer -informal 2"ct, plur. De Dem Dem -formal 3rct, plur. De dem sig 121 T bl 2 I a e 'fi d ntenst te pronouns an d fl re extves. Nominative Acc./Dat. Reflexive 1st, sing. jeg selv Migselv mig selv 2"d, sing du selv Dig selv dig selv -informal 2"ct, sing. De selv Dem selv Dem selv -formal 3rct, sing. han selv ham selv sig selv -masculine 3rd, sing. hun selv hende selv sig selv -feminine 3rct, sing. den selv den selv sig selv - common gender 3rct, sing. ?det selv ?det selv sig selv -neuter I st, plur. vi selv os selv os selv 2"ct, plur. I selv Jer selv jer selv -informal 2"ct, plur. De selv Dem selv Dem selv -formal 3rct, plur. de selv dem selv sig selv Complex and simple reflexives differ in several ways, i.e. phonologically, syntactically, semantically and pragmatically. The proposal defended here is that these differences can be explained as consequences of adnominal intensification. But before we go into the detailed analysis we first need to spell-out the set of assumptions on which our analysis of binding is based. 3.2.2 Semantics of reflexives Unlike pronouns, which are specified for person, number, and gender, Danish reflexives are only overtly specified for person 5 • Semantically the reflexive sig can be thus characterized as 5 Since Danish. unlike Chinese and other languages, does not allow subject reflexives, the only potential case marking ofreflexives would be an ACCUSATIVE vs. DATIVE/OBLIQUE distinction. However. as the examples in (i-ii) show. no such distinction is overtly marked on reflexives or pronouns in Danish: 122 an expression that has an incomplete set of phi-features. That is, we assume sig to function semantically 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 marked for features like person, number and gender and that they can be used deictically, i.e. have lexical content and reference on their own. The assumption that the simple reflexive sig 'REFL' is a featureless variable is supported by the fact that it is the only element that can be bound by an impersonal DP. In certain sentences, a pronoun can be used in the same configurations in which sig can be long- distance bound, see (7). Yet only the reflexive sig 'REFL', but not the pronoun ham 'him', can occur ifthe matrix antecedent is an impersonal DP, such as arbitrary PRO or an indefinite expressions, e.g. man 'one' or enhver 'everybody', see (8) 6 7 . (7) a. b. (i) (ii) Peter, Iader folk tale om ham; /sig,. Peter let-PRES people talk-INF about him/REFL 'Peter lets people talk about him.' Peter, bad mig om at invitere sig/ham, z· Peter ask-PAST me about to invite-INF REFL!him 'Peter asked me to invite him.' a. b. a. Peter vasker sig. Peter washes REFL-ACC 'Peter washes (himself).' Peter kobte sig et hus. Peter bought REFL-OAT a house-ACC 'Peter bough himself a house.' Peter vasker ham. Peter washes him-ACC ·Peter washes himself.' c. Peter kgbte him et hus. Peter bought him-DATa 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. 6 This observation, as well as the examples in (7-8~b ). is adapted from Jakubowicz ( 1994: 133). 7 A similar situation holds in French. In certain contexts, e.g. non-contrastive PPs. the distribution of the reflexive soi 'REFL' and the pronominal/ui/elle 'him/her' overlap, see (i). However, when the antecedent is indefinite then only soi 'RELF' can be used, see (ii). (i) Pierre, estfier de soi/lui,. Peter is proud of REFL/him ·Peter is proud of himself." (ii) Personne n 'estfier de soi,l*/ui;. (8) a. b. PRO; at lade folk tale om sig/ham*;z er kedeligt. PRO to let people talk about REFL!him be-PRES boring 'To let people talk about one/him, is boring.' Man,!Enhver, bad mig om at invitere sig/ham ., z· one/everybody ask-PAST me about to invite-INF REFL!him 'One/Everybody asked me to invite him.' 123 The examples in (9) illustrate that intensifier-adjunction does not change the ability of the sig to be bound by an impersonal DP. This should come as no surprise since the intensifier selv 'self is also a featureless, morphologically invariable particle. (9) a. b. Enhver/man, skalforsvare sig, (selv) mod politiet. everyone/one must defend REFL against police-the 'Everyone/one must defend himself/oneself against the police.' PRO, At prcesentere sig, (selv) for studenterne er kedeligt. PRO to introduce REFL for students-the is boring 'To introduce oneself to the students is boring.' According to proposals by Burzio (1989, 1991), only so-called 'true reflexives' can have impersonal antecedents. In this respect both sig and sig selv qualify as true reflexives 8 • While sig lacks overt morphological marking of number, gender and case, it can still be said to be marked for person since it differs morphologically from 1st and 2nd person forms, see Table 1 in section 3.2.1. Burzio (1991 ), however, argues that the person agreement of sig with a definite DP acting as its antecedent is merely a case of "pseudo-agreement" 9 and that, consequently, true reflexives lack even the person feature. Indeed, as observed by E. Benveniste, in many languages the third person, is the default category, best characterized negatively as the absence of first and second person features, rather than as the presence of a nobody NEG is proud ofREFL/him 'Nobody is proud of himself.' 8 Cf. also Jakubowicz ( 1994: ll6(1a,b)). 9 Cf. Burzio ( 199Lp. 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 of any number and any gender, but not with one which is I" or znd person'" (ibid.). 124 special third person feature. True reflexives can thus be defined as genderless, numberless, and personless, i.e. as lacking phi-features. We adopt Burzio's proposal and assume that both simple sig and complex sig selv classifY as true reflexives 10 in this sense. In the rest of 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 mentioned above, being itself an uninflected particle, the intensifier selv does not add any phi-features to the complex reflexive sig selv. While sig and sig selv are semantically the same in so far as they are both featureless reflexives, they clearly differ semantically in other respects, i.e. with respect to contrastive focus, doppelganger-effects, sloppy vs. strict readings in VP ellipsis constructions, etc. We argue that all these differences between simple sig and complex sig selv follow from semantic properties of the intensifier selv 'self. That is, simple sig and complex sig selv have the same properties with respect to the binding theory and whatever properties sig selv have in addition to simple sig derive from the adjunction of the intensifier selv 'self. As will be shown in the following section, even the different morphological and syntactic behavior of sig and sig selv follow from the intensifier-adjunction of selv to sig which transforms the phonological clitic into a full DP. 3.2.3 Syntactic structure of reflexives Simple and complex reflexives display quite different syntactic behaviors. While the simple reflexives, e.g. sig, display clitic-like behavior, the complex forms of the reflexives, e.g. sig selv, quality as full DP phrases. Kayne (1975) found that the French reflexive clitic se behaves differently from regular DPs with respect to a number of different syntactic processes. Applied to complex and simple reflexives in Danish, these tests show that sig behaves 10 In this respect Danish siglsig selv behaves like French se/soilsoi-meme and Dutch zichlzich zelf 125 similarly to French se in a number of respects while the syntactic behavior of the complex reflexive sig selv patterns with that of full lexical DPs. That is, the complex reflexive sig selv can be stressed, used as answer to a question, coordinated, clefted and topicalized. The simple reflexive sig, on the other hand, cannot undergo any of these processes, see (I Oa-e ). ( 1 0) selsig DP!sig selv!ham/himse(fllui-meme a. stress * OK b. answer to questions * OK c. coordination * OK d. clefting * OK e. topicalization * OK The examples illustrating the properties listed in (lOa-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 examples, while se has to attach to its host verb, sig does not. The example in ( 11) illustrates the fact that French se cannot be separated from the verb. ( 11) Pourquoi se rase-t-il? [CP pourquoi [C' se, rase, [TP il, t 1 •••• ]]] why REFL shave-PRES he 'Why does he shave (himself)?' In the question in ( 11) the clitic se attaches to the verb and moves along with it when it is moved to C. While French se necessarily attaches to its host predicate, this is not the case for Danish sig, which can be separated from its verb by other material. Being a verb-second (V2) language, Danish requires the verb to raise to C in main clauses 11 • The examples in ( 12a,b) show that in such cases the clitic sig remains behind thereby becoming separated from the verb by the subject. 11 The literature on Mainland Scandinavian 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. (12) a. b. Hvorfor barberer han sig? [CP hvorfor [C' barberer, [TP han, t 1 sig, ]]]? why shaves he REFL 'Why does he shave (himself)?' Han barberer sig. [CP han, [C' barberer 1 [TP t, t 1 sig, ]]] he shaves RELF 'He shaves (himself).' 126 The examples in (13) and (14) 12 illustrate another difference between se and sig. The clitic se cannot be omitted in the second of two coordinated verbs with or without auxiliary verb, see (13a,b) 13 . Only when the auxiliary verb is dropped too is this possible, see (13c). Danish sig shows no such constraints and can be freely omitted from all but the last of the coordinated verbs, see (14a,b) 14 • ( 13) a. b. c. (14) a. b. Avant de sortir, Maries 'habille et *(se) peigne soigneusement. before to go out, Marie REFL dresses and (REFL) comb meticulously 'Before going out, Marie dresses and combs herself meticulously.' Maries 'est habillee et *(s ')est peignee soigneusement. Marie REFL is dressed and (REFL) is combed meticulously 'Marie dressed and combed herself meticulously.' Maries 'est habillee et (s 'est) peignee soigneusement. Marie REFL is dressed (REFL is) combed meticulously 'Marie dressed and combed herself meticulously.' Marie klceder (sig) og reder sig omhyggeligt. Marie dresses (REFL) and combs REFL carefully 'Marie dresses and combs herself carefully.' Peter barberede (sig), vaskede (sig) og tgrrede sig. Peter shaved washed and dried REFL 'Peter shaved, washed and dried (himself).' We take the differences between se and sig illustrated in ( 11-14) to mean that while the French reflexive clitic se forms a morphological constituent with its host sig does not. Sig is, however, a phonologically dependent form that is prosodically too light to stand on its own 12 The examples are based on similar examples in Jakubowicz ( 1994:209. ex. (12-3)). 13 The observation that se cannot be dropped in such coordinated constructions is from Kayne (1975). 14 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. 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 Danish, both simple pronouns, e.g. ham 'him', and simple reflexive, e.g. sig 'REFL', differ from full lexical DPs in that they can undergo object shift to a position outside the VP and attach to an appropriate phonologically heavy host, see (15a) 15 , where sig and ham precede the negation ikke 'not'. But unlike sig, the simple pronoun ham 'him' is stressable and can therefore be left in situ if stressed, see (15b), which shows that the stressable simple pronoun ham (and the stressed intensified reflexive sig selv) but not the unstressable sig can follow the negation ikke 'not'. (15) a. Peter, vaskede sig/ *sig, selv I *HAMzl hamzl *bilen ikke.(+object shift) Peter washed REFLI*REFL self/him/*car-the not Peter did not wash (himself)/him/the car.' b. Peter, vaskede ikke *sig,/ sig, selv I HAMzl *ham= I bilen.( -object shift) Peter washed not *REFLIREFL self/him/car-the Peter did not wash himself/him/the car.' Summarizing the above, we follow Halpern (1992) and Jakubowicz's (1994) in concluding "that Danish clitics are syntactically independent prosodically bound words, whereas in modem French, se as well as le are clitics that select for morphological attachment. Thus sig and ham form only a prosodic constituent with their host and not a morphological one"(Jakubowicz 1994, p. 118-9). In brief, a crucial difference between ham 'him' and sig 'REFL' illustrated in (15) boils down to stressability: while sig can never receive stress, see the tests in (10), ham 'him' can, see (15b). When unstressed, both sig and ham are syntactically independent but prosodically bound words. Since the negation ikke 'not' cannot felicitously receive stress it is not a viable host for prosodically dependent clitics. In contrast, the proper name Ida in (!Sa) can be stressed and is thus able to host clitics. The simple 15 Examples as well as discussion thereof are from Jakubowicz ( 1994:218. i-ii). 128 pronoun ham can still be made acceptable in (15b) by attracting sentence stress onto itself, although thereby changing the meaning contribution of the sentence by adding the corresponding focus-generated implicatures (contrast, contextually defined scales of prominence/remarkability, etc.). Since the simple reflexive sig on its own cannot be stressed under any circumstances the only way to save the sentence is to enable sig to receive sentence stress by adnominally intensifying it, see (15b ). Based on the differences between sig and sig selv described above it is clear that these expressions must have different syntactic representations. Adopting Longobardi's analysis of nominal expressions (which goes back to Abney (1987)) we assume that all nominal arguments are projections of 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 of the syntax of nominal expressions in argument positions, see (16). ( 16) a. b. c. [or [or [or kong] [o [o [o -en] ham] sig] 'the king' 'him' 'REFL' As shown in (16) all nominal arguments are assumed to be projections of a head determiner. Pronouns and reflexives are assumed to behave as determiners, ·i.e. they are assumed to head their own DPs, see (16b) and (16c). Besides making it possible to arrive at a unified analysis ofthe syntax ofnominal expressions in this approach has the additional advantage of enabling us to unify the account of adnominal intensification. As discussed in chapter 2 intensified nominals such as kongen selv 'the king himself are best analyzed as simple adjunction structures, see (17a). (17) a. b. c. [or [or [N [or [or [or [or kong ] [ 0 -en ]] [ 0 ham]] [o sig ]] [selv]] [selv]] [ selv ]] 'the king himself 'him self 'REFL self 129 As shown in (17b) and (17c) the complex pronoun ham selv 'him self' and the complex reflexive sig selv 'REFL self' can be analyzed as intensified DPs having the same syntax as intensified R-expressions. In other words, syntactically all types of nominal expressions behave alike with respect to intensification. The fact that both simple pronouns and simple reflexives can undergo object shift follow from their ambiguous X 0 /XP status. In contrast, full lexical DPs as well as intensified reflexives and pronouns, neither of which can undergo object-shift, are unambiguously 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 16 • 3.2.4 Consequences of relegating binding and intensification to different modules: a syntactic approach to binding of reflexives: principle A The central claim defended here is that simple sig and complex sig selv have the same properties with respect to the binding theory and that whatever properties sig selv has in addition to simple sig derive from the intensifier selv 'self'. One often mentioned difference between simple and complex reflexives, which appear to contradict this claim, is their seemingly different behavior with 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 complex reflexive sig selv cannot. ( 18) Peter, bad Jgrgenz barbere sig, zl sig., z selv. Peter ask-PAST Jorgen shave-INF REFL 'Peter asked Jorgen to shave him/himself.' Since Faltz ( 1975) there has been a tendency to link the morphological complexity of reflexive elements to locality. For Danish and similar languages (e.g. the other mainland Scandinavian languages, Dutch, etc.) this lead to the proposal that the simple reflexive sig is anti-local (i.e. 130 specialized for LD-binding) while the complex reflexive sig selv has to be locally bound ( cf. Vikner 1985, among others). As will be shown below, both of these assumptions are wrong. The simple reflexive sig can indeed be a locally-bound, theta-role receiving argument anaphor (as shown by the example in (18)) and, given the right circumstances, complex reflexives may be LD-bound. Notice that this does not amount 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 within binding theory, we divide the explanatory burden between two modules, 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 adnominal intensification, i.e. focus, prosodic differences, doppelganger-effects, etc. For expository reasons, we will limit the following description of binding properties of Danish reflexives to the behavior of simple sig, - the assumption being that the any differences between sig and sig selv follow either directly from the binding-independent principles of the module of intensification or are phonological, syntactic or semantic side-effects of intensifier-adjunction. Before jumping ahead to the analysis of complex reflexives in Danish, let us first consider the relevant descriptive generalizations concerning the binding of the simple reflexive sig. Consider the sentences in (19) which illustrate the locality constraints restricting the distribution of sig. ( 19) a. [Peter, barberede sig, •ol Peter shave-PAST REFL 'Peter shaved (himself).' 16 See also section 3.3.6 where stressability will be discussed in more detail. b 17 • [Peter, bad migz barbere sig, •z]. Peter ask-PAST me shave-INF REFL 'Peter asked me to shave him.' c. [Peter, bad Jgrgen= barbere sig, z]· Peter ask-PAST J0rgen shave-INF REFL 'Peter asked J0rgen to shave him/himself.' d. *[Peter; sagde [atjeg barberede sig;]]. Peteri say-PAST that I shave-PAST REFLi 131 The sentence in (19a) illustrates that while sig may be locally bound 18 , 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 most 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 of sig is only ok as long as the antecedent is still located inside the minimal tensed clause containing sig. Though languages which allow long-distance anaphora seem to differ widely with respect to the type of complement out of which binding of reflexives can occur, typological studies claim to have discovered the implicational universal in (20) (adapted from Huang (2000:92-3), cf. also Burzio 1996, 1998). (20) An implicational universal for long-distance anaphora complement types: NPs > small clauses > infinitivals > subjunctives > indicatives What (20) says is that if a language allows LD-binding of reflexives into a certain type of complement then it will also allow LD-binding of reflexive into all the types of complements lower on the hierarchy. That is, if a language allows LD-binding into indicative complement 17 Note that unlike Chinese ziji §c ·self-self, Danish reflexives are not subject to blocking by intervening I'' or 2nd 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. 18 A number of 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 of local sig are not true reflexive anaphors but 132 clauses then it will also allow binding of reflexives into all the other types of complements. Danish, which allows LD-binding out of infinitivals thus also have LD-binding out of small clauses (and resultatives) and NPs as discussed in section 3.3. In contrast, Italian is reported to allow LD binding out of subjunctives and (cf. Huang (2000:93)) and Old Icelandic is claimed to be of the most inclusive type, allowing LD-binding even out of indicative clauses (cf. Sigurdsson 1990a, quoted in Huang (2000:93)). Many attempts have been made to give a unified account of the various types of LD-binding found in the worlds languages. While the reduction of such cross-linguistic variation to unified account based on a small set of general principles is, of course, the ultimate goal of linguistic theory it remains beyond the scope of the present dissertation. The rest of this chapter will therefore be limited to binding of sig in Danish. That is, our aim will be to explain the following descriptive generalization concerning domain restrictions, see (21 ). (21) Descriptive generalization I: sig must be bound inside the minimal tensed clause 19 • Now let us turn to another property of Danish reflexives, viz. subject orientation. Unlike Eng. himself which appears to allow binding by non-subject antecedents 20 , 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 of such analyses. 19 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 of sig out of non-complement infinitival clauses. The sentence in (i) exhibits LD-binding of sig out of an adverbial clause: (i) Den asketiske munkjlyttede hen til Gangesjlodenfor 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 every morning "The ascetic monk moved to the River Ganges to be able to wash himself in holy water every morning.· (23) PeterJortalte Marie= om sig, *z selv. Peter told Marie about REFL self 'Peter told Marie about himself' This leads us to the more precise descriptive generalization in (24). (24) Descriptive generalization I (revised version): sig must be bound by a subject inside the minimal tensed clause. 133 Chomsky's ( 1981, 1986) binding theory, see (25)-(27), has met with a great deal of criticism over the years. It has been claimed that a number of cross-linguistically wide-spread phenomena cannot be captured properly by these principles, e.g. logophors, overlap in distribution of 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 of Chomsky's syntactic approach to binding are basically sound. That is, the distribution of anaphors and pronouns is controlled by syntactic domain constraints like principles A and B, see (25)-(28). (25) a. b. An anaphor is bound in a local domain. 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-commands beta, and (iii) alpha and beta are co-indexed. (27) alpha c-commands beta if and only if (i) alpha does not dominate beta, (ii) beta does not dominate alpha and (iii) the first branching node dominating alpha also dominates beta. 20 See chapter 5 where it is argued that English self-forms locally bound by non-subject antecedents (as in (22) above) are really reduced forms of intensified pronouns, e.g. herself in (22) is assumed to derive from the adnominally intensified pronominal her herself 134 (28) Binding Domain 21 alpha is a binding domain for beta if and only if alpha is the minimal category (i.e. the smallest DP or IP/S) containing beta, a case-licensor or beta, and a SUBJECT accessible to beta. As discussed in the following, apparent exceptions to these principles should be explained either by reference to parametric variation (e.g. with respect to what counts as binding domain in a given language or with respect to morphological properties of reflexives (i.e. their status as either affixes, clitics, free forms, etc.)) or by reference to binding-independent factors (e.g. semantic/pragmatic factors influencing adnominal intensification of reflexives, prosodic factors (e.g. stressability), etc.) rather than by radically changing the architecture of binding theory. As the descriptive generalization in (24) states, in Danish the local domain is the tensed clause. This is illustrated by the sentence in (29) which shows that any subject within the minimal tensed clause is a potential antecedent for sig. (29) Peter, h0rte Marie" bede sygeplejersken" vaske sig, z o· Peter heard Mary ask-INF nurse-the wash-INF REFF 'Peter heard Marie ask the nurse to wash REFL.' That is, more specifically, in (29) the antecedent of sig can be either the closest subject sygeplejersken 'the nurse', the intermediate subject Marie or the matrix subject Peter. Thus the two facts about sig which any theory must be able to explain is why it must be bound within the minimal tensed clause and why the binder must be a subject. In the following a modified version of Pica's (1984,86) LF movement approach to LD-binding of reflexives will be 21 The "binding domain·· referred to in (28) corresponds what in older versions of the theory used to be 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 lP/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' 135 outlined. The advantages of Pica's LF movement analysis of LD-binding have been argued to be that it provides a unified analysis of three properties considered to be common to all LD- anaphora listed here in (30a-c). (30) a. b. c. LD-binding subject-orientation morphological simplicity Successive cyclic movement from lower to higher INFLs/Ts via COMP allow one to consider LD binding of sig as obeying the locality requirement of principle A. As shown in section 3.2.2, sig lacks phi-features. We therefore assume that it has to adjoin to INFLIT in order to obtain features from the subject via spec-head agreement. This explains why sig is always subject-oriented (cf. Jakubowicz 1994: 119). Though both Danish sig 'REFL' and its French counterpart se 'REFL' are clitics base-generated in argument position, they still differ in important ways as described in section 3.2.3. Unlike Romance clitics, e.g. French se, which cannot be separated from their host, Danish sig can be separated from the verb by other material, see examples (II) and (12) in section 3.2.3. That is, French se moves along with its host verb when it moves overtly to adjoin to IN FLIT ( cf. Pollock 1989), see ( 11 ). In contrast, Danish sig is a syntactically independent prosodically bound word which does not from a morphological constituent with its host 22 , see ( 12). Thus when sig moves from its based- generated position to a higher INFLIT this has to be an instance of covert movement at LF. Long-distance binding of sig can thus be explained as a resulting from cyclic movement of sig from the lower IN FLIT to a higher IN FLIT ( cf. Jakubowicz 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 iC (Chomsky 1986, p. 169). 22 ln contrast to Jakubowicz (1994:123). who assumes that certain Danish verbs (i.e. [+affectednessl verbs) "can incorporate the element projected in their internal argument position'', we do not assume incorporation of sig into the verb in cases of locally bound sig (e.g. Peter forsvarer sig 'Peter defends himself). 136 The claim that only morphologically simple reflexives can be LD-bound stems from the observation by Faltz ( 1975) that, cross-linguistically, simplex reflexives tend to allow LD- binding while morphologically complex reflexives must be locally bound. Since, in the LF movement analysis, only head elements can undergo head-to-head movement, that allegedly explains why, as it is claimed, only morphologically simple reflexive elements can be LD- bound. While we agree that Pica's LF movement analysis provides a satisfactory account of both LD-binding and subjection orientation (30a-b ), we disagree with its account of the difference between simple and complex reflexives (30c ). First of all the descriptive generalization that only simple reflexives can be LD-bound while complex reflexives must be locally bound may not be as solid as it seems. The fact that so-called complex reflexives can be long-distance bound has been brought up in the literature, e.g. the following German and Japanese examples of long-distance binding of the complex intensifier-based reflexives sich selbst 'REFL self', and zibun-zisin 'self-self' discussed in Huang (2000:96, example (2.169c,d)), see (31 )-(32). (31) Willi 1 dachte, daj3 Hans 2 mit Fritz 3 iiber sich selbst 12 gesprochen hat. Willi thought that Hans with Fritz about REFL self spoken has 'Willi thought that Hans has spoken with Fritz about himself.' (32) Takasirwa Hirosi-ga zibun-zisinrni kasite kureta kuruma-o kowasite simatte. Takasi-TOP Hirosi-SUBJ self-self lend give car-OBJ broken ended up 'Takasi has broken the car which Hirosi lent self-self.' Based on such examples we conclude that the LF movement analysis has to be modified so as to allow for the possibility of LD-binding of adnominally intensified reflexives. Note that the complex reflexives in (31) and (32), i.e. sich selbst 'REFL self', and zibun-zisin 'self-self', both qualify as adnominally intensified reflexives in that they are composed of the simple reflexives sich 'REFL' and zibun 'SELF' to which the adnominal intensifiers selbst 'self' and 137 zisin 'self have been adjoined. In this sense they are entirely parallel to the Danish intensified reflexive sig selv 'REFL self. Since we know that adnominal intensifiers can be stranded (Q- floated) by movement of their associate 23 , it is only logical to assume that the same thing may happen in the case of intensified reflexives. That is, we assume that, at LF, the xa reflexive sig may move covertly to a higher !NFL or T 0 node on its own leaving the intensifier behind. For expository reasons detailed discussion of LD-binding of complex reflexives will be postponed till section 3.3.1. The remainder of this section will contain a brief outline of how the version of the LF movement analysis defended here accounts for both local and LD-binding of simple sig. The reflexive is assumed to have interpretable but non-valued features. It thus needs to move to get its features valued, see (33). (33) sig [+interpretable, non-valued] features The structure in (34c) illustrates a sentence containing a locally bound reflexive at spell-out. (34) Local binding of thematic sig with neutral predicates (embedded clauses): a. Array: {at, John, vasker, sig, ikke} b. Sentence: 00 00 at John ikke vasker sig. that John not washes REFL ' ... that John doesn't wash himself.' 23 See the analysis of q-tloated adnominal intensifiers in chapter 2. section 2.2.5. c. CP I \ C' I \ at TP I \ John T' I \ To vP I NEG I I ikke \ vP \ I v' \ I v v' \ VP I \ V DP I I vaskede sig 138 In (34) the simple reflexive sig merges with the neutral verb vaske 'wash' from which it receives both case and the internal theta-role. An embedded clause was chosen for this example in order to avoid the complications of V2 word-order and object-shift - both phenomena which occur only in main clauses. The negation - which is assumed to adjoin to the left edge of the predicate (i.e. vP) does not play any role here and is just added to facilitate comparison with structures discussed in (36). Just like traditional LF-movement analyses of LD-binding a Ia Pica (1984, 1986), we assume that it is the specific featura1 properties of sig which dictate that it must move to adjoin to T 0 at LF in order to get interpreted. That is, sig is assumed to have [+interpretable, -valued] features. In other words, since sig has no lexical content of its own it has to move 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-movement analysis accounts for LD-binding of sig. 139 (35) LD-binding of thematic sig with neutral predicates (embedded clauses): a. Array: {at. John, bad, Hans, vaske, sig} b. Sentence: ... at John bad Hans vaske sig. that John asked Hans wash REFL ' ... that John asked Hans to wash him/himself.' c. kr [c at [ TP John, [ r [To sig,] [ vr bad Hansk b t," [ TP P ROk [r [To t, '] [ vr vaske t,]]]]]]]]] d. kr [c at [ TP John, [ r [T 0 ] [ vr bad Hansk b [rr PROk [r [To sigk] [vr vaske tk]]]]]]]]] Since sig is a head it can undergo successive cyclic movement from lower infinite T 0 to a higher finite To where it may receive lexical content from its antecedent through spec-head agreement. It may also, as shown in (35d), get its features valued by spec-head agreement with the subject of the embedded clause. In main clauses, Danish has V2 word order as well as object shift of personal pronouns and reflexives. Y2 word order entails movement of a constituent (often the subject) and the main verb to spec-CP and co respectively. And object-shift involves the movement of pronouns and reflexives from their argument positions to spec-vP ( cf. Platzack and Holmberg (1995), McGinnis (1999)). Both processes are illustrated in (36) which shows the sentence at spell-out. In order to get interpreted the reflexive sig will still have to move to T 0 at LF as described above. (36) Local binding of thematic sig with neutral predicates (main clause): a. Array: {John, vasker, sig, ikke} b. CP I \ John C' I \ vasker TP I t I t \ T' \ vP I Sig I NEG I ikke \ v' \ v' I \ DP v' I \ v VP I \ V DP I I t t 140 So far only examples with local and LD-binding of the simple reflexive sig in argument position of neutral verbs have been discussed. In the following section the viability of the analysis proposed here will be scrutinized by a closer investigation of how it deals with both simple and complex reflexives in a number of different syntactic environments as well as with different types of predicates. The analysis of non-thematic sig which occurs with certain "deponent"linherently reflexive verbs (e.g. skamme sig '(lit.) shame REFL!be ashamed', bramde sig '(lit.) burn REFLI(accidentally) get burned' will be discussed in chapter 4, section 4.3. 141 3.3 Binding of reflexives in different contexts: testing the independence of binding and intensification. In the following sections, the descriptive and explanatory adequacy of the analysis of Danish reflexives presented in section 3.2 will be put to the test by taking a closer look at the behavior of simple and complex reflexives in different syntactic positions. That is, we discuss morphological complexity and locality (section 3.3.1 ), reflexives as direct objects of 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 between sig and sig selv follow from adnominal intensification rather than binding- related factors. 3.3.1. The independence of intensification and locality constraints. According to accounts of Danish currently found in the literature (e.g. Vikner (1985), Jakubowicz (1994), etc.), the complex reflexive sig selv is a subject-oriented reflexive which cannot be long-distance bound, see (37). (37) McArthur, bad sygeplejerenk give__ en morjinindsprajtning. McArthur asked nurse-the give a morphine injection 'McArthuri asked the nursei to give himself a morphine injection.' a. sig, k REFL b. sig., k selv REFL Jakubowicz ( 1994:130-131) explains the absence of LD-binding of the complex reflexive sig selv as a consequence of its syntactic structure. Unlike sig, which is a head, sig selv is a maximal projection which cannot undergo successive cyclic movement from a lower to a higher INFLIT via COMP without violating the ECP. However, since sig selv is also 142 featureless it must move the closest INFLIT to obtain features which explains why, like sig, it can only be bound by a subject antecedent. As mentioned above, sig selv has so far been assumed to be a local reflexive. However, given the right context it is possible to find examples in which sig selv may (at least marginally) felicitously be LD bound, see (38) and (39). (38) Context: McArthur is an extremely tough general who feels that the lightly wounded soldiers ought to put up with the pain in order to save painkillers for the truly needing. When McArthur himself got a large piece of shrapnel in his thigh he stubbornly refused to take any kind of painkillers. But in the end the pain became too much 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, migk endeligt give en morfinindsprc1jtning. But yesterday, late on night-the asked McArthur me finally give-INF a morphine injection 'But yesterday, late at night McArthur; finally asked mek to give a morphine injection.' a. ?sig, selv REFL; self b. ?sig, REFL; In (38) the local subject mig 'me' is first person and is thus ruled out as a potential antecedent for sig due to lack of (pseudo-)agreement with the reflexive. The only possible antecedent for sig is therefore the matrix subject McArthur. The elaborate context given in (38) also contributes to making the matrix subject the most likely antecedent. Furthermore, the context explicitly contrasts a typical situation, in which McArthur asks the nurse to give one of the other wounded soldiers an injection, with the truly exceptional situation in which he asks the nurse to give himself an injection. In other words, the explicitly contrastive context triggers focusing of the reflexive sig which, due the fact that simple sig cannot be stressed, has to be realized as the adnominally intensified reflexive sig selv. The example in (39) is identical to (38) except for the local subject which has been replaced by the 3rd person DP sygeplejeren 'the nurse' which is a potential antecedent for the reflexive since it (pseudo-)agrees with sig. 143 (39) Context: same as in (38). Men igar, sent pa natten bad McArthur, endeligt sygeplejerenk give en morfinindsprgjtning. But yesterday, late on night-the asked McArthur finally nurse-the give-INF a morphine injection 'But yesterday, late at night McArthur; finally asked the nursek to give a morphine injection.' a. ?sig, selv REFL; self b. sig 1 REFL; c. sigk selv REFLk self (odd reading in this context) d. *sigk REFLk (*because give is anti-reflexive) As shown in (39a) the presence of this potential subject antecedent does make LD-binding of sig selv more difficult to obtain than LD-binding of simple sig (39b ), but it does not exclude it. The'*' in (39d) is due to the anti-reflexivity of the predicate give 'give'. Now, if it is the case that sig selv is even marginally acceptable long-distance bound (given the right context) then Pica and Jakubowicz's accounts of sig selv as a local anaphor cannot be correct. We know from the discussion of q-floated intensifiers in chapter 2, section 2.2.5, that adnominal intensifiers may be stranded in the same way as floated quantifiers when the associate has to move to a higher position, e.g. V2 movement of 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. b. Lcereren havde selv skrevet en stil. [ Lcereren, havde [oP t 1 selv] skrevet en stil] teacher-the had self written an essay 'The teacher had himself written an essay.' Selv har kongen ringe magt. [ Selv, har [oP kongen t 1 ] ringe magt] selfhas king-the slight power 'The king himself has little power.' It is therefore natural to assume that the sig of the intensified reflexive sig selv may leave the adjoined intensifier selv behind when it undergoes cyclic LF movement to get features from the matrix subject in sentences like (38a) and (39a). While such an approach makes it possible to account for LD-binding of sig selv, it does not by itself explain why LD-binding of complex 144 reflexives is so rare as to have been overlooked (or marginalized) by researchers working on Danish reflexives for so long. Within the analysis proposed in this dissertation a straightforward explanation of the rarity of LD-bound sig selv naturally presents itself. As discussed in greater detail in section 3.3.2, one of the main motivations for intensifier- adjunction to simple reflexives is predicate meaning. Anti-reflexive predicates like ajl0se 'replace, relieve (someone as guard)' are not readily compatible with reflexive scenarios. Therefore, when such readings are forced, they have to be construed as relations between the referent of the antecedent and a representation ofthe referent of the reflexive. As discussed in section 3.3.2, this doppelganger-effect triggers the intensification of locally bound reflexives with anti-reflexive predicates, see ( 41 a). ( 41) a. b. Vagten; ajl0ste *sig, I sig, selv. guard-the replaced REFL/REFL self 'The guard replaced himself.' Vagten, bad Peterk aflgse sig, *k I sign, k selv. Guard-the asked Peter replace REFL 'The guard asked Peter replace himself.' When the meaning of the predicate is compatible with reflexive scenarios (e.g. neutral and inherently reflexive predicates) then selv-adjunction to locally bound sig is not mandatory for semantic reasons. In the case of neutral predicates, selv-adjunction to sig thus has to be motivated either by explicitly contrastive contexts or phonological factors (as in the case of certain prepositional predicates). The same applies to LD-bound instances of reflexives in argument position of anti-reflexive verbs, see ( 41 b). That is, as shown in ( 41 b) vs. (38-9), in the absence of an explicitly contrastive context, as for instance, the kind triggered by anti- reflexive predicates ( 41 a), intensification of LD-bound sig selv is not felicitous. Other contexts which license intensification of reflexives are the so-called wax museum contexts involving the use of reflexive pronouns to refer to statues/representations of 145 the individual denoted by the antecedent. First consider the instances of locally bound sig and sig selv in ( 42). (42) Context: Imagine Bill Clinton visiting the wax museum. He notices a statue of himself with an unshaven face. Since he doesn't like the look of the statue he takes out a razor and starts to shave it. A third party witnessing this situation describes it in the following terms: a. Bill Clinton, barberede sig,. b. sig, selv. Bill Clinton shaved RELF/REFL self 'Bill Clinton shaved 0/0 himself.' The sentence with the simple reflexive in ( 42a) can only have the interpretation in which the real Clinton shaves himself(i.e. the real Clinton), see (43a). It cannot have the so-called statue- reading, see ( 43b ), in which the real Clinton shaves a statue of Clinton. In contrast, the sentences with the complex reflexive in ( 42b) can have both reading ( 43a) and ( 43b ). (43) a. b. Clinton<real> shaves himself<real>. Clinton<real> shaves himself<statue>. What these examples show is that in order to get a statue-reading of a reflexive pronoun it must be adnominally intensified. The example in (44) is designed to test what happens when a statue-reading of a LD-bound reflexive is forced. (44) Context: Imagine Bill and Hillary Clinton visiting the wax museum together. Bill notices a statue of himself with with an unshaven face. He happens to have a razor in his inner pocket but since he is temporarily 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 witnessing this situation describes it in the following terms: a. Clinton, bad Hillaryk barbere #sig,. b. (?)sig, selv. Clinton asked Hillary shave REFL/REFL self 'Clinton asked Hillary to shave him.' Although solid judgments are hard to come by in sentences like (44) 24 it seems to be the case that sig selv is more felicitous than sig. Once again, the generalization that sig selv cannot be 24 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 of reflexives. The usual reaction being: "I would never say '"Clinton shaved himself' in such cases but rather "Clinton shaved the statue of 146 LD-bound appears to be too strong 25 • That is, syntactically, sig and sig selv behave alike with respect to binding-related locality constraints. Pragmatically and semantically they differ with respect to which contexts they may or may not occur in. But these differences falls under the theory of intensification, as presented in chapter 2. So while it is true that there is a strong tendency for sig selv to have local antecedents this is due to the pragmatic/semantic constraints on intensification rather than to syntactic locality constraints on the distribution of reflexives and pronouns. 3.3.2 Locally bound reflexives in argument position: complex reflexives = a sub-type of adnominal intensification. In the following four sections (i.e., sections 3.3.2.1-4) an intensification-based account of the distribution of Danish simple and complex reflexives, i.e. sig and sig selv, with different types of predicates will be outlined. Based on what kinds of nominal expressions they allow in object position, Danish verbal predicates can be divided into three main types 26 : (i) "neutral" predicates which allow both simple and complex reflexives (i.e. sig and sig selv) and DPs, see ( 45), (ii) "anti-reflexive" predicates which allow the complex reflexive sig selv and DPs but not the simple reflexive sig, see ( 46), and (iii) "inherently reflexive" predicates which allow only the simple reflexive sig, see ( 4 7). (45) "Neutral" predicates: a. Peter vasker sig I sig selv I bilen. Peter washes REFL I REFL-SELF I car-the 'Peter washes himself I the car.' himself', 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. 25 See also example (I 09b) in section 3.3.5 which clearly shows that intensified possessive reflexives (e.g. sin egen 'POSSREFL own') may be LD-bound. 26 This tripartite typology of predicates is directly inspired by a similar proposal by Zribi-Hertz ( 1995). b. Peter forsvarer sig /sig selv I Marie. Peter defends REFL I REFL-SELF I Mary 'Peter defends himself I Mary.' ( 46) "Anti-reflexive" predicates: a. Peter mistrenker *sig I sig selv I Marie. Peter suspects *REFL I REFL-SELF I Mary 'Peter suspects himself I Mary.' b. Peter misunder *sig I sig selv I Marie. Peter envies *REFL I REFL-SELF I Mary 'Peter envies himself I Mary.' ( 47) "Inherently reflexive" predicates 27 : a. Peter skammer sig /*sig selv /*Maire. Peter shames RELF I *REFL-SELF I *Mary 'Peter is ashamed of himself I Mary.' b. Peter dukker sig I *sig selv I *Marie. Peter ducks REFL I *REFL-SELF I *Mary 'Peter ducks *himself I *Mary.' 147 The distribution of different types of nominal expression in object position with the different verb types is summarized in ( 48). (48) Distribution of nominal expressions in object position 28 : a. Anti-reflexive *sig I b. Neutral sig I c. Inherently reflexive sig I sig selv I sig selv I *sig selv/ DP DP *DP We argue that the so-called "neutral" transitive verbs that allow the simple 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) assume that the so-called "neutral" verbs occur twice in the lexicon: once, as a true transitive 27 The exact definition .. inherently reflexive'' is discussed in chapter 4 where a distinction between inherently reflexive predicates with thematic reflexives vs. inherently reflexive predicates with non-thematic reflexives is introduced. 28 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 *sig I sig selv / Peter sold REFL REFL self (ii) neutral ditransitive : Peter k0bte sig sig selv I Peter bought REFL REFL self (iii) inher. refl. ditransitive : Peter anskaffede sig *sig selv / Peter acquired REFL REFL self ·Peter sold/bought/acquired himself/Mary a house.' Mary Mary Mary Mary *Mary Mary et hus. a house et hus. a house et hus. a house 148 verb (which require the complex reflexive sig selv), and once as an inherently reflexive verb (which does not need to be overtly reflexive-marked by the SELF element of the complex reflexive). These competing views of"neutral" predicates will be compared and evaluated in more detail in chapter 4, section 4.2. With respect to binding, reflexives, pronouns, and DPs have widely different properties. While DPs and pronouns, have semantic content and are able to refer on their own, the reflexive sig does not have phi-features and behaves semantically like a variable which is dependent on its binder for reference. However, from the point of view of intensifier adjunction, reflexives, pronouns, and DPs are all nominal expressions which may be the target of adnominal intensification. In other words, by assuming that the selv in sig selv is the same adnominal intensifier as the one we find in intensified DPs, e.g. Peter selv kom til medet 'Peter himself came to the meeting', 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. Simply observing the distribution of simple reflexives, complex 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 following sections, binding-independent semantic tests which can be used to classify predicates. 3.3.2.1 Anti-reflexives predicates: predicates which are incompatible with reflexive scenarios In this section we show that it is possible to predict whether or not a given verb is anti reflexive by testing its semantic compatibility with reflexive scenarios. To illustrate how this semantic testing of predicate meaning works let us take a closer look at the transitive predicates in ( 49). 149 (49) a. aflgse 'replace (the guard), succeed to' b. bgnfalde 'implore, beseech, entreat, plead' c. efterfgfge 'succeed (e.g. Peter succeeded Kim as director)' d. mistcenke 'suspect (somebody of doing something)' e. misunde 'envy, be envious of' f. barbere 'shave' g. tgrre '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 determine whether they are "anti-reflexive" or "neutral" without looking at the distribution of simple and complex reflexives? As mentioned above, we argue that testing for semantic compatibility with reflexive scenarios makes it possible to predict whether a transitive verb is neutral or anti-reflexive. On the basis ofthis kind of testing, we predict that the verbs in (49f-h) are neutral while those in (49a-e) are anti-reflexive. Based on what they mean, the predicates in (49a-e) simply do not make much sense in reflexive scenarios. For example, how can one ofthe guards standing guard in front of Buckingham Palace possibly replace himself, see (49a) and (50a)? The new well-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 himself in this sense and still remain the same person. The same goes for (49c) efterfg/ge 'succeed'. Strictly speaking it makes little sense to say that King Henry IV succeeded himself as king of England, especially since in traditional monarchies a new king can only be crowned after the death of the old, see (SOb). One way to make sense of such an utterance would be to assume that King Henry IV only pretended to be dead and then dressed up as the crown prince in order to be crowned in his place. In other words, he would have to appear in someone else's guise or under a different representation (i.e. that of the crown prince). That is, the generalization seems to be that anti-reflexive predicates presuppose that the linguistic expressions occupying 150 the positions of external and internal argument do not refer to the same individual, or, as in the sentences in (50), do not refer to the same representation of a given individual 29 • (50) a. b. c. d. e. Vagten ajl0ste *sig I sig selv. guard-the replaced REFL I REFL self 'The guard replaced himself(as guard). Henry IV efterfulgte *sig I sig selv. Henry IV succeeded REFL I REFL self 'Peter succeeded himself (as king of England).' Peter bgnfaldte *sig I sig selv. Peter implored REFL I REFL self 'Peter implored himself.' Peter mistcenkte *sig lsig selv. Peter suspected REFL I REFL self 'Peter suspected himself of doing something.' Peter misude *sig I sig selv. Peter envied REFL I REFL self 'Peter envied himself The predicates b0nfalde 'implore', mistcenke 'suspect (somebody of doing something)' and misunde 'envy' also fall under this generalization in that they only make sense when the arguments are two different individuals, or the same individual appearing under a different representation, see (SOc-e). Take the example of mistcenke 'suspect' in (SOd). Suspecting somebody of murder means that you have reasons to believe that this person committed a crime but are still unsure whether or not he is the culprit because you lack the evidence to prove it. Since people are normally assumed to know whether or not they have committed a crime it does (under normal circumstances) not make sense to say that they suspect themselves of having committed a crime. The uncertainty presupposed by the predicate mistcenke 29 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 afbarber. barberede *siglsig selv. (qua-sentence) Peter, in quality of 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. 151 'suspect', illustrated by the contradictory flavor of the sentence in (51), clashes with the presupposed knowledge of one's own acts and makes the reflexive use of this predicate infelicitous, unless some kind of presupposed representational non-identity is involved. (51) #Peter mistamkte Mary for at have skudt Kennedy og han vidste at hun havde gjort det. #'Peter suspected Mary of having shot Kennedy and he knew that she had done it.' That is, one way to make sense of (SOd) would be to assume that Peter is schizophrenic and thus has multiple personalities, or to assume that he has had memory loss, etc 30 • Let us now turn to the verbs barbere 'shave', torre 'dry' andpynte 'adorn' in (49f-h). Testing them for compatibility with reflexive scenarios indicates that they are neutral rather than anti-reflexive. A proposition like Peter shaved/dried/adorned himself does not have any of the non-sensical or weird flavor found with the sentences in (50a-e). The activities denoted by the predicates in ( 49f-h) are entirely compatible with reflexive scenarios. That is, the verbs barbere 'shave', torre 'dry' and pynte 'adorn' do not evoke any presupposition of (representational) non-identity. A sentence like Peter shaved himse(f can (and usually does) mean that Peter shaved his own face. It does not have to mean that he shaved himself under a different representation (e.g. a statue of himself, etc.). Hence, based on compatibility 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 examples in (52) show that this prediction turns out to be true. (52) a. Peter barberedeltorrede/pyntede sig I sig selv I John. Peter shaved I dried I adorned REFL I REFL self I John. 'Peter shaved/dried/adorned himself/John.' 30 The sentence in (i) might be construed as a counterexample to this account. (i) I suspected myself of 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) 152 Based on the above examples and discussion we conclude that testing a predicate's semantic compatibility with reflexive scenarios can be used to determine whether or not a predicate is "anti-reflexive" 31 • If a transitive predicate is semantically incompatible with reflexive scenarios then it is "anti-reflexive". In other words, an anti-reflexive predicate can be defined as a predicate whose meaning presupposes that its arguments are either referentially different (i.e. have different semantic values) or (if co-extensional) appear under different representations 32 , see (53). (53) Anti-reflexive predicates: A given predicate is anti-reflexive if it evokes a presupposition of (representational) non-identity of its arguments that is not cancelable by context. Conversely, if a transitive predicate is semantically compatible with reflexive scenarios then it is "neutral". It is important to note that the definition of anti-reflexivity defended in this dissertation does not crucially rely on the notion of co-argumenthood. What is important is the relationship between reflexive and antecedent in terms of presupposed representational identity vs. non-identity. Certain contexts trigger a presupposition that reflexive and antecedent are (representationally) non-identical. Predicate meaning is only one among different source of this presupposition. In sections 3.3.3, 3.3.4, and 3.3.5 we show that the meaning of certain resulative constructions, EC, constructions, and possessive constructions, e.g. (100), (104) and (Ill), also trigger a presupposition of representational non-identity of reflexive and antecedent. The definition in (53), as well as the definitions of neutral and inherently reflexive predicates given throughout section 3.3.2, are thus only to be considered definitions of different predicate-types. The notions of anti-reflexivity, neutrality and inherent 31 Cf. E. Konig's notion of"other-directedness'" and "other-directed"" predicates. 32 For more discussion of the notions of identity and representational identity, see section 3.3.2.7. 153 reflexivity are more general and do not rely on the notion of co-argumenthood but rather on presupposed identity vs. (representational) non-identity of antecedent and reflexive. 3.3.2.2 Predicates which are semantically incompatible with non-reflexive scenarios Predicates that are semantically incompatible with non-reflexive scenarios, i.e. predicates which require reflexive scenarios 33 , can be classified as "inherently reflexive," see (54)-(56). (54) (55) (56) a b. Peter, is out of his, /*his, own I *DP, 's mind. Peter, cleared his, I *his, own I *DP, 's throat by saying ahem. Peter solede sig I *sig selv I *Marie. Peter tanned REFL I REFL self I Marie 'Peter was tanning/sunbathing.' a. b. Peter took the knife with Peter tog kniven med Peter took knife-the with 'Peter took the knife with (Danish) him I *himself/ *Mary. sig I *sig selv I *Mary. (Danish) RELF I REFL self I Mary him I *himself I *Mary.' The meanings of the complex predicate 'be out of one's mind' in (54) and the simple verbal predicate sole 'tan' (55) both presuppose the identity of the participants involved in the action/event. Unlike normal bathing, sunbathing can only be performed 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 *Marie/*the child/etc.') intensifier-adjunction to sig is not possible. Similarly, since it not possible to go out of somebody else's mind, intensifier- adjunction of own 3 ~ to the possessive pronoun his in (54a) is not ok. The English and Danish examples in (56) also qualify as inherently reflexive. In the account proposed here the judgments in the examples in (56) are explained in the same way as (54)-( 55). The meaning of 33 Note the difference between the narrow definition of reflexivity (antecedent-anaphor relation between co arguments) and the more loosely defined notion of 'ret1exive 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, washed 0, himself (ii) Peter, cleared [his, throat]z· 154 the whole construction 'X took a knife with Y' is only compatible with reflexive scenarios in which X=Y. All non-reflexive scenarios in which Xi Y necessarily crash because they lead to non-sensical meanings. The theory of adnominal intensification developed in chapter 2 can thus be used to explain why the complex form of the reflexive is not possible. Since the unique object cannot be contrasted with other individuals with whom the knife could have been brought (i.e. *Peter tog kniven med Marie/barnet/etc. '*Peter took the knife with Marie/the child/etc.') intensifier-adjunction to sig is not possible because it would lead to a violation ofthe constrastiveness condition on adnominal intensification. Presupposed identity of the participants in the action denoted by the predicate/construction can thus b~ used to determine whether 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 of identity of its arguments that is not cancelable by context. Since the examples in (54), involving possessive pronouns and intensifiers, obviously do not involve antecedent-anaphor relations between co-arguments of the same predicate, they constitute strong evidence against both the lexical ambiguity accounts of local reflexives ( cf. Reinhart and Reuland 1993) and reanalysis accounts ( cf. Vikner 1985). The examples in (56) also pose a threat to lexical ambiguity and reanalysis accounts since it is not obvious that the Subject DP Peter and him/sig in (56a,b) are co-arguments of the same predicate take/tage 'take'. Unless, of course, one analyzes take_ with_ as a complex predicate. In contrast, since the nominal approach to binding defended here is not based on co-argumenthood, it encounters none of these difficulties in providing a unified account of (54-56). 34 As discussed in section 3.3.5. we assume that own is the suppletive form of the adnominal intensifier himself 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 combining the semantic tests proposed in the previous two sections, a given predicate can now be identified as "neutral" (independently ofbinding-theoretic factors) if it is semantically compatible with reflexive scenarios while at the same time not being semantically incompatible with non-reflexive scenarios. The verbs in (58a-c) meet these requirements and thus qualify as neutral. (58) a. b. c. barb ere t0rre pynte 'shave' 'dry' 'adorn' As discussed above, the verbs in (58a-c) are entirely compatible with reflexive scenarios, i.e. they do not evoke any presupposition of non-identity of the representations of their arguments. A sentence like Peter shaved (himself) can (and usually does) mean that Peter shaved his own face. It does not have to mean that he shaved himself under a different representation. We thus conclude that (58a-c) are not anti-reflexive. Furthermore, these predicates are also compatible with non-reflexive scenarios. That is, sentences like Peter shaved John are perfectly meaningful. This shows that these verbs are not inherently reflexive, i.e. do not evoke any presupposition of identity of their arguments. Based on such semantic testing for anti-reflexivity, see (53), and inherent reflexivity, see (57), we conclude that (58a-c) are neutral predicates. More generally, we define neutral predicates negatively as predicates that are devoid of particular presuppositions concerning the (representational) identity of their arguments. If (58a-c) are neutral predicates then they should allow sig, sig selv and DPs in object position. As (59) shows, this prediction turns out to be true. (59) Peter, barberedelt0rrede/pyntede sig; I Peter shaved /dried I adorned REFL I 'Peter shaved (himself) I John.' sig, selv REFL self I I John. John 156 Notice, however, this kind of semantic testing of neutral predicates is not problem-free. Even if a transitive predicate is logically compatible with reflexive readings this does not necessarily mean that it always allows the simple 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 compatible with reflexive readings. The verb koge 'boil' is an example of such a predicate, which at first glance appear to behave as an anti-reflexive in that it disallows simple sig, see (60), but is nevertheless semantically compatible with reflexive readings. (60) Peter kogte #sig I Peter boiled REFL I 'Peter boiled himself I Mary.' sig selv REFL self I I Mary. Mary While it may be strange or unexpected given what we know about the world, it still does not lead to nonsense to say that someone consciously puts himself in a big water-filled pot on the stove and then turns on the gas in order to boil himself. Semantically both barbere 'shave', see (59), and koge 'boil', see (60), classify as predicates which are compatible with non-reflexive scenarios while at the same time not being incompatible with reflexive scenarios. They should thus both behave as neutral predicates. However, as illustrated by (59) and (60) they differ with respect to the acceptability of simple sig. How do we explain this difference? The fact that predicate meaning is not sufficient to tell the two verbs apart suggests that their differences are to be explained at the level of expectations generated by standard assumptions about the world rather than lexical semantics. Under normal circumstances a person is not expected to boil him- or herself. Predicates of this type, e.g. koge 'boil, can change behavior (e.g. from "anti-reflexive" to "neutral") as a result of changes in the larger context in which it occurs. Compare the way the predicate koge 'boil' behave in a "normal" context, see (60) with how it behaves in science-fiction context, see (61). 157 ( 61) Science-fiction context based on "special" assumptions about the world: Viste du at marsbeboenes hud kan tale langt h0jere temperaturen end vores. F aktisk er det ganske normalt for marsbebeoere at vaske sig i varm olie. Nogle steder paa Mars er det enddog normalt at koge sig i olie med regelma?ssige mellemrum. 'Did you know that the skin of the Martians can endure much higher temperatures than ours. Indeed, it is quite normal for Martians to wash themselves (sig) in hot oil. Some places it is even normal to boil oneself (sig) in oil with regular intervals.' The different contexts in (60) and (61), based on different basic assumptions about the way the world works, give rise to different sets of expectations. Based on what we know about boiling and the consequences it has for the human body we expect the verb koge 'boil' to be other- directed, i.e. under normal circumstances it denotes an activity that is directed at things or people other than oneself. Hence, it is odd (but not impossible) to say that somebody boiled himself. This explains why simple sig is not fully acceptable in (60). Note that sig is not ungrammatical in (60) but merely unacceptable 35 • As indicated in (60), under normal circumstances focusing of the reflexive through adnominal intensification (i.e. sig selv) is necessary to mark the (representational) non-identity evoked by the expectations of 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 assumptions about the world are different and no longer clashes with the reflexive scenario x boiled x. Hence unintensified sig can occur as the object of koge 'boil' without making the sentence unacceptable. Based on the above discussion we conclude that there are two sub-types of "neutral" predicates, see (62). 35 We use the symbol * to indicate grammaticality violations and # to indicate violation of pragmatic/discourse principles. 158 (62) Sub-types of neutral predicates: a. (Normal) neutral predicates: A given predicate is neutral if it lacks presuppositions concerning the (representational) identity of its arguments. Examples: vaske 'wash' barbere 'shave' b. "Hidden" neutral predicates: torre forsvare 'dry' 'defend' A given neutral predicate is a "hidden" neutral verb if normal background assumptions about the world trigger expectations of (representational) non identity of its arguments. Examples: koge dolke filme forgylde 'boil' 'stab' 'film, shoot' 'gild' Both the normal neutrals in (62a) and the "hidden" neutrals in (62b) are defined by their lack of presuppositions of either (representational) non-identity or identity of their arguments. To the untrained eye, however, the predicates in (62b) at first appear to behave like anti-reflexives in that they disallow simple sig under normal background assumptions about the world, see (60). But given the right context (i.e. a different set of background assumptions about the world) they can be shown to accept simple sig in object position, see (61 ). Since their true nature as neutral verbs (i.e. the lack of presuppositions about the identity or non-identity of their arguments) is hidden under normal circumstances, we refer to these verbs as "hidden" neutral predicates. Note that the anti-reflexive predicates discussed in section 3.3.2.1 cannot be made to allow simple sig no matter how much the context is changed. That is, the presupposition of non-identity of the representations ofthe arguments evoked by anti-reflexive predicates cannot be overridden or cancelled by context. That is, even in a science-fiction scenario in which Martians are described as incurable schizophrenics who pass their days suspecting themselves of this and that; even in such scenarios, simple sig is still not ok in sentences like (63). (63) Marsbeboeren mistcenkte *sig I sig selv I Peter. Mars-dweller-the suspected REFL I REFL self I Peter 'The Martian suspected himself I Peter.' 159 The present approach to reflexives can account for the change-of-behavior data in (60-61) without assuming the existence of any specific lexical features distinguishing between "hidden neutral" and "normal neutral" predicates. Instead, it is assumed that the predicate koge 'boil' is one and the same in both (60) and (61) and that it is the changes in context (i.e. background assumptions about the world) which are responsible for generating different sets of presuppositions in (60) and (61). This contrasts with other approaches to reflexives (cf. Reinheart and Reuland 1993) which would have to assume the existence of two homophonous predicates: (i) kogel 'boil (something)' a "normal" 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 meaning and pragmatic factors can be used to predict whether a given verb is anti-reflexive, inherently reflexive, or neutral. Whether a predicate is anti-reflexive or not depends on whether or not it presupposes (representational) non-identity of its arguments, see the definition of anti-reflexivity in (53), repeated here as (64). Whether a predicate is inherently reflexive or not depends on whether it presupposes identity of its arguments, see (57), repeated here as (65). As shown in (66) neutral predicates are characterized by the lack of presuppositions concerning the identity or non-identity of their arguments. Finally, "hidden" neutral predicates, which also lack such presuppositions, are distinguished from (normal) neutral predicates in that they evoke a expectations (based on world-knowledge) ofthe (representational) non-identity of their arguments, see (62b) repeated here as (67). 160 (64) Anti-reflexive predicates: A given predicate is anti-reflexive if it evokes a presupposition of (representational) non-identity (of its arguments) that is not cancelable by context. (65) Inherently reflexive predicates: A given predicate is inherently reflexive if it evokes a presupposition of identity (of its arguments) that is not cancelable by context. (66) Neutral predicates: A given predicate is neutral if it lacks presuppositions concerning the identity (of its arguments). (67) "Hidden" neutral predicates: A given neutral predicate is a "hidden" neutral verb if it normal background assumptions about the world trigger expectations of (representational) non-identity (of its arguments). It is important to remember that the different types of predicates defined in (64-7) are not based on featural differences. Rather the behavior of a given predicate with respect to intensification of reflexives depends directly on its meaning (whether or not it has presuppositions concerning the identity of its arguments) and, in the case of hidden neutrals, on the meaning of the verb plus background assumptions about the world. That is unlike, for example Zribi-Hertz who bases her predicate typology on a [+/-disjoint reference feature], the present account does not rely on binding-related features. Furthermore, as mentioned above, although the definitions in (64)-(67) refer to arguments of predicates, the notion of argumenthood is not crucial for semantic notions of "anti-reflexivity", "neutrality" and "inherent reflexivity". 3.3.2.5 Complex reflexives as arguments of proto-typical neutral predicates. In the previous section, we explained why locally bound reflexives occurring in object position of 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 161 bound intensified reflexives as direct objects and why they do take have intensified reflexives in some contexts by simple reflexives in other. As shown in chapter 2, adnominal intensification of any kind of nominal expression is felicitous only in contrastive contexts, i.e. in contexts in which the intensified DP could have some other referent, see the contrastiveness principle repeated here as (68). ( 68) Contrastiveness condition on adnominal intensification: A nominal expression can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. Ofthe three unsaturated propositions in (69-7lb) both (69b) and (70b) give rise to felicitous sentences when expressions referring to entities other than the subject Peter are saturating the empty argument position indicated by x. In other words, with both neutral and anti-reflexive predicates the referent of the expression filling the internal argument position can potentially be contrasted with other entities. Hence intensification of sig is possible with these kinds of verbs. In contrast, with inherently reflexive predicates (71) the argument position can only be filled by an expression which corefers with the subject Peter, i.e. sig. Hence, the referent of the expression filling the internal argument position cannot be contrasted with other entities. Consequently, ( 68) rules out intensification of reflexives with inherently reflexive predicates. (69) a. Peter mistamkte *sig I sig selv I Mary. Peter suspects *REFL I REFL self I Mary b. Peter suspects x. (70) a. Peter vasker sig I sig selv I Mary. Peter washes REFL I REFL self/ Mary b. Peter washes x. (71) a. Peter tog en kniv med sig I *sig selv I *Mary. Peter took a knife with REFL I REFL self/ Mary b. Peter took a knife with x. As shown in section 3.3.2.1, in the case of anti-reflexive predicates it is the presupposition of (representational) non-identity of their arguments which trigger the intensification of sig. But 162 smce, as argued in the previous section, neutral predicates do not presuppose the (representational) non-identity of their arguments, the intensification of sig with proto-typical neutral predicates has to have some other motivation. In the literature it has sometimes been claimed that intensification of sig is optional with grooming predicates (e.g. wash, dress, clean, etc .. ). However, as the examples in (72)-(73) show, this generalization needs qualification. Intensification of sig with proto-typical neutral predicates is not optional, but rather determined by discourse factors like focus and contrastiveness. That is, intensification of sig has to occur when the referent of sig is focused or explicitly contrasted with some other entity, see (72), and it cannot occur in contexts where sig is not being contrasted or focused, see (73). (72) Q: A: (73) Q: Hvem var det nu at Peter havde vasket? Havde han vasket sin hund? who was it now that Peter had washed had he washed POSSREFL dog 'Who was it that Peter had washed? Had he washed his dog?' Nej. han havde vasket SIG SELV I *SIG. No he had washed REFL SELF I REFL 'No, he had washed himself.' Hvad var det nu at Peter havde g;·artfor han gik i seng? what was it now that Peter had done before he went to bed 'What was it Peter had done before he went to bed?' Havde han barberet sig? had he shaved RELF 'Had he shaved himself?' A: Nej, han havde VASKET sig I #sig selv. no he had WASHED REFL I REFL self 'No, He had WASHED himself.' The example in (73) shows that when the contrast is not on the object of vaske 'wash' 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 more than one instance of focus is allowed in one sentence ( cf. Baker's ( 1995)). It has been argued more that a sentence may contain more than on focused element, e.g. Simpson & Wu (2002). Regardless of whether one or more foci is allowed per sentence, the unacceptability of (73) follows from the lack contextual motivation 163 for focusing the simple reflexive sig, in addition to the predicate. In other words, we do not exclude the possibility that it may be possible to find contexts in which (73) is ok. Based on the above discussion, we conclude that although neutral predicates allow both simple unintensified and intensified reflexives, intensifier-adjunction of reflexives still needs to be licensed. Contrastive focus is one way to license complex reflexives with neutral predicates. In the section 3.3.2.7 additional "contextual" triggers of intensifier adjunction of reflexives with 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 whether a given predicate is neutral or anti-reflexive. Neutral verbs differ from anti-reflexive verbs not only with respect to their meaning (i.e. whether or not they presuppose (representational) non- identity of their arguments) and, consequently, the distribution of simple and complex reflexives, but also with respect to the possibility of taking inalienably possessed direct objects. The examples in (74-75) illustrate the differences between neutral and anti-reflexive with respect to inalienable possession. (74) Neutral predicates allow inalienable possession: (75) Peter vaskede hamderne. Peter washed hands-the. ok(i) 'Peter washed his own two hands.' 0 \ii) 'Peter washed the hands (of the body he was dissecting).' Anti-reflexive predicates do not allow inalienable possession: Peter stjal hcenderne. Peter stole hands-the. *(i) 'Peter stole his own two hands.' o\ii) 'Peter stole the hands (ofthe body he was dissecting).' (inalienable) (alienable) (inalienable) (alienable) (76) "Hidden neutral" predicates tend not to allow inalienable possession: Peter kogte hcenderne. Peter boiled hands-the. \i) 'Peter boiled his own two hands.' (inalienable) ok(ii) 'Peter boiled the hands (ofthe body he was dissecting).' (alienable) 164 For the sentence in (74) both of 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 normal circumstances they are incompatible 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 olie fgrend han lavede mad. Martian-the boiled hands-the in oil before he made food 'The Martian washed his hands in oil before cooking.' (inalienable) The table in (78) sums up the behavior of different types of predicates with respect to inalienable possession. (78) Testing the availability of inalienably possessed DP objects. Predicate type Allows inalienable possession a. Anti-reflexive NO b. Neutral YES c. d. "Hidden Neutral" Inherently reflexive YES/NO (depending on context) (Does not apply) According to L0drup (1997), who analyzes sig as a generalized inalienable, the ability to license inalienable possession is a common property of all neutral predicates. L0drup needs this assumption to support his claim that locally bound simple sig and inalienable possessed objects are found in the same contexts. While we do not share these assumptions we do consider inalienable possession to be a useful (though not fool-proof) method to distinguish between anti-reflexive and neutral predicates. However, rather than analyzing sig as a generalized inalienable, as proposed by L0drup (1997), we adopt Vergnaud and Zubizarreta's (1992) analysis of inalienable 165 constructions which is based on the assumption that inalienable nouns take a possessor argument which is syntactically bound to the external possessor (which in Danish is usually the subject DP). Vergnaud and Zubizarreta argue that while inalienable nouns take a possessor argument, see (79a), alienable nouns do not, see (79b). (79) a. b. hamderne(x) hcenderne 'the hands' 'the hands' They furthermore assume the existence of a lexical redundancy rule relating the two lexical entries in (79), cf. Vergnaud and Zubizarreta 1992, p. 596, 60 I, numbers of examples have been adapted. Adopting this approach enables us to view inalienably possessed nouns to be a kind of anaphoric nominal expressions whose distribution can be accounted for by the principles of binding theory. The fact that inalienable possessed nouns seem to be found in the same environments as simple sig is therefore not surprising. The presupposed (representational) non-identity of anti-reflexive predicates explains why the inalienable reading of hcenderne 'the hands' in (75) is not available. In other words, the presupposition of (representational) non-identity extends to inalienably possessed objects. Since the anti- reflexive meaning of stjcele 'steal' in (75) presupposes representational non-identity of its arguments it also presupposes the representational non-identity of the subject and the anaphoric possessor argument of the inalienable object noun 36 • 36 Note that it could be argued that the analysis defended here would lead to the wrong prediction that intensification of the C:J possessor argument anaphor should be possible, see (i). (i) *Peter stjal C:J egne 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 of possessive reflexives the full paradigm includes three different forms, e.g. (/) (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(/) 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 166 3.3.2. 7 Linking doppelganger effects and anti-reflexivity Intensifier-adjunction to simple reflexives can take place in a number of different contexts. In sections 3.3.2.1 and 3.3.2.3 anti-reflexive (containing presuppositions of (representational) non-identity) and "hidden neutral" predicates (which together with background assumptions about the world generate expectations of (representational) non-identity)) were shown to trigger intensification of simple reflexives. In section 3.3.2.5 contrastive contexts were identified as another trigger of intensification of simple sig with neutral predicates. In this section a number of other contexts triggering intensification of simple reflexives with neutral predicates will be discussed, namely wax museum contexts (statue-readings), qua-sentences, and strict readings of reflexives in VP-ellipsis constructions. In the following we will show that all these intensification triggering contexts, see (80d(i-iv)), involve some kind of 'doppelganger-effects', i.e. (representational) non-identity ofthe reflexive and its antecedent. (80) Complex reflexives (e.g. sig selv): a. anti-reflexive predicates b. hidden neutral predicates c. contrastive contexts (with neutral predicates) d. 'doppelganger-effects' (with neutral predicates): (i) statue-readings (ii) qua-sentences (iii) strict (and sloppy) readings in VP ellipsis Example: (50) (60-1) (72) (81) (84) (86) In other words, we argue that all the phenomena in (80a-d) should be given a unified analysis in terms of (representational) non-identity of the reflexive and its antecedent, the only difference being the locus of the presupposition of representational non-identity: with the anti- reflexive predicates (80a) the presuppositions is part of the meaning of the predicate; in the case of the different types of 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. 167 3.3.2.7.1 Selv as marker of statue-readings Since Jackendofrs (1992) paper "Madame Tussaud meets the binding theory" called attention to the fact that names can be used to refer to statues or representations of their normal referents, it has been known that simple and complex reflexives differ in this respect. First consider the instances of locally bound sig and sig selv in (81 ). (81) Context: Imagine Bill Clinton visiting the wax museum. He notices a statue of himself with an unshaven face. Since he doesn't like the look ofthe statue he takes out a razor and starts to shave it. a. Bill Clinton; barberede sig;. b. Bill Clinton, barberede sig, selv. Bill Clinton shaved RELF/REFL self 'Bill Clinton shaved (himself).' The sentence with the simple reflexive sig in (81 a) can only have the interpretation in which the real Clinton shaves himself (i.e. the real Clinton), see (82a). It cannot have the so-called statue- or doppelganger-reading, see (82b), in which the real Clinton shaves a statue of Clinton. In contrast, the sentence with the complex reflexive sig selv in (81 b) can have both reading (82a) and (82b) 37 • (82) a. b. Clinton<real> shaves himself<real>. Clinton<real> shaves himself<statue>. We argue that this semantic difference between simple/unintensified reflexives and complex/intensified reflexives is a consequence of intensifier-adjunction. As described in chapter 2, adnominal intensification automatically generates a set of alternative referents for the associate of the intensifier. In the case of sig selv it is thus the focus-generated set of alternative semantic values for sig (triggered by adnominal intensification) which gives rise to doppelganger-effects, i.e. statue-readings in wax-museum contexts. Note that a similar 37 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 of sig selv without statue-readings (e.g. (81b) read as (82a))? We suggest that in these cases the se/v adjunction simply serves other purposes, e.g. contrastive focus. see section 3.3.2.5. 168 proposal is found in Konig & Siemund ( 1999:48) 38 • In its unintensified/simple form sig behaves as a variable which has to be referentially and representationally identical to its binder. Thus the present approach achieves a (semantically and morphologically) fully compositional analysis of complex reflexives. That is, while the simple reflexive (e.g. Dan. sig) has to be corefential (and co-representational) with its antecedent, the complex reflexive (e.g. Dan. sig selv) may be either coreferential with its antecedent, or refer to a representation of the antecedent or the referent of the antecedent under a different representation 39 • 3.3.2.7.2. Qua-sentences The examples in (83) illustrate the motivation for analyzing dobbelganger-effects in terms of representational identity/non-identity rather than in terms of referential identity. (83) a. b. ? ? *Peter, working as barber, shaved. Peter, working as barber, shaved himself. 38 Discussing similar examples in Dutch, see (i) below, where the complex reflexive zich zelf '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 staan. 'She looked into a mirror and she saw herself standing in a comer.' b. Ze keek in een spiegle en ze zag zichzelfin een hoek staan. 'She looked into a mirror and she saw herSELF standing in a corner.' 39 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. 169 Imagine a context where a group of soldiers are lost in a forest for weeks. In order to maintain a certain level of hygiene the platoon leader picks Peter to function as camp barber and immediately orders him to shave everybody in the camp including himself. In this context, (83b) is clearly preferred over (83a). Similarly, in the corresponding Danish sentence in (84b) the complex reflexive sig selv is much better than the simple reflexive sig in (84a). (84) a. b. Peter, i egenskab afbarber, barberede ??*sig. Peter in quality of barber shaved REFL 'Peter, qua barber, shaved.' Peter. i egenskab afbarber, barberede sig selv. Peter in quality of barber shaved REFL self 'Peter, qua barber, shaved himself.' Now, in all the sentences in (83a-b) and (84a-b) there is clearly referential identity between the antecedent Peter and the reflexive pronoun. The qua-sentences thus differ from the statue- reading sentences, discussed in the previous section, where the relation between reflexive and antecedent does not involve referential identity but rather a relation between a real person (the referent of the antecedent) and a statue of this person (the referent of the reflexive). So if the dopppelganger-effect is defined as involving lack of referential identity between antecedent and reflexive (as in the case of (82b)) then the mandatory complex reflexive in qua-sentences would 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. Furthermore, at an intuitive level, the sentences in (83b) and (84b) are clearly felt to involve some kind of doppelganger of Peter. That is, the individual Peter is perceived of as occupying two roles at the same time. On the one hand, he is performing the duty of squad barber. On the other, he is the object of the shaving efforts of 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 of the arguments of the predicate (as was the case of statue-reading in wax museum contexts) qua-sentences involve only 170 representational non-identity. The referents of 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 similar to the sentences with anti-reflexive predicates discussed in section 3.3.2.1, e.g. (SOb) repeated here as (85). (85) Henry IV efterfulgte *sig I sig selv. Henry IV succeeded REFL I REFL self 'Peter succeeded himself(as king of England).' Strictly speaking the complex 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 of such an utterance would be to assume that reflexive and antecedent refer to different representations of the same individual, i.e. in this case Henry IV himself vs. Henry IV appearing disguised as the crown prince after feigning his own death. Analyzed 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 of anti-reflexive predicates, it is the meaning of the predicate which presupposes representational non-identity. Since both statue-readings and qua-sentences can be constructed with semantically 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 of statue-readings, representational non-identity is motivated by the wax-museum context, or similar contexts involving an individual and a representation (i.e. statue, painting, picture, photo, tape-recording, video- recording 40 , etc.) of this individual. In the case of qua-sentences, the representational non- identity is introduced by the word qua itself, or one of 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 of representational identity thus makes it possible to arrive at a unified 171 analysis of these three phenomena which have hitherto been considered to be unrelated. In the following section we argue that this analysis should also be extended to another phenomenon involving complex reflexives, i.e. the choice between strict and sloppy reading of 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 of reflexives in VP-ellipsis constructions is due to doppelganger-effects, i.e. it involves representational non-identity. The example in (86) illustrate the difference between simple and complex reflexives with respect to the availability of strict and sloppy reading in ellipsis constructions. (86) a. b. Peter vaskede sig og det gjorde Hans ogsa. Peter washed REFL and it did Hans also 'Peter washed (himself) and so did Peter.' Peter vaskede sig selv og det gjorde Hans ogsa. Peter washed REFL self and it did Hans also 'Peter washed himself and so did Peter.' Note that the same paradigm is found in English, see (87). (87) a Peter washed 0 and so did Hans. b. Peter washed 0 himself and so did Hans. (only sloppy) (strict and sloppy) (only sloppy) (both sloppy and strict) As shown in (86) the simple reflexive sig can only be given a sloppy reading while the complex reflexive sig selv can have both sloppy and strict readings. The sentences in (86) both involve two 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 same sentence, i.e. og det gjorde Hans ogsa, in both cases. The first proposition in both (86a) and (86a) involve referential identity of the arguments of the verb wash, they differ with respect to representational identity. The first conjoint in (86a), i.e. Peter vaskede sig, expresses a self- washing activity. In contrast due to the presence of the intensified selv 'self, the first conjoint 40 See Rooryck and van den Wyngaerd ( 1999) for a number of interesting examples in Dutch. 172 in (86b), i.e. Peter vaskede sig selv, implies that Peter washed somebody and that this somebody happened to be Peter. In this respect, the first conjoint in (86b) is similar to a qua- sentence, i.e. it involves a case of accidental identity of two participants in an activity. That is, in both (86b) Peter washes/shaves somebody who just happens to be himself. In other words, Peter is performing a shaving/washing action on his doppelganger, i.e. a representation of himself. Hence, the necessary presence of the intensifier selv. It has been claimed that long-distance sig has both strict and sloppy readings, see (88). (88) Peter; bad sygeplejersken vaske sig, og det gjorde Hans ogsa. Peter asked nurse-the wash REFL and it did Hans also 'Peter asked the nurse to wash himself, and so did Hans.' Reliable data on the availability of strict and sloppy readings of LD-sig is difficult to come by. Although (88) has been claimed to be ok with a strict reading of 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 make-up a context for (88) most informants assume a scenario in which Peter and Hans are patients in the same ward in a hospital who both ask the nurse to washes themselves. So far none of the informants that I have consulted provides a context in which Peter is the only patient and Hans is a relative who happens to be visiting Peter and who out of concern for his welfare asks the nurse to wash Peter. Furthermore, if forced to provide a sentence with a strict reading, then most speakers prefer the sentence in (89) with the pronoun ham 'him' instead of the reflexive. (89) Peter, bad 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 himself, and so did Hans.' 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 most speakers, preferred over reflexives). 173 Based on the above discussion of reflexives in VP-ellipsis constructions we conclude that the availability of strict readings of reflexives is an instance of the doppelganger-effect (i.e. (representational) non-identity) triggering adnominal intensification of the simple reflexive sig. 3.3.2.8 Complex reflexives and distributivity Complex reflexives have sometimes been claimed to disambiguate in favor of distributive readings, see (90a) vs. (90b). (90) a. Soldaterne forsvarede sig. soldiers-the defended REFL 'The soldiers defended themselves.' (collective/distributive) b. Soldaterneforsvarede sig selv. (distributive) soldiers-the defended REFL self 'The soldiers defended themselves.' However, in a contrastive context in which the soldiers as a group are contrasted with another group of individuals, the collective reading is also possible with complex reflexives, compare (91 b) and (91 b). (91) a. b. (?)Soldaterne forsvarede sig, men overladte civil befolkningen til jjenden. soldiers-the defended REFL but left civil population to enemy-the 'The soldiers defended but left the civilians to the enemy.' Soldaterne forsvarede sig selv, men overladte civil befolkningen til jjenden. soldiers-the defended REFL self but left civil population to enemy-the 'The soldiers defended themselves but left the civilians to the enemy.' The fact that the sentence in (9la) is slightly odd without the intensifier selv is due to the explicitly contrastive context which calls for intensification of the reflexive. The sentence in (91 b) clearly falsifies the claim that the complex reflexive sig selv must have a distributive reading, thus refuting the alleged direct link between distributivity and intensification of reflexives. Furthermore, it seems that in most cases the choice between collective vs. distributive readings of reflexives depends on factors other than intensification of the reflexive. See, for 174 example, the sentences in (92) which show that the distributive reading can be triggered by predicate meaning. (92) a. b. Soldaterne vaskede sig. soldiers-the washed REFL 'The soldiers washed (themselves).' Soldaterne vaskede sig selv. soldiers-the washed REFL self 'The soldiers washed themselves.' (distributive) (distributive) While (90a) the simple 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 between (90a) and (92a) follows from the meaning of the predicates. Soldiers are per definition assumed to work together under unified command to achieve a common goal, e.g. to defend or attack someone or something. It is therefore not surprising that in (90a) with the predicate forsvare 'defend' sig is most readily given a collective reading. In contrast, the default choice of a distributive reading of sig in (92a) follows from the meaning of the grooming verb vaske 'wash'. Under normal assumptions, washing (by showering or sponging) is a self-grooming activity which is performed by each individual on his or her own body. As a consequence, the collective reading of sig in (92a) is blocked by the relative difficulty of imagining a plausible scenario in which the soldiers engage in a collective grooming 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 of reflexives in these examples. The fact that the collective readings of complex reflexives are quite felicitous given the right context, see (91 b), shows that, unlike what has been claimed in the literature, intensification of reflexives does not necessarily trigger distributive readings. Furthermore the fact that the choice between distributive and collective readings of reflexives can be shown to vary from predicate to predicate, as in (90a) 175 vs. (92a), provides additional evidence that intensification of 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 of section 3.3.2 we have discussed the behavior of locally bound reflexives occurring as internal arguments of verbal predicates. According to our proposal that binding and intensification constitute separate modules of the grammar we predict the binding behavior of reflexives sig to follow from principle A as formulated in section 3.2, while intensification of reflexives (i.e. the distribution of the so-called complex reflexive sig selv) should follow from the principles of intensification, i.e. the principle of contrastiveness as formulated in chapter 2. We have shown that this approach to binding and intensification can successfully account for the distribution of simple and complex reflexives found in the contexts summarized in the descriptive generalizations in (93) and (94). (93) Simple reflexives (e.g. Dan. sig): a. Neutral predicates b. Inherently reflexive predicates (94) Complex reflexives (e.g. Dan. sigselv): a. anti-reflexive predicates b. hidden neutral predicates c. contrastive contexts (with neutral predicates) d. doppelganger-effects (with neutral predicates): (iv) statue-readings (v) qua-sentences (vi) strict (and sloppy) readings in VP ellipsis (section 3.3.2.3) (section 3.3.2.2) (section 3.3.2.1) (section 3.3.2.3) (section 3.3.2.5) (section 3.3.2.7) (section 3.3.2. 7. I) (section 3.3.2.7.2) (section 3.3.2.7.3) In the following sections the distribution of simple and complex 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 amenable to the same type of explanation. Section 3.3.6 presents a phonological account of the distribution of complex reflexives in prepositional predicates which explains intensification of reflexives with certain PPs a consequence of the inability of 176 simple reflexives (which behave phonologically like clitics) to carry stress on their own. Section 3.3.6 also contain discussion of the role of stressability of reflexives with respect to adnominal intensification in other languages (French, and German). 3.3.3 Reflexives in resultative constructions As shown in (95-98), resultatives can behave as either neutral (95), anti-reflexive (96), or inherently reflexive predicates (97-98), depending on the meaning of the words involved (cf. Veraart 1996). (95) Neutral resultative construction: a. Peter arbejdede sig b. Peter arbejdede sine forceldre c. Peter arbejdede sig selv Peter worked REFLIREFL self/his parents 'Peter worked himself/his parents rich.' rig. rige. rig. rich (96) Anti-reflexive resultative construction: a. Peter drak #sig under bordet. b. Peter drak Hans under bordet. c. Peter drak sig selv under bordet. Peter drank REFLIREFL self/Hans under table-the 'Peter drank himself/Hans under the table.' (97t Inherently reflexive resultative construction: a. Peter arbejdede sig svedig. b. Peter arbejdede #sine forceldre svedige. c. Peter arbejdede #sig selv svedig. Peter worked REFLIREFLself/his parents sweaty 'Peter worked himself/#his parents sweaty. 41 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 hiirdt. 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 hiirdt 'Peter got his employees to work hard.' 177 (98) Inherently reflexive resultative construction: a. Peter dansede sig til verdensmesterskabet. b. Peter dansede #sine foraddre til verdensmesterskabet. c. Peter dansede #sig selv til verdensmesterskabet. Peter danced REFLIREFLself/his parents to the world championship 'Peter danced himself/#his parents to the world championship.' Notice that the only difference between the sentences in (95) and (97) is the adjective denoting the end result of Peter's work, i.e. rig 'rich' vs. svedig 'sweaty'. In other words, simply replacing one with the other changes the whole resulative from a neutral (95) to an inherently reflexive (97) or vice versa. Likewise, (96) could change from "anti-reflexive" to "neutral" simply by replacing the phrase under bordet 'under the table' with the fuld 'drunk'. In the analysis proposed here the fact that changing the adjective in (95) and (96) can change the behavior of these resultative constructions can be explained as a change in the presuppositions triggered. The resultative construction drink somebody under the table in (96) is semantically 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 of Peter and the reflexive. Hence the neutral behavior shown in (96). In other words (95) and (97) are structurally identical, but differ semantically with respect to presupposed (representational) non-identity of reflexive and antecedent. Unlike purely syntactic accounts of binding, which have difficulties accounting for the different distribution of simple and complex reflexives in such cases, the account adnominal intensification proposed here is sensitive to semantic and pragmatic factors (i.e. presupposed representational non-identity) and, as a result, faces no such difficulties. Unless resultatives are analyzed as complex predicates these examples constitute a serious challenge for predicate-centered theories of binding like Reinhart & Reuland (1993). However, in the nominal account of binding proposed here nothing hinges on whether or not 178 resultatives are complex predicates. Indeed, since binding and intensification are independent of each other, binding theory has nothing to say about the distribution of the intensifier selv 'self which falls under the contrastiveness principle discussed in chapter 2. 3.3.4 Reflexives in ECM constructions As shown in (99-1 00), ECM constructions can behave as either neutral, see (99), or anti- reflexives predicates, see (I 00), depending on the meaning of the words involved. (99) Neutral ECM construction: a. Peter ansa sig for at were intelligent. b. Peter ansa sine forceldre for at vcere intelligente. c. Peter ansa sig selv for at vcere intelligent. Peter considered REFLIREFL self/his parents for to be intelligent 'Peter considered himself/his parents to be intelligent.' ( 1 00) 'Hidden' utral ECM construction: a. Peter ansa #sig for at vcere dod. b. Peter ansa sine forceldre for at vcere dode. c. Peter ansa sig selv for at vcere dod. Peter considered REFLIREFL self/his parents for to be dead 'Peter considered himself/his parents to be dead.' Considering somebody to be intelligent is a neutral activity which can be directed and oneself as well as others, see (99). In contrast, considering somebody 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 assume the existence of some kind of doppelganger, e.g. Peter's soul (liberated from his body at death), Peter's ghost roaming around after the physical death of Peter, etc., who considers the physical Peter to be deceased. Hence ( 1 00) behaves as an anti-reflexive or 'hidden' neutral with respect to intensification of reflexives. That is, the meaning of the ECM construction in ( 1 00) 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), which triggers adnominal intensification of the simple reflexive sig. 179 Once again, binding and intensification are independent of each other. Binding of sig falls under principle A which is satisfied in both (99) and (1 00) since the reflexive are bound by a subject (Peter) inside the minimal tensed clause. The distribution of selv 'self on the other hand, falls under the contrastiveness principles of the module of intensification and is thus directly sensitive to the presuppositions of identity triggered by the meaning of the words involved in the construction (e.g. intelligent 'intelligent', vs. dgd 'dead') and background encyclopedic knowledge 42 • Together with resultative constructions (discussed in section 3.3.3 above) and possessive construction (discussed in the following section), ECM constructions constitute potentially problematic cases for predicate-based approaches to binding (e.g. Reinhart and Reuland (1993)). Since the notions of anti-reflexivity, neutrality and inherent reflexivity invoked in the analysis proposed here do not rely on the notion of co-argumenthood but rather on the presence/absence ofpresuppositions of(representational) (non-)identity ofreflexive and antecedent it does not run into these problems. 3.3.5 Complex reflexives and pronouns in possessor position In this section we show that the distribution of simple and complex possessive reflexives, e.g. sin 'POSSREFL' and sin egen 'POSSREFL', follow the same general pattern as the distribution of reflexives in argument position (i.e. sig and sig selv). That is, the binding 42 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 sa sig/sig selv danse i spejlet. Peter saw REFLIREFL self dance in mirror-the "Peter saw himself dance in the mirror.' b. Peter sa #siglsig selv danse pa video-optagelsen. Peter saw REFLIREFL 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 of a video-recording (which thus trigger a presupposition of representational non-identity). 180 behavior of possessive reflexives (e.g. Dan. sin "POSSREFL') are accounted for by principle A as formulated in section 3.2, and the distribution of the intensifier egen 'own', follows the same principle of contrastiveness as its suppletive variant selv 'self. One of the main strengths of the analysis proposed here is thus that it can be straightforwardly extended to intensified possessive reflexives, which constitute a major problem for most predicate-based accounts, e.g. Reinhart & Reuland ( 1993). The Danish system of nominal expressions in possessor position is given in ( 101 ). Unlike English, Danish has a possessive reflexive, i.e. sin 'POSSREFL!his/her/one's', in addition to the possessive pronouns hans 'his'. (I 01 ) 43 Danish nominal expressions in possessor position: Simple/unintensified Complex/intensified a. Reflexive sin 'his/her/one's' sin egen 'his/her/one's own' b. Pronoun hans 'his' hans egen 'his own' c. DP kongens 'the king's' kongens egen 'the king's own' Except for the suppletive form of the intensifier, see (1 03), and the different case forms (i.e. assuming sin 'POSSREFL' to be the genitive form of sig 'REFL') this system is exactly the same as the ones found in argument position, compare (101) and (102). (I 02) Danish nominal expressions in argument position: Simple/unintensified Complex/intensified a. Reflexive sig 'REFL' sig selv 'REFL self/himself b. Pronoun ham 'him' ham selv 'him himself/himself c. DP kongen 'the king' kongen selv 'the king himself 43 The full paradigm of possessive reflexives and pronouns can be found in Appendix II. (I 03) 44 Suppletive variants of intensifier: a. Argument position: b. Possessor position: selv egen 181 'himself 'own' Like their argument position counterparts (i.e. sig 'him, her, one' and ham 'him') the distribution of sin and hans are constrained by the principles A and B of the binding theory. Sin (like sig) is a reflexive and must be bound by a subject in a local domain and hans 'his' (like ham 'him') is a pronoun and must be free in a local domain, i.e. the minimal tensed clause, see (I 04a-d). (104) a. Peter, sagde at Johnk vaskede 'Peter said that John washed sin., k *z I hans, *k z POSSREFLI his tegnebog. wallet.' b. Peter, sagde at Johnk vaskede sin.; k *z egen I hans, *k *= egen tegnebog. 'Peter said that John vaskede POSSREFL own/ his own wallet.' c. Peter, sagde at Johnk stjal sin., '*k *z I hans, *k z tegnebog. d. 'Peter said that John stole POSSREFLI his wallet.' Peter; sagde at Johnk stjal 'Peter said that John stole sin., k *z egen I hans, *k *z egen tegnebog. POSSREFL own/ his own wallet.' The sentences in (I 05) illustrate LD-binding of the possessive reflexive sin follow the same pattern as LD-binding of sig 'REFL', i.e. its antecedent can be found outside infinitival clauses. (109) a. Peter, bad Hansk vaske sin; k hund. Peter asked Hans wash POSSREFL dog 'Peteri asked Hansi to wash hi silk dog.' 44 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, landomsfulde Hvisken klinge frem i Virkelighedens Lufl, sam vildt udfordrende Lurtoner, sam Brag af Kalleslag paa Tempelmure, sam Hvin af Davidsstene paa Flugt mod Goliathspander og sam sejerssikker Fanfare. Del var sam at hare sig selv tale i fremmede Tunger, medfremmed 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 find 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. 182 b. Peter lad farceldrene save i sin egen seng mens han selv sav pa safaen is stuen. Peter let parents-the sleep in POSSREFL own bed while he self slept on sofa the in living room-the 'Peter let his parents sleep in his own bed while he himself slept on the sofa in the living room.' Though LD-binding of the simple possessive reflexive sin is more frequently found, LD- binding of intensified possessive reflexives is by no means excluded, i.e. given an appropriate context that satisfies the contrastiveness principle of adnominal intensification, see (105b). That is the adjunction of the intensifier egen 'own' does not affect the LD-potential of the possessive reflexive sin. In other words, just like in the case of sig, see section 3.3.1, intensification of possessive reflexives does not directly affect locality constraints 45 • The examples in (I 06-1 08) illustrate that Danish possessive reflexives sin and sin egen exhibit the same overall distributional pattern as the argument reflexives sig and sig selv with respect to "neutral", "anti-reflexive" and "inherently reflexive" constructions/sentences 46 • 45 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 Association, are encouraging people= ez to buy theiri =houses. b. The Housing Association, are encouraging people= ez 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 of his own is pragmatically more acceptable more, see (ic). The importance of 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 of 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 himself whose distribution is controlled by the principles of the module of intensification). The minimal pair (ib,c) also shows that the distribution of the intensifier own is sensitive to semantic/pragmatic factors. See also chapter 5 for more discussion of the analysis of possessives in Modern English. 46 Examples involving I'' person possessives can also be used to illustrate that intensifier adjunction is triggered by something else than a need tor disambiguation, e.g.: (i) Je suis *mon!mon propre ennemi. ·1 am my own enemy.' (ii) I am *my/my own enemy. (iii) Jeg er *min/min egenjjende. (iv) Wo shi *wo I wo ziji de di-ren. 'I am my own enemy.' 'I am my own enemy.' (Fr.) (Dan.) (Ch.) (I 06) "Neutral" constructions: a. Peter, vasker sin /sin egen I John's tegnebog. b. Peter washes POSREF /POSREF own/ John's wallet. 'Peter washes his /his own I John's wallet.' Peter hader Peter hates sin /sin egen I John 's mar. POSREF /POSREF own/ John's mother. 'Peter hates his I his own I John's mother.' (I 07) "Anti-reflexive" constructions: a. Peter er *sin /sin egen I John's .ffende. Peter is *POSREF /POSREFL own/ John's enemy. 'Peter is his I his own I his enemy.' Peter stjal *??sin I sin egen I John's Peter stole *??POSREF/ POSSREF own I John's 'Peter stole his I his own I John's wallet.' (I 08) "Inherently reflexive" constructions: tegnebog. wallet. a. Han varved at gaa ud afsit /?*sit eget/*Peters gode skind afgkede. he was about to go out ofPOSREF/?*POSREF own/Peter's good skin of happiness 'He nearly jumped out of his good skin of sheer happiness.' b. Han hyttede sit I ??sit eget /*Peters skin. (cont. bet. alb) he saved POSREF/ ??POSREF own/*Peter's skin 'He saved his own life.' 183 The importance of the above examples involving Danish reflexive possessives sin and sin egen is that they show that disambiguation is not the prime motivation for intensifier adjunction to possessives. In sentences like the ones in (I 04) with two potential antecedent for the possessive sin only the local antecedent can bind sin. These sentences (unlike the English translations thereof) are not ambiguous; sin can only have John (the local subject) as its antecedent since it is an anaphor which 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 disambiguate between multiple potential antecedents which motivates the presence of egen 'own' on sin in the "anti-reflexive" constructions in (107) it has to be See chapter 4 for more discussion of I'' and znd person pronouns and reflexives. 47 Note that adding the particle back saves the sentences in (ia), see (ib): (i) a. *He, stole his, wallet. (simple predicate: steal_) b. He; stole his, wallet back. (complex predicate:sleal __ back 184 something else 48 . As the contrast between (104a,b), (106) and (104c,d), (107) shows it is the semantic/pragmatic make-up of the sentence (i.e. the presupposed or expected (representational) non-identity of the reflexive and its antecedent 49 ) which triggers intensifier- adjunction to sin. Based on the meaning of the predicates vaske 'wash' and stjcele 'steal' and world knowledge, native speakers know that for a given individual x, the proposition x washes x 's wallet is pragmatically OK. The proposition x steals x 's wallet, however, is pragmatically odd; stealing is per definition an "anti-reflexive" activity. The meaning of the verb steal presupposes that the stealer does not originally possess the stealee. Indeed, anti-reflexivity, i.e. presupposed representational non-identity of the stealer and the original owner of the stolen goods, seems to be an integral part of the meaning of the predicate stjcele 'steal', i.e. given normal circumstance the expectation is that people would not steal their own things. The examples in (I 08) show that Danish also have "inherently reflexive" possessive constructions (usually of idiomatic nature) which disallow all but the simple unintensified reflexive possessive sin. We can therefore conclude that the distribution of Danish simple and complex possessives (sin and sin egen) in "neutral", "anti-reflexive" and "inherently reflexive" predicates follow the same general pattern as the simple and complex argument reflexives sig and sig selv. In this section we have shown that the same generalizations hold for both adnominal intensification of reflexives in argument position (e.g. sig selv 'REFL self) and adnominal intensification of reflexives in possessor position (e.g. sin egen 'POSSREFL own'). This constitutes a major obstacle for analyses of binding which are based on the notions of co- The sentences in (i) thus indicate that some kind of analysis of resultatives as complex predicates is needed. 48 For the example in (Ill b) and similar examples an structural i-within-i filter explanation has been proposed, see Vergnaud and Zubizarreta (1992). 49 If retlexivity 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 of retlexivity. "Retlexivity" is here used more loosely to refer to self-directed predicates/constructions in a broader sense. 185 argumenthood, e.g. Reinhart and Reuland (1993). It seems rather implausible to maintain that the difference between (106a) and (108b) should derive from the a lexical feature (e.g. Reinhart and Reuland's [+/-inherently reflexive], or Zribi-Herts's [+/- disjoint reference]) distinguishing between the predicate vaske 'wash' and the predicate stjcele 'steal'. The structural relationship between the reflexives (i.e. sin and sin egen) and their antecedent has to be assumed to be the same all the sentences in (106-108). And yet, the sentences differ as to whether they allow the simple possessive reflexive or not. While this difference can be accounted for in terms of focus and presuppositions as illustrated, it seems impossible to account for it in terms of any kind of predicate-centered 'reanalysis' or 'lexical ambiguity' account which only applies to co-argument binding 50 • 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 placement - interact with the distribution of intensifier-adjunction to the simple reflexive sig. We argue that the focus-based analysis of intensifier-adjunction proposed here stands a better chance of giving a satisfying account of the nature of this interaction between phonology and syntax than competing analyses of complex reflexives 51 • The rest of this section will be organized as follows. First, in section 3.3.6.1, the clitic-like properties of the simple reflexive sig (e.g. the fact that it cannot be stressed) are illustrated using a number of syntactic tests. Then, in section 3.3.6.2, it is shown that the 50 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. 51 Cf. Siemund (2000). 186 idiosyncrasies of the distribution of simple and complex 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 of reflexives will be discussed as a potential explanation of cross-linguistic variation in intensification of reflexives 52 • 3.3.6.1. Stressability of reflexives: prosodic uses of selv = stress carrying element In this section a set of syntactic tests are used to show that simple reflexive sig is a phonologically 'light' clitic-like element which cannot carry word-stress on its own and therefore requires intensifier-adjunction in order to be able carry stress like a full lexical DP. The different behaviors of the simple and complex reflexives in Danish are summarized in (I 09). Unlike the complex reflexive anaphor sig selv, the simple reflexive anaphor sig is a clitic that cannot be stressed, used to answer questions, coordinated, clefted, or topicalized. (I 09) Comparing simple sig and complex sig selv: Tests\ expressions sig sig selv I DP I ham I ham selv a. Stress NO YES (see (110)) b. Answer to questions NO YES (see (111)) c. Coordination NO YES (see (112)) d. Clefting NO YES (see (113)) e. Topicalization NO YES (see (114)) The sentences in (110)-(114) illustrate the differences between sig and sig selv with respect to the properties in ( 1 09) 53 • ( 11 0) Stess: Peter bad Hans om at vaske*SIG I SIG SELV/ KONGEN. Peter asked Hans about to wash REFLIREFL SELF/ KING-the 'Peter asked Hans to wash *REFLIREFL SELF/the KING.' 52 Cf. Siemund (2000). 53 The tests in (109) and the examples in (110-4) are based on .lacubowicz (1994) who in tum was inspired by Kayne (1975). ( 111) Answering questions: Q: Hvem vasker Peter? A: *Sig I sig selv I Marie Who washes Peter REFL/REFL self/Marry 'Who does Peter wash?' '*REFL I REFL SELF I Mary' ( 112) Coordination: Peter vaskede barnet og *sig I sig selv I hunden. Peter washed child-the and REFL /REFL self /dog-the 'Peter washed the child and *REFL I REFL SELF I the dog.' (113) Clefts: Det var *sig I sig selv I hunden Peter vaskede. it was REFL I REFL self I dog-the Peter washed 'It was *REFL I REFL SELF I the dog that Peter washed.' ( 114) Topicalization: *Sig I sig selv I hunden vaskede Peter ikke. REFL I REFL self I dog-the washed Peter not '*REFL I REFL SELF I the dog Peter didn't wash.' 187 Notice that the verbal predicate vaske 'wash' which is used m the above examples is a semantically neutral predicate which allows the simple reflexive anaphor sig as its internal argument, e.g. Peter vasker sig 'Peter washes (himself)'. The mandatory presence of selv in these examples thus cannot be due to any semantically "anti-reflexive" nature of the d . 54 pre 1cate . We therefore conclude that intensification of sig in these cases must be prosodically motivated, i.e. it must be due to a need to make the clitic sig heavy enough prosodically to receive stress. In this section, we have established that intensifier-adjunction to sig interacts with prosodic factors like stress. Being a phonologically 'light' clitic-like element ( cf. Zwicky 1977) sig cannot carry stress 5 5 • Therefore, intensifier-adjunction is necessary to make the 54 The French reflexive soi 'REFL', which can both be stressed (i.e. SO[) and intensified (i.e soi-meme), illustrates that inability to be stressed is not a universal property of simple reflexives. German sich and Russian sebja are also both stressable. In contrast, Russian ~sja and English @-reflexives are both unstressable, just like Danish sig. 55 There is one idiomatic expression which contains an instance of stressed, simple/unintensified sig, see (i). (i) Peter er ikke sam a//e de andre bern. Haner 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.' 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 of complex reflexives in certain types ofprepositional predicates. 3.3.6.2. Intensifier-adjunction to reflexives in PPs and the interaction with sentential stress The main purpose of this section is to show that in addition to semantic factors (i.e. anti- reflexivity and inherent reflexivity defined in terms of presupposition of identity or (representational) non-identity), prosodic factors (i.e. stress-placement) also plays an important role in determining when reflexives must be adnominally intensified. The claim is that a number of exceptions to the semantic account of intensification of reflexives can be explained as following from special prosodic properties of certain prepositional constructions. Distributionally, most prepositional predicates in Danish exclude the use of the simple reflexive sig while allowing 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/at' c. drgmme om 'dream about' d. lytte til 'listen to' e. sigte pa 'aim at' f. Hans peger pa *sig I sig selv I bilen. 'Hans points at *0 I 0 himself I the car.' At this point we have no account of this fact. Note, however, that some speakers prefer sig selv over stressed S/G in (i). (116) Prepositional nouns: a. et billede af 'a picture of b. en beskrivelse af 'a description of c. en hyldeste til 'a tribute to, an homage to' d. en bog om 'a book about' e. en evaluering af 'an evaluation of f. . ... en evaluering af *sig I sig selv I bilen. ' .... an evaluation of *0 I 0 himself I the car.' ( 117) Prepositional adjectives: a. sikker pa 'sure of b. interersseret i 'interested in' c. optaget af 'occupied by' d. tilfreds med 'satisfied with' e. glad.for 'pleased with, like' f. forskelligfra 'different from' g Peter er tilfreds med *sig I sig selv I bilen. 'Peter is satisfied with *0 I 0 himself I the car' 189 While prepositional predicates pattern with anti-reflexive predicates distributionally, i.e. with respect to intensifier-adjunction to the simple reflexive sig, they do not necessarily all have anti-reflexive meanings which are logically incompatible (or even implausible) with reflexive scenarios. Since Danish sig 'REFL' cannot be stressed, as shown in section 3.3.6.1, we have to look for a language with stressable reflexives to see if the distinction between semantically anti-reflexive and neutral predicates can hold for prepositional predicates. French is such a language, since it has the prosodically strong reflexive soi 'REFL' which can host stress on its own. And as shown in (118) and ( 119), French reflexives in PPs may have both the intensified and the unintensified form depending on the meaning of the predicate, see (118) vs. (119). That is, French prepositional predicates can be classified as either anti-reflexive, neutral or inherently reflexive based on semantic/pragmatic criteria. ( 118) Semantically neutral prepositional predicates (French): a. Personne n 'est fier de soilsoi-memeiMarie. person NE is proud ofREFLIREFL-same/Marie 'Nobody is proud ofhimself/Marie.' b. Personne n 'est content de soilsoi-memeiMarie. person NE is satisfied of REFLIREFL-same/Marie 'Nobody is happy with himself/Marie.' (119) Semantically anti-reflexive prepositional predicates (French): a. Personne n 'estjalou:x de *soi/soi-meme/Marie. person NE is jealous of REFLIREFL-same/Marie 'Nobody is jealous of himself/Marie.' b. Personne ne bavarde avec *soi/soi-meme/Marie. person NE chats with REFLIREFL-same/Marie 'Nobody is chatting with himself/Marie.' ( 120) Semantically inherently reflexive prepositional predicates (French): Personne n 'est hors de soil*soi-memei*Marie defureur. person NE is out of REFLIREFL-same/Marie of fury 'Nobody is beside REFLI*REFL self /*Mary with rage.' 190 The predicates etre fier de and etre content de, in ( 118), are semantically neutral in that they are compatible with both reflexive and non-reflexive scenarios. In contrast, the predicates etre jalou:x de and bavarder avec, in (119), are semantically anti-reflexive in that they are compatible only with non-reflexive scenarios and thus trigger intensifier-adjunction to the simple reflexive soi 'REFL' 56 • Finally, the predicate in (120) presuppose identity of its arguments and is thus inherently reflexive. Since Danish also uses intensifier-adjunction to license reflexive scenarios with predicates whose meaning is logically incompatible (or pragmatically implausible) with reflexive scenarios, as shown in section 3.3.2, one would assume that semantically neutral prepositional predicates corresponding to the French predicates in ( 118) would allow both simple and complex reflexives. This prediction, however, does not hold. As shown in (121- 122), in Danish prepositional predicates, regardless of meaning, need to have a complex reflexive. (121) Semantically neutral prepositional predicates: a. Peter er stolt af*sig/sig selv/Marie. Peter is proud of REFLIREFL self/Maire Peter is proud of himself/Marie.' b. Peter er tilfreds med *sig/sig selv/Marie. Peter is satisfied ofREFLIREFL self/Marie 'Peter is happy with himself/Marie.' 56 This observation is from Zribi-Hertz ( 1995) who invokes a [+/-disjoint reference] feature to capture the difference between PPs like fier de 'proud or vs. jaloux de 'jealous or. ( 122) Semantically anti-reflexive prepositional predicates a. Peter er misundelig pa *sig/sig selv/Marie. Peter is jealous on REFLIREFL self/Maire 'Peter is jealous ofhimself/Marie.' b. Peter sludrer med *sig/sig selv!Marie. Peter chats with REFLIREFL self/Marie 'Peter is chatting with himself/Marie.' 191 Meaning-wise one would expect the predicates vcere stolt af 'be proud of and tilfreds med 'satisfied with' to behave like their French counterparts in (118), but in spite oftheir "neutral" meanings they still behave distributionally like anti-reflexives as shown in (121a,b). So if it is not the semantic anti-reflexivity of the predicates which triggers selv-adjunction to sig in ( 121 a,b ), then what it it? We argue that it is prosodic factors. Simple/unintensified sig is a non-stressable element which 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 of simple reflexives is necessary to make the reflexive able to carry non- contrastive sentence stress which is required on the complement of such prepositional predicates. Hence, only sig selv and DPs are allowed with the prepositional predicates in ( 115- 117). As mentioned above, many of the prepositional predicates in Danish are of the type illustrated in (115-117), which, regardless of their meaning, do not allow simple sig. There are, however, a number of prepositional constructions which allow simple sig. They can be divided into those which only allow simple sig, and those which allow both simple and complex reflexives as well as DPs. Let us first consider the prepositional expressions in (123- 124). ( 123) 57 Prepositional constructions allowing only simple sig: a. at were bange af sig to be afraid of REFL 'to be (naturally) timid' 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 without 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 were om sig to be about REFL 'be enterprising, be busy; have an eye for opportunities' (124) Peter er bange afsigl *sigselv/ *Mary. Peter is afraid of REFL I REFL self I Mary 'Peter is (naturally) timid' 192 The example sentences in (124) illustrate how the expressions listed in (123) behave, i.e. allowing simple 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 meanings of the predicates in ( 115) are clearly formed compositionally from the meaning of the verb plus the meaning of the preposition. Take for instance, tale til 'speak to' which is straightforwardly composed of the verb tale 'speak' plus the preposition til 'to'. As in German and other related languages, many of the predicates ofthe type illustrated in (115) have alternative forms where the preposition is fused with the verb, e.g. tiltale 'address someone, speak to someone' 58 • In contrast, the predicates in (123) are of a more idiomatic nature. In many cases, the meaning of 57 Note that out of the 6 predicates in (123). only (123b) has a reflexive (i.e. onese!fJ 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. 193 the whole expression cannot be deduced compositionally from its component parts, e.g. it is not immediately obvious how (123t) were 'be' +om 'about' comes to mean 'be active; have an eye for opportunities'. In other cases, however, the meaning is more transparent, e.g. ( 123b) have en revoler pa sig 'have a gun on oneself. Finally, none ofthe expressions in (123) have variant forms of the tilt ale 'to-speak, address' vs. tale til 'speak to, address' type which, as discussed above, is frequently found with the expressions in ( 115). That is, compounds such as *omvcere 'about-be', *frabide 'from-bite' are all impossible. In addition to prepositional constructions allowing only simple sig, Danish also has a series of prepositional expressions which allow simple and complex reflexives as well as DPs, see (125). (125) Prepositional constructions which allow both simple and complex reflexives as well as full lexical DPs: a. Peter sa en slange bag sig I sig selv /Mary. Peter saw a snake behind REFL I REFL self /Mary 'Peter saw a snake behind himself I Mary' b. Peter sa en slange foran sig I sig selv /Mary. Peter saw a snake before REFL I REFL self /Mary 'Peter saw a snake in front of himself I Mary' Notice that the prepositions found in (125) are different from both (115-7) and (123). First of all, they are not part of the verbal/adjectival/nominal predicate but function as locative and directional adverbials. Second, the prepositions involved, e.g.foran 'in front of, bag 'behind', over 'over', etc., all assign theta-roles on their own. In contrast, in the predicates in ( 115-117) it is the verbal part of the predicates which assign theta-roles while the prepositions merely 58 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) 194 function as case-assigners. As for the expressions in (123) the thematic status of sig is debatable 59 • Even if considered to be thematic, the simple sig in ( 123) cannot plausibly be claimed to received a theta-role form the verbal component of the expression. Hence, if the instances of sig in ( 123) have a theta-role, it must come from the preposition itself. Finally, in addition to being morphologically more complex, the prepositions in ( 125) are all prosodically strong enough to carry word stress on their own, and, even more importantly, have to be stressed. As discussed below, this difference in stressability will turn out to be crucial. The distribution of different types of prepositions among the three distributional classes of prepositional predicates is summarized in ( 126). (126) a. b. c. Prepositions constructions: Type (115-117): *sig I sig selv I DP Type (123): sig I *sig selv I *DP Type (125): sig I sig selv I DP Prepositions afr-stress] 'or om[-stress] 'about' pa[-stress] 'on' fra[-stress] 'from' i[-stress] 'in' med[-stress] 'with' for(-stress] 'for'' etc. aj[TS!ress] 'of om[i-S(feSS] pd[+stress] 'on' fra(+stress] i[1-stress] bagl •stress 1 foran(+stress] over(+stress] 'about' 'from' 'in', etc. 'behind' 'in front or 'over', etc. As shown in (126), the prepositions found in (115-117)/(126a) also appear in well-formed constructions with the simple reflexive sig (123)/(126b). However, at closer inspection a very important difference between the constructions in (115-117) vs. the constructions in (123) becomes apparent. In the case of the complex predicates in ( 115-117), non-contrastive sentence stress always falls on the main verb/noun/adjective rather than the preposition, which when separated, due to prosodic factors. However, so far we have not been able to find an example. 195 is always unstressed, see (127a), where stressed syllables are marked in bold. In contrast, in the expressions in ( 123), it is the preposition which is unstressed, thus allowing the reflexive to be unstressed, see ( 12 7b ). (127) a. b. Peter pegede Peter pointed 'Peter pointed Jeg havde en revolver I had a gun 'I was carrying a gun.' pa mig. (type illustrated in ( 115-7)) on me at me.' pli mig. (type illustrated in (123)) on me The reason why in (127) first person pronoun/reflexive mig 'me' is used instead of third person forms, is that mig 'me' is stressable while sig 'REFL' is not, thus making it possible to construct a paradigm as in (127) where the only variable is stress placement. Since we know from section 3.3.6.1 that sig is not stressable and thus has to be adnominally intensified in order to carry stress, we can now explain some of the differences between the prepositional constructions in (115-7) vs. (123) as following from the differences in stress-patterns, see (128) and (129). (128) a. (129) b. c. a. b. c. Peter pegede Peter pointed on 'Peter pointed at himself/Marie.' Peter havde en revolver pli Peter had a gun on 'Peter was carrying a gun.' sig oksig selv okMarie REFLIREFL self/Marie ok · szg sig selv Marie REFLIREFL self The unacceptability of ( 128a) follows directly from the fact that sig cannot carry stress on its own. Hence intensification of sig is necessary, as shown in (128b). Furthermore the fact that simple (unstressed) sig is ok in (129a) follows from the fact that in this case the prepositionpa is stressed, and as a consequence, able to host the clitic sig. 59 Cf. Vikner (1985) who claim that the sig in bange af sig 'be timid" and similar "inherently reflexive·· 196 While (128a-b) and (129a) can be explained by reference to the stress patterns, the unacceptability of (129b) cannot. In this case, sig selv should be ok. There is no prosodic reason why sig selv should not be allowed here. Based on what the expressions in (123b)/(129) mean we suggest that it is their semantically "inherently reflexive" nature (i.e. the presence of a presupposition of (representational) identity of sig and its antecedent) which is responsible for the unacceptability of (129b). That the expression in (129) is semantically inherently reflexive is confirmed by the unacceptability of (129c). Notice that even with unstressed pa 'on' and stressed Marie, (129c) would still be unacceptable. Hence, the unacceptability of (129b) follows from the Constrastiveness Condition on adnominal intensification. Based on the above examples we conclude that the presence/absence of intensification of sig in prepositional constructions is triggered by either semantic factors, e.g. mandatory absence of intensification in the "inherently reflexive" expressions in ( 123), ( 129b ), and ( 131 b), or prosodic factors (e.g. mandatory presence of intensification in the distributionally, but not necessarily semantically, "anti-reflexive" expressions in (115-7), (121), (128) and (130). (130) Semantically neutral expression with stress-determined preposition (-stress): a. Peter tog et billede *af sig (prosodic) b. af sig selv c. af barnet Peter took a photo of REFL self/child-the 'Peter took a photo of himself/the child.' ( 131) Semantically inherently reflexive expression: a. Peter er bange af b. *af c. *af Peter is afraid of 'Peter is afraid of himself/the child.' prepositional expressions is non-thematic. sig. sig selv barnet REFL (semantic) (semantic) 197 In ( 130-131) stress is marked in bold. Behind each unacceptable example the type of violation is indicated in parenthesis. So far all of the prepositional predicates involving the prepositions listed in (126a-bt 0 which allow simple sig, i.e. (123) and (129), have been (or at least have meanings which can plausibly be argued to be) semantically inherently reflexive 61 (i.e. they presuppose the (representational) identity of the referents of sig and its antecedent). There are however a small number (/at least one) semantically neutral prepositional predicates involving the prepositions in (126a-b ), which allow simple sig, in addition to sig selv and DPs, see ( 132) where stress is marked in bold and reasons for unacceptability indicated in parenthesis following the example. (132) Semantically neutral exQression with stress-undetermined QreQosition 62 : a. Peter tog tojet af sig. b. *af sig (prosodic) c. *af sig (prosodic) d. *af sig (prosodic) e. *af sig selv (prosodic) f. af sig selv g *af sig selv (prosodic) h. *af sig selv (prosodic) I. *af barnet (prosodic) J. af barnet k. *af barnet (prosodic) I. *af barnet (prosodic) Peter took clothes of REFLIREFL self/child-the 'Peter undressed (himself)/the child.' In ( 132) all the logically possible combinations of stress and absence of stress on both the preposition af 'of and its complement have been spelled out. Only, three combinations, i.e. ( 132a,f,j) are acceptable. The rest are unacceptable due to violations of prosodic well- 60 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. 61 Or, alternatively, to be deponent/unaccusative verbs which are constructed with non-thematic sig, see chapter 4, section 4.3. 62 There appear to be very few examples of neutral constructions of this type. all involving sig as indirect object. 198 formedness constraints 63 • The most interesting aspect of these examples is the variable nature of the preposition af'of. With the simple 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,h,i,k,l). Whether af 'of is stressed or not depends on what type of expression follows: af 'of is stressed when followed by an unstresseable element like sig, and stressed when followed by a stressed expression. Since both the complex reflexive sig selv and DPs have word-stress, af 'of is unstressed in front of these expressions (132f,j). This generalization is further confirmed by the behavior of af 'of in the expression in ( 132) when found with pronominal complements, see ( 133). ( 133) Semantically neutral expression with stress-undetermined preposition: a. Sygeplejersken tog tgjet af ham. b. af ham c. *af ham d. *af ham nurse-the took clothes of him 'The nurse undressed him/him.' The personal pronouns displays a dual behavior with respect to stress. On the one hand they can be unstressed (like simple sig). In this case, the preceding preposition af 'of must be stressed, see ( 133a) vs. ( 133d). On the other hand, they can also carry stress, e.g. when contrastively focused. When that happens, the preposition aj'or is stressless, see (133b) vs. ( 133c ). Examples such as the above amply illustrate the fact the absence/presence of stress on the prepositions in ( 126a-b) vary depending on the following expression 64 • 63 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 afsig selv, ikke med sig selv. He took clothes-the ofREFL self not with REFL self ·He undressed rather than bringing the clothes' 64 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: 199 Now let us return to the prepositions in (126c). As the examples in (134) illustrate these prepositions can also be found in neutral expressions, see also (125). (134t 5 Neutral predicate with mandatorily stressed preposition: a. Han lagde uret bag sig. b. *bag sig c. *bag sig d. *bag sig ( i) Group I 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 fra 'jj;rtarnet 'fra det 'from the lighthouse' i 'Iommen 'i den 'in the pocket' med 'b(Jrnene 'med dem 'with the children' pa 'bordet 'pa det ·on the table' til 'dig 'til dig 'to you· (ii) Group II consists of the following prepositions: bag, efler, foran, forbi, for, (i)gennem, (i)mod, (i)mellem, inden, indtil, tangs, omkring, over (for), siden, uden, under. These are either stressed or unstressed when their complement is stressed: Stressed Unstressed complement complement (')efler 'vatget 'efler det 'after the election' (')for 'krigen 'for den 'before the war' (')fangs 'vejen 'tangs den ·along the road' (')over 'doren 'over den ·over the door' (')uden '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 'efler ogfandt ogsa et egetra:sbord. 'I looked for and found and oak table.' (ii) The prepositional complementary be fronted, leaving the preposition 'stranded' m 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, somjeg ser pa. 'She is the one I'm looking at,' [etc...]" 65 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 mandatorily stressed preposition: a. Hun lagde uret bag ham. b. *bag ham c. bag ham d. *bag ham she put watch-the behind him 'She put the watch behind him/him.' e. f. g h. (*)bag *bag bag *bag sig selv sig selv sig selv sig selv 1. *bag barnet j. *bag barnet k. bag barnet I. *bag barnet he put watch-the behind REFLIREFL self/child-the 'The put the watch behind himself/the child.' 200 The examples in ( 135) illustrate these prepositions can also be found with inherently reflexive expressions. ( 135) 66 Semantically inherently reflexive construction with mandatorily stressed preposition: a. Hun kunne ikke lcegge det bag sig. b. #bag sig selv. c. #bag barnet. she could not put it behind REFLIREFL self/child-the 'She couldn't leave it behind/she couldn't put it behind her(self).' The unacceptability of (135b-c) must be due to the inherently reflexive semantics of the expression lcegge noget bag_ 'put something behind_' rather than the prosodic properties of the preposition bag 'behind'. Unlike the preposition aj'of in the neutral prepositional expression tage t0jet af_ 'take clothes-the of_' illustrated in (132), the preposition bag 'behind' in the neutral expression lcegge uret bag _ 'put watch-the behind _' in (134) and the inherently reflexive predicate in (135) cannot be completely destressed, see (134f,j) vs. (132f,j). We attribute this difference between the prepositions in (126a-b) and (126c) to a difference in syllabic/moraic structure. The prepositions in ( 126a-b) are all mono-syllabic expressions with the syllable structure (C)(C)V, which means that they are all mono-moraic. In contrast, the 66 Notice that the English expression put something behind _ (with the meaning •forget') is also inherently reflexive. see (i): (i) Peter, couldn't put it behind hirn/*hirn/*hirnself/*Mary. See chapter 5 for more discussion of such expressions in English. 201 prepositions in (126c) are bi- or even poly-syllabic with the minimal structure (C)VC(V), which means that they are bi- or poly-moraic, see ( 136). (136) a. b. C. d. e. Syllabic structure Moras Stress-underspecitled prepositions cv VS. Qualify as phonological word Examples Stress I NO ( 126a-b) - 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 of what follows The behavior of different types of prepositional predicates discussed above is summarized in the table in (137). (137) Type I (a) I (b) I (c) II (a) II (b) Ill (a) Ill (b) Pregositional gredicates: (i) (ii) Semantics th-role tr. ofpredic. anti-ref n/hidd.n. inh. ref inh (+th) inh (-th) neut. neut. inh. ref v./adj/n. + + (iii) th-role tr prep. (-/+) + + + (iv) (v) prepos. distribution ofselv +/-stress *siglsig selv!DP *sig/sig selv!DP *sig/.5ig selvi*DP undersp. sigl*sig se/vi*DP undersp. sigl*sig se/vl* DP undersp. siglsig se/v!DP + siglsig se/v!DP + sig/*sig selvi*DP (vi) explanation of [+1-se/v] prosody (/semantic) prosody prosody semantic semantic semantic semantic semantic (vii) prepos. afpa afpa af afpa afpa af (viii) example(s) (122) (121) (138) (123b) ( 123a,c-f) (132) bag.foran (134) bag.foran ( 135) As shown in ( 137) we divide prepositional predicates into three types depending on whether or not the preposition is stressed. Predicates of type I, e.g. ( 115-7), have prepositions which are always unstressed 67 • Semantically these predicates can be either anti-reflexive (sub-type Ia), neutral (sub-type lb), or inherently reflexive (sub-type Ic). Examples of anti-reflexive and neutral predicates of this types were given in (122) and (121) respectively. Inherently reflexive prepositional constructions oftype (137(Ic)) are illustrated in (138). 67 The reason why the prepositions af, pa, etc. ·o[ on. etc.' are always unstressed in I( a-c) but variable in II( 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. ·of, on, etc.'. We leave the question why this should be the case for further research. (138) Peter er ude af*sig/sig selv/*Mary af glcede. Peter is out ofREFLIREFL self/Mary ofhappiness 'Peter is beside himself/*Mary with happiness.' 202 The French example (due to Zribi-Hertz (1995)) in (139) (:=:::;(120)), shows that the stressable reflexive soi 'REFL' can, and must (due to semantic constraints, i.e. meaning of predicate plus contrastiveness condition on adnominal intensification) be unintensified in the semantically inherently reflexive construction etre hors de_ dejoie 'be beside_ with happiness'. (139) Pierre est hors soi/*soi-meme/*Marie dejoie. Pierre is beside REFLIREFL-same/Marie of joy 'Pierre is beside himself/*Marie with happiness.' (see also (126)) As indicated in ( 13 7) column (vi) the presence of selv is mandatory with neutral and even with inherently reflexive predicates. These examples thus clearly show that prosodic factors can override semantic constraints. In (138) the Contrastiveness Condition on adnominal intensification is overridden by the prosodic constraint barring unstressed af 'of to be followed by an unstressed clitic (i.e. sig). Since in these expressions the preposition (e.g. af, pa 'of, on') is mandatorily unstressed, the reason for the presence of selv with semantically neutral and inherently reflexive predicates is due purely to prosodic factors 68 • Predicates of type II have prepositions with variable stress, see (123) and (132). Semantically these predicates are either inherently reflexive (sub-type !Ia) or neutral (sub-type lib). In contrast to predicates of type I, the presence/absence of selv on sig can be explained purely in terms of semantic factors, see (137) column (vi). As shown by the examples in (132), in these predicates the preposition is underspecified with respect to stress and varies according to the following expression. If the expression following the preposition is stressed then the preposition is unstressed and vice versa. Hence it is ultimately semantic factors (i.e. the 68 Cf. Yeraart ( 1996) who proposes an analysis of Dutch reflexives which includes the assumption that prosodic factors can atTect the choice between complex and simple reflexives. 203 semantic difference between neutral and inherently reflexive predicates +the Contrastiveness Condition on adnominal intensification) which determine the distribution of sig and sig selv. Type III predicates have prepositions which always have some degree of word stress. In these cases, it is semantic factors (i.e. the semantic difference between neutral and inherently reflexive predicates + the Contrastiveness Condition on adnominal intensification) which straigthforwardly determine the distribution of sig and sig selv. 3.3.6.3 Can intensification of reflexives be reduced to lack of stressability? In this section the question whether adnominal intensification of reflexives can be reduced to stressability is discussed. First in section 3.3.6.3.1 and 3.3.6.3.2 the similarities between stressed and intensified reflexives in French and German are discussed. Then in section 3.3.6.3.3 the stressability-properties of pronouns and reflexives in Danish are contrasted. Finally in section 3.3.6.3.4 two different hypotheses concerning the relationship between stressability and intensification are discussed. 3.3.6.3.1 Stressable reflexives in French The examples in (140-142) show that the intensified reflexive soi-meme 'REFL-self and the stressed reflexives SO/ 'REFL' behave similarly in that they are both allowed in anti-reflexive constructions like etre jaloux de 'be jealous of ( 141 b,c) and excluded in inherently reflexive constructions like etre hors de 'be out of ' ( 142b,c ). (140) a. b. c. d. Personne n 'estfier de ok · SOl ok • "' soz-meme okSOI okMarie. 'Nobody is proud ofhimself/Marie.' ( 141) a. b. c. d. (142) a. b. Personne n 'est jaloux de soi ok · " soz-meme okSOJ "kMarie. 'Nobody is jealous of himself/Marie.' Personne n 'est hors de oksoi -soi-meme c. SO/ d. Marie de fureur. 'Nobody is beside REFLI*REFL self /*Mary with rage.' 204 As already discussed above, the similar behavior of soi-meme and SOl follow from the contrastiveness principle, as formulated in chapter 2. Both intensification and focus (realized as stress, e.g. ( 140-2c)) require the possibility of creating a contrast set of alternatives. Since this is excluded by the semantics of inherently reflexive constructions both intensified and stressed reflexives are excluded in (142). Conversely, since both focus and intensification trigger the generation of a set of alternatives (and thus allow doppelganger-readings of reflexives, i.e. (representational) non-identity of the referents of 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 similarly there still is a subtle difference between them, viz. generally soi-meme is usually preferred over SO/ in sentences like ( 141 ). 3.3.6.3.2 Stressable reflexives in German Like soi 'REFL' in French, in German the simple reflexive sich 'REFL' is also stressable and, as predicted by the theory of intensification proposed here, the behavior of stressed reflexive (e.g. SIC H) is similar to that of intensified reflexives (e.g. sich selbst), see ( 143-145). (143) a. b. c. d. (144) a. b. c. d. ( 145) a. b. c. d. Peter ist Stoltz auf oksich oksich selbst okSICH okMarie 'Peter is proud of himself/Marie.' Peter ist eifersiichtig auf sic h. oksich selbst okSICH okMarie 'Peter is jealous of himself/Marie.' Peter schiimt 'Peter ashamed.' oksich sich selbst S!CH Marie 205 The examples in ( 143-5) show that the intensified reflexive sich selbst and the stressed reflexives S!CH behave similarly in that they are allowed in anti-reflexive constructions like auf_ eifersiichtig sein 'be jealous of (144), and neutral constructions like stoltz auf'proud of (143), and excluded in inherently reflexive constructions like sich schiimen 'be ashamed' (145). Note that although intensified and stressed reflexives behave similarly there still is a subtle difference between them, viz. sich selbst is better than S!CH, just like, in French, soi- meme is generally preferred over SO/. 3.3.6.3.3 Stressable pronouns in Danish As shown in section 3.3.6.1 simple sig is a syntactically independent but prosodically dependent element which cannot host stress on its own, see ( 146a(iii)). In this respect, it contrasts with German sich and French soi, both of which are stressable, see (146b,c). Notice, however, that Danish object pronouns are stressable, see ( 146d), just like German and French reflexives. 206 (146) (i) unstressed, unintensified (ii) intensified (iii) stressed a. Dan. sig vs. sig selv *SIG b. Fr. soilse vs. soi-meme ;:::;69 SO! c. Ger. sich vs. sich selbst ;:::; SICH d. Dan. ham vs. ham selv ;:::; HAM The similarities of stressed and intensified pronouns (e.g. Dan. HAM 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 of functional equivalence. 3.3.6.3.4 Reducing intensification of reflexives to unstressability Taken together, the functional equivalence of stress and intensification in German (e.g. SICH ;:::; sich selbst), French (e.g. SO!;:::; soi-meme) and Danish (e.g. HAM;:::; ham selv), see sections 3.3.6.3.1-3, and the unstressabi1ity of simple sig, see section 3.3.6.1, may lead one to the hypothesis that intensification of reflexives is ALWAYS prosodically motivated, see ( 14 7). (147) Hypothesis 1: Only one function of intensification stress-carrier element (required to prosodically strengthen unstressable clitics) Basically what hypothesis I says is that adnomina1 intensification of simple sig is always motivated by stress. The reason why stress must fall on sig may vary, as shown in ( 148a-c ), but in all cases intensification of sig takes place in order to make and unstressable simple reflexive stressable (as marked in bold in (148)). ( 148) Stress on sig is triggered by: a. Semantics (representational non-identity => focus => stress => intensification (cf. (94a,b,d) in section 3.3.2.9) 70 69 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. 70 If Hypothesis I 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 Clinton; est fier de SO!, <statue>. Bill Clinton is proud ofRELF (ii) Bill Clinton, raziert S!CH,<statue>. Bill Clinton shaves REFL 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) While hypothesis I may work 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. HAN kom til mr:;det. HE came to meeting-the 'HE came to the meeting.' b. Han selv kom til mr:;det. (150) a. He self came to meeting-the 'He himself came to the meeting.' KONGEN kom til mr:;det. KING-the came to meeting-the 'THE KING came to the meeting.' b. Kongen selv kom til mr:;det. King self came to meeting-the 'The king himself came to the meeting.' As shown in chapter 2, the semantics of intensification and focus are similar, i.e. both involve the generation of a contrast set of alternatives (which, depending on the context, may or may not be scalarly ordered). It is thus not surprising that in many cases they can be found in the same contexts. However, the mere fact that certain languages allow for the existence of both intensified and focused expressions is a strong indication that there must be some semantic/functional difference between the two. Due to space limitations we leave the investigation of this topic to further research. However, if differences between focus and intensification are found, which seems to be likely, then Hypothesis I must be revised. A possible alternative hypothesis is formulated in ( 151 ). (!51) Hypothesis II: Two different functions of intensification a. Contrastiveness/representational non-identity(cf.(94a-d) in section 3.3.2.9)) b. Stress-carrier element (required to prosodically strengthen clitics, (121 ), ( 138)) Due to space limitations we leave this topic for further research. 208 Hypothesis II assumes two different functions of selv: (i) a contrastiveness marker which falls under the contrastiveness condition on adnominal intensification ( 151 a), (i i) a stress-carrier element which is required solely to enable unstressable reflexives to host stress (which is required by non-contrastive sentence stress) ( 151 b). While less elegant and less economical than Hypothesis I, Hypothesis II has the advantage of not predicting focus and intensification to be semantically and functionally identical (a prediction which is probably too strong). Furthermore, the existence of examples like ( 138) clearly shows that, in certain cases, prosodic constraints can outweigh semantic constraints 71 • Ultimately, the choice between Hypothesis I and II is, of course, extremely important. However, since it does not both are compatible with the analysis of intensification and binding proposed here we leave this topic for father research. 3.3. 7 Summary The important lesson to be draw from sections 3.3.1-6 is that binding and intensification are independent, see (152). ( 158) Independence of binding and intensification: a. Independence of locality and intensification b. Predicate-meaning and intensification c. Doppelganger-effects and intensification d. Resultative constructions e. ECM-constructions f. Possessive reflexives g. Reflexives in prepositional predicates and stressability (sect. 3.3.1) (sect. 3.3.2.1-4) (sect. 3.3.2.7) (sect. 3.3.3) (sect. 3.3.4) (sect. 3.3.5) (sect. 3.3.6) Binding of reflexives (i.e. the distribution of (thematic) sig) falls under principle A as formulated in section 3.2. That is, the distribution of the reflexive anaphor sig can be explained by a Pica-style account based on LF movement of the xo reflexive to IN FLIT. As shown such an analysis explains both the LD-behavior and subject-orientation of sig. Independently thereof the distribution of the adnominal intensifier selv 'self and its suppletive 71 See also discussion of inherently reflexive verbs with intensified sig selv in chapter 4. section 4.3.3. 209 variant egen 'own' can be shown to be determined by the non-binding related semantic, pragmatic, and prosodic factors. Furthermore, as shown in sections 3.3.3, 3.3.4, and 3.3.5. the nominal approach to binding adopted here has no difficulties with non-coargument bound reflexives (e.g. in resultative, ECM and possessive constructions) while many predicate centered approaches flounder on such cases. In section 3.4, the consequences of applying the same approach to binding and intensification of simple and complex pronouns are discussed. 3.4 Distribution of simple and complex pronouns In this section we propose an account of simple and complex pronouns in Danish based on the assumption that complex pronouns, e.g. ham selv 'himself/him himself, are the intensified versions of the simple pronouns, e.g. ham. But before we proceed to the discussion of complex pronouns, we need to determine the principles accounting for the distribution of simple pronouns, e.g. ham 'him. This will be done in section 3.4.1 where we argue that a syntactic approach to the distribution of pronouns based on principle B provides the best account. Then in section 3.4.2 and 3.4.3 the special properties of intensified pronouns (e.g. ham selv 'him self) in object and subject position respectively will be discussed in more detail. It will be shown that the differences between simple and intensified pronouns are explained as following from general properties of adnominal intensification. That is, the difference between ham and ham selv should not be explained within binding theory but rather with the module of intensification. Section 3.4.4 contains a brief discussion of the notion of logophoricity which has often been adduced to account for the behavior of intensified pronouns (as well as non-locally bound reflexives). Finally, section 3.4.5. concludes the discussion of pronouns and sums up the chapter. 210 3.4.1 Binding of pronouns: principle B As mentioned above, we argue that the binding of simple 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 Domain: alpha is a binding domain for beta if and only if alpha is the minimal category (i.e. the smallest DP or IP/S) containing beta, a case-licensor or beta, and a SUBJECT accessible to beta. Assuming that the relevant local domain is defined as in (154) in Danish, B principle in (153) accounts for the distribution of pronouns, see ( 155). (155) a. [Peter; barberede ham., ol Peter shave-PAST him 'Peter shaved him*i/o·' b. [Peter, bad migz barbere ham, •zl Peter ask-PAST me shave-INF him 'Peter asked me to shave him, *z·' c. [Peter; bad Jorgenz barbere ham, •zl Peter ask-PAST Jorgen shave-INF him 'Peteri asked Jorgenz to shave himi/*z·' d. [Peter, sagde [at jeg barberede ham,]]. Peteri say-PAST that I shave-PAST him 'Peter said that I shaved him.' Example (155a) illustrates that ham 'him' cannot be locally bound. It also shows that ham 'him' must 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 common with the simple reflexive sig 'REFL'. 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 II: ham must be free inside the minimal tensed clause 211 The sentences in ( 157) and (158), which contain what appear to be instances of pronominals bound inside their binding domains, indicate that we may have to modify the generalization in (156). (157) 72 a. Peter,fortalte Martin= om hamz selv. b. #hamz. c. *sigz (selv). Peter told Martin about him self/him/REFL self (158) a. Peter, gav Martin= 's maleri til hamz selv. b. #hamz. c. *sigz (selv). Peter gave Martin's painting to him self/him/REFL self What (157) and (158) show is that pronouns can have an antecedent inside their binding domains as long as it is not a subject. This leads us to the new generalization in ( 159). ( 159f 3 Descriptive generalization II (revised version): ham must be free from binding by a subject inside the minimal tensed clause While the distribution of pronouns can be accounted for syntactically within binding theory (i.e. principle B), the presence/absence of selv-intensification of pronouns, e.g. (157a)-(158a) vs. ( 157b )-(158b ), has to be explained by reference to the principles controlling adnominal intensification. In the case of examples ( 157-8), a semantic account naturally presents itself. Under normal circumstances one does not give somebody information about himself, since everybody is normally assumed to be well informed about themselves 74 • In the following 72 It would be interesting to test to see if stressed HAM '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. 73 As already shown in section 3.3.5, example (I 04), 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 Olek om b. John,Jortalte Olek om John told Ole about sin, *sink POSSREFL *hansi hansk his kane. kane. wife 212 section we will discuss more evidence showing that the so-called complex pronouns, e.g. ham selv 'him self/him himself, are in fact nothing but adnominally intensified version of their simple counterparts. 3.4.2 Intensified object pronouns. If binding and intensification constitute separate modules of the grammar, as claimed in this dissertation, then adjunction of the adnominal intensifier selv 'self to simple pronouns should follow from the principles of intensification. The examples in ( 160-1) appear to confirm this prediction. (160) 75 a. b. (161) a. Jon, vii giftes med en kvinde sam er stolt af ham, I *ham, selv I *sig, I *sig, selv 76 'Jon wants to marry (passive) with a woman who is proud of_.' Kongen troede at ingen kunne fide ham, I *ham, selv I *sig, I *sig, selv. king-the thought that nobody could like him. [=(2e) in section 1 above] 'The king thought that nobody liked him.' Jon, plejede at hade folk sam var anderledes end ?ham, I ham, selv I *sig, I *sig, selv. 'Jon used to hate people who were different from_.' 74 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) [. . .} NielsJorsogte [. . .}at gore Frithiofz komiskfor hamz selv [. . .}. ( 1980:86) Niels tried to make Frithiof comical for him self 'Niels tried to make Frithiof. comical to himselfz.' (ii) Hans Kjcerlighed, var stcerk som hendes, men den mangled den fine, mandige 0mhed, der vcerner den elskede Kvindez mod hendez selv og vaager over hendesz l'cerdighed. (I 980: 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 that, guards the beloved womanz against herselfz and protects herz dignity.' Although we have yet to find any convincing cases of semantically "neutral'' or "inherently reflexive" constructions with non-subject-bound, local intensitied pronouns, we cannot exclude that such examples might exist. If this is the case. then they may constitute a potential problem for the account of complex pronouns presented here. 75 The examples in (160a) and (16lac.d) are from Safir (M.S.), p. 17-18. (35ab.d,e). 76 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 of 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. b. Carl, sagde at Marie havde snakket med aile andre end ham, I ?ham, selv I *sig 1 I *sig, selv. 'Carl said that Marie had talked to all others than c. Carl, sagde at disse mcend hader aile inklusivelundtagen ham, I ?ham, selv I *sig, I *sig, selv. 'Carl said that the men hate all including/excluding_.' d. Jon, sagde at han aldrig ville tillade sin datter sa me get sam at overveje at gifte sig med en mand helt forskelligfra ?ham, I ham, selv I *sig, I *sig, selv. 'Jon said that he never would allow his daughter so much as to consider to marry herself to a man completely different from_.' Notice that while the examples in (160) only allow the simple pronoun ham 'him', the examples in ( 161) all allow for both simple ham 'him and complex ham selv 'himself (him 213 himself)'. Contrast seems to be the most important characteristic of the contexts allow LD- bound ham selv illustrated in (160). Both the so-called "similarity predicates", e.g. (161a,d), and "exclusion predicates", e.g. (161b,c), are inherently contrastive predicates which trigger adnominal intensification. The term "similarity predicate" is from Safir ( 1992). Safir (M.S., p. 14) defines this class of predicates as follows: "The semantic class of [similarity] predicates [ ... ] very often appears in copular constructions the same way as other adjectives do as in the comparisons in [(162)], yet other adjectives do not induce the same logophoric effects, as illustrated in [(163)]. A list of similarity predicates (from Safir (1992)) appears in [( 164)]". (162) a. b. c. (163) a. a'. b. b'. c. d. Darby is fairly deferential I similar to Selena. Esther is quite estranged I different from Da.IJY. Angela is rather angry at I like Andrea. Sis.sy insists that Darby is fairly similar I *?deferential to herself. Etta attacks people similar I *?deferential to herself. Thea says Esther is quite different I *?estranged from himself. Thor threatens people different I*? estranged from himself. Otto admits that Mary is rather like I *angry at himself. Abraham always abhors someone like I *angry at himse(f ( 164) like, unlike. similar to, different from, identical to, such as 214 "While the "similarity" predicates, all having to do with identity of person or properties in relation to another, all permit the Jogophoric effect, none of the non-similarity predicates do. The R&R account predicts that none of these predicates permit logophoricity if they apply their definition carefully. However 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 presumably they must say the same thing about adjective plus preposition units like deferential to. angry at and estranged from. Either way, they fail to make a significant distinction because they fail to take into account the semantic force of the predicates involved." (Safir (M.S.), p. 14). The term "exclusion" predicate is from Safir ( 1992). Safir (M.S.) describes the properties of exclusion predicates as follows: "unlike similarity predicates exclusion predicates [see (165)] are not possible in copular environments [see (166)], yet [like similarity predicates] they also are typically two 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 I rather than I other than Bill. By the latter criterion, the SELF-forms are syntactic arguments in the R&R account if exclusion predicates are indeed predicates. Yet all of these exclusion contexts permit logophoric behavior [see ( 167)]. ( 167) Dole pointed out that the Republicans would look foolish if anyone other than I rather than I apart from I except I save himse~f were nominated. Safir concludes 'that exclusion predicates are predicates and as such they are just as problematic for R&R as the similarity predicates are" (Safir (M.S.), P. 14-15). The similarity predicates x er anderledes end y 'x is different than y' and x er forskellig fra y 'x is different from y' in ( 161 a, d) both involve directly contrasting two 215 individuals or groups. Similarly, the exclusion predicates alle andre end x 'all others than x', alle medregnet y 'all including y', and aile undtagen y 'all excluding y' in (161b,c) also involve direct contrast of two groups/individuals; i.e. in ( 161 b,c) the group of people whom Marie spoke to is explicitly contrasted with the group of individuals with whom she didn't speak, see ( 168). (168) a. X er anderledes end y 'X is different from Y' b. alle andre end y 'all others than Y' c. aile inklusive/undtagen y 'all including/excluding Y' d. X er forskelligfra y 'X is different from Y' e. X IS CONTRASTED WITH y Inherently contrastive predicates can be defined as predicates which involve explicitly contrasting two individuals or groups (X and Y) with respect to some property or event, see ( 168t) 77 • In examples with inherently contrastive predicates it is thus the meaning of these predicates which satisfies the contrastiveness principle, as formulated in chapter 2, and thereby licenses intensifier adjunction to personal pronouns. Now, let us return to the example in (160a). According to the judgments reported in Sa fir ( 1997), the intensified pronoun ham selv is unacceptable in this example. Notice however, that ham selv can be rendered felicitous, even in the sentence in ( 160a), in explicitly contrastive situations, see (169A). 77 Note that with respect to the distribution of simple and complex reflexives in Danish (i.e. sig vs. sig se/v). inherently contrastive predicates are "anti-reflexive'' (i.e. ''other-directed"). e.g.: (i) Peter er forskelligfra *sig i sig selv I John. 'Peter is different from himselt/John. · (ii) Peter barberede all undtagen *sig I sig selv I John 'Peter shaved all excluding himselt/John.' (169) Q: Hvem er det Jon vil have, at den kvinde han vi! g(ftes med skal vcere stolt af? Vil Jon absolut g!ftes med en kvinde, som er stolt afhansfar? 216 who is it Jon will have that the woman he will marry-PASS with shall be proud of will Jon absolutely marry-PASS with a woman who is proud of his father 'Who is it that Jon wants the woman he is going to marry to admire? Does Jon absolutely want to marry a woman who admires his father?' A: Nej, dit torskehovede! Jon vi! g!ftes med en kvinde sam er stolt afham selv. No you cod-head Jon will marry-PASS with a woman who is proud of him self 'No, idiot! Jon want to marry a woman who is proud of (him) himself.' It thus seems to be the case that ham selv can felicitously occur in sentences like (160a) under the condition that strong contrast is directly expressed or implied by the larger context as in ( 169) (where ham selv, referring to Jon, is directly contrasted with hans far 'his father'). If this condition is not met, e.g. ( 160a), then the presence of selv does indeed seems rather infelicitous. The examples discussed above seem to indicate that the presence of selv in the complex pronoun ham selv can be licensed in at least one of two different ways: (i) by the inherent contrastiveness of its governing predicate, see (161a-d), or (ii) by the contrastiveness evoked by the larger (linguistic or extra-linguistic context), see ( 169). Now compare that with the situation holding for sig selv. As already discussed in section 3.3.2, with "neutral" predicates the presence of intensifier-adjunction is optional and usually only occurs in contexts which involve strong contrast between two individuals/groups, see ( 170), and cannot occur in sentences in which another element has already been focused, see ( 171 ). See also Keenan (200 I). Konig & Siemund ( 1999), Zribi-Hertz ( 1989) for discussion of"contrastive predicates". (170) Q: A: (171) Q: Hvem var det nu at Peter havde vasket? Havde han vasket sin hund? 'Who was it that Peter had washed? Had he washed his dog?' Nej, han havde vasket SIG SELV I *SIG. 'No, he had washed himself.' Hvad var det nu at Peter havde gjort for han gik i seng? 'What was it Peter had done before he went to bed?' A: Han havde VASKET sig I ??sig selv. 'He had WASHED himself.' 217 As illustrated by the contrast between (160a) vs. (169) and (170) vs. (171), selv is often adjoined to the simple reflexive sig and the simple pronoun ham for the purpose of indicating contrast. Indeed, the contrast between ( 160a) vs. ( 169) and ( 170) vs. ( 171) illustrates that optional intensifier-adjunction to sig and ham is felicitous only when contrastiveness is somehow motivated (or at least compatible with) by the larger context. The fact that ham selv can occur freely with all the predicates m (16la-d), even without explicit contrast present in the extra-sentential context, can now be explained simply as following from the fact that all those predicates are inherently contrastive, see ( 168e ), and thus in themselves provide the contrastive context necessary to motivate the presence of selv. This also explains why selv-adj unction is not licensed in those constructions involving predicates like stolt af'proud of which are not inherently contrastive, unless the larger context contains explicit contrast between the argument of the predicate stolt af 'proud of and some other entity, see ( 160a) vs. (169). In the above discussion we have demonstrated that complex 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 adnominal intensification discussed in chapter 2. There are, however, also important differences between ham selv and and sig selv which follow from semantic and prosodic differences between simple sig and ham, some of which are listed in (172). (172) Comparing sig and ham: a. Can carry stress b. Can refer on its own sig NO NO ham YES YES 218 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 examples ( 161 a-d). (173) a. b. c. Carl sagde at Marie havde snakket med alle andre end ham selv Carl sagde at Marie havde snakket med alle andre end HAM. ?*Carl sagde at Marie havde snakket med alle andre end ham. Carl said that Marie had talked to all others than himself/HIM/?*him The examples in (173a,b) show that both intensifier adjunction (ham selv) and focus on HAM 'HIM' can be used to satisfy the contrastiveness requirement of the "inherently contrastive" predicate alle andre end 'all others than' 78 • In contrast, the sentence degrades considerably if the pronoun ham is neither focused nor intensified, see (173c) 79 • The fact that ham has phi-features and thus can have reference on its own, see ( 173b ), is the reason why simple ham gives rise to referential ambiguities which do not occur with simple sig, see ( 174a). (174) a. b. Peter, sagde at arkestret ikke maatte spille med andre end ham; z sam dirrigent. Peter, sagde at arkestret ikke maatte spille med andre end ham,*= selv sam dirrigent. Peter said that orchestra-the not could play with others than him I him self as conductor 'Peter said that the orchestra couldn't play with anybody but himself as conductor.' Simple ham in (174a) may refer either to the matrix subject Peter or to some 3rd person male individual not mentioned in the sentence. The complex ham selv, on the other hand, seems to 78 This is reminiscent of the situation holding in English where stressed HIM and himself are also (to a certain extent) in free variation, see Baker (1995). See also Siemund's (2000) critique of this analysis. 79 Note that the sentence in (i) may potentially constitute a problem for the account of the distribution of ham and ham selv proposed here. (i) Peterx havde snakket med aile andre end *hamx I hamx selv I *sigx I sigxselv. '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). 219 be restricted to referring back to the matrix subject. It thus seems that intensifier-adjunction to ham has the effect of limiting the range of possible antecedents. This effect is what Zribi-Hertz ( 1995) refers to as the endophoricizing effect of intensifier-adjunction. It does not seem implausible that, in certain instances, the use of ham selv rather than simple ham could be motivated by a need/intention to disambiguate the reference of ham. Notice, however, that intensification of pronouns does not always disambiguate between two or more possible antecedents, see (175). (175) a. b. Rune= troede at Peter, havde sagt at orkestret ikke maatte spille med andre end ham, z selv som dirrigent. Rune thought that Peter had said that orchestra-the not could play with others than him self as conductor 'Rune thought that Peter had said that the orchestra couldn't play with anybody but himself as conductor.' Rune= var ikke synderligt bange for at nogen anden skulle usurpere hansz plads som dirrigent pa det kongelige teater. Peter, havde nemlig sagt at orkestret ikke maatte spille med andre end ham 1 = selv som dirrigent. Rune was not particularly afraid for that somebody else should usurp his place as conductor at the royal theater. Peter had actually said that orchestra-the not could play with anybody but himself as conductor. 'Rune was not particularly afraid for that somebody else should usurp his place as conductor at the royal theater. That was because Peter had said that the orchestra couldn't play with anybody but himself as conductor.' In (175a) the antecedent of the intensified pronoun ham selv could be either the Peter or the matrix 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 examples, we conclude that the apparent disambiguation function of selv-adjunction is simply a side-effect of adnominal intensification. The sentence in (176) illustrates that intensified pronouns do not need to be bound by logophoric subjects and that the main function of adnominal intensification is contrastive. 220 (176) For hende, var det. som om dode, lykkelige Dage rejste sig af deres Grav og gik igjen, saa Alting scelsom sodt forvirredes og Fortid og Nutid smelted sammen til en solvsloret, dcemrende Drommedag, hvor hun, elskede Ynglingenz , halvt som hamz selv, halvt som en Andens Skygge, og gav hamz helt sin; halve Sjcel. For her was it, as if dead happy days rose REFL of their grave and went again so everything strangely sweet confused-PASS and past and present melted together to a silver-veiled dim dream-day where she loved the swain half as him self half as a other's shadow and gave him whole POSSREFL half soul 'For her; it was as if dead, happy days rose from their grave and came back hauntingly, so that everything was confused in a strangely sweet way and past and present melted together to a silver-veiled, dim dream-day where she; loved the swain., half as himself., half as somebody else's shadow, and gave himz unconditionally half of her; soul.' (J.P.Jacobsen (1880)Niels Lyhne, p. 81, bold and italic added here) In (176) intensification of the pronoun ham is clearly not motivated by a need for referential disambiguation, but rather by the explicit contrast between Ynglingen 'the swain' and en Andens Skygge 'somebody else's shadow'. We suggest that the primary function of intensification is to indicate some kind of contrast, and that what appears to be a disambiguation function of selv is no more than a secondary effect of intensification. 3.4.2 Intensified subject pronouns: Intensified pronouns as subjects of embedded clauses in English and Danish In the previous section intensification of object pronouns was discussed. This section will present an account of simple and complex subject pronouns. As with complex reflexives and complex pronouns discussed above, we argue that complex subject pronouns, e.g. Dan. han selv 'he himself, are simple adnominally intensified versions of the corresponding simple forms, e.g. han 'he'. In his 1986 Linguistic Inquiry squib "He Himself: Anaphor, Pronouns, Or ... ?" Bickerton claims that the complex form he himself is a special kind of anaphor having the 221 features [+anaphor, +pronominal] in Chomsky's (1982) system 80 • This feature combination would then explain the (alleged) fact it appears to have properties associated with both anaphors and pronouns. Bickerton (1986:347) 81 arrives at this conclusion based on data such as ( 177)-( 182). 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-command it, see ( 181) and (182a), also confirms its pronominal behavior. However, it behaves more like anaphor in other respects. It does not allow extra-sentential antecedents in the presence of potential intra- sentential antecedents, see ( 178). Furthermore, if it is bound inside the sentence, then it must be by a c-commanding antecedent, see (178) and (179). While complex forms like she herself share properties of both anaphors and pronouns they also differ from both with respect to Case-marking and reconstruction effects. Unlike reflexives and object pronouns, they receive nominative case and appear to allow Principle C violations in LF reconstructions, see (182a) vs. (182b). (177) A: How will Mary, do in the exam? B: I don't know, but shei herself says she/ 11 pass. (178) A: How will Mary, do in the exam? B: I don't know, but Susank says that she, k /she herself•; k will pass. 80 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 himself. see (i). (i) a. 0 Yanis; pistevi oti o idhios; tha kerdhisi. (=latridou 1986(6)) John believes that himself will win 'John, believes that he himself will win.· b. 0 Yanis, theli [i Mariana voithisi ton idhioJ. John wants Mary helps himselt; 'John, wants Mary to help him 1 .' c. 0 Yanis, ipe stan C ostak [ oti i Maria aghapa ton idhio, k •nJ. John said to Costa that Mary loves himself 'John, said to Cost~ that Mary loves him, k 'm·' Like English himself, the expression idhios is also used as adnominal intensifier, see (ii). (ii) a. 0 Yanis o idhios pighe sto scholio. b. Aftos o idhios pighe sto scholio. c. pro o idhios pighe sto scholio. ·John I He I pro went to the school himself. • Modern Greek is a pro-drop language and Iatridou argues that (iic) shows that pro can be adnominally intensified by the intensifier idhios 'same, himself. 81 The account of examples ( 177-182) is based on a paraphrase of Bickerton ( 1986:347). 222 (179) a. John, told Bilh 's sister that he, k I he himself *k had been arrested. b. Susan, told everyone who knew Maryk that she, k I she, *k herself was pregnant. ( 180) Even Bill, 's genius has its limits. The problem that Mary just raised is one that he himself admits he can't solve. ( 181) a. b. c. (182) a. b. The essays that Mary, wrote were things that she; herself attached little importance to. Explaining what John, really believes is something that he, himself can 't always do. The ways in which the twins, behave suggest that they, themselves don't always know what they're doing. Which pictures of John, does he, himself/ike t? *Which pictures of John, does he, like t? Unlike Bickerton ( 1986), McKay ( 1991) argues that he himself is not an anaphor. Furthermore, according to McKay the "emphatic appositive use of reflexives", i.e. what in this dissertation is called intensifiers, "require a clearly indicated referent and a relevant contrast or comparison class [ .. ]"(McKay 1991 :368). In other words, McKay essentially proposes that he himself is an intensified pronoun. Syntactically, it behaves like its unintensified counterparts he, him, etc. Pragmatically, it is subject to the same discourse based conditions (i.e. contrastive contexts) as other intensified expressions. In this sense, McKay ( 1991) is the forerunner of Baker ( 1995) as well as the present dissertation. That is, we share McKay's conclusion that "he himself is grammatical whenever he is grammatical, and the same range of antecedent relationships is grammatically possible, though he himself requires a relevant contrast or comparison for pragmatic appropriateness" (McKay 1991:370-371 ). McKay ( 1991) uses examples like (183) to show that he himself is not a new type of anaphoric expression (as argued by Bickerton), but simply an intensified pronouns which is still subject to principle B. (183) A: Mary, has been concerned about her friends. Susank said that several were going to fail the course, and Susank might be right. But Mary, should think more about her, own work. How will Mary, do on the exam? 223 B: I don't know, but Susank says that she, (sheJ (she, herse(f) (shek herself will pass. We agree with McKay that the example in (183) falsifies Bickerton's (1987) claim that he himselfbehaves like an anaphor in that it "cannot have an antecedent outside the sentence if there is a possible antecedent inside". Bickerton's claim was based on examples like (178). McKay's main contribution is to show that, given the right context, as in (183), he himself can be bound by a sentence-external antecedent in spite of the presence of a potential antecedent inside the sentence. That is, the pronominal nature of subject pronouns is not affected by adnominal intensification. In other words, binding and adnominal intensification are independent of each other. As the examples in ( 184-185) show, Danish intensified subject pronouns, e.g. han selv 'he himself, behave similarly to their English counterparts, i.e. they are pronominals falling under principle B of the binding theory. Independently thereof, whether or not they are intensified depends on pragmatic factors (i.e. contrast). (184) A: B: Mary, has been concerned about her friends. Susank said that several were going to fail the course, and Susank might be right. But Mary, should think more about her; own work. How will Mary, do on the exam? Det ved jeg ikke, men Susank siger at hun; /hun; k selv vil best a uden problemer. that know I not, but Susan says that she/she herself will pass without problems 'I don't know, but Susank says that she; 1 k/she; 1 k herself will pass without any problems.' ( 185) Niels Lyhne; havde derfor ogsaa digtet ude fra en cesthetisk Personlighed i a! Almindelighed. der Jandt Foraaret svulmende, Havel start, Kjcerligheden erotisk og D0den melankolsk. Han; selv var ikke kommet videre med denne Poesi, han; bare lavede versene. Niels Lyhne had therefore also composed-poetry out from an aesthetic personality in all generality, which found spring-the swelling, sea-the great, love-the erotic and death-the melancholic he self was not come further with this poetry he merely made verses-the 'Niels Lyhne; had therefore also composed poetry based on a general aesthetic personality who found the spring swelling, the sea great, love erotic and death melancholic. He; himself had not gotten any further with this poetry; he; merely made the verses.' (J.P.Jacobsen (1880) Niels Lyhne, p. 83, bold and italic added here) 224 In (184) Mary is contrasted with her friends and in (185) Niels Lyhne is contrasted with en a;sthetisk Personlighed i al Almindelighed 'a general aesthetic personality'. In brief, intensified pronouns in Danish, e.g. han selv 'he himself, are not a special type of anaphor but simply adnominally intensified versions of their simple counterparts which have to satisfy the contrastiveness condition on adnominal intensification. 3.4.3 Logophors and Iogophoricity vs. intensification and intensified pronouns The term "logophor" is often used loosely in the literature to refer to locally free instances of himself in English 82 • It has also sometimes been used to refer to the intensified pronoun ham selv which tends to be the form used in Danish in contexts were English would have locally free instances of himself This use of the term logophor is, however, potentially misleading since the term logophor is also widely used in a more narrow sense as referring to an author- denoting pronominal element which has to be bound by an attitude operator (cf. Clements ( 1975), Chierchia (1989), Huang&Liu (2000), Schlenker (1999), etc .. ) 83 • This definition of logophoricity was first developed to describe the behavior of morphologically specialized logophoric elements in West African languages (cf. Hagege (1974), and Clements (1975)). The examples in (186-87) (from Clements (1975)) exemplifies the behavior ofthe logophoric element ye in Ewe. (186) a. Kofi be ye-dzo Kofi say LOG-leave 'Kofi said that he (Kofi) left.' b. Kofi be e-dzo Kofi say he/she-leave 'Kofi said that he/she ( tXofi) left.' 82 E.g .. Reinhart & Reuland ( 1993) who refer to all instances of himself 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. 83 Sells ( 1987). Zribi-Hertz ( 1989). Reinhart & Reuland ( 1993). etc. for different definitions of logophoricity. 225 (187) a. *Am a do nku ny~uvi hi dze ye gb:J dyi Ama set eye girl WH stay LOG side on b. Am a do nku ny~uvi hi dze e gb:J dyi Am a set eye girl WH stay pro side on 'Ama set eye on (remembered) the girl who stayed with her' c. Am a gbl:J be ye-do nku ny~uvi hi dze ye gb:J dyi Ama say that LOG eye girl WH stay LOGside on 'Amai said shei remembered the girl who stayed with heri' The example in ( 186a) show the standard case of a logophoric pronoun bound by an overtly realized attitude operator (i.e. the matrix subject). Interestingly, the English and Danish translations of this example do not necessarily involve the so-called "logophors", i.e. himself and ham selv, but rather the simple subject pronouns he and han 'he', see ( 188). ( 188) a. b. Kqfi said that he I!! he himself I *himself left. Kofi sag de at han/ (!han selv I *selv tog af~ted. Kofi said that he I he self/ self took of-place 'Kofi said that he/ (!he himself/*himselfleft.' The intensified pronoun han selv, marked with a (#J in ( 188) is perfectly acceptable when the larger context implies contrast (see the discussion of examples (160a) and (169) in section 3.4.2). In this respect, han selv seems to behave exactly like he himself in English, see also section 3.4.2. The Ewe examples in ( 187) illustrate the fact that logophors cannot occur inside a relative clause, unless the relative clause is itself embedded under an attitude verb. Let us now test to see if Danish ham selv behaves the same way in relative clauses, see ( 189) . (189) a. b. c. Jon vil giftes med en kvinde sam er stolt af h I .,., okh l IU / * . / * . l am · am se v szg szg se v. 'Jon wants to marry (passive) with a woman who is proud ofhimself.' Jon sagde altid at han ville giftes med en kvinde sam var stolt af ham I ""kham selv I *sig I *sig selv. 'Jon always said that he wanted to marry (passive) with a woman who is proud of .' Jon vil giftes med en kvinde sam er anderledes end ?ham I ham selv I *sig I *sig selv. 'Jon wants to marry (passive) with a woman who is different from.' 84 The judgments indicated as '' nkham selv here means that the use of the form ham selv is unfelicitous (i.e. '.m unless when the whole sentence occurs in an explicitly contrastive context (i.e. ok). 226 d. Jon sagde altid at han ville giftes med en kvinde som var anderledes end ?ham I ham selv I *sig I *sig selv. 'Jon always said that he wanted to marry (passive) with a woman who is different from ' As already discussed in the previous section, the sentence (189a), does seem to be infelicitous as it is, i.e. in the absence of an explicitly contrastive larger context. The question is now whether it will behave like its Ewe counterpart (187a,c) and improve when embedded under an attitude verb, see (189b). It is my impression as a native speaker that the acceptability of ham selv in ( 189a) does not improve significantly when the sentence is embedded under an attitude verb whose subject is coreferent with ham selv, see (189b) 85 • While (189b) may be slightly better than ( 189a) it is still the absence vs. presence of an explicitly contrastive context which is the most important factor deciding the acceptability of both ( 189a) and (189b ). The examples 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 of the presence/absence of overt attitude operators and the nature of the context. In other words, unlike the situation holding for Ewe logophors, see ( 186) and (187), in Danish, it is not the presence/absence of an attitude operator which determines the acceptability of the sentences with ham selv in ( 189), but rather the contrastive/non-contrastive nature of the predicate governing ham selv. Based on the above examples we conclude that Danish ham selv is not a logophor in the narrow sense of the term but rather the intensified form of the pronoun ham. 3.4.4 Summary While there are still a number of unsolved problems surrounding the question of the distribution of ham selv in Danish (e.g. the endophorizing/logophorizing function of selv- 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 of selv) and binding (distribution of pronouns han/ham/hans 'he/him/his') are mutually independent. Complex pronouns (e.g. han selv 'he himself, ham selv 'him himself, and hans egen 'his own') composed of a pronominal part (e.g. han/ham/hans 'he/him/his) plus an adnominal intensifier (selvlegen 'self/own'). The pronominal part fall under principle B of the binding theory while the intensifier is subject to the contrastiveness condition on intensification. Consequently, intensified pronouns in Danish and English are neither logophors (i.e. elements which have to be bound by an attitude operator), nor a special kind of anaphor (cf. Iatridou (1986), Bickerton (1986)) as has sometimes been proposed. 3.5 Conclusion The independence of binding and intensification In this chapter we have presented an account of simple and complex reflexives and pronouns in Danish based on the proposal that binding and intensification should be clearly separated although they overlap in the case of complex reflexives and pronouns which are here analyzed as adnominally intensified forms of their simple counterparts. That is, we proposed that the descriptive generalizations in (190)-(193) are best accounted for within a framework which separates binding and intensification into the independent modules summarized in (194) and ( 195) respectively. ( 190) Simple reflexives (e.g. sig): must be bound by subject in a local domain (=minimal tensed clause) (191) Simple pronouns (e.g. ham): must be subject-free in a local domain (=minimal tensed clause) 85 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. 228 ( 192) Complex reflexives (e.g sig selv): a. anti-reflexive predicates b. hidden neutral c. contrastive contexts (with neutral predicates) d. doppelganger-effects (i) statue-readings (ii) qua-sentences (iii) strict (and sloppy) readings in VP ellipsis e. stress-carrier (e.g. after unstressed prepositions) (193) Complex pronouns (e.g ham selv): a. inherently contrastive predicates b. explicitly contrastive contexts (194) Bindingtheory: a. Principle A (accounts for (190)). b. Principle 8 (accounts for(l91)). (195) Contrastiveness condition on adnominal intensification (=(2) chapter 2) (accounts for (192-193): A nominal expression (DP, pronoun, reflexive) can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. Some of the consequences of the present proposal for linguistic theory in general are listed in ( 196) and in ( 197) particular consequences for the analysis of Modern Danish can be found. ( 196) Consequences for binding theory: a. Intensification and binding are independent modules of the grammar. b. Binding is purely syntactic phenomenon. c. The unified account of possessive and argument reflexives is a strong argument in favor of nominal approaches to binding and against current predicate-centered approached (e.g. Reinhart and Reuland (1993)). d. "Doppelganger-effects" (statue-readings, qua-sentences, strict reading in VP ellipsis, etc.) and anti-reflexivity (the triggering of adnominal intensification of reflexives with certain predicates) can be unified as cases of adnominal intensification falling under the contrastiveness condition. (197) Consequences for analysis of Modern Danish (and similar languages): (i) Simple reflexives (e.g. Dan. sig 'REFL') are not anti-local (discussed in more detail in chapter 4). (i) Complex reflexives (e.g. Dan. sig selv 'REFL self) are not mandatorily local. (ii) Complex reflexives, e.g. sig selv, are adnominally intensified reflexives. (iii) The complex object pronoun ham selv is not a logophor but an adnominally intensified object pronoun. 229 (iv) The complex subject pronoun han selv is not a special kind of anaphor (cf.Bickerton 1986) but an intensified subject pronoun. (v) The element selv 'self is not a reflexivizing particle which falls under binding theory (e.g. Reinhart and Reuland 1993), but an intensifier which falls under the module of intensification (see chapter 2) (vi) The element egen 'own' is the suppletive variant of the adnominal intensifier selv 'self. (vii) The argument anaphor sig 'REFL' and the possessive anaphor sin 'POSSREFL' have the same distribution and can both be accounted for by the same principle A of the binding theory. As shown above, the contrast-based account intensification proposed in chapter 2 brings out the core common feature of se/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 of this paper, intensifier-adjunction to reflexives can only take place when the predicate/construction containing sig/sinx 6 allows for contrast with other entities. Likewise, intensifier adjunction to ham has a strong tendency to prefer contrastive contexts. Compared with previous analysis of reflexives and pronouns in Danish, the focus-based analysis of intensifier-adjunction proposed in chapter 2 has a better chance at eventually arriving at a unified account of intensified reflexives, e.g. sig selv, intensified pronouns, e.g. ham selv, and intensified DPs, e.g. kongen selv 'the king himself. Indeed, it seems to be the failure to treat intensification and binding as separate subsystems of the grammar which has made it impossible for many researchers to even conceive of a unified analysis of the function of selv across reflexives, pronouns and DPs. In chapter 4 a few loose ends will be tied up. That is, the thematic status of simple reflexives with neutral and inherently reflexive verbs will be discussed in greater detail and a late-insertion analysis of reflexives and pronouns, which also accounts for 1st and 2nd person 86 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. 230 forms, is proposed. In chapters 5 and 6 similar approaches to binding and intensification in English and Mandarin Chinese will be put to the test. 231 Chapter 4 The syntactic status of sig 4.1 Introduction In this chapter, the hotly debated question of the lexical and syntactic status of the simple reflexive sig 'REFL' (=Nor. seg, Dutch zich, etc.) is discussed. We argue that sig can be a locally-bound theta-marked reflexive pronoun (I a), (2a). Hence, it is not an anti-local anaphor as has sometimes been proposed. We furthermore argue that the lexicon only contains one entry for the element sig and that the difference between thematic and non-thematic uses of sig depends on where in the derivation this elements merges (I b). ( 1) Proposal: (2) a. sig is not anti-local. i.e. thematic sig can have both local and LD-antecedents. b. the lexicon contains only one entry for sig. This element can be either a thematic anaphor, when it occurs in argument position of an active transitive predicate (2a), or a non-thematic grammatical marker of unaccusativity, i.e. in middles or passive, when it occurs in the v of the voice-projection vP (2b). a. Thematic sig: yp I \ V' I \ v sig b. Non-thematic sig: vP I \ v' I I \ VP sig v I \ .... V' I \ v The examples in (3) illustrate thematic uses of sig 'REFL'. (3) a. Peter vasker sig. Peter washes REFL 'Peter washes (himself).' b. Peter lagde bogen bag sig. Peter put-PAST book-the behind REFL 'Peter put the book behind him.' 232 Both instances of sig in (3) are thematic in that they receive thematic-roles from their predicates. In contrast, the examples in (4) illustrate non-thematic uses of sig. (4) a. Peter skynder sig. Peter hurries REFL 'Peter is in a hurry/Peter hurries.' b. Peter skammer sig. Peter shames REFL 'Peter is ashamed.' In section 4.3 we show that the predicates skynde sig 'hurry' and skamme sig 'be ashamed' are not transitive predicates but rather lexically unaccusative intransitive predicates which only assign one theta-role to the internal argument Peter which later moves to the subject position for case reasons. Unlike the transitive verb vaske 'wash' in (3a), they do not denote events or actions involving two objects. Rather, they denote particular states of one individual, i.e. the referent of the theta-marked subject. Hence, since there is no theta-role left for the reflexive, the instances of sig in ( 4) are all non-thematic. Some verbs, e.g. abne 'open' and lukke 'close', have both agentive transitive uses (5) and deagentive unaccusative uses with sig (6). The examples in (4) and (6) thus illustrate lexical and derived unaccusative predicates respectively. (5) a. b. Peter abner d0ren. Peter opens door-the 'Peter opens the door.' Peter lukker d0ren. Peter closes door-the 'Peter closes the door.' (6) a. c. Doren abnede sig . door-the opened REFL 'The door opened.' Doren lukkede sig . door-the closed REFL 'The door closed.' 233 As mentioned in (1), we propose to analyze the difference between thematic sig, see (3), and non-thematic sig, see ( 4,6), as a syntactic rather than as a lexical difference. Lexically there is only one element sig which can take on different functions depending on where it merges in the derivation. Thematic sig merges in a theta-position, e.g. internal argument of verbal predicate (3a) or argument of preposition (3b ), while non-thematic sig merge with the v of the voice projection vP, see (2b ), ( 4) and (6). It is the clitic properties of sig, i.e. its simultaneous Xo and XP status, which enable it to behave differently depending on its position in the derivation. Needless to say, only thematic sig falls under the binding theory, while non-thematic sig falls under a different module of the grammar (i.e. argument structure/voice). We argue that this approach to the analysis of all occurrences of sig in Danish is both more explanatory as well as more economical than previous accounts. According to a number of current accounts of Danish (and similar languages, e.g. Norwegian, Swedish, and Dutch) simple reflexives (e.g. Dan. sig) are anti-local. That is, it is assumed that local sig is never a true anaphor but rather a non-thematic grammatical marker of intransitivity which is part of the lexical make-up of deponent verbs and reanalyzed deagentive unaccusative verbs (cf. Holmberg (1984), Vikner (1985) among others). We argue that although this analysis does apply to some instances of local sig, e.g. (4) and (6), it does not extend to ALL verbs allowing local sig. As will be shown, the large group of verbs referred to here as "neutral" do allow local sig (even when bound by co-arguments) without showing any signs of reanalysis or 234 lexical ambiguity. The fact that the neutral predicate vaske 'wash' has the same meaning 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 predicate vaske 'wash' assigns the same two theta-roles (AGENT and THEME) and thus denotes a relation of washing between two entities (the only difference being that in (3a) the washer and washee happen to be the same). We argue that examples such as (3a) and (7) falsify the anti-locality accounts of sig which would have to claim that the sig in (3a) is non-thematic grammatical marker of intransitivity. Using a number of syntactic and semantic tests we show that there is no evidence for assuming the verb vaske 'wash' to have been reanalyzed as an unaccusative intransitive verb in (3a). In contrast, several tests show that vaske 'wash' is still a transitive verb assigning two theta-roles. As already discussed in chapter 3, based on what kinds of nominal expressions they allow in object position, Danish verbal predicates can be divided into three main types: (i) "neutral" predicates which allow both simple and complex reflexives (i.e. sig and sig selv) and DPs, see (8), (ii) "anti-reflexive" predicates which allow the complex reflexive sig selv and DPs but not the simple reflexive sig, see (9), and (iii) "inherently reflexive" predicates which allow only the simple reflexive sig, see (I 0). (8) "Neutral" predicates: a. Peter vasker sig I sig selv I bilen. b. Peter washes REFL I REFL-SELF I car-the 'Peter washes himself I the car.' Peterforsvarer sig lsig selv I Marie Peter defends REFL I REFL-SELF I Mary 'Peter defends himself I Mary.' (9) "Anti-reflexive" predicates: (thematic sig) (thematic sig). a. Peter mista:nker *sig I sig selv I Marie. (thematic sig selv) Peter suspects *REFL I REFL-SELF I Mary 'Peter suspects himself I Mary.' b. Peter misunder *sig I sig selv I Marie. (thematic sig selv) Peter envies *REFL I REFL-SELF I Mary 'Peter envies himself I Mary.' (I 0) "Inherently reflexive" predicates: a. Peter skammer sig /*sig selv /*Maire. (non-thematic sig) Peter shames RELF I *REFL-SELF I *Mary 'Peter is ashamed of himself I Mary.' b. Peter dukker sig I *sig selv I *Marie. (thematic sig) Peter ducks REFL I *REFL-SELF I *Mary 'Peter ducks *himself I *Mary.' 235 The distribution of different types of nominal expressions in object position with the different verb types is summarized in (II). (II) Distribution of nominal expressions in object position: a. Anti-reflexive *sig I b. Neutral sig I c. Inherently reflexive sig I sig selv I sig selv I *sig selv/ DP DP *DP The relation between the semantic properties of predicates and adnominal intensification of simple reflexives has already been discussed in great length in chapter 3 and will not be repeated here. In the rest of this chapter the focus will be on the types of predicates which allow simple sig, i.e. neutral (8) and inherently reflexive predicates (10). We argue that "neutral" transitive verbs retain their transitivity even when they occur with the simple reflexive sig, see (8) as well as (3a) vs. (7). In contrast, other approaches (e.g. Reinhart and Reuland I993, and others) assume that all the "neutral" verbs occur twice in the lexicon: once, as a true transitive verb which require the intensified reflexive sig selv, e.g. (7), and once as an inherently reflexive verb which does not need to be overtly reflexive-marked by the SELF element of the complex reflexive, e.g. (3a). We argue that such lexical ambiguity is both unwarranted and unnecessary. That is, neutral predicates can allow simple sig as direct object without being "inherently reflexive". In section 4.2. a number of syntactic and semantic tests will be used to compare and evaluate these competing views of"neutral" predicates and it will be argued that the approach adopted here makes it possible to arrive at a simpler, more 236 economical, account of binding which does not need to stipulate double entries for neutral verbs. The intensified form of the reflexive, i.e. sig selv, has been claimed always to be thematic. See section 3.3.6.2 in chapter 3 and section 4.3.3 in chapter 4 for discussion of instances of non-thematic sig selv, where the presence of selv is triggered by prosodic rather than semantic factors. Section 4.3 contains a discussion of different types of "inherently reflexive" predicates, i.e. predicates which can only occur with sig, see (I 0). It will be argued that the large majority of inherently reflexive predicates are of a type of "deponent" unaccusative verbs where sig functions as a non-thematic marker of unaccusativity occurring in the v of the voice projection vP, cf. (2b ), (I Oa). However, there are also a number of inherently reflexive predicates which fail to behave as unaccusatives with respect to different syntactic and semantic tests. These predicates thus appear to be semantically inherently reflexive but still fully transitive predicates which assign two theta-roles but which, due to strict semantic selection restrictions, only allow simple reflexives as internal arguments (2a), (I Ob ). From a binding-theoretical point of view, one of the main interest of this chapter is that it addresses the issue of where to draw the dividing line between thematic and non-thematic uses of reflexives, e.g. while Reinhart and Reuland (1993) consider both instances of sig in (1 Oa,b) to be thematic, we argue that sig in ( 1 Oa) in non-thematic and that the sig in (I Ob) is thematic. We also disagree with Vikner ( 1985) who would consider both instances in (1 Oa,b) as well all the instances of sig with neutral predicates, e.g. (3a,b) and (8a,b ), to be non-thematic grammatical markers 1 • 1 In these respects Jakubowicz ( 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. 237 The problem posed by the fact that there is no formal distinction between reflexives and pronouns in the 1st and 2nd person is taken up in section 4.4 where we present a late- insertion analysis which accounts for these facts without stipulating the existence of multiple lexical entries. Like the adnominal intensifier selv 'self discussed in chapter 2, the reflexive element sig can be found in a number of different uses with what seems to be widely different syntactic and semantic properties. In section 4.5 the full range of uses of reflexive elements (e.g. in reciprocal, middle, passive, etc. constructions) will be briefly discussed. This section is largely descriptive and its main function is to situate the uses of sig as thematic reflexive and non-thematic marker of unaccusativity in the context of the full range of uses of reflexive elements. As mentioned in chapter I, a complete mapping of the full ranges of uses of both the intensifier selv and the reflexive sig is a necessary prerequisite to our understanding of the behavior ofthese elements independently as well as the instances where they combine to form the so-called complex reflexives. Our analysis of complex reflexives as intensified nominal expressions thus relies on carefully distinguishing between the contributions of different modules of the grammar, see (12). (12) a. b. c. Module: intensification binding theory argument structure, voice, etc. Accounts for: distribution of selv distribution of thematic sig distribution of non-thematic sig Finally, section 4.6 summarizes the results of the previous sections and concludes the chapter. 4.2 Testing the lexical/syntactic status of sig with neutral verbs In this section a number of syntactic and semantic tests will be used to show that simple sig occurring with neutral predicates is best analyzed as a theta-role receiving internal argument of a transitive predicate. As defined in chapter 3, neutral predicates are predicates which are 238 semantically compatible with both reflexive and non-reflexive scenarios, and which therefore are neutral with respect to what types of nominal expressions can occur as internal arguments. The grooming activity predicates in (13), exemplify proto-typical neutral verbs. (13) Grooming verbs: a. barb ere 'shave' b. vaske 'wash' c. frisere 'dress one's hair' d. bade 'bathe' e. rede 'comb one's hair' f. tgrre 'dry' g. pynte 'adorn' h. klcede _pa 'dress' i. klcede af 'undress' J. afluse 'delouse' k. b0rste 'brush' I. sminke 'make up' m. frottere 'rub (with a towel)' Verbs denoting non-translational movement and verbs denoting transformation also constitute core groups of neutral predicates, see (14) and (15). (14) Movement verbs: a. strcekke 'stretch' b. bevcege 2 'move (non-translationally)' c. flytte 'move (translationally)' d. dreje 'turn' e. b0je 'bend' (15) Transformation verbs: a. forvandle 'transform' b. udvikle 'develop' c. cendre 'change, alter; modify' d. for andre 'change, alter' There are also a number of neutral verbs falling into neither of these categories, see (16). 2 Only DPs denoting inalienably possessed object can occur as internal arguments of bevcege ·move'. see (i). In contrast,flytte ·move allows both alienable and inalienable objects, see (ii). (i) Peter bewegede handen!fodenlstolenlglasset. ·Peteri moved his hand/hisi foot/* the chair/*the glass.' (ii) Peter bevcegede handen!fodenl*stolenl*glasset. ·Peter, moved his hand/hisi foot/the chair/the glass.' 239 (16) a. befri 'liberate' b. forsvare 'defend' c. redde 'save' d. etc. Finally, the class of neutral predicates also include all the "hidden neutral predicates" which tend to pattern as "anti-reflexives" (i.e. disallowing simple sig) although they do not presuppose (representational) non-identity of their arguments, see discussion of koge 'boil' in chapter 3, section 3.3.2.3. By using syntactic and semantic tests showing that neutral verbs allow locally bound thematic sig we argue against analyses which are - in one way or the other - based on the alleged anti-locality of sig (e.g. Vikner 1986, etc.), as well as analyses based on the assumption that all neutral verbs are lexically ambiguous (e.g. Reinhart & Reuland 1993). 4.2.1 Testing for agentivity Testing for agentivity is one way to determine whether a given predicate is causative transitive or decausative inchoative/unaccusative. As observed by Waltereit (1999:269), the French transitive verb tuer 'kill' can be used with the reflexive se 'REFL' in two different ways, i.e. (i) as a causative transitive verb (17b,c), or as (ii) a deagentive inchoative (17a). (17i a. b. Pierre s 'est tue dans un accident. Pierre REFL is killed in an accident 'Pierre has died in an accident.' Pierre s 'est tue. Pierre REFL is killed ([-agentive], unintentional death) ([+agentive], intentional death) 'Pierre killed himself/'Pierre committed suicide.' c. Pierre a tue lejuge. ([+agentive], intentional death) Pierre has killed the judge 'Pierre has killed the judge.' In (17a) se tuer occurs in a context which excludes agentivity, i.e. the adverbial phrase dans un accident 'in an accident', and as consequence the verb tuer+se takes on the deagentive 3 These examples in and the following discussion thereof is closely inspired by Waltereit ( 1999). 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 meaning '(intentionally) kill (somebody/oneself)' just like the non-reflexive scenario tuer+DP in (17c). In French, the difference between the true thematic reflexive (17b) and the deagentive inchoative uses of verb non-thematic se (17a) can usually only be determined by the presence vs. absence of 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 Danish ( 18a) which have causative transitive counterparts ( 18b,c ). (18) a b. Peter slog sig (pa bordkanten). ([-agentive], unintentional event) Peter hit REFL on table-edge 'Peter (accidentally) hurt himself(on the edge ofthe table).' Peter slog sig selv. ([+agentive], intentional event) Peter hit REFL self 'Peter (intentionally) hit himself.' c. Peter slog Hans. ([+agentive], intentional event) Peter hit Hans 'Peter (intentionally) hit Hans.' Notice that unlike French, which uses the same form of the reflexive, viz. se 'REFL', in both the deagentive inchoative sentence in ( 17a) and the causative transitive reflexive construction in (17b), Danish tend to use the simple form sig 'REFL' in the deagentive inchoative sentence in (18a) and the complex form sig selv 'REFL self in the causative transitive reflexive construction in ( 18b ). The preference for sig selv over simple sig in causative transitive reflexive constructions involving the predicate sla 'hit' is due to the fact that this particular predicate is a so-called "hidden neutral" predicate, i.e. a predicate which - although not semantically incompatible with reflexive scenarios - is most often used as an other-directed predicate and thus triggers the expectation that its arguments denote (representationally) different entities. The verbs sla 'hit' in (18) and brcende 'bum' in (19) thus belong to a subset 241 of verbal predicates which display the same meaning difference between verb+sig (= unintentional, [-agentive]), and verb+sig selv/DP (=intentional, [+agentive]) as the French examples in ( 17a-c) discussed above. (19) a Peter brcendte sig pa ovnen. ([-agentive], unintentional event) Peter burned REFL on oven-the 'Peter (accidentally) burned himself on the oven. b. Peter brcendte sig selv. ([ +agentive ], intentional event) Peter burned REFL self 'Peter (intentionally) burned himself.' c. Peter brcendte Hans. ([+agentive], intentional event) Peter burned Hans 'Peter (intentionally) burned Hans.' Notice furthermore that, although complex sig selv is preferred with the hidden neutral predicates sla 'hit' and brcende 'bum' in (18-19b), it is possible to find agentive, transitive uses of these predicates with simple sig. Adding an instrumental (and/or locative) phrase to the sentences in ( 18-19b) makes simple sig more acceptable even with the agentive reading, compare ( 18b) vs. (20a) and ( 19b) vs. (20b ). (20) a. b. Peter slog sig (i hovedet) (med en hammer). (agentive, intentional) Peter hit REFL (in head-the) with a hammer 'Peter (intentionally) hit himself (in the head) with a hammer.' Peter brcendte sig (pa armen) (med en cigaret). (agentive, intentional) Peter burned REFL (on arm-the) with a cigarette 'Peter (intentionally) burned himself(on his arm) with a cigarette.' Examples like ( 18-20) illustrate the difference between deagentive inchoative verbs denoting unintentional events and their causative transitive counterparts which denote activities involving two entities, usually an AGENT doing something to a PATIENT/THEME. Since the sig in ( 18-19a) occurs with the intransitive predicates sta/brcende 'hit/bum', it is non- thematic and should be analyzed as in (2b ). In contrast, the occurrences sig selv"' (18-19b) and 4 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 of non-thematic intensified reflexives are discussed. 242 sig in (20a-b) are thematic and receive a theta-roles from the transitive versions of the predicates sla/brcende 'hit/burn'. As mentioned above, Holmberg (1984) and Vikner (1985) assume simple sig to be anti-local. That is, it can only be thematic when LD-bound. For them all instances oflocal sig thus have to be non-thematic. Consequently, in their system, only complex reflexive (e.g. sig selv) can be locally bound. They thus explain the difference between ( 18a, 19a) and ( 18b, 19b) by reference to a difference in thematic structure. In ( 19b, 19b and (18c, 19c) the subject DP Peter receives the AGENT theta-role and the internal arguments sig selv/Hans receive the THEME theta-role. In the (a) cases the verbs brcendelsla only have one theta-role to assign, i.e. the THEME theta-role assigned to the subject DP Peter. In other words, their analysis is based on the assumption that local sig is "some sort of detransitivizing element that somehow prevents the assignment of the theta-role that would otherwise have been assigned to the subject. In accordance with Burzio's generalization (cf. e.g. Burzio (1981), Haegeman (1985)) this in tum prevents the deep structure object from getting object case, and it therefore has to move to subject position to be case-marked, taking along its object theta-role, in this case THEME, parallel to the analysis of passive in LGB: 124ff."(Vikner (1985:50, footnote 8)). While 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 of local sig. As shown in (20), agentive readings of the predicates sla/brcende 'hit/bum' can be found even with locally bound sig. Additional counter-examples to their theory are provided by numerous cases of sig with neutral predicates, e.g. vaske 'wash' in (21 ). (21) a. b. c. Peter vaskede sig. ([+agentive], intentional event, [-contrast]) Peter vaskede sig selv. ([+agentive], intentional event, [+contrast]) Peter vaskede Hans. ([+agentive], intentional event) Peter washed REFL/REFL self/Hans 'Peter washed 0/himself/Hans.' 243 The difference between (21 a) and (21 b) is not a difference of agentivity/intentionality - as would be predicted by Vikner's (1985) approach- but rather one of [+/-contrastive focus] 5 or [+/-doppelganger-effect]. In other words, Holmberg's and Vikner's reanalysis account of sla sig and bra::nde sig cannot be extended to the neutral verb vaske 'wash' which has [+intentional, +agentive] readings with both the simple reflexive sig and the complex reflexive sig selv. The sentence in (21 a) does not mean that Peter accidentally experienced a washing event that somehow 'happened' to him. Proof of this is provided by the fact that while the sentences in (21 a-c) are all compatible with adverbials like med vilje 'deliberately' or purpose clauses like for at 'in order to' only the (b) and (c) examples in (J 8) and (19) combine felicitously with such expressions. Vikner (1985) acknowledges that such examples constitute a potential problem for his approach. In his analysis local sig is a non-thematic, detransitivizing element which reduces the number of theta-roles assigned by a predicate by one. While this analysis is justified for the examples in ( 18a) and (19a), there is no evidence supporting its extension to cases like (21) and (22). (22) a. b. c. .... at [s Peter, AGR satte ' ... that [s Peter AGR set sig, sig; selv Michaelz ned] down]' Vikner's account of(22) is as follows: "the difference in meaning is that where [(22a)] means straightforwardly "Peter sat down", [(22b)] means something like "Peter sat himself down" (pragmatically a bit odd, as if he had set himself down by means of e.g. a crane). This difference would seem to be accounted for by assuming the following theta-roles: in [(22b)] Peter is AGENT, sig selv is THEME (parallel to [(22c)]), whereas in [(22a)] Peter is AGENT, and there is no THEME as such. If [(22a)] is compared to [(18/19a)], they have in common that sig ensures that only one theta-role is assigned, but they differ in which one it is. This difference is supported by my intuition that [(22a)] which has an AGENT may be followed by a purpose clause (e.g. "in order to impress everyone in the room"), which is not possible for the AGENT-less [(18/19a)]. Similarly [(2la)] 5 For discussion of adnominal intensification of reflexives used to express contrastiveness or doppelganger-effects. see chapter 3, section 3.3.2.5 and 3.3.2.7 respectively. 244 but not [( 18/19a)] may contain an adverbial like med vilje "deliberately"." (Vikner ( 1985:50-51 ,footnote 9) Vikner's analysis seems rather ad hoc. While assuming that the reduction in arity and the absence of agent in ( 18a) and ( 19a) to be due to Burzio's generalization seems well justified, Vikner does not provide any evidence supporting the assumption of a different arity reduction rule responsible for the alleged absence of the THEME rather than the AGENT role in (22a). Given what we now know about doppelganger-effects, it seems quite obvious that the meaning difference between (22a) and (22b) is due to the doppelganger reading triggered by the intensified reflexive sig selv, not ad hoc deletion of the THEME theta-role. This analysis thus eliminates the need for stipulating any additional ad hoc arity-reduction rule responsible for eliminating the THEME theta-role while preserving the AGENT theta-role. Just like with any other transitive neutral verb, e.g. vaske 'wash' in (21 ), the theta-grid of sa:tte _ ned 'sit _ down' remains the same (i.e. AGENT, PATIENT/THEME) regardless of what kind of nominal expression occupies the position of 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 THEME theta-role to the direct object. In other words, there is no special arity reduction rule at work in (22a). The simple reflexive sig simply receives the theta-role THEME while the subject gets the AGENT theta-role. Similar arguments can be leveled against Lidz (1996) who assumes local simple sig to be a non-thematic, verbal reflexive. Like Vikner, Lidz runs into trouble with the neutral predicates which, as shown above, allow locally bound thematic sig. On the basis of such examples we conclude that neutral predicates (e.g. vaske 'wash', etc.) remain agentive causative transitive predicates even when the simple refl~xive sig 'REFL' occupies the position of direct object. In other words, in spite of what has been the accepted account for decades, simple sig CAN be an theta-role receiving argument even when it is locally bound by a co-argument. 245 4.2.2 Testing for compatibility with formation of agentive nouns in -er Formation of agentive nouns from verbal roots has sometimes been used to test whether a verb is unaccusative, e.g. Evereart ( 1986). It has been claimed that verbs taking the simple reflexive, e.g. Danish sig, Dutch zich, etc., also do not allow -er affixation, thus indicating that they are unaccusative ( cf. Lidz (I 996: 132) following Evereart ( 1986)) Lidz ( 1996) bases his argument on examples such as (23)-(26) which show that unaccusative verbs disallow -er suffixation, while transitive and unergative verbs allow it. (23) Unaccusative: a. Han er faldet. he is fallen 'He fell.' (24) Transitive: a. Han spiser. he eats (25) Unergative: a. Han danser he dances b. b. b. *Enfalder. a faller en (hvidfggs)spiser a (garlic )eater en danser a dancer (26) a. Hanforspiste sig (i salat). he over-ate REFL in salad 'He overate on the salad.' b. *enforspiser an overeats Lidz claims that verbs taking sig behave like unaccusatives by disallowing the formation of- er nouns. While 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), which allow formation of agentive nouns in -er in spite of the fact that they can all be found with locally bound simple sig, see (28). (27) a. en vasker b. (nggen-)bader c. en ajluser d. en (falck-)redder e. en befrier f. en forsvarer 'a washer' 'a (naked-)bather, a skinny-dipper' 'a delouser' '(lit.) a saver, i.e. a salvage-corps man' 'a liberator' 'defender, defense player (football) defense lawyer' 246 (28) Peter vaskerlbader/afluser/etc. sig. Peter washes/bathes/delouses REFL 'Peter washes/bathes/delouses/etc. (himself)' Now, if one accepts the lexical ambiguity hypothesis ( cf. Reinhart & Reuland (1993)), one could object that the verbs in (27) all have transitive forms and that the -er nouns in (30) are formed from the transitive fonn of the verb rather than the inherently reflexive form of the verb, which is the one found in (28). The problem with this explanation (besides the obvious problem of having recourse to unmotivated lexical ambiguity) is that it does not work for verbs like (29), which do allow the formation of -er nouns, see (30), in spite of their inherently reflexive nature, i.e. they allow nothing other than sig in object position, see (29). (29) Peter bosatte *0/sig/*sig selvi*Hans i Afrika. Peter settled REFLI*REFL self/*Hans in Africa. 'Peter settled 0/*himself/*Hans in Africa.' (30) Peter er en af de indiske bosrettere i Afrika. Peter is one ofthe Indian settler-s in Africa 'Peter is one ofthe Indian settlers in Africa.' In other words, the -er test does NOT allow us to conclude that all verbs taking local sig behave like unaccusatives. While it appears to work for the verbs of the type illustrated in (26), it does not work for (27-30t Furthermore, as discussed in section 4.3.2, the fact that a given verb does not allow formation of agentive nouns in -er does not necessarily mean that it is unaccusative. Many anti-reflexive transitive verbs do not allow formation of agentive nouns in -er, e.g. mistamke 'suspect', *en mistcenker '*a suspecter'. In other words, there is no guarantee that the reason why a given verb cannot fonn agentive nouns in -er is because it is unaccusati ve. 6 The question of how to analyze ~er nominalizations is also discussed in chapter 7. Agentive nouns of 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. bosretter ·settler' vs. *selvbosretter 'self-settler'. Notice that unlike ~er nouns derived from anti-reflexive verbs. e.g. morder 'murderer'. bosretter ·settler· only has reflexive readings (i.e. refers to someone who settles (himself) 247 4.2.3 Non-compositional meaning change and the question test Neutral verbs have the same meaning when they are used in reflexive constructions as when they are used in non-reflexive transitive contexts, compare (31 a) and (31 b). (31) a. b. Peter vasker/barberer Hans. 'Peter washes/shaves Hans.' Peter vasker/barberer sig. 'Peter washes/shaves (himself).' (neutral) (neutral, thematic sig) That is, the meaning of the verb+reflexive in (31 b) can be derived compositionally from the meaning of the verb and the meaning of the reflexive sig in the same way as the meaning of the verb+DP in (3la) can be derived compositionally from the meaning of the verb and the meaning of the DP in the object position. In contrast, the anti-reflexive predicate bcere 'carry' does not have the same meaning in (32a) and (32b). (32) a. Peter bar kufferten. Peter carried suitcase-the 'Peter carried the suitcase.' (anti-reflexive) b. Peter bar sig med star vcerdighed. (anti-refl. reanalyzed as inher. refl., Peter carried REFL with great dignity non-thematic sig) *(i) 'Peter<real> carried himself<statue>.' *(ii) 'Peter<real> carried himself<real>.' (Mlinchhausen story) o\iii) 'Peter behaved carried himself/behaved with great dignity.' c. Peter bar sig selv. (anti-reflexive, thematic sig) Peter carried REFL self ok(i) 'Peter<real> carried himself<statue>.' ok (ii) 'Peter<real> carried himself<real>.' (Mlinchhausen story) *(iii) 'Peter behaved carried himself/behaved with great dignity.' In (32a) the predicate bcere 'carry' occurs in its normal use as an anti-reflexive transitive verb meaning to 'carry (something/somebody)'. In (32b) it is used together with the reflexive sig in the meaning 'carry oneself; behave'. Unlike (31 b) this meaning cannot be obtained compositionally from the basic meaning of the verb+reflexive pronoun. Such non- compositional meaning differences have often been used as argument in favor of assuming that the verb+sig combination 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 248 that bcere is lexically ambiguous between a transitive verb with the meaning 'carry' and an inherently reflexive with the meaning 'behave.' An additional argument in favor of assuming bcere sig 'carry oneself; behave' to be reanalyzed as a reflexive verb is that, unlike the reflexive pronoun sig in (31 b), the sig in (32b), is not thematic. In both (32a) and (31 a, b) Peter is actually washing an object. In (31 b) this object being washed is identical to the washer himself; in (31 a) it is different. In (32a) the direct object DP kufferten 'the suitcase' serving as direct object of bcere 'carry' does refer to a particular entity in the universe of discourse, namely 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 of discourse; in particular it clearly does not share the reference of its antecedent Peter. Had this been the case the sentence would have meant that Peter carried himself (on his back), as in a Mlinchhausen story, see (32c), which is clearly not the meaning of(32bf_ Waltereit ( 1999:260) proposed to use questions to test whether sig is thematic or non- thematic in a given verb+sig combination. When the referent of the reflexive can be questioned, as in (33), then the reflexive is thematic. Conversely, when replacement of the reflexive with a question word yields unacceptable sentences then the reflexive is non- thematic, see (34 ). (33) 8 Marie se regardedans Ia Marie REFL looks m 'Marie looks at herself in the mirror. (34) Jean-Luc s 'est rendu Jean-Luc REFL returned 'Jean-Luc went to the station. glace. Elle regarde qui? the mirror she looks whom? a Ia gare. to the station Whom does she look at?' #Il a rendu qui? He has returned whom #He returned whom?' inherently reflexive predicate bosa:tte ·settle' is also discussed in section 4.3.2. 7 For more discussion of the relationship between Munchhausen readings/doppelganger-effects and intensified reflexives, see chapter 3 section 3.3.2.7. 8 These examples are adapted from Waltereit (1999:260), examples (2) and (4). 249 As shown by the examples in (35) and (36), the same question test, can be used to test the thematic status of sig in Danish. (35) Marie vaskede sig i varmt vand. Hvem var det hun vaskede? Marie washed REFL in warm water who was it she washed? 'Marie washed (herself) in warm water. Who did she wash?' (36) Marie bar sig med star vrerdighed. #Hvem bar hun med star vrerdighed? Marie carried REFL herself with great dignity who carried she with great dignity 'Marie carried herself with great dignity. #Who did she carry with great dignity?' An inherent weakness to this test lies in the fact that one can only ask questions about the parts of a sentence which can be focused. Since inherently reflexive verbs only allow expressions that are coreferential with the subject to occupy the position of direct object, it is in principle impossible to ask a meaningful question about the referents of the expression (which, due to the semantic selections of the predicate, will always be a reflexive pronoun) filling the object position of an inherently reflexive predicate. 4.2.4 Testing for compatibility with impersonal passivization To further support the claim that verbs taking local simple sig behave as unaccusatives Lidz ( 1996) adduces examples such as (37-39) to show that, just like uaccusatives (37-38), "verbs which take [sig] also disallow impersonal passivization [, see (39)]" (p. 133). The observation that unaccusative verbs disallow impersonal passivization while unergatives allow it is due to Perlmutter ( 1978). Lidz's Dutch examples have here been translated into Danish for ease of exposition. (37) a. (38) a. Han Iober. he runs 'He runs.' Hanfalder. he falls 'He falls.' b. b. Der bliver lobet. there becomes run 'People are running' ( unergati ve) *Der bliver faldet. (unaccusative) there becomes fallen 'People are falling.' (39) a. Hanforspiste sig. He overate REFL 'He overate.' b. *Der bliver forspist. there becomes overeaten 'People are overeating.' 250 On the basis of such data, Lidz concludes that the occurrences of sig in examples such as (39) are not anaphors but non-thematic, non-referential verbal reflexives. While we agree that this holds for the verbs in (39) it is certainly not the case for neutral verbs. Neutral predicates allow impersonal passivization, see ( 40), although they also allow locally bound thematic sig, see ( 41 ). ( 40) Der blev vasket/badet/etc. there became washed/bathed/ 'There was washed/bathed.' ( 41) Peter vasker/bader/ajluser/etc. sig. Peter washes/bathes/delouses REFL 'Peter washes/bathes/delouses himself.' These examples clearly show that impersonal passivization cannot be used to prove that ALL verbs which allow local simple sig are unaccusative. The generalization that verbs taking simple sig also disallow impersonal passivization could, of course, be saved by assuming that neutral verbs like vaske 'wash' 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 4.2.5-1 0, there are a number of facts which go against this assumption and which are more readily explainable under the assumption that neutral verbs like vaske 'wash' and bade 'bathe' are entered only once into the lexicon, i.e. as causative transitive verbs consistently assigning the theta-role THEME to their internal objects (be they DPs, pronouns or the simple reflexive sig). The example in ( 42), originally from Hell an ( 1988), which Lidz uses to argue that verbs which take simple sig are unaccusative, is rather curious. (42) *Der blev vasket sig (afmanden). there became washed REFL (by man-the) 251 This example appears to be flawed. Ifthe point is to show that vaske 'wash' is not a causative transitive verb which may allow sig as a true thematic anaphor in argument position, then one should not use a sentence like (42). Indeed ifvaske 'wash' is truly a causative transitive verbs in all its uses (i.e. even with simple sig) then one would expect it to be ungrammatical in impersonal passive constructions WITH the reflexive pronoun, since all causative transitive verbs behave in the same way, see ( 43). (43) a. b. *Der blev dolket sig (selv). there was stabbed REFL self *Der blev spist/vasket sig (selv). there was eaten/washed REFL self (dolke 'stab'= anti-reflexive) (vaske/spise 'wash/eat' =neutral) In other words, rather than showing that vaske 'wash' behaves as an unaccusative verb, ( 42) shows that it behaves just like any other transitive verbs, see also ( 44), which furthermore underlines the parallels between vaske 'wash' and the other transitive verbs. (44) a. Der blev spist/vasket. there was eaten/washed b. Der blev spist/vasket kartofler. there was eaten/washed potatoes Furthermore, as discussed in section 4.3.2, the fact that a given verb does not allow impersonal passivization does not necessarily mean that it is unaccusative. Many anti-reflexive transitive verbs, e.g. ( 45a), cannot felicitously form impersonal passives, e.g. ( 45b) (45) a. b. Peter mistcenkte *sig I sig selv !Marie. Peter suspected REFL I REFL self I Marie 'Peter suspected himself I Marie.' *Der blev mistcenkt. there was suspected In other words, there is no guarantee that the only reason a given verb cannot undergo impersonal passivization is because it is unaccusative. So this test should be used with caution when applied to inherently reflexive verbs. That is, while it seems plausible to assume that the 252 inherently reflexive verb forspise 'overeat' is unaccusative, the mere fact that it cannot form the impersonal passive (39b) cannot be used as solid proof of its unaccusativity. 4.2.5 Auxiliary-selection A further indication that sig-form verbs do not necessarily behave as unaccusatives comes from auxiliary selection in Danish. Unaccusative verbs take at were 'to be' as auxiliary, see (46), while unergatives and transitive verbs take at have 'to have', see (47) and (48t ( 46) Unaccusative predicates: Han er ankommet. he is arrived 'He has arrived.' 9 Alan et al. ( 1995) provide a conveniently concise description of the use of the verbs have ·have' and va>re '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 la>st bogen. 'I have read the book.' Vi har brugt pengene. ·We have spent the money.' Hun har vente! pa dig. "She has waited for you.' Vasen har staet her. 'The vase has stood here.' Notice that have and va>re both use have as their auxiliary. Vi har haft en dejligferie. 'We have had a lovely holiday.' Vi har va>ret i Italien. ·we have been to Italy.' (ii) Intransitive verbs expressing some kind of motion use va>re as their auxiliary (but sec (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 aj? ·Has he run away?' Notice that blive "be. become' uses va>re as its auxiliary: Haner blevet professor. ·He has become a professor.' Del 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 har jlyttet bordet. (trans.) 'I have moved the table.' Deer jlyttet til A rhus. (intrans.) "They have moved to Arhus.' Han har jlejet en helikopter. (trans.) "He has flown a helicopter.' Haner jlejet 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 va> re): Haner gaet 10 kilometer. Nu er han gaet. Hun har svommet over Kana/en. ·He has walked I 0 kilometers.' "Now he has gone/left.' 'She has swum the Channel.' Hun er svemmet va>k. 'She has swum away.' Note- The verb begynde ['begin'] can be used with either have or va>re. Have is used when there is a direct object, e.g.: Jeg har begyndt oversa>ttelsen. 'I have begun the translation.' Jeg er begyndt pa oversa>ttelsen ... (Alan et al. (1995:263-264)) ( 4 7) Unergative predicates: Han har talt. he has spoken 'He has spoken.' 253 However, none of the sig-form verbs, regardless of whether sig is thematic ( 48) or non- thematic ( 49), take at were 'to be' as auxiliary; they all take at have 'to have'. ( 48) Causative transitive predicates: (49) a. Peter har vasket sin bil. Peter has washed POSSREFL car 'Peter has washed his car.' b. Peter har vasket sig. Peter has washed REFL 'Peter has washed himself.' Inherently reflexive predicates: a. Peter har forspist sig i cebler. Peter has over-eaten REFL in apples 'Peter overate in apples.' b. Peter har ofte skammet sig. Peter has often shamed REFL 'Peter has often been ashamed.' (compare with (26b) and (39b)) As shown in ( 49), with respect to auxiliary selection inherently reflexive predicates, regardless of whether they have transitive counterparts ( 49a) or not ( 49b ), all behave like unergative ( 4 7) and causative transitive predicates ( 48) which take have 'have', rather than like unaccusative predicates like ankomme 'arrive' in (46), which take vcere 'be'. The auxiliary selection test for unaccusativity was originally developed for Romance languages, e.g. French where unaccusative predicates, as well as all reflexive predicates, take the auxiliary etre 'be', while causative transitive predicates form composite tense forms with the auxiliary avoir 'have'. Clearly, the discrepancies between Germanic (e.g. Danish examples discussed above) and Romance languages show that auxiliary-selection cannot be used as a cross-linguistically reliable test for unaccusativiti 0 • 254 4.2.6 Ledrups tests According to Lodrup ( 1999), there is clear evidence for the status of simple anaphors as theta- role receiving direct objects with neutral predicates. In Norwegian, for instance, a referential seg 'REFL' (cognate of Danish sig 'REFL') can be modified by the modifier hele 'whole', see (50a-b),just like full lexical DPs (SOc). (50) Norwegian: a. Hun vasker sig. She washes REFL 'She washed (herself).' b. Hun vasker hele sig. She washes whole REFL 'She washed all of herself.' c. Hun vasker hele bilen. She washes whole car-the 'She washed all of the car.' Furthermore, sentences with thematic seg do not allow existential constructions, in contrast to sentences with inherently reflexive verbs which do, see (51 a) vs. (Slb,c). (51) Norwegian: a. * Det vasket seg en dame i sjon. it washed REFL a lady in sea-the 'There was a lady washing in the sea.' b. Det lukket seg en dor. it closed REFL a door 'A door closed.' c. Det i nfannt seg en dame pa kontoret. it appeared REFL a lady on office-the 'A lady appeared in the office.' Based on data such as (50) and (51) Lodrup concludes that local seg 'REFL' may function as a theta-role receiving internal argument. Due to minor parametric variation the examples in (50)-( 51) cannot be replicated in Danish. However, Danish and Norwegian (especially bokmal Norwegian) are very closely related languages. In almost all other respects, the distribution of the simple reflexive seglsig 1 ° Cf discussion of auxiliary selection in Borer (2002) and Schein (2002) in G. Preyer and G. Peter (eds.) Logical Form and Language. Oxford University Press. Oxford. 255 follow the same patterns in the two languages. Hence, the Norwegian data may be used as indirect evidence supporting the idea that local sig can be thematic with neutral verbs. 4.2. 7 German sich is not anti-local Konig & Siemund (1999:63) also argue strongly against the idea that simple reflexives are anti-local: "To group these referential uses of the simplex anaphors together with the inherently reflexive verbs and analyze them as cases of reanalysis and detransitivization is particularly absurd for a language like German, where sich is practically the only reflexive marker and is only combined with the intensifier selbst in the most extreme cases of other- directed 11 predicates[, see (52)]." (52) German: a. Paul stolperte iiber #sichlsich selbst. Paul stumbled over REFLIREFL self 'Peter stumbled over himself.' b. Maria traf #sichlsich selbst. Maria met REFLIREFL self 'Maria met herself.' As illustrated in (53)-(54), verbs which are anti-reflexive or 'hidden neutrals' in Danish, may behave as neutral predicates in German. (53) Danish: Paul hader *sig I sig selv. Paul hates REFLIREFL self 'Paul hates himself.' (54) German: Paul hasst sich I sich selbst. Paul hates REFLIREFL self 'Paul hates himself.' (Danish, 'hidden neutral') (German, neutral) It thus appears, as observed by Konig and Siemund ( 1999:64 ), languages may vary with respect to the kind and number of predicates they classify as anti-reflexives: "[ ... ] There are languages like Dutch where simplex zich is restricted to only a few verbs and there are 256 languages like German, where a reinforcing selbst is rarely needed" (Konig and Siemund, 1999:64). In other words, a certain amount of cross-linguistic variation with respect to which verbs are classified as anti-reflexive, neutral, and inherently reflexive will have to be assumed. In order to account for German sich, Reinhart & Reuland (1993) would have to extend the lexical ambiguity analysis arbitrarily to include almost all predicates. In contrast, the analysis of anti-reflexivity in proposed in chapter 3 enables a more natural account of such cross-linguistic variation. At first glance the fact that a predicate like hate behaves differently in Danish and German seems to constitute problem for the account of intensifier-adjunction to reflexive pronouns given in chapter 3. If intensification of reflexives is triggered by the semantics of the predicate (i.e. the presupposition of non-representational identity of internal and external arguments) then one would expect a verb like hate to behave alike in different languages, and especially in the case of closely related languages like Danish 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 made between anti-reflexives and "hidden neutral predicates", like koge 'boil', which appear to behave like anti-reflexives but which (unlike the true anti-reflexives) can be made acceptable with simple reflexives given the appropriate context. In the case of true anti-reflexive predicates, e.g. stolpern uber 'stumble over' and treffen 'meet' in (52), the implicature of non-representational identity of the internal and external arguments is conventionalized (i.e. triggered by the lexical meaning of the predicate) and cannot be overridden by context. That is, it is impossible to imagine a person meeting himself or stumbling over himself without assuming some type of doppelganger scenario (e.g. schizophrenia or wax museum contexts). Hence, all languages which distinguish between unintensified and intensified reflexives like Danish and German are predicated to require intensified reflexives with semantically anti-reflexive predicates like (52)-(53) and (55). 11 The original passage has "'non-other-directed"'. This is clearly a typo and has been emended as ""other-directed··. (55) Danish: a. Peter snublede over *sig I sig selv. Peter stumbled over REFL I REFL self 'Peter stumbled over himself.' b. Peter traf*sig I sig selv. Peter met REFL I REFL self 'Peter met himself.' 257 In the case of the "hidden neutral", like hassen 'hate' in (54), the implicature of non- representational identity is triggered by pragmatically determined usage patterns (i.e. it is common/expected, but not logically necessary, for people to use a certain verb to express commonly 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 semantically vs. pragmatically triggered presupposition of non- representational identity makes it possible to explain why strongly anti-reflexive predicates require intensification of reflexives in both German and Danish while the "hidden neutral" predicates may vary cross-linguistically. This account is thus much less ad hoc than Reinhart and Reuland's ( 1993) account which do not explain why a certain predicates classified as anti- reflexive rather than neutral or inherently reflexive 12 • Another source of cross-linguistic variation is due to the syntactic and prosodic properties of the reflexive pronouns themselves. Danish sig is a syntactically free but prosodically bound form which cannot carry stress on its own and which has to be adnominally intensified by selv-adjunction in order to be heavy enough to carry stress, see 12 Note that there are a number of problems for this account of the differences between Danish and German, notably examples like (i) and (ii) which appear to constitute counter-examples to the generalization that semantically anti-reflexive 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 I sig selv. (Danish. anti-reflexive) Paul suspects REFLIREFL self 'Paul suspects himself.' (ii) Paul verdachtigte sich I sich selbst. (German, neutral) Paul suspects REFLIREFL self 'Paul suspects himself.' 258 chapter 3, section 3.3.6.1. In contrast, German sich 'REFL', is stressable, see section 3.3.6.3.2. See also discussion of stressability, focus, and intensification in chapter 3 section 3.3.6.3.4. The fact that German reflexives can be focused without adnominal intensification, may be one of the reasons why they may potentially obey anti-reflexivity and doppelganger requirements without adnominal intensification. See section 3.3.6.3.1-4 where we discuss the hypothesis that focus (realized as stress) and adnominal intensification are functionally equivalent, i.e. may occur in the same contexts with more or less the same meaning contribution. Of the two hypotheses concerning the differences between Danish an German the latter, i.e. the prosodic account also discussed in section 3.3.6.3.2-4 appear to be the most promising. However, due to space limitations we leave the elaboration of this topic for future research. 4.2.8 Constructions involving coordination ofverbs 13 • In this section we argue that data related to optional deletion of (all but the last) instance of simple sig with 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. While coordinated anti-reflexive and neutral verbs can optionally delete all but the last instance of simple sig, see (56), inherently reflexive verbs do not seem to be able to, see (57). (56) a. Peter b!ide vaskede (sig) og barberede *(sig) pa mindre end 5 minutter .. Peter both washed (REFL) and shaved REFL on less than 5 minutes We have also fond at least one native speaker who disagrees with Konig & 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. 13 This section is inspired by Jakubowicz ( 1994:209, (12, 14)). (57) 'Peter both washed (himself) and shaved *(himself) in less than 5 minutes' b. Peter havde vasket (.~ig) og t0rret *(sig) pa mindre end 5 minutter. Peter had washed (REFL) and dried REFL on less than 5 minutes 'Peter had washed (himself) and dried in less than 5 minutes.' c. Peter bad mig barbere (sig) og sminke *(sig). a. Peter asked me shave and make-up REFL 'Peter asked me to shave and put make-up on him.' Peter soledede *(sig) og boltrede *(sig) pa 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) pa stranden. Peter had tanned REFL and frolicked REFL on beach-the 'Peter had tanned and fro) icked on the beach.' 259 Indeed, inherently reflexive verbs with non-thematic sig are expected to behave in this way since the include sig as part of their lexical formatives. In contrast, since the neutral verbs in (56) behave like transitives they are expected to be able to omit all but the last instance of identical object DPs with coordinated verbs, compare (56a) and (58a), which show that the simple unintensified sig behave just like the DP kartojlerne 'the potatoes' in such constructions. (58) a. Peter blide vaskede og skra?llede kartojlerne pa mindre endfem minutter. Peter both washed and pealed potatoes-the on less than five minutes 'Peter both washed and pealed the potatoes in less than five minutes.' The fact that optional deletion of all but the last instance of sig is not allowed in the examples in (57) indicate that these instances of sig with the inherently reflexive verbs sole sig 'tan' and boltre sig 'frolic' have a different status from the sig in the examples in (56). That is, the sig in (57) is not a true thematic anaphor filling the argument position of two-place predicates, but rather a verbal reflexive left-adjoined to little v of the vP projection 14 • More examples of the 14 Note, 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 trom 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 trolicking on the beach.· 260 impossibility of deleting simple sig with coordinated inherently reflexive verbs are given in (59). (59) a. Peter brcekkede *Mg) og kom *(sig). Peter broke REFL and came REFL 'Peter vomited and recovered.' b. Peter havde brcekket *(sig) og kommet *(sig). Peter have broken REFL and come REFL 'Peter had vomited and recovered.' The fact that neutral verbs taking simple sig can be coordinated without repeating simple sig constitutes a serious problem for reanalysis accounts (e.g., Vikner 1985) and lexical ambiguity accounts of local sig ( cf. Reinhart and Reuland 1993), which are both based on the assumption that neutral verbs taking simple 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 when they take simple sig. While optional deletion of sig with coordinated predicates can be used to test whether a verb is inherently reflexive, it should be noted that there are cases where deletion of sig with conjoined inherently reflexive predicates may be marginally acceptable, or even quite felicitous, see (61 a). (61) a. (??)Peter havde tit bade skammet og cergret sig over sin mangel pa takt. Peter had often both shamed and annoyed REFL over his lack of tact 'Peter had often been both ashamed and vexed over his lack of tact.' b. #Peter skammer og cergrer sig. Peter shamed and chagrins REFL 'Peter is ashamed and vexed.' b. Hvorfor bra'kkede Peter sig? Why broke Peter REFL Why did Peter vomit?' 261 Notice however, that even in such cases, coordination of the same predicates is much less acceptable in the present tense, see (6\a), than in the composite tense in (60) 15 • See also section 4.3.2 for more discussion of the reliability of 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 main types, see (62)-(64). (62) Anti-reflexive ditransitive: a. Peter solgte *sig et hus. b. sig selv c. Mary 'Peter sold a house.' (63) Neutral ditransitive: a. Peter k0bte sig et hus. b. sig selv c. Mary 'Peter bought a house.' (64) Inherent))' reflexive ditransitive: a. Peter anskaffede sig et hus. b. *sig selv c. *Mary 'Peter acquired __ a house.' According to the reanalysis/lexical ambiguity approaches (cf. Vikner 1985, and Reinhart & Reuland 1993, Lidz 1996, etc.) one would have to argue that the verb in (63a) has somehow been reanalyzed as an inherently reflexive ditransitive predicate, while it remains a 'regular' di-transitive predicate in (63b-c). Notice, however, that there is no evidence supporting such claims. That is, there is no detectable difference in the meaning of the verb k0be 'buy' in (63a) vs. (63b,c). Nor are there any other indications that any 'reanalysis/lexical ambiguity' might have taken place in (63a) but not in (63b,c). We therefore conclude that these examples constitute strong evidence against all varieties of the 'reanalysis/lexical ambiguity' 15 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 of se improves when the conjoined 262 approaches. In contrast, the proposal defended here, namely that the distribution of intensified and unintensified forms of the reflexive follow directly from a combination of predicate meaning, pragmatics and world knowledge (see chapter 3), is strongly supported by the data in (62)-(64). That is, given what we know about selling transactions, namely that it is a proto- typical anti-reflexive/other-directed activity, it is highly unexpected that anyone would sell anything to himself. Hence, the intensified form ofthe reflexive is necessary to license (either through contrastive focus or doppelganger-effects (e.g. memory-loss or schizophrenia contexts)) the reflexive scenario, as shown in (62b). In contrast, buying is a neutral activity in that it is equally likely for someone to buy something for himself as it is for him to buy something for somebody else. Finally the impossibility of having non coreferential DPs in object position with inherently reflexive di-transitive predicates like anskaffe 'acquire', see (64c), explains why the adnominally intensified form of the reflexive, i.e. sig selv, is not possible either, see ( 64b ). 4.2.10 Non-co-argument bound local sig All the instances of locally bound thematic sig discussed above have co-argument antecedents. Depending on how the local domain relevant for binding is defined, locally bound simple reflexives, which are not bound by a co-argument, can be argued to be found in two contexts in Danish: (i) locational PPs, see (65), and (ii) possessive constructions, see (66). (65) Max, lagde bogen bag sig,. Max put book-the behind REFL 'Max put the book behind him( self).' In (65) relevant local domain 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 domain. That local binding predicates are in the past participle. see example ( 13c) chapter 3. section 3.2.3. 263 is possible in sentences like (65) falsifies the claim that simple reflexive pronouns in Danish are anti-local 16 • (66) Peter, vasker sin, hund. Peter washes POSSREFL dog 'Peter washes his dog.' According to the analysis of binding proposed in chapter 3, both sig 'RELF' in (65) and sin 'POSSREFL' in (66) are locally bound and thus constitute evidence supporting the claim that thematic simple reflexives can be locally bound. As argued in length in chapter 3, section 3.3.4 the existence of a morphologically specialized series of reflexive possessives (e.g. Dan. sin 'POSSREFL') constitute a major problem for predicate-based approaches to binding like Reinhart and Reuland (1993). 4.2.11 Summary: Falsifying the standard analyses of sig as anti-local In this section we have shown that a number of the tests which have been claimed to show that predicates taking simple sig are unaccusative do not hold up to scrutiny. In some cases, they turned up to be less reliable than previously assumed, in other cases they turned out to show the opposite of what they were originally intended to show. Furthermore, we have introduced new tests and phenomena which indicates that local sig can be thematic anaphors (e.g. with neutral predicates, in certain PPs and in possessive constructions). Taken together, the data and tests discussed in sections 4.2.1-1 0 thus adds up to a strong refutation of the claim that the simple 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. Distributionally, they can only be found with the simple reflexive sig, see (67). 16 Cf. Lidz ( 1996: 115-6) who uses similar arguments to argue against assuming Dutch zich 'REFL' to be anti-local. 264 (67) Peter skynder sig I *sig selv I *Hans. Peter hurries REFLIREFL self/Hans 'Peter hurries.' The verbal predicates listed in (68-71) are all distributionally inherently reflexive in that they are only found in the sig-form 17 • (68) Malefactive verbs denoting actions which involve hurting oneself by OVERDOING something: a. forbygge sig 'overbuild, build beyond one's means' b. fordriste sig 'venture to, make so bold as to' c. forhaste sig 'be in too great a hurry' d. for kg be sig 'overbuy (oneself), overpay' e. forlofle sig 'overstrain' f. forregne sig 'miscalculate' g. forslide sig 'overwork (oneself)' h. forsluge sig 'overeat (lit. over-swallow)' i. forslcebe sig 'overstrain oneself' J. forsnakke sig 'give oneself away (by revealing too much)' k. forspise sig 'overeat' I. forsynde sig 'offend, sin (against)' m. fortale sig 'make a slip ofthe tongue, give oneself away' n. forvilde sig 'lose one's way, go astray' o. forvrovle sig 'get into a muddle, bungled' p. forcede sig 'overeat' q. overanstrenge sig 'overexert oneself' (69) a. b. c. d. e. f. g. h. Verbs ofthinking, feeling: affinde sig med bekvemme sig til besinde sig bestrcebe sig *(forlp!i) betakke sig (for) bryste sig (aj) driste sig (til at) dummesig 'come to terms with, accept' 'strive to, endeavor' 'collect oneself, regain one's composure' 'strive to, endeavor to' 'decline (with thanks)' '(fig.) throw out one's chest, strut, swagger' 'venture to' 'make a fool of oneself, make a blunder' 17 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 I 000 dollars for bilen. he paid I 000 dollars for car-the ·He paid $1000 for the car.' In many cases the existence/absence of transitive counterparts seems like an arbitrary criterion. The verb be lobe sig til ·amount REFL to' seems to be of the same type as bet ale sig ·pay. be worthwhile' in (ia) above, in spite of the fact that it has no transitive counterpart, e.g. *Del beloher $/00 'it amount $100'. In this respect, the majority of all inherently reflexive verbs behave like the so-called deponent (s-torm) verbs discussed in section 4.4.5. In other words. verbs like belobe sig til 'amount to' can be characterized as deponent sig-fom verbs, which differ from s form deponent verbs only in the overt realization of the rc11exive element left-adjoined to the v of the vP projection. i. J. k. I. m. n. 0. p. (70) dy sig (for at inf) jlotte sig forbarme sig forelske sig (i) forscette sig (at) forgabe sig *(i) opfgre sig (godt/darligt) skamme sig 'behave oneself, contain oneself 'spread oneself, do it in style' 'take pity over' 'fall in love with' 'decide, resolve' 'fall in love with' 'behave well/misbehave' 'be ashamed' 265 a. b. c. d. e. f. g. Verbs of movement, location: albue sigfrem 'elbow one's way' 'ally oneself with' 'be; feel' (= resultative, see sect. 3.3.3) alliere sig med befinde sig begive sig boltre sig boscette sig skynde sig 'go, travel to' 'frolic, gambol' 'settle, set up house' 'be in a hurry' (71) "Middles" (with not tr. counterpart, opp. lukke 'close') only [-animate] subjects: a. forgrene sig 'branch of, ramify' b. ansamle sig 'gather' c. fortone sig 'loom; fade out of sight' d. befgbe sig (til) 'amountto' ( comp. betale sig vs. be tale ( tr.)) The division of the above inherently reflexive verbs in to four groups is impressionistic rather than based on rigorously defined semantic criteria. Furthermore, the lists in (68)-(71) are not exhaustive. Indeed, they were established simply by looking through all the verbs from A to F in a comprehensive Danish-English dictionari 8 (+a few additional verbs discussed above). The verbs in (68-71) all take [+animate] subjects while the verbs in (71) stand apart by only allowing [-animate subjects]. The verbs in ( 68) stand out both morphologically and semantically. Morphologically, they are all compounds composed of a prefix (either for- (=Ger. ver-) or over- (=Ger, iiber-)) 19 +a verbal stem. Semantically, they constitute a type of negative benefactives or malefactives, i.e. they denote actions/events which result in a negative experience on behalf of the subject. The semantic difference between (69) and (70) can be 18 Vinterberg. H, and C.A. Bodclsen (1966) Dansk-Engelsk Ordbog. Gyldendalske Boghandel, Nordisk Forlag. Copenhagen. 266 defined, admittedly in rather vague/impressionistic terms, as a difference between verbs involving cognitive activity on behalf of the subject (69) and verbs involving physical movement/location ofthe individual denoted by the subject (70). Rather than pursuing in any more depth the project of discovering semantically defined groups of inherently reflexive predicates, let us instead return to the more important question of the thematic status of sig with inherently reflexive predicates. That is, is it a thematic anaphor which occurs in argument position of an active transitive predicate (72a), or a non-thematic grammatical marker of unaccusativity occurring in the v of the voice-projection vP (72b). (72) a. Thematic sig: b. Non-thematic sig: VP I \ V' I \ V sig vP I \ v' I \ I \ VP sig v I \ .... V' I \ v In the following, we argue that the vast majority of inherently reflexive predicates have non- thematic sig (72b). In a sense, it would be more precise to refer to these verbs, e.g. skynde sig 'hurry', as deponent sig-form verbs, see section 4.4.5 for discussion of deponents-form verbs. In contrast, the small number of inherently reflexive predicates which may be argued to have thematic sig (72a), e.g. boscette sig 'settle', are simply transitive predicates whose selection restrictions only allow reflexive anaphors to occur in object position. 19 Which indicates that they probably are loan words from Lower German dialects which entered the language during the Hansa-era. 267 4.3.1 Inherently reflexive predicates with non-thematic sig Most of the predicates in ( 68)-(71) have non-thematic simple reflexives rather than locally bound thematic sig. As shown in (73-74), in many cases, the inherently reflexive predicate (73-74b) is derived from either agentive (transitive or unergative) verbs by productive morphological operations, e.g. in these cases by adding the prefix/or- 'over-' 20 • (73) a. b. (74) a. b. Han spiste (salat). he ate salad 'He ate salad.' Hanforspiste sig (i salat). he over-ate REFL in salad 'He overate on the salad.' Han talte. he talked 'He talked.' Han fortalte sig. he over-talked REFL (causative transitive) (inherently reflexive) ( unergati ve) (inherently reflexive) 'He made a slip of the tongue/gave himself away.' Lidz adopts Everaert's (1986) account of these examples, arguing that there "is a semantic effect corresponding to the morphological change [between (73-74a) and (73-74b)]. The (b) cases have an interpretation in which "an involuntary effect ... has taken hold of the agentive subject"(Everaert (1986:52)). This[ .. ] suggest[s] that [sig] may actually be a verbal reflexive [ ... and] that these verbs only have one semantic argument. [ .. ] In (73b), for example [sig] is not the object of overeating. That is, the sentence does not mean that the subject ingested himself. Typically, the original verb undergoing such morphological operations is intransitive [e.g. (74a)]. If, however, the original is transitive [,e.g. (73a)], the direct object is turned into a prepositional object, similar to [ .. ] anti-passives [, e.g. (73b) ... ]."(Lidz 1996: 130). Overall, we adopt this account for the predicates in ( 68-71 ). As mentioned above, and discussed at length m section 4.2.1, a number of anti- reflexive verbs have inherently reflexive/deagentive inchoative counterparts, e.g. Peter slar 268 Hans 'Peter hits Hans' vs. Peter slar sig 'Peter gets hurt'. Interestingly, there are also a few inherently reflexive verbs which are derived from unaccusative verbs, see (75a) which is derived from (75b) and (76a, b). The fact that inherently reflexive verbs can be derived from unaccusative verbs is another indication of the non-thematic status of these occurrences of simple sig. (75) a. b. (76) a. b. Peter kom sig Peter came REFL 'Peter recovered (from a disease).' Peter kom. Peter came 'Peter came.' Peter blcerer sig. Peter blister REFL 'Peter brags, shows off Peter's finger blcerer. Peter's finger blisters 'Peter's finger blisters.' (inherently reflexive) ( unaccusative) (inherently reflexive) ( unaccusati ve) Unlike the inherently reflexive predicates derived by prefixation of for- 'over-' in (73-74b), the meaning of the derived predicates in (75-76) do not appear to be related to the verbs from which they are derived in any systematic 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 While it may be tempting to conclude that all predicates labeled here as "inherently reflexive" take non-thematic sig, such a conclusion may be premature. First of all, as discussed in chapter 3, the are a number of constructions involving possessive reflexives which classify as inherently reflexive on semantic criteria, see (77). 2 ° Cf. Everaert ( 1986) for a similar analysis of Dutch. (77) "Inherently reflexive" constructions a. Han varved at gaa ud af sit/?* sit egeti*Peters gode skind af glrede. he was about to go out ofPOSREF/?*POSREF own/Peter's good skin of happiness 'He nearly jumped out of his good skin of sheer happiness.' b. Han hyttede sit I ??sit eget /*Peters skin. (cont. bet. a/b) he saved POSREF/ ??POSREF own/*Peter's skin 'He saved his own life.' 269 As discussed in chapter 5, similar inherently reflexive possessive constructions can be found in English, see (78). (78) a. He was out ofhis/*his own!* Mary's mind of happiness. b. He said ahem several times to clear his/*his own!* Mary's throat. Clearly the possessive reflexive sin 'POSREF' in (77) and the possessive pronoun his in (78) cannot be analyzed as non-thematic grammatical marker of unaccusativity occurring in the v of the voice-projection vP, see (72b). Hence, as argued in chapter 3 section 3.3.5, the semantic definition of inherently reflexive predicates/constructions is still needed, see (80). (80) Definition of inherent reflexivity: (=(65) chap. 3, sect. 3.3.2.4) A given predicate is inherently reflexive if it evokes a presupposition of 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 of the possessives sin/his is infelicitous. Due to the meaning of these predicates/constructions no contrast set of 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 which distributionally behave like inherently reflexive predicates (by only allowing the simple unintensified reflexive sig/sin) but which, nevertheless, appear to have thematic sig, e.g. (81a,b). (81) a. Peter dukkede sig I *sig selv I *Marie. Peter ducked REFL I REFL self I Marie 'Peter ducked 0 I *himself I *Marie.' b. Peter bosatte sig I *sig selv I *Marie i Afrika. Peter settled REFL I REFL self I Marie in Africa 'Peter settled 0 I *himself I *Marie in Africa.' 270 The examples in (82)-(86) are designed to test the thematic status of sig with the verbs dukke 'duck' and bosrette 'settle' using the tests discussed in sections 4.2.1-8 above. (82) Testing for agentivity (section 4.2.1): a. Han dukkede sigfor at undga at blive slaet i hovedet. he ducked REFL for to avoid to become hit in head-the 'He ducked to avoid being hit in the head.' b. Han bosatte sig i Argentina for at undga retsforfolgelse i USA. he settled REFL in Mexico for to avoid prosecution in the US 'He settled in Argentina to avoid prosecution in the US.' (83) Testing for compatibility with formation of agentive nouns in -er (section 4.2.2): a. *en dukker *a ducker b. en bosretter a settler (84) Thematic vs. non-thematic sig(section 4.2.3): a. *Hvem dukkede Peter? Who ducked Peter 'Who did Peter duck?' b. *I#Hvem bosatte Peter? Who settled Peter 'Who did Peter settle?' (85) Impersonal passivization (section 4.2.4): a. *Der blev dukket. there became ducked 'There was ducked b. *Der blev bosat. there became settled 'There was settled.' (86) Deletion of sig with 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. (?lok)Peter bosatte og integrerede sig i detfremmede. Peter settled and integrated REFL in the foreign 'Peter settled and integrated himself abroad.' 271 The results of the tests in (82-(86) are summarized in (88) which also includes the results of the same tests run on different inherently reflexive predicates. (87) Tests: (i) Agentivity (section 4.2.1) (ii) Agentive -er) (section 4.2.2) (iii) Referential ity (section 4.2.3) (iv) Impersonal passive (section 4.2.4) (v) Coordination (section 4.2.8) (88) Inherently reflexive: (i) (ii) (iii) (iv) (v) a. forspise 'overeat' * * * * * b. skarn me 'be ashamed' * * * * (?) c. skynde 'be in a hurry' ok * * * * d. dukki 1 'duck' ok * (*) * (*) e. boscette 'settle' ok ok * * (?/ok) As shown in (88) the verbs classified as inherently reflexive (based on the fact that they allow only sig as internal argument) do not behave uniformly with respect to the five tests. Compatibility with agentive phrases, test (i), and the agentive suffix -er (ii) vary considerably. Only test (iii) and (iv) seem to apply uniformly to all inherently reflexive verbs. The fact that some inherently reflexive predicates are compatible with agentive subjects (test (i)), and even allow formation of agentive nouns in -er (test (ii)), may (but need not) be an indication that some occurrences of simple 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 some inherently reflexive verbs are really transitive neutral predicates whose selection restrictions exclude anything but thematic reflexive pronouns from occurring in object position. Based on these tests it may be argued that dukke 'duck' and boscette 'settle' have thematic sig and should be analyzed as in (72a) rather than as non-thematic sig in (72b ). However, before any final conclusion is drawn let us first return to the issue of the reliability of these tests (already briefly discussed in sections 4.2.1-4,8). The tables in (89) and 21 In addition to the simple reflexive sig the verb dukke 'duck' can take one other DP as internal argument. viz. the inalienably possessed DP hovedet 'the head', see (i). 272 (90) listed the result of applying the same five tests in (87) to different anti-reflexive and neutral verbs. (89) Anti-reflexive: (i) (ii) (iii) (iv) (v) a. hade 'hate' ok ok ok ok ok b. efterfolge 'succeed' ok ok ok * (?) e. bonflade 'implore' ok * ok (*) (ok) b. mistcenke 'suspect' * * ok * (ok) c. misunde 'envy' * * ok * (ok) (90) Neutral: (i) (ii) (iii) (iv) (v) a. vaske 'wash' ok ok ok ok ok b22. klcede _pa '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 homogenous group with respect to the five test. Some verbs, are incompatible with agentive phrases or formation of agentive nouns in --er, e.g. mistcenke 'suspect' and misunde 'envy'. Others cannot form impersonal 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 of a given inherently reflexive verb does not necessarily mean that they have non-thematic sig. Though the question of whether sig with inherently reflexive verbs is thematic and referential or non-thematic and non-referential, is highly interesting in itself it is not crucial to the topic discussed in the present dissertation. Since our disagreement with the standard accounts (i.e. the reanalysis/lexical ambiguity accounts) of local sig can be settled simply by looking at the behavior of sig with neutral predicates we do not need to answer this question 22 Note that even neutral verbs which can undergo optional sig-deletion in some contexts, e.g. barbere 'shave' in (i), cannot in other contexts. e.g. (ii). (i) Peter vaskede. barberede og tgrrede 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.' 273 here and will simply leave it for further research. For simplicity we will use the term "inherently reflexive" to refer to predicates/constructions which only allow the simple unintensified forms of the reflexive, i.e. sig 'REFL' and sin "POSS REFL', regardless of the thematic status of the reflexive. 4.3.3 Grammatical uses of inherently reflexive verbs with sig selv In the literature on binding in Danish it is often claimed that the complex reflexive sig selv 'REFL self is always thematic. While it is certainly true that sig selv is thematic in the vast majority of instances, there do in fact appear to exist contexts in which non-thematic complex reflexives are required. As observed by Veraart ( 1996), complex reflexives, e.g. sig selv, can be used with unaccusative inherently reflexive verbs like skamme sig in meta-linguistic or didactic contexts like (64). (91) a. Learner: b. Teacher: *Peter skammer Marie. Peter shames Marie 'Peter shames Marie.' Nej. Peter kan ikke skamme andre. Han kan kun skamme sig selv. No Peter can not shame others he can only shame REFL self 'No, Peter cannot shame others he can only be ashamed.' As shown in the previous section, skamme sig 'be ashamed' is one of the clearer cases of non- thematic sig. So if, skamme does in fact take non-thematic sig, then how come it is intensified in (91 b)? We suggest that the answer lies in the prosodic properties ofsig, i.e. the fact that sig is a clitic which cannot carry stress on its own. In (91 b) sig selv seems to be licensed by the explicitly contrastive context. Since, no doppelganger-readings are available (due to the non- thematic status of sig with skamme), and since the inherently reflexive meaning of skamme sig does not allow for the generation of a focus-triggered contrast set, one has to argue that in (64b) it is simply the stress on sigthat motivates the presence ofselv. As discussed in chapter 274 3, section 3.3.6 the simple reflexive sig is prosodically/syntactically too weak to receive stress on its own, consequently selv is adjoined to make the reflexive strong/heavy enough to carry stress. In other words, the sig selv in (91 b) has all the characteristics of being a prosodically motivated instance of the intensified reflexive sig selv. Notice that the existence of complex reflexives with inherently reflexive verbs cannot be explained within the frameworks of Vikner (1985) and Reinhart and Reuland (1993), but receives a straightforward explanation in the present focus-based account of adnominal intensification. 4.3.4 Conclusion Based on the discussion of inherently reflexive predicates in the preceding sections, we conclude that predicates which classify distributionally as "inherently reflexive" should be divided into two groups: unaccusative/deponent predicates with non-thematic sig in v ofvP transitive predicates with thematic sig as internal argument. Many "inherently reflexives" are lexically unaccusative, e.g. skamme sig, as opposed to derived unaccusative sig-form predicates, e.g. abne sig 'open'. Furthermore, there is probably only a few true, transitive, "inherently reflexive" predicates, e.g. boscette sig 'settle'. In addition to the two types of inherently reflexive predicates distinguished above, we also have inherently reflexive possessive constructions. The semantic definition of inherent reflexivity (80) thus only applies to verbs with thematic sig and possessive constructions which only allow the simple/unintensified possessive reflexive sin as possessor argument. Furthermore, only thematic sig and the possessive reflexive sin 'POSREF' both fall under the binding principle A discussed in chapter 3. The non-thematic sig found with most inherently reflexive 275 verbs is not a reflexive anaphors but rather a grammatical marker of voice/unaccusativity which fall under a different module of the grammar. As mentioned above, the question of whether sig with inherently reflexive verbs is thematic and referential or non-thematic and non-referential, is highly interesting in itself. However, it is not of vital importance to the topic discussed in the present dissertation, since, as shown in section 4.2, the argument that local sig can be thematic can be made on the basis of neutral verbs alone. We therefore simply leave the question of the exact status of sig with different types of"inherently reflexive" predicates for future research. 4.4 Late-insertion analysis of reflexives and pronouns in Danish In this section a late-insertion analysis of Danish reflexives and pronouns is proposed (section 4.4.1 ), which has the advantage of being able to account for 1st and 2"d person pronouns and reflexives (section 4.4.2). The late-insertion analysis thus provides the morphological component of the syntactic LF-movement (Cf. Pica 1984) analysis of reflexives presented in chapter 3. 4.4.1 Lexical and syntactic properties of pronouns and reflexives Personal pronouns and both thematic and non-thematic reflexives all start out as the same unspecified determiner, 0° {a person, a number, etc.}, see (92a-c(i)). Personal pronouns differ from reflexives in that their phi-features are specified before merge, see (92a(ii)), which explains why they are able to refer on their own and as well as why they are subject to principle B. Both thematic and non-thematic reflexives have their phi-features specified after merge, see (92b-c(ii)). Thematic and non-thematic reflexives get their features valued in different ways. For thematic reflexives, which merge into argument position where they 276 receive their theta-roles (92b(iii)), it happens via binding relations, i.e. an A-chain created by movement of reflexive from theta-position to To (92b(v)). As argued in chapter 3, the subject- orientation of Danish reflexives supports the assumption that they move to T 0 • Non-thematic reflexives merge directly into v" (92c(iii)). While their features also ultimately get valued by movement to T 0 they are not interpretable (92c(iv)) and are thus not bound by the subject but rather enter into a clitic-doubling chain with it (92c(v)). (92) Third person pronouns/reflexives: a. Pronominal ham 'him', (93a), (95c): (i) starts out as unspecified determiner Do {a person, a number, etc.} (ii) phi-features are specified before merge (iii) inserted into argument position at merge, assigned theta-role (iv) phi-features are interpretable (v) no chain is formed (principle B effects) (vi) morpho-/phonological realization spelled out from valued features by phonological rule. b. Thematic sig 'REFL', (93b): (i) starts out as unspecified determiner Do {a person, a number, etc.} (ii) phi-features are specified after merge (iii) inserted into argument position at merge, assigned theta-role (iv) phi-features are interpretable (v) phi-features are valued via movement to T 0 , i.e. via A-chain. (vi) morpho-/phonological realization spelled out from valued features by phonological rule. c. Non-thematic sig 'REFL', (94a), (95a): (i) starts out as unspecified determiner D 0 {a person, a number, etc.} (ii) phi-features are specified after merge (iii) adjoined to vat merge, 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 0 ) via clitic-doubling chain 23 (vi) morpho-/phonological realization spelled out from valued features by phonological rule. The sentences in (93-5) illustrate the pronominal and reflexive elements in (92a-c). (93) a. Peter, vaskede sig,. Peter washed REFL 'Peter washed himself.' 23 Cf. Clitic-doubling in Spanish: (i) (thematic sig, phi-feat specified after merge, A-chain, agentive, (92b)) Lo, vimos a Juan,. him, see-IPL-PRES to Juan, ·we see Juan.' b. c. (94) a. b. (95) a. b. c. Peter, vaskede Hansk. Peter wash-PAST Hans 'Peter washed Hans.' Peter, vaskede hamk. Peter washed him 'Peter washed him.' Doren, abnede sig,. door-the opened REFL 'The door opened.' Peter abnede doren. door-the opened REFL 'The door opened.' Hans; slog sig;. Hans hit-PAST REFL 'Hans got hurt.' Peter, slog Hansk. Peter hit-PAST Hans 'Peter hit Hans.' Peter, slog hamk. Peter hit him 'Peter hit him.' (thematic DP, phi-feat specified before merge, no chain) (thematic ham, phi-feat specified before merge, no chain, (92a)) (non-thematic sig (in V 0 ), phi-features specified after merge, valued via clit- doubling, non-agentive, (92c)) (thematic DP, phi-feat specified before merge, no chain) 277 (thematic sig (in V 0 ), phi-feat specified after merge, valued via clit-doub., non-agentive, (92b)) (thematic DP, phi-feat specified before merge, no chain) (thematic ham, phi-feat specified before merge, no chain, (92a)) As the examples in (93b-c), (94b) and (95b-c) show, the verbs vaske 'wash', abne 'open' and sla 'hit' are all causative transitive verbs which assign theta-roles to DP and personal pronouns occurring as internal arguments and take agentive readings when occurring in active sentences. The most interesting aspect of (93-5) is the thematic status of sig. As already shown in 4.2.1 the verbs vaske 'wash' and sla 'hit' in (93) and (95) are both neutral but differ in how they interact with simple sig. While the predicate vaske 'wash' always takes thematic sig and thus always gives rise to agentive readings, sla 'hit' can take either thematic or non-thematic sig, compare (20a) and (18a) in section 4.2.1. In the above late-insertion analysis the overt realization of pronouns and reflexives is determined by where in the derivation (i.e. before or after merge) the {3rd person, singular} features are specified, see (96). (96) a. b. Third person forms: Do {3'd person, sg., -specified before merge} Do {3'd person, sg., +specified before merge} => => Overt realization: /sig/ /ham/ 278 Features which are specified in the lexicon (i.e. before merge) are per definition interpretable (e.g. pronominals, and DPs). Features which are valued in syntax (via A-chain binding, or clitic-doubling chains) may or may not be interpretable depending on whether or not they are features of a thematic element. Thematic 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 sin, kane. every married-man loves POSS-REFL wife 'Every married man loves his wife.' Since simple sig is a clitic (i.e. simultaneously Xo and XP) it may be ambiguous between thematic (in argument position) and non-thematic in V 0 of vP. The vP is a voice/diathesis projection which determines whether the predicate is an (active) agentive, causative transitive predicate (assigning two (or more) theta-roles, or a (passive), deagentive/non-agentive, intransitive predicate (only assigning one theta role, e.g. THEME to the internal object). (98) 24 In Danish V 0 has an EPP feature requiring the presence of a Do element. This Do element may be realized in one of the following two ways: a. as a (thematic) DP in [Spec, vP], or b. as non-thematic sig merging with V 0 • As shown in (99), (98a) gives rise to agentive, transitive predicates. (99) Peter abner doren. Peter opens door-the 'Peter opens the door.' As shown in (I 00), (98b) gives rise to non-agentive, unaccusative predicates. (I 00) Doren abnede sig. door-the opened REFL 'The door opened.' 279 All predicates start out as mono-thematic (unaccusative) predicates assigning only one theta- role (THEME) to the internal object. When sig merges in argument 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 AGENT theta-role of the vP projection and the resulting predicate will be an agentive (causative) transitive predicate, see (93b). When sig merges directly with V 0 the EPP feature (98) is already satisfied. Hence, the DP which merges in argument position of the predicate is assigned the THEME 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 81 and 2"d person: the implications of person asymmetries The late-insertion account presented above is particularly good for analyzing 1st and 2"d person forms which otherwise would necessitate the postulation of forms which are ambiguous between reflexives and pronouns (e.g. (102a) vs. (102b)). Languages like English which seem to distinguish between reflexives and object pronouns in the 1st and 2nd person forms, see ( 10 I), are fairly rare. (1 01 i 5 a. b. I, wash Peter, washes myself *myself I *me, *zl I me•,z· 0,. Danish, like most other languages in the world, does not distinguish between pronominal and reflexive fonns in the I ' 1 and 2nd persons 26 • That is the distinction between the ham 'him' and sig 'REFL' is only found in the 3rd person, see (102-3). 24 T 0 features are the number and person endings found (amalgamated) with tense endings m inflectional languages). Unlike 0° features, P phi-features are never interpreted. 280 (102) a. Peter; vaskede migk. Peter washed me 'Peter washed me.' ( +th mig, phi-feat specified before merge, no chain) b. (103) a. Jeg; vaskede mig;. I washed me 'I washed me.' Jeg, slog mig;. I hit-PAST me 'I got hurt.' ( +th mig, phi-feat specified after merge, A-chain) ( -th mig (in V 0 ), phi-feat specified after merge, clit-doubling, non-agentive) b. Peter; slog migk.( +th mig, phi-feat spec before merge, no chain) Peter hit-PAST me 'Peter hit me.' Syntactically the first and second person forms can be divided into the same three types as the third person forms in (92). Pronominal mig 'me', which can refer independently and which is subject to principle B in the same way as ham 'him', is illustrated in ( 1 02a) and (1 03b ). The sentences in (102b) and (103a) illustrate thematic reflexive mig 'me' and non-thematic mig respectively. Since the overt distinction between pronominal and reflexive first and second person forms is neutralized, the relevant phonological rules do not have to be sensitive to the point at which features are values - which was necessary in the rules for the third person forms, see (104). (I 04) Phonological spell-out of first and second person reflexives/pronouns: a. {1st person, sg.} => mig b. {2nd person, sg.} => dig C. { 1st person, pJ.} => OS d. {2nd person, pl.} => jer 25 The differences between I wash and I wash myself will be taken up in chapter 5 where it will be argued that English has @-reflexives, i.e. the sentence I wash is really I wash CJ, and that I wash myself is really I wash 0 myself with and adnominally intensified 0-reflexive. 26 There is a functional motivation often evoked for the absence of morphological differentiation of I st and 2"d person reflexives and pronouns, i.e. the absence of potential ambiguity between coreference and disjoint reference readings of I st person and 2"d person forms. 281 4.5 Reflexive and non-reflexive uses of reflexive elements: How many sig's do we have? The purpose of this section is two-fold. First, it contains a survey of all the so-called non- reflexive uses of reflexive elements in Danish. This survey will show that the simple reflexive sig (and its counterparts in other languages, e.g. Ger. sich, Fr. se, etc.) has a number of uses outside its use as reflexive anaphor. In contrast, English x-selfand 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 argument that while Danish sig is a true reflexive, English x-selfand Chinese ziji are always intensifiers. Second, we show that the reflexive element-s in Danish, is functionally equivalent to non-thematic sig in certain contexts, e.g. deponent verbs. In many languages which have reflexive elements 27 these may have many different uses, e.g. reflexives, reciprocals, middles, passives, and lexical formatives of deponent 28 verbs. The French and Spanish examples in (105) illustrate some ofthe different uses ofthe reflexive element se in Romance. Likewise the examples in (I 06) illustrate a similarly wide range of the simple reflexive sich 'REFL' in German. (105) French (105a,b,c,e), Spanish (105d): a. Pierre se rase. (reflexive) Pierre REFL shaves 'Pierre shaves (himself).' b. Pierre et Paul se rasent (l'un !'autre). (reciprocal) Pierre and Paul REFL shave the one the other 'Pierre and Paul shave each other.' 27 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 of Modern English proposed here which is based on the assumption that Modern English only has phonologically unrealized a-reflexives. 28 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 of the set of "inherently reflexive" verbs. c. d. C es lunettes se nettoient facilement. These glasses REFL clean easily 'This pair of eyeglasses cleans easily.' Aqui se habla espagnol. here REFL speaks Spanish 'Spanish is spoken here.' 282 (middle) (impersonal passive) e. Pierre s 'evanouit. (deponent/inherently reflex.) Pierre REFL faints 'Pierre faints.' (106) German: a. b. Peter raziert sich. Peter shaves REFL 'Peter shaves (himself).' Peter und Hans razieren sich. Peter and Hans shave REFL 'Peter and Hans shave eachother/themselves.' (reflexive) (reciprocal/reflexive) c. Die Tur 6./Jnet sich ohne Muhe auch ohne Brechstange (middle) the door open REFL without trouble also without crowbar 'The door opens without trouble even without a crowbar.' d. 29 Die Kiste bricht sich Ieicht mit einer Brechstange auf (passive) the coffer breaks REFL easily with a crowbar open 'The coffer can easily be broken open with a crowbar.' e. Peter erholt sic h. (deponent/inherently reflex.) Peter relax REFL 'Peter relaxed Given the fact that reflexive elements are frequently used in non-reflexive constructions, it is clear that in order to arrive at an account of the binding behavior of reflexive elements in a given language one must first determine which instances of these elements count as true thematic/referential reflexive anaphors and which do not. While in the Romance languages and German the same reflexive element may be used in all of the five interpretations in (I 05- I 06a-e ), in Danish the labor is divided between the reflexive pronoun sig and the verbal suffix 29 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 (I 06d) can be paraphrased as: (i) a. Die Kiest kann leich aufgebrochen werden. the cotTer can easily open-broken become 'The coffer can easily be broken open.' b. Die Kiste ist Ieicht auftubrechen. the cotTer is easy to break open 'The coffer is easy to break open.' but not (ii) Die Kiste wird Ieicht aufgebrochen ... ONGOING, EVENT the coffer becomes easily open-broken PASSIVE 283 -s, which is historically derived from encliticized versions of the reflexives pronoun sig, i.e. Old Icelandic sik 'REFL', in earlier stages ofthe language 30 , see (107a-e). (107) a reflexive sig: b. reciprocal -s: Peter barberer siglsig selv. Peter shaves REFL self 'Peter shaves (himself).' (i) Peter og Marie kysses. Peter and Marie kiss-S 'Peter and Marie are kissing (each other). (ii) Peter og Marie kysser hinanden. Peter and Marie kiss-PRES each other 'Peter and Marie are kissing (each other). c. "middle" sig/-s: (i) Lceberne losnede sig. munden blev storre. d. passive-s: (ii) Lceberne losnedes, munden blev storre. lips-the loosen REFLI-S mouth-the became bigger 'The lips parted, the mouth grew bigger.' (iii) lige i det samme skiltes Elverhoj ad right in the same split-S Elfin-Hillock apart 'at the same time Elfin-Hillock split open.' (i) Maskinen skiltes ad for at blive renset. machine-the split-S apart for to become cleaned 'The machine was taken apart to be cleaned.' (ii) Maskinen blev skill ad for at blive renset. machine-the became split apart for to become cleaned 'The machine was taken apart to be cleaned.' (iii) Her tales spansk. here speak-S Spanish 'Spanish is spoken here.' (impersonal passive) e. sig/-s in deponent vb: (i) Peter skammer sig. (inherently reflexive) 30 See Larsen ( 1969). Peter shame REFL 'Peter is ashamed.' (ii) Peter vcemmes blot ved tanken om mad. Peter disgust-S merely by thought-the about food 'Peter is disgusted by the mere thought of food.' 284 The sentences in (107a) illustrate the by now familiar use of sig and sig selv as thematic/referential reflexive pronouns. See section 4.5.1 for a brief discussion of reflexive use of reflexive pronouns. The sentences in ( 1 07b) illustrate the two ways to form reciprocals in Danish: (i) the non-productive morphological reciprocal formed by adding the -s morpheme to a small, lexically determined, group of verbs, and (ii) the productive syntactic reciprocal which is formed by letting the reciprocal pronoun hinanden 'each other' occupy the object position of any transitive verb. In contrast, to reflexive pronouns in the Romance languages as well as German sich, the Danish reflexive pronoun sig is not used reciprocally. Instead, as shown in ( 1 07b(i)), the reflexive/reciprocal/passive suffix -s (historically derived from a encliticized reflexive) is used. The reciprocal constructions will be discussed in more detail in section 4.5.2. The sentences in (107c) may be dubbed "middles" or "reflexive middles" 31 • Actually, Danish cannot use the reflexive elements to form true middle constructions of the Romance type illustrated by the French sentence in (lOSe). In order to render the meaning of(105c) in Danish one would have to resort to a paraphrase of the type Disse briller er lette at vaske 'These glasses are easy to wash/These glasses wash easily.' That is, unlike English, Danish does not have 0-middles of the type These glasses wash easily. In this respect, Danish is like Dutch. Note that if one adopts the presence of an implicit agent as criterion for middle constructions (as opposed to inherently reflexive/deponent verbs) (cf. Zubizarreta, 1982) then the sentences in ( 1 07c) 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, 31 The traditional French term for this group of verbs is "verbes pronominaux neutres·· ( cf. Ruwet, 1972. p. I 07). 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 of the vaske-type). 285 inherently reflexive verbs derived from their transitive counterparts by lexical rules. Notice that in examples like (107c(i-ii)) sig and -s may be used with no difference in meaning. In many cases whether sig or -s is used is a matter of dialectal variation. Diachronically determined variation (i.e. in some 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 of such case of apparently free variation between-sand sig-forms. The sentences in (107d) illustrate the use of the -s suffix to form the so-called s passive also sometimes referred to as 'impersonal passive' due to the lack of overtly expressed agents, see (107d(iii)). Note that in certain cases the s-form of the same verb (e.g. skille 'part, divide, take apart') may be interpreted either as an inherently reflexive/middle in one context, see (107c(iii)), or as an s-passive in another, see (94d(i)). The sentence in (107d(ii)) illustrates the periphrastic passive formed by the auxiliary blive 'be, become'+ the past participle of the main verb. Note that the meaning of what is called passive constructions in ( 1 07d) overlaps considerable with the meaning of the Romance middle, i.e. "se moyen", constructions in ( 1 05d). Both the s-passive and the periphrastic passive are very productive constructions in Modem Danish, see section 4.5.4 for more detailed discussion of the s-passive. Finally, the sentences in (107e) illustrate the use of both sig and -s as part of the lexical formatives of deponent verbs. The verbs of this group form a closed set. As in the case ofthe examples in (107c), there appear to be no meaning difference between the -s-forms and the sig-forms. In fact, as will be argued in section 4.5.5, one ofthe main difference between, on the one hand verbs of the type illustrated by losne 'loosen' (107c(i-ii)) and verbs ofthe type 286 skamme sig 'be ashamed' !va?mmes 'be disgusted' (I 07e) on the other, is that the former have transitive counterparts while the latter does noe 2 • The remainder of this chapter will be devoted to a comparison of the semantic, syntactic and morphological characteristics of reflexive constructions, reciprocal constructions, middle constructions, passive constructions and deponent verbs. Anticipating somewhat table I summarizes some ofthe result of this comparison. Table 1. Reflexive Reciprocal Middle Passive Depon/"lnher.Refl." (a) non-thematic sig: sig sig sig (b) thematic sig: sig (sig) 11 (c) -s -s -s -s -s (d) sig se/v sig selv (s1g selv/' (e) DP object DP (DP)Ihinanden (DP) DP (f) 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 I indicates the type(s) of morphological realization the different 'diathetic' categories 35 , listed in the five columns, may take. Row (b) shows that only reflexive constructions may have thematic sig, i.e. sig as a true theta-role receiving anaphor occupying the direct object position of the a neutral transitive verb. As discussed in chapter 3, 32 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. 33 The existence of inherently reflexive verbs taking thematic sig is discussed in section 4.3.2. 34 The existence of non-thematic intensified reflexives is discussed in section 4.3.3. 35 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 287 the intensified reflexive sig selv can only be thematic, see row (d) in table I above. The close affinity between deponent verbs and the co-called middles is illustrated by the fact that they have the same 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 which may enter the other 'diathetic' categories in table I also have transitive uses in with full lexical DPs occurring in object position. Row (g) refers to examples illustrating the different constructions and rows (g-h) show that in German and the simple reflexives (se/sich) can be used in all the five columns in table I. 4.5.1 Use of reflexive elements in reflexive constructions From a semantic point of view, the event denoted by a verb is said to be reflexive if it is the case the individual referred to by the subject DP engages in an activity that is directed at him- /herself, see (I 08b ). (I 08) a. b. Peter hit James. Peter, hit 0, himself (A-> B) (A-> A) The sentence in (I 08a) denotes a non-reflexive event in which the direction of the hitting action goes from Peter (as the AGENT) to James (as the PATIENT). The sentence in (108b) above denotes a reflexive event in that the action of hitting both originates from Peter (as the AGENT) and ends with Peter/himself(as the PATIENT). In English the intensified 0-reflexive, 0 himself is always thematic. 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 I. 288 reflexive sig selv 'REFL self is also always thematic 36 • In contrast, their simple counterparts (i.e. Eng. 0 and Dan. sig) have both thematic, see (l 08b ), and non-thematic uses, see ( 1 09a). (109) a. Peter slog sig. Peter hit REFL 'Peter got hurt.' (deagentive inchoative) b. Peteri slog sigi selv (med en hammer). (agentive causative transitive) Peter hit REFL self with a hammer 'Peter hit himself(with a hammer).' Semantically, the sentence in (109b) denotes a reflexive event involving a hitter (AGENT), i.e. Peter, and a hittee (PATIENT). In contrast, the sentence in (109a) describes a non-reflexive scenario only involving one participant, viz. Peter (THEME). Hence, we characterize the use of the simple reflexive sig 'REFL' in (109a) as an "non-reflexive"/non-thematic use of sig. The examples in (110) show that simple sig 'REFL' may also be a true thematic reflexive. (II 0) a. Peter; vasker James,. (A-> B) Peter washes James 'Peteri washes Jamesi.' b. Peter, vasker sig,. (A-> A) Peter washes REFL 'Peteri washes 0i.' In Danish, both the simple/unintensified reflexive sig 'REFL' and the complex/intensified reflexive sig selv 'REFL self may be used as true thematic, reflexive anaphors 37 • While historically derived from an encliticized reflexive pronoun, the verbal suffix -s can nevertheless not be used as a thematic reflexive in Danish. That is, the s-forms of verbs all represent non-reflexive uses ofthis reflexive element. 4.5.2 Use of reflexive elements in reciprocal constructions In some languages the reflexive elements can be used in reciprocal constructions. Unlike its French and German counterparts, the Danish reflexive pronoun sig cannot be used with 36 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. 289 reciprocal meaning. Only the s-form, which is historically derived from a phonologically reduced encliticized reflexive, of a small lexically defined set of verbs can receive a reciprocal interpretation, see (Ill). ( 111) Reciprocals-forms: Transitive form: a. modes 'meet' mode 'meet' b. trce.ffes 'meet' trce.ffe 'meet, encounter' c. ses 'see each other, meet' se 'see' d. slas 'fight' sla 'hit, strike' e. sla:endes 'quarrel' (sla:ende pa _) 'scold' f. modes 'meet' m0de 'meet' g. trcettes 'quarrel' (trcette 'tire, exhaust') h. brydes 'clash, wrestle' bryder 'break' I. kysses 'kiss' kysse 'kiss' j. skiftes (til at . .)'take turns .. -ing.' skifte (_ud) 'exchange; change sby's diapers' k. fglges (ad) 'accompany (each other)' fglge 'follow' I. hjcelpes (ad) 'help (each other)' hjcelpe 'help' m. tales ved 'talk' tale 'talk' n. snakkes ved 'talk, chat' snakke 'talk, chat' An small number ofthe s-reciprocals do not have transitive counterparts. The verbs in (112) exemplify this group of deponent reciprocal verbs. (112) Reciprocals-forms: a. enes 'agree, get on' Transitive form: *ene b. forliges 'become reconciled' c. kappes 'compete' *forlige *kappe d. kives 'bicker' *kive While the reciprocal s-forms may have been productive in an earlier stage of the language (probably when the s- was still formally identical to the reflexive pronoun), it certainly isn't productive anymore. Only in a few cases do the reciprocal s-form and the ordinary transitive use of the verb have the same meaning, see (llla,b) and (113a,b). ( 113) a. b. Peter m0der ofte Hans pa gaden. (meet accidentally 'run into') Peter meet-PRES often Hans on street-the 'Peter often meets Hans in the street.' Peter og Hans mgdes ofte pa gaden. Peter and Hans meet-S often on street-the 'Peter and Hans often meet in the street.' ((i) 'run into', (ii) 'meet with', (iii) passive 'are encountered') 37 See also discussion of semantic and syntactic properties of sig in chapter 3. 290 In most cases, the meanings of the reciprocal s-forms are different, to a greater or lesser extent, from meanings ofthe transitive counterparts, see (114) and (115). (114) a. Peter s!Qr Hans. Peter hits Hans. 'Peter is hitting Hans.' b. Peter og Hans sills. ( 115) a. b. Peter and Hans hit-S 'Peter and Hans are fighting.' Peter skifter luftfilter (pa sin nye bil). Peter exchanges air filter on POSS-REFL new car Peter changes the air filter in his new car. Peter og Marie skiftes til at vaske op. Peter and Marie exchange-S to to wash up 'Peter and Marie take turns doing the dishes.' Such non-compositional meaning differences between the ordinary transitive uses, e.g. (a) examples in (114)-(115), and the derived reciprocal forms, e.g. (b) examples, is an strong indication that the reciprocal s-forms are lexicalized as such. Further evidence of lexicalization comes from the fact that certain verbs are particle verbs in their reciprocal uses but not in their transitive forms, see (111 k,l), or vice versa, see (111 e). Formally the s-reciprocals appear to be identical to the s-passive forms, see (116a) and (116b). (116) 38 a Peter og Marie modes ofte pa gaden. (reciprocal) Peter and Marie meet-Sreciprocai often on street-the 'Peter and Marie often meet each other in the street.' b Peter og Marie modes ofte pa gaden. (passive) Peter and Marie meet-Spassive often on street-the 'Peter and Marie are often met (by others) in the street.' However, in at least one case the two forms differ in pronunciation 39 , see (117a) vs. (117b). 38 The superscripts ··reciprocal'' and "passive•· qualifying the diflerent 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 of 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. 291 ( 117) a. Bornene sl!is i skolen. (s/G [ sla' 11 ] +reciprocal -s = [ sJ!'s]) children-the hit-Sreciprocat in school-the 'The children are fighting at school.' b. Bornene s/Gs i skolen. (sla [sla' 11 s] +passive-s= [sla' 11 s]) children-the hit-Spassi\e in school-the 'The children are being beaten (by the teachers) at school.' The difference in pronunciation between the reciprocal s-form sliis [sl"'s] and the passive s- form sliis [sla' 11 s] thus indicates a difference in morphological structure. The long vowel in the infinitival form s/G [sla'"] is preserved in the s-passive form slas [sla' 11 s] due to the presence of a strong inflectional morpheme border between the verb stem and the suffix which prevents the formation of a closed syllable and the accompanying shortening of the vowel. Notice that bla also changes its pronunciation from [bla' 11 ] to [bl"'] in other derivational processes such as the formation of compound nouns 40 • All this indicates that the s-reciprocals are lexicalized in the same way as compound nouns, i.e. the phonological rules operating on the reciprocal s- forms are the same as the ones which operates on the output of lexical formation processes h d . 41 sue as compoun mg . The fact that somes-reciprocals are derived from intransitive verbs, see (111m) above, and that yet others have no transitive counterparts, see (112a-d), further illustrates the 39 Cf. Hansen (1967). 40 The difference in vowel length and quality between the infinitival form sla [sllr•], pronounced with long vowel and stod (i.e. pitch accent realized as creaky voice), and the reciprocal form slas [sV's], which is pronounced with a slightly higher (=more closed) short version of the long vowel [aa], is also found elsewhere in the language, sec the examples in (i) and (ii). The examples in (i)-(ii) illustrate the fact that the vowel (A] is the regular shortened form of the long vowel [aa] which is generally assumed to occur only in open syllables. (i) a. en bla bil (bla [bla'•], common gender form) b. (ii) a. b. ·a blue car· et blat hus ·a blue house' en bla musling a blue mussel en blamusling a common mussel (Mytilus edulis) (bla [bla'"] + -t (neuter)= [bl/\t]) (bla [bla'•]. common gender form) (bta- [bJA], compound noun) 41 The above argumentation is based on the assumption that the deponents-form verb omgas [/\mga'•s] 'handle. mix with' should be analyzed as an s-passive rather than as an s-reciprocal form. 292 idiosyncratic nature of s-reciprocals and can thus be added to the list of arguments in favor of assuming s-reciprocal to be lexicalized as such. Some s-reciprocals, e.g. kysses 'kiss (each other)' in (llli) have already become obsolete among younger speakers who now only use the pronominal reciprocal with this verb, e.g. de kysser hinanden 'they kiss each other'. This kind of idiosyncratic diachronic variation further underscores the lexicalization hypothesis. Some of the s-reciprocals have alternative forms with the reciprocal pronoun hinanden 'each other', see (118). (118) a. Peter og Hans moder ofte hinanden pa gaden. (compare w. (116a)) Peter and Hans meet-PRES often each-other on street-the 'Peter and Hans often meet each other in the street.' b. Bornene slar hinanden i skolen. (compare w. ( 117a)) children hit-PRES each-other in school-the 'The children often hit each other at school.' While in the case of ( 116a) and (118a) the s-reciprocal and the pronominal reciprocal have the same meaning, this is by no means always the case. Indeed, most ofthe s-reciprocals which have alternative pronominal reciprocals exhibit some meaning differences between the two forms. As the examples in (117a) and (118b) illustrates, when there is a meaning difference, then the verb forming the reciprocal with hinanden 'each other' always have the same meaning as it has in its transitive use. While the s-reciprocal slas, see (117a), refers to any kind of physical fighting activity (i.e. wrestling could qualify) involving more than one person, the transitive and reciprocal uses in (114a) and (118b) necessarily involves a hitting activity (i.e. boxing or fighting with clubs/sticks could qualify but wrestling could not). The examples in (119) exemplify the kind of meaning differences often found between transitive forms of a verb and its reciprocal and inherently reflexive forms. 293 ( 119) Meaning differences between lexicalized and productive forms: Lexicalized Meaning Transitive productive use: a. Reciprocal: De slas. # De slar hinanden. they hit-S they hit each other 'They fight.' 'They hit each other.' b. Reflexive: Han slar sig. he hits REFL 'He gets hurt.' Han slar sig selvlhunden he hits REFL self/dog-the 'He hits himself/the dog.' On the basis of the above discussion of reciprocals we conclude that, unlike French and German, Danish cannot use the simple reflexive pronoun, i.e. sig, to form reciprocal constructions 42 • We furthermore conclude that the only presently productive way to form the reciprocal in Modern Danish is by using the reciprocal pronoun hinanden 'each other'. That is, in Modern Danish reciprocals formed with the reflexive element -s are limited to a small (closed) set of lexically reciprocal verbs. 4.5.3 The use of reflexive elements in middle constructions English can productively form 0-middles, see (120a), from most transitive verbs. This kind 0-middles are not possible in Danish, see ( 121 a). Even the finite forms of the s-passive cannot felicitously be used in this way, see (121b). The most productive way to form the middle is the DP'uh,ecl is/are easy to verb-INFINITE construction exemplified by the sentence in (121c). (120) a. b. ( 121) a. b. The LAP D officers bribed easily. Peter easily bribed the LAPD officers. *LAPD betjente bestikker let. LAPD officers bribe-PRES-ACT easily 'LAPD officers bribe easily.' */?? LAPD betjente bestikkes let. LAPD officers bribe-PRES-Spassive easily 'LAPD officers bribe easily.' 42 Interestingly, Russian is like Danish in that reciprocals are formed using the suffix --sja. but cannot be formed using the free reflexive sebja. cf. Israeli ( 1997:74-78). c. LAPD betjente er lette at bestikke. LAPD officers be-PRES easy to bribe 'LAPD officers are easy to bribe.' d. LAPD betjente kan let bestikkes. LAPD officers can easily bribe- Spassive 'LAPD officers can easily be bribed.' 294 The modal verb+s-passive construction of the type exemplified by the sentence in (121 d) is another frequently used, and fully productive, way to translate the meaning of English 0- middles into Danish. The use of the s-passive in ( 121 d) is in certain respects similar to the so- called se-moyen in French, see (122). (122) a. b. Ce veston se lavefacilement. this west REFL washes easily 'This west washes easily.' Ces livres se vendent rapidement. These books REFL sell fast 'These books sell fast.' Notice, however, that in this kind of construction the s-form has to be an infinite form following a modal verb. This is not the case in French where the verb usually occurs in the finite se-moyen form, compare ( 122) and ( 123). (123) a. b. ??Ce veston peut se laverfacilement. this west can REFL wash easily 'This west washes easily.' ??Ces livres pouvait se vendre rapidement. these books can REFL sell rapidly 'These books sell rapidly.' Based on the above examples we therefore conclude that the sentences ofthe type illustrated in (12ld) has more in common with passives-forms than with the French middle constructions in (122). While Danish does not use the sig-forms to form middle constructions of the English and French types illustrates in (120a) and (122a,b), it does use sig-forms to form deagentive inchoatives, see abne sig '(unintentionally) open' ( 124b) which is derived from the original 295 causative form ofthe verb abne '(intentionally) open something' in (124a). The sentences in (125)-(126) illustrate similar causative-inchoative pairs in English and French. (124) a. b. (125) a. b. (126) a. b. Peter abner doren. (agentive causative transitive) Peter opens door-the 'Peter opens the door.' Doren abner sig. (deagentive inchoative I "middle" or pseudo-middle) door-the opens REFL 'The door opens.' Peter opens the door. The door opens. Pierre ouvre laporte. Peter opens the door 'Peter opens the door.' La porte s 'ouvre. the door REFL opens 'The door REFL opens.' Such deagentive inchoatives involving sig-forms (se-forms in French) which we have referred to here as "middles" or pseudo-middles, will be shown to have a great deal in common 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 of the -s suffix to form the so-called s-passive, also sometimes referred to as the impersonal passive due to the lack of an overtly expressed agent. (127) a. Maskinen skiltes ad for at blive renset. (=(I 07d(i))). machine-the split-S apart for to become cleaned 'The machine was taken apart to be cleaned.' b. Maskinen blev skilt ad for at blive renset. (=(107d(ii))) machine-the became split apart for to be come cleaned 'The machine was taken apart to be cleaned.' (128) a. Se habla espana!. (=(105d)) REFL speaks Spanish 'Spanish is spoken here.' b. Her tales spansk. here speak-S Spanish 'Spanish is spoken here.' (129) a. Lige i det samme skiltes Elverhaj ad right in the same split-S Elfin-Hillock apart 'At the same time Elfin-Hillock split open.' 296 (=(107c(iii))) In certain cases the s- form of the same verb (e.g. skille 'part, divide, take apart') may be interpreted as a "middle"/pseudo-middles in one context, see (129a), and as an s-passive in another, see (127a). The sentence in (127b) illustrates the periphrastic passive formed by the auxiliary blive 'be, become' +the past participle of the main verb. Note that the meaning of what is called passive constructions in ( 128b) overlaps considerably with the meaning of the Romance middle constructions in (128a). Both the s-passive and the periphrastic passive are productive in Modem Danish. The morphology of the so-called s-form/s-passive is illustrated in ( 130). (130) The formation of the the s-form/s-gassive 43 : Infinitive Present Past Past Part Meaning a. Act. made made-r mad-te mad-t 'meet' Pass. mode-s mode-s mad-te-s mad-te-s b. Act. lcegge lcegger lcegde lagt 'lay, place' Pass. lcegges lcegges lcegde-s lagde-s c. Act. sla sla-r slog staet 'hit' Pass. sla-s sla-s slog-es slog-es 'fight' d. Dep. vcemmes vcemmes vcemmedes vcemmedes 'feel disgusted' Simply put, the s-form is formed by adding -s to the active form, with an epenthetic schwa, spelled with the letter 'e', inserted as required by phonological well-formedness constraints. The three main uses of the s-form of verbs are: (a) the passive-s form, (b) deponent verbs, and (c) reciprocal verbs. The reciprocal have already been discussed in the precious section and the so-called deponent verbs will be discussed in section 4.5.5. 43 Cf. Allan, eta!. ( 1995:310-311 ). 297 A clear distinction between a passive reading and an intransitive "middle"/pseudo- middle reading can be found with the verb skilles ad 'be taken apart; burst, go to pieces', see (131a)vs. (13lb). (131) a. maskinen skiltes ad for at blive renset (passive) machine-the separate-S apart for to become cleaned 'the machine was taken apart in order to clean it.' b. lige i det samme skiltes Elverhej ad (middle, intransitive) right in the same separated-S Elfin-Hillock apart 'at the same moment Elfin-Hillock (a famous hill/mound) split open.' The s-passive is productive and can thus be formed on the basis of any transitive verb. Needless to say, when occurring with a full lexical object DP, such verbs never have a passive meaning. Notice also that the s-passive can even be formed from intransitive verbs, e.g. (132). (132) a. b. Der dansedes tillangt ud pa natten there dance-S to long out on night-the 'Dancing was going on until late at night' Skulle der dees sa dede man uden at klage (Cf. Hansen 1966) should there die-S then died one without to complain 'If one had to die, then one died without complaining'). 4.5.5 Deponent verbs. Allan, eta!. (1995:311) define deponent verbs as follows: 'Deponent verbs are verbs that have a passive form (-s form) but active meaning. They may be transitive or intransitive', see the examples in ( 133). (133) a. b. c. Der findes mange dyrearter. there find-S many animal-species 'There exist many species of animals.' Hun mindedes sin ungdom. she recall-S POSS-REFL youth (how should the deponents be glossed?) 'She recalled her past.' Jeg synes, at deter en god ide. I think-S that it is a good idea 'I think that it's a good idea.' The deponent verbs constitutes a small closed list of forms, most of which are given in (134). 298 (134) ExamQles of deQonent verbs: a. dages 'dawn' b. fattes 'lack' c. fin des 'be, exist' d. fa:rdes 'move, travel' e. gmnnes 'become green' (inchoati ve ) 44 f. kendes ved 'acknowledge' g. la:nges 'long' h. lykkes 'succeed' 45 I. mindes 'remember' j. mislykkes 'fail' k. om gas 'handle, mix with' 46 I. rygtes 'be rumored' m. synes 'think, seem' n. trives 'thrive' (=Eng.) 0. va:mmes 'feel disgusted' p. a:ldes 'become older' (inchoative) Per definition deponent verbs do not possess a form without -s. Where a verb with such a form exists e.g. finde/findes, minde/mindes, see examples in (135), there will usually be a marked difference in meaning from the deponent verb, and they must be regarded as two separate verbs, see ( 135-6). (135) a. b. Han mindede mig om modet. he reminded me about meeting-the 'He reminded me ofthe meeting.' Jeg mindes ikke hans tale. I remind-S not his speech I can't remember his speech. 44 lnchoative verbs can be subsumed under passives by assuming that they represent a passivized forms of originally causative transitive predicates. 45 In contrast to its English equivalent ·succeed'. lykkes cannot have an inanimate subject in Danish (cf. Allan, eta!. (1995:311 ): (i) a. b. c. Forseget er lykkedes. Del lykkedes ham at besta eksamen. Deter lykkedes hende at fa etjob. 'The attempt was successful.' 'He succeeded in passing the exam.' 'She has succeeded in getting a job.' 46 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) Active S-passive a. sla [sla" 1 ]'hit' s/a-s [sla'•s] 'be hit' b. ro [ro' 0 ] ·row' ro-s [ro' 0 s]/ *[r/\s] 'rowing is going on' c. DNA omgas [sla'"sl 'handle, mix with' (ii) Active tr. Active reciprocal a. sla [sla'"J'hit' s/a-s [sl/\s] · [they] fight' (productive s-pass.) (productive s-pass.) (lexicalized depon. verb) (lexicalized s-reciprocal) (136) a. b. Jeg kan ikke finde mine briller. I can not find my glasses 'I can not find my glasses.' Der kan ikkefindes dyreliv pa Mars. there can not find-S animal life on Mars 'There cannot exist animal live on Mars.' ( 137) Examples of deponent verbs with transitive counterparts: 299 a. fattes 'lack' fatte 'understand, comprehend' b. findes 'be, exist' finde 'find' c. kendes ved 'acknowledge' vedkende sig 'acknowledge' d. mindes 'remember' minder om 'remind sby about sth.' e. synes 'think, seem' syne 'inspect' The verbs in ( 138) are interesting in that in addition to the regulars-forms they also have sig- forms. Furthermore the s-form and the sig-form have the same meaning and can be used interchangeably. These alternate sig-forms thus bear witness to the fact that the -s in the s- forms originated from an encliticized reflexive, e.g. (Old Norse) -sik >(Iceland.) -st >(Dan.) -s. (138) s-form: sig-form: transitive form: a. kendes ved 'acknowledge' vedkende sig 'acknowledge' (ved)kende 'know,recognize' b. det forstas 'understand 47 ' det forstar sig 'understand' forst a 'understand' c. det hcends at 'it happens that' det hcender sig 'it happens that' det hcender 'it happens that' 48 In substandard varieties of Danish certain verbs (which are not found in the -s form in the standard language) can occur in the -s form with the intransitive/middle meaning, see (139a,b). In some cases a sig-form is also found with the same meaning as both the active form and the (deponent) passives-form, compare (139a,b and c) and (140). (139t 9 a. Det hcender at, .... (Standard Danish) (active) it happens( active) that .. 'It happens that .... ' 47 Notice that the same idiomatic use of reflexives with verbs of understanding is found in French: Dan. Detforstas/Detforstiir sig "'Fr. ('a se comprends. 48 Notice that the same idiomatic use of reflexives with verbs of 'happening' is found in French: Dan. Det hamds at! det hamder sig at co Fr. il se passe 'it happens ... .'. 49 These examples are from Hansen. 196 7. b. c. ( 140) a. b. Det hcend(e)s at,... (Substandard Danish) (s-passive) it happened-S that .... ' 'It happens-S (s-passive) that ... Det hcendte sig at, ... (Substandard Danish) (sig-form) it happened REFL that ... 'It happened that ... ' Detforstas I Detforstar sig. (Danish) it understand-S it understand REFL 'It goes without saying.' ('a se comprend. It REFL understands 'It goes without saying.' (French) 300 Notice also that the inherently reflexive verb par excellence (i.e. the example most often mentioned in the literature, e.g. Reinhart & Reuland (1993)), viz. skamme sig 'be ashamed' has two variants in Norwegian dialects, see ( 141 ). (141) a. b. c. skamme seg shame REFL 'be ashamed' slqemme-s shame-S 'be ashamed' skamme sig shame REFL 'be ashamed' (Bokmal Norwegian) (Ny-norsk Norwegian) 50 (Modern Standard Danish) As discussed in section 4.3, many inherently reflexive verbs are in fact simply deponent verbs with sig form rather than s-form. That is, the sig in skamme sig -be ashamed' (141c) fulfils the same grammatical function as the-sin vcemmes 'feel disgusted' (134o), both merge in v with unaccusative verbs. Superficially, there are three morphological classes of deponent verbs: (i) those which take only -s, see (134), (ii) those which are found with both-sand sig (138-141 ), and those which take only sig, e.g. skynde sig 'hurry', (68-71 ). However, semantically/grammatically there is only one class of deponents (i.e. -s and sig have the same function in v), see (2b). 5 ° Cf. Jan Terje Faarlund. Svein Lie, and Kjell Ivar Vannebo (1999) Norsk Referanse-Grammatikk. Universitetsforlaget. Oslo, p. 511. 301 4.6 Conclusion The different reflexive and non-reflexive uses of reflexive elements in Danish and English have been summarized in the table in ( 142), to which has been added a similar survey of the uses of reflexive elements in French, German and Chinese. (142) Typological survey ofreflexive and non-reflexive (and non-thematic) uses of reflexive elements: Danish English French German Chinese a. retl sig b. recp. -s c. p-mid sig d. mid d. pass. -s e.dep -s/sig f int himself selsoi sich (ta)ziji himself se se se se se sich sich sich sich sich (Ia) ziji Interestingly, neither English nor Chinese have any non-reflexive 51 uses of their so-called reflexives, viz. himself and ziji respectively, see (I 42b-e ). While French se, Danish sig and German sich all classify as true reflexives, himse(f does not since it is not featureless. Chinese ziji, however, is (like the Danish intensifier selv) unmarked for person, number, and gender, etc. and could thus be a featureless "true" reflexive. Notice, however, that it behaves just like English himself in that it cannot be used in any of the non-reflexive constructions. Furthermore, both English himse(f and Chinese ziji behave like intensifiers, something the reflexives in Danish, French, German and Russian are unable to do, see ( 142t). So except for the reflexive uses, English himself is basically in complementary 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, English himself and Chinese ziji display most of the uses of the Danish intensifier selv, see (143). 51 Needless to say. the logophoric uses of himself 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-locally bound instances of himself are best analyzed as reduced form of underlying intensified pronouns himself< him himself(Cf. Siemund 2002:81 ). 302 (143) Typological survey of different uses ofadnominal intensifiers: Danish English French Chinse a. Adnom. inten. kongen selv the king himself le roi (lui-)me me guo-wang (ta) ziji b. Complex retl sig selv [0 himself] soi-meme/lui-meme [ta ziji/0 ziji) c. Prosodic inten. sig selv [HIMSELF/0 himself] [SOli soi-memeJ [0 ziji] d. Doppelgrenger-mark sig selv [0 himself] [soi-meme] [0 ziji] e. lnten. pron./logophor ham selv [(him) himself] [(lui) lui-meme) [(ta) ziji/ta ziji) f Exclus.adv. int selv himself lui-meme ziji g lnclus. adv. int selv himself Jlui-meme] ye h. Add. focus part. selv even me me !ian _ye j Nominal uses selv(jeg selti'*I *meme/moi k. Deverbal compound selv *himself/self *meme-/auto-/sui-/etc .. *ziji/zi-/-ji I. Positive adjective selve very/actual[meme] 'same (kind ot)' m. Superlative adjective selveste We take the complementary distribution of himself and sig, see (142), as well as the similar distribution of himself and selv, see ( 143), as evidence that himself is always an intensifier. The apparent reflexive use of himself in (142a) can be explained if we assume the English have 0-reflexives which can be intensified by the adnominal intensifier himself. These assumptions makes it possible to arrive at the following revised version of the typological survey in (142), see (144). (144) Typological survey of reflexive and non-reflexive (and non-thematic) uses ofreflexive elements: Danish English French German Chinese a. retl. sig 0 selsoi sich -I@ b. recp. -s @ se sich c. p-mid sig @ se sich d. mid @ se sich d. pass. -s (@) se sich e. dep -sisig g se sich Combined, (143) and (144) draw a picture of Modern English in which himse(fis consistently analyzed as an intensifier which may adnominally intensify all types of nominal expressions, i.e. 0-reflexives, pronouns, and DPs. In other words, the absence of non-reflexive/non- thematic uses of himself plus its consistent behavior as an intensifier provide strong support for 303 the analysis of intensified and unintensified nominal expressions m Danish and English proposed in chapter 2, see (145). (145) Systems of intensified and unintensified nominal expressions in Danish and English. Simple/unintensified Complex/intensified a. Reflexives sig 0 sig selv 0 himself b. Pronouns ham him ham selv himself (<him himself) c. DPs kongen the king kongen selv the king himself In this chapter the hotly debated question of the lexical/syntactic status the simple 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 shown that local sig can be a true theta-marked reflexive pronoun even when bound by a co-argument since. That is, the large group of verbs defined here as neutral do allow local sig (even when bound by co-arguments) without showing any signs of reanalysis/lexical ambiguity. THE INDEPENDENCE OF BINDING AND INTENSIFICATION VOLUME II by U ffe 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 U ffe Bergeton 11 Dedication Til mine forreldre Henn·ing A. V. Larsen og Inger J. Larsen 111 Acknowledgements This dissertation is built on a large body of accumulated insight into the workings of human language in general and reflexives and intensifiers in particular found in the very rich literature on these topics. While I have tried to give credit where credit is due I fear that, in many cases, I may have failed to explicitly state in what respect and to what extent I depend on the research of others. My hope is that these omissions may be forgiven. In addition to the existing linguistic literature I am also profoundly indebted to a large number of people who have provided me with invaluable support and assistance, both personal and professional, throughout the process of writing this dissertation. Since this acknowledgement section is too brief to adequately express my gratitude to everybody, I hope that those not mentioned here will be able to forgive me and trust that the lack of mention should not be translated into lack of gratitude. Needless to say, while all of the people mentioned here have contributed to the dissertation in one way or another, directly or indirectly, and share the credit ofthe better parts, only I am responsible for any inconsistencies, errors and other shortcomings it may still contain. First of all I'd like to thank my faculty advisors at USC, Roumyana Pancheva, Philippe Schlenker, and Jean-Roger Vergnaud. Beginning in chronological order, Jean-Roger Vergnaud has been an invaluable source of both academic and administrative support ever since I came to USC. Since he guided me as faculty advisor during the initial stages of my research the impact of his input cannot not be overestimated. Philippe Schlenker, who was my advisor from 1999 to 200 I, and co-chair of my Ph.D. committee from 2002 to 2004, has been pivotal to the development and refinement of my ideas. IV Without his sharp reasoning most of the ideas in the dissertation would still be entangled in conceptual confusion and unclarity. Roumyana Pancheva, who accepted the task of being my committee chair after Philippe left USC in 200 I, has been no less important in helping me turn vague intuitions into coherent and explicit analyses. The late-insertion analysis of reflexives in chapter 4 and the analysis of the evolution of English reflexives in chapter 7 were developed in close collaboration with her. Many other parts of the analysis of both reflexives and intensifiers were worked out during our many meetings over the last few years which have, without a doubt, contributed the most to making the final product into a readable and fairly coherent whole. In addition to my advisors, a number of other faculty members at USC have also had a large impact on my work. Among them are, of course, my Ph.D. Committee Members, Joseph Aoun, James Higginbotham, Audrey Li, and Edward Slingerland. I thank them for their helpful comments and encouragement and not the least for having had the patience to read through earlier drafts of the dissertation. Special thanks also go to Jack Hawkins, Barry Schein, and Maria Luisa Zubizarreta for being in my screening paper committee and for their classes at USC which have contributed to forming my thinking as a linguist. Finally, I also want to thank Jean Lowenstamm, my faculty advisor at the University of Paris VII for his instruction. Without his suggestion that I apply for graduate school in the Unites States I would never have come to USC in the first place. And without his inspiring approach to the syntagmatic study of the sound structure of human language (i.e. Government Phonology) I would never have decided to pursue a career in formal linguistics. The student community in the Department of Linguistics and my circle of friends (including Shadi Ganjavi, Aaron Sonnenschein, Junichi Hayashishta, Barbara Lohse, Karine v Megerdoomian, Lina Choueiri, and many others) have provided the social setting without which I would not have thrive in Los Angeles. Last but not least I am deeply indebted to my girlfriend Iris Yim (t±}~) for having had the patience to live with a graduate student working on his dissertation. Without her unfailing support and encouragement I could never have completed this work. Table of Contents Dedication Acknowledgements Abstract Volume I Chapter 1 Introduction 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.10 I. II 1.12 1.13 1.14 1.15 The failure of most current binding theories to correctly separate binding and intensification Independence of binding and intensification in Danish Revisiting the analysis of simple and complex reflexives in Danish Adnominal intensification Predicate meaning and intensification of reflexives Intensification and the meaning difference between of simple and complex reflexives Logophoricity and intensified pronouns Possessive reflexives, pronouns and intensifiers Refuting anti-locality Self-nominal izations Unification of all uses of adnominal expressions Extending the analysis to Modern English Historical evolution of English reflexives and intensifiers Extending the analysis to Mandarin Chinese Conclusion VI II Ill xvi 2 5 9 12 15 18 20 22 26 30 30 31 34 35 39 VII Chapter 2 Intensification 2.1 Introduction 41 2.2 Adnominal intensification 45 2.2.1 The morphology of adnomina1 intensification 45 2.2.2 The syntax of adnomina1 intensification 48 2.2.2.1 Possible syntactic positions of adnominally intensified 48 nominal expressions 2.2.2.2 The syntactic relation between intensifier and its focus: 50 adjunction 2.2.2.3 Unified analysis of adnominally intensified expressions 54 2.2.3 Semantic and pragmatic aspects of adnominal intensifiers 55 2.2.3.1 The semantic contribution the adnominal intensifier selv 56 'himself' 2.2.3.2 Semantic and pragmatic constraints on adnominal 60 intensification 2.2.3.2.1 Unique identifiability 60 2.2.3.2.2 Prominence, centrality or contrastiveness 62 2.2.4 Is adnominal intensification of PRO possible? 67 2.2.5 Selv used as q-tloated adnominal intensifier 72 2.2.6 Fronted selv 85 2.2.7 Adnominal intensification: unification of different sub- 88 cases 2.3 Selv in secondary predication constructions 89 2.4 Selv used as scalar additive focus particle meaning 'even' 91 2.4.1 Syntax of the scalar, additive focus particle selv 'even' 95 2.4.2 Semantics of the scalar, additive focus particle selv 'even' 97 V111 2.5 Selv used as a noun 102 2.6 Adjectival forms of selv: selve and selveste 104 2.7 Selv 'self in nominalizations 108 2.8 Idiomatic uses of the element selv 109 2.9 Cross-linguistic perspective on the range of uses of 110 intensifiers 2.10 Conclusion Ill Chapter 3 Binding: Reflexives and Pronouns in Modern Danish 3.1 Introduction. Binding of reflexives and pronouns in 1\5 Danish 3.2 Basic properties of Danish reflexives 119 3.2.1 Morphological characteristics of reflexives in Danish 119 3.2.2 Semantics of reflexives 121 3.2.3 Syntactic structure of reflexives 124 3.2.4 Consequences of relegating binding and intensification to 129 different modules: a syntactic approach to binding of reflexives: principle A 3.3 Binding of reflexives in different contexts: testing the 141 independence of binding and intensification 3.3.1 The independence of intensification and locality 141 constraints 3.3.2 Locally bound reflexives in argument position: complex 146 reflexives= a sub-type of adnominal intensification 3.3.2.1 Anti-reflexives predicates: predicates which are 148 incompatible with reflexive scenarios 3.3.2.2 Predicates which are semantically incompatible with non- 153 reflexive scenarios 3.3.2.3 Neutral predicates: predicates which are semantically 155 compatible with reflexive scenarios and which are not semantically incompatible with non-reflexive scenarios IX 3.3.2.4 Summary: Presuppositions triggered by predicate 159 meaning 3.3.2.5 Complex reflexives as arguments of proto-typical neutral 160 predicates 3.3.2.6 Using inalienable possession to test whether 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 Selv as marker of statue-readings 167 3.3.2.7.2 Qua-sentences 168 3.3.2.7.3 Strict readings of reflexives in VP-ellipsis 171 3.3.2.8 Complex reflexives and distributivity 173 3.3.2.9 Summary 175 3.3.3 Reflexives in resultative constructions 176 3.3.4 Reflexives in ECM constructions 179 3.3.5 Complex reflexives and pronouns in possessor position 180 3.3.6 Binding of simple and complex anaphors in PPs and 185 prepositional predicates: phonological factors affecting intensification of reflexives 3.3.6.1 Stressability of reflexives: prosodic uses of selv =stress 186 carrying element 3.3.6.2 Intensifier-adjunction to reflexives in PPs and the 188 interaction with sentential stress 3.3.6.3 Can intensification of reflexives be reduced to lack of 203 stressabi I ity? 3.3.6.3.1 Stressable reflexives in French 203 3.3.6.3.2 Stressable reflexives in German 204 3.3.6.3.3 Stressable pronouns in Danish 205 3.3.6.3.4 Reducing intensification of reflexives to unstressability 206 X 3.3.7 Summary 208 3.4 Distribution of simple and complex pronouns 209 3.4.1 Binding of pronouns: principle B 210 3.4.2 Intensified object pronouns 212 3.4.2 Intensified subject pronouns: Intensified pronouns as 220 subjects of embedded clauses in English and Danish 3.4.3 Logophors and logophoricity vs. intensification and 224 intensified pronouns 3.4.4 Summary 226 3.5 Conclusion: The independence of binding and 227 intensification Chapter 4 The syntactic status of sig 4.1 Introduction 231 4.2 Testing the lexical/syntactic status of sig with neutral 237 verbs 4.2.1 Testing for agentivity 239 4.2.2 Testing for compatibility with formation of agentive 245 nouns in -er 4.2.3 Non-compositional meaning change and the question test 247 4.2.4 Testing for compatibility with impersonal passivization 249 4.2.5 Auxiliary-selection 252 4.2.6 L0drups tests 254 4.2.7 German sich is not anti-local 255 4.2.8 Constructions involving coordination of verbs 258 4.2.9 The status of sig as indirect object of ditransitive verbs 261 4.2.10 Non-co-argument bound local sig 262 4.2.11 Summary: Falsifying the standard analyses of sig as anti- local 4.3 The status of sig with inherently reflexive verbs 4.3.1 Inherently reflexive predicates with non-thematic sig 4.3.2 "Inherently reflexive verbs" with thematic sig 4.3.3 Grammatical uses of inherently reflexive verbs with sig selv 4.3.4 Conclusion 4.4 Late-insertion analysis of reflexives and pronouns in Danish 4.4.1 Lexical and syntactic properties of pronouns and reflexives 4.4.2 Binding theory in 1st and 2nd person: the implications of person asymmetries 4.5 Reflexive and non-reflexive uses of reflexive elements: How many sig's do we have? 4.5.1 Use of reflexive elements in reflexive constructions 4.5.2 Use of reflexive elements in reciprocal constructions 4.5.3 The use of reflexive elements in middle constructions 4.5.4 Use of reflexive elements in passive constructions 4.5.5 Deponent verbs 4.6 Conclusion Volume II Chapter 5 Independence of intensification and binding in English: synchronic and diachronic perspectives 5.1 Introduction XI 263 263 267 268 273 274 275 275 279 281 287 288 293 295 297 301 304 xii 5.2 Intensifiers, 0-reflexives and intensified pronouns in 308 Modern English and the independence of intensification and binding 5.2.1 Binding of 0-reflexives in different contexts: testing the 309 independence of binding and intensification in Modern English 5.2.1.1 Locally bound 0-reflexives in argument position: 309 complex reflexives = a sub-type of adnominal intensification 5.2.1.1.1 Anti-reflexive predicates 311 5.2.1.1.2 Inherently reflexive predicates 312 5.2.1.1.3 Neutral predicates 314 5.2.1.1.4 Non-optionality of intensifier-adjunction to 0-reflexives 318 with proto-typical neutral predicates: focus and doppe I ganger -effects 5.2.1.1.5 Ditransitives 319 5.2.1.1.6 Linking Doppelganger-effects and anti-reflexivity 320 5.2.1.1.6.1 X-self as marker of statue-readings 320 5.2.1.1.6.2 Qua-sentences 321 5.2.1.1.6.3 VP-ellipsis and the sloppy vs. strict reading of reflexives 321 5.2.1.1.7 Summary 322 5.2.1.2 Resultatives and the 0-reflexive analysis of English 324 5.2.1.3 ECM constructions the 0-reflexive analysis of English 325 5.2.1.4 Adnominal intensifiers in possessive constructions 326 5.2.1.4.1 Absence of inalienable possession in English 328 5.2.1.5 0-reflexives in PPs and prepositional predicates in 329 English 5.2.1.6 Potential problem: stressed and unstressed forms of 332 himself X Ill 5.2.2 Non-reflexive uses of 0-reflexives in English 335 5.2.2.1 Reciprocals 335 5.2.2.2 Middles 336 5.5.2.3 Inherently reflexive/deponent verbs 337 5.2.2.4 Summary 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 of intensified and unintensified 340 pronouns 5.2.4 Conclusion 340 5.3 Historical development of Danish and English intensifiers 341 5.3.1 The historical development of the Germanic/Modern 341 Danish intensifier selv 5.3.2 Previous account ofthe evolution of the Modem English 343 se{fforms 5.3.3 A new account ofthe evolution of the Modern English 348 self-forms 5.3.3.1 Stage one: intensification and binding in Old English 348 5.3.3.2 Stage two: intensification and binding in Early Modem 354 English 5.3.3.3 Stage three: intensification and binding in Modern 363 English 5.3.3.4 Additional evidence from the evolution of pronominal 363 reciprocals into 0-reciprocals 5.3.4 Summary 364 5.4 Conclusion 367 XIV Chapter 6 Independence of Binding and Intensification in Mandarin Chinese 6.1 Introduction 369 6.2 Setting the stage: evidence supporting the assumptions 369 that ziji is always an intensifier and Chinese has 0- reflexives 6.3 Internal structure of adnominal intensifiers in Chinese 380 6.3.1 Is ziji really mono-morphemic? 380 6.3.2 Ta ziji: complex intensifier, complex reflexive 0 ta ziji, or 383 intensified pronoun ta ziji? 6.4 Analyzing what appears to be uses of ziji as reflexive 387 anaphors as adnomina\ intensifiers 6.4.1 Locally bound @-reflexives in argument position: 387 complex reflexives= a sub-type of adnomina\ intensification 6.4.1.1 Anti-reflexive predicates 387 6.4.1.2 Neutral predicates 389 6.4.1.3 Inherently reflexive predicates 394 6.4.1.4 Predicate types: summary 395 6.4.2 LD-bound ziji: complex pronoun or complex reflexive? 396 6.5 Intensifier analysis of local and LD ta ziji 396 6.5.1 Locally bound ta ziji: complex pronoun or complex 397 reflexive? 6.5.2 LD-bound ta ziji: complex pronoun or complex reflexive? 398 6.6 Intensified and unintensified pronouns in Mandarin 400 Chinese 6.6.1 Simple/unintensified pronouns in Mandarin Chinese: 400 Principle B 6.6.1.1 Overlapping distribution of simple ta and ziji/ta ziji 401 6.6.1.1.1 Possessive constructions 401 XV 6.6.1.1.2 Embedded subject pronominals 402 6.7 Conclusion 405 Chapter 7 The element selv 'selr in nominalizations 7.1 Introduction 407 7.2 Different uses of the morpheme selv 'self in derived 409 nouns and adjectives 7.2.1 Word-internal selv 'self appearing to behave as a 409 "reflexive" while it really is an adnominal intensifier 7.2.2 Word-internal selv 'self behaving as an adverbial 413 7.2.3 Problematic cases 414 7.3 Cross-linguistic perspective on use of intensifiers in 416 derived words 7.3.1 French 416 7.3.2 Chinese 418 7.3.3 Summary 420 7.4 Outline of analysis of selv 'self in derived nouns and 420 adjectives Chapter 8 Conclusion 424 References 431 Appendix I Danish se/v-compounds 447 Appendix II Intensified and unintensified pronouns and reflexives in Danish 469 XVI Abstract A new approach to the interaction of binding and intensification is advanced. While most current approaches to (e.g. Reinhart&Reuland 1993, Huang&Liu 2001, among others) take the selv element in so-called SELF-anaphors (e.g., Dan. sig selv) to be itself a reflexive, we argue that it is in fact an adnominal intensifier (cf. Baker 1995, and Konig&Siemund 2000). Thus we achieve a unified account of all types of intensified nominal expressions, be they intensified DPs (e.g., Dan. Peter selv 'Peter himself'), intensified reflexives (e.g., Dan. Peter hader sig selv 'Peter hates REFL self'), or intensified pronouns (e.g., Dan. Peter sagde at Mary dansede med aile andre end ham selv "Peter said that Mary danced with everyone except himself.') - something most 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 which binding is defined as the interaction of intrinsic properties of nominal expressions and syntactic locality constraints. We also argue that the Danish reflexive sig (= Dutch zich, Norwegian seg, etc.) is not an anti-local anaphor specialized for long-distance binding, thus falsifying those binding theories which crucially rely on this assumption. Finally, concerning intensification, we argue that the semantic contribution of adnominal intensifiers (e.g. Danish selv, Eng. himselj) is not similar to that of scalar focus particle ( cf. Eckardt 2001) or reducible to centrality-effects ( cf. Konig (1997) and Siemund (2000)), but rather very similar to that of contrastive focus, which - depending on the context- may or may not involve scalar ordering of focus-generated alternatives. The analysis is extended to English and Chinese where himself and ziji 'self-self' are argued always to be intensifiers (::::: Dan. selv) modifying 0-reflexives (::::: Dan. sig), e.g. Peter, shaved 0, himself We show that this analysis provides answers to hitherto unanswered questions related to the evolution of Modern English XVll reflexives and intensifiers from Old English which had a monomorphemic selfintensifier and no reflexive pronouns. Chapter 5 Independence of intensification and binding in English: synchronic and diachronic perspectives 304 5.1 Introduction In this chapter we argue that the analysis of Danish based on the independence of intensification and binding presented in chapters 1-4 can be extended to Modem and Old English as well as to the evolution of reflexives and intensifiers between these two stages of the language. The analysis of Modern English which will be presented in section 5.2 is summarized in ( 1 ). (1) a. b. c. d. e. f. English x-selfforms (e.g. himself, herself, etc.) are not reflexive anaphors but rather adnominal intensifiers (Eng. x-self-;:::; Dan. selv 'self'), see (2a). Modern English has 0-reflexives (Eng. 0-;:::; Dan. sig 'REFL'), see (2b-e). What looks like locally bound reflexives is really locally bound adnominally intensified 0-reflexives, (Eng. 0 x-self-;:::; Dan. sig selv 'REFL self'), see (2b,c,e). The distribution of unintensified and intensified 0-reflexives is controlled by the same semantic/pragmatic factors which control the distribution of sig and sig selv in Danish, e.g.: (i) Predicate meaning: 1. Anti-reflexive predicates require 0 himself, (2c ). 2. Neutral predicates can take either 0 or 0 himself, (2b ). 3. Inherently reflexive predicate require 0, see (2d). (ii) Doppelganger-effects: Statue-readings require intensified 0 himself, see (2e). What looks like locally free reflexives (or "logophors") 1 are really intensified object pronouns whose pronominal part is not realized phonetically (due to a morpho-/phonological deletion rule) (Eng. [him} 2 himself-;:::; Dan. ham selv 'him self), see (2t). Complex subject pronouns (e.g. he himself, I myself, etc.) are intensified subject pronouns, not as special kind of anaphor ( cf. McKay 1991) (Eng. he himself-;::; Dan. han selv 'he self'), see (2g). 1 The term '"logophor" is used here informally to refer to the intensified object pronoun ham selv in Danish and the corresponding English cases of non-locally bound [him} himself Strictly speaking Danish ham selv and English {him} himself are not a Jogophors but intensified pronouns, see chapter 3. section 3.4.3 for a more detailed discussion of ham selv in Danish and the notion of Jogophoricity. 2 Square brackets are used to indicate the phonetically unrealized pronominal. (2) a. b. c. d. e. f. g. The king, himsel/ came to the meeting. The king, washed 0,.1 0, himself I DP 2 • The king, suspected *0,.1 0; himself I DP 2 • The king, rested 0,./ *0, himself I *DPz. Bill Clinton, shaved *0, statue .I 0, statue himself The king, said that the orchestra could not play with anyone other than [him,] himself as conductor. He himself passed the exam. 305 The assumptions in (I) makes it possible to unify the analysis of intensifiers and reflexives in English and Danish, see examples in (3) which show that the distribution of Danish selv, sig, sig selv, ham selv, and han selv mirrors that of English himself, 0, 0 himself, [him} himself, and he himself (3) a. b. c. d. e. Kongen selv kom til model. king-the self came to meeting-the 'The king himself came to the meeting.' Kongen, vaskede sig, I sig, selv I DP 2 • king-the washed REFL I REFL self I DP 'The king, washed 0, I 0, himself I DP 2 • Kongen, mistamkte *sig, I sig, selv I DPz. king-the suspected REFL self 'The king suspected *0, I 0, himself/ DPz. Kongen; hvilte sig, I *sig; selv I *DP 2 • king-the rested REFL self 'The king rested 0, I *0, himself/ *DP=. Bill Clinton, barberede *sig, statue I sig, statue selv. Bill Clinton shaved REFL I REFL self 'Bill Clinton, shaved *0, statue I 0, statue himself' (:::o(2a)) (:::o(2b)) (:::o(2c)) (:::o(2d)) (:::o(2e)) f. Kongen, sagde at orkestret ikke matte spille med andre end ham, selv sam g. dirrigent. ( :::o(2f)) king-the said that orchestra-the not must play with others than him self as conductor 'The king said that the orchestra could not play with anyone other than himself as conductor.' Han selv bestod eksaminen. he self passed exam-the 'He herself passed the exam.' (:::o(2g)) The English x-seljforms are consistently analyzed as adnominal intensifiers whose distribution are determined by contrastiveness condition of the module of intensification just like the Danish intensifier selv 'self', see chapters 2-3. Likewise, the behavior of the English 0- 3 Bold italics are used here only for purposes of highlighting and does not necessarily indicate stress. 306 reflexive mirrors that of sig 'REFL' in Danish, see chapters 3-4. Finally, [him] himself and he himself are analyzed as intensified pronouns whose pronominal parts fall under principle B of the binding theory while the intensifier part, i.e. himself, fall under the contrastiveness condition on intensification. These parallels between the Danish and English systems of reflexives, pronouns, and intensifiers is summarized in (4)-(5). (4) Nominal expressions in Danish: Simple/Unintensified 4 Complex/Intensified a. Reflexive sig 'himself/herself/oneself' sig selv 'himself/herself/oneself' b. Pronoun ham 'him' ham selv 'him himself/himself' c. DP kongen'the king' kongen selv 'the king himself' d. Intensifier selv 'self'(uninflected particle) (5) Nominal expressions in English: a. Reflexive b. Pronoun c. DP d. Intensifier Sim ple/U nintensified himself him the king' Complex/intensified 0 himself [him} himself(him himselj) 5 the king himself himself(inflected for gender, number and person) 6 As shown in (6), form a superficial point of view the English system of pronouns and intensifiers appears to have a much higher degree of morphological opacity that the Danish system. 4 As mentioned above the terms '"simple·· and "complex'" are used interchangeably with the terms "unintensified'' and ·'intensified'". Although from the point of view of this paper the terms "unintensified'' and '·intensified" are more correct, the terms '"simple .. and "complex'" occasionally serves as more convenient theory-neutral terms. 5 While English has an intensified form of subject pronouns which can be analyzed as a combination of pronoun + adnominal 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 Bickerton ( 1987), latridou (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 himself that I have been able to locate, see (i). (i) In his inscriptions Kirder has little to say of doctrinal matters, being concerned rather with observances, church discipline, conversions, and the discouragement of infidels. He proclaims, however, the existence of heaven and hell, and the latter part of the inscription at Sar-Mashad (1.57 ff.) is taken up with an account, only partly legible, of a vision of the hereafter seen by him himself (Boyce 1979/200 I: 115) 6 The English intensifier has 8 frequently found forms: myself, yourself, ourselves, yourselves, himself, 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. Siemund, 1999:8). 307 ( 6) Morphological opacity of English vs. transparency of Danish: Eng. himself ~ Dan. selv, sig, sig selv, ham selv One of the main advantaged of the analysis outlined in (1) is that it enables us to analyze the system of intensified and unintensified nominal expressions in English as having the same degree of morphological transparency as the Danish system in (5), see the tables in (7) and (8) which spell-out the complete English pronoun systems as analyzed here. (7) s· 1 c t "fi d) Imp e I.e. umn ens1 Ie pronouns an d fl re . M d ex1ves m o em E r h ngJIS . Nominative Acc./Dat. Reflexive 1st' sing. I me 0 2"a, sing. you you 0 3'd . , smg. masc. he him 0 3'a, sing. fern. she her 0 3'a, sing. neut. it it 0 1 s\ plur. we us 0 2"a, plur. you you 0 3'a, plur. they them 0 (8) Complex (i.e. intensified) pronouns and reflexives in Modem English. Nominative Acc./Dat. Reflexive ! 51 ' sing. !myself [me]myself 0 myself 2"ct, sing. you yourself [you ]yourse(f 0 yourself 3'a, sing. masc. he himself [him]himself 0 himself 3'ct, sing. fern. she herself [her ]herself 0 herself 3'a, sing. neut. it itself [it]itself 0 itself 1'\ plur. we ourselves [we]ourselves 0 ourselves 2"ct, plur. you yourselves [you]yourselves 0 yourselves 3'ct, plur. they themselves [them]themselves 0 themselves Section 5.3. deals with on the historical development of both the element -self, which is the adnominal intensifier in most Germanic languages, and the Modem English x-selfforms (e.g. himself, herself, etc.). It will be shown that the assumptions that Modem English has 0- reflexives and that the x-self forms are ALWAYS intensifiers provide the key to a number of 7 Although the forms you yourself. it itself, and you yourselves are perfectly acceptable. they seem to appear only in subject position and never in object position. 8 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 det 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. 308 hitherto unsolved problems related to the evolution of Modern English complex intensifier x- self( e.g. himself, herself, etc.) from the simple intensifier seolfin Old English. Finally section 5.4 concludes the chapter by providing a summary of the main results obtained and their consequence for linguistic theory. While the analysis of English outlined above has the advantage of enabling an elegant unification of the account of intensified and unintensified pronouns and reflexives in Danish and English, it also raises a host of questions, a number which will be addressed in the following sections. One of the most recalcitrant problems is the existence of unstressed himself with anti-reflexive verbs. If himselfis assumed always to be an intensifier and if intensifiers are always stressed (as has been claimed in the literature (cf. Siemund (2000) among others) then the existence of sentences like (9) with unstressed himself constitutes a potential problems. (9) Peter suspects himself As discus~ed in section 5.2.4, this and other facts, lead us to the conclusion that the 0- reflexive hypothesis may be better suited to account for earlier stages of Modern English than the language of present day native speakers. 5.2 Intensifiers, lJ-reflexives 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 of simple and complex reflexives (e.g. 0 vs. 0 himself) in English outlined above is discussed in more detail. Section 5.2.2 parallels section 4.5 in chapter 4 by surveying the different reflexive and non- reflexive use of 0-reflexives in English (e.g. reciprocals, middles, etc.). Section 5.2.3 takes a closer look at the analysis of simple and complex pronouns (e.g. he/him vs. he himselfl[him]himselj). Finally, in section 5.2.4 the results of the present analysis are summarized. 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 of the analysis of English 0- reflexives presented above will be put to the test by taking a closer look at the behavior of simple and complex reflexives (i.e. 0, and 0 himself) in different syntactic positions. That 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 of 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 x-selfforms 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 According to the analysis of English proposed above, the distinction between the English simple and complex reflexives 0 and 0 x-self is assumed to be of the same kind as the distinction between the simple (unintensified) reflexive sig in Danish and its complex (intensified) counterpart sig selv. One would therefore expect to find distribution of simple and complex reflexives in object position of "neutral", "anti-reflexive" and "inherently reflexive" predicates 9 in English to correspond to the distribution of simple and complex reflexives in Danish. The examples in ( 1 0-12) support this prediction. (I 0) Anti-reflexive verbs: (i) pron.: (ii) unint. refl. (iii) int. refl. a. He, suspects *him, I *0, I 0, himself. b. Han, mistcenker *ham, I *sig; I sig; selv. he suspects him REFL REFL self 9 See chapter 3. sections 3.3.2.1-3 for definitions of the terms "anti-reflexive". "neutral" and "inherently reflexive''. 310 ( 11) Neutral verbs: a. He, washes *him, I 0, I 0 1 himself b Han, vasker *ham, I sig, I sig, selv. he washes him REFL REFL self ( 12) Inherently reflexive verbs: a. He rests *him I 0, I *0, himself b. Han hviler *ham I sig, I *sig, selv. HE rests him REFL REFL self Based on examples such as (10-12) we conclude that predicate meaning influences the choice of 0 vs. 0 himself in the same way as it influences the choice of sig vs. sig selv in Danish. The predicates disallowing unintensified 0-reflexives, exemplified in (1 0), are anti-reflexive in that they presuppose (representational) non-identity of their arguments. The predicates in ( 11) are neutral since they are compatible with both reflexive and non-reflexive scenarios. Finally, the predicates in (12) are inherently reflexive in that they require the argument DPs to be coreferential 10 • Intensifier-adjunction is thus clearly sensitive to semantic/pragmatic factors: (i) it is only allowed when contrast with other entities is possible ( cf. Baker 1995), see (I 0-1 c,d) vs. ( 12c,d), (ii) intensifiers adjoin to 0-reflexives with anti-reflexive predicates since these presuppose the (representational) non-identity of their arguments. In following seven sections (5.2.1.1-7) we address the question of how far the equations Eng. 0 ;:::; Dan. sig and Eng. 0 Dan. x-self;:::; sig self can be pushed. That is, do all semantically anti-reflexive predicates in English require the complex form of the reflexive 0 himself? Do all semantically inherently reflexives predicates in English require the simple form of the reflexive 0? And do all neutral verbs allow both the simple and complex form of the 0-reflexive, i.e. both 0 and 0 himself? 10 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. 311 5.2.1.1.1. Anti-reflexive predicates The theory proposed here predicts that semantically anti-reflexive predicates in Danish, which allow only complex reflexives (e.g. sig selv) and DPs in object position and do not allow simple unintensified reflexives (e.g. sig), should have English counterparts which likewise allow only complex reflexives (e.g. 0 himself) and DPs and disallow the simple unintensified reflexive (i.e. 0). The list of Danish anti-reflexive predicates and their closest English equivalents in (13) was arrived at by flipping through a large Danish-English dictionary 11 • Based on the verbs listed in (13) it seems possible to conclude that Danish anti-reflexive predicates (i.e. *sig I sig selv I DP) always have anti-reflexive English counterparts (i.e. * 0 I 0 himself I DP). No exceptions to this generalizations were found. That is, none of the English predicates in ( 13) allow 0-reflexives; e.g. *Peter insulted 0 cannot mean 'Peter insulted himself, etc. (13) a. b. c. d. e. f. g. h. I. J. k. I. m. n. 0. p. q. r. s. t. u. V. Anti-reflexive predicates in Danish and their English equivalents: absolvere 'absolve, give absolution' adlyde 'obey (fx one's superior's) adoptere 'adopt' afbryde 'interrupt' affcerdige 'dismiss, brush aside' ajh0re 'examine (witness), interrogate' aflaste 'relieve the pressure on; releive (for from)' ajlevere 'deliver' ajl0se 'relieve (the guard), replace, succeed to' b0njalde 'implore, beseech, entreat' efterfglge 'follow (upon), succeed, follow' eje eskortere forfglge forfgre frygte invitere }age kvcele kvcerke lagre lede Xi fristelse 'own' 'escort' 'persecute' 'seduce' 'fear' 'invite' 'hunt, stalk' 'choke, suffocate, strangle, asphyxiate' 'throttle (=kill)' 'store; season mature (wine, cheese)' 'lead X into temptation' 11 Cf. Vinterberg & Bode1sen ( 1966. 1986). w. X. y. z. mistcenke X for at misunde mode strangulere 'suspect somebody of doing something' 'envy, grudge; be envious of' 'meet' 'strangle' 5.2.1.1.2 Inherently reflexive predicates 312 Clear cases of English inherently reflexive predicates seem very hard to find. Duck 0 and its Danish counterpart dukke sig 'duck' is one of the few plausible pair of cognate inherently reflexive predicates we have been able to find, see (14). (14) Danish: English: a. b. c. Peter dukkede sig *Peter dukkede sig selv *Peter dukkede DP Peter ducked 0 *Peter ducked 0 himself *Peter ducked DP And yet, how can we be sure that duck is really a dyadic "inherently reflexive" predicate rather than simply a common intransitive (unaccusative) verb. It is difficult to find objective critieria for deciding whether the English verbs frolic and gambol, which translate the Danish "inherently reflexive" predicate boltre sig, true "inherently reflexive" predicates rather than simple intransitive predicates, see (15). (15) boltresig 'frolic gambol' -sig *- sig selv *-DP frolic 0 I gambol 0 *frolic 0 oneself I *gambol 0 oneself *frolic DP I *gambol DP Other potentially true inherently reflexive predicates in English include overeat, see (16) and acquire, see (17c). ( 16) forcede sig -sig 'overeat' *- sig selv *-DP ( 17) anskaffe sig no get - sig noget *- sig selv noget *-DP noget overeat 0 *overeat 0 oneself *overeat FP 'acquire_ something' acquire 0 something [TEST] * acquire 0 onese(f something * acquire DP something 313 There are basically two types of mismatches between Danish 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) opfgre sig 'behave (oneself)' a. opfgre sig behave 0 b. * opfgre sig selv behave 0 oneself c. * opfgre DP *behave DP (19) se sig gal pa nogen 'see oneself mad at somebody' a. se sig gal pa nogen *see 0 mad at somebody b. *se sig selv gal pa nogen see 0 oneself mad at somebody c. *se DP gal pa nogen *see DP mad at somebody The Danish predicates in (18) and (19) both behave as "inherently reflexives" in that they allow only sig in object position. Their English counterparts, however, 'misbehave' in that they do not allow DPs, see ( 18c) and ( 19c) in spite of the fact that they allow the intensified reflexive 0 oneself This is a violation of the contrastiveness condition on adnominal intensification, see (62) in chapter 2, repeated here as (20). (20) Contrastiveness condition on adnominal intensification: A nominal expression (DP, pronoun, reflexive) can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. (= (62), chapter 2) If we assume English himselfto always behave as an intensifier, then the English examples in ( 18-9b) constitute blatant counterexamples to the condition in (20). As shown in chapter 3, section 3.3.6 counterexamples to (20) in Danish can be explained as being due to prosodic factors, i.e. the lack of stressability of sig which leads to prosodic reinforcement by adnominal intensification when prosodic rules dictate that this element must have non-contrastive word- stress. See section 5.2.1.5 below for a similar account of the exceptions to (20) in English. As mentioned above, it is very hard to establish with certainty that English has true inherently reflexive predicates at all. Even if we assume that English has some true inherently reflexive predicates, it still seems safe to say that they are rarer in English than in Danish. In 314 (21) is listed a few predicates which are inherently reflexives in Danish but which do not seem to have inherently reflexive English counterparts. (21) a. forelske sig i 'fall in love with b. skamme sig 'be ashamed' c. skynde sig 'hurry, be in a hurry' d. etc ... Note furthermore that in Danish simple sig also has a number of non-reflexive uses, see (22). (22) a. Doren lukkede sig. door-the closed REFL 'The door closed.' b. Doren abnede sig. door-the opened REFL 'The door opened.' c. En lille pol dannede sig. a small puddle formed REFL 'A small puddle formed.' In section 5.2.2.2 we discuss the possibility of analyzing English middles such as close and open in (22a) as involving non-reflexive use of 0-reflexives. 5.2.1.1.3 Neutral predicates The list of Danish and English neutral predicates in (23) was established by flipping through a large Danish-English dictionari 2 • (23) Neutral Qredicates in English and Danish denoting grooming activities: a. bade ;:::: bathe b. barbere ;:::: shave c. borste ;:::: brush d. forklcede ;:::: dress up (as), disguise e. klcede _pa ;:::: dress f. klcede _af ;:::: undress g. klcede udsom ;:::: dress _ up like - h. klo _ (pa ryggen) ;:::: sctratch _ (on the back) I. torre ;:::: dry J. vaske ;:::: wash 12 Cf. Yinterberg & Bodelsen (1966. 1986). 315 (24) Neutral predicates in Danish and English: verbs denoting mental state: a. bekymre ::::: worry b. cengste ::::: worry (25) Neutral predicates in Danish and English: nontranslational 13 movement: a. baje ::::: bend b. strcekke :::::: stretch c. dreje _ (rundt) ::::: turn_ (around) d. flytte ::::: move e. vende (om) ::::: turn (round/over) (26) Neutral predicates in Danish and English: verbs denoting internal change: a. cendre ::::: change b. forandre ::::: change c. forbedre ::::: improve d. forberede ::::: prepare e. forvandle _ til noget ::::: turn _ into something While all the verbs listed in (23-6) behave as neutral predicates in both languages (i.e. allowing 0/sig, 0 himselflsig selv, and DP in argument position), there is also a number of mismatches between the two languages. That is there is a number of "neutral" predicates in Danish which have no straightforward English equivalents taking intensified reflexive pronouns. As discussed below, these cases do not necessarily have to be considered as counterexamples to the proposed parallelism between Danish and English. It is indeed quite normal for two different languages to lexicalize the same or similar concepts in different ways, see (27). (27) Neutral predicates in Danish with no direct English equivalent: frisere 'dress somebody's hair' - sig dress/do ones hair - sig selv dress/do ones own hair - DP dress/do DP :5 hair The verb frisere in (27) has no transitive counterpart in English and has to be paraphrased as 'do 'shair'. 13 The term "non-translational movement"' is from Kemmer's ( 1995, 1993) classification of '"middle verbs'" into different semantic classes. 316 There is, however, a different set of mismatching predicates which are potentially more problematic, i.e. Danish neutral predicates with English equivalent which do not allow 0-reflexives, see (28). (28) English equivalents of Danish neutral predicates which do not allow simple 0-reflexives: 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. etablere 'establish' Peter fed 0 on onions Peter fed 0 himself on/with onions Peter fed DP with onions - sig (sam tandlcege) *establish 0 as a dentist /set 0 up as a dentist' - sig selv (sam tandl.) establish 0 oneself as a dentist /set 0 oneself up as a den - DP (sam tandl.) establish DP as a dentist /set DP up as a dentist d.fornoje sig 'amuse, enjoy' - sig *amuse 01 *enjoy 0 - sig selv amuse 0 oneself I enjoy 0 oneself - DP amuse DP I enjoy DP e.forsyne 'provide, procure, get' - sig (med) *provide 0 (with vegetables) I get 0 (vegetables) - sig selv (med) provide 0 oneself (with vegetables) I get 0 oneself (veg.) - DP (med) provide DP (with vegetables) I get DP (veg.) f. hcenge 'hang' (compare w. lynche 'lynch' which is "anti-reflexive) - sig *hang@ - sig selv hang onself -DP hangDP g. isolere 'isolate, withdraw' - sig *isolate 0 - sig selv isolate oneself - DP isolate DP - sig withdraw 0 (die: he withdraws .from other people) - sig selv withdraw 0 oneself - DP withdraw DP h. more 'amuse, enjoy' -sig - sig selv -DP i. mure X inde - sig inde - sig selv inde -DP inde *amuse 0 amuse 0 oneself amuse DP 'immure' *immure 0 immure 0 oneself immure DP I *enjoy 0 I enjoy 0 oneself I enjoy DP (d(fferent meaning) 317 All the Danish predicates in (28) are neutral, but their closest equivalents in English are behave as anti-reflexives. Should they therefore be considered counterexamples to the proposed theory of 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 counterexample we take this to be a idiosyncratic fact of the two languages. As discussed in chapter 3, verbs that, under normal circumstances, allow only DPs and complex reflexive (but not simple reflexives) can be divided into true anti-reflexives, and 'hidden neutrals', see (29). (29) a. b. Anti-reflexive predicates: semantically presupposed representational non identity, non-cancelable by context. Examples: ajlase 'relieve (the guard), replace, succeed to' e.fterfalge 'succeed' mistcenke 'suspect (sby. of doing sth.)' misude 'envy, be envious of' 'Hidden neutrals': pragmatically implied/presupposed representational non-identity, cancelable by context. Examples: koge dolke filme forgylde 'boil' 'stab' 'film, shoot' 'gild' The group of predicates in (29a) consists of predicates which can never change status from anti-reflexive to neutral no matter what context is imagined. The second group consists of predicates which may change status depending on the context, e.g. koge 'boil'. Cross- linguistically we predict that in any language which 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, may behave as neutral by default in one language, while "anti-reflexive" by default in another. The Danish verb lynche 'lynch', for example, while close to hcenge 'hang' in meaning behaves as an anti-reflexive predicate, just like its English counterpart lynch, see (30). (30) lynche 'lynch' *lynche sig lynche sig selv lynche DP *lynch 0 lynch 0 oneself lynch DP 318 Furthermore, due to slight meaning differences and differences in usage it may in many cases be difficult to determine which out of several candidates is the closest English counterpart of a Danish predicate, e.g. (28c) erncere which translates into English as either 'support', 'live on' or 'feed'. Depending which one of these translations is chosen as THE English counterpart of this Danish predicate, we end up with either a mismatch or a match. Examples such as these illustrate the relative uselessness of the whole project of finding English 'counterparts' of Danish predicates. A better approach would be to simply consider each language separately in order to see if any language internal counterexamples tum up. Under this second approach, none of the 'mismatches' in (28) above would count as counterexamples. All they would allow us to conclude is that Danish seems to have a greater number of "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 similarities between the English 0-retlexive and the Danish reflexive sig are not limited to the sensibility to predicate meaning illustrated in (1 0-12). Like sig, which has to be intensified every time it is focused, the English 0-retlexive also cannot be focused without being adnominally intensified. That is, the choice of unintensified 0-retlexive vs. intensified 0 himself in (31 a,b) is not optional, as often claimed, since it depends on focus, e.g. (31 b-,c,d) can answer the question Who did Peter wash?, but (31 a) cannot. (31) Peter 1 washes a. 01 b. 0 1 himself c. him·~= 319 As discussed in chapter 3, section 3.3.2.5, the focus-sensitivity of 0 vs. 0 himself is also found in Dan. sig vs. sig selv, e.g. (32b,c,d) but not (32a) can answer the question Who did Peter wash? (32) Peter; vasker a. sig, b. sig, selv c. ham.,z 5.2.1.1.5 Ditransitives The difference between (33), (34) and (35) shows that even for ditransitive predicates it is the semantics of the predicate which determines whether it behaves as an anti-reflexive, neutral predicate or inherently reflexive predicate. (33) a. b. (34) Neutral di-transitive predicates: bygge 'build' bygge sig et hus bygge sig selv et hus bygge DP et hus kobe 'buy' kobe sig et hus kobe sig selv et hus kobe DP et hus Anti-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 'acquire_ something' - sig noget acquire 0 something *- sig selv no get * acquire 0 oneself something *- DP noget * acquire DP something 320 5.2.1.1.6 Linking Doppelganger-effects and anti-reflexivity In this section we show that intensifier-adjunction to simple 0-reflexives in English take place in the same contexts where intensifier-adjunction to simple sig takes place in Danish, i.e. the contexts in (36). (36) Complex reflexives (e.g. 0 himself): a. anti-reflexive predicates b. hidden neutral predicates c. contrastive contexts (with neutral predicates) d. 'doppelganger-effects' (with neutral predicates): (i) statue-readings (ii) qua-sentences (iii) strict (and sloppy) readings in VP ellipsis Example: (10,13) (29b) (31-2) (37-8) ( 40-1) ( 42-3) In sections 5.2.1.1.1 and 5.2.1.1.3 above anti-reflexive (containing presuppositions of (representational) non-identity) and hidden neutral predicates (which together with background assumptions about the world generate expectations of (representational) non-identity)) were shown to trigger intensification of simple reflexives. In section 5.2. 1.1.4 contrastive contexts were identified as another trigger of intensification of simple unintensified @-reflexives with neutral predicates. In this section a number of other contexts triggering intensification of simple reflexives with neutral predicates will be discussed, namely wax museum contexts (statue-readings), qua-sentences, and strict readings of 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 some kind of 'doppelganger-effects', i.e. (representational) non-identity ofthe reflexive and its antecedent. 5.2.1.1.6.1 X-self as marker of statue-readings The adnominal intensifiers himself and selv are used as marker of statue-readings in the same way in Danish and English. The English and Danish intensified reflexives (0 himselj!sig selv) 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) Peter, washes a. 0, b. 0, himself (38) Peter, vasker a. sig, b. sig, selv (39) a. 'Peter<real> washes PeteLreal>·' b. 'Peter <real> washes Peter <statue>· (statue-reading) 5.2.1.1.6.2 Qua-sentences As argued in chapter 3 section 3.3.2. 7.2, the qua-sentences in ( 40-1) illustrate the motivation for analyzing dobbelganger-effects in terms of representational identity/non-identity rather than in terms of referential identity. Again the distribution of intensified and unintensified reflexives is the same in English and Danish. (40) a. b. (41) a. b. ? ? *Peter. working as barber. shaved. Peter, working as barber, shaved himself ?? *Peter. i egenskab af barber, barberede sig. Peter in quality of barber shaved REFL 'Peter, qua barber, shaved.' Peter, i egenskab af barber. barberede sig selv. Peter in quality of barber shaved REFL self 'Peter, qua barber, shaved himself.' In order to facilitate the processing of the above qua-sentence they should be read in a context where a group of soldiers are lost in a forest for weeks. In order to maintain a certain level of hygiene the platoon leader picks Peter to function as camp barber and immediately orders him to shave everybody in the camp including himself. In this context, (40-lb) is clearly preferred over ( 40-1 a). 322 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 of reflexives in VP- ellipsis constructions is due to doppelganger-effects, i.e. it involves representational non- identity. The examples in (42-3) illustrate the difference between simple and complex reflexives with respect to the availability of strict and sloppy reading in ellipsis constructions. Again, the distribution of simple and complex reflexives is the same in English ( 42) and Danish (43). (42) (43) a b. a. b. 5.2.1.1.7 Peter washed 0 and so did Hans. Peter washed 0 himself and so did Hans. Peter vaskede sig og det gjorde Hans ogsa. Peter washed REFL and it did Hans also 'Peter washed (himself) and so did Peter.' Peter vaskede sig selv og det gjorde Hans ogsa. Peter washed REFL self and it did Hans also 'Peter washed himself and so did Peter.' Summary (only sloppy) (both sloppy and strict) (only sloppy) (strict and sloppy) In the previous sub-sections of section 5.2.1 we have discussed the behavior of locally bound 0-reflexives occurring as internal arguments of verbal predicates. According to our assumption that binding and intensification constitute separate modules of the grammar we predict the binding behavior of 0-reflexives to follow from principle A as formulated in chapter 3, while intensification of reflexives (i.e. the distribution of complex reflexives, e.g. Dan. sig selv and Eng. 0 x-selj) should follow from the principles of intensification, i.e. the principle of contrastiveness as formulated in chapter 2. We argue that this approach to binding and intensification can successfully account for the distribution of simple and complex reflexives found in the contexts summarized in the descriptive generalizations in ( 44-5). (44) Simple reflexives (e.g. Eng. 0;:::; Dan. sig. ): a. Neutral predicates b. Inherently reflexive predicates ( 45) Complex reflexives (e.g. Eng. 0 himself;:::; Dan. sig selv): a. anti-reflexive predicates b. hidden neutral predicates c. contrastive contexts (with neutral predicates) d. doppelganger-effects (with neutral predicates): (i) statue-readings (ii) qua-sentences (iii) strict (and sloppy) readings in VP ellipsis 323 (section 5.2.1.1.3) (section 5.2.1.1.2) (section 5.2.1.1.1) (section 5.2.1.1.3) (section 5.2.1.1.4) (section 5.2.1.1.6) (section 5.2.1.1.6.1) (section 5.2.1.1.6.2) (section 5.2.1.1.6.3) In the following sections the distribution of simple and complex 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 of the distribution of complex reflexives in prepositional predicates which explains intensification of reflexives with certain PPs a consequence of the inability of simple reflexives (which behave phonologically like clitics) to carry stress on their own. While there are some apparent "exceptions" to the correspondence between sig and 0 these are either due to lexical differences (e.g. Jack of exact semantic correspondence between predicates in two languages, cf. the discussion of neutral predicates in section 5.2.1.1.2), or can be explained by reference to other aspects of grammar or language use (e.g. phonetic factors: 0-reflexives behave differently from overly realized clitics, etc., see sect. 5.2.1.5 below). The remaining set of more recalcitrant counterexamples is fairly restricted and does, in our opinion, not in itself warrant abandoning the viability of the general approach binding and intensification proposed here. 324 5.2.1.2 Resultatives and the @-reflexive analysis of English If the 0-reflexive is the English equivalent of 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. (47) a. b. c. (48) a. b. c. Peter arbejdede sig Peter arbejdede sine foraddre Peter arbejdede sig selv Peter worked REFLIREFL self/his parents 'Peter worked himself/his parents rich.' Peter drak #sig Peter drak Hans Peter drak sig selv Peter drank REFLIREFL self/Hans 'Peter drank himself/Hans under the table.' rig. rige. rig. rich under bordet. under bordet. under bordet. under table-the Peter dansede sig til verdensmesterskabet. Peter dansede #sine fora:ldre til verdensmesterskabet. Peter dansede #sig selv til verdensmesterskabet. Peter danced REFLIREFLself/his parents to the world championship 'Peter danced himself/#his parents to the world championship.' The English counterparts of(46-8) are given in (49-51). As shown in (49a), (50a) and (51a), contrary to expectations, English resultatives of the neutral and inherently reflexive types do not allow the 0-reflexive. This constitutes a potential problem for the analysis proposed here, which is based on the assumption that the English 0-reflexive corresponds to the Danish simple/unintensified reflexive sig 'REFL'. (49) a. Peter worked *0 rich. b. Peter worked his parents rich. c. Peter worked 0 himself rich. (50) a. Peter worked *0 across the ocean. b. Peter worked #his parents across the ocean. c. Peter worked 0 himself across the ocean. (51) a. Peter dansed *0 to the word championship. b. Peter dansed #his parents to the word championship. c. Peter dansed 0 himself to the word championship. 325 At this point, we have no fully-fledged solution to this problem. A possible solution may come from comparison with Romance languages, which lack (or have only very few) resultative constructions 14 of the type exemplified in (46-51 ), see the French example in (52). (52) *Pierre se travaille riche. Peter REFL works rich The grammatical constraints barring reflexives in resultatives in French, see (52), may also be responsible of the absence of unintensified 0-reflexives in English resultative constructions. As discussed in more detail in section 5.2.1.5, phonological constraints on the distribution of certain types of clitics and zero-elements may also be at play. In other words, the absence of 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 modules of the grammar (i.e. phonology, etc.). 5.2.1.3 ECM constructions the 0-reflexive analysis of English If the 0-reflexive is the English equivalent of 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) (54) a. b. c. a. b. c. Peter ansa sig for at vcere intelligent. Peter ansa sine forceldre for at vcere intelligente. Peter ansa sig selv for at vcere intelligent. Peter considered REFLIREFL self/his parents for to be intelligent Peter ansa #sig for at vcere dod. Peter ansa sine forceldre for at vcere dode. Peter ansa sig selv for at vcere dod. Peter considered REFLIREFL self/his parents for to be dead The English counterparts of (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 14 Cf. Lidz & Williams (2002). 326 analysis proposed here, which is based on the assumption that the English 0-reflexive corresponds to the Danish simple/unintensified reflexive sig 'REFL'. (55) a. Peter considered *0 (to be) intelligent. b. Peter considered his parents (to be) intelligent. c. Peter considered 0 himself (to be) intelligent. (56) a. Peter considered *0 (to be) dead. b. Peter considered his parents (to be) dead. c. Peter considered 0 himself (to be) dead. As in the case of the resultative constructions discussed in the previous section, the grammatical constraints responsible for the absence of unintensified 0-reflexives in English resultative constructions may be unrelated to both binding and intensification. As discussed in more detail in section 5.2.1.5, interference of phonological constraints on the distribution of certain types of clitics and zero-elements may be to blame. In other words, the absence of unintensified @-reflexives in resultative constructions does not necessarily constitute a fatal problem for the present analysis since it be explainable as being due to other modules of the grammar (i.e. phonology, etc.). 5.2.1.4 Adnominal intensifiers in possessive constructions If, as assumed here, reflexives and pronouns are in complementary distribution then the local domain must be defined differently in Danish and English. In English the possessive pronoun his can be locally bound, see (58c). This is not possible in Standard Modern Danish, where the locality difference between pronouns and reflexives is maintained in possessive constructions, see (57a) vs. (57b). (57) a. b. c. Peter, vaskede sin, *z pung. Peter, vaskede hans., z pung. Peter, vaskede Marie Sz pung. Peter washed POSSREFL!his/Mary's wallet (58) a. b. c. d. Peter, washed *0, wallet. Peter, washed *0, own wallet. Peter; washed his, z wallet. Peter washed his, ( Jz own wallet. 327 If, as argued above, 0-reflexives are the equivalent of unintensified sig in Danish, then it would be natural to expect English 0-reflexives to be able to occur in possessor position just like the Danish possessive reflexive 15 sin 'POSSREFL' in (57a,b). As shown in (58a,b), this is not the case. We do not, however, consider this mismatch in the distribution of possessive reflexives and pronouns in Danish and English to constitute a serious problem for the present approach since it can be explained quite simply a foJiowing from a morphological differences between the two languages. Languages differ with respect to how many and which grammatical categories are captured overtly by the morphology. While the morphology of Danish and Latin differentiate overtly between possessive reflexives and possessive pronouns, languages such as English and French do not. Whatever the linguistic principles are which are at plays in determining the overt morphology of a language, they do not directly affect the present approach to binding and intensification proposed here. The difference between the binding domains 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 argument against the proposal defended here. 15 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 of sin and hans): a. Peter, vaskede sin; 'z pun g. b. Peter, vaskede hans, z pung. Peter washed POSSREFL/his wallet (ii) Danish Dialect (complementarity of sin and hans): a. Peter, vaskede sig, 'z· b. Peter, vaskede ham., z· Peter washed REFL!him (ii) Micro-parametric variation: DPs containing a subject constitute a local domain for Binding of anaphors: a. English YES b. Standard Danish NO c. Dial. Danish YES/NO??? 328 5.2.1.4.1 Absence of inalienable possession in English In this section we argue that the differences between English and Danish with respect to inalienable possession, i.e. the fact that Danish allows inalienable possession of the type illustrated in (59) while English does not (60), do not constitute a serious problem for the present approach to binding and intensification. (59) (60) Inalienable possession in Danish: Peter vaskede hcenderne. Peter washed hands-the ok(a) 'Peter washed his own two hands.' ok(b) 'Peter washed the hands (ofthe body he was dissecting).' Absence of inalienable possession in English: Peter washed the hands. *(a) 'Peter washed his own two hands.' ok(b) 'Peter washed the hands (of the body he was dissecting).' (inalienable) (alienable) (inalienable) (alienable) As discussed in chapter 3, section 3.3.2.6, we follow Vergnaud & Zubizarreta in assuming that the inalienable possessed DPs in (59a), contains an empty possessor argument. That is, the readings in (59a) vs. (59b) are obtained by the representations in (61a) and (61b) . I 16 respective y . (61) a. b. Peter, vaskede hcenderne(xJ Peteri washed hands-the( xi) Peter, vaskede hcenderne Peteri washed hands-the If inalienable possession is analyzed as binding of phonetically unrealized possessor arguments (i.e. 0-refiexives), and if English is assumed to have 0-refiexives, then why does English not allow inalienable possession as shown in (60a)? While this question may, at first, seem to constitute a serous problem for the present approach, it most likely is not. Languages simply differ (parametrically) with respect to whether (and where) they allow inalienable possession. French, which is similar to English in that it does not have possessive reflexives 16 Yergnaud and Zubizarreta ( 1992) suggest that hcenderne(x) is derived from hcenderne by a lexical redundancy rule. 329 (e.g. in the sentence Pierre 1 mange son,= pain 'Peteri eats hisi/z bread' son/sa 'his, her' is ambiguous in the same way as English his), patterns with Danish in allowing inalienable possession, see (62). (62) Inalienable possession in French: Pierre lave les bras. ok(a) 'Peter washed his own two hands.' ok(b) 'Peter washed the hands (ofthe body he was dissecting).' (inalienable) (alienable) As this cross-linguistic data indicates, the presence/absence of inalienable possession in a given language is independent of the presence/absence of 0-reflexives in argument 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. Furthermore, in spite of the fact that Danish has a possessive reflexive (which is different from the possessive pronoun) it still allows inalienable possession. 5.2.1.5 lJ-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 with the distribution of simple and complex reflexives in certain prepositional predicates. In this section we argue that the absence of simple unintensified 0-reflexives in prepositional predicates can be explained along similar lines. The 0-reflexive analysis of Modern English proposed here predicts that the 0-reflexive should behave like the simple unintensified reflexive sig 'REFL'in Danish. As shown in (63-5) this predication is supported by the behavior prepositional predicates in English. 330 (63) PreQositional verbs in English and Danish: a. tale til 'speak to' b. pegepa 'point to/at' c. dr0mme om 'dream about' d. lytte til 'listen to' e. sigte p!i 'aim at' f. Hans peger pa *sig I sig selv I bilen. Hans is pointing at *0 I 0 himself I the car (64) PreQositional nouns in English and Danish: a. et billede af 'a picture of' b. en beskrivelse af 'a description of' c. en hyldeste til 'a tribute to, an homage to' d. en bog om 'a book about' e. en evaluering af 'an evaluation of' f. . ... en evaluering af *sig I sig selv I bilen . . . . . an evaluation of *0 I 0 himself I the car (65) PreQositional adjectives in English and Danish: a. sikker p!i 'sure of' b. interersseret i ' interested in' c. optaget af 'occupied by' d. tilfreds med 'satisfied with' e. glad for 'pleased with, like' f. forskellig fra 'different from' g Peter er tilfreds med *sig I sig selv I bilen. Peter is satisfied with *0 I 0 himself I the car In both English and Danish, most prepositional constructions allow complex reflexives but exclude simple unintensified reflexives. There is, however, a number of mismatches between the languages. As mentioned in chapter 3 section 3.3.6, a small group of prosodically heavy prepositions ('snake'-prepositions) do allow simple sig, e.g. bag 'behind' in (66a). (66) a. b. Han plantede flaget bag He planted the flag behind sig *0 I I sig selv I 0 himself! Mary Mary The Danish verb+PP construction plante X bag Y 'plant X behind Y' in (66a) behaves as a neutral construction. Assuming parallelism between Danish sig and English 0-reflexives, we would expect the English verb+PP construction plan X behind Y to behave similarly. As the ungrammaticality of the 0-reflexive in the English example in (66b) shows this prediction does not hold. While in (63-5) the absence of simple unintensified 0-reflexives in can be 331 argued to be due to the same prosodic factors excluding simple unintensified sig in Danish, the mismatch between English and Danish illustrated in (66) points to a difference between English 0-reflexives and Danish sig. That prosodic explanation of the absence of @-reflexives with certain prepositions is supported by the difference between unstressable se 'REFL' and stressable soi 'REFL' in French: only the the latter is found in prepositional predicates, see (67) (67) a. b. Pierre est fier de soi. *Pierre est fier de se. Peter is proud of REFL 'Peter is proud of himself.' Like its English counterpart, the simple 0-reflexive, the French reflexive clitic se 'REFL' also never occurs with any prepositions, see (68). (68) Pierre a plante le drapeau derriere *se lsoi-(meme). Peter has planted the flag behind SE I SE-(same) 'Peter planted the flag behind himself.' As discussed in chapter 3, section 3.2.4, the differences between Danish sig and French se follow from the fact that the former is syntactically independent prosodically bound word, which does not form a morphological constituent with its host verb, while the latter fuses morphologically with the verb with which it does form a constituent. The fact that English 0- reflexives behave like French se in never occurring with prepositions, indicate that they may be of the same nature, i.e. clitics which need to fuse morphologically with a verbal host with which they then form a constituent. The above data show that in many cases where the similarities between Dan. sig and Eng. 0 break down 0 actually patterns with the French reflexive clitic se 'REFL'. The leads us to the conclusion that 0-reflexives have more in common with Romance reflexive clitics 17 17 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. 332 than with Mainland Scandinavian sig/seg and Dutch 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 morpho-syntactic peculiarities of phonologically unrealized elements. 5.2.1.6 Potential problem: stressed and unstressed forms of himself It has often been claimed that the uses of x-self forms as reflexives and intensifiers can be distinguished based on stress. X-self forms used as adnominal intensifiers are stressed, as opposed to x-seljforms used as reflexive anaphors which are unstressed. Siemund (2000:82-3, (3.173)) proposed the generalization in (69). (69) English x-sel(is ambiguous between: a. a reflexive anaphor, which is unstressed (x-selj) and b. an intensifier[ ... ], which is stressed (x-SELF). According to Siemund (2000), the assumption that x-selfforms are ambiguous as indicated in (69) makes it possible to reduce binding conditions to (70). (70) a. b. An anaphor (i.e. unstressed x-selj) is bound in co-argument position. A pronominal is free in a co-argument position. Siemund, following Baker ( 1995) 18 , suggests that locally free x-SELF forms should be analyzed as intensified 0-pronouns, e.g. (71). (71) John; believes that letter was sent to everyone but [himJ himSELF. This analysis is based on the assumption that himself is stressed in sentences like (71) but not in sentences like (72). (72) Peter, envies 0, himself b. (ii) a. b. Peter and Maryfought 0 Peter and Mary pointed at *0 Peter and Mary are proud of *0 I I each other. each other. each other. 18 Baker ( 1995) in tum is inspired by the analysis of certain instances of locally free himself as deriving from underlying him himself proposed in Ross ( 1970). 333 While there appears to be a tendency for x-selfforms to be stressed in (3) but not in (4), we will argue that this does not warrant the jump to the conclusion (69) that adnominal intensifiers are always stressed while anaphors are not. Siemund himself acknowledges this and cites a number of counterexamples, see (73) (73) a. b. John noticed a picture of himself in the post office. Bill likes stories about himself In spite of the fact that the x-self forms in (73) are predicted by Siemund to be intensifiers intensifying 0-pronouns by Siemund, they are nevertheless perfectly ok with out stress. Rather than abandoning the analysis of the x-selfforms in (73) as intensified 0-pronouns, we suggest to save the analysis by questioning the very generalization that adnominal intensifiers ALWAYS have to be stressed. Rather than being an intrinsic property of intensifiers, stress placement is determined by a number of factors, e.g. focus structure of the sentence, non- contrastive sentence stress, stressability of lexical elements (e.g. Dan. sig, Fr. se and English 0-retlexives, 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 of unraveling the mysteries of the interactions between stress placement and intensification is suggested by the fact that, as observed in Siemund (2000:83), in the German translations of (73) the reflexive sich can occur without the intensifier selbst. In contrast, the Danish translations of (73) would require sig to be intensified by adjunction of selv. As argued in chapter 3, section 3.3.6 19 , this difference between Danish and German follow from the fact that Danish sig is unstressable while German sich is stressable. As argued in 5.2.1.5, English 0-retlexives 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 adnominally intensified to do so, see also chapter 3, section 3.3.6. The question of 19 And as suggested in Siemund (2000). 334 whether this approach necessitates the assumption of two different kinds of intensifiers: (i) intensifiers whose presence is semantically motivated, and (ii) intensifiers whose presence is prosodically motivated, is discussed in section 3.3.6. For now, suffice it to say that the relationship between stress and intensification is far from clear and certainly not clear enough to warrant the assumption that all adnominal intensifiers must be contrastively stressed. Awaiting further, more conclusive, research on the matter we tentatively assume that the generalization that adnominal intensifiers MUST have stress (cf. lb) is too strong. As argued here, the interrelations between stress and the distribution of adnominal intensifiers is far to complex to be captured by the simple generalization that intensifiers must be stressed. Indeed, to proceed from the assumption that intensifiers must be stressed as a god-given fact, amounts to taking the solution to the problem under investigation as one's point of departure. The whole point of the present dissertation is to show that the dividing line between intensifiers and reflexives is far from clear and that a great deal of misunderstandings have been generated by mistaking traditional assumptions about what is a reflexive and what is an intensifier for solid facts. If, as proposed here, all instances of x-self forms in English are adnominal intensifiers, then the assumption that intensifiers are all mandatorily stressed must be wrong and should be revised. Logically, there is no way to decide which assumption (that all intensifiers must be stressed, or that all x-self forms are intensifiers) is correct. It is, of course, an empirical question which must be decided by looking at the facts. At this point, while the final answer is still up for grasp, we hope to have presented enough data to show that the assumption that all x-seljforms are intensifiers is at least as plausible as, if not more plausible, than the assumption that x-self forms are ambiguous between unstressed reflexives and mandatorily stressed intensifiers. Furthermore, we hope to have called attention to the need to distinguish between non-contrastive stress 20 contrastive stress. 20 As discussed in chapter 3. section 3.3.6. all bi-moraic words are assumed to required (by phonological rules 335 5.2.2 Non-reflexive uses of 0-reflexives in English Pushing the equivalence of English 0-reflexives and Danish sig to the extreme, we should expect to find a large degree of degree similarities in their non-reflexive uses. In this section we argue that 0-reflexives function the same way as simple 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. Romance languages and German) the simple unintensified reflexive (selsilsich) can be used as both reciprocal and reflexive can be used to bolster both our analysis of 0-reflexives in Modern English, see (74). (74) a. [Peter and Mary], kissed I each other,. b. [Peter and Mary]Jought I each other,. See chapter 4, section 4.5.2 for discussion of the use of the reflexive suffix -s, which is derived from an encliticized reflexive -sik, to form reciprocal forms of certain verbs in Danish. The evolution of 0-reciprocals in Modern English can also be used bolster our analysis of the historical development of self-intensifiers, see also section 5.3 below. In the same way Old English (OE) locally bound pronouns were replaced by 0-reflexives in Early Modern English (EME), see (75a), so were also locally bound pronouns with 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 of phonological material, e.g. by adjunction of 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 of intensification (e.g. Dan. sig se/v) does not necessarily require the intensifier to be contrastively stressed. 336 (75) Old English: >> (Early} Modern English: a. Locally bound sing. pron. >> 0-reflexive He, washed him,. >> He; washed 0, (himself}. 'He washed himself' b. Locally bound plur. pron >> 0-reci p roc a! They, kissed them,. >> They, kissed 0,. 'They kissed 0/each other.' In other words, locally bound expressions in Old English (be they reflexive or reciprocal) were all realized as pronouns. The change from OE to EME can thus be capture saying that locally bound pronominals turned 0-anaphors in EME, see (76). (76) a. (OE) pronominal reciprocals >> b. (OE) locally bound pronouns >> (EME) 0-reciprocals (EME) 0-retlexives This kind of evidence illustrates the close affinity between reflexives and reciprocals and provides additional support for the analysis of the assumption that Modern English has 0- anaphors (which are functionally equivalent to Dan. sig, Fr. se and Ger. sich) as well as for the historical evolution of 0-reflexives/reciprocals proposed in section 5.3. 5.2.2.2. Middles Danish and French use simple unintensified reflexives (i.e. sig and se) to form deagentive inchoatives, see abne sigls 'ouvrir '(unintentionally) open' (77b) and (79b) which is derived from the original causative transitive form of the verb abnelouvrir '(intentionally) open something' in (77a) and (79a). The sentences in (78) illustrate that similar causative- inchoative pairs an be found in English, where 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. D0ren abner sig. (deagentive inchoative) door-the opens REFL 'The door opens.' (78) (79) a. b. a. b. Peter opens the door. The door opens 0. Pierre ouvre laporte. Peter opens the door 'Peter opens the door.' La porte s 'ouvre. the door REFL opens 'The door REFL opens.' 5.2.2.3 Inherently reflexive/deponent verbs 337 The sentences in (80-1) illustrate typical deponent inherently reflexive verbs in Danish and English, i.e. verbs which only allow simply unintensified reflexives in object position. (80) (81) a. b. c. a. b. c. Peter anskaffede sig et hus. *Peter anskaffede sig selv et hus. *Peter anskaffede Marie et hus. Peter acquired 0/0 self I Mary a house 'Peter acquired (*himself/* Mary) a house.' Peter acquired 0 a house. *Peter acquired 0 himself a house. *Peter acquired Mary a house. While in Danish the presence of the simple reflexive is beyond doubt since it is phonetically realized, the existence of an unintensified 0-reflexive in (81 a) is harder to prove. Keyser and Roeper (1992) claim to have found a binding-independent way to decide whether a given predicate has zero arguments. According 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 arguments, e.g. *re-come, *re-arrive. 5.5.2.4 Summary The fact that English 0-reflexives pattern with overt unintensified reflexives in German, Danish, and French supports the argument that they are functionally equivalent, see (82). 338 (82) Typological survey of reflexive and non-reflexive uses of reflexive elements: 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 -slsig se sich As shown in chapter 2, English x-selfforms have more in common with adnominal intensifiers (e.g. Dan. selv, Ger. selbst, Fr. (lui-)meme), see (83). (83) Typological survey of different uses of adnominal 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 selbst 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. lnclus. adv. int. himself selv lui-meme selbst Combined, the evidence summarized in (82-3) provide strong support for the analysis based on the assumptions that English has 0-reflexives and that English x-self forms are ALWAYS intensifiers. 5.2.3 Intensified and unintensified pronouns in English In this section we propose an account of simple and complex pronouns in English, e.g. he himself and him himse(fl[him] himself, based on the assumption that complex pronouns are formed by adjunction of the adnominal intensifier himself, in the same way as intensified DPs, e.g. the king himself Both subject and object pronouns fall under the same principle B of the binding theory, i.e. they have to be locally free. Furthermore, simple and complex pronouns behave alike with respect to the binding theory. Any differences between them are assumed to follow from the module of intensification rather than from binding related factors. As mentioned in the introduction, the present analysis thus builds on and extends similar ideas expressed in earlier work by Ross (1970), Zribi-Hertz (1989,1995), McKay (1991), Baker ( 1995) and Konig & Siemund ( 1999) and others. 339 5.2.3.1 Intensified subject pronouns Our analysis of intensified subject pronouns, e.g. he himself and Dan. han selv 'he self', as formed by a general, binding independent process of adnominal intensification has already been presented in chapter 3, section 3.4.2 and will therefore not be repeated here. 5.2.3.2. Intensified object pronouns In this section we argue that the analysis of intensified object pronouns in Danish, e.g. ham selv 'him self', see (85), can be extended to English, see (84). (84) Peter, said that Mary danced with everyone except fhimJhimself. (85) Peter, sagde at Marie dansede med aile andre end ham selv. Peter said that Mary danced with all others than him self 'Peter said that Mary danced with everyone except himself.' In both languages the intensified object pronouns is composed of a pronominal + the adnominal intensifier. The only difference is that the pronominal part of the English intensified pronoun is made inaudible by some process of haplological reduction, while it is phonetically realized in Danish. In Danish, the complex pronoun ham selv 'himself, e.g. (3d), can be straightforwardly analyzed as an intensified form of the object pronoun ham 'him', see (3e). While English has an intensified form of subject pronouns which can be analyzed as a combination of pronoun + adnominal intensifier, e.g. he himself, the corresponding form for the object pronoun, which would be him himself, is extremely rare (though not unattested) and is generally not accepted by native speakers. Instead of him himself, the reduced form himself is usually used 21 22 • See chapter 3, section 3.4.2 for discussion of the semantic/pragmatic principles governing the distribution of intensifier-adjunction to object pronouns. 21 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 x-self 340 5.2.3.3 Summary: binding of intensified and unintensified pronouns While only a rough sketch, the analysis of intensified pronouns in English outlined above, suggest that it may be possible to extend the analysis of Danish pronominals presented in chapter 3, section 3.4 to English, and thus supports the viability of the particular separation of binding and intensification proposed here. 5.2.4 Conclusion In the above sub-sections of section 5.2 we have discussed the cross-linguistic viability of the analysis binding and intensification in Danish proposed in chapter 2-4, by showing to what extent it can be extended to account for similar phenomena in Modem English. While a number of differences between the two languages were noted, e.g. the absence of simple unintensified 0-reflexives in ECM constructions and resultatives, we argued that similarities outweighed the differences. We also argued that most differences can be accounted for by reference to factors not related to either binding or intensification, e.g. prosodic differences 22 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 */??me myself, */??him himself 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 myself and us ourselves. Baker (1995) suggests that the absence of intensified object pronouns might be a consequence of his prominence condition (see (100) in section 3.5.1 ). 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 himself 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 himselt' "The king himself (ii) Huang-di ziji Huang-di ta ziji (Chinese) emperor self emperor him self "The emperor himself "The emperor himself.' 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) (iv) him same him him-same ta ziji him self *ta ta ziji him him self (Chinese) 341 between overt (sig) and non-overt reflexive elements (0-reflexives), inalienable possession, etc. All in all, in spite of the fact that the analysis of English outlined here may have raised more questions that it answers, we hope to have shown that this approach is at least a viable alternative which deserves to be explored further. In the following section, we show that the 0-reflexive analysis of English presented here allows for interesting new solutions to hitherto unanswered questions concerning the historical development of reflexives and intensifiers from Old English (OE) to Early Modern English (EME) and eventually Modern English. 5.3 Historical development of English intensifiers The Germanic languages have had intensifiers of the self-type for as long back as historical records go. For this reason very little is known for certain about the early historical development of self-type intensifiers. Section 5.3.1 provides a brief summary of current theories of the historical development of Germanic intensifiers. Similarly, section 5.3.2 contains an overview of previous accounts of the historical development of Modern English x selfforms. In section 5.3.3 it is argued that the assumptions that English has 0-reflexives and that the compound himse(f is always an intensifier provide the key to a coherent, unified account of a number of the hitherto unanswered questions concerning the development of the Modern English x-self form intensifiers. Finally, section 5.3.4 summarizes 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 of selv 'self' in other Germanic languages and in (86b) the reconstructed Germanic form can be found. (86) 23 (a) Cognates of selv in the Germanic languages 24 : Danish selv Swedish sjalv Old Danish sialf Old Norse sjalfr Faroese sjalvur Norwegian (bokmal) selv Norwegian (nynorsk) sjol En~ish ~elf Old English se(o)lf, sylf, self German selbst Old High German selp Dutch zelf Afrikaans self Gothic silba (b) Reconstruction form of Common Germanic origin: Germ. *selba- . *selban 342 In many languages the morphological material of which the intensifiers (as well as reflexive elements) are made can be shown to derive from expressions for 'body parts', see (87). (87) 25 a. Georgian b. Okinawan c. Arabic d. Hausa e. Hungarian f. Japanese g. Haitian Creole h. Classical Chinese tv it on du:na nafs ni daikana mag a }i-s in ti!t-Emile %shen 'body' 'body' 'soul' 'I with my head' 'seed' 'own body' (lit.)'head Emile' ='Emile himself' 'body' According to E. Konig (1996: I 0), the cross-linguistic evidence in (87) provides support for Grimm's suggestion that German selb-st originally derived from the possessive construction si-lib 'his body', see (88). (88) selv (Dan.) <<< *selba- (Germ.)<<< *si + 'his' + *lib-, *liba 'life, body' 23 The examples in (86a) as well as the reconstructions in (86b) are partly from Niels Age Nielsen (1989) Dansk Etymologisk Ordbog. Ordenes Historie. Gyldendalske Boghandel, Nordisk Forlag A/S, p. 363, partly from Siemund (1999:22). 24 Cf. Siemund (1999:22) ·A notable exception is Yiddish, which has lost SELF and uses aleyn (cf. E. alone) as the intensifier.' The loss of selftype intensifiers in Yiddish could, perhaps, by due to influence from neighboring Slavic languages also have adnominal intensifiers meaning 'alone', e.g. Russian sam 'alone, himself. 25 Cf. Moravcsik (1972), Schladt (1995). 343 Based on this kind of data Konig furthermore suggests that intensifiers play a prominent role in the historical developments of reflexive pronouns, see the path of grammaticalization illustrated in (89). (89) 26 body parts > intensifiers > reflexive anaphors Konig thus takes intensifiers to be the (missing) link between words for body parts and reflexive anaphors. The fact that intensifiers and reflexives are closely related historically in many languages does not contradict our basic assumption that binding and intensification belong to separate modules of the grammar. Instead, as discussed in the following 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 self-forms Modern English reflexives and intensifiers constitute a typical example of apparent convergence of the formal realization of intensifiers and reflexives and can thus be used to illustrate the urgent need to relegate binding and intensification to separate modules of 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 of the evolution of Modern English x-self forms. To the best of our knowledge, all the existing studies of the historical development of the compound self-forms are based on the assumption that these elements function as reflexive pronouns in Modern English. As argued in the previous sections, there is plenty of data indicating that this assumption may be wrong. That is, as argued in chapter 5, Modern English has clitic-like 0-reflexives which may be intensified by adjunction of the appropriate x-self adnominal intensifier, e.g. 0 himself, 0 ourselves, etc. In other words, the x-self forms are never true reflexives but rather consistently function as adnominal intensifiers in all contexts. 26 Cf. Konig ( 1996: I 0). 344 Before we develop our analysis of the evolution of 0-reflexives and self-form in Modern English let us first briefly summarize the status quo of the history of English himself Basically most (if not all) existing analyses agree that the desire to disambiguate between coreference readings and disjoint reference readings of local pronouns was (one of) the main motivation for the development of the ME self-forms. In this respect, Siemund's (1999:25-30) account is quite representative 27 • Old English did not have reflexive pronouns and thus used personal pronouns instead. This yielded a situation in which third person personal pronouns were ambiguous between co- reference and disjoint reference interpretations, see (90). (90i 8 a. b. hine he beweradh mid wcepnum 'he defended himself with weapons' dha 29 behydde Adam hine & his wif eac swa dyde 'and Adam hid himself and his wife did the same' [ JEGram 96.11] [Gen 3.9] According to most traditional accounts the intensifier seolf'self' was added to the ambiguous local pronouns for the sake of disambiguation, see (91-2). (91 ) 30 a. b. (92) a. se Hcelende sealde hine sylfnefor w,.3 1 [JELet 4 1129] 'The Saviour gave himself for us.' He [Moses} sceawode hine selfne, & pinsode, dha dha him dhuhte dhcet he hit doon ne meahte, ... [CP 7.51.14, translation provided] 'He contemplated himself, and thought that he could not do it, ... ' he ... seldh Gode his cehta, & hine selfne diobule [CP 44.327.23,] 'He gives his possessions to God and himself to the devil.' 27 There is a large number of 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 Sicmund 1999, (2.44). 29 The letter sequences 'dh' and 'th' are used here to transcribe the Old English letters for the voiced and voiceless (inter-)dental fricatives respectively. 30 These examples are from Siemund 1999 (2.45-6). 31 Bold face is added to Siemund's examples to help the reader locate the relevant intensified expressions. The numbers ofSiemund's examples have also been modified to fit into the present context. b. Hannibal ... hine selfne mid atre acwealde 'Hannibal killed himself with poison.' 345 [Or 4 11.11 0.2] While 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 disambiguation of local pronouns was not the only function of intensifiers. As argued in chapter 2 and 3 above, we argue that intensifier adjunction to locally bound elements is due to other semantic factors (i.e. presupposed representational non-identity) rather than directly linked to binding and that such an account allows for the potential unification of all used of intensifiers. (93) 32 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 wish ye to be wise for good and simple for evil. And again, the Lord spoke about the same thing through himself to his elect: "Be cunning as adders and simple as pigeons."' While the traditional disambiguation account seems very persuasive, we thus argue that it is false since it is based on wrong assumptions. Siemund himself admits that such disambiguation accounts run into a number of problems. Notably, they do not explain why the fusion of self and pronominal elements also took place in the first and second persons where there could be no ambiguity between coreference and disjoint reference. As discussed in Siemund ( 1999:25-30), "analogy" has often been adduced as answer to the question of why self-adjunction happened in theist and 2nd (cf. Penning (1875:13)). In contrast, the analysis proposed here provides a straightforward explanation of why the self-forms emerged in the I st and 2nd persons as well as in the 3'd person. In a nutshell, our proposal is that self-adjunction to pronouns in OE was not primarily motivated by a need to disambiguate between coreference and disjoint reference but rather by a need to adnominally intensify locally bound simple 32 This example is from Siemund 1999 (2.47). 346 pronominals occurring in object position of anti-reflexive or 'hidden neutral' verbs, e.g. (91-2). That is we propose that self-adjunction to locally bound pronominal in OE follows the same pattern as selv-adjunction to simple reflexives with anti-reflexive predicates in Modem Danish (e.g. Jeg mistcenker *mig I mig selv 'I suspect *me I me self'), see discussion of Danish reflexives in chap. 3. Another problem encountered by previous accounts is the lack of explanations for why the compound self-forms came to be used as adnominal intensifiers (e.g. ME the King himse(f) and why the OE simple intensifier self was abandoned. The examples in (9) below show that Old English used the simple form self as adnominal intensifier. If that is the case, then why did OE abandon the use of self as adnominal intensifier in favor of the self-forms, which according to standard account developed as a complex reflexive pronoun? (94) 33 a. b. se cyning sylfa 'the king himself Swa swa Crist sylf cwcedh 'as Christ himself said' [GD 14.131.3] [JELet 3 173] As discussed below, this problem becomes much easier to understand once we realize that the self-forms did NOT develop as complex reflexive pronouns but rather were adnominal intensifiers right from the first moment the pronominal and the adjoined self fused into one word. Finally, let us take a closer look at an often overlooked aspect of Old English, which we, following suggestions by Siemund (2000), believe to be crucial to the development of 0- reflexives in Modern English. Old English had obligatory pronouns in many contexts where Modern Danish, German and French have non-thematic reflexives, i.e. with deponent inherently reflexive verbs, see Old English examples in (95) and German and Danish examples in (96), see also the discussion of non-reflexive uses of 0-reflexives in section 5.2.2 above. 33 This example is from Siemund 1999 (2.48). (95) 34 a. b. c. (96) 35 a. b. c. d. . .. dhcet dhu dhin scamige, Sidon ' ... that you be aschamed, Sidon.' hit is cyn dhcet we ure scomigen 'It is appropriate that we be ashamed' he gereste hine on dhone seofothan dceg 'He rested on the seventh day.' Nero wandte sich /*sich selbst an Paul. Nero turned REFLIREFL self to Paul 'Nero turned to Paul.' Er schamt sich/*sich selbst. He shamed REFLIREFL self 'He was ashamed.' Peter skammede sigl*sig selv. Peter shamed REFLIREFL self 'Peter was ashamed.' Peter hvilte sig/*sig selv. Peter rested REFLIREFL self 'Peter rested.' 347 [CP 52.409.33] [CP 52.407.15] [Gen 2.2] (German) (German) (Danish) (Danish) Siemund (1999:25-30) draws attention to an often overlooked aspect of the evolution of Modern English, namely the loss of locally bound pronouns in object position ofverbs like the ones exemplified in (95). He also suggests that we need to take the question "why English did not start anew to develop pronouns with these functions" into consideration. As will be shown below, the single assumption that at a certain point in history a recent ancestor of Modern English developed a 0-retlexive, provides the key to a coherent, unified account of all the four questions raised in the passages from Siemund ( 1999) quoted above and summarized in (97). (97) Questions concerning the development of the x-selfforms in Modern English: (a) Why was the fusion of pronoun and self not confined to the third person, but was extended to the first and second persons? This remains a mystery for all the disambiguation accounts since no ambiguity arises in these cases. (b) Why did the newly developed self-compound forms also come to be used as intensifiers? (c) Why did English abandon its original intensifier self? Since Old English self was used in much the same way as its modern equivalent and there seems to have been no obvious reason to replace it with a new intensifier. 34 These OE examples as well as the German examples in (96) are from Siemund (2000). 35 The absence of intensified forms of the reflexive in these examples is, of course, due to the fact that they involve inherently reflexive predicates. 348 (d) Why 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 self-forms In the following we will outline an alternative picture of the evolution of the English system of reflexives and intensifiers which provides new answers to the questions listed in (97). As will be shown, the problem with most of these questions is that they are based on the wrong basic assumptions and consequently make us ask the wrong questions. Oversimplifying the immense complexity of the evolution of Modem English somewhat we have decided to cut the period from spanning from Old English to Modem English into three stages: (i) Old English, (ii) Early Modem English, and (iii) Modem English. For ease of presentation a few other simplifications have been made as well, see (98). (98) Simplifications adopted in the following sections: (a) In order to facilitate comparison between different stages, Modem English spelling and morphology has been used even in the Old English examples. (b) Case complications have been ignored. That is, the fact that first and second person forms (i.e. myself, yourself, ourselves, yourselves) are composed of a pronoun in the genitive plus self whereas the third person forms (i.e. himself, herself, itself, themselves) are composed of pronouns in accusative or dative case plus selfhas been ignored. 5.3.3.1 Stage one: intensification and binding in Old English Old English was a language without morphologically specialized reflexive pronouns. This meant that pronouns could do double duty as either reflexives or pronouns. See the example in (90), repeated in (99). (99) hinei/z he, beweradh mid wcepnum 'hei defended himself;/himz with weapons' [JEGram 96.11] (=Siemund 1999, (2.44)) As shown in (99), locally bound pronouns were potentially ambiguous between coreference and disjoint reference readings. According to most existing accounts of the evolution of x-self 349 forms (see section 5.3.2), self was adjoined to locally bound pronouns to disambiguate in favor of coreference readings, see (92) repeated in (100). (I 00) Hannibal, ... hine, *z selfne mid atre acwealde 'Hannibal, killed himself,*= with poison.' [Or 4 11.110.2] (=Siemund 1999 (2.46b )) Unlike, (99) where the pronoun is ambiguous, (I 00) tend to have only the coreference reading of the locally bound object pronoun hine 'him'. In contrast to the standard accounts, according to which disambiguation was the main motivation for self-adjunction to local pronouns, we argue that self-adjunction was motivated by a need to mark representational non-identity of the antecedent and the locally bound pronoun with anti-reflexive (or "hidden" neutral predicates, e.g. (I 00)) in exactly the same was as se/v-adjunction to simple reflexives in Modern Danish, see chapter 3. The distribution of Danish pronouns and reflexives (e.g. ham 'him', sig 'REFL 3 person', mig 'me') with these different verb types is illustrated in (101)-(102). (101) Modern Danish third person singular pronoun ham 'him': a. Peter, mistcenkte ham*iz· (anti-reflexive predicate) 'Peteri suspected him*ilz·' b. Peter, vaskede ham*iz· (neutral predicate) 'Peteri washed him*i/z·' c. Peter, hvilte *ham.i *z· (inherently reflexive predicate) '*Peter rested him.' (102) Modern Danish third person singular reflexive sig 'REFL': a. Peter mistcenkte *sig/sig, *z selv. (anti-reflexive predicate) 'Peteri suspected himsel~;•z·' b. Peter vaskede sig, •/sig, *z selv. (neutral predicate) 'Peteri washed himsel~;*z·' c. Peter, hvilte sigi •l*sig selv. (inherently reflexive predicate) 'Peter rested *??himself.' Unlike Old English, Danish has a distinction between reflexives (e.g. sig 'REFL') and pronouns (e.g. ham 'him'). As discussed in chapter 3, the distribution of sig and ham fall under principle A and principle B respectively: sig 'REFL' is an anaphor which must be bound in a local domain (viz. the tensed clause), and ham 'him' is a pronoun which must be free in the same local domain. As (I 01 c) shows, inherently reflexive verbs do not allow pronouns in 350 object position. See section 3.3.2.2 in chapter 3 for discussion of the semantics of inherently reflexive predicates and ofthe reasons why unintensified reflexives are required in such cases, see (102c). In contrast, Old English allowed pronouns in such positions, see (95). Note furthermore, that in spite of the fact that sig 'REFL' in ( 1 02a) is not ambiguous between a co-referential and a disjoint reference reading, the presence of the intensifier selv 'self' is still mandatory as a marker of presupposed representational non-identity with anti- reflexive predicates. Note also that selv-intensification of sig is optional with neutral predicates in (102b) and impossible with inherently reflexive predicates in (102c). This distribution of selv clearly shows that selv-intensification of reflexive is not triggered by a need for disambiguation in Modern Danish, but rather by the semantics of the predicate: anti- reflexive predicates require intensification of reflexives to mark the reflexive reading (in contexts which presuppose representational non-identity, see section 3.3.2.1 in chapter 3) and se/v-intensification is excluded with inherently reflexive predicates since in those contexts generation of a contrast set of alternatives is blocked by the meaning of the predicate which presupposes the identity of its arguments. Our analysis of the evolution of himself in Modern English is based on the assumption that in Old English self-intensification functioned like selv- intensification does in Modern Danish, i.e. it was triggered more by a need to license representational non-identity than by a need to disambiguate. The first person examples in (103) further corroborate this account of Modern Danish. (103) Modern Danish first person pronoun/reflexive mig 'me': a. Jeg mistcenker *mig/mig, *z selv. (anti-reflexive predicate) 'Ii suspect myselfi:•z·' b. Jeg vasker mig, •/mig, *z selv. (neutral predicate) 'Ii wash 0/myselfi:•z·' c. Jeg hviler mig, •/*mig selv. (inherently reflexive predicate) 'I rest (*myself).' 351 Danish does not distinguish between pronouns and reflexives in the first and second persons. In spite of this fact, no ambiguity is - for obvious pragmatic reasons - possible. But even so, the presence of selv is still mandatory with anti-reflexive verbs, optional with neutral verbs, and impossible with inherently reflexive verbs, see (I 03a) and (I 03c) above. As the examples in (I 03) illustrate the fact that the creation x-selfforms took place in I stand 2"d persons as well as in the 3rd person is no mystery in the account proposed here which, consequently, does not need to recur to vague notions of 'analogy'. Note that the semantic/pragmatic approach of intensification proposed here stands a better chance of explaining why self-adjunction to locally bound pronouns did not always result in unambiguously coreferential readings, see (93) repeated in (1 04). (104) Be dham cwa:dh se a:dhela lareow sanctus Paulus: Jc wille dha:t ge sien wise to gode & bilwite to yfele. Ond eft be dha:m cwa:dh Dryhten dhurh hine selfne to his gecorenum: Beo ge swa ware sua sua na:dran & sua bilwite sua culfran. [CP 3S.237.18, translation provided] 'Therefore the noble teacher St. Paul said: "I wish ye to be wise for good and simple for evil. And again, the Lord spoke about the same thing through himself to his elect: "Be cunning as adders and simple as pigeons."' (=Siemund 1999 (2.47)) The examples ( 1 OS)-( I 06) illustrate the predictions this approach makes for the interpretation of locally bound object pronouns with the three main types of predicates, i.e. anti-reflexive predicates (e.g. suspect), see (lOSa) and (106a), neutral predicates (e.g. wash), see (lOSb) and (106b), and inherently reflexive predicates (e.g. rest), see (lOSe) and (106c). Note that the some of the OE data in ( 1 OS) are based on extrapolations rather than attested examples. Of the sentences in (lOS), we only have attested examples corroborating the judgments in (lOSe). The judgments in (IOSa,b) are based on the behavior of other typical anti-reflexive/'hidden' neutral predicates (e.g. give and kill in 91-2) and other typical neutral predicates (e.g. defend in (90)). Needless to say, more careful and exhaustive studies of the behavior of different verb types in OE is still needed. At this point the analysis of OE presented here is thus more an 352 outline of a new approach than a fully-fledged analysis whose empirical coverage has been carefully researched. ( 1 05) Stage one: unintensified locally bound 3rd person pronouns in Old English: a. Peter, suspected himm, z· (anti-reflexive)( compare with ( 100)) b. Peter, washed him, z· (neutral) (see (99)) c. Peter, rested him, *z· (inher. reflex.) (= (95c)) (106) Stage one: unintensified locally bound 1 51 person pronouns in Old English: a. I, suspected mem, *z· (anti-reflexive)(compare with ( 100)) b. I, washed mei *z· (neutral) (see also (99)) c. Ii rested me, *z· (inher. reflex.) (see (95c)) As shown in (105a), anti-reflexive predicates are predicted to require self-adjunction to the simple pronouns in order to mark the presupposed representational non-identity. This also applies to the "hidden" neutral 36 predicate acwealde 'kill' in (100). Neutral predicates, which evoke no presuppositions concerning the representational identity of their arguments, may or may not display self-adjunction, see (105b) and (99). Finally, the since they presuppose identity of their arguments, inherently reflexive predicates do not allow non-coreferential DPs in object positions. Therefore, since no focus-based contrast set of alternatives can possibly be generated, non-coreferential readings of locally bound object pronouns is not allowed, see (lOSe). The examples in (I 07)-( I 08) illustrate how the intensifier-adjunction to simple pronouns function to license representational non-identity, viz. in the cases of the anti- reflexive predicates in (I 07 -8a,d). Note also that, since intensification requires that the generation of a contrast of alternatives to the referent of the adnominally intensified nominal expression, intensification of simple pronouns in object position of inherently reflexive verbs 36 See section 6.3. for a detinition of"hidden·· neutral predicates. 353 is under no circumstances allowed, see (1 07-8c,f). Finally, intensification of simple pronouns with neutral verbs is optional, see (107-8b,et. ( 1 07) Stage one: intensified locally bound 3rd person pronouns in Old English: Intensified pronouns: Unintensified pronouns: a. Peter, suspected him, 'z se(f d. Peter, suspected him,, z· b. Peter; washed him,'= self e. Peter, washed him, z· c. Peter, rested him, *z *self f. Peter, rested him, *z· ( 1 08) Stage one: intensified locally bound 1st person pronouns in Old English: Intensified pronouns: Unintensified pronouns: a. I, suspected me, *= self d. I, suspected me, *z· b. I; washed me; *z self e. I, washed me; *z· c. I, rested me, *z *self f. I, rested me, *z· ( 1 09) Stage one: intensified and unintensified DPs and subject pronouns in Old English: Intensified: Unintensified: a. the king self, etc. c. the king b. he self. I self, etc. d. he, I, etc. At this stage the primary function of self-adjunction was not to disambiguate third person pronouns but to mark representational non- identity with in anti-reflexive contexts, cf. the Modern Danish examples in (I 02a) and (I 03a). Note also the existence of examples like (8), which bear witness to the fact that even intensification of third person pronouns was not always triggered by a desire to disambiguate. In other words, we propose that the so-called disambiguating function is derivative of the primary intensifying function illustrated by the contrast between (107b) and (107e). Note that a this stage we assume that there still has been no fusion of him+self The fact that self-intensification was motivated by a need to license co- reference scenarios which would otherwise be infelicitous with anti-reflexive predicates, e.g. (I 07a,d) and ( 1 08a,d), and not by a need to disambiguate 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 wrong question to ask in the first place. The system of intensified and 37 As discussed in chapter 3. section 3.3.2.5. intensifier-adjunction to simple retlexives with neutral predicates is never truly optional, but rather dependent on focus and other discourse factors. 354 unintensified nominal expression in Old English (stages one and two) is summarized in the table in (11 0). ( 11 0) Stage one: system of intensification of nominal expressions in Old English: Unintensified Intensified a. Loc. bound pr. him him self b. Subj.pronoun he he self c. Obj.pronoun him him self d. DP the king the king self As shown in table 1, Old Eng! ish had a morphologically transparent and fully compositional system of intensification in which all kinds of nominal expressions (DPs and pronominals) were intensified through the same process of self-adjunction. In this respect Old English is similar to Modem Danish, see the table in (111 ). ( 111) System of intensification of nominal expressions in Modem Danish: Unintensified Intensified a. Reflexive sig 'himself/herself/etc.' sig selv 'himself/herself' b. Subj. pronoun han 'he' han selv 'he himself' c. Obj. pronoun ham 'him' ham selv 'him himself/himself' d. DP kongen 'the king' kongen selv 'the king himself 5.3.3.2 Stage two: intensification and binding in Early Modern English The schemas in (112)-(114) summarize the changes which took place between Old English (OE) and Early Modem English (EME). In EME the locally bound pronouns of OE were replaced by 0-reflexives while locally free pronouns remained, see ((112-3), and the OE intensifier se?fhas been replaced by himselfas intensifier across the board, see (114). (112) The evolution ofadnominal intensifiers: a. (OE) self >> b. (EME) himself ( 113) The evolution oflocally bound pronouns: a. (OE) him >> b. (EME) 0-reflexive 355 (114) The evolution oflocally free pronouns: a. (OE) him >> b. (EME) him The three changes in ( 112-4) is all that is needed to account for how the Early Modern English system of binding and intensification, see (115), evolved out of Old English. Notice that the 0- reflexive analysis proposed here allow us to account for EME as having the same kind of morphologically compositional system of intensification as the ones found in Old English and Modern Danish, see ( 11 0-1 ), repeated as ( 116-7). (115) Stage two: System of intensification of nominal expressions in Early Modern English: Unintensified Intensified a. Reflexive 0 0 himself b. Subj.pronoun he he himself c. Obj.pronoun him him himself d. DP the king' the king himself (116) Stage one: system of intensification ofnominal expressions in Old English: Unintensified Intensified a. Loc. bound pr. him him self b. Subj.pronoun he he self c. Obj.pronoun him him self d. DP the king the king self ( 117) System of intensification of nominal expressions in Modern Danish: Unintensified Intensified a. Reflexive sig 'himself/herself/etc.' sig selv 'himself/herself' b. Subj. pronoun han 'he' han selv 'he himself' c. Obj. pronoun ham 'him' ham selv 'him himself/himself' d. DP kongen 'the king' kongen selv 'the king himself In other words, the 0-reflexive approach allows us to account of the evolution of the EME system in (115) from the OE system in (116) without stipulating any shift from one type of system to another. Basically the morphological compositionality of the system of intensification remains same modulo a change in the form of the adnominal intensifier from the simple x-selfform to the complex x-selfforms 38 • Furthermore, a new reflexive pronoun, 38 Notice that this change from simple to complex form of the intensifier closely mirrors the (still ongoing) change of the older simple intensifier -meme into the complex intensifier lui-meme in Modern French, e.g. le roi meme 'the king self >> le roi lui-meme 'the king himself. Notice also the close parallels in behavior between the clitic- 356 the 0-retlexive was introduced to replace the locally bound pronouns in OE. We argue that these two changes are intimately connected. That is, we claim that the analysis proposed here can explain the link between the loss of locally bound pronouns and the formal change of the adnominal intensifier from selfto himself, see questions (b-d) in (118). First let us illustrate the changes described in (118-121) with concrete examples 39 . ( 118) The evolution of unintensi tied locally bound pronouns: Old English Early Modern English a. ??*Peter 1 suspected him 1 • >> *Peter 1 suspected 0 1 • b. Peter 1 washed him 1 • >> Peter 1 washed 0 1 • c. Peter 1 rested him 1 • >> Peter 1 rested 0 1 • d.. Peter 1 took a knife with him 1 • >> *Peter 1 took a knife with 01. ( 119) The evolution of intensified locally bound pronouns: Old English Early Modern English a. Peter 1 suspected him 1 self >> Peter; suspected 0 1 himself b. Peter 1 washed him 1 self >> Peter 1 washed 0 1 himself c. *Peter 1 rested him 1 self >> *Peter 1 rested 0 1 himself d.. * Peter 1 took a knife with him 1 se(f >> *Peter 1 took a knife with 0 1 himself (120) The evolution of unintensified locally free pronouns: Old English Early Modern English a. Peter 1 suspected him 2 • >> Peter 1 suspected himz. b. Peter 1 washed himz. >> Peter 1 washed himz. c. *Peter 1 rested himz. >> *Peter 1 rested himz. d.. *Peter 1 took a knife with himz. >> *Peter 1 took a knife with himz. ( 121) The evolution of unintensified locally free pronouns: Old English Early Modern English a. ...Peter 1 [.}except himz self >> ... Peter;[.} except (him)z himself The examples in (118-9) illustrate the consequences of the shift from locally bound pronouns to 0-retlexives. As shown, the distribution of unintensified and intensified forms is the same in both stages of the languages, i.e. anti-reflexives (due to the presupposition of like @-reflexive in English and the French reflexive clitic se 'RELF'. Similarly the simple and complex pronouns lui 'him' and lui-meme 'himself also closely mirror the behavior of their English counterparts him and locally free/logophoric himself Interestingly, the shift from locally bound pronouns to 0-reflcxivcs 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 (:7-rcflexivcs was due to influence from French. 39 As mentioned above. these OE "examples" arc based in large part on extrapolations from similar data. More work is needed to find attested examples for all these j udgmcnts. 357 representational non-identity) require intensification, see (118a) and (119a), neutrals are compatible with both intensified and unintensified forms, see ( 118b) and ( 119b ), and inherently reflexives are only compatible with unintensified forms, see (118c-d) and (119c-d). The only unexpected judgment is ungrammaticality of ( 118d) with simple unintensified 0- reflexive. However, this "exception" can be explained as a consequence of the clitic-like behavior of 0-reflexives, i.e. clitics can only occur in direct argument-positions (cf. Fr. Pierre estfier de *se/soi 'Pierre is proud ofREFL'). Since the 0-clitic cannot is barred in (188d) for morphological reasons, and since the inherently reflexive semantics of the expression take a knife with _ excludes intensifier-adjunction, the personal pronoun him is the only overt realization offered by the morphology of the language. Note that Danish has simple unintensified sig in such expressions, see chapter 3, section 3.3.6. The examples in (120-1) illustrate the fate of locally free pronouns in OE. Unlike locally bound pronouns, locally free pronouns remained unchanged in EME. Likewise, adnominal intensification of pronouns took place in the same environments in OE and EME, i.e. with in contrastive contexts, e.g. Peter [said that Mary danced with everybody] except himself, see (121b). Note that in EME locally free intensified pronouns could take the form him himself, her herself, us ourselves, etc. which is banned for phonological reasons (i.e. possibly haplology) in most versions of Modem English. These forms thus clearly show that the form ofthe intensifier had changed from selfto himself Note also that the total absence of examples like (122) in the history of English, is a witness to the fact that the change locally bound him >> 0-reflexive took place simultaneously with (or just prior to but not after) the change se(f>> himself (122) *Peter, washes/suspects him; himself 358 In other words, the evolution of OE into EME must have taken place as follows. First the simple intensifier began to be adjoined to locally bound pronouns in OE to mark representational non-identity in anti-reflexive contexts. Since the majority of verbs are either anti-reflexive or "hidden" neutral predicates this happened in a large number of contexts. Hence the combination 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. As a certain point the association between unintensified pronouns and disjoint reference readings became so strong that the deletion of 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 time the combination him+self was reinterpreted as an adnominal intensifier, see (123). As mentioned above, the emergence of himself as intensifier is corroborated by the simultaneous emergence of intensified (locally free) pronouns of the form him himself, and intensified DPs of the form the king himself (123) The creation of compound 'reflexives'/intensifiers: him+self me+self etc. ==> himself myself 0 , etc. (=='reflexives'/intensifier which could only be used in contrastive contexts) (123) illustrates the fusion of him+self me+self, etc. into the so-called compound intensifiers himself, myself, etc., which took place fusion between OE and EME. The compound x-self forms which arose through this fusion process have a very peculiar and somewhat surprising property: they are never found in inherently reflexive contexts (which is exactly the kind of context where one would expect to find reflexives rather than pronouns, see Danish examples in ( 125)), see (124 ). ( 124) Peter took a knife with him/* himself. 40 As mentioned above, we do not take the case complications into account here. ( 125) Peter tog en kniv med sigl*sig selv. Peter took a knife with REFLIREFL self 'Peter took a knife with him/*himself.' 359 The fact that the x-selfforms are found mainly in contrastive situations confirms our analysis of them as adnominal intensifiers. That is, they have the same meaning contribution and (to a large extent) the same distribution as the simple intensifier selv 'self' in Danish and the simple intensifier self in Old Eng! ish, see ( 125-7). The assumption that English developed 0-reflexives is also necessary to explain why sentences like (126a) are bad, while its OE and Modem Danish/Dutch counterparts in (126b,c) are ok. (126) a. b. c. !shave *me. Jeg barberer mig. 'I shave me.' Ich scheere mich. 'I shave me.' (Danish) (Dutch) The examples in (127-132) illustrate that correspondence between English 0-reflexives and simple unintensified reflexives in Danish, column (iii) in (127-132). Note that while both Danish and English distinguish between 3rd person pronouns and reflexives, compare columns (i) and (iii) in (127-129), only English distinguish formally between 1 51 (and 2nd) person pronouns and reflexives, i.e. me vs. 0-reflexive, see column (i) and (iii) in (130-132). As shown in columns (i) and (iii) in (130-132) Danish 1st and (2nd) person pronouns and reflexives share the same formal expression, e.g. mig 'me' in columns (i) and (iii) in (130-132). (127) Anti-reflexive verbs: (i} gron.: (ii} 0-gron. (iii} unint. refl. (iv} int. refl. c. I, suspect *me; I *0z I *0, I 0, myself d. Jeg, mistcenker (mig,) I *0z I *mig, I mig, selv. (128) Neutral verbs: a. I, wash *me; I @z I 0; I 0, myself b Jeg,vasker (mig) I 0: I mig, I mig, selv. 360 (129) Inherently reflexive verbs: a. I, rest *mei I *0= I 0, I *0, myself b. Jeg, hviler (mig,) I *0= I mig, I *mig, selv. (130) Anti-reflexive verbs: (i) Qron.: (ii) 0-Qron. (iii) unint. refl. (iv) int. refl. e. He, suspects *himi I *@z I *0, I 0, himself f. Hani mistcenker *ham/ *@z I *sig, I sig, selv. (131) Neutral verbs: a. He, washes *him, I @z I 0, I 0, himself b Han, vasker *ham, I @z I sig, I sig, selv. ( 132) Inherently reflexive verbs: a. He, rests *him, I *@z I gi I *0i himse(f b. Han, hviler *hami I *@z I sigi I *sig, selv. Assuming the English system of intensified nominal expressions to follow the Danish pattern illustrated in (127-132), yields the picture of Early Modern 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 of the appropriate form of the complex intensifier himself( e.g. 0 himself). (133) Stage two (third Qerson): Intensified reflexives: a. Peter, suspects 0, *z himself b. Peter, washes 0, *z himself c. Peter, rests 0, *z *himself (134) Stage two (first person): Intensified reflexives: a. Ii suspect 0i *z myself b. I, washe 0, *z myself c. I, rest 0, *z *myself (135) Stage two (intensified DPs, subject pronouns): Intensified exQressions: a. the king himself c. b. he himself, I myself etc. d. U nintensified reflexives: d. Peter, suspects 0., *z· e. Peter, washes 0, *z· f. Peteri rests 0, *z· Unintensified reflexives: d. I, suspect 0., *z· e. I, wash 0, *z· f. I, rest 0, *z· Unintensified expressions: the king he, I. etc. Note that the neat compositional character of the system of intensified and unintensified nominal expressions of EME illustrated in ( 115) is perfectly paralleled in the modern Danish 361 system in (117). That is, in both systems the difference between intensified and unintensified forms is signaled by the presence/absence of the adnominal intensifier selv/himself EME differs from Modern English in that it allows for intensified object pronouns, e.g. him himself(= Dan. ham selv 'him himself'). That these intensified pronouns are pronominal expressions is demonstrated 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) (137) Stage two (third person pronouns/ 1 Intensified pronouns: a. Peter, suspects him., z himself b. Peter, washes him•; z himself c. Peter, rests *him; *z *himself Stage two (first person pronouns) Intensified pronouns: a. He, suspects me., z myself b. He, washes me.,= myself c. He, rests *me., *z *myself Unintensified pronouns: d. Peter, suspects him., z· e. Peter, washes him., z· f. Peter, rests *him., *z· Unintensified pronouns: d. He, suspects me•; z· e. He, washes me., z· f. Peter, rests *me., *z· Note that, unlike Old English, EME does not allow simple unintensified pronouns in object position of inherently reflexive verbs, see (136c,f) and (137c,f). We now see that question (97c) (i.e. 'Why 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?') asks the wrong question. English did develop a new kind of pronominal element to replace the simple pronouns in object position of inherently reflexive verbs, VIZ. the simple unintensified 0- reflexive which thus can be seen as the English counterpart of the simple unintensified reflexive sig in Danish, see tables 3-4. The examples in ( 138)-( 139) illustrate the distribution of intensified and unintensified forms of the 1st and 3rd person pronouns in Modern Danish. As described above, close comparison of the English stage four pronominal system in and the Danish system in (127)-(132) reveals one interesting difference. While English chose 41 The judgments in (56-59) are based on the assumption that these sentences are placed in the appropriate "logophoric' contexts where the intensified pronouns are bound by a non-local antecedent referring to a subject of consciousness (cf. Zribi-Hertz ( 1989), Kuno ( 1987), Huang & Liu (200 I), etc.). 362 to replace all (i.e. 1 5 \ and 2nd person forms as well as 3rd persons forms) locally bound simple pronouns with the zero reflexive, Danish chose (at some point in history) to live with a system in which the reflexive and pronominal forms in the 1st and 2nd persons are formally identical, compare the ungrammatical sentence I wash me in (128a) versus the grammatical sentence jeg vasker mig '(lit.) I wash me' in (128b). Notice also that the same form mig 'me' ofthe personal pronoun is used in the non-reflexive sentence in (139e). Furthermore notice that Modern Danish does not have zero reflexives, compare ( 128a) and (128b ). This fact strongly supports the hypothesis that English zero reflexives correspond to Danish simple unintensified fl · (. E n fl · D . 'REFL' . , reflexive, d. , reflexive') S I re ex1ves I.e. ng. u re ex1ve = an. szg , mzg me , zg you . ee a so chapter 4, section 4.4.2 for a late-insertion analysis of I stand 2nd person forms in Danish. (138) ( 139) Modern Danish third person pronouns: Intensified pronouns: a. Peter, mistcenker ham., z selv. 'Peter; suspects him*ilz himself.' b. Peter, vasker ham., zselv. 'Peter; washes him•;;, himself.' c. Peter hviler *ham selv. 'Peter; rests him;;•, *himself.' Modern Danish first person pronouns: Intensified pronouns: a. Han, mistcenker mig., z selv. 'He; suspects me•;1z myself.' b. Han, vasker mig., z selv. 'He; washes me•;1z myself.' c. Han, hviler *mig selv. 'He; rests *me•;;•, *myself.' Unintensified pronouns: d. Peter, mistcenker ham•; z· 'Peter; suspects him•;;,.' e. Peter, vasker ham., z· 'Peter; washes him•;;,.' f. Peter hviler *ham. 'Peter; rests 0;;•,.' Unintensified pronouns: d. Han, mistcenker mig., z· 'He; suspects me•;;,.' e. Han, vasker mig., z· 'He; washes me•;;,.' f. Han, hviler *mig. 'Peter; rests *me•;;•,.' Note 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 become impossible in Modern English. 363 As illustrated by (136)-(137) and (115), stage two (EME) is represented by Older Modem English (e.g. Jane Austen's English 42 ) which does allow intensified object pronouns (e.g. him himself). Later, in stage five, a morphological redundancy rule eliminated such forms. 5.3.3.3 Stage three: intensification and binding in Modern English Stage five is characterized by the introduction of a morpho-(phono-)logical rule 43 banning repetition of person, number and gender features in intensified object pronouns: him himself =>[him] himself, me myse?f=> [me] myself, etc., see the table in (140). (140) System of intensification of nominal expressions in stage five (=Modem English). Unintensified Intensified a. Reflexive 0 0 himself b. Subj. pronoun he he himself c. Obj. pronoun him [him] himself (<him himself) 44 d. D P the king' the king himself 5.3.3.4 Additional evidence from the evolution of pronominal reciprocals into 0- reciprocals The observation that in many languages (e.g. Romance languages and German) the simple unintensified reflexive (se/silsich) can be used as both reciprocal and reflexive is highly relevant to our analysis of the historical development of self-intensifiers and 0-reflexives in Modem English. In the same way OE locally bound pronouns were replaced by 0-reflexives in EME so were also locally bound pronouns with reciprocal readings, see ( 141 a), replaced with 0-reciprocals, see ( 141 b). 42 See Baker (I 995) for a detailed description of Jane Austen's English. 43 Cf. Siemund's (I 999:81) rule: "lftwo expressions Eland E2 form a complex expression Ec, and ifthe semantic features ofEI are a subset of those of E2. then E I is superfluous and can be dispensed with." 44 The fact that the form him himself is 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. 364 ( 141) a. (OE) They, kissed them,. >> b. (EME) They, kissed 0,. 'They kissed 0/each other.' 'They kissed (each other).' (142) (OE) pronominal reciprocals >> (EME) 0-reciprocals This kind of evidence illustrates the close affinity between reflexives and reciprocals and provides additional support for the analysis of the historical evolution of 0- reflexives/reciprocals proposed above. 5.3.4 Summary The above discussion of the development of Modern English (ME) himself has shown that the assumptions that: (i) ME has 0-reflexives which are functionally equivalent of Dan. sig, Fr. se, Ger. sich, etc. (ii) ME x-se(fforms ofthe type himself are always intensifiers. (iii) predicate meaning (presupposed representational non-identity) affects the distribution of intensified and unintensified reflexives (e.g. Dan sig vs. sig selv, and Eng. 0 vs. 0 x-selj). The above approach to the evolution of x-self forms and 0-reflexives offer interesting new answers to all of the hitherto unanswered questions listed in (97) and repeated in (143), concerning the evolution of the compound intensifiers/reflexives himselfin Modern English. (143) Questions concerning the development ofthe x-sel[intensifier in Modern English: (a) Why was the fusion of pronoun and self not confined to the third person, but was extended to the first and second persons? This remarkable in so far as no ambiguity arises in these cases. (b) Why did the newly developed compound form also come to be used as an intensifier? (c) Why did English abandon its original intensifier self? Old English selfwas used in much the same way as its modern equivalent and there seems to have been no obvious reason to replace it with a new intensifier. (d) Why 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 wrong way to begin with. Since the adjunction of se(f to pronouns in Old English was not primarily motivated by a need for disambiguation (but rather by a need to mark representational non-identity in anti-reflexive, 365 see (91-2), or inherently contrastive contexts, see (93), it is not at all surprising but, indeed, expected that the fusion of pronoun and self should be extended to the first and second persons. Indeed, as shown be the Danish examples discussed above, even in the absence of ambiguity (between coreference and disjoint reference readings) selv-adjunction to pronouns and reflexives is still necessary to mark representational non-identity with anti-reflexive and 'hidden neutral' predicates. There is therefore nothing remarkable in the fact that the fusion of self and pronoun was took place across the board regardless disambiguation issues. As for questions (143b) and (143c) the answer lies in the 0-reflexive approach. As argued in section 5.3.3.2, the fact that the combination him+self came to be associated with reflexive readings lead to the unintensified pronouns being associated with non-reflexive, disjoint reference readings. At a certain point the association between unintensified pronouns and disjoint reference readings became so strong that the deletion of locally bound pronouns was enforced, e.g. pronouns could no longer be used with inherently reflexive predicates. Thus the 0-reflexive were created. And at the same time, the combination him+self was reinterpreted as an adnominal intensifier. The tables in (144-147), neatly summarizes the account of historical development of reflexives and intensifiers in Modem English and its parallels to the system of intensification of nominal expression in Modem Danish. (144) Stage one: system of intensification ofnominal expressions in Old English. Unintensified Intensified a. Loc. bound pr. him him self b. Subj.pronoun he he self c. Obj.pronoun him him self d. DP the king the king self 366 (145) Stage two: system of intensification of nominal expressions in Early Modern English. Unintensified Intensified a. Reflexive 0 0 himself b. Subj.pronoun he he himself c. Obj.pronoun him him himself d. DP the king· the king himself (146) Stage three: system of intensification of nominal expressions in Modern English. Unintensified Intensified a. Reflexive 0 0 himself b. Subj.pronoun he he himself c. Obj.pronoun him (him) himself (<him himself) d. DP the king' the king himself ( 147) System of intensification of nominal expressions in Modern Danish. Unintensified Intensified a. Reflexive sig 'himself/herself/etc.' sig selv 'himself/herself' b. Subj. pronoun han 'he' han selv 'he himself' c. Obj. pronoun ham 'him' ham selv 'him himself/himself' d. DP kongen 'the king' kongen selv 'the king himself Finally, the question in (143d) receives a straightforward explanation if we assume that the pronouns in object position of inherently reflexive verbs were replaced by 0-reflexives. Further evidence for the existence of @-reflexives is provided by the @/himself alternations found with neutral predicates, see (68). (148) Neutral verbs: a. b. !wash Jegvasker 0 I 0 myself mig/ mig selv. Rather than being optional (as is often claimed in the literature), the alternation between simple unintensified reflexive and complex intensified reflexive is motivated by the same semantic/pragmatic factors (i.e. contrastive focus, doppelganger-effects, anti-reflexivity, etc.) in both English and Danish (i.e. Eng. 0 vs. 0 myself= Dan. mig 'me' vs. mig selv 'me self'). Only the assumption that English have 0 reflexives allows for the unification of the analysis reflexives and intensifiers in English and Danish. Indeed, this assumption is, as we have argued above, the very key which unlocks that mysteries surrounding the evolution of the compound forms of the intensifier himselfin Modern English. 367 5.4. Conclusion While potentially raising more questions than it answers, we hope that the analysis of English intensifiers and reflexives outlined above has convincingly shown that the analysis of Danish based on the independence of intensification and binding presented in chapters 1-4 can be extended to both Modern and Old English. The articulation of binding and intensification proposed here has far reaching consequences for our understanding of both modules. First, in contrast to both traditional grammars and modern linguistic accounts of binding, English x-self forms are not ambiguous between intensifiers (e.g. Peter himself) and reflexives (Peter succeeded himself), but rather ALWAYS function as intensifiers. In this respect, Eng. x-self forms have more in common with the Danish intensifier selv than with the simple reflexive sig, see (149). (149) Typological survey of different uses of adnominal intensifiers: English Danish French German a. Adnom. inten. the king himself kongen selv le roi (lui-)meme der Konig selb.~t b. Complex retl. 0 himself sig se/v soi-meme sich selbst d. Doppelganger-mark 0 himself sig selv soi-meme sich selbst e. lnten. pron./logophor [him} himself ham selv [lut} lui-meme ihn selbst f. Exclus.adv. int. himself selv lui-meme selbst g. Inclus. adv. int. himself selv lui-meme selbst Furthermore, the existence of 0-reflexives in Modem English finds support in the fact that they have largely the same distribution as simple unintensified reflexives in Danish, see (150). (150) Typological survey of reflexive and non-reflexive uses of reflexive elements: 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 One of the advantages of the 0-reflexive analysis of English is that locally free reflexives no longer constitute potential exceptions to principle A. That is, the distribution of anaphors (0- reflexives and reciprocals, both 0-reciprocals and each other) is now limited to the local domain (principle A), while pronominals have to be free in the same domain. Finally, as 368 discussion in section 5.3, the 0-reflexive analysis of Modern English allows for a more straightforward analysis of the evolution of reflexives and intensifiers. Chapter 6 Independence of binding and intensification in Mandarin Chinese 369 6.1 Introduction In this chapter we discuss to what extent the analysis based on the independence of intensification and binding presented in chapters 2-5 can be applied to reflexives and intensifiers in Mandarin Chinese. Applied to Chinese this approach gives rise to the hypotheses in (I). (I) a. b. Chinese ziji E1 a 'self-self I is always an intensifier, never a reflexive. Chinese has 0-reflexives which can be adnominally intensified by adjunction of ziji. What appears to be reflexive uses of ziji are really adnominally intensified 0-reflexives, e.g. Zhangsan hen 0 ziji 'Zhangsan hates 0 himself. As will be shown below, the two hypotheses in (I) yields a radically new approach to the analysis of reflexives and intensifiers in Mandarin Chinese. 6.2 Setting the stage: evidence supporting the proposals that ziji is always an intensifier and that Chinese has 0-reflexives In this section the main properties of the element ziji will be introduced in order to set the stage for the following sections. The main focus will be on showing that there is a large body of data supporting the proposals in (1 ). The Chinese intensifier element ziji has a great deal in common with the intensifier element himselfin Modem English and it will be shown that these similarities can be used to argue that the 0-reflexive analysis of Modem English proposed in chapter 5 can be extended to Chinese. First, like the English adnominal intensifier himself, the Chinese adnominal intensifier ziji also appears to be able to function as a reflexive anaphor, see (2)-(3). 1 In the linguistic literature on Chinese El c ziji is usually glossed as 'self. In section 6.3.1 we argue that 'self self is a better gloss since it is a hi-morphemic word composed of the morphemes El zi and cji which both mean 'self. (2) (3) £wE!c3R7o Huang-di ziji lai-le. Emperor self-self come-PERF 'The emperor himself came.' ~'m·r:&E! tJ 0 Huang-di hen ziji. Emperor hate self-self 'The emperor hates himself.' 370 (ziji = adnominal intensifier) (himself= adnominal intensifier) (ziji =reflexive) (himself= reflexive) From the beginning of modern formal syntax in the sixties and seventies till today, binding-theoretical accounts of himself and ziji have tended (With a few exceptions, e.g. Moyne (1974), among others) to treat these elements purely as reflexive anaphora (3) without taking into consideration the fact that they can also be used as intensifiers as illustrated in (2). We argue that this approach is too narrow in scope and that any viable theoretical account of these elements has to take their use as intensifiers into account. Indeed, in many cases the behavior of ziji and himseffin what seems to be reflexive uses, e.g. (3), follow from the fact that they are first and foremost adnominal intensifiers. One of the peculiarities of intensifier-based reflexives like Chinese ziji and English himself is that they do not have any of the non-reflexive uses 2 (e.g. reciprocal, middle/unaccusative verbs, medio-passives, deponent verbs, etc.) displayed by simple reflexive elements 3 in many other languages (e.g. Danish sig, and -s 4 , German sich, French se, etc.), see (4). 2 Needless to say, the logophoric uses of ziji and himself (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 himself are best analyzed as reduced forms of underlying intensified pronouns himself< him himself(cf. Siemund 2002:81). In section 6.2.2 a similar approach oflogophoric uses of ziji and ta ziji will be discussed. 3 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 tor more discussion of the different uses of reflexive elements in Danish. 4 The passive suffix-sis 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. 371 (4) 5 Typological survey ofreflexive and non-reflexive uses ofreflexive elements: (i) Chinese (ii) English (iii) French (iv) German (v) Danish a. reflexive ziji himself se/soi sich sig b. reciprocal se sich -s c. middle se sich d. passive se sich -s e. deponent verbs se sich -slsig f intensifier ziji himself (lui-)meme selbst selv Notice that neither English nor Chinese have any non-reflexive uses of their so-called reflexives, viz. himself and ziji, see ( 4b-e ). While French se, Danish sig and German sich all classify as true reflexives 6 , himself does not since it is not featureless. Chinese ziji, however, is (like the Danish intensifier selv) unmarked for person, number, and gender, and could thus be a featureless "true" reflexive. In spite of this, it still behaves just like English himselfin that it cannot be used in any of the non-reflexive constructions in (4). Furthermore, both English himself and Chinese ziji can function as intensifiers, something the reflexives in Danish, French, and German are unable to do, see (4t). So except for the reflexive uses, Chinese ziji 'self-self and English himself are basically in complementary distribution with the Danish reflexive sig (as well as with Ger. sich, Fr. se, etc.). On the other hand, as discussed in Chapter 2, Chinese ziji and English himself display most of the main uses of the Danish intensifier selv, see (5). 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. 6 The term "true reflexive" is from Burzio ( 1991 ), see also chapter 3. 372 Typological survey of different uses of adnominal intensifiers: a. Adnominal intensifier b. Complex reflexive Chinese English French guo-wang ziji the king himself le roi (lui-)meme 0 ziji 0 himself soi-memellui-meme Danish kongen selv sig selv c. Doppelganger-marker d. Intensified pronoun e. Exclusive adv. inten. f. Deverbal compound 0 ziji 0 himself soi-meme {ta] ziji Ita ziji [him] himself [lui] lui-meme ziji himself lui-meme *ziji/zi-1-ji *himselj!self *meme-lauto-lsui-letc .. sig selv ham selv selv selv We take the complementary distribution of ziji and himselfvs. sig in ( 4-5) as strong support of the assumption that both ziji and himself are intensifiers rather than reflexive anaphors. The apparent reflexive uses of ziji and himself in (3) and ( 4a) can thus be explained if we assume that Chinese and English both have 0-reflexives 8 which can be intensified by the adnominal intensifiers ziji and himself respectively, e.g. 0 himself and 0 ziji in (5b) and (6a). Combined, ( 4) and (5) thus draw a picture of Chinese and Modem English in which ziji and himself are consistently analyzed as intensifiers which may adnominally intensify different types of nominal expressions, i.e. 0-reflexives, pronouns, and DPs. In other words, the absence of non-thematic uses of ziji and himself( 4b-e) plus their consistent behavior as intensifiers (5a-e) provide strong support for the analysis of intensified and unintensified nominal expressions in Chinese outlined in (6) where ziji is consistently analyzed as an intensifier (like selv in Danish and himselfin English) while 0-reflexives correspond to the simple reflexive sig in Danish. (6) Systems of intensified and unintensified nominal expressions in Chinese, Danish and English. Simple/unintensified Complex/intensified a. Reflexives (i) Chinese 0 0 ziji b. Obj. Pron. (ii) Danish sig sig selv (iii) English 0 0 himself (i) Chinese (ii) Danish (iii) English ta ham him (ta) ham [him/ ziji selv himself 7 The table in ( 4) is repeated from chapter 2 where example sentences illustrating the different uses of intensifiers can be found. 8 See chapter 5 for more discussion of 0-ref1exives in English. 9 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. 373 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 Most Pica-style LF movement approaches to LD-anaphora tend assume that Chinese ziji and Danish sig share some properties since they can both be LD-bound. Most importantly they are both assumed to be heads since only heads are able to undergo the successive cyclic movement which is necessary in order to be LD-bindable. Usually the comparison of properties of sig and ziji ends there. It is, however, important not to overlook that, as shown above, sig and ziji are different in almost all other respects and that ziji has far more properties in common with the English intensifier himself (and Danish selv 'self) than with the simple reflexive sig in Danish. Himself and ziji shares at least five properties which Danish sig does not have, see (7). (7) (i) (ii) (iii) (iv) (v) Common properties of ziji and himself Both ziji and himself can: be focused (e.g. in answers to questions, as associates of focus particles, etc.) receive both sloppy and strict interpretations in sentences with verb ellipsis receive both near-reflexive vs. pure-reflexive readings be used as adnominal intensifier of DPs be used as exclusive adverbial intensifier. Note that the simple reflexive sig in Danish has none of the properties in (7). In contrast, the intensified reflexive sig selv 'REFL self has (7i-iii) and the adnominal selv 'self on its own has the properties (7iv-v). In other words, both ziji and himselfbehave more like intensified reflexives and intensifiers than as unintensified reflexives. As the sentences in (8-12) show, assuming the existence of 0-reflexives (which may or may not be intensified) makes it possible to consistently analyze himself and ziji as adnominal intensifiers on a par with Danish selv 'self.' The sentences in (8-12) illustrate the properties in (7i-v). (8) Absence of focused unintensified 0-reflexives in Chinese (7i): Q: 5:&~"-:tl~71~18"J5fj5fja,~ Zhangsan tui-jian le ni-de di-di ma Zhangsan recommend PERF your brother ? 'Has Zhangsan recommended your brother?' A: 15I1'-r' ftP.FUlll7 §l c 0 Mei-you, tai zhi tui-jian-le lJ i ziji!*lJ; not-have he only recommended PERF self-self 'No, hei only recommended 0i himself/*0j .' (9) 10 Ziji can have both sloppy and strict readings in elliptical constructions (7ii): '* -· tb*JJ:9~ §l c¥~~s"JH 0 Zhangsan bi Lisi wei ziji bianhu de hao. Zhangsan than Lisi for self-self defend DE good 374 (i) 'Zhangsan defended himselfbetterthan Lisi defended himself.' (ii) 'Zhangsan defended himself better than Lisi defended him.' (=sloppy) (=strict) ( 1 0) 11 Ziji have both <statue> and <real> (i.e.both pure- and near-reflexive) readings (7iii): ~1~i-BIHtgEJc!f~~7 o Mao Ze-dongi ba zijii qiang-bi le. Mao Ze-dong BA self-self shoot PERF (i) 'Mao Ze-dong<real> shot himself<statue>.' (ii) 'Mao Ze-dong<real> shot himself<real>.' ( 11) Ziji can function as adnominal intensifier (7iv): (=Mao vandalized his statue) (=Mao committed suicide) ~'$ El c*7 ° (ziji = adnominal intensifier) Huang-di ziji lai-le. yellow-emperor self-self come-PERF 'The emperor himself came.' (12) Ziji as exclusive adverbial intensifier (7v): '* ~ ::f EJ c f'FJ}J~* 0 Zhangsan bu ziji zhuo gong-ke Zhangsan not self-self do homework (himself adnominal intensifier) 'Zhangsan doesn't do the homework himself.' (i.e. somebody else helps him) The Danish examples in ( 13-17) illustrate that in all these respects ziji has more in common with the Danish intensified reflexive sig selv (see ( 13-15)) and the adnominal 10 This example is adapted from Lidz (2001a:239, (25)). 11 This example is adapted from Lidz (2001a:239. (24)). 375 and adverbial intensifiers selv 'self, see (16) and (17), than with the unintensified reflexive sig. ( 13) Sig selv can be focused, sig cannot (7i): Q: Did Peter wash your brother? A: Nej, han vaskede *sig I sig selv. No, he washed REFL I REFL self 'No, he washed himself.' ( 14 ) Sig selv has both sloppy and strict readings, sig only has sloppy readings (7ii): a. Peter,forsvarede sig. Og det gjorde Hansk ogsa. Peter defended REFL and it did Hans too 'Peter defended himself and so did Hans.' (i) Hansk defended himselfk. (sloppy) *(ii) Hansk defended himi. (strict) b. Peter forsvarede sig selv. Og det gjorde Hans ogsa. Peter defended REFL self and it did Hans too 'Peter defended himself and so did Hans.' (i) Hansk defended himselfk. (ii) Hansk defended himj. (sloppy) (strict) ( 15) Sig selv has both near- and pure-reflexive readings, sig only has pure-reflexive readings (7iii): a. Peter barberede sig. Peter shaved REFL 'Peter shaved 0.' (i) 'Peter<real> shaved 0<real>.' *(ii) 'Peter<real> shaved 0<statue>.' b. Peter barberede sig selv. Peter shaved REFL self 'Peter shaved himself.' (i) 'Peter<real> shaved himself<real>.' (ii) 'Peter<real> shaved himself<statue>.' (16) Selv 'self can be used as adnominal intensifier, sig 'REFL' cannot (7iv): Peter selvl *sig kom til m0det. Peter self REFL came to meeting-the 'Peter himself attended the meeting.' ( 17) Selv 'self can be used as exclusive adverbial intensifier, sig 'REFL' cannot (7v): Peter fgste opgaven selv I *sig. Peter solved task-the self 'Peter solved the problem himself.' 376 We conclude that ziji should be treated as an intensifier (or intensified 0-reflexive) since it has five properties in common with the intensifier selv and the intensified reflexive sig selv 'REFL self' but only one property in common with sig (i.e. LD-binding). The analysis of Chinese outlined in (6), which yields a morphologically transparent system of adnominal 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. b. c. d. Chinese ziji is not a reflexive anaphor but rather an adnominal intensifier (e.g. huang-di ziji 'the emperor himself). Chinese has @-reflexives, e.g. Peter, xi-le lJ; 'Peter washed 0/, see ( 19) 12 • What looks like locally bound reflexives, e.g. Peter, hen ziji, 'Peter hates himself', is really locally bound adnominally intensified @-reflexives, e.g. Peter, hen lJ; ziji 'Peter hates 0; himself, see (20). What looks like locally free reflexives (also called "logophors") is really adnominally intensified pronominals, e.g.: Peter~}( Mary ~_&~ 7 [ @] El cJ-:J-:)l}pfTi=fB"JAJJhn o Peter, shuo Mary gen chu-le [taJ ziji yi-wai suo-you de ren tiao-wu P say M with except he self-selfiNST-outside that-have DE people dance 'Peteri said that Mary danced with everyone except [him;] himself.' The sentences in (19-24) illustrate the elements predicted to exist by the analysis sketched in ( 6) and ( 18): ( 19) testifies to the existence of @-reflexives (or at least phonetically null arguments which may receive a reflexive interpretation), (20) illustrates what is meant by intensified @-reflexives, (21 )-(22) illustrate unintensified and intensified object pronouns, and (23)-(24) illustrate unintensified and intensified subject pronouns. Adnominally intensified DPs have already been illustrated in (2). 12 Note that the absence of overtly realized objects in sentences of this type does not have to be construed as evidence that Chinese has 0-retlexives. 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-retlexives. Needless to say. more work is needed to determine the exact nature of such empty categories in Chinese. One possibility is that the 0-retlexive 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. (19) Unintensified 0-retlexives in Chinese: Q: {fl\1fijt1§15t#fe Ni you-mei-you xi-zao? you have not-have have wash-bath 'Have you bathed' A: 1§, ~t.=A~i5t-I 0 You, wo, yi-jing xi 0; le. have I already wash PERF 'I have, I have already washed 0.' (20) 13 Adnominally intensified 0-retlexives in Chinese Q: fjJ\:ffi5t#:ft~ftr\8"J~~ Ni you-mei-you tui-jian ni-de tong-xue? you have-not-have recommended your classmates 'Have you recommended your classmates?' A: 15t:ff, ~fUtt~7 §a o Mei-you, wo, zhi tui-jian-le O;ziji. not-have I only recommended PERF self-self 'No, I only recommended 0 myself.' (21) Unintensified/simple object pronouns in Chinese Q: ftr\.ft¥. '88"JB~fl*ff15t:ff;@-¥U~w? Ni qu huang-gong de shi-hou you-mei-you kan-dao huang-di,? you go emperor palace DE time have-not-have see-reach emperor 'Did you see the emperor when you went to the palace.' A: :ff, ~1§:-g¥U{tiJ. o You, wo you kan-dao Ia;. have I have see-reach him 'I have, I saw him (there).' (22) 14 Adnominally intensified/complex object pronouns in Chinese 377 13 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 wo;ziji. I already recommend PERF I self-self 'I already recommended myself.' 14 The answer in (22) sounds a bit awkward with the intensified pronoun ta 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 ;$::)\. ben-ren '(lit.) root-person, in person' instead of ziji would also yield a more felicitous sentence: (ii) fl};1fi"¥Uftfl;;ts:J\. o Wo you kan dao ta ben-ren. Q: {fri-*~'88'JB~fl9c:f_f;w¥U£1\1? ~JikR:f_f;w¥Uf-lli8'Jf~w1? Ni qu huang-gong de shi-hou you kan-dao huang-di, Hai-shi zhi you kan-dao ta-de shi-wei? you go emperor palace DE time have see-reach emperor or is only have see reach he DE serve guard 378 'Did you see the emperor when you went to the palace? Or did you merely see his guards?' A 15 : f~:f_f~¥Uf-lli~ a 0 Wo you kan-dao ta ziji;. I have see-reach him self-self 'I saw him himself (there).' (23) Unintensified/simple subject pronouns in Chinese '*=--mmil5tf-lli::f~tx;w~~ 0 Zhangsan, chang-chang shuo ta, bu xi-huan kan dian-yin g. Zhangsan often-often say he not like see movies 'Zhangsan often said that he didn't like movies.' PJ;@B'fl(f-tQ~;g "The Lord of the Rings" Ke-shi zuo-tian ta; qu kan '"The Lord of the Rings". but yesterday he go see "The Lord of the Rings" 'But yesterday he went to see "The Lord of the Rings"' I have see reach he root-person 'I saw him himself.' Needless to say an exhaustive account of adnominal intensification in Chinese should include discussion of other adnominal intensifiers than ziji, e.g. /$:A. ben-ren, 2js:Jlr ben-shen, etc. However. since such a study would largely exceed the scope of this dissertation, we leave that for further research. 15 The Chinese informants consulted for these sentences were speakers of Standard Taiwanese Mandarin Chinese. This explains why the verb you :fj '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 docs not directly bear on the behavior of reflexives and intensifiers. (i) ttw¥n1-t!2s2 ° Wo kan-dao leta ziji;. I see-reach PERF him self-self 'I saw him himseif(there).' (24) 16 Adnominally intensified/complex subject pronouns in Chinese ~:& ~ 'l%'1%ll3tfm/f~~~X~~~ 0 Zhangsan, chang-chang shuo ta, bu xi-huan kan dian-ying. Zhangsan often-often say he self-self not like see movies 'Zhangsan often said that he didn't like movies.' "BJ~A'F~{m~c:t;;g "The Lord ofthe Rings" Ke-shi zuo-tian fa; ziji qu kan "The Lord of the Rings ". but yesterday he self-self go see "The Lord of the Rings" 'But yesterday he himself went to see "The Lord ofthe Rings'" 379 In the following sections the hypotheses listed in (18) will be discussed in more detail as the present analysis of Chinese is compared to previous analyses. While it may still be too early to reach a definitive verdict, argue that the present analysis has the advantage of making it possible to explain the intensifier-behavior of ziji and ta ziji and thereby unify the analysis of adnominally intensified DPs (e.g. huang-di ziji 'the emperor himself) and uses of ziji which have hitherto been analyzed as reflexive anaphors and logophors. Other analyses of ziji appear to be incapable of achieving a unified account of all uses/properties of ziji. Furthermore, most existing accounts of Chinese (e.g. Huang and Liu 2001) suffer from the drawback that they need to assume 17 the existence of two, three, or even more different ziji's: (i) ziji 1 = adnominal intensifier (e.g. huang-di ziji 'the emperor himself), (ii) ziji 2 =locally bound reflexive anaphor subject to a syntactic principle A, and (iii) ziji 3 =locally free reflexive/logophor 18 • 16 To 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. (i) 'Jlt=m-H;~>t82:f'§l}:;g~~ c 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.' 17 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's 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 ([Ia] ziji). In other words, there is only one ziji which may intensify different nominal expressions. see sections 6.4-6.6 below. 380 In this section we have discussed data which support analyzing intensifiers and reflexives in Mandarin Chinese based on the assumptions in ( 1) and ( 18). The following sections will be dedicated to more detailed discussion of different aspects of this analysis. 6.3 Internal structure of ad nominal intensifiers in Chinese In this section the internal structure of adnominal intensifiers in Chinese will be discussed. In section 6.3.1, which focuses on the morphology of ziji, it will be argued that this element is bi- morphemic rather than mono-morphemic as is often claimed in the literature. Section 6.3.2. contains a discussion of whether Chinese has complex intensifiers of the English type, e.g. ta zji 'he/him self-self which is composed of a pronominal element (ta 'he/him') plus an intensifier element (ziji 'self-self). 6.3.1 Is ziji really mono-morphemic? In the literature on binding, ziji has often been claimed to be mono-morphemic. The distinction between simple/mono-morphemic reflexives and complex reflexives dates back to Faltz ( 1977) who observed that in many languages mono-morphemic reflexives allow LD- binding (i.e. clause-external binding) while complex/multi-morphemic reflexives have to be locally bound. Since ziji allows clause-external antecedents it is naturally to assume that it is mono-morphemic. While it is certainly true that ziji is morphologically simpler than ta ziji, this does not mean that it is mono-morphemic. Except for a few loan words Chinese lexical 18 To this list of different ziji' s most analyses would have to add one more. namely the so-called exclusive adverbial intensifier. see ( i). (i) ~.!F'"::f El c {'f-r}J~* o 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 of intensifiers. As discussed in chapter 2. the approach to intensification defended in this dissertation makes the unification of all uses of intensifiers seem less utopian. 381 morphemes are all monosyllabic. The fact that ziji is hi-syllabic is thus strong indication that it consists of two morphemes. The hi-morphemic structure of ziji is further confirmed by the fact that both the ~ zi- and the c -ji parts of ziji can be used independently, see (25-6). (25) (26) a. El~~ zi-qi self-cheat 'self-deceit' b. E=l ?Jt a. zi-ai self-love 'have regard for oneself; self-respect(ing)' ~ua zhiji know self (i) 'bosom/intimate friend', (ii) 'be intimate/close', (iii) 'know oneself, b. t'Jc±.~ liji zhu yi benefit self master thought 'egoism (lit. 'self-benefitism ')' Since both the !3 zi- and the c -ji parts of ziji can be used independently, ~ c ziji is obviously a hi-morphemic compound, composed of two different morphemes which both mean 'self 19 • Now the question is whether its hi-morphemic status affects its binding properties. 19 In Classical Chinese [3 zi and cji also both mean ·self but they had different syntactic functions. While 8 zi was used as pre-verbal reflexive clitic indicating coreference between two arguments of a predicate, see (i), c ji 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) rEH§t: wang zi sha king selfkill 'The king killed himself.' (ii) /f,\!tA;;;::_:fi35;D bu huan ren zhi buji zhi not worry people DE not self know '[I] do not worry that people do not know me.' Notice, that the typical word order of the Classical language (i.e. El zi being pre-verbal while c ji is post-verbal) still tends to be preserved in Modern Chinese compounds. 382 According to the Pica-style LF movement approach to LD-anaphora only reflexives which can undergo successive cyclic head-movement are able to be LD-bound. This is how the alleged difference between simple and complex reflexives in Danish, i.e. sig and sig selv, has often been accounted for (cf. Jakubowicz (1994) discussed in Chapter 3). Since sig is a mono-morphemic phonological clitic it automatically qualifies as a head. The complex reflexive sig selv, on the other hand, claimed to be unable to undergo head movement since it behaves syntactically as an XP. Similarly the difference between ziji (which can be both locally and LD bound) and ta ziji (which allegedly 20 has to be locally bound) has been explained as following from the morphological properties of these elements, i.e. the simple reflexive ziji being able to function as a head while the complex reflexive ta ziji is blocked from head movement due to its status as XP ( cf. Cole, Hermon and Sung 1991, and others). Notice that the question whether ziji is mono- or bi-morphemic 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 mono-morphemic. In other words, even though ziji is clearly bi- morphemic and is thus not a clitic it may still function syntactically as a head. In other words, the bi-morphemic status of ziji does not by itself suffice to exclude a Pica-style analysis ofthis element as a head reflexive able to undergo successive cyclic head movement. However, as shown in section 6.2, there is amble evidence elsewhere in the grammar showing that ziji behaves more like himself (i.e. complex reflexive/adnominal intensifier) than like the simple reflexive sig in Danish. 20 There is plenty of evidence suggesting that this generalization is wrong. Sentences with LD-bound ta ziji will be discussed below. 383 6.3.2 Ta ziji: complex intensifier, complex reflexive fJ ta ziji, or intensified pronoun ta ziji? It has been claimed that Chinese has two forms of the adnominal intensifier, viz. (i) ziji 'self- self (27a) and (ii) ta ziji 'he/him self (27b). (27) (28) a. ~1!1Elc%7o Huang-di ziji lai le. (ziji = adnominal intensifier) Emperor self-self come PERF 'The Emperor himself came.' b. £wf-t12.Elc%7o Huang-di ta z(ji lai le. (ta ziji = adnominal intensifier?) Emperor he self-self come PERF 'The Emperor himself came.' a. ~'r'fft&E1a 0 Huang-di hen ziji. (0 z(ji =intensified reflexive) emperor hate self-self 'The emperor hates himself.' b. ~wtEHlliEl a 0 Huang-di hen ta ziji. emperor hate he self-self 'The emperor hates himself.' (0 ta ziji = intensified reflexive) While the existence of complex intensifiers of the ta z(ji type in Mandarin Chinese, see (27b ), is still debated 21 , the existence of languages which have both simple and complex forms of intensifiers is beyond dispute. Some registers of French allow two different forms of the adnominal intensifier, the simple form meme 'same, self, see (29a), and the complex form lui- meme 'himself, see (29b), which is composed of the pronoun lui 'him'+ meme in the same way as Modern English himself was formed by adding the Old English simple adnominal intensifier self' self to the object pronoun him. (29) a. Le roi meme est venu. the king same is come 'The king himself came.' 21 People who do not recognize the existence of ta 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 ta intensified by ziji 'self-self. b. Le roi lui-meme est venu. the king him-same is come 'The king himself came.' 384 It is therefore not unrealistic to imagine that, during the transition period from Old English to Modern English, English also may also have had both a simple and a complex form of the intensifier. However, unlike French which appears to still be in such a transition period, Modern English now only allow the complex form of the intensifier, see (30a) vs. (30b). (30) a. b. *The king self came. The king himself came. One would expect both the simple and the complex forms of the adnominal intensifier in languages like French and Chinese to be able to intensify any kind of nominal expression. This is, however, not the case. There seems to be a strong tendency to avoid using the complex form of the intensifier to intensify object pronouns, see the Chinese examples in (3la,b). (31) a. c. Peter §ft Mary ~R~;F I Peter, shuo Mary gen chu-le ta; 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 Mary danced with everyone except [himi] himself.' Peter; shuo Mary gen chu-le ta ta; ziji yi-wai suo-you de ren tiao-wu Peter §}t Mary ~R~;F 7 Peter, shuo Mary gen chu-le ziji; yi-wai suo-you de ren tiao-wu P say M with except self-self INST -outside that-have DE people dance 'Peteri said that Mary danced with everyone except himselfi.' English displays a similar tendency to avoid overly redundant forms. That is repetitive forms like him himself are usually simplified to himself, see (32a-b ). (32) a. Peter, said that Mary danced with everyone except b. c. himselj; */'?'?h' h' If .. zm, zmse . *self 385 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 *!??me myself, */??him himselftend 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 such an explanation seems to run into trouble with forms like me myself and us ourselves 22 • In any case, some rule must be at play reducing him himselfto himself Whatever the rule is, the same phenomenon appears in Fr., compare (32a-c) and (33a-c). (33) a. b. Pierre bavardait avec lui-meme *lui lui-meme c. *meme Peter chatted with him-same/him him-same/same 'Peter chatted with himself.' In the case of Chinese, one can imagine several different accounts of why *ta ta ziji does not occur. One way would be to assume the existence of a morpho-phonological rule simplifying ta ta ziji to [ta] ta ziji in the same way as him himself is reduced to [him] himself in English. Another approach would be to assume that only the simple adnominal intensifier ziji is allowed to intensify the object pronoun ta. English and French illustrate that the absence of object pronouns intensified by complex intensifiers (e.g. *him himself, and *lui lui-meme) is not in itself enough to falsify the hypothesis that a language has complex intensifiers. That is, the absence of forms such as ta ta ziji does not by itself suffice to falsify the hypothesis that Chinese has complex intensifiers of theta ziji type. We thus conclude that the existence of complex intensifiers (Huang-di ta ziji 'the Emperor himself) in Modern Chinese is at least a theoretical possibility. 22 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 himself are ok in the right context. See the critique of Baker's prominence condition in chapter 2, section, 2.2.3.2.2. (56). 386 Critics of this analysis might claim that the sequence Huang-di ta ziji in (27b) should be analyzed as the DP Huang-di in topic position resumed by the intensified pronoun ta ziji in subject position. While such an analysis of (27b) is possible it is worth remembering that the topic position is characterized by being separated from the rest of the sentence by a brief intonational pause. Since (27b) can be pronounced without intonational pause we conclude that Huang-di ta ziji can be analyzed as a complex DP. Independent evidence for this comes from other complex post-nominal modifiers, e.g. yi-ge-ren 'alone' 23 , which are known to be able to form complex DPs, compare (34) and (35) (34) Adnominally intensified DP: DP I \ DP ta ziji I huang-di (35) DP modified by yi-ge-ren 'alone': DP I \ DP yi-ge-ren I huang-di In section 6.5.1 below we argue that locally bound ta ziji is in fact a 0-reflexive intensified by the complex intensifier ta ziji, i.e. 0 ta ziji, see (36). (36) Intensified DP: DP I \ DP ta ziji I 0 23 Cf. Siemund (2000) who also discusses the close similarities between adnominal intensifiers and other post nominal modifiers, e.g. Ger. a/ein 'alone' in Der Konig alein 'the King alone' vs. der Konig selbst 'the king himself. 387 6.4 Analyzing what appears to be uses of ziji as reflexive anaphors as adnominal intensifiers In this section we outline analyses of what appears to be (and is generally assumed to be) uses of ziji as locally (section 6.4.1) or LD-bound (section 6.4.2) simple reflexive anaphors as complex intensified 0-reflexives (i.e. 0 ziji). Section 6.4.3 concludes the section by summarizing the results. 6.4.1 Locally bound 0-reflexives in argument position: complex reflexives= a sub-type of ad nominal intensification In this section we investigate whether the distribution of locally bound unintensified 0- reflexives and intensified 0-reflexives in Mandarin Chinese can be accounted for within the framework assumed here. Is the distribution of locally bound unintensified 0-reflexives and intensified 0-reflexives (i.e. 0 ziji) determined by predicate meaning in the same way as the distribution of sig vs. sig selv in Danish and 0 vs. 0 himselfin English? Do we have evidence of a distinction between anti-reflexive, neutral, and inherently reflexive predicates in Chinese? 6.4.1.1 Anti-reflexive predicates According to the analysis proposed here, all semantically anti-reflexive (or "hidden" neutral) predicates should potentially require the intensified reflexive 0 ziji in object position 24 • As far as we can tell at this point (not having done an exhaustive survey), this prediction is borne out by facts. That is, the anti-reflexive verbs in (37) do seem to require the intensified form of the 0-reflexive, see (38a,b), (39a,b) and (40). 24 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. (37) a. b. c. d. e. (38) a. tN hen ~~ qi-pian §1~ ti-dai ili zhui etc. *'*_::.tN o Zhangsan hen 0;. Zhangsan hate 0 '*Zhangsan hates 0i.' 'hate' 'cheat' 'replace, substitute for' 'follow, court, pursue' b. 5K=J:&EJa o (39) a. Zhangsan hen 0, ziji. Zhangsan hate 0 self-self 'Zhangsan hates 0i himself.' *'*~W:,~~ 0 Zhangsan qi-pian 0i. Zhangsan cheat 0 '*Zhangsan cheats 0i.' b. '* ~ ttx,~ 13 a 0 ( 40) 25 a. Zhangsan qi-pian 0i ziji. Zhangsan cheat 0 self-self 'Zhangsan cheats 0i himself.' Jj\GAt:JJ~~effi5U~Wt¥7 o Xiaoming1 yi-wei lao-shi2you yao ze-guai 0 le. Xiaoming think teacher again will blame CSR 'Xiaoming1 thinks that the teacher2 will blame (me/you/him 1/3/her/himself/herself/us/you/them ... ) again.' b. !J\BAtJJ~~gffixw:itt£E:J a 7 ° Xiaoming 1 yi-wei lao-s hi you yao ze-guai 0-ziji le. Xiaoming think teacher again will blame self-selfCSR 'Xiaomingl thinks that the teacher2 will blame (himtlhimself2) again.' 388 The examples in (38) and (39) show that local unintensified 0-retlexives are ungrammatical with anti-reflexive predicates. The example in (40a) (which is from Huang (2000:86) furthermore illustrates that null objects in Chinese can receive a number of different 25 While finding it marginally acceptable, none of my informants liked the sentence in ( 40a), which Huang (2000:86) finds to be ok. 389 interpretations. However, in spite of the range of possible interpretations of the null object in ( 40a), it cannot (or only very marginally) be interpreted as a 0-reflexive bound by the local subject. Adnominally 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~ f~~ ba _qiangbi 'shoot b. f~ ~I±\~ ba _ diu-chu-qu 'throw out' (42) a. ~ ¥8¥5{ wei bianhu 'defend b. ~R §ft§li gen_ shuo hua 'speak with_' In this respect they behave like Danish prepositional predicates which also do not allow the simple unintensified reflexive. And similar to the case of Danish prepositional predicates, the absence of unintensified 0-reflexives with ba-constructions or co-verb + verb sequences may follow from the morpho-phonological properties of 0-reflexives (i.e. lack of stressability), rather than being due to the semantic properties of the predicate. 6.4.1.2 Neutral predicates As described in chapter 3-5 many of the neutral predicates in English and Danish are either grooming verbs, or verbs denoting movement 26 , see ( 43-44). (43) (44) Neutral grooming verbs: a. He, washes DP= b Han, vasker DPz I I I I Neutral verbs denoting non-translational movement: a. He, bends DP= I 0, I b Han, bojer DP= I sig, I 0, himself sig, selv. 0, himself sig, selv. 26 Some verbs denoting emotional states may also qualifY as neutral in Danish. e.g. bekymre ·worry', etc. See chapter 4 for more discussion of different types of neutral predicates. 390 One would therefore expect verbs of these semantic types to qualify as neutral predicates in Chinese as well. This, however as shown in ( 45)-( 46), appears not always to be the case. ( 45) Predicates denoting grooming activities: a. 17t#f~ *17t#~ §3 c b. c. xi-zaa xi-zaa ziji wash-bath wash-bath self-self 'wash (oneself), take a bath, bathe (oneself)' tJrEITJ[ *tfrEITJ[ §3 c shu-tau comb-head 'comb (oneself) 5iJ!lT- shu-tau ziji comb-head *17t#~! DP xi-zaa DP wash-bath DP *tJrEITJ[ DP shu-tau DP comb-head DP gua hu-zi gua hu-zi ziji gua hu-zi DP (46) cut beard-NOM cut beard-NOM cut beard-NOM 'shave (oneself).' Predicates denoting non-translational movement: a. ~T§~ *~T§~§3c b. c. wan xiayaa bend down waist 'bend down, duck' 1.t£9J[ di tau low head 'duck' 1$~%ft wan xia yaa ziji bend down waist self *1.t£ITJ[EJ c di tau ziji low head self-self *~~~DP wan xiayaa DP bend down waist DP *1.[£9~ DP di tau DP low head DP shen zhan shen ti shen zhan shen ti ziji shen zhan shen ti extend spread body body 'stretch oneself Unlike the typical grooming verbs in English and Danish allow reflexives in object position, the Chinese counterparts are often complex predicates composed of a verbal element denoting the action+ either a nominal element denoting an inalienable body-part, e.g. tau 'head' in shu- tau '(lit. comb head) comb' as in (45b) and (46a,b,c), or a cognate object as in (45a). Since these cognate or inalienably possessed objects saturate the position of internal argument there is no room for any more direct objects. This accounts for the judgments in ( 45-46); the starred sentences are ungrammatical because they contain two internal arguments where there is only room for one thematically 391 One way to contrast the activity of combing oneself with combing somebody else is to add an overt possessor to the otherwise inalienably possessed object, see ( 47). (47) a. 'J:&~t.t,: ~ !IL 91[ 0 Zhangsan shu 0 (de) tau. comb head 'Zhangsan combs 0.' b. (?)')fk ~,f:i< • ~ }IL ~a i'I"J 9§ 0 Zhangsan shu Zl)l de tau. comb-head self-self DE head 'Zhangsan combs 0 himself.' c. (?)'!.& ~ ,ffrt ~ i'I"J 91[ 0 Zhangsan shu wa de tau. comb DE head 'Zhangsan combs me.' Notice, however, that the sentences in (47b,c) are only marginally acceptable. The idiomatically correct way to say "Zhangsan combs me' would be by using a circumlocution with the verb m bang 'help' as in (48). (48) a. Zhangsan shu bang wa Zhangsan help me 'Zhangsan combs me.' ,ffrt9J[ 0 shu-tau. comb-head In the case of the verb+cognate object xi-zaa 'wash', the only way to make this verb transitive is to use the bang-construction illustrated in (48), see (49) 27 • (49) a. *'!.&=: 17t ~ i'I"J ~* 0 Zhangsan xi wa de zaa Zhangsan wash I DE bath b. '*=- m ~ 17t1l 0 Zhangsan shu bang wa xi-zaa. Zhangsan help me wash-bath 'Zhangsan bathes me.' Likewise most verbs denoting non-translational bodily movement also cannot be found (or are highly infelicitous) in the genitive construction illustrated in (47), see (50l 8 • 27 Notice, however, that strongly contrastive contexts can, in many cases, improve judgments. (50) a. *'*-· 1~ 1X 8~ 91[ 0 Zhangsan di WO de tau. Zhangsan lower I DE head b. ??'*~ ~""f 1X 8~ ~0 Zhangsan wan xia WO de yao. Zhangsan bend down I DE waist 'Zhangsan bends my waist down.' = 'Zhangsan bends me down.' Notice, however, that the bang-constructions are all anti-reflexive 29 , see (51). (51) Zhangsan shu bang 0/0 ziji I Lisi Zhangsan help 0 /0 self-self I Lisi 'Zhangsan bathes himself/Lisi.' xi-zao. wash-bath 392 Another way to achieve a reflexive reading of an empty object position is by omitting the cognate object, see (52A). (52) Unintensified 0-reflexives in Chinese: Q: 1fl\1'f1~:fi15t1t? Ni you-mei-you xi-zao? you have not-have have wash-bath 'Have you bathed' A: :fi, 1X8*~15tT o You, wo,yi:iing xi OJe. have I already wash PERF 'I have, I have already washed 0.' Finally, notice that the ability to have inalienably possessed objects is one way to test whether a verb is neutral or anti-reflexive. Just like in Danish, Chinese neutral verbs are compatible with inalienably possessed object while anti-reflexive and 'hidden" neutral verbs are noe 0 , see (53). 28 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 (SOb). 29 As anti-reflexive bang-constructions should in principle be able to take (/) 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 ba-constructions are distributionally anti reflexive because of the phonological properties of {/}-reflexives. {/}-reflexives cannot carry stress on their own and are thus unable to occur in PPs. In this respect, {/}-reflexives behave like the simple reflexive sig in Danish, see chapter 3 section 3.3.6 and chapter 5 section 5.2.1.5. 30 See chapter 3, section 3.3.2.6 for discussion of inalienable possession in Danish. (53) a. ilttfrE9J'[ 0 wa shu tau. I comb head (lit.) 'I comb my head'= 'I am combing (myself).' b. fltt~9J'[ 0 wa hen tau. I hate head *(i) 'I hate my head' ok(ii) 'I hate heads' I 'I hate the head' (i.e. somebody else's head) 393 We conclude that the absence of neutral predicates of the English/Danish type (e.g. Peter washes @/shaves 0/ducks 0/etc.) in Chinese (except for forced examples as (52A), which probably involve deleted objects rather than 0-reflexives) does not constitute a serious problem for the present proposal since it is the formal properties of predicate-formation in Chinese which are responsible for the absence of mono-syllabic neutral transitive verbal predicates 31 • That is, it appears to be the case that neutral predicates (i.e. grooming verbs, movement verbs, etc.) in Mandarin Chinese are characterized by the mandatory presence of either cognate objects (e.g. xi-zaa 'wash-bath') or inalienably possessed objects (e.g. shu-tau 'comb-head'). Hence the absence of mono-syllabic neutral verbs of the type xi 0 'to wash oneself can not be used as proof that Chinese does not have neutral predicates which take 0- reflexives. All we need to do is to classify predicates as anti-reflexive, neutral or inherently reflexive depending on whether they allow DPs, intensified 0-reflexives, and unintensified 0- reflexives in possessor position, see (54). (54) Predicate tyQe: Distribution: a. Anti-reflexive: DP /0 ziji b. Neutral: DP /0 ziji c. Inherently reflexive: *DP 1*0 ziji /*0 /0 /0 ExamQies: (55) (47) (58) The sentence in (55) illustrates the type of "anti-reflexive possessive predicate" referred to in (54). 31 It would thus be interesting to test whether hi-morphemic neutral verbs of the tui-jian 'recommend' type can take a-reflexives. (55) a. ~~ftu-[*[7]8'-Ji~~ 0 Zhangsan tou-le Lisi de qian-bao. Zhangsan steal-PERF Lisi DE money-wrap 'Zhangsan stole Lisi' s wallet. b. ~-~--flfrl 7 0 § as"J~~ 0 Zhangsan tou-le ziji de qian-bao. Zhangsan steal-PERF self-self DE money-wrap 'Zhangsan stole his own wallet. c. *~={1fr17 0i~~ 0 Zhangsan tou-le 0 qian-bao. Zhangsan steal-PERF money-wrap 'Zhangsan stole wallet. 6.4.1.3 Inherently reflexive predicates 394 In chapters 3, 4 and 5 we discussed inherently reflexive expressions in Danish and English, see (56-7). (56) a. Peter skynder sig I *sig selv I *Hans. Peter hurries REFL I REFL self I Hans 'Peter is in a hurry.' b. Peter tog en kniv med sig I *sig selv I *Hans. Peter took a knife with REFL I REFL self I Hans 'Peter took a knife with him.' c. Peter dukker sig I *sig selv I *Hans. (57) a. b. c. Peter ducks REFL I *REFL self I *Hans. 'Peter ducks 0 I *0 himself I *Hans.' Peter is out of his I *his own I *Hans's mind. Peter took a knife with him I *himself I *Hans. Peter ducks 0 I *0 himself I *Hans. Notice that the intensified forms of the reflexives (i.e. sig selvlsin egen and 0 himself/his own) are never allowed with inherently reflexive predicates. This (surprising) behavior of himself can be explained by assuming that it is an intensifier. Thus the absence of himself in those construction can be seen as following from the contrastiveness condition on adnominal intensification (see (62) in chapter 2, section 2.2.3.2.2) which states that intensifiers can only occur in positions where contrast with other entities is possible. Our analysis therefore predicts that Chinese should not have any predicates/constructions which are semantically 395 inherently reflexive while still allowing ziji. To the best of our knowledge this prediction seems to be true: Chinese does not (appear to) have any inherently reflexive expressions which allow ziji. The predicate{1t9J[ di-tau 'lower head/duck', discussed above (see (50)), may be a potential candidate for an inherently reflexive predicate, see (58). (58) a. if.&=; {It 0, R§o Zhangsan, di 0, tau. Zhangsan low head 'Zhangsan lowers his head/ducks.' b. *ffi= "-~-i {It 0, E3c i'!'j 9§ 0 Zhangsan; di 0; self-self de tau. Zhangsan lower self-self DE head c. *']:& -- i {It fltz i'!'j 91[ 0 Zhangsan, di Waz de tau. Zhangsan lower DE head 6.4.1.4 Predicate types: summary On the basis of the discussion in this section we conclude that Chinese data relating to the behavior of empty object positions and locally bound ziji do not empirically falsify the proposals in (I), and (18) on which the present framework is based. While there is a great deal of 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 @-reflexives is harder to come by. Some evidence of 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 more precisely what exactly Chinese @-reflexives are. This and other related questions will be taken up in the following sections. 396 6.4.2 LD-bound ziji: complex pronoun or complex reflexive? As described in Huang and Liu (2001) (among others) local ziji and LD ziji differ from each other in important ways. In their terminology, local ziji is a reflexive anaphor whose distribution is determined by the syntactic principles of the binding theory. LD ziji, on the other hand, is a logophoric element which has to be bound by some kind of attitude operator and whose distribution thus fall under both syntactic AND pragmatic principles. While we agree with their account of the facts, we disagree with their analysis which is based on the assumption that the lexicon contains two entries for ziji: (i) ziji 1 =a reflexive anaphor subject to principle A of the binding theory, and (ii) ziji 2 = a logophoric element bound by attitude operator and susceptible to pragmatic factors. Instead of proposing two 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 between the phonetically unrealized elements that ziji is intensifying in the two cases. In the case of "local ziji", the element ziji is intensifying a 0- reflexive (e.g. Zhangsan hen 0 ziji 'Zhangsan hates 0 himself), which falls under principle A ofthe binding theory. In contrast, we suggest that in the case of "LD ziji", ziji is intensii).ring a phonetically unrealized pronominal [ta] 32 (e.g. [ta] ziji '[him] himself(= Dan. ham selv)) which falls under principle B of the binding theory. See section 3.4 in chapter 3, and section 5.2.3 for discussion of intensified object pronouns in Danish and English. Intensified pronouns in Mandarin Chinese will also be taken up in section 6.6 below. 6.5 Intensifier analysis of local and LD ta ziji In this section, the consequences of extending the intensifier-based analysis proposed here to local (section 6.5.1) and LD ta ziji (section 6.5.2) will be discussed. 32 We follow the convention. introduced in chapter 5. to indicate phonetically unrealized material with square brackets. Instances of locally free himself in Modern English, are thus represented as [him] himself 397 6.5.1 Locally bound ta ziji: complex pronoun or complex reflexive? At first sight the fact that Chinese allow the complex form ta ziji to be locally bound may seem like a serious counterexample to the analysis proposed here, see (59b). (59) a. b. c. *Zhangsan, ai ta,. ta, ziji. 0i ziji. Zhangsan love him/him self-self/self-self 'Zhangsan loves himself.' The simple unintensified pronoun ta 'he/him' in (59a) behaves as expected, i.e. it cannot be locally bound without violating principle 8 33 • As shown in (60a,b ), intensification of pronouns in Danish does not affect their binding properties, i.e. they remain pronouns subject to principle B whether they are intensified or not. (60) a. b. Peter, vasker *ham,. *ham, selv. Peter washes him/him self The same applies to pronouns in Modern English, see (61a-b). (61) a. b. c. Peter, washes *him,. *him, himself himse(f Ifta ziji were an intensified pronoun ofthe same type as Danish ham selv 'him self then (59b) should be just 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 adnominally intensified pronoun, then what exactly is it? Traditionally it has been analyzed as a complex reflexive which due to its status as an XP has to be locally bound, cf. Cole, Hermon & Sung (199?) following Pica (1987). Since this solution implicitly assumes that adding ziji to the pronominal ta changes the binding properties of the latter it cannot be adopted in the present framework which is based on the independence of binding and intensification. One option 33 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. 398 consists in analyzing the occurrence of ta ziji in (59b) as a 0-retlexive adnominally intensified by the complex intensifier ta ziji, see (62a) and (63a). (62) a. 0-retlexive +complex intensifier: DP (e.g. (63a)) I \ DP ta ziji I 0 b. 0-retlexive +simple intensifier: DP (e.g. (63b)) I \ DP ziji I 0 (63) a. Zhangsan, ai 0i ta ziji. (see (62a)) Zhangsan loves REFL he self self 'Zhangsan loves himself.' b. Zhangsani ai 0, ziji. (see (62b)) Zhangsan loves REFL selfself 'Zhangsan loves himself In other words we need to assume that Modem Chinese has both simple and complex intensifiers which can intensify 0-retlexives, compare (62a) and (62b). As discussed in section 6.3.2 there is independent evidence supporting the existence of both simple and complex adnominal intensifiers in Mandarin Chinese. 6.5.2 LD-bound ta ziji: complex pronoun or complex reflexive? In spite of what has often been claimed in the literature, the complex form ta ziji, which is composed of the third person pronoun ta 'he/she/it' +the adnominal intensifier ziji 'self, can be long-distance bound, see the following example from Huang (2000:96(2.169a)) 34 , see (64) (64) ;j\B~§ft§§§-te:HtP.Ei clfT ~JJ!jt o Xiaoming] shuo lei-sheng ba ta ziji 1 xiao le yi tiao. Xiaoming say thunder-sound BA 3SG self-selffrighten PERF one jump 'Xiaoming said that the loud crash of thunder had given him a fright.' 34 See also Pan ( 1998) for more such examples. 399 Such examples has sometimes been construed as evidence that complex reflexives can be LD- bound, contrary to Faltz's ( 1985) generalization that LD reflexives are morphologically simplex. Notice, however, that one does not have to assume ta ziji '3SG self-self to be a complex reflexive in the first place. In the framework adopted here, it could also be considered an intensified pronoun on a par with Danish intensified pronouns like ham selv 'him self, etc. In others words, the apparent LD-binding of ta ziji follows from the pronominal character of ta '3SG'. See (65) which illustrates that the pronominal ta can be bound by the matrix subjectXiaoming. (65) ;j\BJj§}t§§§:t~{ili!ifT ~lf3t o Xiaoming 1 shuo leisheng ba ta 1 xiao le yi tiao. Xiaoming say thunder-sound BA 3SG frighten PERF one jump 'Xiaoming said that the loud crash ofthunder had given him a fright.' In summary, LD ta ziji is not a complex reflexive but rather a pronominal ta 'he/him' which has been intensified by adjunction of the adnominal intensifier ziji 'self-self, see (66b). (66) a. ta +complex intensifier: DP (e.g. (64)) I \ DP ta ziji I [ta] b. ta + simple intensifier: DP (e.g. (64)) I \ DP ziji I ta Just like with the pronominals in Danish (ham 'him) and English (discussed in chapter 3 and chapter 5) adnominal intensification (e.g. ham selv 'him self) does not alter their binding properties, i.e. they still fall under principle B requiring them to be locally free 35 • 35 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. 400 6.6 Intensified and unintensified pronouns in Mandarin Chinese In this section we investigate the behavior of intensified and unintensified pronouns m Mandarin Chinese. According to the assumption that binding and intensification are independent of each other, we predict that intensification of pronominals should not interfere with the binding properties of 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 following we test to what extent this analysis can be extended to Chinese. 6.6.1 Simple/unintensified pronouns in Mandarin Chinese: Principle B The sentences in (67-8) demonstrate the existence of principle B effects with third person personal pronouns in Chinese. (67) (68) a. b. a. b. Zhangsanihen Zhangsan hate 'Zhangsan hates ta., o· z!jii *o· him/self-self him/himself.' Zhangsan,jiao Lisik da ta, *k o· z!ji, k *o· Zhangsan ask Lisi hit him/self-self 'Zhangsan asks Lisi to hit him/himself.' The sentences in (67-8a) illustrate the fact that the pronominal ta 'he/him' must be locally free. Now the question is whether intensification of pronominals affect their binding properties. The sentences in (69) seem to suggest that it does. (69) a. b. Zhangsan, hen ta, • 0 ZIJI. 'Zhangsan hates him/himself.' Zhangsan,jiao Lish da ta, k *o z!ji. 'Zhangsan asks Lisi hit him/himself.' The complex form ta z!ji 'him self-self seem to follow the pattern of z!ji in (60-61 b). Traditionally this has been taken as evidence that ta z!ji is a complex reflexive which must be 401 locally bound. However, as discussed in sections 6.3 and 6.5.2 above several different analyses of ta ziji 'him self-self are possible, see (70). (70) a. Intensified pronoun: DP I \ DP ziji I ta b. Intensified 0-reflexive: DP (=(69a),( 69bk)) I \ DP ta ziji I 0 c. Intensified pronoun: DP (=(69b,)) I \ DP ta ziji I [ta] Notice furthermore, that the sentence m (70b,) illustrate LD-binding of ta ziji, see also discussion of (64-66). 6.6.1.1 Overlapping distribution of simple ta and ziji/ta ziji The distribution of simple ta vs. ziji/ta ziji overlaps in two contexts: (i) possessive construction (section 6.6.1.1.1 ), and (ii) embedded subject position (section 6.6. I .1.2). 6.6.1.1.1 Possessive constructions The complementarity of simple pronominais (e.g. ta f-tl1'helhim ')and zijilta ziji breaks down in possessive constructions, see (71-2). (71) a. b. c. Zhangsan; ai ta," de gou. ziji, *o ta ziji; *o (72) a. b. c. Zhangsan, shuo Lisik ai fa; k" de gou. ziji, ')k *o ta ziji, k *o 402 The existence of two different possessives in Chinese, i.e. ta and ziji, has sometimes led to Chinese being compared to Danish which distinguishes between pronominal and reflexive possessives, see hans 'his' vs. sin 'POSSREFL' (73). (73) a. Peter, siger at Hansk hader b. 'Peter says that Hans hates hans, *k" (egen) hund. his (own) sin., k *o (egen) POSSREFL (own) his/POSSREFL (own) dog:' Notice, however that, unlike Chinese, the complementarity between pronominals and reflexives is maintained in Danish regardless of whether the intensifier egen 'own' is present or not. We therefore conclude that the Chinese system is closer to the English (or French) where both his and his own (Fr. son vs. son propre) can be locally or LD-bound, just like Chinese ta and [ta] ziji/ta ziji. 6.6.1.1.2 Embedded subject pronominals In Mandarin Chinese the subject of embedded clauses can be realized in three different ways: (i) a a simple unintensified pronoun ta 'he' (74a), as an intensified pronoun [ta] ziji (74b), or as an a pronoun intensified by the complex intensifier [ta] ta ziji (74c). (74) a. Zhangsani shuo tai/o kanjian-le Lisi. b. ziji, *o c. ta ziji, *o Zhangsan say he/self-self/he self-self saw-Perf Lisi 'Zhangsan says that he/he himself saw Lisi.' In this section we explore to what extent the analysis of intensified Danish and English pronouns proposed in chapter 3 and 5 can be extended to embedded subject pronominals in Mandarin Chinese. 403 As shown in (75-76) simple unintensified pronouns behave alike (i.e. obeying principle B) in both Chinese and English. (75) Joan, believes that Joan, xiang-xin ta, z Joan; believes 'Joan; believes that she,= will win. hui ying. she;1z will win. she;;z will win.' Since the simple pronoun does not display any surprising behavior we will focus our attention on the two types of intensified subject pronouns in Chinese, i.e. [ta] ziji and [ta] ta ziji. The examples in (77-8) illustrate both [ta] ziji and [taj ta ziji behave like their English counterpart she herself in out-of-the-blue contexts. (77) Joan, believes that she, •z herse~f will win. (78) a. Joan ,ffl{§Ylli §a Wtli o Joan, xiang-xin [ta} ta ziji; huiying. b. *[ta} ta zijiz c. [ta] ziji, d. [ta} *ziji= Joan; believes she herself/herself will win. 'Joan; believes that she herself will win.' In his (1991) article, McKay showed that, given the right context, intensified pronouns can be bound by an extra-sentential antecedent, see (79) (79) Ann, wants to interview the winner. Joank believes that she, k herself will win. As shown in (80) the judgments of the direct translations of (79) differ from those of their English counterparts. (80) a. *Ann, xiang yaofang-wen de-jiang-zhe. Joank xiang-xin ta ziji, hui ying. b. Ann, xiang yao fang-wen de-jiang-zhe. Joank xiang-xin ta zijh hui ying. c. *Ann, xiang yao fang-wen de:jiang-zhe. Joank xiang-xin ziji, hui ying. d. Ann, xiang yao fang-wen de-jiang-zhe. Joank xiang-xin ZlJlk hui ying. Ann think want interview receive-prize-NOM. Joan believe self-self will win 'Ann; wants to interview the winner. Joan believes that she•;;k herself will win.' 404 It thus appears to be the case that [ta] ziji and [ta] ta ziji are more likely to be bound the closest subject than intensified pronouns in English. However, we do not want to rule out the possibility that Chinese intensified pronouns may be bound extra-sententially. As shown by (81-2) and (83-4) the intensified pronoun [ta] ta zij"i may indeed have an extra-sentential antecedent. (81) How will she herself do on the exam? (82) a. (83) (84) b. a. b. Ta zij"i kao-shi hui kao-de zen-me yang? (??I*)Zij"i 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." Mary, has been worrying about the grades of her friends. How will she, herself do on the exam? Mary, zui-jin hen dan-xin ta de peng-you de cheng-ji. Mary recently very worry she DE friend(s) DE grade(s) 'Mary has been worrying about the grades of her friends.' Ke-shi ta, zij"i hui kao-de zhen-me yang? *zij"i, but she self-self/self-self will exam DE how type 'But how will she herself do on the exam?' The asymmetry between the (a) and (b) examples in (81-4) show that subject [ta] zij"i is more restricted in its choice of antecedent than [ta] ta ziji. At this point we have no explanation of this difference. The contrast between the (85-6) and (87-8) show that, in both English and Chinese, the choice of context does make affect the ability of intensified pronouns to skip a closer non- subject antecedent in favor of a more distant matrix subject antecedent. (85) Susan, told everyone who knew Maryk that she, herse(f was pregnant. (86) a. b. c. d. e. f. Susan, gao-su suo-you ren-shi Maryk de ren Susan tell all know Mary DE people 'Susan told everyone who knew Mary that 405 ta, ziji huai-yun le. *takziji ??ta; tak ziji, * zijh pregnant LE __ 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, xiang yao bao-fu. Mary Maryk always lectured pregnant DE girls. Susani think want get-even 'Maryk always lectured at the pregnant girls. Sus ani did something to get even.' a. Susan, gao-su suo-you ren-shi Maryk de ren ta, ziji huai-yun le. b. takziji c. d. e. f. Susani told all know Maryk DE people that 'Susani told everyone who knew Maryk that ta, tak ziji, * zijh was pregnant. was pregnant.' The contrast between (86b) and (88b) is entirely due to the richer context of (88). This kind of data (which shows that intensification of subject pronouns (e.g. [ta] ta ziji) is subject to semantic/pragmatic factors) is evidence that intensification and binding of subject pronouns are independent in similar ways in Mandarin as in Danish and English. Note however, that ziji (which we analyze as an intensified pronoun [ta] ziji) behaves in an unexpected way. As mentioned above, at this point we have nothing to say about sentences such as (86f) and (88f) and simply leave the matter for further research. 6. 7 Conclusion In this chapter we have outlined an analysis of Mandarin Chinese based on the proposal that binding and intensification are mutually independent modules of the grammar. While a number of problems remains we hope to have showed that this approach is at least a viable 406 option which deserves to be explored in greater detail. Among the evidence adduced in its favor were the fact that ziji has more in common with adnominal intensifiers in other languages than with true reflexives, see ( 4-5) repeated here as (89-90). (89) Typological survey of reflexive and non-reflexive uses of reflexive elements: (i) Chinese (ii) English (iii) French (iv) German (v) Danish a. reflexive ziji himself se/soi sich srg b. reciprocal se sich c. middle se sich d. passive se sich e. deponent verbs se sich f intensifier ziji himself (lui-)meme selbst (90) Typological survey of different uses of adnominal intensifiers: Chinese English French a. Adnominal intensifier guo-wang ziji the king himself le roi (lui-)meme b. Complex reflexive 0 ziji 0 himself soi-meme/lui-meme c. Doppelganger-marker 0 ziji 0 himself soi-meme d. Intensified pronoun [ta] (ta) ziji [him] himself [lui] lui-meme e. Exclusive adv. inten. ziji himself lui-meme -s -s -slsig selv Danish kongen selv sig selv sig selv ham selv selv The main advantage of the present proposal is that it enables a unified account all uses of the intensifier element ziji 'self-self. In contrast, most existing analyses of Chinese are forced to assume ziji to be lexically ambiguous between (i) a simple reflexive anaphor, (ii) an adnominal intensifier, and (iii) a locally free logophor. 407 Chapter 7 The element selv 'self' in nominalizations 7.1 Introduction In this chapter the word-internal uses of the element selv 'self' will be discussed. It has been argued that selv 'self' functions as a reflexive in nominalizations such as selvrespekt 'self- respect'. That selv 'self' should function as a reflexive in nominalizations (rather than the simple reflexive sig 'REFL', e.g. *sig-respekt 'REFL-respect', has been taken as evidence in favor of analyzing selv 'self' as a reflexivizing element (cf. Reinhart and Reuland (1993)) and could thus be construed as an argument against the analysis defended here, namely that binding of reflexives, e.g. sig, and the distribution of intensifiers, e.g. selv 'self', are controlled by separate modules of the grammar. In this chapter these issues will be discussed and an alternative analysis will be proposed in which the element selv 'self' is consistently assumed to be an intensifier even in its word-internal uses. In the remainder of this introductory section the main facts concerning nominalizations containing the element selv 'self' will be introduced. Section 7.2 will contain more detailed description of the different functions of the element selv 'self' in nominalizations. Section 7.3 offers a cross-linguistic perspective on the use of reflexives and intensifiers in nominalizations and also takes a brief look at the diachronic aspects of word formation processes and lexicalization. Finally, in section 7.4, the exact status of word- internal uses of selv 'self' will be discussed (i.e. is it an intensifier or a reflexive, or something else?) and a possible analysis of selv 'self' in nominalizations will be outlined. For expository reasons an alphabetically ordered list of all the morphologically complex words containing the element selv 'self' occurring in one of the most comprehensive Danish-English dictionaries has been placed in Appendix I. 408 The element selv 'self occurs in a number of derived nouns, see (1 ). (1) a. selvrespekt 'self-respect' 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. Jorsvare 'to defend' => forsvar 'defense' The verbal meanings of the words in (2) are preserved in their respective nominalizations in ( 1 ). The element selv 'self occurring at the first part of these nominalizations could thus potentially be interpreted as receiving the theta-role associated with the direct object of the verbs in (2). In other words, the meanings of the nominalizations in ( 1) correspond roughly to the meanings of the corresponding reflexive clauses in (3). As also shown in (3), the external theta-role of the verbs forming the base of the nominalizations 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. himself.' 'Peter's se If-respect/ etc . .' At first glance, the close correspondence between clauses and nominalizations illustrated by the examples in (3) may appear to indicate that the element selv 'self functions as a reflexive element inside nominalizations in the same way that the reflexive pronoun sig (selv) functions as a reflexive element in full clauses. However, as will be proposed in section 7.4, one may also analyze the selv element in nominalizations to be an intensifier adjoining to a 0 reflexive. 409 7.2 Different uses of the morpheme selv 'self' in derived nouns and adjectives In this section, different uses of the morpheme selv 'self in derived nouns and adjectives will be described. Two main uses can be distinguished: (i) what appears to be "reflexives" uses, which we will argue are really adnominal 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 of idiomatic expressions. 7.2.1 Word-internal se/v 'self' appears to behave as a "reflexive", yet it really is an adnominal intensifier As mentioned in section 7.1, the close correspondence between clauses and nominalizations illustrated by the examples in (3) may at first glance appear to suggest that the element selv 'self functions as a reflexive element inside nominalizations in the same way that the reflexive pronoun sig (selv) functions as a reflexive element in full clauses. Almost all of the nominalizations in which the element selv 'self 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' 'se I f-o bservati on' 'self-criticism' 'to deceive' 'to observe' 'to criticize' Peter bedrager *sig/sig selv/Marie. Peter deceives REFLIREFL self/Marie 'Peter deceives himself/Marie.' Peter iagttager #siglsig selv!Marie. Peter observes REFLIREFL self/Marie 'Peter observes himself/Marie.' Peter kritiserer #siglsig selv!Marie. Peter criticizes REFLIREFL self/Marie 'Peter criticizes himself/Marie.' (anti-reflexive) (hidden neutral) (hidden neutral) (anti-reflexive) (hidden neutral) (hidden neutral) 410 The nominalizations in ( 4) are based on the verbs in (5) whose anti-reflexive or hidden neutral nature is illustrated by the examples in (6). The anti-reflexive meaning of these predicates generates a presupposition of representational non-identity of their arguments. Therefore adnominal intensification of the simple reflexive is necessary to mark the reflexive as representationally non-identical to its antecedene. In other words, though the instances of selv 'self' in ( 4) appear to function as reflexives, they are really adnominal intensifiers intensifying a word-internal 0-reflexive (see the analysis proposed in section 7.4 below). There are, however, also a few of examples of such nominalizations derived from neutral verbs, see (7)-(9). (7) a. b. selvforsvar selvforsorgende 'se If-defense' 'self-supporting, independent' (8) a. b. forsvare fors(1rge 'defend' 'support' (9) Peter forsvarer/forsorger sig/sig selv/Marie. Peter defends/supports REFLIREFL self/Marie 'Peter defends/supports himself/Marie.' (neutral, transitive verb) (neutral, transitive verb) The nominalization in (7a) is based on the neutral verbs in (8) whose neutral nature is illustrated by the examples in (9). As expected there are no self-nominalizations based on "grooming" verbs, e.g. *selvvaskning 'self-washing' 2 • 1 See analysis of intensified reflexives in chapter 3. 2 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 af sig selv 'by itself/automatically', see (i), rather than from a reflexive pronoun, see (ii). (i) Ovnen kan rense sig af sig selv. oven-the can clean-INF REFL ofREFL self 'The oven can clean itself automatically/by itself.' (ii) Ovnen rensede sig (selv). oven-the clean-PAST REFL self 'The oven cleaned itself.' 411 There are, however, somewhat unexpectedly, a couple of nominalizations based on inherently reflexive verbs, in which the element selv 'self seems to play the role of reflexive, see ( 1 0)-( 12). (10) a. b. selvbesindelse selvudvikling 'collectedness, self-communion; self-reflection' 'self-development; spontaneous evolution (medical terminology)' ( 11) a. b. besinde sig udvikle sig 'collect oneself 'develop' (12) Peter besinder sig/*sig selv/*Marie. Peter collects REFLIREFL self/Marie 'Peter collects himself/Marie.' (inherently reflexive) (inherently reflexive/transitive) The nominalizations in (10) are based on the inherently reflexive verbs in (11) whose inherently reflexive nature is illustrated by the examples in (12). The fact that an inherently reflexive verb like besinde sig 'collect oneself is able to form nominalizations with selv- is rather surprising. As discussed in chapter 4, there is ample evidence in favor of assuming the sig occurring with (most) inherently reflexive verbs to be a non-thematic grammatical marker rather than a theta-role receiving nominal argument. That is, besinde sig 'collect oneself is assumed to have been reanalyzed as an intransitive verb unable to assign more 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 nominalizations based on these verbs, the element selv- 3 receives the internal theta-role, see (13) and (14). (13) a. Peters kritik af sig selv. 'Peter's criticism ofREFL self b. Peters kritik af Marie. 'Peter's criticism ofMarie.' (14) a. *Peters selvkritik af sig selv. 'Peter's self-criticism ofREFL self b. *Peters selvkritik af Marie. 'Peter's self-criticism ofMarie.' 3 Alternatively, as proposed in section 7.4, it is a @-reflexive to which the intensifier selv is adjoined which receives the theta-role from the verbal stem. c. ok Peters selvkritik Peter's self-criticism 412 In (13a,b) both theta-roles of the deverbal noun kritik 'criticism' are assigned: one to the possessor phrase Peters 'Peter's' and the other to the nominal expression in the prepositional complement af _ 'of_'. Likewise, in (14c) both theta-roles of the deverbal noun kritik 'criticism' are assigned: one the possessor phrase Peters 'Peter's' and the other to the element selv 'self. The examples in ( 14a,b) can be construed as evidence that the internal theta-role of the verbal stem kritik- in the nominalization selvkritik 'self-criticism' has been assigned to the element selv-. That is, the ungrammaticality of (14a,b) can be explained as a consequence of the DPs in the prepositional complement af _'of_' in (14a,b) receiving no theta-role, since the two theta-roles of the deverbal noun kritik 'criticism' have already been assigned to other elements. However, if it is assumed that the element selv- in nominalizations always receives a theta-role from the verbal root it is prefixed to, then how do we explain the presence of selv in nominalizations based on inherently reflexive verbs (see (IOa,b))? As discussed below, there are even cases of selv-compounds based on intransitive verbs. But in those cases, e.g. selvdod 'self-dead (used to describe animals which have died of disease or other natural causes rather than having been killed)', the element selv is interpreted as an adverbial phrase meaning 'by itself, automatically, spontaneously, etc' rather than as an argument. Note that it is possible to adopt a similar adverbial analysis of selv in the compound selvbesindelse 'collectedness, self- communion; self-reflection' (I Oa). By consistently analyzing self-nominalizations based on inherently reflexive verbs (of which there is no more than a handful out of the total of 235 compounds listed in Appendix I) as adverbial self-nominalizations we arrive at the generalization that reflexive self-nominalizations only can be formed on the basis of anti- reflexive and neutral verbal stems, see list of self-nominalization in Appendix I. The term 413 'reflexive self-nominalizations' refers to deverbal nominalizations in which selv appear to function as the internal argument (but, as we argue, really is an adnominal intensifier). In contrast, the term 'adverbial self-nominalizations' refers to deverbal nominalizations in which selv functions as an adverbial. 7.2.2 Word-internal selv 'self' behaving as an adverbial Some transitive verbs are not, or only marginally, compatible with reflexive scenarios. In nominalizations of such verbs the selv element tends to take on an adverbial meaning 'on one's own, by oneself, without help, all alone, etc.', see (15). In other words, the occurrence of the element selv in the complex noun in ( 15) should not be interpreted as the direct object of the verbal element bygge 'build' but rather as the manner adverbial selv 'by himself, etc." occurring sentence-finally in the example in ( 16). (15) selvbygger self-build-er 'person who builds his house with his own hands.' (16) Peter byggede huset (helt) selv. ('exclusive' adverbial use of selv 'self) Peter built house-the all self 'Peter built the house (all by) himself' Other examples of adverbially used word-internal instances of the element selv are listed in ( 17) which contain adjectivizations/nominalizations based on the intransitive verbs listed in ( 18). (17) a. selvdrJd self-dead (Adjective used to describe animals which have died of disease, old age or other natural causes rather than having been killed) b. selvklatrende self-climbing '(bot.) climbing, creeping, trailing (plant)' ( 18) a. b. drJ klatre intr. 'die' intr. climb, clamber 414 The adjectivizations/nominalizations in (19) are based on the transitive verbs listed in (20). But unlike (15), the element selv here means 'by itself, automatically' rather than 'all by itself, without help.' (19) a. (20) b. c. d. a. b. c. d. selvspillende self-playing (adj. used to describe a piano which can play by itself/automatically) selvbinder self-bind-er (adj. used to describe hay-raking machine which binds the hay into bundles on its own/automatically) selvha:vende self-rising 'self-rising (fx flour)' selvlukkende self-closing 'self-closing (fx door)' !lpille (et instrument) binde (ha) ha:ve lukke 'play (an instrument' 'bind (hay)' (i) tr. 'raise (fx one's arms; one's glass); lift, lift up' (ii) intr. 'swell, tumefy; (dough) rise' 'close (a door/gate/window/etc.)' The main difference between (16), which allows exclusive adverbial selv, and (21a), which doesn't, lies in the nature of the predicate. Bygge 'build' in (16) is a causative transitive predicate. In this case the exclusive selv can associate with the agentive subject Peter. (21) a. b. *Dejen ha:vede (helt) selv. (Compare (21a) and (16)) dough-the rose all self 'The dough rose all by itself.' Dejen ha:vede (helt) af sig selv. dough-the rose all ofREFL self 'The dough rose all by itself Since the verb ha:ve 'rise' in (21) is unaccusative, it does not assign any AGENT theta-role. Hence exclusive adverbial selv can not be used and the phrase af sig selv 'of REFL self is found instead. 7.2.3 Problematic cases It is unclear how to characterize the exact function/meaning of selv 'self in the more or less idiomatic adjectival/nominal compounds in (22)-(26). 415 (22) a. selvklog adj. 'self-opinionated; conceited' b. klog adj. wise (fx a wise 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. selvmal n. '(sports) own goal' b. mal n. goal (fx in football) (25) a. selvsamme pron./ adj. 'self-same; the very same' b. sam me adj. same (26) a. selvtredje adv. 'with the two others' b. tredje num. 'third' Most of these 'problematic' cases are usually NOT deverbal nominalizations but rather compounds based on non-verbal nominal or adjectival roots which do not take arguments, or express actions (hence cannot take either adnominal, or exclusive adverbial selv). The adjectives klog 'wise, intelligent' and retfcerdig 'just, righteous' in (22-23) do not allow internal arguments. It is thus impossible to suggest that in the self-compounds in (22a) and (23a,b) the element selv functions as an internal argument. The self-compounds in (24-26) seem even more idiosyncratic. In at least one case the adnominal intensifier selv and its suppletive form in possessive constructions, egen own', can occur in the same compound without changing the meaning, see (27). (27) egenrisiko own-risk 'own risk' selvrisiko self-risk 'own risk' (in insurance terminology) The words in (27) are interesting since they may potentially be used as evidence that selv 'self' and egen 'own' are suppletive variants ofthe same element, an assumption which is crucial to the analysis defended in this dissertation, see section 3.3.5 in chapter 3 for more discussion of the possessive intensifier egen 'own'. 416 Another set of idiomatic uses of selv is given in (28), where it may be possible to argue that the selv in the compound in (28a) corresponds to the complex reflexive in the idiomatic phrase in (28c) though the exact analysis of both (28a) and (28c) remains to be worked out. (28) a. b. c. selvsagt self-said adj. 'self-evident, obvious, plain' sige vt. say det siger sig selv it says REFL self 'It is obvious/it goes without saying/etc.' 7.3 Cross-linguistic perspective on use of intensifiers in derived words In this section it will be argued that while word-externally (syntactically) used adnominal intensifiers have the same basic properties in most languages (i.e. as markers of representational non-identity, markers of contrastive focus, etc., see (29), word-internally used intensifiers seem to display a great deal more cross-linguistic variation which are mainly due to the effects of lexicalization and blocking. (29) Typological survey of different uses of adnominal intensifiers 4 : a. Adnom. inten. b. Complex retl c. Prosodic inten. d. Doppelganger-mark e. lnten. pron./logophor f Exclus.adv. int g. lnclus. adv. int h. Add. focus part j. Nominal uses k. Deverbal compounds I. Positive adjective m. Superlative adjective 7.3.1. French Danish English French kongen selv the king himself le roi (lui-)meme sig selv [0 himself] soi-meme/lui-meme sig selv [HIMSELF/0 himself] [SOl/ soi-memel sig selv [0 himself] [soi-meme] ham selv [(him) himself] [(lui) lui-meme] selv himself lui-meme selv himself [lui-meme] selv selv/jeg selv selve selveste even self!*! *himself/self very /actual [ me me *meme/moi *meme-/auto-/sui-/etc .. meme] 'same (kind of)' Chinse guo-wang (ta) ziji [ta ziji/0 ziji] [0ziji] [0ziji] [(ta) ziji] ziji ye lian_ye *zi.Ji/zi-1-ji The nominal intensifier used to mark non-guise-identity in French is -meme, see (30). 4 See chapter 2 for examples illustrating the different uses of intensifiers referred to in this table. (30) Use of -meme to mark non-guise-identity. a. Jl travaille contre soi*(-meme). b. Jl travaille pour soi(-meme). (anti-reflexive predicate) (neutral predicate) 417 Even though the adnominal intensifier -meme is used to mark non-guise-identity in the same way is the adnominal intensifier selv in Danish the Fr. morpheme is not used as the first part of any derived nouns and adjectives. It apparently only occurs in one compound, i.e. the adverb memement 'likewise; especially'. This clearly shows that the morpheme used as adnominal 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 Greek reflexive/intensifier autos 'self) is more productive than it is in the Germanic languages and thus is able to take over some of the functions of -meme in derived nouns and adjectives, see (31 ). (31) Nominalizations in auto- Danish: French: a. selvanklager self-accuser auto-accusateur *meme-accusateur auto-accuser b. selvforsvar self-defense c. selvportrcet self-portrait d. selvplageri self-torture auto-defense auto-defense auto-portrait auto-portrait auto-punition auto-punishment *meme-defencse *meme-portrait *meme-punition The examples in (32) and (33) illustrate that Germanic self-nominalizations often correspond to French nominalizations in which the internal argument is relegated to a prepositional complement. (32) Paraphrases with PP + soi-meme Danish: French: a. selvbebrejdelse grief contre soi-meme b. selvbegrcensning limitation de soi-meme c. selvkritik critique de soi-meme/autocritique (33) Paraphrases with PP + soi(-meme) Danish: French: a. selvbehag contentement de soi, ??complaisance de soi auto-complaisance b. selvtilfreds content de soi[-meme] c. selvforsvar defense de soi-meme/autodefense 418 The word-internal distribution of soi and soi-meme in nominalizations of the types illustrated here seem to neatly follow the word-external distribution of soi and soi-meme and of lui and lui-meme with non-contrastive prepositional predicates. That is, anti-reflexive predicates require the complex forms lui-meme/ soi-meme, see (30a) and (32), while neutral predicates allow both simple soi/lui and complex lui-meme/ soi-meme, see (30b) and (33). Finally, notice that in neither French nor Danish is it possible to form compounds with the reflexives se and sig, e.g. *se-punition, *sig-plageri. Indeed, it seems to be impossible to find 'true' reflexives used in compounds in this way in any language. The absence of nominalizations with incorporated reflexive pronouns may be due to a constraint on what kind of elements may undergo incorporation. That is, being DPs, reflexive pronouns cannot incorporate, while the morpheme selv being anN is allowed to incorporate. 7.3.2 Chinese In Modem Mandarin Chinese the reflexive pronoun and the intensifier appear to have the same form, i.e. ziji I§ c 'self', see (34). (34) a. '*=-t&E!c 0 Zhangsan hen ziji. Zhangsan hates self 'Zhangsan hates himself.' b. 5.& -- E~ at~r~JJ:9 ° Zhangsan ziji hen Lisi. Zhangsan selfhates Lisi 'Zhangsan himself hates Lisi.' 419 Like French -meme, ziji Elc 'self does not seem to be used in derived nouns and adjectives. Instead, the morphemes zi El 'self andji c 'self, which both have the basic meaning 'self, are used independent!/, see (35) and (37). In Classical Chinese both zi El 'self and ji c 'self could be used as independent words. The word zi El 'self was mostly used as a verbal prefix, and was therefore usually only used as local reflexive. Ji c, on the other hand, could be used both as local and as long-distance reflexive. In Classical Chinese the noun shen Jif 'body' was also used as emphatic reflexive/adnominal intensifier 6 • All three words now occur in the Modern Chinese counterparts of English/Danish words containing the element selv 'self, see (35)-(37). (35) zi El 'self + verb 7 8 a. zi-bei § l2f. b. zi-qi [] !t,X c. zi-ai §~ d. zi-wei E31¥r (36) verb + shen :5} 'self, body' 'self-abasement' 'self-deception' 'self-love' 'self-defense' (see (49a-b)) a. hu-shen lll~ 'self-defense' b. fang-shen i37]~ 'self-defense' (37) verb+ ji c 'self a. zhi-ji ~De 'know oneself b. zhi-ji ~De n. 'bosom/intimate friend, be intimate/close' 5 Note incidentally. that ziji § c ·self-self is often wrongly assumed to be mono-morphemic, see chapter 6 for more discussion of the intensifier ziji § c ·self·self. 6 As in wu shen ]§=~(lit.) "I body'= '1 myself. 7 zi §'self +verb is a frequently found type of self-compound in Modem Chinese. The prefix zi §'self is only found in a few. idiomatic, constructions which are not based on deverbal nouns, e.g. zi-wo §Ilt (lit. 'self-me') 'self: oneself: ego'. 8 Note that verbal compounds containing zi§ 'self abound in Chinese (e.g. zi-sha§ *li '(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 I. 420 The formation of many of the nominalizations now found in modem Chinese, e.g. (35-37), occurred long time ago at a time when ziji § c as reflexive/intensifier had not yet been formed. It could therefore be argued that the existence of such nominalizations in the lexicon is now blocking the productive formation of compounds containing ziji § c, e.g. *ziji-ai * § c'£ 'self-love', compare with (35c). 7.3.3 Summary The above mini typological survey of the behavior of the word-internal uses of intensifiers shows that the elements occurring (word-externally) as reflexives/intensifiers in clausal constructions need not be identical to the ones occurring in same functions word-internally. Furthermore, even if the same elements are used there is no guaranty that they will behave the same way when used word-intemaJJy as when used in clausal constructions. The fact, that the element selv 'self' 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 'selr 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 same way, i.e. either as an adnominal intensifiers adjoining to a simple reflexive or as an adverbial phrase modifying a predicate. The distribution of the word-external adnominal intensifier with difference verb types is given in (38). (38) a. Neutral predicates: sig I sig selv I DP b. Anti-reflexive: *sig I sig selv I DP c. Inherently reflexive: sig I *sig selv I DP 421 (39) Neutral predicates: a. Peter forsvarer sig selv. b. Peter forsvarer sig. 'Peter defends REFL I REFL self' As stated in (38a) and illustrated in (39a,b) se/v-adjunction to the simple reflexive sig is optional with neutral predicates like forsvare 'defend'. Likewise in 'reflexive' nominalizations based on neutral predicates the presences of the element selv is optional, see (40a,b). ( 40) Neutral predicates: a. [ [0 selv] [forsvar ]] b. [ [0] [forsvar ]] 'self-defense' 'defense (of oneself unless otherwise specified)' As shown in ( 40), we assume nominalizations of neutral verbs like forsvar 'defense' and selvforsvar 'self-defense' to contain a 0 reflexive which may or may not be intensified by the adnominal intensifier selv. Even when the intensifier selv is not present the nominalization still has a reflexive reading (see (4l(i)) below) as its default reading. Only when a non- reflexive argument is overtly present (e.g. His defense of Mary) is a non-reflexive reading possible. ( 41) The USC football team had a problem with their defense. (i) their defense (of themselves) Now let's turn to the anti-reflexive predicates. As stated in (38b) and illustrated in ( 42a,b) se/v-adjunction to the simple reflexive sig is mandatory with anti-reflexive predicates like kritisere 'criticize'. ( 42) Anti-reflexive predicates: a. Peter kritiserer sig selv. b. Peter kritiserer *sig. 'Peter criticizes REFL I REFL self.' Likewise in 'reflexive' nominalizations based on anti-reflexive predicates the presence of the element selv is mandatory, see ( 43a,b ). 422 (43) Anti-reflexive: a. [ [0 selv] [ kritik]] 'self-criticism' b. * [ [0] [ kritik]] 'criticism (of somebody other than oneself unless otherwise specified)' As shown in (43), in nominalizations based on anti-reflexive verbs the 0 reflexive has to be intensified to get the reflexive reading. When the intensifier selv is not present the nominalization cannot have a reflexive reading, see ( 44(i)). (44) a. The USC football team had a problem with their criticism. *(i) their criticism of themselves ok(ii) their criticism of someone else/other teams b. The USC football team had a problem with their selfcriticism. 0 \i) their criticism of themselves * (ii) their criticism of someone else/other teams That is, the default reading of nominalizations of anti-reflexive verbs is a non-reflexive reading, see ( 43b ), and to obtain the reflexive reading the intensifier must be adjoined to the word-internal 0 reflexive. The fact that a few (only one or two) nominalizations of inherently reflexive verbs allow the optional presence of 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 af sig self collectedness of REFL self >> >> selvbesindelse 'collectedness' As indicated in (38c) inherently predicates do not allow the complex reflexive. Consequently one would expect the nominalizations of such verbs to exclude the presence of the adnominal intensifier selv. A possible way out might be to assume the selv in ( 45a) to be an adverbial modifier rather than the adnominal intensifier adjoined to a 0 reflexive, see ( 45c ). 423 As already mentioned in section 7.3.1 above, in neither French nor Danish is it possible to form compounds with the reflexives se and sig, e.g. *se-punition, *sig-plageri. Indeed, it seems to be impossible to find 'true' reflexives used in compounds in this way in any language, see ( 46). ( 46) Generalization: True reflexives (e.g. Dan. sig, Fr. se, Ger. sich, etc.) do not occur in nominalizations, e.g. Fr. *se-punition, Dan. *sig-plageri Since reflexives cannot be used in compounds, it has been suggested that the element selv functions as a reflexive word-internally. In this section, we have outlined an alternative analysis based on the assumption that word-internal reflexives are phonologically unrealized elements that may or may not be intensified by the adnominal intensifier selv. As suggested above, this analysis makes it possible to unify the account of both word-external and word-internal occurrences of the adnominal intensifier selv. In addition to its use as an adnominal intensifier, the element selv can also be used as an adverbial modifier. Again, the word-internal uses mirror the word-external uses, see (38) vs. (39), and (4lc) vs. (43b). Chapter 8 Conclusion 424 In this dissertation we have presented an account of simple and complex reflexives and pronouns based on the proposal that binding and intensification should be clearly separated. The two modules overlap in the case of complex reflexives and pronouns which are here analyzed as adnominally intensified forms of their simple counterparts. That is, we proposed that the descriptive generalizations in (1 )-( 4) are best accounted for within a framework which separates binding and intensification in to the independent modules summarized in (5) and (6) respectively. (1) Simple reflexives (e.g. Dan. sig): are bound by subject in local domain (=minimal tensed clause) (chapter 3, section 3.2.4) (2) Simple pronouns (e.g. Dan. ham): are subject-free in local domain (=minimal tensed clause) (chapter 3, section 3.4.1) (3) Complex reflexives (e.g sig 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) 425 (iii) strict (and sloppy) readings in VP ellipsis (section 3.3.2.7.3) e. stress-carrier (i) after unstressed prepositions (ex. ( 144), section 3.3.6.1) (ii) meta-linguistic contexts (with inherently reflexive) (chapter 4, section 4.3.3, ex. (91)) ( 4) Complex 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 (with neutral predicates) (section, 3.4.2, ex. (169) (5) Binding theory (see chapter 3): a. Principle A, see (chap. 3, sect. 3.2.4, (25-28)) b. Principle B, (chap. 3, sect. 3.2.4, (25-28)) (6) Theory of intensification (see chapter 2): Contrastiveness condition on adnominal intensification: A nominal expression (DP, pronoun, reflexive, etc.) can only be intensified adnominally if it can be contrasted with other expressions in the context in which it is found. (chapter 2, ex. (2/62)) The syntactic binding principles in (Sa) and (5b) account for the distribution of reflexives and pronouns in (1,3) and (2,4) respectively. The pragmatic/semantic principle of contrastiveness in (6a) is the basis of the account of the distribution of the intensifier selv in the contexts listed in (3) and (4). 426 The approach to binding and intensification proposed here has a number of important consequences for linguistic theory in general, see (7), and for the analysis of Danish in particular, see (8). (7) Consequences for binding theory: a. intensification and binding are independent modules of the grammar b. binding is a purely syntactic phenomenon c. unified account of possessive and argument reflexives d. doppelganger-effects (statue-readings, qua-sentences, stict reading in VP ellipsis, etc.) and anti-reflexivity (the triggering of adnominal intensification of reflexives with certain predicates as well as in other semantically/pragmatically anti-reflexive contexts (e.g. resultatives, ECM, and possessive constructions)) can be unified. (8) Consequences for analysis of Modem Danish (and similar languages): a. sig not anti-local b. sig selv is not mandatorily local (cf. his own). c. complex 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 of anaphor ( cf. Bickerton 1986) but an intensified subject pronoun. f. selv 'self is not a reflexivizing particle which falls under binding theory (e.g. Reinhart and Reuland 1993), but an intensifier which falls under the module of intensification. 427 g. egen 'own' is the suppletive variant of selv 'self. h. the possessive reflexive sin 'POSSREFL' is a morphological variant of sig 'REFL' and falls under the same principle A of the binding theory (in contrast, Reinhart and Reuland ( 1993) have to assume sin to be fundamentally different from sig, since in their analysis reflexives have to be arguments of predicates), In chapters 5 and 6 similar approaches to English and Mandarin Chinese were put to the test, see (9)-( 11 ). (9) Consequences for analysis of Modern English: a. English has 0-reflexives (;::;Dan. sig) b. English x-selfforms are not reflexives but ALWAYS intensifiers c. English have both simple and complex reflexives, i.e. 0 vs. 0 himself d. The distribution of simple (i.e. 0) and complex reflexives (i.e. 0 himself) in Modern English mirrors that of simple (i.e. sig) and complex reflexives (i.e. sig selv) in Danish, see (3) d. Locally free himselfis not a logophor but an intensified object pronoun e. he himself is not a special kind of anaphor ( cf. Bickerton 1986) but an intensified subject pronoun ( cf. McKay 1991) f. -self is not a reflexivizing particle which falls under binding theory (e.g. Reinhart and Reuland 1993), but part of complex intensifiers, i.e. the x-self forms, which falls under the module of intensification g. own is the suppletive variant of the adnominal intensifier himself 428 ( 1 0) Consequences for the analysis of the evolution of English reflexives and intensifiers: a. Old English (OE) locally bound pronouns (e.g. he, washes him,) were replaced by 0-reflexives in Early Modern English (EME) (e.g. he, washes 0). b. The fact that the loss of locally bound pronouns took place across the board explains the loss of the pronoun in OE he rested him which turned into he rested 0 in EME and ME. That is, Eng. 0-reflexive occurs where Dan. has simple sig, e.g. han hvilte sig 'he rested'. c. OE self fused with its associate him and was reanalyzed as an adnominal intensifier adjoined to a 0-reflexive, e.g. he washed 0 himself. d. After emerging as the new intensifier EME x-self replaced the simple OE intensifier seolfin all contexts (e.g. 0 himself, Peter himself, etc.) e. Since intensifier-adjunction to simple reflexives is motivated by semantic/pragmatic factors (e.g. predicate meaning, presuppositional context) rather than a need to disambiguate between disjoint reference and coreference, the fact that it also happens in the 1st and 2nd persons is no longer a mystery. ( 11) Consequences for analysis of Mandarin Chinese: a. Chinese has 0-reflexives (;::; Dan. sig) b. § c ziji is not a reflexive but ALWAYS an intensifier c. Chinese have both simple and complex reflexives, i.e. 0 vs. 0 ziji d. the distribution of simple (i.e. 0) and complex reflexives (i.e. 0 ziji) in Chinese mirrors that of simple (i.e. sig) and complex reflexives (i.e. sig selv) in Danish, see (3) e. complex reflexives, e.g. 0 ziji, are intensified reflexives 429 f. Locally free ziji is not a logophor but an intensified object pronoun g. {tt!. !3 a ta ziji in subject position is not a special kind of anaphor but an intensified subject pronoun. h. Locally bound ta ziji is an intensified 0-reflexive, 0 ta ziji. I. !3 a ziji is not a reflexivizing particle which falls under binding theory (e.g. Reinhart and Reuland 1993), but rather an intensifier which falls under the module of intensification. One of the main lessons to be learned from the analysis defended here is that the conventional wisdom of traditional grammars cannot always be taken for granted. Just because Eng. x-self forms and Ch. ziji have traditionally been considered to be first and foremost reflexive anaphors this does not have to be taken for a good-given truth. Traditional grammarians were inclined to look for matches to categories found in Latin grammar. Since Latin has overt reflexive pronouns (which differ morphologically from object pronouns), and since Modern English x-self forms and Chinese ziji do occur in many contexts where Latin reflexives are found, it is not surprising that x-se/fforms were taken to be reflexive pronouns. However, as we have seen in chapter 5 and 6, there is a large body of data which suggest that 0-reflexives are the true Eng. and Ch. counterparts of Romance (and Latin) reflexives and that x-se/fforms and ziji are really intensifiers. Finally, the present dissertation stresses the need for comprehensive accounts which take the interaction between several different modules into account. The behavior of elements like Dan. sig, selv, and sig selv, is extremely complex exactly because they are found at the intersection of many different modules, e.g. intensification, binding, focus, phonology, argument structure of predicates, etc. A central theme of this dissertation has been to show 430 that many current accounts of binding fail to recognize the importance oftaking intensification into account when explaining the behavior of complex reflexives (e.g. Dan. sig selv) and pronouns (e.g. Dan. ham selv), see chapter 2-3. We have also shown that correct understanding of the thematic status of reflexive elements is a necessary prerequisite of any binding theory, see chapter 4. Finally, we have shown that the role played by stressability and non-contrastive stress is crucial to prosodically motivated instances of complex reflexives. Narrow accounts of binding phenomena that are limited to only to strictly binding related principles are thus likely to have correspondingly limited descriptive and explanatory adequacy. The best way to improve our understanding of both binding and intensification (as well as the other modules mentioned here, i.e. focus, argument structure, prosody, etc.) lies in the development of multi-modular models that are able to capture complex modular interaction. It is our hope that by developing an alternative comprehensive approach to binding and intensification, which differ radically from current mainstream theories of binding, the present dissertation constitutes a small step in this direction. 431 References Abraham, Werner (1995) Diathesis: The Middle, Particularly in West-Germanic. What Does Reflexivization have to do with Valency Reduction? In Abraham, W, T. Givon, and Sandra A. Thompson (eds.) Discourse Grammar and Typology. Papers in Honor of John WM Verhaar. Studies in Language Companion Series Vol. 27, pp. 3-47. John Benjamins Publishing Company. Amsterdam/Philadelphia. Abraham, Werner (2000) The structural and lexical space between reflexive binding and logophorics: Sundry paradigms of reflexives and anaphora. In Reflexives: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volume 40, pp. 104-124. Amsterdam: John Benjamins. Abush, Dorit (1989) "Reflexives, Reference Shifters, and Attitudes", in E.J. Fee and K. Hunt (eds.), WCCFL 8, Stanford Linguistics Association, Stanford, pp. 1-13. Aissen, Judith (1982) "Valence and coreference." In Paul Hopper and Sandra A. Thompson (eds.), Studies in Transitivity, 7-35 [Syntax and Semantics 15.] New York: Academic Press. Allan, R. & Holmes, P. & T. Lundskrer-Nielsen (1995) Danish: A Comprehensive Grammar. Routledge. London and New York. Anagnostopoulou, Elena, and Martin Everaert ( 1999) Toward a more complete typology of anaphoric expressions. Linguistic Inquiry 30:97-118. Anderson, S.R. (1972) "HOW TO GET even". Language 48.4:893-906. Anderson, S.R. (1986) "The typology of anaphoric dependencies: Icelandic (and other) reflexives, in L. He11an and K. K. Christensen ( eds. ), Topics in Scandinavian Syntax. (Studies in natural language and linguistic theory.) Dordrecht: D. Reidel, pp. 65-88. Anscombre, J.C. (1973) Meme le roi de France est sage. Un essai de description semantique. Communications 1973, t. 20, pp. 40-82. Aoun, Joseph (1985) A Grammar of Anaphora. Linguistic Inquiry Monographs. The MIT Press, Cambridge, Massachusetts. Ariel, Mira ( 1988) "Referring and accessibility". Journal of Linguistics 24:65-87. Ariel, Mira (1990) Accessing noun-phrase antecedents. London: Croom Helm. Ariel, Mira (1994) "Interpreting anaphoric expressions: a cognitive versus a pragmatic approach" . .!. Linguistics. 30:3-42. Baker, C.-L. (1995). Contrast, discourse prominence, and intensification, with special reference to loca11y-free reflexives in British English. Language 71.1. 63-101. 432 Barker, Chris (1995) Possessive descriptions. Dissertations in Lunguistics. CSLI Publications: Stanford, California. Barker, Stephen (1991) "Even, still and counterfactuals". Linguistics and Philosophy 14:1-38 Barss, Andrew and Howard Lasnik (1986) "A Note on Anaphora and Double Objects", Linguistic Inquiry 17:347-354. Battistella, Edwin and Yonghui Xu ( 1990) Remarks on the reflexive in Chinese. Linguistics 28:205-240. Beas, Omar (200 1) "The Derivation of Meaning and Binding Properties of SAME-elements in Spanish: Anaphors and Non-Anaphors". MS. USC. Beas, Omar (2002a) "On the Role ofClitics in Spanish Reflexive Doubling". MS. USC. Beas, Omar (2002b) "On the Role of Clitics in Spanish Reflexive Doubling". Hand-out for Student Workshop, April29, 2002. USC. Belletti, Adriana and Luigi Rizzi ( 1988) "Psych-Verbs and theta-Theory", NLLT 6:291-352. Benveniste, Emile (1966) Problemes de linguistique genera/e. Paris: Gallimard. Bickerton, Derek ( 1987) He himself; Anaphor, pronoun, or ... ? Linguistic Inquiry 18, 345-48. Bolinger, Dwight L. (1961) "Contrastive Accent and Contrastive Stress". Language 37.1:83- 96. Borkovec, Vera Zanda (1973) Grammatical and Stylistic Uses ofCertain Reflexive Forms of Verbs in Czech and in Russian. Georgetown University Ph.D. Brinton (1995) "Non-anaphoric reflexives in free indirect style". In Dieter Stein and Susan Wright (eds.), Subjectivity and Subjectivization in Language. Cambridge: Cambridge University Press, 193-194 Brown, Karen Leigh (1996) "Reflexivity and Belief De Se". In Jerry Seligmann and Dag Westerstahl ( eds.) Logic. Language and Computation, pp. 109-120. Stanford University. Blihring, Daniel (1997) The Meaning of Topic and Focus. The 59' 11 Street Bridge Accent. Routledge Studies in German Linguistics. Routledge: London and New York. Burzio, Luigi (1991) The morphological basis ofanaphora Journal of Linguistics 27:81-105. Burzio, Luigi ( 19??) "On the non-existence of disjoint reference principles". 3-27. Burzio (1996a) "The Role of the antecedent in anaphoric relations". In Robert Freidin ( ed. ), Current Issues in Comparative Grammar. Dordrecht: Kluwer, 1-45. 433 Burzio (1996b) "Anaphora and soft constraints". Proceedings of the Workshop on Optimality in Syntax "Is the Best Good Enough?". MIT Working Papers in Linguistics. Can<;ado, Maria ( 1999) "Exceptional binding with psych verbs?". LI ( 1999: 133-143). Cantrall, William R. (1973) "Why Would I Relate Own, Emphatic Reflexives, and Intensive Pronouns, My Own Self'. In Papers form the ninth regional meeting, Chicago Linguistic Society, April 13-15, 1973. Chicago Linguistic Society. Cantrall, William R. (1974) Viewpoint. Reflexives, and the Nature of Noun Phrases. The Hague: Mouton. Carroll, S. ( 1986a) "On non-anaphor reflexives". Revue quebecoise de linguistque 15:135-66. Carroll, S. (1986b) "Reflexives and the dependence relation 'R'". The Canadian Journal of Linguistics 1:1-43. Chafe, Wallace (1976) "Givenness, Contrastiveness, Definiteness, Subjects and Topics". In Charles N. Li (ed.) Subject and Topic. New York. Academic Press. Chierchia, (1989) Anaphora and Attitudes De Se. In R. Bartsch, J. van Benthem, and P. van Emde Boas, eds., Semantics and Contextual Expression: 1-31. Dordrecht: Foris. Chomsky, N. ( 1980) On binding. Linguistic Inquiry 11: 1-46. Chomsky, N. (1981) Lectures on government and binding. Dordrecht: Foris. Chomsky, N. (1986) Knowledge of language: Its nature, origin, and use. New York: Praeger. Chomsky, N. ( 1995) The Minimalist Program. Com bridge, Mass.: MIT Press. Chomsky, N. (2000) Minimalist inquiries: The framework. In Step by step: Essays on minimalist syntax in honor of Howard Lasnik, ed. Roger Martin, David Michaels, and Juan Uriagereka, 89-155. Cambridge, Mass.: MIT Press [Also available as MIT Occasional Papers in Linguistics 15, MITWPL, Department of Linguistics and Philosophy, MIT, Cambridge, Mass.] Cinque, Guilelmo (1993) A null theory of phrase and compound stress. Linguistic Inquiry 24(2):237-97. Clements, George, N. ( 1975) The logophoric pmoun in Ewe: Its role in discourse. Journal of West ~frican Languages 10, 141-77. Cole, Peter, Gabriella Hermon, and Li-May Sung (1990) "Principles and parameters of long distance reflexives". Linguistic Inquiry 21:1-22. Cole, Peter and Li-May Sung (1994) "Head Movement and Long-Distance Reflexives". Linguistic Inquiry 25:355-387. 434 Cole, Peter, Chengchi Wang (1996) "Antecedents and Blockers of Long-Distance Reflexives: The Case of Chinese ZijC'. Linguistic Inquiry 27.3:357-390. Dacygier, Barbara (1997) Reflexive markers in Polish: Parcipiants, Metaphors, and Constructions. In Verspoor, Marjolijn, Kee Dong Lee and Eve Sweetser (eds) Lexical and syntactical constructions and the construction of meaning. Prooceedings of the bi-annual ICLA meeting in Albuquerque, July 1995. Amsterdam studies in the theory and history of linguistic science. Volume 150,pp. 311-327. Dirven, R. ( 1973) "Emphatic and Reflexive in English and Dutch". Leuvense Bijdragen, Tijdschrift voor Germaanse Filologie 63:285-299. Eckardt, Regine (1999) 'Focus with Nominal Quantifiers', in P. Bosch and R. van der Sandt (eds), Focus & Natural Language Processing, vols. 1-3, Heidelberg: IBM, 281-290. Eckardt, Regine (2000). Reanalysing "selbst", Handout, Sinn und Bedeutung V, Amsterdam, December, 2000. Eckardt, Regine (200 I). Reanalysing "selbst ", MS. University of Konstanz .. Edmondson, J.A. & Plank, F. (1978). Great expectations: an intensive self analysis. Linguistics and Philosophy 2. 373-413. Erteschik-Shir, N. ( 1973) On the nature of island constraints. Doctural dissertation, MIT. Erteschik-Shir, N. ( 1981) "On extraction from noun phrases (picture noun phrases). Theory of Markedness in Generative Grammar. Proceedings of the GLOW conference. Pisa: Scuola Normale Superiore, pp. 147-69. Evans, Gareth (1980) Pronouns. Linguistic Inquiry 11:337-362. Everaert, Martin ( 1986) The ,1,yntax of reflexivization. Dordrecht: Foris. Everaert, Martin ( 1991) "Contextual determination of the anaphor/pronominal distinction". In J. Koster & E. Reuland (eds.) Long-distance Anaphora, Cambridge: Cambridge University Press. Everaert, Martin (2000) Types of anaphoric expressions: Reflexives and reciprocals. . In Reciprocals: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volume 41,63-83. Amsterdam: John Benjamins. Faarlund, Jan Terje, Svein Lie, and Kjell Ivar Vannebo (1997) Norsk Referansegrammatik. Universitetsforlaget. Oslo. Faltz, Leonard M. (1977) Reflexivization: A study in universal syntax. Doctoral dissertation, University of California at Berkeley. [Distributed by University Microfilms International, Ann Arbor, Mich., and London.] 435 Faltz, Leonard M. ( 1985) Reflexivization: A study in universal syntax. New York: Garland. Fauconnier, Gilles (1975) "Pragmatic Scales and Logical Structure". Linguistic Inquiry Vol.VI, Number 3:353-375. Ferro, Lisa (1993) "On 'self as a focus marker". ESCOL '92: Proceedings of the Ninth Eastern States Conference on Linguistics. Ithaca, NY: Cornell University, 68-79. Fiengo, Robert, and James Higginbotham (1981) "Opacity in NP". Linguistic Analysis 7.4:395421. Fiengo, Robert, and Robert May (1994) Indices and identity. Cambridge, Mass.: MIT Press. Fowler, George (1993) "A Syntactic Account of Derivational -sja in Russian." American Contributions to the Eleventh Congress of Slavists. Eds. Robert A. Maguire and Alan Timberlake. Columbus, Ohio: Slavica. 270-284. Frajsyngier, Zygmunt (2000) Domains of point of view and coreferentiality: System interaction approach to the study of reflexives. In Reflexives: Forms and functions, ed. Frajzyngier, Zygmunt and Traci Curl. Typological Studies in Language (TSL) Volume 40, 104-124. Amsterdam: John Benjamins. Frajzyngier, Zygmunt and Traci Curl (ed.) (2000) Reflexives: Forms and functions. Typological Studies in Language (TSL) Volume 40. Amsterdam: John Benjamins. Frajzyngier, Zygmunt and Traci Curl (ed.) (2000) Reciprocals: Forms and functions. Typological Studies in Language (TSL) Volume 41. Amsterdam: John Benjamins. Fraser, Bruce ( 1971) "An Analysis of "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 of Earlier English. In Reflexives: Forms andfunctions. Typological Studies in Language (TSL) Volume 40. Amsterdam: John Benjamins. Geniusiene, Emma (1987) The Typology of Reflexives. Berlin: Mouton de Gruyter. Giorgi, Alessandra (1984) "Toward a theory of long distance anaphors. A GB approach." The Linguistic Review 3:307-361. Grice, H. P. (1975) Logic and Conversation. Morgan and Cole (eds) Syntax and Semantics, Vol. 3, Speech Acts. Academic Press, 1975. 436 Grimshaw, Jane (1982) "On the lexical representation of Romance reflexive clitics". In Joan Bresnan (ed.), The Mental Representation of Grammatical Relations. Cambridge, Mass.: The MIT Press, 87-148. Grimshaw, Jane (1990) Argument Structure. MIT Press, Cambridge, MA. Grodzinsky, Yosef, and Tanya Reinhart (1993) The innateness of binding and coreference. Linguistic Inquiry 24:69-101. Gussenhoven, Carlos ( 1984) On the Grammar and Semantics of Sentence Accents, Dordrecht: Foris. Hagege, Claude (1974) Les Pronoms Logophoriques, Bulletin de la Societe de Linguistique de Paris 69, 287-310. Haiman, John (1995) Grammatical Signs of the Divided Self. A Study of Language and Culture. In Abraham, W, T. Givan, and Sandra A. Thompson (eds.) Discourse Grammar and Typology. Papers in Honor of John WM Verhaar. Studies in Language Companion Series Vol. 27, pp. 213-234. John Benjamins Publishing Company. Amsterdam/Philadelphia. Hale, Kenneth & Samuel Jay Keyser ( 1993) "On Argument Structure and the Lexical Expression of Syntactic Relations", inK. Hale & S.J.Keyser (eds.) The View from Building 20, Cambridge: MIT Press. Halle, Morris (1973) "Prolegomena to a Theory of Word Formation". Linguistic Inquiry 4.1:3- 16. Hansen, Aage (1967) Moderne Dansk. I-III. Grafisk Forlag: K0benhavn. Heim, Irene ( 19??) "On the Projection Problem for Presuppositions". Proceedings of the Second West Coast Conference on Formal Linguistics. Heim, Irene ( 1993) Anaphora and semantic interpretation: A reinterpretation of Reinhart's approach. Sf'S report 07-93. University of Tilbingen. [Reprinted in MIT working papers in linguistics 25: The interpretive tract, 205-246. MITWPL, Department of Linguistics and Philosophy, MIT, Cambridge, Mass.] Heim, Irene (1993) "Puzzling reflexive pronouns m de se reports." Hand-out, Bielefeld conference March 1994. Heim, Irene, and Angelika Kratzer (1998) Semantics in Generative Grammar. Blackwell Publishers. Helke, Michael (1971) The grammar of English reflexives. Unpublished dissertation. MIT, Cambridge Mass. Helke, Michael (1979) The Grammar of English Reflexives. New York: Garland Publications. 437 Hellan, Lars (1986) On anaphora and predication in Norwegian. In L. Hellan and K. Koch Christensen (eds.), Topics in Scandinavian Syntax, 103-124. Hell an, Lars ( 1988). Anaphora in Norwegian and the theory of grammar. Dordrecht: Foris. Hell an, Lars ( 1991) Containment and connectedness anaphors. In Long-distance anaphora, eg. Jan Koster and Eric Reuland, 27-49. Cambridge: Cambridge University Press. Hellan, Lars, and K. Koch Christensen (eds.) (1986) Topics in Scandinavian Syntax, Studies in Natural Language and Linguistic Theory. D.Reidel Publishing Company. Dordrecht: Kluwer. Hestvik, Arild (1991) "Subjectless Binding Domains", Natural Language and Linguistic Theory 9:455-496. Hestvik, Arild, and William Philip ( 1997) Reflexivity, anti-subject orientation and language acquisition. In NELS 27, 171-185. GLSA, University of Massachusetts, Amherst. Higginbotham, J. (1985) On semantics. Linguistic Inquiry 16.4. 547-593. Higginbotham, James ( 1980) "Reciprocal interpretation". Journal of Linguistic Research 1 :97- 117. Hopper, Paul J. and Sandra Thompson (1980) "Transitivity in grammar and discourse." Language 56:251-99. Horn, L. R. ( 1969) A presuppositional analysis of only and even. Papers from the fifth regional meeting Chicago Linguistic Society. 98-1 07. Horn, Laurence R. (1998) "Presupposition and Implicature". In ???? (eds.) Handbook of Semantics. Blackwell. H C T J ( ????)D. "b . . dR ,n . . I ????? uang, .. ames . . . . zstn utzvzty an eJ,exzvzty. n ..... . Huang, C.T. James & C.S. Luther Liu (2001) Logophoricity, attitudes and ziji at the interface. In Syntax and semantics, Volume 33. Huang, Yan (2000) Anaphora. A Cross-linguistic Study. Oxford Studies in Typology and Linguistic Theory. Oxford University Press. Hyman, Larry M. and Bernard Comrie ( 1981) "Logophoric Refence in Gokana." Journal of African Languages and Linguistics 3:19-37. latridou, Sabine (1986) An Anaphor Not Bound in Its Governing Category. Linguistic Inquiry 17, pp. 766-772. latridou, Sabine (1988) "Clitics, anaphors and noun phrase interpretation". Linguistic Inquiry 19:698-703. 438 Israeli, Alina (1997) Semantics and Pragmatics of the "Reflexive" Verbs in Russian. Verlag Otto Sagner. Mi.inchen. Jackendoff, R.S. (1972) Semantic Interpretation in Generative Grammar, Cambridge: Cambridge University Press. Jackendoff, R.S. (1987) "The Status of Thematic Relations in Linguistic Theory", Linguistic Inquiry 18:369-411. Jackendoff, R.S. (1992) Mme Tussaud meets the binding theory. Natural Language & Linguistic Theory 10: 1-33. Jaeggli, Osvaldo A. ( 1986) "Passive". LI 17.4:587-622. Jakubowicz, C. ( 1992) "Sig en Danois:syntaxe et acquisition". In H.G.Obenauer and A. Zribi Hertz ( eds.) Theories du Liage et structue de la phrase, Presses Universitaires de Vincennes. Jakubowicz, C. (1994a) "On the morphological specification of reflexives: implications for acquisition." In NELS 24:205-219, Merce Gonzalez ( eds ), Univ. of Massachusetts Amherst. Jakubowicz, C. (1994b) "Reflexives in French and Danish: Morphology, Syntax, and Acquisition". Janda, Laura A. (1993) "Cognitive Linguistics as a Continuation of the Jakobsonian Tradition: the Semantics of Russian and Czech Reflexives." American Contributions to the Eleventh Congress of Slavists. Eds Rober A. Maguire and Alan Timberlake. Columbus, Ohio: Slavica. 310-319. Jayaseelan, K.A. ( 1997) Anaphors and pronouns. Studia Linguistica 51:186-234. Joseph, Brian D. (1979) "On the agreement of reflexive forms in English." Linguistics 17:519- 523. Karde\a, Henryk ( l9xx.) "Target: Emphatics. A note on government, binding and case assignment in Po\ish. Societas Linguistica Europaea. Mouton Pub\ishers, The Hague. Karttunen, L. & Peters S. (1979) Conventional Imp I icature Syntax and Semantics 11, 1-56. Keenan, Edward ( 1971) Names, quantifiers and a solution to the sloppy identity problem. Papers in Linguistics 4.2. Keenan, Edward ( 1976) "Towards a Universal Definition of "Subject"". In Charles N. Li ( ed) Subject and Topic ( 1976). Academic Press, N.Y. Keenan, Edward (1988) On semantics and the binding theory. In Explaining language universals, ed. John Hawkins, I 04-155. Oxford: Blackwell. 439 Keenan, E. (1994) Creating Anaphors: An Historical Study of the English Reflexive Pronouns. MS. UCLA. Keenan, E. ( 1998) The Historical Creation of Anaphors in English. MS. UCLA. Keenan, (2000) A historical explanation of English binding theory. Paper presented at SHELL 1, UCLA, May 27. Keenan, (2000) An Historical Explanation of Some Binding Theoretic Facts in English. MS. UCLA, November 12, 2000. Keenan, E, (200 1) Explaining the Creation of Reflexive Pronouns in English. Handout. UCLA. Keenan, E, (200 1) Explaining the Creation of Reflexive Pronouns in English. MS. UCLA, May 2001. Kemmer, Suzanne (1993) The Middle Voice. Amsterdam and Philadelphia: John Benjamins. Kemmer, Suzanne (1994) Middle Voice, Transitivity and the Elaboration ofEvents. In Fox, Barbara, and Paul J. Hopper (eds.) (1994) Voice: Form and Function. John Benjamins. Amsterdam Philadelphia, pp. 179-230. Kemmer, Suzanne ( 1995) "Emphatic and reflexive -self: expectations, viewpoint and subjectivity". In Dieter Stein and Susan Wright ( eds. ), Subjectivity and Subjectivization in Language. Cambridge: Cambridge University Press, 55-82. Kibrik, Aleksandr. A., and Ekaterina Bogdanova (1995) "Sam kak operator korrektsii ozidanij adresata". [Russian sam as operator of correction of hearer's expectations]. Voprosy jazykoznanija 3:4-47. Klaiman, M.H. (1992) "Middle verbs, reflexive middle constructions, and middle voice". Studies in Language 16:1, 35-62. Klenin, Emily ( 1975) "The Pronoun sebja, Particle sebe, and Affix -sja." Slavic and East European Journa/19.2: 188-199. Klenin, Emily ( 1980) "Sentential and Discourse Prominence: The case of the Emphatic Pronoun ."Russian Linguistics 4:269-280. Koenig, Jean-Pierre (1998) "French body-parts and the semantics of binding". Natural Language and Linguistic Theory????, pp. 219-265. Konig, Ekkehard ( 1989). On the Historical Development of Focus Particles. In Weydt, H. ( ed) Sprechen mit Partikeln. Wiater de Gruyter. Berlin. New York. Konig, Ekkehard (1991 ). The meaning of focus particles: a comparative approach. The Linguistic Review 3. 307-361. 440 Konig, Ekkehard (1991 ). The meaning of focus particles: a comparative perspective. Routledge. London and New York .. Konig, Ekkehard ( 1993) Distribution und Bedeutung von Reflexivpronomina im Eng1ishen. Versuch einer historischen Erk1iirung. Lecture, Free University of Berlin. Konig, Ekkehard (1996a) Towards a Typology of Intensifiers (Emphatic Reflexives), MS. FU Berlin. Konig, Ekkehard ( 1997a) "From expressions for body parts to reflexive anaphors: Semantic change in the development of intensifiers". In Gerald F. Carr et al. (eds.), Interdigitations: Essays for Irmgard Rauch. Frankfurt: Lang. Konig, Ekkehard (1997b) "Towards a typology of intensifiers". XVIth International Congress of 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 Siemund (1998) On the development of reflexive pronouns in English: A case study in grammaticalization. ms. Institut fiir Englishe Philologie, Freie Universitiit Berlin. Konig, Ekkehard & Peter Siemund ( 1998) Towards a Typology of Intensifiers (Emphatic Reflexives). Paper given at the XVIth International Congress of Linguistics, Paris July 1997. Konig, Ekkehard & Peter Siemund (1999) Intensifiers and reflexives: a typological perspective. In Reflexives: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volume 40,41-74. Amsterdam: John Benjamins. Koopman, Hilda, and Anna Szabolcsi (2000) Verbal complexes. Cambridge, Mass.: MIT Press. Koster, Jan (1987) Domains and Dynasties. Dordrecht: Faris. Koster, Jan (1991) "Toward a New Theory of Anaphoric Binding". In Koster, Jan and Eric Reuland ( eds.) ( 1991) Long-distance anaphora. Cambridge: Cambridge University Press.Dordrecht: Faris. Koster, Jan and Eric Reuland ( eds.) ( 1991) Long-distance anaphora. Cambridge: Cambridge University Press. Kuno, Susumo ( 1972) Pronominalization, Reflexivization, and Diriect Discourse. Linguistic Inquiry 3(2): 161-195. Kuno, Susumo (19??) "The Syntax of Comparative Clauses".?????? 441 Kuno, Susumo (1987) Functional Syntax: Anaphora, discourse and empathy. Chicago: The University ofChicago Press. Kuno, Susumo, and E. Kaburaki (1977) "Empathy and syntax". Linguistic Inquiry 8:627-672. Lakoff, George (1971) 'Presupposition and Relative Well-formedness', in D. Steinberg and L. Jacobovits (eds) Semantics- An Interdisciplinary Reader, Cambridge: Cambridge University Press. 329-340. Lambrecht, Knud ( 1994) Information Structure, Cambridge: Cambridge University Press. Larsen, Erling Georg ( 1969) Narron Grammatik. Universitetsforlaget. Oslo. Lebeaux, David ( 1983) "A distributional difference between reciprocals and reflexives". Linguistic Inquiry 14: 723-30. van der Leek, Frederike (1991) "Iconicity and two-form reflexive sysmtems". Chicago Linguistic Society 27: 445-463. Levin, Beth, and Maika Rappaport Hovav (1995) Unaccusativity at the syntax-lexical semantics interface. Linguistic Inquiry Monographs 26. The MIT Press, Cambridge, Massachusetts. London, England. Levinson, S. (1991) Pragmatic reduction of the binding conditions revisited. Journal of Linguistics 27. 107-61. Levinson, S. (2000) Presumptive Meanings. MIT Press. Cambridge, Massachusetts. Lichtenberk, Frantisek (1994) "Reflexives and reciprocals." In R.E. Asher and J.M.Y. Simpson (eds.), The Encyclopedia of Language and Linguistics, vol. 7. Oxford: Pergamon Press, 3504-9. Lidz, Jeffrey ( 1995) Morphological reflexive marking: Evidence from Kannada. Linguistic Inquiry 26: 705-710. Lidz, Jeffrey (1996) Dimensions of reflexivity. Doctural Dissertation, University of Delaware, Newark. Lidz, Jeffrey (200 I) Condition R. Linguistic Inquiry 32, 123-140. Lobel, Elisabeth (2000) "Copular verbs and argument structure: participant vs. non-participant roles". Theoretical Linguistics 26:229-258. L0drup, H. (1999). Inalienables in Norwegian and binding theory. Lyutikova, Ekaterina A. (2000) Reflexives and emphasis in Tsaxur (Nakh-Dagestanian). In Reflexives: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl, 227-255. Amsterdam: John Benjamins. 442 Maling, Joan (1984) Non-clause bounded reflexives in Icelandic. Linguistics and Philosophy 7:211-41. Maldonado, Ricardo (2000) Conceptual distance and transitivity increase in Spanish reflexives. In Reflexives: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volume 40, 153-185. Amsterdam: John Benjamins. Martin, R. ( 1975) Sur I 'unite du mot meme. Travaux de linguistique et de litterature. Centre de philologie et de litteratures romanes del 'universite de Strasbourg. XIII, 1. p. 227-243. McKay, Thomas J. (1991) He himself: Undiscovering an anaphor. Linguistic Inquiry 22, 368- 73. Megerdoomian, Karine (2002) Beyond words and phrases: A un(fied theory of predicate composition. Ph.D. dissertation, USC. Milner, Jean-Claude (1989) Introduction a une science du langage. Editions du Seuil. Paris. Moravcsik, Edith A. ( 1972) Some Cross linguistic Generalizations about Intensifier Constructions. Chicago Linguistic Society 8:271-77. Moyne, J.A. (1971) Reflexive and Emphatic. Lg. 47. 141-63. Muysken, Pieter (1993) Reflexes of Ibero-Romance reflexive clitic + verb combinations in Papiamentu: Thematic grids and grammatical relations. In Focus and grammatical relations, ed. F. Byrne and D. Winford. Amsterdam: John Benjamins. Muysken, Pieter, and Norval Smith (1994) Reflexives in the creole languages: An interim report. In Creolization and language change, ed. D. Adone and I. Plag. Ti.ibingen: Max Niemeyer Verlag. Nielsen, Niels Age (1989) Dansk Etymologisk Ordbog. Ordenes Historie. Gyldendalske Boghandel, Nordisk Forlag A/S. Pan, Haihua ( 1997) Constraints on rejlexivization in Mandarin Chinese. Garland Publishing: New York & London. Pesetsky, David (2000) Phrasal movement and its kin. Cambridge, Mass.: MIT Press. Pica, Pierre (1987) On the nature of the reflexivization cycle. In Proceedings of NELS 17, 483-499. GLSA, University of Massachusetts, Amherst. Pica, Pierre (1991) On the interaction between antecedent-government and binding: The case of long-distance reflexivization. In Long-distance anaphora, eg. Jan Koster and Eric Reuland, 27-49. Cambridge: Cambridge University Press. 443 Plank, F. (1979) "Exclusivierung, Reflexivierung, Indentifizierung, relationale Auszeichnung: Variationen zu einem semantisch-pragmatischen Thema", in I. Rosengren (ed.), Sprache und Pragmatik. Lund: CWK Gleerup, pp. 330-54. Pollard, Carl, and Ivan Sag ( 1992) Anaphors in English and the scope of the binding theory. Linguistic Inquiry 23:261-305. Primus, Beatrice (1992) "Selbst-variants of a scalar adverb in German". In Joachim Jacobs (ed.), Informationsstruktur und Grammatik, Opladen: Westdeutscher Verlag, 54-88. Quirk, R. Greenbaum, S., Leech, G. & Svartik, J. (1985). A comprehensive grammar of the English language. London: Longman. Reinhart, Tanya (1983) Anaphora and semantic interpretation. London: Croom Helm. Reinhart, Tanya (2000) Strategies of anaphora resolution. In Interface strategies, ed. Hans Bennis, Martin Everaert, and Eric Reuland, 295-325. Amsterdam: Royal Academy of Arts and Sciences. Reinhart, Tanya (To appear) The theta system: Syntactic realization of verbal concepts. Cambridge, Mass.: MIT Press. Reinhart, T. & Reuland, E. (1991) Anaphors and Jogophors: An argument structure perspective. In Long-distance anaphora, eg. Jan Koster and Eric Reuland, 27-49. Cambridge: Cambridge University 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 of Massachusetts, Amherst. Reuland, Eric (1998) Structural conditions on chains and binding. In NELS 28, 341-356. G LSA, University of Massachusetts, Amherst. Reuland, Eric (2000a) Anaphors, logophors and binding. In Long-distance reflexives, ed. Peter Cole, Gabriella Hermon, and C.-T. James Huang, 343-370. Syntax and Semantics 33. San Diego, Calif.: Academic Press. Reuland, Eric (2000b) The fine structure of grammar: Anaphoric relations. In Reflexives: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl, 227-255. Amsterdam: John Benjamins. Reuland, Eric, and Tanya Reinhart (1995) Pronouns, anaphors and Case. In Studies in comparative Germanic syntax, ed. Hubert Haider, Susan Olsen, and Sten Vikner, 241-269. Dordrecht: Kluwer. 444 Reuland, Eric, and Sigrf8ur Sigurjonsdottir (1997) Long distance "binding" in Icelandic: Syntax or discourse? In Atomism in binding, ed. Hans Bennis, Pierre Pica, and Johan Rooryck, 323-340. Dordrecht: Faris. Rizzi, Luigi (1990) Relativized Minimality, Cambridge, Mass.: MIT Press. Rochemont, Michael (1986) Focus in Generative Grammar, Amsterdam/Philadelphia: John Benjamins. Rooryck, J. & G. Vanden Wyngaerd (1998) 'The Self as Other, A Minimalist approach to zich and zichzelfin Dutch' NELS 28, 359-373. Roath, Mats (1985) Association with Focus, PhD dissertation, University of Massachusetts at Amherst. Roath, Mats (1992). A theory of 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, Waltham, Mass.: Ginn & Co. Rothstein, Robert A. ( 1970) "Reflexive "Reflexive Verbs" in Polish", Slavic and East European Journal [SEEJ], Vol. XIV, No.2, pp. 194-197. Rullmann, Hotze (1988) Binding, Stress and the Distribution of Pronouns in Dutch, ms. Groningen University, the Netherlands. Ruwet, Nicolas ( 1972) "Les constructions pronominales en Frans;ais. Restriction de selection, transformations et regles de redondances." Le Fran9ais Moderne 40:102-25. Safir, K. (1992) Implied non-coreference and the pattern of anaphora. Linguistics and Philosophy 15: 1-52. Safir, K. ( 1996). Semantic atoms of anaphora. Natural Language and Linguistic Theory 14. 545-589. Safir (1997) Symmetry and Unity in the Theory of Anaphora. M.S. Saxon, Leslie (1991) "On one's own: The semantics and pragmatics of reflexives". In Carol Georgopoulos and Roberta Ishihara (eds.), Interdisciplinary Approaches to Language: Essays inhonorofS.-YKuroda. Dordrecht: Kluwer, 501-17. Schachter, Paul ( 1976) "A Nontransformational Account of Gerundive Nominals in English", Linguistic Inquiry 7:205-241. Schladt, Mathias (2000) The typology and grammaticalization of reflexives. In Reflexives: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volume 40, 104-124. Amsterdam: John Benjamins. 445 Schlenker, P. (1999) Propositional Attitudes and Indexicality: A Cross-Categorical Approach. Ph.D. Dissertation MIT. Sells, Peter ( 1987) Aspects of logophoricity. Linguistic Inquiry 18.445-79. Sells, Peter, Annie Zaenen, and Draga Zec (1987) "Reflexivization variation: Relations between syntax, semantics and lexical structure". In Working papers in grammatical theory and discourse structure, ed. Masayo Iida, Stephen Wechsler, and Draga Zec, 169-238. Stanford, Calif.: CSLI Publications. [Distributed by Cambridge University Press.' Siemund, Peter (2000) Intensifiers in English and German. A Comparision. Routledge. Sola, Jaume (1994) A uniform analysis for SELF elements. Unpublished manuscript. University ofGroningen.[Have not been able to procure copy ofthis paper] Thrainsson Hoskuldur ( 1991) Long-distance reflexives and the typology of NPs. In Long distance anaphora, eg. Jan Koster and Eric Reuland, 27-49. Cambridge: Cambridge University Press. Timberlake, Alan (1979) "Reflexivization and the Cycle m Russian." Linguistic Inquiry 1 0. I : I 09-141. Townsend, Charles E. (1967) "Voice and Verbs in -sja." Slavic and East European Journal 11.2: 196-203. Valfells, Sigridur (1970) "Middle voice in Icelandic." In Hreinn Benediktsson (ed.), The Nordic languages and modern linguistics I, 551-572. Reykjavik: Visindafelag lslendinga. van Gelderen, Elly (1996) "Self in the history of English". Germanic Linguistics Annual Conference 2. Arizona State University, Handout. Vat, Jan (1980) "Zich en Zichzelf', in Linguistics in the Netherlands 1980:127-138. Veraart, Fleur ( 1996) On the Distribution of Dutch Reflexives. MA thesis, MIT. Vergnaud, J.-R. & Zubizarreta, M.-L. (1992). The definite determiner and the inalienable constructions in French and in English. Linguistic Inquiry 23. 595-652. Vinterberg, H, and C.A. Bodelsen ( 1966) Dansk-Engelsk Ordbog. Gyldendalske Boghandel, Nordisk Forlag. Copenhagen. Waltereit, Richard (2000) What it means to deceive yourself: The semantic relation of French reflexive verbs and their corresponding transitive verbs. In In Reflexives: Forms and functions, ed. Zygmunt Frajzyngier and Traci Curl. Typological Studies in Language (TSL) Volume 40, 104-124. Amsterdam: John Benjamins. Wehrli, Eric (1986) "On some properties of French clitic se". In Hagit Borer (ed.), The Syntax of Pronominal Clitics [Syntax and Semantics 19]. New York: Academic Press, 263-83. 446 Williams, Kemp (1988) "Exceptional Behaviour of Anaphors in Albanian". Linguistic Inquiry 19:161-8. Williams, Edwin ( 1985) "PRO and Subject of NP", Natural Language and Linguistic Theory 3, 297-315. Zarechnak, Michael (1971) "-sja Verbs in Russian." Slavic and East European Journal 15.2:199-209. Zribi-Hertz, Anne (1980) "Con~ferences et pronoms reflechis: notes sur le contraste lui/lui meme en franyais", Linguisticae lnvestigationes IV: I. Zribi-Hertz, Anne (19??) "Economisons-nous: A propos d'une classe de formes reflexives metonymiques en franyais." In????? vol. ??, pp. 104-128. Zribi-Hertz, Anne ( 1987) Le reflexivite ergative en franyais moderne. In Le Franc;ais Moderne vol. 55, pp. 23-54. Zribi-Hertz, Anne (1989) Anaphor binding and narrative point of view: English reflexive pronouns in sentence and discourse. Lg. 65.695-727. Zribi-Hertz, A. ( 1995) 'Emphatic or reflexive? On the endophoric character of French lui meme and similar complex pronouns', J Linguistics 31 (1995), 333-374. Zribi-Hertz (1996) "Some wondering remarks on the development of syntactic thories: The case oflong-distance reflexives". Recherches Linguistiques de Vincennes 25. Zubizarreta, Maria-Luisa (1998) Prosody, Focus, and Word Order. Linguistic Inquiry. Monograph Thirty-Three. MIT Press. Cambridge, Massachusetts. 1.1 Abbreviations Appendix I Danish self-compounds 447 The following list of words containing the element selv is based on Vinterberg, H, and C.A. Bodelsen ( 1966) Dansk-Engelsk Ordbog. Gyldendalske Boghandel, Nordisk Forlag. Copenhagen. Abbreviations: ANT/a NEU/neu INH/i DO/do ADV/adv 10/io DITR/ditr INT/intr. VI vt n adj anti-reflexive (transitive) neutral (transitive) inherently reflexive (transitive) selv ~ direct object of verb selv ~adverbial modifier of verb selv ~ indirect object of verb ditransitive verb intransitive intransitive verb transitive verb noun adjective The abbreviations ANT/a, NEU/neu and INH/inh indicate the type of the verb on which the self-compound is based. INT/intr and DITR!ditr further describes the verb type by indicating whether the verb is intransitive or ditransitive. It is assumed that all verbs described as either ANT/a, NEU/neu, or INH/inh are either transitive or ditransitive. The abbreviations DO/do, 10/io, and ADV/adv indicate the syntactic role played by the element selv 'self inside the se({compound. That is, is it the direct object of the verb (e.g. Peter respects himself, Peter's self-respect), an adverbial element (e.g. Peter stored the goods himself, self-storage), or perhaps the indirect 448 object. Below each self-compound all the additional compounds based on the same verbal base are listed, e.g.: ANT NEU INH INT DO selvmord selvmorder se/vmorderisk selvmordsmani selvmordspilot a myrde mord n. 'suicide' n. '(person committing) suicide' adj. 'suicidal' n. 'suicidal mania' n. 'suicide pilot' vt. 'kill, murder' n. 'murder' do ADV 10 DITR Below the list of related self-compounds the base word(s) from which the self- compound is derived are listed, see entry for selvmord 'suicide' above. When the base word(s) has/have more than one meaning, the relevant meaning has been highlighted in bold fonts. Finally, on the last line of the entry the type of the verbal base is described and the function of the element selv is indicated. In the case of selvmord 'suicide' above, the original verb myrde 'kill, murder' is a transitive, anti-reflexive(a) verb. And selv can be interpreted as direct object (do). In the statistical surveys, se(f-compounds like selvmord 'suicide' are counted only once. 1.2 The morphology of derived nouns in Danish In Danish most of the compounds containing selv 'self are either nouns or adjectives derived form a verbal stem. The following lists of derivational suffixes and examples are intended to illustrate the word formation processes involved in the formation of the selv-compounds found in the list in Appendix I. That is word formation processes not encountered in the words in this list are not included in the following descriptive survey of derivational morphology. 449 ( 1) NOUN formation derivational suffixes added to verbal stems: (2) a. -else b. -(n)ing c. -eri d. -(a)tion -er e. f. g. h. -(e)lig-hed -som-hed -0 Examples of deverbal listed in (1): Verb a. bedmnme 'judge' b. begrcense 'limit' c. bygge 'build' d. amputere 'amputate' e. bygge 'build' (verb+ -Zig=> adjective, see (8b)) (verb+ -som =>adjective, see (8d)) nominalizations containing the derivational suffixes Ending -else -ning -eri -a lion -er Nominalization 'judgment' 'limitation' f virke 'work. function' -som-hed bedemme/se begrcensning byggeri amputation bygger virksomhed behagelighed be drag 'the activity of building· 'amputation· 'builder· 'business' 'pleasantness' 'deceit' g. behage 'please' -lig-hed h. bedrage 'deceive' -0 (3) Nominalizations ofthe types illustrated in (1) containing the element selv: a. selvbedommelse 'self-judgment' b. selvbegrcensning 'self-knowledge' c. selvbyggeri 'building one's house with one's own hands' d. selvamputation 'spontaneous amputation' e. selvbygger 'person who builds his house with his own hands' f. selvvirksomhed 'self-activity' g. selvbehagelighed '(self-)complacency, self-satisfaction, smugness' h. selvbedrag 'self-delusion, self-deceit, self-deception' ( 4) NOUN formation derivational suffixes added to adjectival stems: a. -hed (adj.+ -hed =>noun) (5) Examples of 'de-adjectival' nominalizations containing the derivational suffixes listed in (4): Verb Ending Nominalization a. b. virkesom 'working, functional' -hed behage/ig 'pleasant' -hed virksomhed behagelighed 'business' 'pleasantness' (6) Nominalizations of the types illustrated in (5) containing the element selv: a. selvvirksomhed 'self-activity' b. selvbehagelighed '(self-)complacency, self-satisfaction, smugness' ((6a,b) = ((3f,g)) 450 (7) ADJECTIVE formation derivational suffixes added to verbal stems: a. -ende (gerund/present participle) b. -(e)lig c. - (e)tldlen (past participle) d. - som e. -(a)tiv (8) Examples of deverbal 'adjectivizations' containing the derivational suffixes listed in (7): Verb Ending Adjective a. beskue 'look at, observe' -ende beskuende 'observing b. behage 'please' -fig behagelig 'pleasant. pleasing' C. erhverre 'acquire -I ejhvervet 'acquired' d. virke 'work, function -sam virksom 'working, functional· e. forme 'form' -(a)tiv formativ 'formative' (9) 'Adjectivizations' of the types illustrated in (8) containing the element selv: a selvbeskuende 'introspective; narcissistic' b. selvbehagelig '(self-)complacent, self-satisfied, smug' c. selverhvervet 'self-acquired, self-taught (fx knowledge)' d. selwirksom 'active' (I 0) ADJECTIVE formation derivational suffixes added to nominal stems: a. - isk (II) Examples of adjectivizations containing the derivational suffixes listed in ( 1 0): Noun Ending Adjective a. morder 1 'murderer' -isk morderisk 'murderer-like' ( 12) Adjectivizations of the types illustrated in (II) containing the element selv: a. selvmorderisk 'suicidal' 1.3 List of compounds containing the morpheme selv 'self' ANT NEU INH INT DO ADV 10 DITR se/vajbryder n. 'automatic cutout' a a ajbryde I interrupt, break (fx the silence); 2 interrupt (fx conversation); cease to do, interrupt, break otT (fx work); 3 (electric current) interrupt, cut off, switch off; 4 (teleph) cut otl disconnect [ajbryde sig selv switch-otT REFL self 'switch-off itself] [afryde strgmmen afsig se/v do switch-otT current-the of REFt self adv 'switch off the current by itself] 1 The agentive noun morder 'murderer'. in (II a) above, is derived from the verb myrde 'to kill' through both (apophonic) vowel change (y > o) and adding the agentive suffix --er. selvafladning aflade n. 'self-discharge' I (naut) ship; unload; 2 discharge (fx a battery) [ ajlade s1g selv discharge REFL self a do selvajlceggende adj. 'self-dump(fx hayrake)' 'discharge itself] selvajla:gger n. 'self~rake' (note: afla:gger = '(agricult.) side-delivery reaper' afla:gge I put down; 2 take off, discard; 3 drop, leave off, discard; 4 (gardening) layer, lay [ajla:gge heel af sig selv drop hay-the of REFL self 'drop the hay by itselfl a ~v selvagte/se n. 'self-respect, self-esteem' agte I esteem, respect; reverence, revere, venerate; 2 intend; mean; 3 regard as, consider [agte sig se/v esteem REFL self 'esteem oneself] a ~ selvamputation n. 'spontaneous amputation' amputere [amputere sig selv I amputate amputate REFL self do a [amputeredes af sig selv amputated-PASSIVE ofREFL self adv 'spontaneously' a selvanalyse n. 'self-analysis' analysere I analyze r analysere sig selv analyze REFL self do a selvanden adv. 'with one other (person)' I (num .. ) second; (adj) other i 'JO and en 'other' is not a verb} selvangive selvangive/se v. 'file one's (income) tax return' n. 'confession; tax return; tax form' 'amputate itself] 'was amputated by itself'] 'analyze oneself] angive I state; note; 2 indicate; 3 allege; profess; 4 inform against, denounce. [ angive skatten sell' state tax-the self 'state the tax oneself] a [an give sig selv adv denounce REFL self do a selvanklage selvanklagende selvanklager n. 'self~accusation · adj. 'self~accusing, self-accusatory' n. 'self-accuser' anklage [anklage sig selv I accuse. accuse REFL self do a selvansvarlig adj. 'responsible· ansvarlig I adj. responsible [ansvarligfor sig selv responsible for REFL self {0° ansvarlig 'responsible' is not a verb} selvanta:ndelig selvanta:ndelse selvanta:nding selvanta:ndt adj. 'spontaneously combustible' n. 'spontaneous combustion, self~ignition' n. 'spontaneous combustion, self~ ignition' adj. 'ignited spontaneously' 'denounce/inform against oneself] 'accuse oneself] 'responsible for oneself] anta:nde I set on fire; set fire to, kindle, ignite (fx ignite the petrol in a motor) [ anta:nde sig selv a selvarbejde arbejde n. 'self-activity' I work intr. ignite REFL self 'ignite itself"] (do) (adv) 'spontaneously' adv (??) 451 selvavl selvav!et a avle selvbebrejdelse bebrejde a selvbedrag bedrage a selvbedommelse bedgmme a selvbeji-ugtende selvbefrugtning befrugte a n. 'spontaneous generation, autogenesis' adj 'generated spontaneously' I grow, raise; 2 breed; 3 beget, procreate, engender, generate (do) (adv) 'spontaneously' adj. 'selt~reproach' I reproach do n. 'self-delusion, self-deceit, self~dcception' I deceive, cheat do n. 'self-Judgment' I judge; estimate; mark do adj 'selt~fertilizing, seltlng' n. 'self-fertilization' I (bioi) fertilize, fecundate; (artificially) inseminate. do n. 'self~knowledge' (" 0 ) selvbegrcensning begrcense begrcense sig I limit; 2 reduce, restrict I keep within reasonable bounds. a selvbehag selvbehagelig selvbehagelighed behage do n. '(self~ )complacency, selt~satisfaction, smugness' adj. '(selt~)complacent, selt~satisfied, smug' n. '(selt~ )complacency, selt~satisfaction, smugness' I please, appeal to; 2 deign (to do ... ) 452 a (do) (although behagelig is an adj. it is derived from a verb behage 'please' l selvbeherskelse n. 'selt~control, self-command, selt~restraint' selvbehersket adj. 'selt~possessed, selt~controlled' beherske I rule over, govern; 2 control, dominate, be master of; 3 master (fx a language) beherske sig I control oneself, restrain oneselC keep one's temper (a) (i) do selvbehold beholde a selvbekendelse bekende a selvbekrcefielse bekrcejie a n. 'retention' I keep, retain. do n. '(voluntary) confession' I confess meronym1c (do) (adv) ml n. 'self-affirmation' I certify; 2 affirm, contlrm do selvbesindelse n. ·self-retlection; self-communion, collectedness' besinde sig I collect oneself, collect one's thoughts; 2 regain one's composure; 3 think twice, pause, deliberate selvbeskatning beskatte a selvbeskikket beskikke a (do) (adv) n. 'selt~taxation' I tax; rate; assess do adj. 'selt~appointed, selt~constituted' I appoint; 2 allot do selvbeskuelse selvbeskuende beskue a n. 'introspection, self-examination, self-observation' adj. 'introspective; narcissistic' I contemplate, behold, view, gaze at. do selvbeska!fligelse n. 'self-activity; self-communion' beska!flige I occupy, engage; 2 employ beska!flige sig (med) I be occupied with, occupy oneself with, devote one's time to. (a) (n) (i)( 70 ) do se/vbesmittelse n. 'self-abuse, self-pollution, masturbation, onanism' besmitte I defile, pollute; sully (fx one's name), contaminate (fx c. sby's morals) a ~ selvbespejling bespejle spejle spejle sig n. ·narcissism' NO VERB I reflect (fx the lake retlects the trees); mirror; 2 fry (fx an egg) I be reflected, be mirrored (fx in the water) ( 70 there is no verb *bespejle} selvbestaltet bestaltet (a) se/vbestemmelse selvbestemmelsesret bestemme a se/vbestovende selvbestover se/vbestevning bestove a selvbesaning be sa a adj. 'self:appointed, self-constituted' adj. duly appointed NO VERB (do) n. 'self-determination' n. '(right of) se]t:determination' I decide, make up one's mind, determine, resolve; 2 determine, fix (fx the price) adv adj. 'autogamous, self-compatible' n. 'autogamous plant' n. 'autogamy, se]t:pollination' I pollinate (do) (adv) n. 'selt:seeding' I sow (fx a field) (do) (adv) se/vbetjenende adj. 'automatic, self-acting' selvbetjening n. 'selt:service' selvbetjeningsvaskeri n. 'launderette' betjene I serve; attend; wait on; 3 operate, work ( fx machine) a ~ selvbetragtning betragte a se/vbeundring beundre a se/vbevidst se/vbevidsthed bevidst po there is no verb} n. 'introspection, self:communion, se]t:observation' I look at, contemplate; view do n. 'se]t:admiration' I admire do adj. 'selt:conceited, selt:opnionated, self-important, arrogant' n. 'self-conceit, sell: importance, arrogance' I conscious; 2 self-confident; selvbeva!gelig adj 'sea: propelling (fx guns)' se/vbeva!gende adj. 'se]t:propelling (fx guns)' beva!ge I move; agitate; 2 move (tx the scene moved him); affect, touch beva!ge sig I move, stir (a)?" (n)?O Cim (do) (adv) 453 selvbindende selvbinder selvbindersejl bin de adj '(- mejemaskine) see selvbinder' n. 'reaper-binder, harvester, self-binder' n. 'canvas' I bind, tie; tie up (fx a dog); 2 tie (fx a knot); 3 bind (fx this promise binds me) 454 a adv {the machine is not binding itself, but biding hay by itself} selvbiograf selvbiograjisk biograf n selvblegende blege a n. 'autobiographer' adj. 'autobiographic( al)' I n. biographer (do) adj '(gart)- b/adse/leri selt~blanching celery' I bleach, blanch, whiten do selvbuden adj. '-g<est intruder; gate-crasher' byde!bodlbudt(buden) 1 ask, invite; 2 order, command indbyde 1 invite a ~ selvbygger selvbyggeri selvbygget bygge a n. '[person who builds his house with his own hands' n. '[building one's house with one's own hands' adj '[built with one's own hands)' 1 build, construct (fx a house) adv selvbcerende adj. 'standing on its own base' bcere 1 carry (compare w. bcere s1g 'behave ') a do selvcentrerende centrere a selvdanne/se selvdanne/sesevne selvdannet danne danne sig (a) adj. 'selt~centering' 1 center n. '(biol) = se/vav/' do n. 'power of spontaneous generation' adJ. 'selt~created' (i) 1 form, make (up), frame, shape, mould; 2 form, found (fx a society); 3 make, constitute; 4 form; produce; generate (fx heat) 1 form (itself) (fx a red spot formed on his hand) (do) (adv) 'spontaneously' selvdisciplin n. 'selt~disciplin, discipliner e 1 vt disciplin a ~ selvdreven drive a selvdyrke/se dyrke a selvded de adj. 'selt~propelled' 1 drive, move, drift (fx cattle); 4 drive, operate, work, (- frem fx car, ship, airplane) propel do n. 'self-worship' 1 worship (fx a god), idolize (fx a hero) do adj 'dead from accident or disease (about animals)' 1 die, pass away INTRANS int adv selvdemme n. '[the right of the plaintitfto decide the penalty to be paid by the defendant]'adv demme I judge, decide, a (do) selve adj 'the very' {?? selve =adjectival torm ofse/v, no verbal base) selvegen egen {?? no-verbal basel selveje selve;endom se/vejende selvejer se/vejerbil selvejerbonde selvejergard se!vejen,ogn a one eje selverfaring erfare a se/verhverv se/verhvervende selverhvervet erhverve adj 'original, particular, peculiar, specific' I adj. own n. 'private property; freehold' n. 'private property; freehold" adj. '-institution private foundation, -landmand owner-farmer' n. 'owner; (afland) freeholder, owner-occupier' n 'private car' n. 'freeholder, yeoman (farmer), peasant proprietor' n. 'freehold farm' n. 'private carriage' I own, possess 455 adv { se/veje does not mean that one owns oneself but rather that n. 'personal experience' I experience n. 'independent employment' adJ. 'selt~supporting' adv owns something by oneself, cf. selvbygger. se/vbinder, self storage, etc. Seems to be a productive way to form self nominalizations l adj 'self~acquired; self-taught (fx knowledge)' NOTR. VERB erhverve sig I acquire; gain, get, obtain; achieve ( fx status) erhverve sit udkomme I earn one's living. erhverv I n. occupation, trade, pursuit, calling a adv ditr se/verkendelse n. 'selt~knowledge' erkende I admit (fx admit that you are right), acknowledge (fx one's mistakes); recognize 2 understand, realize; (phil) apprehend, know, arrive at a cognition of a ~ selveste adj the very, the highly esteemed, etc. { 7 'J superlative of selve, no verbal base J selvforagt foragte a selvforblindelse forb Iinde a n. 'selt~contempt' I despise, disdain, scorn, hold in contempt; do n. 'delusion, I (fig) blind, dazzle, infatuate. do se/vforbrcending n. '(see selvantcendelse)' forbrcende I vt/vi. burn (fx wood burns easily); 2 bum (fx one's hands); 3 scorch, parch. mtrans. a (do) (adv) se/vfordybelse n. 'selt~communion' fordybe fordybe sig i I deepen, make deeper; 1 become deeply absorbed in, become immersed in, bury oneself in a se/vfordCJjelse fordo;e a selvfordgmmelse fordCJmme a (i) n. '(fysiol) autolysis' I vt. digest n. 'self-condemnation' ?? (the nominalization is derived fromfordybe sig i rather thanfordybe) (io)?? (adv) {DO does not make sense here l I vt condemn (fx c. sby. for his conduct); denounce (fx he denounced his conduct publicly) do selvforgiflning forgifle a n. 'autointoxication' I vt poison. do 456 selvforglemme/se n. 'forgetting oneselt" selvforglemmende adj. 'selt:forgetting, self-denying (fx love)' forglemme I forget; ikke at- not forgetting. forglemme sig I forget oneself (fx he torgot himself as far as to strike her) (a) (n) do sehforgude/se forgude forgudelse a n. 'self-worship' I vt idolize, dote on. I n. idolization do n. 'self-exaltation. selt:glorification' selvforher/igelse forherlige forherlige/se I vt glorify, exalt. extoL I n. glorification. a sehforklarende fork/are a selvformal formal selvfornedrelse fornedre a sehforncegtelse selvforncegtende forncegte (a) (n) sellforpleJning forpleje a selvforringelse forringe a se hforsi krer selvforsikret selvforsikring forsikre a selvforskyldt forskyldt ( oo no verbal base) selvforstcerkende forstcerke a do adj. 'selt:explanatory' I 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)O'I n. 'end in itself I n. aim, end, object, purpose, objective. n. 'selt:abasement' I vt degrade, debase, abase; hvo sig selv ophcyer skal -s. og hvo sig selv -r skal oph<Jjes (bib/) whosoever shall exalt himself shall be abased; and he that shall humble himself shall be exalted. do n. 'selt:denial' adj. 'selt:denying' I 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) han -de sig ikke T he ran true to form; - sig selv (bib!) deny oneself (fx let him deny himself and take up his cross and tollow me) han -r ikke sig se/v that is JUSt what he would do, that is him all over. do n. '[supplying oneself with provisions]' I vt board, feed; cater for. do n. 'selt:disparagement' I vt reduce (fx the possibilities of.); impair (fx our influence abroad); 2 depreciate, detract from the value of; 3 disparage (=forklejne) n. 'sclt:insurer' adj 'selt:insured' n. 'selt:insurance · do I vt insure; 2 assure ( fx he assured me that I was wrong) (do) (adv)O'l adj. 'selt:inflicted' I adj.fa I<Jn sam- get one's deserts adv adj. 'selt:increasing' I vt strengthen (fx the foundations of a building); make stronger. fortify; 2 vt augment reinforce; 3 intensify, heighten, enhance (fx the etfect); 4 (radio) amplify do 457 n. 'self-knowledge' selvforstaelse forsta I vt & vi understand, comprehend, perceive, grasp~ realize, sec; 2 know ( fx he knows how to .. ) a do selvforsvar n. 'selt~defence' forsvare (Jorsvare sig) I vt defend (fx one's country, one's interests, a fortress) I defend oneself, put up a defense n do selvforsynende adj. 'selt~sufficient' selvforsyning n. 'self-sufficiency' selvforsyningspolitik n. 'policy of self-sufficiency, autarkic policy" forsyne I vt supply sby with sth, provide sby with sth, furnish forsyne sig (med) I 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 trom sby n selvforsergende forserge adj. · sel f~supporting, independent' 10 ditr (a) forserge sig se/v (n) I vt support, keep, provide for, maintain,- enfamilie support a family support oneself, be self-supporting (do) (adv) sehforyngelse n. '(forst) natural reproduction, selt~sowing' forynge I vt rejuvenate~ (forst) regenerate foryngelse I n. rejuvenation, rejuvenescence; (forst) regeneration. kunst(glnatur/ig Joryngelse (forst) artificial/natural regeneration a (do) (adv) selvfremkaldt adj. '-sar self-inflicted wound' fremkalde I cause, bring about give rise to. produce, call forth, provoke, evoke~ engender~ 2 (photo) develop~ 3 give a call (fx actor atler the end of play) a adv selvfrugtbar selvfrugtbarhed frugtbar adj f')'H \··/ selvfoldende selvfolder fvlde a selvfelelse sehfe/ende a fole fele/se selvfe/ge sehfe/gelig selvfe/gelighed fg/ge a selvgjort ggre a selvglad glad adj {??) adj. 'autogamous, self-fertile' n. 'autogamous, selt~fertilization' adj fertile, rich (soil)~ 2 fruitful, productive (fx co-operation) adv adj. 'self-tilling (fx fountain-pen)' n. 'self~ tilling fountain-pen, self-filler' a 1 vt fill (fx fill the glass, fill the pen), till up~ 2 refill, replenish~ 3 inflate, blow up, fill (do) (adv) n. 'selt~esteem~ egotism' adj. 'conceited, self-satisfied, selt~important~ proud' I vt feel~ sense (fx I sensed a certain hostility in his manner)~ perceive, be sensible of n. feeling~ sentiment; 2 feeling, sensation (fx sensation of heat); 3 sense of touch; etc .. (do)?' 1 n. 'matter of course' adj. 'natural, inevitable; obvious' n. 'matter of course; truism~ naturalness' I vt follow (advm adJ. 'of one's own making, of one's own creation' I vt do, make adv ( cf selvbygger, etc.) adj 'pleased with oneself, self~satisfied~ complacent; smug, adj. happy, (io)?'i selvglcede n 1')')\ t··1 glcede selvgod selvgodhed god n. 'self:satisfaction, complacency' n. happiness (io) ?'> adj. 'selhighteous, priggish, pharisaric(al), smug' n. 'self-righteousness, pharisaism, smugness' adj good 458 selvgroet adj. 'self-sown, seJt:set; in its natural shape (abt wood); (fig.) peculiarly one's own (Gruntvig's mode of thought was peculiarly his own and came from withm himself' gro selvgcering gcere selvgaende ga selvhelbredelse helbrede I vi grow; 2 vt grow intr adv n. 'anaerobic respiration' vi ferment; (tlg) ferment, work (fx the suggestion was beginning to work in his mind) intransitive intr adv adj. '(mining)- maim self-fluxing ore' vi walk, go, intr adv n. 'spontaneous healing' I vt cure, restore (to health); (bib! & poet about miracle cures) heal helbredelse n. cure, restoration to health (fx to cure sby); 2 recovery (fx sby recovered) n (do) (adv) 'spontaneously' selvhengive/se n. 'se]t:devotion, hengive hengive sig til I vt sacrifice, give up; abandon oneself to, give oneself up to; indulge in (fx emotion, pleasures, etc.) (io)?? selvhersker selvherskerdgmme selvkerskermagt hersker a i jC)')\ I··J n. 'autocrat" n. 'autocracy' n. 'autocracy' vi reign; 2 rule; 3 be (fx there was great confusion in the house); be prevalent; prevail (do) (adv) selvhjulpen adj 'resourceful; (=selvjorsc1rgende); (abt. one who has succeeded) seJt:made' selvhjcelp n. 'self-help' selvhjce/pssalg n. '[sale by an unpaid vendor in order to ascertain the amount of his damages]' hjcelpelhjalplhjulpet(-en) vt help sby; a do selvhcejiende hcefle (a) selvhcerdende hcerde selvhcevde/se selvhcevdende hcevde a selvhcevende hceve selviagttagelse iagttage a adj. '-vildvin (Parthenocissus tricuspidata) Japanese ivy, Boston ivy' vt tlx, fasten, attach, stich; 2 stick, paste; 3 stitch, sew; 4 vi stick (the paper will not stick); cling. (intr) (do) (adv) adj '-stat seJt:hardening steel' vt. harden, intr n. 'sclt~assertion' adj 'self:asserting, assertive' assert, claim, do adj 'se]t:raising (fx tlour)' adv raise (about flour, dough, etc ) intr adv n. 'selt~observation, introspection; personal observation , vt. observe, do selvi/cegger n. '(agricult) self-feeder' (ilregge ooooo; lregge _ i) (no verbal base} {no verbal base} adv selvimprcegnering n. '(forst) self-impregnation' imprcegnere vt impregnate, proof(fx- wood) a (do) (adv) selvimpulserende adj. 'selt: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· ind/ysende adj (no verbal base} adv {? 0 } (cf se/vfelge. selvformal, etc.} selvindstillelig adj. 'selt:adjusting' indstillelig adj. ajustable indstille vt adjust ( fx apparatus, an engine, a brake) a selvindtrrede selvindtrcedelse selvindtrceden selvindtrceder selvindtradt do v. 'to contract on one's own behalf n. 'contracting on one's own behalf' n. 'contracting on one's own behalf n. 'commission agent contracting on his own behalf' n. 'contracting on one's own behalf' ENESTE verbum!! indtrcede vi. I enter; 2 commence, set in (fx a crisis set in); take place intr adv selvinduktion n. 'selt:induction' selvinduktiv adj. 'selt:inducive' induktion n. (physics) induction inducere vt (physics) induce a do (adv) ?? selvinfektion adj 'auto-infection, selt:infection · inficere a selvinteresse interesse interessere (do) (adv) 'spontaneously" n. 'self-interest (fx enlightened self: interest)' n. interest vt interest a interesse} adv {0°} ( = sehformal. Selvinteresse is derived from the noun selvironi selvironisk ironi n po no verbal base} selvisk seiVIskhed adj. P''l selvjustits n 1'7')1 l··J just its selvklatrende klatre selvklog klog n. 'self:irony' adj. 'selt:ironic' n. irony adj. 'selfish· n. 'selfishness· n. '(=se/vtcegt)ljur.) self:help' adv O? n. justice, administration of justice adj '(bot) climbing, creeping, trailing' vi climb, clamber, intr adj. ·selt:opinionated; conceited' adj smart. intelligent adv adj ( oo no verbal base} (='irony on one's own behalf) 459 selvklcebende 'adj. adhesive (a latex-gummed envelope)' klcebe vi. stick. cleave, adhere (til. ved to) klcebe sig til vt. stick. glue, paste; - sammen 'glue together' 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 { oo} 'on one's own behalf selvkontrol n. (=selvbeherskelse) 'selt~control. kontrollere vt. control a ~ selvkritik n. 'selt~criticism' selvkritisk adj. ·selt~critical' kritisere a selvkrejende kreje kroge kroge sig selvkorende kere selvkiiren kare a selvlade selvladende selv/adepistol selvladevaben lade vt. criticize do adj. '-mol/e mill automatically keeping in the wind' (spelling variant of kroge { 00 }) vi. bend bend, stoop intr adv adj 'selt~propelling, selt~propelled (fx gun)' vi. drive. go, (abt car) run, go, travel intr adv adj. (=selvbestaltet) adj. 'self-appointed, selt~constituted' vt. choose. elect, do adj- pretix-'automatic (pistol (lit.) sel1~loading)' adj. 'automatic' n. 'automatic pistol' n. 'automatic firearm' vt. load (a gun). etc 460 a (do) (adv) {does the gun load itself. or does it load (middle) by itself/} selvlavet lave a selvlede lede {no verbal basel selvlemlcestelse lemlceste a selvlossende selvlossepram losse a selvlukke selvlukkende selvlukker lukke lukke sig adj. 'home-made, of one's own making' vt. make. adv n. 'selt~loathing' n. disgust. loathing; distaste for. (do)?" n. '(jur.) selt~mutilation' vt. mutilate do adj 'self-discharging' n. 'self-discharging lighter, hopper barge' vt. (naut.) unload, discharge. unship (do) (adv) n. 'med- selt~closing' adj. 'self-closing (fx door)' {proof that lukke is not a true inherently reflexive but a middle 0 } n. 'selt~closing device; (for dor) automatic door closer' vt. close (fx a door) (a) lukke sig af sig selv (n) vi. close (fx a door closes) (e.g. the door closes by itself) (do) (adv)O? se/vlyd n. 'vowel; sonant' selvlydsrim n. 'assonance' selvlydssammenstgd n. 'hiatus' lyd n. sound {no verbal basel se/v/ysende lyse se/v/a:gende /a:ge a se/v/a:nsende la:nse a se/vla:rd /a:rd la:re /a:re adv? 0 'by itseW adj. 'luminous; luminescent' vi. shine, give out light intr adv 'by itself adJ. '(med.) self-healing' vt. heal. cure vi. heal (up) do intr adj. '(naut.) self-bailing' {is the verbal base vt. or vi.} adv vt!vi empty, discharge; (naut.) free (fx a boat) (do)OO adj. 'selt~taught, auto-didact' adj. learned, erudite; scholar, vt learn vt. teach 461 a (do) (adv) {is the verbal source lrere 'learn', or lrere 'teach'') On ly the second selvlrert adj. 'selt~taught' la:rd adj. learned, erudite; scholar, /a:re vt learn /a:re vt. teach a second se/vmader made a se/vmagnetiserende se/vmagnetiseret se/vmagnetisering (do) n. '(pa ta:rskeva:rk) self~ feeder' vt. feed do adj. 'selt~exciting (fx dynamo)' adj 'self-excited' n. 'selt~excitation' magnetisere vt. magnetize a se/vmedlidende adj. 'self-pitying' se/vmedlidenhed n. 'self-pity' med/idenhed n. compassion /ide vi. suffer {no verbal basel se/vmodsigelse se/vmodsigende modsige a n. 'selt~contradiction' n. 'self-contradicting' n. 'suicide' vt. contradict do n. '(person committing) suicide' adj 'suicidal' n. 'suicide club' (adv) (adv) se/vmord se/vmorder se/vmorderisk se/vmorderklub selvmorderske se/vmordsbatalj'on selvmordsdrama se/vmordsepidemi se/vmordsjlyver n. '(female person committing) suicide' n. 'suicide squad' n. 'suicide tragedy' n. 'epidemic of suicides' n. 'suicide flyer' verbal base allow the (do) reading} {is the verbal source /a:re 'learn', or /a:re 'teach 'o Only the verbal base allow the (do) reading} io (0°) { medlidenhed med sig selv} se/vmordsforsf!g n. 'suicide attempt' se/vmordshyppighed n. suicide rate" se/vmordskandidat n. 'would-be suicide' se/vmordsmani n. 'suicidal mania' se/vmordspilot n. 'suicide pilot' se/vmordstanke n. 'omgas med- -r contemplate suicide' se/vmordstilbojelighed n. '(sociol.) trend in suicide rates' myrde vt. kill, murder mord n. murder a se/vmf!der mCJde mCJde selvmal mal {no verbal base} se/vom selvopgivelse selvopgivende opgive se/vopholdelse se/vopho/delsesdnft do n. 'ljur.) 'plaintiff vi appear, attend. vt. meet (sby). intr n. '(sports) own goal' n. goal conj. 'even though· n. 'despair, defeatism' adj. 'despairing, defeatist' vi. give up n. 'self-preservation' n. 'instinct of self: preservation' vt. detain, delay, keep adv adv opholde opholdelse {derived from nominal base) n. - af livet sustenance do selvophfJjelse ophldje a se/voplevelse se!vop/evet op/eve a se/voplldsning op!CJse a selvopofre/se se/vopofrende opofre a n. 'self-exaltation· vt. exalt do n. 'personal experience· adj. 'personally experienced' vt. experience sth. n. '(fysiol) autolysis' vt. dissolve do n. 'selt:sacritlce, self-denial' adv adj 'self-sacrificing, seJt:denying; devoted' vt. sacrifice do {exclusive adverbial reading ofselv} se/voptaget se/voptagethed optage adj 'seJt:centered, selt:absorbed, wrapped up in oneself' n. 'self-centeredness, selt:absorption' vt. take up, pick up, etc. vrere optaget af {adjectival base) adj. be busy with sth, be engaged in sth, be absorbed in sth io { cf selvfordybelse/vcege fordybet i. etc.} selvoptrcekkende adj. '(about watch) self-winding' optrcekkeltrcekke _ op vt. wind (a clock, watch) a do selvopvarmning n. 'spontaneous heating' opvarme vt. heat, a (do) (adv) 'spontaneously' 462 selvovervindelse n. 'self-conquest; resignation' oven,inde vt. beat, conquer, defeat, vanquish, a ~ selvovervurdering n. 'exaggerated opinion of oneself overvurdere vt. overestimate a do selvpasser j<)')l t··j selvpejler a itself! (1?passe) pej/e selvpensionering selvpensionist pensionere a selvpineri selvpinse/ a pinsel pine selvpisker selvpiskning piske a selvplage selvplager selvplageri selvplagerisk a plage plage selvportrcet portrcet {no verbal base} n. '(mill) automatic governor' {is there a corresponding verb?) n. '(tlying) (wireless) direction tinder' vt. take bearing, take bearings: 2 sound (fx the pumps, the tanks): 3 locate adv {a selvpejler t!nds the direction on its own, it does sound n. 'contributory pension scheme: providing for one's old age' n. 'person who, in the absence of a pension has to provide for his old age' vt. pension, grant a pension to n. 'scJt:torture' n. 'seJt:torture' do n. torture, torment vt. torture, torment do n. 'flagellant' n. 'flagellation' vt. whip n. 'se]t:torture' do n. 'self:tormentor; tlagellant' n. 'selt:torture' adj. 'selt:tormenting' n. nuisance, atlliction, plague, curse vt. torment do n. 'self:portrait' n. portrait do { se/vportrcet = portrcet af sig se/v} selvprove/se n. 'se]f:cxamination' prove n. test prove vt. test, a do selvpa/cegger n. '(typigraph.) automatic feeder' (pa/cegge) selvransage/se n. 'selt:cxamination' ransage a selvregistrerende registrere selvregulator selvregulerende selvregulering regulere vt search (fx one's conscience, one's heart, one's memory); examine do adj 'self-registering' vt. register n. 'self-regulator' adj 'selfregulating' n. 'selt:regulation' vt. regulate 463 selvrejsende rejse selvrensende selvrenser se/vrensing rense rense sig n adj 'selhighting' v adj 'self-purifying' n. 'automatic purifying apparatus' n. 'natural purification' vt. clean clean oneself do se/vrespekt n (= selvfagtelse)'self-respect, seJt:esteem' respektere vt. respect a selvre!fi:erding selvretfcerdighed retfcerdig se/vrisiko se/vrisikoklausu/ risiko ( n. no verbal base: selvros rose a do adj 'selhighteous' n. 'self-righteousness' adj. just, righteous n. 'franchise; own risk' n. 'coinsurance clause' n.risk n. 'selt:praise' vt. praise do adv adj 'self-willed_ wilful, headstrong, stubborn, obstinate' n. 'self-willedness, wilfulness, stubbornness, obstinacy' NEllTRAL VERB se/vradig se/vradighed rade rade over s1g se lv vb 1 advise, counsel; 2 be master, command (fx who commands here?), rule 'be master over oneself(??) do (adv)?O selvsagt adj. 'selt:evident, obvious, plain' sige vt. say del siger sig selv it goes without saying, it is obvious a adv se/vsamme sam me se/vsikker selvsikkerhed sikker pron./ adj. ·self-same; the very same' adj same adj. '(self-)assured, selt:contldent, cocksure' n. '(selt:)assurance, self-confident. cocksureness' adj. sure, contldent { adj ?? no verbal base: (adv) se/vskabt skabe a adj. 'selt:created, of one's own making' vt. create adv ( cf selvbygger, etc ) se/vskiflende skifle n. '(jur)- arvinger heirs who distribute an estate without recourse to a court' vt - et bo administer an estate ?? selvskildring skildre a se/vskreven skrive a n. 'description of oneself vt. depict, describe do adv adj. 'in one's own handwriting; eminently qualified; obvious natural' vt. write adv 464 se!vskud se/vskygge ~? skyde skyde sig skud skygge skygge se/vskyldner sky/de ?? se/vsmf!rende selvsmerer se/vsmering sm<Jre a n. '(hunting) lcegge- for set a spring-gun for' vt shoot vr. shoot oneself n. shot n. '(lightening techique) shadow' n. shade; shadow vt & vi. give shade; 2 shade, cast (its) shade upon; 3 shade, shelter intr adv n. '(jur) surety' v. ditr. owe sth to sby. adv adj. '(tech ) selt~lubricating' n '(tech ) automatic lubricator' n. '(tech ) automatic lubrication, self~ lubrication' vt lubricate do ditr se/vspej/ing n. '(psycho!) narcissism; (abt. lit perfonnance) reflection of the writer's personality' ?? spejle spejle sig se/vspillende spi/le a selvspcegelse vt reflect; mirror look at oneself in a mirror do adj.'- klaver player-piano' vt play n. 'asceticism' adv spcege sig spcege sit k<Jd inh vr. mortify the flesh/the body mortify the flesh/the body do se/vspcendende se/vspcending spcende spcende a se/vspcerrende spcerre a adj.(= selflilspcendende) 'self-tightening' n. 'self-tightening' vt tighten vi. be tight do adj. 'self-closing, self-locking ' vt close, block (do) (adv) selvstabilitet n. '(flying) inherent stability' stabi/isere vt stabilize a ~ se/vstagende stage adj '(naut) selt~supporting (fx mast)' stage sigfrem vt pole, punt (fx a boat) pole, punt a selvstarter a se/vsteril {adj ??: starte starte steril ?? n. 'self-starter, starting engine' vi. start vt start intr adj. '(bot) selt~sterile' adj. sterile do ?? adv (compare selvlukkende. selvabnende) INHERENTLY REFLEXIVE?? {=the engine starts itself) {=the engine starts by itself): ?? ?? 465 selvstudium studere a selvstyre selvstyrende selvstyreparti selvstyrer selvstyret selvstyring a a styre styre s1g n. 'self-tuition, private study' vt. study adv n. 'self-government, autonomy' adj. 'selt~governing, autonomous' n. 'autonomist party' n. '(naut) automatic steering' adj 'automatically controlled' n. '(naut) automatic steering' vt. govern vr. control oneself do adv {govern themselves} (govern the country by themselves} selvstamdig adj. 'independent, self-supporting,(! it.) selt~standing' selvstcendiggere v. 'make independent' selvstcendighed n. 'independence' selvstcendighedsfelelse n. 'love of independence· selvstcendighedstrang n. 'love of independence' stii vi. stand intr selvsuggestion n. 'autosuggestion' suggestion n. adv suggerere vt. suggestionize; hypnotize a ~ selvsupp/erende selvsupplering supplere adj. ·self-elective' n. 'co-option· vt. supplement; 2 replace, bestyrelse der- sig selv committee with power to add to its numbers/with power to co-opt do a selvsvikkende adj '-melle self-regulating windmill' selvsvikning n. 'self-regulation' { ~ 0 no verbal or nominal base. Maybe the verb svikke existed before but is now lost, has become obsolete} selvsvingende selvsvinger svinge adj. 'a stable multivibrator' n. '(radio) self-oscillator' vi. swing intr adv selvsymmetrisk adJ. '(math) symmetric( a!)' symmetrisk adj. symmetrical { 0 ? no verbal base} selvsyn se syn a selvsiiet selvsaning sa a selvtagen tage {??} selvti/bedelse tilbede a n. 'personal inspection (or experience)' vt. see n. sight; inspection do adj 'self-sown· n. 'self-sowing, selt~seeding' vt. sow (do) (adv) 'spontaneously' adj 'selt~assumed' vt. take n. 'selt~worship' vt. worship do { cf. selvklog, etc.) ?? ?? 466 se/vti/freds adj. 'selt~satisfied' se/vtilfredshed n. 'self-satisfaction, (self-)complacency, smugness' tilfreds (med) adj. satisfied with {adj. no verbal basel do se/vtil/id se/vtillidsfuld tillid { n. no verbal basel n. '(self-)confidence' adj '(selt~ )confident' n. confidence, trust 7? se/vti/spamdende adj. 'self-tightening, self-adjusting' se/vtilspcending n. 'automatic feed' (lilspcende) vt. tighten (up) a ~ selvti/strcekkel(ghed n. 'self-sufficiency' tilstrcekkelig adj {??) ( cf selvklog. selvglad, etc.) se/vtonende tone a se lv transpor ta be I adj. '(photograph.) -papir selt~toning paper' tone (photogr.); tone (fx a wall) (do) (adv) adj. 'selt~propelled' transportere vt. transport do a se/vtredje se/vtugt a tredje tugte adv. 'with the two others' adj. third n. 'selt~discipline' vt. do se/vtcegt n. 'Uur) selt~help' tregtedag court day, session (the noun tcegt has either never existed or has been lost/is now obsolete) { ?? no verbal basel adv ( 77 ) se/vtcendende se/vtcending tcende tcende a se/vtcenkning tcenke a adj. 'seJt:igniting' n. 'self-ignition' vt. ignite vi. ignite intr do n. 'independent thought' vi. think se/vudfoldelse n. 'self-realization' udfo/de vt. udfolde sig/folde sig ud vr. a ~ adv intr se/vudlcegger udlcegge n. '(typ) tlyer, automatic delivery' vt. lay out (cables) a do adv se/vud/oselig se/vud/oser udlose adj. '-ti/spcending (tech) disengaging feed mechanism' n. '(photo) self-timing release, self-timer' vt. release, a do ?? ?? 467 selvudslette/se n. 'selt:effacement' selvudslettende adj. 'seif:effacing' uds!ette vt. eradicate, extinguish a do se/vudvikling udvikle udvikle sig n n. 'self-development' vt. develop sth develop oneself do (adv) 'by itselt/without help' 468 selvumyndiggorelse n. '(jur) [declaration that one is incapable of administering one's own affairs]' umyndiggore vt. have sby declared incapable of managing his own affairs a ~ selvvalg selvvalgt n. '(tlt) direct distance calling; co-option' adj. 'seif:elected' va!lge!valgte!valgt vt. choose. elect, a selvvanding selvvandingsanla!g van de a selvvarme varme varme selvvending vende vende (a) selvvirkende selvvirksom selvvirksomhed virke selvvokset vokse selvvoldt volde a do n. 'automatic sluicing' n. 'automatic sluicing plant' vt. water (fx the flowers), irrigate n. 'selt:generated heat' n. warmth, heat vt. heat, warm sth up adv 'spontaneously' n. 'spontaneous turning (of the foetus)' vi. turn vt. turn (intr) (do) adj. 'seJt:acting, automatic' adj. 'active· n. 'self-activity' (adv) vi. work (fx the radio works). function, intr. adv adj. 'of spontaneous growth' vi. grow mtr adJ. ' = selvforskyldt' vt. cause adv adv selvvundet adj. ·selvvundne erfaringer selt:experience' vindeivandt!vundet vt. win; acquire a selvvurdering vurdere a se/wJdela!ggelse Gdela!gge a n. 'self-estimation' vt. est1mate do n. 'self-destruction· vt. destroy do adv adv selvabnende adj. '(tech) self-opening' (comp selv/ukkende) abne vt. open a do {the irrigation system is not irrigating itself, but rather irrigating the tlelds by itselt/automatically} Appendix II Intensified and unintensified pronouns and reflexives in Danish Table I. Unintensified pronouns and reflexives. Nominative Acc./Dat. I st, sing. Jeg Mig 2" 0 , sing Du Dig -informal 2"a, sing. De Dem -formal 3ra, sing. Han Ham -masculine 3ra, sing. Hun Hende -feminine 3ra, sing. Den Den - common gender 3ra, sing. Det Det -neuter I 5 \ plur. Vi Os 2"a, plur. I Jer -informal 2"a, plur. De Dem -formal 3ra, plur. De Dem 469 Reflexive Mig Dig Dem Sig Sig Sig Sig Os Jer Dem Sig 470 Table 2. Intensified pronouns and reflexives. Nominative Acc./Dat. Reflexive I st, sing. jeg selv mig selv mig selv 2nd, sing du selv dig selv dig selv -informal 2no, sing. De selv Dem selv Dem selv -formal 3rd' sing. han selv ham selv Sig selv -masculine 3n\ sing. hun selv hende selv Sig selv -feminine 3rd, sing. den selv den selv Sig selv - common gender 3rct, sing. ?det selv ?det selv Sig selv -neuter I 5 \ plur. vi selv os selv os selv 2nd, plur. I selv jer selv Jer selv -informal 2nd, plur. De selv Dem selv Dem selv -formal 3n\ plur. de selv dem selv Sig selv 471 Table 3. Unintensified possessive pronouns and reflexives Common gender Neuter Plural 1 st · , smg. Min mit mine 2nd, sing. Din dit dine -informal 2 nd · , smg. De res De res De res -formal 3 rd · , smg. Pron: hans Pron: hans Pron: hans -masculine Reflex: sin Reflex: sit Reflex: sine 3 rd · , smg. Pron: hen des Pron: hendes Pron: hendes -feminine Reflex: sin Reflex: sit Reflex: sine 3 ra · , smg. Pron: dens Pron: dens Pron: dens - common gender Reflex: sin Reflex: sit Reflex: sine 3 ra · , smg. Pron: dets Pron: dets Pron: dets -neuter Reflex: sin Reflex: sit Reflex: sine 1 5 \ plur. vores (vor) vores (vorl) vores (vore) 2na, plur. Jeres jeres jeres -informal 2nd, plur. De res De res De res -formal 3rd, p1ur. De res de res de res 472 Table 4. Intensified possessive pronouns and reflexives Common gender Neuter Plural 1 st · , smg. min egen mit eget mine egne 2nd, sing. din egen dit eget dine egne -informal 2 nd · , smg. Deres egen Deres eget Deres egne -formal 3 rd · , smg. Pron: hans egen Pron: hans eget Pron: hans egne -masculine Reflex: sin egen Reflex: sit eget Reflex: sine egne 3ra, sing. Pron: hendes Pron: hendes eget Pron: hendes -feminine egen Reflex: sit eget egne Reflex: sin egen Reflex: sine egne 3 ra · , smg. Pron: dens egen Pron: dens eget Pron: dens egne - common gender Reflex: sin egen Reflex: sit eget Reflex: sine egne 3 ra · , smg. Pron: dets egen Pron: dets eget Pron: dets egne -neuter Reflex: sin egen Reflex: sit eget Reflex: sine egne 1 5 \ plur. vores (vor) egen vores (vort) eget vores (vore) egne 2nd, plur. jeres egen jeres eget jeres egne -informal 2n°, plur. Deres egen Deres eget Deres egne -formal 3ra, plur. deres egen deres eget deres egne

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The independence of binding and intensification

PDF

Ryūkyūan language history

PDF

The Ilias Latina: a study of the Latin Iliad, including translation, commentary, and concordance

PDF

Functional and inflectional morphology problems of projection, representation and derivation

PDF

Morphosyntactic feature chains and phonological domains

PDF

Syntactic reanalysis in early English

PDF

Post-verbal phenomena in colloquial Persian syntax

PDF

Form and meaning: negation and question in Chinese

PDF

Seeing ghosts: readings on the effectivity of phantasms

PDF

Assertion and modality

PDF

The nine songs: a reexamination of shamanism in ancient China

PDF

Maryknoll in Manchuria, 1927-1947: a study of accommodation and adaptation

PDF

Modernization and standardization in Somali press writing

PDF

The Arab theater: A quest for unity and identity

PDF

Removing the body: representations of animal skins on Greek vases

PDF

Taxpayer perceptions of complexity and the effect of complexity on reporting positions

PDF

The effects of race, gender, and family background on children's educational attainment: contemporary patterns and historical change

PDF

The development of organization and disorganization in the social life of a rapidly growing working-class suburb within a metropolitan district

PDF

Religiosity and its relationship to self-concept, locus of control, and dogmatism

PDF

The structural geology and ages of deformation of a portion of the southwest Columbia Plateau, Washington and Oregon

Asset Metadata

Creator Bergeton, Uffe (author)

Core Title The independence of binding and intensification

Degree Doctor of Philosophy

Degree Program Linguistics

Defense Date 05/01/2004

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

Tag OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Digitized by Interlibrary Loan Department (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-493403

Unique identifier UC11298298

Identifier etd-BergetonVolumes1and2-.pdf (filename),usctheses-c3-493403 (legacy record id)

Legacy Identifier etd-BergetonVolumes1and2-.pdf

Dmrecord 493403

Document Type Dissertation

Format application/pdf (imt)

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