University of Southern California Dissertations and Theses

"The unaccusative trap": L2 acquisition of English intransitive verbs.

(USC Thesis Other)

"The unaccusative trap": L2 acquisition of English intransitive verbs.

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the origmal or copy submitted. Thus, some thesis and dissertation copies are in typewriter fiice, while others may be from any type o f computer printer. The quality' o f this reproduction is dependent upon the quality o f the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photogrsq>hed in one exposure and is included in reduced form at the back o f the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Ifigher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations spearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Com patQr 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. "THE UNACCUSATIVE TRAP": L2 ACQUISITION OF ENGLISH INTRANSITIVE VERBS by Hiroyuki Oshita 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) August 1997 Copyright 1997 Hiroyuki Oshita Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9816056 Copyright 1997 by Oshita, Hiroyuki All rights reserved. UMI Microform 9816056 Copyright 1998, hy UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeh Road Ann Arbor, MI 48103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. U N IV ERSITY OF SOUTHERN C A LIFO R N IA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 9 0 0 0 7 This dissertation, written by E i r o y u k i O s h i t a under the direction of h .^ .^....... Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY CTVÎ ............................. Dean of Graduate Studies Date DISSERTATION COMMITTEE Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 This thesis is dedicated to Jean K. Glasser Karen E. Schmitt and Maya E. Oshita. Each taught me an important lesson in life. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. lU Acknowledgm ents My graduate student days at USC and life in Los Angeles have been supported and enriched by the firiendship and generosity of a great number of people. First of all, I would like to thank the members of my dissertation guidance committee for their continuing support. Bill Rutherford, the committee chair, has been a great teacher, warm friend, understanding mentor and enthusiastic supporter of my work. I owe so much to him both academically and personally that no words can adequately express my gratitude. My most sincere thanks go to him. Masha Polinsky has generously shared her time to discuss the topics of the dissertation and provided me with words of encouragement during the writing. Maria Luisa Zubizarreta’ s continuing interest in my dissertation work carried me through difficult times. Mario Saltarelli responded to my last-minute requests for help, even with his busy schedule. My thanks also extend to Maryellen MacDonald, a member of my qualifying paper guidance committee, who, at a critical moment, encouraged me to pursue topics that really interested me. This dissertation would not be possible without the friendship and assistance of Joe Allen. It was during our informal series of reading sessions in campus cafeterias that my interest in unaccusativity developed, and the corpus- based research included in this dissertation would not have been possible without his generous assistance and stimulating comments. I was also fortunate to receive enormous support from many individuals in the process of planning and implementing the grammaticality judgment test for this work. Gloria Montebruno helped me by translating the test materials into Italian. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV Hiroshi Aoyagi, Mihoko Zushi and their colleagues in Nagoya, Japan conducted the test at Nanzan University and at Aichi Prefectural University. Paola Giunchi and her colleagues carried out the test at Université degli Studi di Roma "La Sapienza," providing critical data on Italian subjects. John Holland, Shadi Ganjavi and Kristi Keller made it possible for me to collect data from their freshman writing classes at USC. I would like to thank all the students who participated in the study for their cooperation. I also owe a great deal to Dr. Harold Urman of Vital Research for help with the statistical analyses of the data. Makiko Hirakawa, Richard Ingham, Eric Kellerman, Robert Kluender, Steven Matthews, Virginia Yip, Helmut Zobl and S.-Y. Kuroda each read a part of my dissertation and gave me important suggestions. Alan Hyun-Oak Kim, Jong- Bok Kim, Keith and Becky Pharis, and Philip Hubbard helped me with their responses to my E-mail inquiries. Karin Stroms wold, Beth Levin, Yaehoon Yeon, Soowon Kim, Luigi Burzio, Janet Randall and Christina Tortora kindly made their work available to me. I am grateful to all of them. I would also like to thank the people at USC's American Language Institute, which provided me with an opportunity for employment in a very supportive and nurturing environment. Among them, Dave Eskey, Lisa Patriquin-Esmaili, Paul Bmthiaux, David Bycina, Daryl Kinney, Cheryl Kraft, Linda Neumaier and Phil Spradling especially deserve my thanks. I am also thankful to my friends and colleagues in East Asian Languages and Cultures: Peter Nosco, Gianina Carver, Hazel Bates, Sean O’ Connell, Akemi Kagawa, Irene Katsui, Yuki Matsuda, Keiko Miyagawa, Aki Mizuno, Chie Paik, Chisato Shuyama, Maki Watanabe, and Shin Watanabe. It was great fun to work with them. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. My thanks also go to the people in Linguistics for their support, encouragement and friendship: Linda Culver, Kathy Stubaus, Don Bui, Jessica Hamilton, Jose Camacho, Fred Field, Cynthia Hagstrom, Hajime Hoji, Miao-Ling Hsieh, Sechang Lee, Karine Megerdoomian, Hong-Keun Park, Magdalena Romera and Liliana Sanchez— to name but a few. Laura Reiter has been an indispensable person for me at USC. Without her, like many students in our department, I would not have come to USC, would not have survived, and would not be writing these acknowledgments. My heartfelt thanks go to her for her unfailing commitment to the welfare of the department's students during their best and worst times as well as hers. Kaoru Horie, Dwight Atkinson and Yumiko Shiotani gave me a lot of advice and encouragement during the first couple of years of my stay at USC when I needed it most. On a more personal note, Jean Glasser deserves special gratitude for always supporting me in virtually every aspect of my life over the last twenty years. Without her, I would not be here. Alain Martinossi and Ed Liegey have given me and my family much joy and warmth throughout our life in Los Angeles. Kanzo and Sayuri Takemori, with their daughter, Shiori, taught me what generosity and kindness really mean. Katsuaki Takeda and his family have always been my friends and supporters since we met in Christchurch, New Zealand. Namie Saeki's free- spirited attitude to life has been an inspiration to me. I thank them all for their unchanging "long-distance friendship." I am grateful to the Department of Linguistics and the Graduate and Professional Student Council of USC for providing me with conference travel Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VI funds to attend academic conferences both in and outside the United States. During the last year of my study at USC, I was financially supported with a Dean's Fellowship from Letters, Arts and Sciences, USC and with a graduate fellowship from the Josephine de Kârmân Fellowship Trust. I would like to thank both for their generous support. I am also grateful to Longman Publishing Company, which has granted me, through Bill Rutherford, the use of the Longman Learners Corpus for the research that has become an integral part of this dissertation. Finally, my very special thanks go to my family. I am grateful to my father, Kazuyoshi, and mothers, Yuriko and Marguerite, for their love and trust. My wife, Karen, has more than once saved me from falling apart by reminding me of the importance of keeping a balanced perspective on life. Our daughter, Maya, who came into our life last September, has opened my eyes to so many things that I never expected to see. Without their unconditional love and constant prodding— intentional or otherwise, this work would never have been completed. Karen and Maya, WE'VE MADE IT! Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. v u Table o f Contents 1. Introduction...................................................................................................1 1.1. Intransitivity and the Unaccusative Hypothesis.................................1 1.2. Framework of Analysis.................................................................... 5 1.3. Goals and Organization...................................................................... 11 2. Two Classes of In tran sitiv es.................................................................. 15 2.1. English............................................................................................... 16 2.1.1. Inchoatives and Causatives.................................................16 2.1.2. Nominalization with -er...................................................... 19 2.1.3. Passive and Perfective Adjectives......................................23 2.1.4. (Compound) Adjectives with -ing......................................27 2.1.5. A rb itr^ Pronoun they......................................................31 2.1.6. Transitive-based Resultative Construction......................... 33 2.1.7. Unergative-based Resultative Constructions..................... 39 2.1.8. "X's Way" and Cognate Object Constructions.................. 44 2.2. Italian................................................................................................. 48 2.2.1. Inchoatives and Causatives................................................ 48 2.2.2. Passive and Perfective Nominals.......................................52 2.2.3. Afe-Cliticization..................................................................54 2.2.4. Auxiliary Selection............................................................ 60 2.2.5. Linear Word Order............................................................ 64 2.2.6. Past Participial Adjectival Clause....................................... 67 2.2.7. Past Participial Adverbial Clause.......................................71 2.3. Spanish...............................................................................................72 2.3.1. Nominalization Based on Present Participles..................... 72 2.3.2. Bare NP Postverbal Subject...............................................74 2.3.3. Past Participial Adjectival Clause.......................................75 2.3.4. Past Participial Adverbial Clause.......................................79 2.3.5. Arbitrary Null Plural Pronominal....................................... 80 2.3.6. Effects on Sentence Processing..........................................87 2.4. Japanese............................................................................................. 88 2.4.1. Inchoatives and Causatives................................................ 89 2.4.2. Construction with -te-iru.................................................... 92 2.4.3. NV Compounds................................................................ 95 2.4.4. W Compounding............................................................. 98 2.4.5. Case Marker Drop..............................................................101 2.4.6. "Floating" Numeral Quantifier............................................103 2.4.7. Quantifier takusan..............................................................108 2.4.8. PRO....................................................................................109 2.4.9. Resultative Construction.....................................................110 2.4.10. Indirect Passive.................................................................114 2.4.11. Passivized Causative.........................................................117 2.5. Korean................................................................................................ 120 2.5.1. Cognate Object Construction and "Nominal + ha" Construction...................................................................... 120 2.5.2. Construction with -e-iss-ta.................................................121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VUl 2.5.3. W Compounding.............................................................. 124 2.5.4. Case Marker Drop firom a Numeral Quantifier...................126 2.5.5. Resultative Constructions................................................... 126 3. Unaccusative and Inchoative Verbs in L2 English...........................132 3.1. Structural Variations.......................................................................... 132 3.1.1. "Passivized" Unaccusative-Inchoatives..............................132 3.1.1.1. Burt and Kiparsky 1972 and Richards 1973.......133 3.1.1.2. Hubbard 1983, Hubbard and Hix 1988, and Hubbard 1994..............................................135 3.1.1.3. Zobl 1989........................................................... 137 3.1.1.4. Yip 1989, 1994, 1995........................................ 139 3.1.1.5. Hirakawa 1995................................................... 142 3.1.2. Avoidance of Inchoative Verbs in the NP-V Order............143 3.1.2.1. KeUerman 1978..................................................143 3.1.2.2. Yip 1989, 1994, 1995........................................ 145 3.1.2.3. Hirakawa 1995................................................... 146 3.1.3. Transitivization of Unaccusative Verbs............................. 147 3.1.3.1. Rutherford 1987..................................................147 3.1.3.2. Yip 1989, 1994, 1995........................................ 147 3.1.3.3. Hirakawa 1995................................................... 149 3.1.3.4. Zobl 1989........................................................... 149 3.1.4. Postverbal NP Structure.................................................... 150 3.1.4.1. Zobl 1989........................................................... 150 3.1.4.2. Rutherford 1989..................................................151 3.1.5. Summary............................................................................153 3.2. Unresolved Questions........................................................................154 3.2.1. Range of Structural Variations........................................... 154 3.2.2. Questions Raised by Suggested Accounts..........................156 3.2.2.1. postverbal NP structures.................................... 156 3.2.2.2. transitivization.....................................................157 3.2.2.3. "passivization".................................................... 157 3.2.3. Developmental Variations on X L Structures........................ 165 3.2.4. Summary............................................................................170 4. Corpus-Based R esearch............................................................................172 4.1. Materials.............................................................................................173 4.2. Method.............................................................................................. 173 4.2.1. Selection of Verbs.............................................................. 173 4.2.2. Extraction of Token Sentences........................................... 175 4.2.3. Classification of Token Sentences......................................177 4.3. Results................................................................................................180 4.4. Discussion.........................................................................................183 4.4.1. "Passivization".................................................................. 183 4.4.1.1. LI transfer..........................................................184 4.4.1.2. passive and perfective adjective..........................185 4.4.1.3. lexical transitivization..........................................186 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX 4.4.1.4. identification of the "passive" morphosyntax with the lack of logical subject.................................................................189 4.4.1.5. marMng of syntactic NP-movement................... 189 4.4.2. Postverbal NP Structures................................................... 192 4.4.2.1. arguments for and against null expletives........... 194 4.4.2.2. on the psychological reality of null expletives...........................................................203 4.4.3. NP-V Structure..................................................................209 4.4.3.1. previous accounts...............................................209 4.4.3.2. clues to a new account........................................211 4.5. A New Hypothesis...........................................................................213 4.5.1. Background.......................................................................213 4.5.1.1. structure of natural language grammar................ 215 4.5.1.2. linking rules........................................................215 4.5.2. Stages of Interlanguage Development................................ 219 4.5.2.1. first stage: lexical confusion...............................219 4.5.2.2. second stage: syntactic confusion.......................222 4.5.2.3. third stage: out of confusion...............................224 4.5.3. Reorganization of Interlanguage Grammar........................ 225 4.5.3.1. to the first stage...................................................225 4.5.3.2. to the second stage............................................. 226 4.5.3.3. to the third stage................................................. 231 4.5.3.4. sununary............................................................233 4.5.4. The "Unaccusative Trap" and U-shaped Development 234 5. Grammaticality Judgment T est................................................................238 5.1. Introduction.......................................................................................238 5.1.1. Research Hypotheses........................................................ 239 5.1.1.1. "passivization"....................................................239 5.1.1.2. NP-V word order...............................................243 5.1.1.3. postverbal NP structures.................................... 246 5.1.1.4. transitivization....................................................248 5.1.1.5. raising verbs.......................................................250 5.1.1.6. proficiency levels and U-shaped development.......................................................253 5.2. Method............................................................................................. 253 5.2.1. Subjects..............................................................................253 5.2.2. Materials............................................................................ 256 5.2.3. Procedures..........................................................................259 5.2.4. Scoring and Analyses........................................................ 260 5.3. Results/Discussions........................................................................... 262 5.3.1. "Passivization"................................................................... 262 5.3.2. NP-V Word Order............................................................. 265 5.3.3. Postverbal NP Structures...................................................269 5.3.4. Transitivization................................................................... 275 5.3.5. Raising Verbs.................................................................... 280 5.3.6. Proficiency Levels and U-shaped Development................ 284 5.4. Concluding Remarks........................................................................ 287 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6. The "Unaccusative Trap" and Related Issues.................................... 291 6.1. The "Unaccusative Trap" in L2 Japanese......................................... 293 6.2. Potential Contrasts between LI and L2 Acquisition.........................299 6.2.1. Verbal Classification and Initial Syntactic Representation...................................................................299 6.2.1.1. LI acquisition..................................................... 299 6.2.1.2. L2 acquisition..................................................... 301 6.2.2. Subsequent Development of Word Order...........................303 6.2.2.1. LI acquisition..................................................... 303 6.2.2.2. L2 acquisition..................................................... 305 6.2.3. Status of the "Unaccusative Trap" in LI Acquisition.........306 7. Conclusion..................................................................................................310 R eferences..........................................................................................................315 A ppendices........................................................................................................ 326 Appendix A: Distribution of Sentences Built on Unaccusative Verbs 326 Appendix B: Sentences with Unaccusative Verbs................................... 329 Appendix C: Cloze Test............................................................................335 Appendix D: Grammaticality Judgment Test............................................340 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XI List of Tables 4.1 Overall Structural Distribution of Unaccusative Verbs ...........................180 4.2 Italian/Spanish versus Korean/Japanese..................................................182 5.1.1 Italian Subjects.........................................................................................254 5.1.2 Japanese Subjects.................................................................................... 255 5.1.3 Native English-Speaking Subjects.......................................................... 256 5.2 Cloze Results........................................................................................... 262 5.3.1 Unaccusatives: NP-be+Ven.................................................................. 263 5.3.2 Inchoatives: NP-be+Ven ...................................................................... 263 5.3.3 Unergatives: NP-be-t-Ven...................................................................... 263 5.4.1 Unaccusatives: NP-V.............................................................................266 5.4.2 Inchoatives: NP-V................................................................................. 266 5.4.3 Unergatives: NP-V................................................................................ 266 5.5 Results on Unaccusatives........................................................................ 266 5.6 Results on Inchoatives............................................................................267 5.7.1 Existence/Appearance: there-V-NP........................................................270 5.7.2 Existence/Appearance: it-V-NP.............................................................270 5.7.3 Existence/Appearance: 0- V-NP............................................................. 271 5.8 Results on there- V-NP Structure............................................................271 5.9 Results on/r-V-iVP Structure.................................................................. 272 5.10 Results on 0-V-NP Structure.................................................................. 272 5.11.1 Unaccusatives: NPi-V-NPi.................................................................. 275 5.11.2 Inchoatives: NP1-V-NP2 ...................................................................... 276 5.11.3 Unergatives: NP1-V-NP2 ..................................................................... 276 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XU 5.12 Results on NPi- V-NP2 Structure............................................................276 5.13.1 (Dis)appearance: NPi-V-NPi................................................................278 5.13.2 Directed change: NPi V -N P i................................................................278 5.13.3 Existence: NPi-V-NPi ..........................................................................278 5.14.1 Raising: NP V to inf Ht V that...............................................................281 5.14.2 Raising: NP-be+Ven to in f................................................................... 281 5.14.3 Raising: It be+Ven that.......................................................................... 281 5.15 Results on NP-be+Ven-to infinitive Structure........................................282 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. x m List of Abbreviations for Glosses ACC Accusative CAUS Causative CL CUtic CLF Classifier CPL Copula DAT Dative DEC Declarative GEN Genitive FEM Feminine ICL Impersonal Clitic INCH Inchoative LOC Locative MAS Masculine NEC Negative NOM Nominative NOML Nominalizer NQ Numeral Quantifier PART Participle PASS Passive PAST Past PFV Perfective PL Plural PRES Present Q Question RES Resultative SEP Sentence Final Particle SO Singular Spec Specifier TOP Topic 3PL 3rd Person Plural 3SG 3rd Person Singular Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XIV Abstract This study investigates the ways in which the division between unergatives and unaccusative-inchcatives within the general class of intransitive verbs affects the acquisition of English by adult nonnative speakers. Special attention is given to potential effects of learners' mother tongues, structural characteristics of interlanguage English, and developmental sequences. Primary data come from published studies in the L2 literature, a research corpus culled from the Longman Learners Corpus, and a grammaticality judgment test. The learners' Lis include, but are not limited to, Italian, Spanish, Japanese and Korean. Previous research identified some non-target stmctures and phenomena such as the "passivization" and transitivization of unaccusatives and learners' reluctance to accept inchoatives in the canonical NP-V word order. The corpus- based research has confirmed this general pattern and also found that unaccusative verbs are used in non-target postverbal NP structures such as it-V-NP mostly by native speakers of Italian and Spanish but not by those of Japanese and Korean. In order to account for these observations, I propose a three-stage developmental model, termed the "unaccusative trap" hypothesis, in which each stage is characterized in terms of the status of the interlanguage lexicon and syntax. Specifically, I argue (i) learners, irrespective of their mother tongues, initially misanalyze the lexico-syntactic properties of unaccusative-inchoatives as those of unergatives, and (ii) the non-target structures and phenomena emerge when relatively advanced learners correctly differentiate the two subclasses of intransitive verbs lexically but resort to non-target syntactic operations, which may be influenced by their Lis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XV A number of predictions made by the "unaccusative trap" hypothesis are further tested against data obtained fix>m the grammaticality judgment test conducted on native Italian and Japanese speakers. The results confirm that unaccusative- inchoatives, in contrast to unergatives, present unique learning difficulties to L2 learners of English. This study strongly suggests that the "unaccusative trap" is also present in other adult L2 acquisition contexts and that the existence and lack of "unaccusative trap" may indicate a significant fundamental difference between adult L2 and child LI acquisition. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1. Introduction One introductory textbook in the field of second language acquisition states fundamental goals in research on the L2 lexicon as follows: The major task of second language lexical research is to discover what second language learners know about the lexicon of the second language, how they learn it, and why this particular path of development is followed. (Gass and Selinker 1994:272) As widely acknowledged (e.g. Gass and Schachter 1989:201, Gass and Selinker 1994:270, Juffs 1996:1-2), however, the grammatical analysis of acquisition of the L2 lexicon traditionally has attracted relatively little research interest. In this thesis, I try to answer the call of Gass and Selinker by investigating how English verbs, intransitives in particular, are acquired by normative speakers. The choice of intransitive verbs as a topic of an acquisition study may at first appear rather misguided. After all, intransitive verbs require only one obhgatory grammatical element, namely a subject, and typically appear in the canonical NP-V word order in English. One may assume, therefore, that the acquisition of this class of verbs is unproblematic and will present no major difficulties to L2 learners. To prove that nothing is farther from the truth is the ultimate goal of the current thesis. 1.1. iNTRANSmVTTY AND THE UNACCUSATIVE HYPOTHESIS Traditionally, intransitive verbs have been defined in contrast to transitive verbs. The former do not require a grammatical object, either nominal or clausal, while the latter do. As Rutherford (1997) notes, however, when defined this way, verbs that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. exhibit surface characteristics of intransitivity are actually a mixed bag. Consider, for instance, the italicized verbs in (1). (1) a. Many people b . John reads whenever he has time. c. This kind of rice cooks easily. d. The ice me/ted. e . A few guests arrived at the hotel. The verbs in (la-e) all appear with a preceding noun phrase that serves as a sentential subject. Differences among them, however, become apparent when the sentences in (2) are taken into consideration as well. (2) a. *The comedian laughed many people. b . John reads books whenever he has free time. c. John often cooks this kind of rice. d . The sun melted the ice. e. *The taxi arrived a few guests at the hotel. In (la), the subject many people is responsible (or accountable) for the event denoted by the verb laugh because it is the subject that causes this event. The subclass of intransitives like laugh that take immediate causers as their subjects such as smile, walk, swim, etc. are called unergative verbs (Levin and Rappaport Hovav 1995). As shown in (2a), unergative verbs do not normally take a grammatical object.* * This does not mean that prototypical unergative verbs never take an object. When they do, however, their objects still maintain some control over their action: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The verb read in (lb) lacks an overt grammatical object but is still interpreted as if it had one. For this reason, it is called pseudo-transitive. Other prototypical transitive verbs such as eat, write and drive also appear as pseudo transitive verbs. The missing objects in pseudo-transitive sentences are usually interpreted as typical objects that the verbs would take as transitives. For instance, they may be books and novels for read and food and meals for eat. (Ic) is a middle construction, whose subject is a noun phrase that would normally appear as an object of the verb. The syntactically unexpressed causer of the event is implied and interpreted as a generic agent. Accordingly, the middle construction conveys an inherent property reading about its subject. Transitive verbs that take an "affected" object may be used in this construction. The verb melt in (Id) is an inchoative verb, which alternates with its causative use in the same form as in (2d). For this reason, inchoatives are sometimes called anti-causatives (e.g. Zubizarreta 1987). As in the middle construction, the subject of an inchoative verb is a noun phrase that would normally appear as an object in its causative use. The inchoative sentence differs from the middle construction, however, in that the ultimate external cause of the event (e.g. heat, time, gravity, etc.) is not only syntactically unexpressed but also semantically not implied. Neither does it have a property reading as the middle sentence does. Verbs such as break, open, close, move, stop, etc. also exhibit the causative- inchoative alternation. (i) a. Tony walked the dog down the street (cf. The dog walked down the street.) b. Tony jumped the horse over the fence. (cf. The horse jumped over the fence.) (Levin and Rappaport Hovav 1992:263, (30)) Contrast these examples with causative-inchoative pairs such as (2d) and (Id) above. Also see section 2.1.8 on the cognate object construction based on unergative verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The verb arrive in (le) is called an unaccusative verb.^ As indicated by (2e), unaccusatives do not have causative use. Linguistically speaking, however, they share both semantic and syntactic properties of inchoative verbs. Semantically, for example, both lack an immediate internal causer of the event and in this respect, contrast with unergatives. Other examples of unaccusative verbs in English are appear, disappear, exist, happen, remain, and so forth.^ The idea that prototypical intransitive verbs fall into two major subclasses, namely unaccusative-inchoatives and unergatives, originates in Perlmutter (1978), who proposed, within the framework of Relational Grammar, that the surface subject of an unaccusative-inchoative verb is an object at the initial syntactic representation while the subject of an unergative verb remains a subject throughout derivation. Furthermore, Perlmutter argued that the distinction between the two subclasses of intransitive verbs is semantically determinable. His insightful proposal known as the Unaccusative Hypothesis'^ has generated enormous research interest and has been tested in numerous languages. ^ Clarification of the terminology may be in order. In the literature, verbs such as those in (Id) and (le) are often referred to as ergative verbs (e.g. Burzio 1986). Since the term "ergative" does not distinguish between the verbs that have transitive use and those that do not, some researchers (e.g. Yip 1995) call them "paired" and "unpaired" ergatives. The term "paired ergative," however, simply refers to the class of alternating verbs and fails to specify the transitive and intransitive use o f them. In this work, therefore, the term "unaccusative" is used for unpaired ergatives (Belletti 1986, Hale and Keyser 1986, 1987) and the term "ergative" is reserved only for paired ergative verbs. In further reference to the transitive and intransitive use o f the ergatives, I use the terms "causative" and "inchoative" respectively. Consequently, the alternation exhibited by ergatives as in (Id) and (2d) is referred to as ergative alternation or causative-inchoative alternation. ^ For a summary o f general diagnostics of these verb classes, see Rutherford 1997. ^ I will distinguish unaccusativity from unaccusative verbs. The former refers to the syntactic phenomenon in which a lexical head which lacks a thematic external argument fails to assign Accusative Case to an NP that it governs. Accordingly, unaccusativity is manifested not only by unaccusatives and inchoatives but also by passives, raising verbs, raising adjectives, some types of reflexive verbs and even some nominals (see Cinque 1990; Tsujimura 1990a, 1990b, Kageyama 1993, Haegeman 1994). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. With respect to the proper characterization of the two subclasses of intransitive verbs, there has been a lively debate among linguists. For instance, researchers such as Van Valin (1987, 1990) and Kishimoto (1996) have advocated, within the framework of Role and Reference Grammar (e.g. Van Valin 1993), a primarily semantic account of the phenomena related to the distinction.^ In contrast, Rosen (1984), for example, has proposed a primarily syntactic account, noting difficulties in finding crosslinguistically valid semantic characteristics of unaccusative verbs. In this work, I follow a middle road, adopting the view that the difference between the two kinds of intransitive verbs is semantically determined but syntactically represented (Perlmutter 1978, Levin and Rappaport 1989, Levin and Rappaport Hovav 1992, 1995). Although the theoretical debate on the nature of unaccusativity is not a primary concern of this work, I believe that various kinds of L2 phenomena dealt with in the following chapters clearly support the view that the distinction between the two classes of intransitive verbs is both semantic and syntactic. 1.2. F r a m e w o r k o f a n a l y s is The analytical framework that I adopt for discussions in the present work is the Government-Binding (GB) Theory of Chomsky 1981, which was further elaborated by subsequent developments on lexical theory in the work of Zubizarreta (1987), Jackendoff (1990), Grimshaw (1990), Levin and Rappaport Hovav (1995) and others.® Perlmutter's Unaccusative Hypothesis was adopted, reinterpreted and ® For strong arguments against the primarily semantic approach, see Legendre 1992. ® For a good introduction to GB theory, see Haegeman 1991, 1994. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. extended to wider linguistic phenomena within the GB framework by Burzio ( 1986). The overall outline in (3) illustrates a generally accepted GB framework. (3) Structure of Grammar^ Lexicon: LSp. (Lexico-Semantic Representation) <— Linking Rules A-Structure (Lexico-Syntactic Representation) I I <— X'-theory, theta criterion, linearity parameters Syntax: D-Structure < Move Alpha (overt movement) S-Structure (Phonological Form) (Logical Form) As shown in (3), grammar consists of two major components: the lexicon and syntax. The lexicon contains, among other things, the semantic and syntactic information of lexical items. Lexical Semantic Representation (LSR) is where grammatically relevant semantic information is represented and Argument Structure (A-Structure) contains the syntactically relevant information extracted from LSR. The two levels in the lexicon are related by Linking Rules. The syntactic component of grammar also consists of two representational levels: D-Structure and S-Structure.* D-Structure is the initial syntactic representation to which arguments of A-Structure are projected. S-Structure representation is obtained by movement of constituents from one syntactic position to another. ^ Details irrelevant to the topics discussed in the current thesis are omitted in this scheme. * Phonological Form and Logical Form, enclosed in parentheses, do not concern us in this work. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Let us see how the verb classes that are central to the topics of this work are represented in this framework. Rrst, the prototypical unergative verb laugh has the following representations. (4) The Unergative Verb (e.g. laugh) LSR: [\xLAUGH\ CAUSE_[_________ ]] X <-----Immediate Cause Linking Rule A-Structure: (x (0)) I I <---- X'-theory, theta-criterion, linearity parameters D-Structure: jjp [yp NP[v V]]] <-----Move Alpha S-Structure: [ % ? NPi [yp ti [y V]]] In (4), the entity that laughs is indicated by x contained in the activity phase of the eventuality described by the verb. Since laughing normally does not cause a state or change of a state, the second (change of) state phase is shown blank. The lexico-semantic information contained in LSR is connected to A-Structure by one of the four linking rules proposed by Levin and Rappaport Hovav (1995). (5) Immediate Cause Linking Rule The argument of a verb that denotes the immediate cause of the eventuality described by that verb is its external argument. (Levin and Rappaport Hovav 1995:135) The Immediate Cause Linking Rule projects x in LSR to the external, rather than internal, argument position in A-Structure. In (4), the external and internal argument positions are indicated by the single and double closures of parentheses respectively. The external argument (x) is projected as an NP in the Spec of VP at D-Structure in accordance with X'-theory, theta-criterion and the values of the linearity parameters, particularly, that of the specifier parameter (Radford 1990:59- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 60) y This NP does not receive Case in situ and moves to the Spec of IP where it can be assigned Nominative Case by Infl. Otherwise, the representation results in ungrammaticali^, violating Case Filter which mandates that every overt NP be assigned abstract Case (Haegeman 1994:193). In order to project arguments in the A-Structure of the prototypical causative verb break in (7), we need a second linking rule that determines the syntactic role of an entity in the (change of) state phase of LSR. The Directed Change Linking Rule in (6) does this job. (6) Directed Change Linking Rule The argument of a verb that corresponds to the entity undergoing the directed change described by that verb is its internal argument. (Levin and Rappaport Hovav 1995:146) As shown in (7), the Directed Change Linking Rule makes y the internal argument ((y)) while X becomes the external argument (x) according to the Immediate Cause Linking Rule. Both arguments are projected as an NP in syntax: NPi in the Spec of VP corresponds to (x) and NP2 within V to ((y)). While NPi moves to the Spec of IP at S-Structure to receive Nominative Case, NP2 is assigned Accusative Case in situ by the verb. ^ Following the VP-internal subject hypothesis (e.g. Fukui and Speas 1986, Kuroda 1988), I assume in this work that a verb's external argument is projected within its maximal projection. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (7) The Causative Verb (e.g. break) LSR: [[x DO-SOMETHING] CAUSE {y BECOME BROKEN]] X <---- Immediate Cause and Directed Change Linking Rules A-Structure: (x (y)) I I <----X’ -theory, theta-criterion, linearity parameters D-Structure: [ip [yp NPi[v’V NP2]]] J, <---- Move Alpha S-Stnicture: [n * NPii [yp ti [y V NP2]]] The inchoatiye verb break differs from its causative counterpart in that there is no entity specified in the activity phase of its LSR representation.The entity y in the (change of) state phase of LSR, however, becomes the internal argument ((y)) by the Directed Change Linking Rule just as in (7). (8) The Inchoative Verb (e.g. break) LSR: [[0 DO-SOMETHING] CAUSE [y BECOME BROKEN]] X <-----Directed Change Linking Rule A-Structure: (0 (y)) I I <----X'-theory, theta-criterion, linearity parameters D-Structure: [n > [yp [y V NP]]] X <-----Move Alpha S-Structure: Q p N?i [yp [y V ti]]] The internal argument ((y)) is projected as an NP within V at D-Structure. Unlike NP2 in (7), howeyer, this NP fails to receive Accusative Case in situ due to a general linguistic condition known as Burzio's Generalization. (9) Burzio's Generalization i. A verb which lacks an external argument fails to assign accusative case. Here I disagree with Levin and Rappaport Hovav (1995:108-109) who assume that such an entity is present in LSR but is prevented from undergoing a linking operation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 ii. A verb which fails to assign accusative case fails to theta-mark an external argument. (Burzio 1986 as summarized in Haegeman 1994:321) Clearly (9i) matches the A-Structure representation in (8). Consequently, the NP moves to the Spec of IP at S-Structure and receives Nominative Case from Infl, thus avoiding violation of Case Filter. Finally, the prototypical unaccusative verb exist has the following representations. (10) The Unaccusative Verb (e.g. exist) LSR:________[[_________] CAUSE [y BE (Floe z)]] I <----- Existence Linking Rule A-Structure: (0 (y)) I I <-----X'-theory, theta-criterion, linearity parameters D-Structure: [n > [yp [v V NP]]] I <----- Move Alpha S-Structure: Q p NPi [yp [y V ti]]] As shown in (10), the actiyity phase of the LSR of exist is blank since activity is irrelevant to its meaning. Within the (change of) state phase, however, the existing entity y and its spatio-temporal location z are contained. • • The entity y is connected to the internal argument position by the Existence Linking Rule. (11) Existence Linking Rule The argument of a verb whose existence is asserted or denied is its direct internal argument. (Levin and Rappaport Hovav 1995:153) ^ * We ignore the syntactic realization of z since it is not relevant to the topics discussed in this thesis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 The internal argument ((y)) is projected as an NP within V at D-Structure. This NP, however, is Caseless because the verb exist which lacks an external argument also fails to assign Accusative Case to its object Consequently, the NP moves to the Spec of IP and receives Nominative Case there, avoiding violation of Case Filter. Notice that both inchoative verbs like break and unaccusative verbs such as exist fall into the group of verbs that match the condition of Bruzio's Generalization. This is because both kinds of verb lack an external argument in their A-Structure representation. As a result, despite their different LSR representations, the two kinds of verb share the identical D-Structure and S- Structure representations as shown in (8 ) and ( 10). It is in this sense that unaccusative-inchoative verbs are said to form one major subclass of intransitive verbs and contrast with unergatives such as laugh. 1.3. G o a l s a n d O r g a n iz a t io n Due to many intricate matters that are involved in it, the Unaccusative Hypothesis has generated a large number of theoretical studies over the years. Some of these important linguisitc issues are, for example, the relation between lexical semantics and lexical syntax, representations of D-Structure and S-Structure, the causal relation between Case requirement and syntactic movement, and so forth. Although the actual amount of research done in the field of second language acquisition has been far more limited, the Unaccusative Hypothesis has also provided L2 researchers with fertile grounds on which to investigate many interesting questions (e.g. Zobl 1989, Yip 1989, 1994, 1995, Hirakawa 1995). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 In particular, Sorace (1993a, 1993b, 1995) has demonstrated the relevance of the division between unaccusative-inchoatives and unergatives to second language acquisition by investigating the way near-native speakers internalize syntactic ramifications of this verb class. For instance, focusing on acquisition of auxiliary selection (see section 2.2.4), she has found that speakers of LI English and those of LI French show a qualitative difference. The contrast, termed as "incompleteness" and "divergence" by her, is that English speakers exhibit indeterminate judgments on Italian sentences that involve auxiliary selection while French speakers show determinate but sometimes nontarget-like judgments. Sorace has also shown that as far as the unaccusative-inchoative verb class is concerned the acquisition of L2 French is harder for native speakers of Italian than the acquisition of L2 Italian is for native speakers of French. Thus, Sorace has clearly demonstrated that the acquisition of unaccusative-inchoative verbs in L2 contexts is influenced both by the semantically characterizable universal unaccusative hierarchy and by the way the division between unergatives and unaccusative-inchoatives is drawn along the hierarchy in the learner's LI. In the current thesis I also investigate the acquisition of unaccusative- inchoative verbs in L2 contexts. The focus of this work, however, is not on the acquisition of a particular property of the target language such as ne-cliticization (see section 2.2.3) and auxiliary selection (see section 2.2.4) but on the structural variations in which this subclass of intransitive verbs are used by nonnative speakers. I show that unaccusative-inchoative verbs present a unique challenge, not shared by unergatives, to adult learners who attempt to master English as a second language. It is argued that this challenge originates from the mismatch between the lexico-syntactic property and the dominant surface syntax of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 unaccusative-inchoatives. Using three sources of data, namely the existing L2 literature, a large computerized corpus of L2 English and the results obtained from a grammaticality judgment test, this work demonstrates how issues familiar to L2 researchers such as overgeneralization, simplification, first language transfer, lexical reanalysis and so forth play significant roles within intricate interactions of lexical and syntactic components of grammar when unaccusative-inchoative verbs are acquired in L2 contexts. Following this introductory chapter. Chapter 2 presents some evidence that the division within the intransitive verb class proposed by the Unaccusative Hypothesis is a linguistically and psycholinguistically real one. The languages covered in the chapter are English, Italian, Spanish, Japanese, and Korean. Some topics discussed in this chapter become relevant in subsequent chapters when we take into consideration the grammatical properties of learners' Lis with respect to a few phenomena related to the L2 acquisition of English unaccusative-inchoative verbs. Chapter 3 reviews the existing L2 literature on the acquisition of unaccusative-inchoative verbs in order to clarify both the issues involved and the questions that need to be addressed. Three important aspects of acquisition discussed in the chapter are structural variations, crosslinguistic influences, and developmental patterns. Chapter 4 presents a study conducted on a research corpus culled from the Longman Learners Corpus. The results obtained from this research and the review of the L2 literature lead to a hypothesis which aims to account for the complex phenomena surrounding L2 acquisition of unaccusative-inchoative verbs. The hypothesis, termed the "unaccusative trap, " suggests that unaccusative- inchoative verbs are initially acquired as unergative verbs in adult L2 contexts as far as their lexico-syntactic representation is concerned. It is further argued that the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 learners who produce typical errors on unaccusative-inchoative verbs are in fact relatively advanced learners who have come to correct the initial lexical misanalysis. Chapter 5 further investigates the structural, crosslinguistic, and developmental predictions that the "unaccusative trap" hypothesis makes by analyzing the data obtained from a grammaticality judgment test conducted on native speakers of Italian and Japanese. Further implications of the "unaccusative trap" hypothesis in other adult L2 contexts and in child LI acquisition are considered in Chapter 6. The study concludes with Chapter 7 which discusses significant potential differences between child LI and adult L2 acquisition that emerge firom the perspective of the "unaccusative trap" hypothesis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 2. Two Classes of Intransitives In this chapter, I review the literature for evidence for the two classes of intransitive verbs, focusing on the following five languages: English, Italian, Spanish, Japanese and Korean. The reviewed phenomena are morphological, semantic, syntactic and psycholinguistic in nature. These phenomena, however, are not meant to serve as watertight "unaccusative diagnostics." As often pointed out as "unaccusative mismatches" (e.g. Levin and Rappaport 1989, Levin and Rappaport Hovav 1992, 1995), it is well-known that many so-called unaccusative diagnostics are sensitive to fine semantic subclassifications within the unaccusative-inchoative class and do not always work in the same manner crosslinguistically. My purpose in presenting a wide-ranging, though far firom comprehensive, overview of linguistic phenomena pertinent to the Unaccusative Hypothesis is to show that the division within the general intransitive verb class, which is not very obvious at first, is nonetheless real and has significant ramifications for various aspects of the linguistic system of human language. Because this division may not be very familiar to many students of second language acquisition, I hope that this chapter will also serve as a general introduction to the linguistic background of the topic of this thesis. This chapter as a whole should present strong evidence for the following points: (1) a. There is a crosslinguistically observable distinction between two major subclasses of intransitive verbs, namely, unergatives and unaccusative- inchoatives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 b. The surface subject of an unergative verb shares some characteristics of the subject of a transitive verb while the surface subject of an unaccusative-inchoative verb shares characteristics of the object of a transitive verb. c. The division is most naturally accounted for by assuming the subject of an unergative verb is an external argument at A-Structure, which is projected to subject position at D-structure while the subject of an accusative-inchoative verb is an internal argument projected into object position at the initial syntactic level. The crosslinguistic similarities and differences observed with respect to the two classes of intransitive verbs should also serve as background necessary for the analysis of the L2 acquisition of English unaccusative-inchoative verbs, which is the focus of the rest of the thesis. 2.1. E n g l is h 2.1.1. Inchoatives and Causatives As briefly mentioned in Chapter 1, English vocabulary contains a large number of ergative verbs which alternate between (intransitive) inchoative use and (transitive) causative use over identical forms. The following is just a small sample of this type of ergative. (2) blacken, break, broaden, bum, change, close, crystallize, darken, decrease, degenerate, deteriorate, double, drop, dry, expand, fossilize, freeze, grow, increase, intensify, harden, melt, move, open, shatter, shift, sink, slide Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 Semantically, most ergatives fall into one of the two semantic classes, i.e. the verbs of "change of state" such as break and freeze and the verbs of "change of location" such as drop and move. As illustrated in the following examples, it is the subjects of the inchoative sentences in (3) and the objects of the corresponding causative sentence in (4) that undergo either a change of state or change of location. (3) a. The vase broke. b . The ice cream froze. c. The hospital has moved to a new location. (4) a. The baby broke the vase. b . John froze the ice cream by mistake. c. The city has moved the hospital to a new location. This consistency in the thematic correspondence between the inchoative and causative usage is a characteristic of ergative verbs. (For the subclassification of these verbs and a more extensive list, see Levin 1993:sec. 1.1.2.1.) Another type of alternating verb that English vocabulary contains consists of those in (5). (5) fall - fell lie - lay rise - raise The verbs in each pair in (5) are not identical in form but there is still an obvious morphological resemblance between them mediated by alternate vowel qualities. In this respect, the verbs in (5) are between alternating ergatives and non-altemating unaccusatives. Although this kind of causative-inchoative alternation based on non identical but morphologically related verb forms is uncommon in English, the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 pattern varies in different languages. (See, for example, sections 2.2.1 and 2.4.1 for discussions on the causative-inchoative alternations in Italian and Japanese.) Some examples of English unaccusative verbs are in shown in (6). (6) appear arise arrive be come die disappear emerge enter exist go happen occur remain stay Semantically, the unaccusative verbs in (6) fall into one of three subclasses: verbs of existence (e.g. be, exist, remain and stay), verbs of appearance and disappearance (e.g. appear, arise, die, disappear, happen and occur), and verbs of inherently directed motion (e.g. arrive, come, enter and go). (For more detailed discussions, see Levin 1993:sec. 6.) The unaccusative-inchoative verbs in (2), (5) and (6) contrast with unergative verbs such as those in (7). (7) dance dream laugh shout skip sleep smile smirk swim talk telephone walk yawn yell As mentioned in Chapter 1, semantically speaking, unergative verbs are generally verbs of internally caused events (Levin and Rappaport Hovav 1995). In the following sections, 1 will present evidence that the distinction between the unaccusative-inchoative class and the unergative class has significant ramifications in various areas of English granunar. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 2.1.2. Nominalization with -er The distinction of the two kinds of intransitive verbs in English is relevant to the derivational morphology involving the nominals with the suffix -er.*2 One major source of -er nominals is the transitive verb class; (8) attacker, cleaner, extinguisher, grinder, killer, promoter, supporter, washer Semantically speaking, many -er nominals may appear to refer to animate entities but this is not always the case. For example, cleaner and grinder can be a machine or a tool as well as a human. What is significant, however, is that they correspond to the subject, not the object, of a sentence in which the respective verb appears as its main predicate. Another major source of -er nominal is the class of intransitive verbs as illustrated in (9). (9) beeper, dancer, dreamer, glider, jumper, runner, speaker, swimmer, walker These examples also refer to an entity that can be the subject of a sentence whose predicate is the base verb. However, not all intransitive verbs can appear as a base of an -er nominal (Levin and Rappaport Hovav 1995:139): (10) *appearer, *ariser, *arriver, ?comer, *dier, *disappearer, *enterer, ♦exister, ?goer, *happener, *occurer, *remainer, *vanisher‘^ Napoli (1988:132-133) is critical o f the diagnostic use of the -er suffixation for the unaccusative vs. unergative distinction. Her arguments are based on verbs such as stick and shine which she considers to be unaccusatives. In my view, however, these verbs can be either middle or unergative. If so, her criticism seems to lose much of its force. The -er nominals shown with a question mark do appear either in compounds like late comer and church-goer or in a noun phrase such as early riser. These cases, however, do not really change the general tendency that unaccusatives are not normally suffixed with the -er. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 For instance, a ghost that appears in an attic of an old mansion cannot be called an appearer. Nor can a rare kind of plant that exists only in the rain forest of the Amazon be an exister. The examples in (10), therefore, present a problem to the generalization that the referent of an -er nominal is a potential subject of a sentence whose predicate is its base verb. Moreover, the contrast between (9) and (10) is an indication that what we know as "intransitive verbs" are not a monolithic class, but consist of at least two subclasses. Since the sentence-level grammatical functions such as subject and object cannot be adequately utilized to account for the data above, we need to look for a descriptive generalization somewhere else. One solution is to account for the data at the level of argument structure (Oshita 1995:185-189). In argument structure terms, the transitive verb class has both external and internal arguments. This can be represented as in (11). (11) Transitive V: (x(y)) Here, the external argument represented as (x) is placed within one set of parentheses and the internal argument ((y)) within two sets of parentheses. The suffixation of -er can be thought of as an argument structure alteration process, which makes the referent of a derived noun semantically correspond to the external argument of the base verb. Let us represent the referent by R. The formula (12) schematizes this argument structure alteration. (12) Nominal Formation with the -er Suffix i: R-bind the external argument (shown as "R=x") [obligatory] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 1 il: Nullify the argument structure of the base verb (shown as [obligatory] The rule in (12) consists of two independent sub-processes, one of which applies to a particular type of argument and the other to the whole argument structure. Although they are numbered for convenience, the order of application is not important at ail. It should be noted as well that the way the process is expressed in (12) makes the rule completely blind to the classification of a base verb to which the suffix is affixed. In other words, the rule need not specify which verb classes it can and cannot apply to. It is automatically determined by the process itself. (For a more detailed formulation of argument alteration processes, see Oshita 1994.) When applied to a transitive verb, (12) works in the way described in (13). (13) Transitive V: (x(y))— > Nom (-er): R=x<(x(y))> The right side of the arrow indicates that the referent for the derived nominal is semantically associated with the external argument of the base verb. Although the internal argument is left untouched, this argument no longer needs to be obligatorily projected because the original argument structure is now nullified. In the case of the grammatical -er nominals based on intransitive verbs in (9), if we assume the argument structure representation of (14) for the base verbs, (12) works without any problem as shown in (15). (14) Unergative V: (x (0)) (15) Unergative V: (x (0)) — > Nom (-er): R=x <(x (0))> Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 In ( 14), the lack of an internal argument is indicated by ((0)) in the two sets of parentheses. Since it is the external argument (x) that has to be R-bound, the lack of ((y)) does not affect the suffixation process. Consequently, the derived -er nominals are perfecdy grammatical. The ungrammaticality of the examples in (10) suggests that their base verbs have a significantly different A-Structure firom that of (14). Since the base verbs in (10) also require one argument, we assign this to the internal argument position. (16) Unaccusative V: (0 (y)) This time, the process of R-binding fails to bind an external argument simply because there is no (x) in (16). Since R-binding is obligatory as indicated in (12i), the derived nominal turns out to be ungrammatical. (17) Unaccusative(0 (y)) ~> Nom (-er): *R=0 <(0 (y))> *[no external argument to R-bind] In this way, the difference in grammaticality between (9) and (10) is naturally accounted for by assigning distinct A-Structures to the base verbs and by seeing the suffixation as a process of A-Structure alteration. Incidentally, the A-Structure alteration process in (12) naturally accounts for Burzio's observation (1986:161) on the interpretation of the -er nominals derived from ergative verbs. Consider, for instance, words such as developer, freezer and opener. Burzio notes that although the base verbs of these nominals can be either transitive or intransitive, the nominals themselves correspond only to the subject of the transitive sentence. Since ergative verbs alternate between causative and inchoative uses, they have two potential argument structures, one for transitive Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 verbs in (13) and the other identical to the unaccusative one in (16). As we have seen above, the R-binding of the -er suffixation fails to apply with the unaccusative- inchoative A-Structure. Consequently, the reading based on the inchoative sense is excluded from the nominals based on ergative verbs. 2.1.3. Passive and Perfective Adjectives Another type of morphological evidence for the distinction of unergative and unaccusative-inchoative classes within the general class of intransitive verbs concerns the formation of deverbal adjectives that have the same form as the past participle of the base verb (Bresnan 1982, Levin and Rappaport 1986, Levin 1993:sec. 5.4, Oshita 1994, 1995). Very often these adjectives are derived from regular transitive verbs and have a passive reading. (18) a. rejpecref/leaders b. difurnished toom. d. 2i tested V D & Û io d e. o . published c. typed manuscripts f. written texts In contrast, the examples of adjectives in (19) are derived from intransitive verbs and have a perfective reading. ( 19) a. elapsed time b. d. a risen Christ 2i fallen leaf c. an undescended testicle e. a stuck window f. the drifted snow g. a lapsed Catholic h. 2l failed writer (excerpted from Bresnan 1982:30, (47)) i. wilted lettuce j. an escaped convict k. a collapsed tent 1 . burst pipes m. rotted railings n. sprouted wheat Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 o. swollen feet p. a rusted screen q. vanished civilizations (excerpted from Levin and Rappaport 1986:654, (101)) However, not all intransitive verbs form grammatical adjectives. For instance, all the examples in (20) are ungrammatical as English adjectives. (20) a. *a run man b. *a coughed patient c. *swum contestants d. *flown pilots e. *cried babies f. * exercised athletes g. *sung singers h. *yawned students i. *laughed audience (based on Levin and Rappaport 1986: 654, (102)) Here again, we have a situation in which some intransitive verbs behave like transitive verbs while others do not. The formation of passive and perfective adjectives*'^ can be formalized in the A-Structure alteration framework: (21) Passive and Perfective Adjective Formation with the -en Suffix i: R-bind the internal argument (shown as "R=y") [obligatory] ii: Delete the external argument (shown as "x=0") [obligatory only if there is x] When applied to a transitive verb, (21) works as follows. (22) Transitive V: (x (y)) — > Adj (-en): R=y (x=0 (y)) The deverbal adjectives discussed in this section are often called "adjectival passives" (e.g. Bresnan 1982, Levin and Rappaport 1986). This term, however, is rather misleading because it emphasizes "passive" more than "adjectival". As Zubizarreta (1987), Oshita (1994) and this section show, it is doubtful that all the adjectives in this class are actually derived from passive participles. In this thesis, therefore, I use the term "passive and perfective adjectives", which seems to me morphologically and semantically more appropriate. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 Notice that the internal argument ((y)) is associated with the referent of the derived adjective while the external argument (x) is deleted. The deletion of (x) is necessary because unlike the regular passivization of a transitive verb, in which the external argument is simply "absorbed" or "suppressed" and remains potentially available as "argument adjunct" such as PRO or an argument of the preposition by (Jaeggli 1986b, Grimshaw 1990), the external argument of the base verb for the adjective formation is not available in any way. Compare (23) and (24). (23) The room was furnished (by the landlord). [verbal passive] (24) This is a furnished room (*by the landlord). [passive adjective] Now let us consider the difference between (19) and (20). We can account for the grammaticality of the examples in (19) by the A-Structure alteration process of (21). (25) Unaccusative V: (0 (y)) — > Adj (-en): R=y (0 (y)) Since unaccusatives do not have an external argument in the first place, the deletion of the external argument applies vacuously, which is allowed by (21ii).*5 However, the internal argument ((y)) is R-bound, as required, and the derived adjectives are grammatical. In contrast, (21) automatically screens out the passive and perfective adjectives based on unergative verbs. (26) Unergative V: (x (0)) — > Adj (-en): *R=0 (x=0 (0)) *[no internal argument to R-bind] This does not mean, however, that all unaccusative verbs can become felicitous passive adjectives. Levin and Rappaport (1989) note that the atelic unaccusatives such as remain and exist do not form passive adjectives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 Since unergative verbs lack an internal argument in their A-Structure representation, the obligatory R-binding fails to apply. This is why hypothetical words such as coughed cannot be used as grammatical adjectives although a noun phrase such as coughed men in the sense of 'men who have coughed' may appear to be a logical possibility. Ergative verbs present an interesting case for the interpretation of passive and perfective adjectives based on them. Unlike their -er nominal counterparts, the adjectives in (27) can be derived from both causative and inchoative A-Structures. If the derivation process involves a causative, the derived adjective takes on the passive sense. If the derivation is based on an inchoative, the adjective has the perfective sense. (27) a. a broken vase b. closed doors c. frozen ice cream d. decreased amount e. a shattered dream f. melted sugar For example, a broken vase in (27a) can be either a vase that was broken' or a vase that broke.' The examples in (18) based on nonaltemating transitive verbs do not share this ambiguity, allowing, for instance, only the passive reading of 'manuscripts that were typed' for typed manuscripts in (18c). Incidentally, in this respect, the verb boil presents a particularly interesting case. When its argument is a liquid, this verb can be used either as a causative or an inchoative. We can say, for instance, both Someone boiled water and Water boiled. Boiled water, therefore, can be interpreted as either 'water that has been boiled' or 'water that has boiled.' When the argument is "the eggs," however, the verb denotes a certain way of cooking and can be used only as a causative. Someone boiled the eggs is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 perfectly grammatical, but The eggs boiled with the relevant sense of boil does not sound very natural. Consequently, the boiled eggs always means the eggs that have been boiled,' not the eggs that have boiled.' 2.1.4. (Compound) Adjectives with -ing A third kind of morphological evidence for the unergative versus unaccusative- inchoative distinction in English is presented in Oshita 1994, 1995. In these papers, I have argued that one function of the derivational suffix -ing is to form adjectives, which are distinct from superficially similar present participles used as modifiers. One obvious difrerence between them, for instance, is the lack of progressive sense in the -ing adjective, which is a characteristic of the present participle. When this adjectival suffix is added to the base verb, the derived -ing is ungrammatical unless the original internal argument, if there is one, is projected as a lefthand constituent of a compound adjective. For instance, if an adjective is derived from a regular transitive verb, the derived adjective normally turns out to be ungranunatical unless the original internal argument is also projected as a constituent of a compound. (28) a. *(odor-)destroying insoles b. *(labor-)inducing salad c. *(water-)dropping helicopters d. * (money-)making machines e. *(mind-)broadening anecdotes f. *(flag-)waving conservatives g. *(pain-)relieving drugs h. *(adjective-)forming affixes (partly based on Oshita 1994, 1995) 1 6 Psychological causative verbs such as frighten and discourage are exceptions to this generalization. For details, see, among others, Grimshaw ( 1990) and Oshita ( 1994, 1995) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 8 (28) also illustrates that the nouns which are modified by the derived (compound) adjectives are potential external arguments of the base verbs. In our A-Structure alteration framework, the suffixation of the adjective-forming -ing can be formalized as (29). (29) Adjective Formation with the -ing Suffix i: R-bind the external argument (shown as "R=x"). [obligatory only if there is x] When the -ing adjective is derived from the base transitive verb, the A-Structure alteration is formally presented as: (30) Transitive V: (x (y)) > Adj (-ing): R=x (x (y)) Because the referent of a derived -ing adjective corresponds to the external argument of the base verb, this is indicated by the R-binding of the external argument (x). (30) clearly shows that the internal argument ((y)) is still intact. This is why it has to be projected as a constituent of the compound. When the base verb is an unergative, the derived -ing adjective can appear without forming a compound: (31) a. yZymg saucers b. S/eepmg Beauty c. a walking encyclopedia d. a traveling salesman e. a praying mantis f. flying fish‘d '7 At a glance, these -ing words may appear to be the first constituent of a compound noun, e.g. flying saucer raxher than an independent adjective. For the argument that they are in fact adjectives and that a structure like flying saucer is a lexlcalized noun phrase, which contrasts with a real compound noun such as flying ticket, please see Oshita (1994, 1995). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 Our A-Structure alteration account works perfectly well in this case, too. (32) Unergative V: (x (0)) — > Adj (-ing): R=x (x (0)) In (32), the external argument of the base verb is R-bound as required and there is no other argument requirement. The prevalence of derived adjectives such as (28) and (31) in English are exemplified, for instance, by the following definition of otter in a popular dictionary.*® (33) a swimming fish-eating animal with beautiful brown fur In (33), the first adjective swimming can appear as an independent word because it is based on an unergative verb. The head of the compound adjective fish-eating, however, is derived from a transitive verb and has to appear with the lefthand constituent which serves as its intemal argument. Now, observe the following two kinds of -ing adjectives. The examples in (34) are derived from unergative verbs and ungrammatical. This is because the potential external arguments of the base verbs, which should be referents of the derived -ing adjectives, are constituents of the compound adjectives. (34) a. *patient-walking miracle medicine b. *child-crying horror movies c. * audience-smiling auditoriums d. *friend-grinning weddings (partly based on Oshita 1994, 1995) • ® Longman Dictionary o f Contemporary English, New Edition (1987) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 In contrast, those -ing adjectives derived from unaccusative-inchoatives are grammatical as long as the words in parentheses appear as lefrhand constituents of the compound adjectives. (35) a. *(mouth-)watering garden vegetables b. *(jaw-)dropping additional suggestions c. *(pulse-)pounding psychological thrillers d. *(spine-)tingling sexy tales e. *(knee-)shaking, *(voice-)quavering anxiety (based on Oshita 1994, 1995) The contrast between (34) and (35) cannot be accounted for by resorting to the use of g ram m atical functions such as subject because the lefthand constituents of the compounds are potential subjects of the base verbs in both sets of examples. Instead, we should note that the grammaticality condition on the derived -ing adjectives is exactly the same in (28) and (35). That is, the condition for gram m aticality is the same between transitive-based and unaccusative-inchoative- based -ing adjectives. Now let us apply (29) to the A-Structure of unaccusative-inchoative verbs. (36) Unaccusative-lnchoative V: (0 (y)) ~> Adj (-ing): R (0 (y)) Here, there is no R-binding because there is no external argument in the A-Structure of unaccusative-inchoative verbs. This is not a problem, however. R-binding is not obligatory when the A-Structure of the base verb does not have an external argument. On the other hand, the intemal argument ((y)) is still intact and has to be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 projected. This is why the -ing adjectives in (35) are all head constituents of compound adjectives which require the lefthand constituents in parentheses. 2.1.5. Arbitrary Pronoun they Jaeggli (1986a) shows that the distinction between unergatives and unaccusative- inchoatives also becomes apparent in the interpretation of the third person plural pronoun they. Observe the ambiguity in the following sentences: (37) a. They rarely dance in Japanese weddings. b . They walk faster in big cities. c. They sell cigarettes at all gas stations. d. They don't allow dogs on the beach. (Jaeggli 1986a, (27a-b)) It is possible to have either a definite referential reading or an indeterminate arbitrary reading for the subject pronoun they, i.e. as long as the referents are understood to be human (Jaeggli 1986a:62). When the third person plural pronoun appears in object position of a transitive verb, however, it cannot have an arbitrary reading: (38) a. I haven't seen them in gas stations. (Jaeggli 1986a, (28a)) b . John meets them every morning. c . We really like them in this country. d . The police will arrest them. In (38), the pronoun them can only have a definite referential reading. Therefore, there appears to be a crucial difference in interpretive possibilities between the subject and object usage of the third person plural pronoun in English. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 This generalization, however, is incorrect. Observe the following examples of passive sentences. (39) a. They are arrested all the time by the police. (Jaeggli 1986a, (29a)) b. They are seen on the beach every Sunday morning. c. They are admired in this country. d . They will be found in the hotel lobby. Although the pronouns in (39) all have a nominative form and are in subject position, they do not allow an arbitrary reading. What is of interest to us is that this situation is also the same with unaccusative sentences: (40) a. They exist without any water on this planet. b. They arrive really tired after such a long trip. (Jaeggli 1986a, (29d, e)) c. They always appear without prior notice. d. They remained in their home country. In (40), we have three semantic subtypes of unaccusative verbs: verbs of existence in (40a, d), a verb of inherently directed motion in (40b), and a verb of appearance in (40c). None of them allows an arbitrary reading. Jaeggli (1986a) accounts for this distributional restriction of the arbitrary third person plural pronoun by (41). (41) An arbitrary plural pronominal cannot be in a chain that is Case- or theta- marked directly by a verb. (based on Jaeggli 1986a, (34)) (9 Jaeggli's data and analysis aie far more detailed than my discussion here. For detail, please refer to Jaeggli (1986a). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 The non-arbitrary interpretation of them in (38) is clearly accounted for by (41) since the pronoun is both Case-marked and theta-marked by the verb in its base generated position. The unavailability of the arbitrary interpretation for they in the passive and unaccusative sentences in (39) and (40) is also covered by (41) because the surface subjects in these examples are generated in object position where they are theta-marked by the verb. The generalization of (41) also leaves room for the ambiguity between the definite and arbitrary readings of they in (37) because in the unergative and transitive sentences, the subject arguments are neither Case-marked nor theta-marked directly by the verb. The interpretations of the third person plural they^ therefore, clearly indicates the distinction of the two kinds of intransitives. And if Jaeggli's generalization is on the right track, the difference between them is not simply semantic but also syntactic in nature. 2.1.6. Transitive-based Resultative Construction Some semantic and syntactic characteristics shared by the sentences in (42) were noticed by Simpson (1983). (42) a. John wiped the table c/eon.(Takezawa 1993:48, (9b)) b. I painted the car a pale shade of yellow. c. I cooked the meat to a cinder. d. The boxer knocked John out. (Simpson 1983:143, (2)-(4)) These sentences all contain the basic transitive word order of SVO and an additional descriptive phrase. Needless to say, all but (42d) are grammatical with or without the respective additional phrase. When it is there, however, each descriptive phrase induces the "resultative" reading, describing a state into which the object NP is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 brought as a result of the event expressed by the main verb. The type of phrase can vary: for example, clean in (42a) is an adjective, a pale shade o f yellow in (42b) is a nominal phrase, to a cinder in (42c) is a prepositional phrase with an object, and out in (42d) may be seen as an intransitive preposition that does not require an object. What remains the same, however, is the semantic and syntactic characteristics of the transitive-based resultative construction: a resultative phrase functions as a secondary predicate that describes a state to which the object NP comes as a consequence of the event expressed by the main predicate. Simpson (1983) also has shown that the NP of which a resultative phrase is predicated does not need to be adjacent to it. In the following examples, the italicized constituents are all predicated of the subject of the respective sentences. (43) a. The table was wiped c/ean. b. The car was painted a pale shade of yellow. c. The meat was cooked to a cinder. d. John was knocked out. At a glance, (43a-d) may appear to pose a challenge to the generalization based on (42) that a resultative phrase is predicated of the sentential object. This, however, is not a real problem since the surface subject NPs of the passive sentences have arguably moved out of object position, leaving traces behind. (43a), for instance, has S-Structure representation: (44) the tablei was wiped ti clean t _____________ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 The grammaticality of (43a-d) is still accounted for because the resultative phrases are predicated of the traces left by the surface subjects. Finally, Simpson (1983) has noted the contrast apparent in "conative alternation" (Levin 1993:sec. 1.3): (45) a. I shot John {dead). b. I shot at John {*dead). (cf. Simpson 1983:147) (46) a. Karen cut the bread {into a few slices). b. Karen cut at the bread {*into a few slices). (47) a. Maya pushed the little toy table {to a comer of the room). b. Maya pushed at the little toy table {*to a comer of the room). In (45)-(47), both the regular transitive and the conative sentences are perfectly grammatical on their own. When a descriptive phrase is added to them, however, the regular transitive clauses remain grammatical and allow a resultative reading but the conative clauses are rendered ungrammatical. In other words, the NP whose resultative state is described cannot be a prepositional object but has to be a verbal object of the main predicate. This syntactic characteristic of the resultative construction (Simpson 1983:146, Levin and Rappaport Hovav 1995, Takezawa 1993) may be summarized as: (48) Direct Object Condition: A resultative phrase must always be predicated of an NP or its trace in direct object position. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 With this background on the general semantic and syntactic characteristics of the resultative construction, now observe (49) and (50). (49) a. The fish burned (black), (based on Takezawa 1993:49, (12b)) b. The ice cream froze (hard). c. The butter melted (to a liquid). d. The vase broke (into little pieces). (based on Simpson 1983:143, (6b), (7b) and (8b)) (50) a. I laughed (*sick). b. The students danced (*tired). c. The baby cried/sobbed (*to sleep). d. I shouted/screamed/yelled/bellowed (*hoarse). (based on Simpson 1983:145) The contrast between the two groups of intransitive sentences is obvious. The verbs in (49) allow the addition of a resultative phrase at the end of the sentence, while those in (50) clearly do not. Notice that the unacceptability of the resultative construction in (50) cannot be attributed to the implausibility of a resultative reading based on events typically expressed by unergative verbs. There is no reason in the real world, for instance, why people should not get tired as a result of dancing. In fact, such a situation often occurs and, as we see later, it is possible to linguistically encode it using other syntactic constructions. Therefore, the ungrammaticality observed in (50) cannot be attributed to a pragmatic or logical anomaly. The unaccusative hypothesis can account for the contrast between (49) and (50) as a natural consequence of the Direct Object Condition in (48). In (49), the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 inchoative verbs such as bum have an internal argument in their argument structure representation and this argument is syntactically projected in object position at D- Structure. The surface subjects have moved in the preverbal position from the initial object position, leaving traces behind. The Direct Object Condition is satisfied because the resultative phrases are predicated of these traces. For example, (49a) has S-Structure representation: (51) the fishi burned tj black Î______ The sentences in (49), therefore, share the crucial derivational characteristic with the passive sentences in (43). In contrast, the surface subjects in (50) are all projected in subject position at D-Structure since they are external arguments in the A- Structure representation. Consequently, there are no traces in syntactic object position of which the intended resultative phrases in (50) can be predicated.^o Simpson (1983) also points out that although it is rare, some unaccusative verbs show up in the transitive-based resultative construction: -0 This, however, does not mean that all the unergative sentences with a descriptive phrase are ungrammatical. The sentences like (i) are perfectly wellformed with a "depictive" reading. (i) a. John danced sweaty. b. John walked tired. (Takezawa 1993:49, (15)) Unlike a resultative phrase, a depictive phrase like those in (i) describes the subject's condition when the event took place. Therefore, (ia), for instance, means John was sweaty while he danced' not John got sweaty as a result of dancing.' It also should be pointed out that a depictive phrase can be predicated of the direct object as well as the subject, (iia) means that the fish was raw when John ate it while (iib) means that John was nude when he ate the fish. (For more detailed discussion, see Takezawa 1993.) (ii) a. John ate the fish raw. b. John ate the fish nude. (Takezawa 1993:48, (10)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 (52) a. He flushed/blushed (jred). b. *I flushed/blushed him {red). (based on Simpson 1983:144, (9)) In (S2a), the resultative phrase red is semantically related to the surface subject although as indicated in (52b) there is no grammatical transitive counterpart for the v&ihs flush and blush with the relevant senses. The examples, therefore, show that unaccusative verbs can appear in the resultative construction. This should not be a surprise at all. After all, the argument structure of an unaccusative verb is identical to that of an inchoative. Both have a direct internal argument with no external argument. Both, therefore, can satisfy the syntactic requirement of the Direct Object Condition when their argument is projected into syntax. However, English unaccusative verbs that appear in the resultative construction are rather rare. This is because there is a semantic constraint on the type of verbs that can be used in the resultative construction. Since the semantics of this construction pertains to the state that arises as a result of an event, verbs with an inherent sense of "a change of state" are most frequently used in this construction. In English, many such verbs tend to fall in the class of ergative verbs that have both causative and inchoative usage. Most unaccusative verbs, in contrast, tend to have a sense of "existence," "(dis)appearance," or "inherently directed motion." All these meanings are not compatible with the change of state meaning, which the semantics of the resultative construction conveys. Instead, when a descriptive phrase is added to a typical unaccusative predicate, it often leads to "depictive" reading. In the following set of examples in (53), for instance, the meanings of the main predicates are "existence" in (53a), "appearance" in (53b) and "inherently directed motion" in (53c, d): Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 (53) a. Carla remained in the country bored. (Levin and Rappaport Hovav 1995:56, (51a)) b. evaexgtd bedraggled. (Simpson 1983:147, (28)) c. John fell (down) dead. (Simpson 1983:147, (29)) d. Willa arrived breathless. (Levin and Rappaport Hovav 1995:56, (5 lb)) Notice that the descriptive phrases in italics depict the states which the surface subjects are already in when the events of the main predicates take place. In sum, despite the semantic constraint which excludes most English unaccusative verbs from appearing in it, the transitive-based resultative construction still distinguishes the unaccusative-inchoative verb class and the unergative verb class. The argument structure level contrast of these two subclasses of intransitives results in different D-Structure representations, accounting for the distinct behaviors of the two verb classes with respect to the Direct Object Condition relevant to the resultative construction. 2.1.7. Unergative-based Resultative Constructions The preceding section has shown that the unergative verbs do not participate in the transitive-based resultative construction because the lack of a direct internal argument in their argument structure representation ultimately violates the Direct Object Condition, a syntactic constraint on this construction. The unergative class of verbs, however, appear in slightly different constructions which also have resultative reading. In these constructions, unergatives, which do not normally subcategorize an object, are followed by a postverbal nominal and an additional descriptive phrase. The non-subcategorized nominals fall in one of three types: the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 subject's reflexive pronoun (also known as "fake reflexives"), an NP inalienably possessed by the subject, and an independent NP (Simpson 1983, Levin and Rappaport Hovav 1995). I call all three types of constructions unergative-based resultative constructions in order to differentiate them from the transitive-based resultative construction discussed in the preceding section. The first type of postverbal NPs that show up in the unergative-based resultative construction are "fake reflexives" (Levin and Rappaport Hovav 1995). (54) a. 1 laughed myself *isick). b. The students danced themselves *{tired). c. The baby cried/sobbed herself *(fo sleep). d. 1 shouted/screamed/yelled/bellowed myself *{hoarse). (based on Simpson 1983:145, (16)-(19)) In (54), the appearance of a reflexive pronoun and the descriptive phrase that follows it are mutually dependent not only for the grammaticality of the sentences but also for their resultative reading. That is, without either one, the sentences do not have the intended resultative reading. Each descriptive phrase is understood as a state of the preceding reflexive as a result of the event expressed by the main predicate. Since the reflexive and the subject of the sentence are coreferential, the resultative state is ultimately the one in which the subject finds itself as a result of its own action.2* Simpson (1983) observes that this particular type of unergative resultative construction accommodates transitive verbs, such as eat and drink in (i) and (ii), which optionally omit their objects. (i) a. I ate all day. b. I ate myself *isick). (ii) a. He drank all through the evening. b. He drank himself *iinto the grave). (based on Simpson 1983:145, (20) and (21)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 The second type of unergative-based resultative construction involves "non- subcategorized inalienably possessed NPs" (Levin and Rappaport Hovav 1995). (55) a. Sylvester cried his eyes b. Sleep your wrinkles *(away). c. Valentino strutted his life *(awayi). d. You need not stitch your poor fingers *{to the bone). (based on Levin and Rappaport Hovav 1995:36-37, (8) and (9)) As in the case of the construction with a fake reflexive, the appearance of an inalienably possessed NP and a resultative phrase are mutually dependent for the grammaticality of the sentences and their resultative reading. Again, each resultative phrase is interpreted as describing the state of the preceding NP. One semantic difference, however, is that the impact on the subject by the predicate's event is not as direct as in the case of the fake reflexive construction. This is in fact quite natural because the relation between the subject and the postverbal NP is not coreference, but one of the whole and a part. The third type of unergative resultative construction involves "independent NPs" (Levin and Rappaport Hovav 1995). (56) a. The joggers ran the pavement *{thin). (based on Levin and Rappaport Hovav 1995:53, (50)) b. The dog barked him *(awake). c. You may sleep it [i.e. the unborn baby] *(quiet). (based on Levin and Rappaport Hovav 1995:36, (6) and (7)) d. The clock ticked the baby *iawake). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 e. The phone rang me *{out o f my slumber). (based on Levin and Rappaport Hovav 1995:67, (76) and (77)) That the grammaticality of the sentence depends on the appearance of both the postverbal NP and the descriptive phrase is exactly the same as in the previous two types of unergative-based resultative constructions. In this third type, however, there is no semantic relatedness between the subject and the postverbal NP. Consequently, only the postverbal NP is interpreted to be affected by the event expressed by the main predicate. In this respect, this particular type of unergative- based resultative construction appears very similar to the transitive-based resultative construction such as (42), (45a), (46a) and (47a).^ The three types of unergative-based resultative constructions appear to share some fundamental properties of the transitive-based resultative construction. In both, there are a postverbal nominal (or trace) and a resultative phrase following it, and the latter is a secondary predicate predicated of the former.^ What is of immediate importance for us is the fact that none of the three unergative-based resultative constructions is available to unaccusative-inchoative verbs, suggesting the appropriateness of the distinction of the unergative and the ^ Transitive verbs with optionally omitted objects also appear in this construction. Notice that the postverbal NPs are not subcategorized by these verbs. (i) a. I ate him *(out o f house and home). b. She drank him *(under the table). (based on Simpson 1983:146, (23)) c. They drank the tea pot *(dry). d. Drive your engine *{clean). (Mobil ad.) (based on Levin and Rappaport Hovav 1995:37-38, (10) and (12)) ^ The syntactic position of the postverbal NPs that show up in the unergative-based resultative constructions is controversial in the literature. Levin and Rappaport Hovav (199S:chap. 2) discuss two possibilities, object position and position governed by the verb heading the VP, but leave the question open for further research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 unaccusative-inchoative classes. Observe the ungrammaticality of the fake reflexive construction built on inchoative verbs: (57) a. The fish burned (*itself) black. b. The ice cream froze (*itself)/mrd. c. The butter melted (*itself) to a liquid. d. The vase broke (*itself) into little pieces. Similarly, the inalienably possessed NP construction is not grammatical on inchoatives: (58) a. The fish burned (*its tail) black. b. The ice cream froze (*its top) hard. c. The butter melted (*its surface) to a liquid. d. The vase broke (*its bottom) into little pieces. The independent NP constmction involving inchoatives is also ungrammatical: (59) a. The fish burned (*the room smoky). b. The ice cream firoze (*the kid angry). c. The snow melted (*the road slushy). (Levin and Rappaport Hovav 1995:39, (20c)) d. The vase broke (*the auctioneer speechless). It also should be noted that the patterns in (57), (58) and (59) remain the same with the two unaccusative verbs that Simpson pointed out with respect to the transitive- based resultative construction. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 (60) a. He flushed/blushed (*hiniself) rcf/. (cf. Levin 1993:sec. 4.1) b. He flushed/blushed (*his cheeks) red. c. He flushed/blushed (*the young girl) embarrassed. The fact that the three unergative-based resultative constructions can be used with unergatives but neither with unaccusatives nor with inchoatives can be attributed to Burzio's Generalization (section 1.2). Burzio observes that only the verbs that can assign thematic role to their external argument can assign Accusative Case. Therefore, the non-subcategorized nominals that follow unergative verbs in the unergative-based resultative constructions can receive their Case from the verbs. Unlike unergative verbs that have a thematic external argument in their argument structure, unaccusative-inchoative verbs without a thematic external argument cannot assign Accusative Case. After all, this is the reason why the preverbal subject of an unaccusative-inchoative sentence is analyzed as an argument moved for a Case theoretic reason. According to this line of reasoning, additional postverbal NPs such as those in (57)-(59), therefore, inevitably render the sentence in question ungrammatical because there is neither Case assigner nor Case- assigning mechanism for them. 2.1.8. "X's Way" and Cognate Object Constructions The difference in the potential capacity to assign Accusative Case also accounts for the contrastive behaviors of the unergative and unaccusative-inchoative classes concerning two other syntactic constructions, namely "X's way " and "cognate object" constructions. Following are some examples of the "X's way" construction built on unergative verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 (61) a. They shopped *(their way) around New York. b. He worked *(his way) through the book. c. She talked *(her way) out o f the class. (Levin I993:sec. 7.4) d. Bill belched *(his way) out o f the restaurant. e. Harry moaned *(his way) down the road. f. Sam joked *(his way) into the meeting. (based on JackendofP 1990:211, (1)) There are two characteristics of the "X's way " construction that are similar to the unergative-based resultative constructions. One is that the main verbs do not subcategorize the "X's way" NP and the other is that the appearance of this NP and the following PP are interdependent for the sentence's grammaticality. In the "X's way" construction, however, the PP serves as a Path through which the subject NP moves (Jackendoff 1990) and the main verb conveys the sense of "directed motion" (Levin and Rappaport Hovav 1995). Considering this meaning shift which the unergative verbs undergo, the ungrammaticality of the "X's way" constmction based on the unaccusative verbs in (62) may appear quite natural because of their semantic anomaly. (62) a. *Jill remained her way to a ticket to the show. b. * Andrea appeared her way to fame. c. *The explosion occurred their way onto the front page. (Levin and Rappaport Hovav 1995:150, (39) and (40)) After all, the unaccusatives in (62) have "existence " and "appearance" senses and are semantically incompatible with the "X's way" constmction. However, the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 incompatibility of this construction and the unaccusative-inchoative class cannot be simply attributed to a semantic mismatch. Look at the following examples in which the unaccusative and inchoative verbs have the sense of "inherently directed motion" but are still incompatible with the "X's way" construction. (63) a. *The oil rose its way to the top. b. *The apples fell their way into the crates. c. *She arrived her way to the front of the line. (Levin and Rappaport Hovav 1995:148. (32)) Semantically speaking, there appears to be no inherent contradiction in using the verbs in (63) in the "X's way" construction. This suggests that there is a fundamental syntactic problem in (62) as well as (63). The lack of Case-assigning capacity inherent in the unaccusative-inchoative verb class can account for this ungrammaticality, by reducing it to a violation of Case Filter caused by the additional "X's way " N P.^'^ One last construction whose grammaticality can be attributed to the potential Case-assigning capacity of the unergative verb class is the "cognate object" construction: Levin (1993:sec. 7.4) also lists examples of "X's way" construction based on transitive verbs: (i) a. She stipulated her way out of the problem. b. The boy pushed his way through the crowd. c. The explorers cut their way through the jungle. Clearly, the "X's way" NPs in (i) are not subcategorized by the respective verbs, but the sentences are as grammatical as those based on unergative verbs. This seems to support the account of the ungrammaticality o f unaccusative-based "X's way " construction in terms of their Case-assigning capacity. Burzio's Generalization in fact predicts that unergatives and unaccusatives should pattern together in contrast to unaccusatives if the grammaticality o f the "X's way" construction depends on Case-related factors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 (64) a. The couple waltzed the most elegant waltz in the competition. b. Nancy danced an exotic dance. c. Wolfy hoped Lisa would dream a peaceful dream. d. Louisa slept a restful sleep. e. Malinda smiled her most enigmatic smile. (Levin and Rappaport Hovav 1995:40) Levin (1993:sec. 7.1) observes that in this construction the cognate object itself does not appear to make much semantic contribution to the whole sentence but the prenominal modifier that often accompanies a congnate object functions like an adverbial. What is important to us is that although not all unergatives take a cognate object, absolutely no unaccusative-inchoative verb seems to appear in this construction: (65) a. *The mirror broke a jagged break. (Levin and Rappaport Hovav 1995:147, (28)) b *She arrived a glamorous arrival. c. *The apples fell a smooth fall. (Levin and Rappaport Hovav 1995:148, (31a-b)) d. * Karen appeared a striking appearance at the department party. e. *Phyllis existed a peaceful existence. (Levin and Rappaport Hovav 1995:148, (32a-b)) The verbs in (65) vary in their core meanings: "a change of state" in (65a), "an inherently directed motion" in (65b, c), "appearance" in (65d) and "existence" in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 (65e). Since paraphrasing such as She arrived glamorously for (65b) seems possible with all the other examples as well, it is difficult to think of any semantic or pragmatic reason for the imgrammaticality of (65a-e) and it is more than reasonable to appeal to the lack of Case-assigning property inherent in the unaccusative-inchoative verb class for a grammatical account of them. 2.2 . ITALIAN 2.2.1. Inchoatives and Causatives As in English, the intransitive verbs which manifest syntactic unaccusativity are unaccusatives and inchoatives. Following are some examples of Italian unaccusative verbs. accadere to occur' andare 'to go' arrivare to arrive' cadere to fall' emergere to emerge' entrare to enter' esistere to exist' essere 'to be' morire to die’ nascere to be bom' partite to leave' restare 'to remain' sorgere to arise' stare to stay'^ uscire to go/come out' venire to come' (cf. Burzio 1986:159, Haegeman 1994:330) ^ Mario Saltarelli (personal communication 1997) informs me that verbs such as restare and stare require an additional temporal argument in addition to the NP argument. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 Semantically speaking, most of these verbs denote "existence," "appearance" or "inherently directed motion" of their sole NP argument and generally correspond to their counterparts in English (see section 2.1.1). Again as in English, ergatives, i.e. verbs that manifest causative-inchoative alternation, are also found in Italian. For our immediate interest, there are two subtypes of ergative verb: one has a "bare inchoative" while the other has a "si- inchoative." The first type of alternation is not morphologically marked and both causative and inchoative forms are identical. This type of ergative, therefore, manifests what Burzio (1986) refers to as "AVB/BV " alternation. (67) is an example of this type of alternation. (67) a. L'artiglieria affondo due navi nemiche. the artillery sank two ships enemy The artillery sank two enemy ships,' b . Due navi nemiche affondarono. two ships enemy sank Two enemy ships sank.' (Burzio 1986:25, (13a-b)) An argument due navi nemiche appears as the object of the verb in (67a) but is the subject in (67b). The subject status of the latter is clearly seen in the subject-verb agreement manifested in the verb form c^ondarono. Other verbs that Burzio (1986:73, footnote 5) lists as examples of the same type of ergative are shown in (68). (68) aumentare to increase" ingrassare to fatten" diminuire "to diminish" migliorare "to improve" Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 peggiorare to worsen' raffreddare to cool' rinverdiie to turn green' sprofondare to sink in' The second type of causative-inchoative alternation found in Italian is the V - si V pairs. The verb rompere is an example of this type of ergative. (69) a. Giovanni rompe il vetro. Giovanni breaks the glass 'Giovanni breaks the glass.' b. n vetro si rompe, the glass CL breaks The glass breaks.' (Burzio 1986:38, (46)) Unlike the morphologically unmarked alternation above, the clitic si appears in the inchoative use of rompere in (69b). The clitic si is considered as a morphological reflex of the 'loss' of subject theta-role by Burzio. Following are some more examples of verbs like rompere which have si V inchoative counterparts (Burzio 1986:75, footnote 24). (70) allargare to widen' attorcigliaie to twist' capovolgere to capsize' dividere to divide' liquefare to liquefy' radunare to gather' riempire to fill up' rovesciare to spill' In this way, the morphological marking of the causative-inchoative alternation in Italian appears to be more systematic than the few similar cases in English (section 2 . 1. 1). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 Unaccusative verbs and two types of inchoative verbs in Italian contrast with unergative verbs such as those listed in (71). (71) ballare to dance' camminare to walk' dormire to sleep' ghignare 'to smirk' gridare to shout' nuotare to swim' parlare to talk' ridere to laugh' saltare to skip' sognare to dream' sorridere to smile' telefonare to telephone A sole NP argument of these verbs is normally an instigator of the event denoted by the verbs. Although movement verbs are found among both unaccusative- inchoatives and unergatives, significant semantic differences exist between the two classes of verb. For unaccusative-inchoative verbs, the essential sense of movement is a change of location apparent in verbs such as arrivare and partira. In contrast, the critical semantics for unergative verbs is the manner of movement. In fact, a change of location is unnecessary for the core semantics of unergative verbs such as camminare and nuotare. Think of a person moving her legs on a conveyor belt of an exercise machine. Even though her activity does not result in any change of her location, we can still call it "walking." In contrast, the same action may never be referred to as "going." The following sections discuss how the unergative versus unaccusative- inchoative distinction within the general class of intransitive verbs is manifested in some areas of Italian grammar. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 2.2.2. Passive and Perfective Nominals There is an aspect of Italian derivational morphology which reflects the distinction between unergatives and unaccusatives. It is the formation of passive and perfective nominals (Burzio 1986:195): (72) I'arrestato the arrested (one)' i reclusi 'the confined (ones)' The grammatical nominals in (72) are formed on the past participles of the base transitive verbs and have a passive sense. In contrast, similar nominals based on intransitive verbs are generally not acceptable: (73) *i telefonati those who telephoned' *i camminati those who walked' However, some nominals derived from a restricted group of intransitive verbs are in fact possible. Burzio lists, for example: (74) I'ultimo arrivato the last (one) arrived' I'ultimo venuto the last (one) come' i caduti the fallen (ones)' i nati the bom (ones)' i soprawissuti 'the survived (ones)' gli accorsi those who rushed in' I'accaduto the happened (thing)' The derived nominals in (74) in general have a perfective sense. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 The grammaticality pattern evident in the word formation of (72)-(74) is a familiar one by now. Transitive verbs and a group of intransitive verbs on the one band and many other intransitive verbs on the other manifest a contrastive grammatical behavior. We have observed a similar pattern in some aspects of English derivational morphology (sections 2.1.2 to 2.1.4). Using the argument structure alteration schema, we can describe the Italian passive and perfective nominal formation as (75). (75) Passive and Perfective Nominal Formation with the -to Suffix i: R-bind the internal argument (shown as "R=y"). [obligatory] ii: Nullify the argument structure of the base verb (shown as "<...>"). [obligatory] Notice that (75) is a kind of hybrid of the passive and perfective adjective formation process (see section 2.1.3) and the -er nominal formation (see section 2.1.2) in English morphology. As the former, the process in (75) R-binds the internal argument and as the latter, it nullifies the argument structure of a base verb. When applied to a transitive base verb, (75) can account for why the examples in (72) and (74) are grammatical while those in (73) are deemed ungrammatical. (76) Transitive V: (x(y))~> Nom (-to): R=y <(x (y))> (77) Unaccusative V: (0 (y)) ~> Nom (-to): R=y <(0 (y))> In (76) and (77), the base verbs have an internal argument semantically associated with the referent of the derived nominal. Whether or not the base verb's argument Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 structure has an external argument requirement is irrelevant to the word formation process since its argument structure requirement is nullified by (75). (78) Unergative V: (x (0)) — > Nom (-to): *R=0 <(x (0))> *[no internal argument to R-bind] The unergative verb class, however, do not contain an internal argument in their argument stmcture representation. Without an internal argument, the R-binding in (75i) fails to apply, which renders any passive and perfective nominal derived from an unergative verb ungrammatical because there is no way to semantically associate the newly formed word with a referent. 2.2.3. Ne-Cliticization In this section, we review ne-cliticization, one of the most often cited linguistic phenomena as evidence for the distinction between the unaccusative and the unergative classes in Italian (e.g. Burzio 1986, Haegeman 1994, Levin and Rappaport Hovav 1995). First, as background, let us note the relative freedom of word order in Italian sentences. (79) a. Molti esperti arriveraimo. 'Many experts will arrive.' b. Molti esperti telefoneraimo. Many experts will telephone.' c. Molti esperti esaminerarmo il caso Many experts will examine the case.' (Burzio 1986:21, (4ia, iia, iiia)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 The examples in (79) contain an unaccusative verb arrivare, an unergative verb telefonare, and a transitive verb esaminare. Clearly, the NP molti esperti serves as the surface subject of each sentence, but (80) shows that the occurrence of the surface subject is not limited to a preverbal position. (80) a. Arriverano molti esperti. b . Telefoneranno molti esperti. c. Esamineranno il caso molti esperti. (Burzio 1986:21, (4ib, iib, iiib)) In (80), molti esperti appears postverbally. This type of subject-predicate permutation is very common in Italian and as long as the discourse context is appropriate virtually any type of preverbal subject may appear to the right of the verb (Burzio 1986:21). Consequently, it is not possible to separate verbs, including the three classes in question, into different groups based solely on the subject-verb word order. A difference among the unaccusative, unergative and transitive verbs in (79) becomes apparent in (81) where part of the postverbal subjects are cliticized with the quantifier molti still in postverbal position. (81) a. Ne arriveraimo molti. of-them will arrive many 'Many of them will arrive.' b. *Ne telefoneranno molti. of-them will telephone many 'Many of them will telephone.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 c. *Ne esamineranno il caso molti. of-them will examine the case many 'Many of them will examine the case.' (Burzio 1986:31, (30 a-c)) There is a clear distinction between (8 la) on the one hand and (8 lb, c) on the other. In (81), ne-cliticization is possible for the postverbal subject of the unaccusative verb arrivare, but not for the subjects of the unergative verb tel^onare and the transitive verb esaminare. In order to understand the contrast between (81a) and (81b, c), let us observe syntactic environments where the clitic ne is typically found. Look at (82) and (83). (82) a. I bambini mangiano molte pesche. the children eat many peaches The children eat many peaches.' (based on Roberts 1997:73, (40a)) b . Giacomo ha insultato due studenti. Giacomo has insulted two students 'Giacomo insulted two students.' c. Giacomo passa tresettimane a Milano. Giacomo passes three weeks in Milan Giacomo passes three weeks in Milan.' (Haegeman 1994:324, (49), (51)) (83) a. I bambini ne mangiano molte. the children of-them eat many The children eat many of them.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 b. Giacomo ne ha insultato due. Giacomo of-them has insulted two Giacomo insulted two of them.’ c. Giacomo ne passa tre a Milano. Giacomo of-them passes three in Milan Giacomo passes three of them in Milan.' (83) shows that part of the postverbal NPs in (82) can be cliticized. These postverbal NPs are clearly the objects of the main verbs. Therefore, the clitic ne is syntactically related to the object position of the main verb. This observation gains further support from additional data such as the following. (84) a. Giacomo ha parlato a due studenti. Giacomo has spoken to two students Giacomo spoke to two students." b. Giacomo resta tre settimane a Milano. Giacomo stays three weeks in Milan "Giacomo stays three weeks in Milan." (Haegeman 1994:324, (50), (52)) (85) a. *Giacomo ne ha parlato a due. Giacomo of-them has spoken to two "Giacomo spoke to two of them." b. *Giacomo ne resta tre a Milano. Giacomo of-them stays three in Milan "Giacomo stays three of them in Milan." Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 The ungrammaticality of (85a) in contrast to the grammatical (84a) indicates that the postverbal NP whose head may be cliticized cannot be an object of a preposition. In other words, the NP has to be the direct, not indirect, object of the verb. Similarly, (85b) in contrast to (84b) is also ungrammatical. There is no preposition involved here, but unlike its function as the object of the verb passare in (82c), tre settimane in (84b) is clearly not the object of the verb restare, but rather functions as an adverbial. These examples lead us to a generalization that ne-cliticization is possible not on any postverbal NP, but on the direct object of the verb. Furthermore, this generalization works for the /le-cliticization of the surface subject of a passive sentence as illustrated in (86). Notice that molti esperti in (86a) is originally the direct object of the verb invitare and that (86a) in this respect is derived from a structure that corresponds to (86b), which is also a grammatical sentence in Italian. (86) a. Molti esperti saranno invitati. many experts will be invited 'Many experts will be invited.' b. Saranno invitati molti esperti. will be invited many experts c. Ne saranno invitati molti. of-them will be invited many 'Many of them will be invited.' (Burzio 1986:23, (8a-c)) The observed grammaticality of (86c) is predicted by the generalization we have come to adopt and therefore gives more credence to the generalization itself. When these facts are taken into consideration, the grammaticality pattern in (81) strongly Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 suggests that the NP argument of an unaccusative verb is base-generated in syntactic object position while the subject argument of an unergative and a transitive is not. Burzio further shows that the two types of inchoative verbs in Italian, namely bare inchoatives such as qffondare (which alternates with its causative version without morphological change) and «-inchoatives such as rompere (whose inchoative version is morphologically marked with the clitic si) also manifest the same pattern as the unaccusative verbs. In both (87) and (88), the (a) sentences contain causative verbs, the (b) sentences inchoatives, and the (c) sentences inchoatives with nc-cliticization. The grammaticality of (87c) and (88c) is seen as evidence that the surface subjects in (87b) and (88b) are in fact the D-Structure objects generated in the same structural position as the objects in (87a) and (88a). (87) a. L'artiglieria affondo due navi nemiche. The artillery sank two ships enemy The artillery sank two enemy ships.' b. Due navi nemiche affondarono. Two enemy ships sank.' (Burzio 1986:25, (13a-b)) c. Ne affondarono due. of-them sank two Two of them sank.' (Burzio 1986:25, (14)) (88) a. Giovanni rompe il vetro. Giovanni breaks the glass.' b . D vetro si rompe. The glass breaks.' (Burzio 1986:38, (46b)) (Burzio 1986:38, (46a)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 c. Se ne rompono molti. themselves of-them break many 'Many of them break.' (Burzio 1986:38, (47)) To summarize the discussion in this section, the surface subject of an Italian sentence may appear preverbally or postverbally. When studied in terms of ne- cliticization, superGcially similar subjects show a marked difference. Specifically, we have observed that unlike the surface subject of an unergative or transitive verb, the surface subject of an unaccusative verb syntactically behaves like a direct object of a transitive verb and a surface subject of a passive predicate, which indicates the base generation of the argument of an unaccusative verb as the sister of V. The NP arguments of the two types of inchoative verbs also manifest the same syntactic behavior with respect to nc-chticization, an indication that these arguments are D- Structure objects as well. 2.2.4. Auxiliary Selection Another often cited syntactic property of Italian relevant to the unergative versus unaccusative-inchoative distinction is the selection of aspectual auxiliaries, avere have' and essere 'be'. The sentences in (89) in simple present tense contain unaccusative, unergative and transitive verbs. (89) a. Maria arriva. Maria arrives.' b. Maria telefona. Maria telephones.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 c. Maria compra il tavolo. 'Maria buys the table.' When the sentences are in the present perfect, they select different auxiliaries. (90) a. Maria è arrivata. Maria is-3SG arrived (SG/FEM ) 'Maria has arrived.' b. Maria ha telefonato. Maria has-3SG telephoned Maria has telephoned.' c. Maria ha comprato il tavolo. Maria has-3SG bought the table Maria has bought the table.' In (90), the fam iliar pattern has emerged with respect to the auxiliary selection. The unaccusative verb in (90a) that selects essere contrasts with the unergative and transitive verbs in (90b, c) which select avere. It should also be noted that the past participle in (90a) agrees with the surface subject in number and gender while those in (90b, c) do not. As in the previous section on ne-cliticization, the syntactic significance of this contrast becomes clear once we pay attention to the data on passive sentences. (91) a. Molti studenti furono arrestati. Many students were-3PL arrived (PL/MAS) 'Many students were arrested.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 b . Furono arrestati molti studenti. 'Many students were arrested.' (Haegeman 1994:326, (55a-b)) (9 la) shows the same pattern as (90a) concerning the auxiliary selection and the subject and past participle agreement. Since the surface subject of a passive sentence is generated in object position at D-Structure, this is a crucial piece of evidence that indicates that the distinction between (90a) and (90b, c) is not simply lexical and/or semantic, but has to be syntactic as well. Incidentally, it should be noted that the argument molti studenti can appear postverbally and still induces the selection of essere and the agreement of the past participle. Without going into detailed discussions, let us assume, with Burzio (1986), Haegeman (1994), etc. that the auxiliary selection chooses essere over avere when there is a chain between the subject position and the object position (including the position of some clitics) of the verb. When an internal argument stays in the object position and acquires Nominative Case, agreeing with the verb (and the past participle), we will assume that this argument is in a chain relation with a null expletive in subject position (see section 4.4.2.1). The generalization that the auxiliary verb essere is selected when there is a chain between the subject and the object positions can account for the cases that involve the two kinds of inchoative verbs in Italian previously mentioned, i.e. bare inchoatives and «-inchoatives. When verbs like qffondare and rompere function as causative verbs as in (92a) and (93a), avere is selected as the appropriate auxiliary. When they function as inchoative verbs, however, essere is the right selection. The selection of avere is natural because subject and object positions do not form a chain in (92a) and (93a). Such a chain would clearly violate Theta Criterion (Haegeman Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 1994:54), rendering the derivations ungrammatical. In (92b) and (93b), the surface subjects are considered to have moved out of the object position where they are assigned a theta role appropriate for an internal argument. Therefore, there is a grammatical chain between the subject and the object position. This derivational history correctly accounts for the selection of essere in (92b) and (93b). (92) a. L'artiglieria ha affondato due navi nemiche. the artillery has-3SG sunk two ships enemy "The artillery has sunk two enemy ships.' b . Due navi nemiche sono affondate. two ships enemy are-3PL sunk (PL/FEM) Two enemy ships have sunk.' (Burzio 1986:54, (80a-b)) (93) a. Maria harotto latazza. Maria has-3SG broken the cup 'Maria has broken the cup.' b. La tazza si è rotta. the cup CL is-3SG broken (SG/FEM) The cup has broken.' (Burzio 1986:57, (91a)) In sum, the auxiliary selection data in Italian illustrate that the unergative versus unaccusative-inchoative distinction is syntactically significant. Among other things, it clearly shows that the sole NP argument of an unaccusative-inchoative verb behaves syntactically like the object of a transitive verb while the argument of an unergative verb does not. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 2.2.5. Linear Word Order Burzio (1986:sec. 1.8) presents further data on syntactic position of surface subject to support the hypothesis that the sole NP argument of an unaccusative verb is generated in object position. In this section, I briefly review a part of his discussion that concerns the position of an argument with respect to the verb and an adverb. As the auxiliary verb selection indicates, sparato (from sparare to shoot ) in (94) is not unaccusative (possibly, pseudo-transitive or unergative) while caduto (from cadere to fall') in (95) is unaccusative. (94) Un carabiniere ha sparato addosso al dimostrante. a policeman has fired upon to-the demonstrator 'A policeman has fired on the demonstrator.' (Burzio 1986:67, (116)) (95) Un carabiniere è caduto addosso al dimostrante. a policeman is fallen upon to-the demonstrator 'A policeman has fallen on the demonstrator.' Superficially, the two sentences appear identical except the auxiliary + verb sequence. In both, it is possible to cliticize the prepositional phrase al dimostrante to the auxiliary verb. Therefore, the superficial similarity between the two sentences is still maintained in (96) and (97). (96) Un carabiniere gli ha sparato addosso. a policeman to-him has fired upon A policeman has fired on him.' (Burzio 1986:67, ( 117)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 (97) Un carabiniere gli è caduto addosso. a policeman to-him is fallen upon 'A policeman has fallen on him.' A potential contrast between the two, however, becomes apparent in (98) and (99) where the NP arguments appear postverbally. (98) ??Gli ha sparato un carabiniere addosso. to-him has fired a policeman upon 'A policeman has fired on him.' (Burzio 1986:67, ( 118b)) (99) Gli è caduto un carabiniere addosso. to-him is fallen a policeman upon 'A policeman has fallen on him.' (Burzio 1986:67, ( 119)) In (98) and (99), the surface subject un carabiniere immediately follows the verb phrase but precedes the adverb addosso. As a result, (98) becomes less acceptable while (99) remains perfectly grammatical. Burzio points out that the contrast between sparare and cadere that becomes apparent in (98) and (99) can be elucidated by comparing (99) to (101), which is a passive counterpart of the active sentence in (100). (100) La folia gli spinse un carabiniere addosso. the crowd to-him pushed a policeman upon 'The crowd pushed a policeman over him.' (Burzio 1986:68, (121b)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 6 (101) Glifuspinto un carabiniere addosso. to-him was pushed a policeman upon 'A policeman was pushed over him.' (Burzio 1986:68, (121a)) In (100), un carabiniere is clearly the sister of V since it is the object of the transitive verb spinse (from spingere to push ). In passive where the past participle spinto does not assign Accusative Case to it, this NP may move into subject position where it is assigned Nominative Case by Infl. Then, we obtain the passive sentence in (102). ( 102) Un carabiniere gli £u spinto addosso. a policeman to-him was pushed upon 'A policeman was pushed over him.' (101), on the other hand, is considered to be a passive sentence in which NP movement has not applied to the syntactic object. This kind of passive construction is very rare in English because the lack of NP movement normally leaves the syntactic object Caseless, violating Case Filter.-^ As we have observed in previous sections, however, in Italian this is not always the case. Italian allows the NP argument to remain in object position, receiving Nominative Case in a chain relation with a null expletive in subject position. Seen in this way, un carabiniere in (99) is arguably base-generated in this position. That is, as far as the position of un carabiniere is concerned, (99) and (101) are identical. The low acceptability of (98) also can be seen as evidence of the impossibility to move an argument base generated in subject position into object position since this kind of movement is a Some semantic classes o f English verbs are exceptions to this generalization. See Levin (1993:93). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 lowering operation and results in violation of Empty Category Principle (Chomsky 1986a: 17), which requires a trace created by movement to be properly governed. 2.2.6. Past Participial Adjectival Clause Burzio (1986:sec. 3.2) shows that the distinction between the unergative and the unaccusative verbs is also relevant to the past participial clause construction in Italian. ( 103) a. Un ragazzo [invitato alia festa] non conosceva Maria, a guy invited to the party did not know Maria b. Un ragazzo [arrivato poco fa] conosce Maria a guy arrived a while ago knows Maria c. *Un ragazzo [telefonato a Maria] non puo venire alia festa a guy telephoned to Maria cannot come to the party (based on Burzio 1986:193-194, (40a-c)) While the past participial constructions based on a transitive and an unaccusative in (103a) and (103b) are both grammatically incorporated into whole sentences, the one based on an unergative in (103c) is clearly ungrammatical. Where does the difference come from? Burzio sees the answer to the question in the morphology of Italian passive past participle formation and some universal syntactic principles. First, he argues that the past participial phrases in question are small clauses functioning as relative clauses, which have the following S-Structure representations. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 ( 104) a. [un ragazzo [sc PROi invitato tj alia festa]] non conosceva Maria. a guy invited to the party did not know Maria b. [un ragazzo [sc PROi arrivato ti poco fa]] conosce Maria a guy arrived a while ago knows Maria c. *[un ragazzo [sc [e] telefonato a Maria]] non puo venire alia festa a guy telephoned to Maria cannot come to the party (Burzio's 1986:193-194, (40a-c)) As indicated by the traces and coindexation, (104a) and (104b) are assumed to contain PRO which is base-generated in post-participial object position and later moves to pre-participial position. PRO needs to originate in object position because this is where it receives a theta-roie assigned to an internal argument. Its movement to pre-participial position is motivated by the PRO Theorem, which requires PRO to be ungovemed (Haegeman 1994:272-275,285). The object position is governed by the past participle and consequently PRO cannot remain in situ. The moved PRO receives its reference from the preceding noun phrase in a Control relation, which makes the interpretation in (103a-b) analogous to a regular relative clause construction. This scenario, however, does not work for (103c). The difference between this and the other two is ultimately due to the morphology of the passive past participle. The impact on the argument structure of this process is generally considered to "absorb" the theta-role normally assigned to the external argument (Jaeggli 1986b) or to "suppress" it (Grimshaw 1 9 9 0 ) The argument structure alteration can be described as (105). For our purpose, there is no difference between 'absorption' and suppression'. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 ( 105) Italian Passive Past Participle Formation with the -to Suffix i: Suppress the external argument (shown as "x=%"). [obligatory only if there is an external argument] When applied to a transitive verb, (105) changes its argument structure as in (106). ( 106) Transitive V: (x (y)) — > Passive Past Participle (-to): (x=% (y)) ( 106) shows that the passive past participle of a transitive verb still has an internal argument to project even after its external argument is absorbed. The unaccusative verb undergoes a similar argument structure alteration: ( 107) Unaccusative V: (0 (y)) — > Passive Past Participle (-to): (0 (y)) Here, however, there is no external argument to begin with. Because ( 105) is obligatory only when there is an external argument, it has no observable effect on the unaccusative argument structure. Or in Burzio’ s term, the process vacuously applies to the unaccusative verb class. The crucial point, though, is that in (107) the internal argument is still intact and has to be projected. Now, let us see how ( 105) changes the unergative argument structure. (108) Unergative V: (x (0)) ~> Passive Past Participle (-to): (x=% (0)) In (108), the external argument is absorbed and there is no argument remaining to be projected. This is why no PRO is assumed to be either in the subject position or in the object position of the past participle in (104c). Incidentally, Burzio notes the grammaticality difference between (109) and ( 110). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 ( 109) a. ?*A student arrived yesterday was accepted in the program.^» (Burzio 1986:191, (35a)) (110) *A student applied to the program arrived yesterday. (Burzio 1986:191, (34a)) For most native speakers, both (109) and (110) are clearly less acceptable than ( 111). (111) A student accepted in the program arrived yesterday. (Burzio 1986:189, (30a)) To some, however, (109) is considerably better than (110). Burzio also suggests that there is a similar difference between the French and Spanish phrases in (112). (112) a. Un étudiant arrivé hier soir .... (French) A student arrived last night.... b. ?Un estudiante recientemente llegado de Francia... (Spanish) A student recently arrived from France ... (Burzio 1986:197, (47a-b)) Burzio attributes these interpersonal and crosslinguistic differences to the degree in which the vacuous absorption of the external theta-role is tolerated by speakers of these languages. The degree of such tolerance may be paraphrased as a question of obligatoriness of a rule like (105). Clearly, in Italian, French and possibly Here, the judgment is Burzio's. A sentence o f this kind, however, may not be so uncommon in English. Consider, for instance: (i) Immigrants recently arrived in this country are subject to new laws. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 Spanish, the absorption is not obligatory when there is no external argument. In contrast, it is always obligatorily applied in the grammar of many English speakers, rendering the passive past participle formation on the unaccusative verb class ungrammatical since these verbs do not contain an external argument that the process can absorb. 2.2.7. Past Participial Adverbial Clause Finally, the contrast between transitive and unaccusative verbs on one hand and unergative verbs on the other is also relevant to the grammaticality of past participial adverbial clause constructions, also known as "participial absolutes." In (113), the subject of the main clause, Paolo, corresponds to the internal argument of the past participle arrestato in the adverbial clause. The construction is grammatical. (113) Arrestato dalla polizia, Paolo ha subito un lungo interrogatorio. Arrested by the police, Paul underwent a long interrogation.' (Sorace 1993b, (5a)) In (114), however, Gianni corresponds to the external argument of lavorato, and the sentence is ungrammatical. ( 114) *Lavorato per trent'anni, Gianni andô in pensione. 'Worked for thirty years, John retired.' (Sorace 1993b, (5c)) (115) shows that the past participle caduto of an unaccusative verb cadere forms a grammatical adverbial clause. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 (115) Caduto dal seggiolone, il bambino si è messo a urlare. "Fallen from his chair, the baby started to scream." (Sorace 1993b, (5b)) The examples in (113)-(15), therefore, illustrate that the unaccusative class behaves in the same way as the transitive verb class with respect to the past participial adverbial construction. This fact serves as additional support to the idea that the sole NP argument of an unaccusative verb is an internal argument that corresponds to the object argument of a transitive verb. 2.3. S p a n i s h 2.3.1. Nominalization Based on Present Participles One type of morphological evidence for the existence of two kinds of intransitive verbs in Spanish is presented in Bever and Sanz (1997). As shown in (116), the Spanish present participle suffix -ante (or -ente) can also function as a nominalizer. (116) el fabricante de munecas th e manufacture-PRES PART o f d o lls "the manufacturer of dolls" (Bever and Sanz 1997, (18)) In (116), e/ fabricante corresponds to the nuinitfacturer though it is literally more like the manufacturing one with one incorporated in it. This kind of nominalization also creates well-formed nouns such as those in (117) based on unergative verbs. (117) a. elcantante th e sing-PRES PART the singing man, the singer" Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 b. elviajante th e travel-PRES PART 'the traveling man, the [commercial] traveler* c. la b ella durm iente th e beautiful sleep-PRES PART th e sleeping beau tifu l w o m an , (sleeping beauty)' (Bever and Sanz 1997, (20)) In contrast, a similar nominalization based on unaccusative verbs results in ungrammatical words. (118) a. *el Uegante th e arrive-PRES PART 'the arriving m an' b. *el muriente th e die-PRES PART th e d y in g m an' c. *el apareciente th e appear-PRES PART 'th e appearing m an' (Bever and Sanz 1997, (19)) The contrast between (117) and (118) can be expressed in the A-Structure alteration framework of morphological changes: (119) Nominal Formation with the -ante/-ente Suffix i: R-bind the external argument (shown as "R=x") [obligatory] Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 ii: Nullify the argument structure of the base verb (shown as [obligatory]^^ Both transitive and unergative verbs have an external argument which can be R- bound by the process in (119i), but by hypothesis, unaccusative verbs do not have an external argument Since R-binding of an external argument is obligatory, nominals derived from unaccusative verbs are deemed to be ill-formed. 2.3.2. Bare NP Postverbal Subject The first syntactic evidence we review is concerned with the position of bare NPs. Bever and Sanz (1997) show that Spanish allows NPs with no article in object position but not in subject position. (120) a. Los gatos comen ratones. the cats eat mice Cats eat mice.' b. *Gatos comen ratones. cats eat mice (Bever and Sanz 1997, (6)) (120a) and (120b) minimally contrast with each other in that the subject of (120a) contains an article but that of (120b) does not. In general, bare NPs in Spanish can appear only in governed positions such as the object of a verb or a preposition. When an NP that corresponds to the internal argument of the base verb appears as part of a PP as in (116), I consider it as modification rather than the projection of an obligatory argument. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 This generalization is quite important in accounting for the difference between (121) and (122), both of which contain intransitive verbs followed by postverbal subjects. (121) a. *Anidan cigiienas. shelter storks 'Storks shelter.' (Bever and Sanz 1997, (7a)) b . *Hablaran representatives manana. (will) talk some representatives tomorrow Some representatives talk tomorrow.' (Sorace 1995, (5b)) (122) a. Crecen flores. grow flowers 'Flowers grow.' (Bever and Sanz 1997, (7b)) b. Llegaron invitados a la fiesta, arrived some guests to the party Some guests arrived to the party.' (Sorace 1995, (5a)) Bever and Sanz ( 1997) argue that the grammaticality of ( 122) indicates that the apparent postverbal subjects of the unaccusative verbs are actually in object position while those of the unergatives in (121) are not. 2.3.3. Past Participial Adjectival Clause Schroten (1986) and Bever and Sanz (1997) show that Burzio's (1986:sec. 3.2) past participial clause construction test discussed in section 2.2.6 also works as a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 diagnostic of Spanish unaccusative verbs. As in Italian, the past participial adjectival clause construction is typically based on a transitive verb. (123) a. La casa [construida con mucho cuidado] resistio al huracàn. The house built with much care resisted the hurricane.' (Schroten 1986, (15)) b . Un estudiante [sobomado] suspendiô el examen. 'A student bribed failed the exam.' (Bever and Sanz 1997, ( 1 lb)) A noun phrase modified by a past participial clause such as la casa in (123a) corresponds to the internal argument of the base transitive verb of the past participle. When a modified noun phrase corresponds to the external argument of a transitive verb as in (124) or to the external argument of an intransitive verb as in (125a-c), the whole sentence is rendered ungrammatical. (124) *Los chicos [construidos la casa] descansaron largo rato. The boys *(who) built the house rested a long time.' (Schroten 1986, (14b)) (125) a. *Un nino [llorado] pinto los dibujos. 'A child *(who) cried made the paintings.' (Bever and Sanz 1997, (9b)) b . *Un estudiante [dormido por la manana] suspendiô el examen. 'A student slept in the morning failed the exam.' (Bever and Sanz 1997, (13b)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 c. *Los chicos [gritados a las nueve] Iloraron a las dlez. 'The boys *(who) shouted at nine o'clock cried at ten o'clock.' (Schroten 1986, (14a)) Examples (126a-b), however, show that some intransitive verbs can appear in the past participial adjectival clause construction. (126) a. Los chicos [salidos de la casa a las nueve] no ban telefoneado. The boys *(who) went out of the house at nine o'clock have not telephoned.' (Schroten 1986, (11)) b. Un estudiante [llegado por la manana] suspendiô el examen . 'A student (who) arrived in the morning failed the exam.' (Bever and Sanz 1997, (14b)) In (126a), for instance, the intransitive verb salir appears in its past participle form, followed by a prepositional phrase de la casa a las nueve. The whole participial clause modifies the preceding noun phrase los cA/coj just as a relative clause would. The observations made on the relation between the modified noun phrases and the modifying past participial clauses in the examples (123)-(126) clearly indicate that the subjects in (123) and (126) and the subjects in (124) and (125) contrast with each other. The former correspond to the internal arguments of the respective participles while the latter correspond to the external arguments of the participles. Schroten accounts for the grammaticality pattern evident in (123)-(126) along the line of Burzio's explanation for the Italian data in section 2.2.6. First, the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 grammaticality of (123) and (126) can be accounted for by assuming the existence of PRO, which is base-generated in object position of the past participles but moves out to subject position, avoiding the violation of the P/ÎO-theorem. The PRO receives its reference from the subject NP of the sentence in a Control relation. Second, the ungrammaticality of (124) and (125) can be attributed to the lack of PRO within the past participle clauses in these examples. PRO cannot be assumed to exist for the external argument of the participles because an external argument is absorbed by the morphological process that forms the passive past participle. PRO caimot be an internal argument of the past participles either because there is already such an argument in (124), i.e. la casa, and the unergative verbs in (125) do not have an internal argument by definition. Applying the past participial clause construction as a diagnostic, Schroten identifies the following unaccusative verbs. bajar 'to go (down)' caer to fall' emigrar to emigrate' llegar to arrive' salir 'to go (from)' saltar 'to jump' subir 'to go (up)' pasar to pass' venir 'to come' volver to return' (Schroten 1986, (16)) He regards these unaccusative verbs semantically as verbs of movement. Noting the existence of verbs such as those in (128) which also pass the past participial clause construction test, however, he concludes that not all unaccusative verbs are movement verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 (128) brotar 'to sprout' ocurrir to happen' nacer to be bom' pasar 'to happen/overcome' (Schroten 1986, (17)) The verbs in (128) correspond to the semantic class of "appearance" verbs. In short, the verbs in (127) and (128) together squarely fall in the category of unaccusatives semantically as well as syntactically. 2.3.4. Past Participial Adverbial Clause Past participles in Spanish can form adverbial constructions such as (129a)30 In these constructions, the past participles are followed by an object with which they agree in gender and number. If the subject argument is expressed by the por phrase (equivalent to the English by phrase) as in (129b) or if it agrees with the past participle as in (129c), the constructions are ungrammatical. (129) a. Examinado el caso eneljuzgado, exam ined-PAST PART-MAS the case-M AS in the court la m agistrada pronuncio la sentencia. the female judge pronounced the sentence 'Once she had examined the case in court, the judge pronounced the sentence.' b . * E xam inado el caso por la m agistrada, exam ined-PA ST PART-MAS the case-M AS by the fem ale ju d g e pronuncio la sentencia. (she) pronounced the sentence. 30 See a similar but slightly different construction in Italian in section 2.2.7. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 c. *La magistrada examinada el caso, the fem ale j udge exam ined-PAST PART-FEM the case-MAS pronunciô la sentencia. (she) pronounced the sentence. (Bever and Sanz 1997, (8)) As shown in (130) and (131), past participial constructions of this kind are also possible with unaccusative verbs but not with unergative verbs. (130) L legado el m inistro, com enzo la reunion. arrived-PAST PART-MAS the m inister-M AS started the m eeting T h e m in ister having arrived, the m eetin g started.' (Bever and Sanz 1997, (10a)) (131) *U n a vez llorado el nino, vino su m adre. on ce cried-PAST PART-MAS the child-M AS cam e his m other 'O nce the c h ild cried, h is m other cam e.' (B ever and S anz 1997, (9a)) The contrast between (130) and (131), therefore, not only shows the distinction of the two kinds of intransitive verbs in Spanish, but also indicates that the surface subject of unaccusative verbs shares a syntactic property with the object of transitive verbs. 2.3.5. Arbitrary Null Plural Pronominal In section 2.1.5, we have discussed Jaeggli's (1986a) observation concerning the interpretive possibility of the English third person plural pronoun they. Specifically, we have seen that in a certain syntactic environment, this pronoun may obtain both definite and arbitrary interpretations with respect to its potential referent. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 Carefully analyzing three arbitrary pronominal constructions, Jaeggli presents convincing arguments that the null plural pronominal pro is the Spanish counterpart of the arbitrary they in English. Jaeggli's analysis has significant implications on the existence of two subclasses of intransitive verbs in Spanish. First contrast (132) and (133) below. (132) Elios llam an a la puerta. they call-3PL at th e door. 'They are knocking at the door.' (Jaeggli 1986a, (5)) (133) pro llam an a la puerta. call-3PL at th e door 'Arb^^ is knocking at the door.' (Jaeggli 1986a, (3)) The sentences (132) and (133) minimally contrast with each other in that the former has an overt pronominal subject ellos while the latter lacks it. Theoretically, however, it is commonly assumed that there is a phonetically null pronominal element in subject position in (133) as well, which is known as pro (= small pro). This superficial contrast, however, carries over into the interpretive possibility of these two sentences. On the one hand, (132) is an unambiguous sentence, allowing only the interpretation of ellos referring to a definite group of people (at least two individuals, excluding the speaker) knocking at the door. On the other hand, the null subject pro in (133) is ambiguous, allowing both definite and arbitrary readings. That is, in addition to the reading of (132), (133) can have another reading in which the agent of the action is left unspecified. In this arbitrary reading Following the glossing convention of Jaeggli (1986), I indicate the referent of a potentially arbitrary pro as Arb in the translation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 of pro, Jaeggli observes, the referent does not even need to be a group of individuals but could be a single person. Here, the feature [+ plural] apparent in the verb form is seen as purely "formal", giving no information concerning the number of the referent. The arbitrary pro may appear quite naturally in Spanish sentences such as the following ones. (134) a. pro forzaron lapuerta para entrar a robar. forced-3PL the door for to-enter to rob 'Arb forced th e d o o r to enter to rob.' b. pro vendieron diarios e n la e s q u in a . sold-3PL new spapers in the com er 'Arb sold new spapers at the com er.' (based on Jaeggli 1986a, (I4 a -b )) (135) a. Aquf pro durmieron hace dos dias. here slept-3PL ago tw o days H ere Arb slep t tw o days ago.' b. pro llam aron pro teléfono para avisar que la m ercaderfa esta lista. call-3PL by phone for to-say that the m erchandise is ready 'Arb telephoned to say that the m erchandise is ready.' (Jaeggli 1986a, (12a-b)) The verbs in (134a-b) and (135a-b) are transitives and unergatives respectively. The third person plural pro is in subject position in each sentence and can have both Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 definite and arbitrary readings. The arbitrary pro is also observed to appear in various types of embedded clauses as long as it is in the subject position of such clauses (Jaeggli 1986a, (6a-d)). The syntactic restriction of the arbitrary pro to subject position appears to be a good first approximation, especially in light of additional data such as the following. (136) a. Juan los vio [pro, 3 p l ] CL saw Juan saw them.’ b . Juan los considéra [[p ro , 3PL tontos]. CL considers stupid Juan considers them stupid.’ c. Juan los vio [[pro, 3PL] salir del banco]. CL saw to-Ieave of-the bank Juan saw them leave the bank.’ (Jaeggli 1986a, (7a-c)) In ( 136), each pro in object position is associated with a clitic pronoun on the verb. Unlike (133)-(135), the arbitrary reading is not allowed for the pros in object position. They are only interpreted as definite in reference. Therefore, the examples in (136) at first seem to support the ’ ’ subject-only” restriction concerning the syntactic environment where an arbitrary pro can appear in Spanish. As is the case in English, however, the ’ ’subject-only” condition for the arbitrary pro turns out to be inadequate in light of further data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 (137) a. Fueron asesinados por criminales. were-SPL assassinated by criminals 'They were assassinated by criminals.' b. Fueron arrestados porlapciicia. were-3PL arrested by the police They were arrested by the police ' (Jaeggli 1986a, (8a-b)) The sentences (137a-b) are both passives in which the verbs' internal arguments, i.e. the arguments that correspond to 'they' in the translations, are not phonetically salient. Jaeggli shows that these sentences cannot be interpreted as having arbitrary pro, but that instead, they are understood as having a definite reference, as if they contained a lexical pronominal subject elles 'they'. The "subject-only" condition fails in (137a-b) because we carmot obtain the arbitrary pro reading for these sentences although we can safely assume that there is a derived subject in them. Similarly, the arbitrary pro cannot appear as a derived subject in unaccusative sentences either: (138) a. pro Uegan cansados despues de un viaje tan largo. arrive-3PL tired after of a trip so long They arrive tired after such a long trip.' b. pro salen mucho de noche durante el verano. go out-3PL much of night during the summer They go out at night a lot during the summer.' c. pro mueren en defensa de la democracia. die-3PL in defense of the democracy They die in defense of democracy.' (Jaeggli 1986a, (lOa-c)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 Here again, the only interpretation available for pro is a definite reference reading as if there were a lexical subject elles in each one of these sentences. The "subject- only" condition, therefore, fails in these examples, too. Moreover, Jaeggli argues that the unavailability of an arbitrary reading in (138) has nothing to do with some unknown semantic properties of the unaccusative verbs by demonstrating that the arbitrary pronominal interpretation is possible in the impersonal se construction such as (139). (139) a. Se llega cansados despues de un viaje tan largo. ICL arrives-3SG tired after of a trip so long 'Arb arrives tired afte r su ch a long trip.' b . Se sale mucho de noche durante el verano. ICL goes out-3SG m uch o f night during the su m m er 'Arb goes o u t at night a lo t d u rin g the sum m er.’ c. Se muere en defensa de la democracia. ICLdies-3SG in d efen se o f the dem ocracy 'Arb dies in defense of democracy." (Jaeggli 1986a, (lla-c)) The examples in (139) strongly suggest that the unavailability of the arbitrary pro interpretation in (138) as well as (137) is not a semantic problem but a syntactic one. So far we have reviewed Jaeggli's (1986a) observation that the arbitrary plural pro in Spanish is allowed in active transitive and unergative sentences while it is excluded from passive and unaccusative structures. To account for this Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 distributional characteristic of the arbitrary pro, Jaeggli resorts to the same syntactic restriction as in the case of the English arbitrary they, namely: ( 140) An arbitrary plural pronominal cannot be in a chain that is Case- or theta- marked directly by a verb. (based on Jaeggli 1986a, (34)) (140) correctly predicts that in cases such as (133), (134) and (135) above, the null plural pronominal will obtain an arbitrary reading because the external argument of the transitive and unergative verbs is neither Case-marked nor ±eta-marked directly by the verb. (140) accounts for the unavailability of an arbitrary reading for the pros in (137) and (138) as well. In the passive and unaccusative sentences, the surface subjects are originally internal arguments of the verbs and have moved into surface subject position from their initial syntactic position within V at D-Stnicture. Although the subjects are Case-marked by Infl, the theta role is directly assigned to the tail of the respective chain by the verbs. In this way, Jaeggli's theory can give a unifying account to both English they and Spanish pro with respect to their interpretive possibility .^ 2 What is critically important for us is that his account of arbitrary plural pronominals illustrates the crucial difference between the unergative and the unaccusative classes in terms of their syntactic properties. In short, Jaeggli's observation clearly shows that the argument of the unergative class patterns with the subject of a transitive verb while the argument of the unaccusative class shows a similarity with the object of a transitive verb. 22 In Jaeggli's theory, the difference between the overt third person plural pronoun ellos and the arbitrary pro with respect to their interpretive possibility is accounted for in terms of the value of the feature [+/- dependent]. For detail, the reader is referred to Jaeggli (I986a:sec. 5.1). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 2.3.6. Effects on Sentence Processing Finally, there is some psycholinguistic evidence which indicates that unergative and unaccusative verbs have distinct syntactic and conceptual representations. Bever and Sanz (1997) studied the processing of sentences with both types of intransitive verbs. The subjects were asked to read sentences in segments shown on a computer monitor and answer whether a word presented immediately afterwards was contained in the preceding sentence. The recognition words were adjectives that were part of the subject NPs of the experimental sentences. In the study, two types of experimental subjects were identified according to the processing strategies they primarily employed. One group of subjects primarily accessed a syntactic representation of the sentence's form, and the other group accessed a conceptual representation of its meaning. Bever and Sanz called the syntactic and conceptual scanners "sequence-sensitive" and "sequence-insensitive" subjects respectively. Although no significant difference was found between the two groups of subjects on the processing of sentences with postverbal subjects, categorically different processing patterns were obtained when they answered on sentences with preverbal subjects. The "sequence-sensitive " subjects answered significantly faster in recognizing the adjectives contained in the subjects of unaccusative verbs than those contained in the subjects of unergative verbs. The results obtained from the "sequence-insensitive" subjects were the complete opposite. Bever and Sanz attribute the performance of the " " sequence-sensitive" subjects to the trace facilitation effects, which have been supported by previous research such as Bever and McElree 1988 and MacDonald 1988. The idea is that the trace created by NP-movement in constructions such as the to«g/i-sentence and the verbal passive has priming effects because it provides an extra representation of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 its antecedent and consequently a separate instance of its referent in the syntactic representation. The preverbal subject of an unaccusative verb originates in object position but moves to surface subject position leaving a trace behind. Therefore, the recognition speed for adjectives contained in the preverbal subjects of unaccusative verbs is expected to be faster because the antecedent phrases themselves are more salient in the mental representation of the sentence structures. This expectation was borne out by the result that Bever and Sanz obtained from the "sequence-sensitive" subjects. As for the reversed pattern of performance of the "sequence-insensitive" subjects, Bever and Sanz argue that this is because the conceptual representation of unaccusative verbs is more complex than that of unergatives. The event structure of unergative verbs consists of the activity phase alone, but that of unaccusatives consists of the state or change of state phase subsumed by the activity phase (Grimshaw 1990, Levin and Rappaport Hovav 1995). If the "sequence- insensitive" subjects accessed the conceptual representations of the experimental sentences by bypassing their syntactic representations, the more conceptually complex event structure of unaccusative verbs could delay the subjects' attempt to locate the recognition words when they were part of the subjects of unaccusative verbs. In sum, Bever and Sanz (1997) present psycholinguistic evidence that unergatives and unaccusatives contrast with each other both in their syntactic and conceptual representations. 2 .4 . J a p a n e s e The distinction between unergatives and unaccusative-inchoatives in Japanese is not very obvious at first. Japanese is an SOV language which has a rigid verb-final Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 condition on a sentence. In an intransitive sentence, therefore, a nominal argument always precedes the verb, masking the effect of movement, if any, on the sentential word order. Japanese also lacks a salient syntactic phenomenon such as auxiliary selection in Italian. In fact, the distinction of the two kinds of intransitive verbs has hardly received serious attention in traditional study of the Japanese language (Kageyama 1993:42,48). Recently, however, researchers in the generative linguistic tradition have turned their attention to a number of subtle linguistic phenomena which show that the intransitive verbs in Japanese also fall into two distinct classes. 2.4.1. Inchoatives and Causatives First of all, as in English and Italian, the Japanese lexicon contains ergative verbs which have both inchoative and causative usage: (141) inchoative causative hiraku 'to open/unfold' hiraku to open/unfold' tqjiru 'to close/fold' tojiru to close/fold' However, ergative pairs with an identical form as in (141) are rather rare among the native Japanese.^^ Most inchoative-causative pairs have similar but non-identical forms (Miyagawa 1989:97, Tsujimura 1996:430). Some of the pairing patterns are shown in (142).^^ 33 Ergative pairs o f this kind are more common among Sino-Japanese compound verbs that contain the light verb suru. Some examples are kakudai-suru to expand', syukusyoo-suru to shrink', kaizen-suru to improve', kasoku-suru to increase in speed', etc. 34 The correspondence expressed in each pair of the above examples is a morphological one, and contrasts with the (morpho-)syntactic causatives such as (i) below. (i) a. aruk-u 'walk' aruk-aseru 'cause to walk' b. hasir-u 'run' hasir-aseru 'cause to run' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 inchoative causative aku to open' akeru to open' agaru to go up/rise’ ageru to raise' sagani to go down' sageru to lower' simaru to close/shut' simeru to close/shut' tamaru to gather' tameru to save' kowareru to break' kowasu to break' kuzureru to crumble' kuzusu to demolish' taoreru ■ to fair taosu to fell/push down kireru to snap' kiru to cut' wareru to break/split' waru to break/split' moreru leak' morasu 'leak' tokeru 'dissolve' tokasu dissolve' okoru occur' okosu 'cause' okiru 'wake up' okosu wake up' c. tabe-ru eat' tabe-saseru 'cause to eat' d. nom-u 'drink' nom-aseru 'cause to drink' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 (141) and (142) show that Japanese ergatives differ firom English in having very few verbs alternating over an identical form, and firom Italian in displaying quite irregular morphological patterns for ergative pairs.^s Just as with English and Italian ergative pairs, however, the semantic role carried by the object argument in the causative sentence remains the same in the corresponding inchoative sentence in the Japanese ergative pairs. That is, as shown in ( 143)-(145), the change in grammatical roles firom object to subject does not affect the semantic role expressed by each NP argument. (143) a. Kodomoga madoo kowasita. child NOM window ACC broke (CAUS) A chUd broke a window.' b. Mado ga kowareta. window NOM broke (INCH) 'A window broke.' (144) a. Taiyooga yukio tokasita. sun NOM snow ACC melted (CAUS) The sun melted snow.' b. Yuki ga toketa. snow NOM melted (INCH) 'Snow melted.' For possible effects of this irregularity on the classification of verbs by a language acquirer, see Rispoli (1988). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 (145) a. Otokoga doao aketa. man NOM door ACC opened (CAUS) 'A man opened a door.' b. Doa ga aita. door NOM opened (INCH) 'A door opened.' Strictly speaking, the existence of inchoative-causative correspondence itself does not affirm or deny the kind of unaccusative hypothesis that we have adopted in this study, i.e. Levin and Rappaport's definition ±at the unaccusativity is semantically determined but syntactically represented. Such a correspondence, however, does show that there are a group of intransitive verbs whose nominal arguments share semantic characterization of the direct internal argument of the morphologically identical or related causative verbs. 2.4.2. Construction with -te-iru The two types of intransitives contrast in the interpretation of a verbal phrase (Miyagawa 1989, Tsujimura 1991). The Japanese verb iru contrasts with aru in that the former means exist' when the subject of the sentence is an animate entity while the latter is used for the same sense when the subject is inanimate. Another common usage of iru is that of an auxiliary or secondary verb, which follows the - te form of the main verb, a nonfinite form similar to the English present participle. The ordinary interpretation of this verbal phrase with a regular transitive verb is that of a progressive event.36 36 What is relevant in the following examples is the most neutral reading of the sentences. The interpretation of V-te iru is sometimes influenced by both contextual and linguistic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 ( 146) a. Taroo ga bon o yonde iru. T. NOM book ACC read exist Taroo is reading a book.’ b. Kodomo ga ari o mite iru. cbild NOM ant ACC look exist 'Tbe cbild is looking at ants.' c. Torakku ga ookina oto o tatete iru. truck NOM big noise ACC make exist "A truck is making big sound." Wben tbe main verb is an intransitive verb sucb as those in ( 147), tbe sentence is also interpreted as a progressive one. (147) a. Taroo ga aruite iru. T. NOM walk exist Taroo is walking.' b. Kodomo ga waratte iru. cbild N O M smile exist Tbe cbild is smüing.' c. Torakku ga ugoite iru. truck NOM move exist 'A truck is moving.' factors such as the existence of adverbs and other modifiers (Jacobsen 1992). Since we are concerned here with classification of verbs, we look at neutral linguistic contexts in which the verbs' lexical properties are reflected most straightforwardly. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 Not all the intransitive sentences, however, show the same pattern of interpretation, concerning the V-te iru verbal phrase. (148) a. Tukiga (mada) dete iru. moon NOM (still) go out exist The moon is (still) out.' b. Okyakusantati ga (moo) tuite iru. guests N O M (already) arrive exist The guests have (already) arrived.' c. Mado ga warete iru. window NOM break (INCH) exist 'The window is broken.' In ( 148), the V-te iru verbal phrase does not mean the action expressed by the main verb is an ongoing one. That is, tuki ga dete iru does not mean 'The moon is coming out.' Nor does okyakusantati ga tuite iru mean 'The guests are arriving.' Instead, the V-te iru phrases in (148) all indicate a state that is brought about when the event expressed by each main verb is completed. This is clearer with adverbs such as mada 'still' and moo 'already', but even without them the natural interpretation of (148a-c) is stative. What is relevant in (146)-(148) appears to be the event structure of the main verb. If the main verb is a regular transitive verb or an unergative verb, the external argument of the verb corresponds to a participant of activity phase of its event structure. What is immediately affected by the V-te iru verbal phrase is this activity phase and the result is the progressive reading. In contrast, if the verb does not have a participant argument in the activity phase of its event structure as in the case Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 of unaccusative-inchoative verbs with the change of state/location sense, V-te iru instead focuses on the state/location change where the internal argument of the verb corresponds to its participant. This interpretation also explains why V-te iru is not feUcitous with simple "existence" verbs such as aru "be/exist (inanimate)' and iru "be/exist (animate)' but perfectly appropriate with other "existence" verbs such as nokoru remain/stay behind' and todomaru remain/stay behind'. Unlike the simple existence' verbs which do not presume any change of location, both noroku and todomaru have a sense of staying put when the moving to another location is expected. Therefore, verbal phrases such as nokotte iru and todomatte iru are perfectly grammatical in expressing the lack of locative change. 2.4.3. NV Compounds The distinction between unergative and unaccusative-inchoative verbs also manifests itself in the formation of two kinds of compounds in Japanese. The first is the formation of deverbal nominal compounds in the form of [N V-stem] (Miyagawa 1988, 1989, Kageyama 1993). NV compounds are abundant in Japanese and typically consist of a noun and the stem of a transitive verb: (149) a. ne-age price raise 'price-raising' d. doobum-ijime animal abuse animal-abusing' b. hito-sagasi c. mati-zukuri person search town make 'person-searching' community-building' e. hon-yomi f. nihon-tataki book read Japan hit book-reading' 'Japan-bashing' (Kageyama 1993:50, (14a)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 As illustrated in (149), in this type of compound, the nominal constituent semantically corresponds to not an external, but an internal argument of the transitive verb. NV compounds, however, are not limited to the compounds whose righthand constituents are transitive verbs. a. kokoro-gawari b. mune-yake c. ji-suberi heart change chest bum earth slide change of mind' heartburn' landslide' d. gake-kuzure e. ama-mori f. ne-agari cliff collapse rain leak price rise 'landslide' a leak in the roof 'price-hike' g. sina-gire h. yuki-doke goods cut snow melt 'out of stock' 'thawing of snow' (Kageyama 1993:50, (15a), Miyagawa 1988) St (150) with (151). a. *tori-tobi b. *inu-boe c. *kodcmo-odori bird fly dog howl child dance bird flying' dog howling' 'child dancing' d. *niwatori-aruki e. *tomodati-warai chicken walk friend laugh 'chicken walking’ 'firfend laughing’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 Unlike the compounds in (150), those in (151) are all ill-formed although both groups of compounds are made of a noun and an intransitive verb. Therefore, as long as we regard the intransitive verbs as a monolithic class, this difference is difficult to account for. However, once the verbal constituents in (150) and (151) are recognized as inchoatives and unergatives respectively, not only the distinction between them but also the commonality between (149) and (150) become clear. When an NV compound is well-formed, the nominal constituent corresponds to the internal argument of the verbal constituent. The well-formedness of the examples in (149) and (150) are justified because both transitive and unaccusative verbs have an internal argument in their A-Structure representation.^^ Since unergative verbs do not have an internal argument in the first place, NV compounds derived from them cannot be grammatical.^» 37 The contrast between (150) and (151) may appear to be due to the animacy factor. The idea is hard to argue against because the external argument o f an unergative tends to be animate and the internal argument o f an inchoative is often inanimate. The well-formedness of NV compounds, however, cannot be totally attributed to the animacy o f the lefthand constituent because of the grammatical compounds such as (149b, d). If we adopt the semantic account for the intransitive-based compounds, therefore, we need an additional explanation for the transitive-based ones. The dual account is clearly less economical. 3» Incidentally, there are some NV compounds based on unergative verbs, which may appear to be apparent counter-examples to the above statement (i) a. kaeru-oyogi b. inu-kaki c. usagi-tobi frog swim dog scratch hare jum p frog swimming' dog swimming' rabbit jump' (Kageyama 1993:50, (15b)) These cases, however, do not constitute true counter-examples against the general account presented here. Unlike the compounds in (150), these examples do not denote actual events but instead refer to the characteristic manner in which the activities expressed by the base verbs are performed. For example, while ne-agari and yuki-doke in (150) refer to the events in which the price goes up and the snow thaws, kaeru-oyogi and usagi-tobi do not mean actual events in which a frog swims and a hare jumps. Instead, these compounds refer to styles of swimming and jumping that may be performed by other animals, typically humans, when the manner of the activity reminds us of these creatures. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 2.4.4. W Compounding The second type of compound, which is relevant to the unergative versus unaccusative-inchoative distinction, is W compound. Kageyama (1993:50-52) illustrates how three verbs, namely, mawaru move about' (unergative), naosu 'fix/correct' (transitive), and otiru 'fall' (unaccusative), behave as the righthand constituent of W compounds with respect to the type of a lefthand constituent verb. As (152) shows, the unergative mawaru and the transitive naosu can form a well-formed compound either with an unergative verb or a transitive verb. In contrast, the unaccusative otiru can form a grammatical compound only with another unaccusative verb. (152) a: mawaru (unergative^ move about/around' unergative + mawaru: abare-mawaru act violently rampage' transitive + mawaru: sagasi-mawaru look for look around for' unaccusative + mawaru: *fuki-mawaru (wind) blow blow around' aruki-mawaru walk walk around' kai-mawaru buy 'buy' *oti-mawaru fall fall around' mge-mawaru escape run for cover' arasi-mawaru ruin ransack' *nagare-mawaru flow 'flow about' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 b: naosu (transitive^ 'fix/correct' unergative + naosu: suwari-naosu sît 'sit again’ transitive + naosu: tutumi-naosu wrap 'wrap again' unaccusative + naosu: *korobi-naosu tumble 'tumble again' c: otiru (unaccusative) 'fall' unergative + otiru: *hasiri-otiru run 'run-fair transitive + otiru: *arai-otiru wash 'wash-fair 'wipe-fall' ne-naosu sleep 'sleep again' tukuri-naosu make make again' *oti-naosu faU 'fall again' *odori-otini dance 'dance-fair *nugui-otiru wipe narabi-naosu wait in line 'wait in line again' nuri-naosu paint 'paint again' *tuki-naosu arrive 'arrive again' *tobi-otiru jump jump-fair *kiri-otiru cut cut-fair Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 0 unaccusative + otiru: kobcre-otiru Icuzure-otiru moe-otiru spill collapse bum 'spill-fair 'collapse-fall* 'bum-fall' (excerpted from Kageyama 1993:51, (18)) Clearly, in (152), unergative and transitive verbs form a group from which unaccusative verbs are excluded. Kageyama attributes the above observation on W compounds to the fact that both unergative and transitive verbs have an external argument while unaccusatives do not. That is, in the combination of two unergatives, two transitives, or an unergative and a transitive, the two constituents of the compound can share the external argument This is not possible when one is an unaccusative and the other is an unergative or a transitive. The unaccusative verb, by definition, does not have an external argument and therefore cannot share the external argument with either an unergative or a transitive. Two unaccusatives on the other hand can share their internal argument and can form a grammatical W compound. In addition to this lexico-syntactic account, there appear to be conceptual (or lexico- semantic) reasons for the patterns of well-formedness in ( 152a-b). When the unergative verb mawaru or the transitive verb naosu is the righthand constituent of a W compound, it modifies the activity expressed by the lefthand constituent verb. In the case of mawaru, it adds a sense of manner while naosu conveys a sense of redoing for a better result. However, unaccusative verbs do not have a participant in the activity phase of their event structure. Therefore, they cannot be modified by either mawaru or naosu. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 1 2.4.5. Case Marker Drop The unergative versus unaccusative-inchoative distinction in Japanese is also manifested in the morphosyntactic realization of Case marking. Japanese exhibits what is sometimes called Case Marker Drop (Saito 1983, Kageyama 1993). This phenomenon usually refers to the fact that the Accusative Case particle o is more naturally omitted compared to the Nominative Case particle ga, particularly in colloquial Japanese as in (153).39 (153) Kodomotati *(ga) hon (o) yomu no o mitayo.'‘ ° children (NOM) book (ACC) read NOML ACC saw SFP I have seen children read books.' (based on Kageyama 1993:56, (24). Glossed and translated by HO) Now observe the contrast that becomes apparent between (154) and (155) when the Nominative Case Marker ga is omitted. (154) a. Tanaka-san (ga) nakunatta no o siranakatta. T . (NOM) die-PAST NOML ACC know-NEG-PAST (I) did not k n o w th a t T. h a d d ie d .’ b. K inoo, kazan (ga) bakuhatusuru no o m ita yo. yesterday, volcano (NOM) erupt NOML ACC saw SFP 'Y esterday, (I) saw a v o lc an o eru p t.' (based on Kageyama 1993:56, (25a), glossed and translated by HO) The following examples are all embedded under a matrix sentence in order to avoid the reading of an NP without ga as the topic of the sentence. ^ The contrast in question becomes clearer when one of the Case markers is retained while the other is dropped. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 2 (155) a. Osiego *(ga) katuyakusita no o siranakatta o ne's student (NOM) did-w ell NOML ACC know-NEG-PAST (I) d id n 't know that m y stu d en t h ad d o n e w ell.' b . T yuu k ak u h a *(ga) d em o su ru no o m ita yo. extrem e leftist group (NOM) dem o n strate NOML ACC saw SFP '(I) saw T yuukakuha dem o n strate.' (based on Kageyama 1993:56, (25), glossed and translated by HO) The verbs, nakunatta and bakuhatusuru in (154) and katuyakusita and demosuru in (155), are clearly intransitive verbs because they require only one obligatory nominal argument. The contrast is naturally accounted for once we realize that nakunatta and bakuhatusuru are unaccusatives and katuyakusita and demosuru are unergatives. Therefore, what these examples show is that the subject of an unaccusative verb may appear without the Nominative Case particle while the subject of an unergative verb carmot. Kageyama further argues that the proper description of the Case Marker Drop cannot be stated in terms of the type of Case marker, since in (153) it is the Accusative particle o that is omitted while in (154) it is the Nominative particle ga. Instead, following Saito (1983), Kageyama (1993) argues that what is crucial is the syntactic position of the nominals. In both (153) and (154), the nouns without a Case particle are presumably in the sister position of the verb at S- Structure where they are base-generated. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 2.4.6. "Floating" Numeral Quantifier Still another type of evidence for the division of the two kinds of intransitives comes from the behavior of Numeral Quantifiers (NQs) (Miyagawa 1988, 1989, Tsujimura 1996). In Japanese, an NQ is made of a numeral and a classifier suffix such as -satu for volumes of books and -nin for the number of people. As shown in (156), the NQ normally appears in the position immediately after the nominal with which it is semantically associated. (156) a. (Kinoo) gakusei ga hono3-satu katta. yesterday student NOM b o o k ACC 3-CLF b o ught 'A stu d en t b o ught three books (y esterday).' b . (K inoo) gakusei ga 3-nin hon o katta. yesterday student NOM 3-CLF b o o k NOM bought T h ree students b o ught books (y esterday).' In (156), the NQs 3-satu and 3-nin are semantically associated with the immediately preceding object argument hon and the subject argument gakusei. The syntactic relation between the argument and the NQ is considered to be one of predication (Miyagawa 1989). This is clearly reflected in the English translation given under each example. Observe, however, the following contrast between (157a) and (157b). (157) a. H ono kinoo gakusei ga 3-satu katta. book ACC yesterday student NOM 3-CLF bought 'A student bought three books yesterday.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 b. *Gakusei ga kinoo hon o 3-nin katta. student NOM yesterday book ACC 3-CLF bought Three students bought books yesterday.’ c. Gakusei ga kinoo 3-nin hono katta. student NOM yesterday 3-CLF book ACC bought Three students bought books yesterday.' Assuming the VP-intemal subject hypothesis (Fukui and Speas 1986, Kuroda 1988) and the VP-extemal generation of the time adverb such as kinoo, the object hon and the subject gakusei are scrambled out of their respective base positions in ( 157). Consequently, neither NQ immediately follows the nominal argument with which it is semantically associated. Despite these shared derivational characteristics, however, (157a) is grammatical while (157b) is not. (Notice, however, (157c) is perfectly grammatical with the intended sense of (157b).) In order to account for this contrast, Miyagawa (1988, 1989) proposes a mutual c-command condition for the argument and the NQ as a crucial structural requirement for grammaticality: (158) Mutual C-Command Requirement: For a predicate to predicate of a NP, the NP or its trace and the predicate or its trace must c-command each other. (Miyagawa 1989:30, (30)) The sentences (I56a-b) clearly satisfy the condition of (158) since the NP and the predicated NQ c-command each other in the respective sentences as shown in (159). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 (159) a. (kin o o ) [vp gakusei g a [ v [hon o 3-satu] katta]] b . (kin o o ) [vp [gakusei g a 3-nin] [ v h o n o katta]] Similarly, in (160a) below, which corresponds to (157a), the trace of the scrambled hon and the NQ mutually c-command each other. In (160b), on the other hand, the mutual c-command requirement is not met. Contrast (160b) with (160c), which corresponds to the grammatical (157c). (160) a. h o n i o kinoo [ v p gakusei g a [v’ [ti 3-satu] katta]] b . g a k u s e iig a kinoo [y p ti [ y h o n o 3-nin katta]] c. g akuseii g a kinoo [yp tf 3-nin [ y hon o katta]] In this w ay, (158) can account fo r the d ata su ch as (156) and (157). Furthermore, the interpretation of a direct passive sentence with an NQ gives additional support to Miyagawa's mutual c-command requirement in (158). (161) a. Takaihonga doroboo ni 3-satu nusum-are-ta. expensive book NOM thief by 3-CLF steal-PASS-PAST Three expensive books were stolen by a thief.' b . Gakusei ga anootokoni 3-nin koros-are-ta. student NOM that man by 3-CLF kill-PASS-PAST Three students were killed b y that man.' c. Kodomoga torakku ni 3-nin hik-are-ta. c h ild NOM tru ck b y 3-CLF run over-PASS-PAST T h re e ch ild ren w ere ru n o v er by a truck.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 6 d. Tomodatiga senseini 3-nin sikar-are-ta. friend NOM teacher by 3-CLF scold-PASS-PAST 'Three friends (of mine) were scolded by the teacher.’ The grammaticality and interpretation of the examples in (161) is clearly accounted for by the mutual c-command requirement in (158) since the surface subjects in these sentences are arguably the D-Structure objects which have moved to their surface position. Now with this background on the interpretation of Japanese NQs in mind, observe the following contrast between (162) and (163) (Tsujimura 1996, Miyagawa 1989). (162) a. * gakusei ga issyookenmei 4-nin student NOM very hard 4-CLF 'Four students studied very hard.' b. *Otoko ga jibun no kane de 5-nin man NOM selfs money with 5-CLF Five men telephoned their own money.' c. *Onna ga geragera to woman NOM loudly Three women laughed loudly.' d. * Kodomo ga wa ni natte child NOM in a circle 'Eight children danced in a circle.' benkyoosita. studied denwasita. telephoned 3-nin waratta. 3-CLF laughed 8-nin odotta. 8-CLF danced Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 (163) a. Gakuseiga ofisuni student NOM office to 'Three students came to the office.' b. Otoko ga baa ni man NOM bar to 'Four men entered a bar.' c. Kodomo ga butai ni child NOM stage to 'Five children went up to the stage.' d. Kyakuga ryokanni guest NOM Japanese inn to 'Six guests arrived at the Japanese inn.' 3-nin kita. 3-CLF cam e 4-nin haitta. 4-CLF entered 5-nin agatta. 5-CLF w ent up 6-nin tuita. 6-CLF arrived In both (162) and (163), the predicate verbs are all intransitives and the subject nominals are separated from the semantically associated NQs by an intervening manner adverb such as issyookenmei or a goal phrase such as ofisu ni. Despite the similarity, there is a clear difference in grammaticality between the sentences in ( 162) and those in (163). The contrast is easily accounted for by distinguishing the unergatives in (162) on the one hand and the unaccusatives and the inchoative in ( 163) on the other hand. Again assuming that manner adverbs and goal phrases are base-generated in positions higher than the direct internal argument but lower than the external argument, only the argument of an unaccusative or an inchoative can leave a trace that both c-commands the relevant NQ and is c-commanded by it. In other words, the contrast between (162) and (163) not only supports the existence of two kinds of intransitive verbs in Japanese but also clearly Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 demonstrates that the argument of one class shares a syntactic position with the subjects of transitive verbs while the argument of the other class does so with their objects. 2.4.7. Quantifier takusan The interpretation of the quantifier takusan lots' in a sentence also serves as evidence for the two kinds of intransitive verbs in Japanese. The quantifier takusan can mean either "many" or "much" and unlike a regular numeral quantifier made of a numeral and a classifier, it can be used without indicating what it is semantically related to. Taking advantage of the semantic and syntactic properties of takusan, Kageyama (1993:54-55) points out that unergative verbs share the pattern of interpretation with transitive verbs while unaccusative-inchoative verbs do not. (164) a. (X ga) (Y o) takusan nonda/tabeta/yonda/tukutta/etc. lots drank/ate/read/made/etc. 'X drank/ate/read/made/etc. lots of Y(s).' b. (X ga) takusan nakunatta/kowareta/arawareta/etc. lots died/broke (incho)/appeared/etc. 'Lots of X died/broke/appeared/remained/etc.' c. (X ga) takusan asonda/aruita/oyoida/odotta/etc. lots played/walked/swam/danced/etc. X played/walked/swam/danced/etc. a lot.' (based on Kageyama 1993:54, (21). translated by HO) In the case of the transitive verbs in (164a), takusan means a large number or quantity of the unexpressed object Y, not that of the unexpressed subject X. An Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 interesting contrast exists between (164b) and (164c). In the case of (164b), what is modified by takusan is the unexpressed subject of the sentence. In the case of (164c), however, the unexpressed subject of the sentence, for example, the people who walked, is not modified by takusan. What is modified instead is the amount of activity in which the unexpressed subject was engaged. Therefore, Takusan oyoida does not mean 'Many people swam' but 'Someone/some people swam a lot.' Kageyama argues that it is the internal argument that is modified by takusan and when such an argument does not exist takusan quantifies the amount of activity. 2.4.8. PRO PRO Theorem (Haegeman 1994:272-275, 285), which precludes the appearance of PRO in a governed position, is used by Kageyama (1993) in order to tease out unergatives and unaccusative-inchoatives. Although PRO can appear as a subject of a transitive or a passive sentence as in (165a-b), it cannot appear as an object of a transitive sentence as in (165c). (165) a. PRO kodomo o sodateru no wa muzukasii. child ACC bring up NOML TOP difficult PRO to bring up a child is difficult.' b. (Sibaide) PRO umaku nagurareru no wa muzukasii. (in plays) w ell slap-PASS NOML TOP difficult It is d ifficu lt to be slapped w ell (in plays).' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 0 c. *Oyaga PRO sodateru no wa muzukasii. parents NOM bring up NOML TOP difficult Tor parents to bring up PRO is difficult.' (Kageyama 1993:57, (26). Glossed and translated by HO) The position of PRO in (165c) is governed by V at S-structure or LF where Kageyama assumes PRO Theorem applies. After establishing the fact in (165), he illustrates that the same pattern exists with respect to the two types of intransitive verbs. (166) a. *Yonakani PRO arawareru koto ga am. the night during appear NOML NOM exist PRO occasionally appears during the night.' b. Yonaka ni PRO sawagu koto wa kinsi sareteim. the night during make noise NOML TOP being prohibited 'It is prohibited PRO to make noise during the night.' (Kageyama 1993:58, (28). Glossed and translated by HO) The verb arawareru is an unaccusative while sawagu is an unergative. The contrast between (166a) and (166b), therefore, suggests that the argument of an unaccusative verb is in object position while the subject of an unergative is not. 2.4.9. Resultative Constmction Resultative constmctions in Japanese also can shed light on the distinction between unergatives and inchoatives (Takezawa 1993, Tsujimura 1996). The italic phrases Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I ll in (167) describe resultative states, which are predicated of the objects, not the subjects, of the sentences. (167) a. Tarooga kabeo masiro-ni nutta. T. NOM wall ACC white painted Taro painted the wall white.' b. Tarooga teeburuo kirei-ni huita. T. NOM table ACC clean w iped. Taro wiped the table clean.' (based on Takezawa 1993, (16a-b)) c. Tarooga yukao pikapika-ni migaita. T. NOM floor ACC shiny polished Taro polished the floor shiny.' Just like the transitive-based resultative construction in English (see section 2.1.6), the passive sentences in (168) show that a resultative phrase does not need to be adjacent to the NP of which it is predicated. (168) a. Kabega wall NOM Taroo niyotte massiro-ni T. by white The wall was painted white by Taro.' b. Teeburu gaTaroo niyotte kirei-ni table NOM T. by clean The table was wiped clean by Taro.' c. Yuka ga Taroo niyotte pikapika-ni floor NOM T. by shiny The floor was polished shiny by Taro.' nurareta. paint-P ASS-PAST hukareta. wipe-PASS-PAST m igakareta. polish-PASS-PAST Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 2 ( 168a-c) are all grammatical because the respective resultative phrase in italics can be predicated of the trace left in the base position by the surface subject of each passive sentence. A similar situation holds with inchoative verbs. In (169), the resultative phrases are predicated of the NPs marked with Nominative Case marker ga. The resultative phrases may be adjacent to the NPs of which they are predicated as in ( 169a-b) or may be separated by modifiers such as an adverb in (169c) and a locative phrase in (169d). In (169a-b), it is not clear whether the surface subjects are in their base positions or have moved out. In the latter, however, the surface subjects clearly have moved out of the VP, leaving a trace behind. It is this trace that the respective resultative phrase is predicated of. (169) a. Sakanaga makkuro-ni kogeta. fish NOM black burned The fish burned black.' (Takezawa 1993, (20b)) b. Danboorubako ga petyanko-ni tubureta. cardboard box NOM flat crushed The cardboard box crushed flat.' c. Kabinga totuzen konagona-ni kowareta. vase NOM suddenly to pieces broke The vase suddenly broke to pieces.' d. Aisukuriimu ga huriizaade kotikoti-ni kootta. ice cream NOM freezer inside hard froze The icecream froze hard in the freezer.' (based on Takezawa 1993, (20a)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 The plausibility of this analysis of the inchoative resultative constructions in (169) becomes particularly clear when they are contrasted with the corresponding causative resultative constructions in (170). (170) a. Hanakoga sakanao makkuro-ni kogasita. H. NOM fish ACC b lack b u m Hanako burned the fish black.' (based on Takezawa 1993, (19b)) b. Hanakoga danboorubakoo petyanko-ni tubusita. H. NOM cardboard box ACC flat crushed Hanako crushed the cardboard box flat.' c. Hanakoga totuzen kabino konagona-ni kowasita. H. NOM suddenly vase ACC to pieces broke Hanako broke the vase to pieces.' d. Hanako ga huriizaa de aisukuriimu o kotikoti-ni kooraseta. H. NOM fieezer inside ice cream ACC hard froze Hanako froze the ice cream hard in the freezer.' (based on Takezawa 1993, (19a)) However, this kind of resultative construction is not possible with unergative verbs. If the ending of the resultative phrase is changed to -de as in kutakuta-de, the "depictive" reading obtains. (i) a. Taro ga kutakuta-de odotta. T. NOM exhausted danced Taro danced (when he was) exhausted.' b. Taro ga dorodarake-de asonda. T. NOM muddy played T aro played (when he was) muddy.' (based on Takezawa 1993, (25a-b)) The "depictive " reading is also possible with transitive verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 (171) a. *Tarooga kutakuta-ni odotta. T. NOM exhausted danced Taro danced (and got exhausted).' b. *Taroo ga dorodarake-ni asonda. T. NOM m uddy played Taro played (and got muddy).' (based on Takezawa 1993, (22a-b)) The contrast between (169) and (171) can be accounted for once we realize that in the grammatical resultative constructions in (167)-(170), all the resultative phrases are predicated of a direct object or its trace. Unlike inchoative verbs, unaccusative verbs do not project their NP argument as the sister of V. Therefore, the surface subject of an unergative sentence has no trace of which a resultative phrase can be predicated. In this way, the fundamental lexico-syntactic contrast between unergatives and unaccusatives can account for the grammaticality pattern of the resultative constructions based on them. 2.4.10. Indirect Passive There are two kinds of passive structures which illustrate the difference between unaccusative-inchoative verbs and unergative verbs (Kageyama 1993). First, the (li) a. Taro ga sakana o nama-de tabeta. T. NOMfish ACC raw ate Taro ate the fish raw.' b. Taro ga hadaka-de sakana o tabeta. T. NOM naked fish ACC ate Taro ate the fish nude.' (based on Takezawa 1993: (41a-b)) Notice that in (iia) it was the fish that was raw but in (iib) it was Taro, not the fish, who was naked. In both, the depictive phrases end in -de. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 indirect passive, also known as the "adversative passive," can be formed based on transitive verbs. (172) a. Dorobooga saihuo nusunda. thief NOM wallet ACC stole The thief stole the wallet.' b. Taro ga dorobooni saihuo T. NOM thief by wallet ACC nusum areta. steal-PASS-PAST Taro got his wallet stolen (by the thief).’ (173) a. Senseiga kodomo o hometa. teacher NOM child ACC praised 'The teacher praised the child.' b. Taro ga senseini kodomo o T. NOM teacher by child ACC Taro had his child praised (by the teacher).' hom erareta. praise-P ASS-PAST Similarly, indirect passive sentences can contain an unergative verb in the embedded clause. (174) a. Kodomo ga sawaida. children NOM made a loud noise The children made a loud noise.' b. Hanakoga kodomo ni sawagareta. H. NOM children by make a loud noise-PASS-PAST Hanako suffered from the children making a loud noise.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 (175) a. Akanbooga naita. b ab y NOM cried 'T h e b ab y cried ' b . H anako g a akanboo ni nakareta. H. NOM baby by cry-PASS-PAST 'Hanako suffered from the baby's crying.' However, indirect passives are not well-formed when they are based on unaccusative-inchoative verbs. (176) a. Tomodatiga nakunatta. friend NOM pass away 'A friend passed away.' b. *Taroo ga tomodatini nakunarareta. T. NOM friend by pass away-PASS-PAST Taro suffered from his friend's death.' (177) a. Niwanokiga taoreta. garden GEN tree NOM fell The tree in the garden fell.' b. *Taroo ga niwa no ki ni taorerareta. T. NOM garden GEN tree by fall-PASS-PAST Taro suffered from the tree's falling in the garden.' (178) a. Yuureiga arawareta. g h o st NOM appeared 'A g h o st appeared.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 b. *Taroo ga yuurei ni arawarerareta. T . NOM g h o st by appear-P a s s - p a s t T a ro su ffe re d from a g h o st's ap p earan ce.’ Kageyama points out that the NP marked by ni needs to be the external argument of the verb in the embedded clause. Furthermore, he takes sentences such as (176) to (178) as evidence that in Japanese the internal argument of an unaccusative verb does not necessarily move to the syntactic position normally occupied by the external argument. If that were the case, he argues, indirect passives based on unaccusative verbs would be grammatical (Kageyama 1993:60). 2.4.11. Passivized Causative Similarly the passivization of causative sentences also points to the distinction between the two kinds of intransitive verbs. First, observe (179). (179) a. Gakuseiga hono katta. stu d en t NOM book ACC bought 'T he stu d en ts bought th e bo o k .' b . S e n s e ig a gakusei n i h o n o kaw aseta. teacher NOM student DAT b o o k ACC buy-CAUS-PAST T h e te a c h e r m ade the stu d en ts b u y the book.' c. Gakuseiga senseini(yotte) hono kawasareta. stu d en t NOM teacher by b o o k ACC buy-CAUS-PASS-PAST T h e stu d en ts w ere m ade to b u y the book by the teacher.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 d. *Hon ga sensei ni(yotte) gakusei ni kawasareta. book NOM teacher by student DAT buy-CAUS-PASS-PAST The book was made to be bought by the students by the teacher.' (based on Kageyama 1993:60, (30)) ( 179) shows that a passive sentence can be formed if the subject of the embedded clause, gakusei in (179b), becomes the main clause subject as in (179c), but not if the object of the embedded clause, hon, is made the main clause subject as in (179d). Now observe how unergative verbs behave with respect to this type of passivization. (180) a. Kodomo ga janpu-sita. child NOM jum p ed 'A ch ild ju m p ed .' b. Tarooga kodomoo/ni janpu-saseta. T. NOM child ACC/DATjump-CAUS-PAST T aro m ade a c h ild ju m p .' c. Kodomo ga Taroo ni(yotte)janpu-saserareta. child NOM T . by jump-CAUS-PASS-PAST 'A child w as m a d e to ju m p by T aro.' (181) a. Kodomotati ga hataraita. children NOM labored T h e children labored.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 b . S o n o o y a ta tig a kodom otati o/ni hatarakaseta. those parents NOM children ACC/DAT labor-CAUS-PAST T h o se parents fo rc e d th eir children to labor.' c . K odom otati g a sono oyatati ni(yotte) hatarakasareta. children NOM tho se parents by labor-CAUS-PASS-PAST 'The ch ild ren w e re forced to labor by th eir p aren ts.' ( 180c) and (181c) show that the NP argument of an unergative verb can be made the main clause subject in a passivized causative sentence. Unaccusative verbs contrast with unergatives in two respects. (182) a. Mizuga joohatu-sita. water NOM evaporated 'Water evaporated.' b . S e n se ig a m izu o /* n i jo o h atu -saseta. teacher NOM w ater ACC/DAT evaporate-CAUS-PAST The teach er m a d e w ater evaporate.' c. *Mizu ga sensei ni(yotte) joohatu-saserareta. w ater NOM teacher by evaporate-CAUS-PASS-PAST T he w ater w as m a d e to evaporate by the teacher.' (183) a. Kanzyaga siboo-sita. patient NOM died The patient died.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 0 b . S o n o is y a g a k a n z y a o /* n i sib o o -saseta. the doctor NOM p atien t ACC/DAT die-CAUS-PAST T h e doctor cau sed th e p atien t to die.' c . * K an zy ag a so n o isy a ni(yotte) sib o o -saserareta. patient NOM th e d o cto r b y die-CAUS-PASS-PAST 'T he patient w as c a u s e d to die by the doctor.' First, (182b) and (183b) show that unlike the argument of an unergative verb, the argument of an unaccusative verb cannot be marked with the Dative ni in the embedded clause of a causative sentence. Second, and more importantly, the passivization of a causative sentence is not possible when it is based on an unaccusative verb as shown in (182c) and (183c). The contrast between unaccusatives and unergatives on this point is strikingly clear. As in the case of the indirect passive, Kageyama argues that ungrammatical sentences such as (182c) and (183c) suggest that the internal argument of an unaccusative verb may remain in situ at S-Structure and gets Nominative Case in that position (Kageyama 1993:62).‘ * ^ 2 .5 . K o r e a n 2.5.1. Cognate Object Construction and "Nominal + ha" Construction As for the distinction between unergatives and unaccusatives in Korean, Lee (1989:478) points out that unergatives with agentive argument may take a cognate object or form a nominal + ha do' construction, creating expressions equivalent to to laugh a laugh, to do running', to do a study', etc. These types of constructions are normally unavailable with unaccusatives, rendering to do an The surface unaccusativity, the phenomenon that the argument of an unaccusative verb remains in situ at S-Structure, is also argued for Japanese data in Nakayama and Koizumi (1991). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 1 arrival' impossible. Young-Joo Kim (1990:85, footnote 23), for example, lists the following cognate object nominals, all of which are derived from unergative verbs. (184) verbal stem cognate object nominal ca- to sleep* ca-m 'sleep' ttwy- to jump' ttwy-m'jump' col- to doze' col-um 'drowsiness' chwu- to dance' chwu-m dance' ket- to walk' kel-um walk' ssawu- to fight' ssawu-m 'fight' kkwu- to dream' kkwu-m dream' kich(i) to cough' kichi-m 'cough' At first, nominal + ha 'do' expressions such as concayha- to exist', cungkaha- to increase' and tochakha- to arrive' may appear to be counter-examples to the above statement. These expressions, however, contain Sino-Korean nominals instead of nominals derived from native Korean vocabulary. Therefore, Lee's generalization should be interpreted to apply only to native Korean verbs.^^ 2.5.2. Construction with -e-iss-ta As the Japanese verbal phrase V-te iru distinguishes many unaccusative-inchoative verbs from unergative and transitive verbs by assigning perfective reading to the The same situation is observed in Japanese. Cognate object nominals are used only with unergative verbs as in uta o utau to sing a song’ and odori o odoru to dance a dance". In addition, sometimes cognate nominals are found in expressions such as hito-hataraki sum 'to do one-work’ and i-nemuri sum to doze off. Native unaccusative verbs do not take a cognate object or appear by themselves in the sum ’ do’ construction. Apparent counter-examples to this generalization such as sonzai sum to exist’, zooka sum to increase’ and tootyaku sum ’ to arrive’ all contain Sino-Japanese expressions. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 2 former, the Korean verbal phrase V-e~iss-ta also expresses a perfective sense when used with many unaccusative-inchoative verbs. The following list of verbs that can take -e iss-ta for a perfective construction is from Young-Joo Kim (1990:91, (49)). (185) Examples of verbs that can take e-iss- tteleci- fall' elso-/olska- go up/come up' mimeci- 'collapse' ppaci- be omitted' nemeci- 'stumble' sos 'rise, tower up' ttena- 'leave' toy- 'become' tolso-Ztolska- return' nathana- 'appear' nam- remain' peleci- happen' concayha- 'exist' tulena- be exposed' theci- break out' phi- bloom' kicelha- 'faint' situl- 'wither* cwuk 'die' nok- 'melt' thayena- be bom' el- 'freeze' tochakha- 'arrive' ane- 'sit' michi- go crazy' se- 'stand' ka- 'go' nwup- 'lie' o- 'come' tah- touch' (involuntary) tuleo-Ztuleka- enter' kitay- lean against' say- 'leak' ssek- rot' puphule- 'swell up' na- 'occur' kkay- 'wake up' sokha- belong' kkuthna- end' (intr.) nulena-Zcungkaha- 'increase' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 As Young-Joo Kim notes, there is an overwhelming match between these verbs and those that are picked out by Italian essere. However, Young-Joo Kim, who semantically defines unaccusatives as a class of verbs with the feature [- agentive],'*^ concludes that the V-e-iss-ta construction is more sensitive to another semantic feature [4- teiic] and therefore cannot be a perfect diagnostic for unaccusatives. She points out, for example, that [+ stative] verbs do not take -e-iss-ta for perfective reading while verbs such as sit', stand' and lie', which are ambiguous in Korean with respect to agentivity, do take this verbal phrase. The first objection has already been dealt with in terms of the Japanese verbal phrase -te iru (section 2.4.2). Perfective reading is possible only when there is potential change involved in the semantics of the verbs in question. This is why genuine [+ stative] verbs are excluded from the perfective constructions both in Korean and Japanese. Young-Joo Kim's second concern is also addressed by Levin and Rappaport Hovav (1995). According to them, verbs such as 'sit' and stand' are in fact unaccusative verbs, irrespective of the agentivity of their argument, when they convey "assume position" and "simple position" s e n s e s .4 5 Therefore, despite the exclusion of genuine [+ stative] verbs, the V-e-iss- ta construction in Korean seems to be able to separate many unaccusative- inchoatives from unergatives. 44 Young-Joo Kim (1990) uses the feature [+ agentive] in the sense of "immediate causer of the event". Her semantic definition of unaccusatives, therefore, appears to be very close to that of Levin and Rappaport Hovav (1995). 45 When they have "maintain position" sense. Levin and Rappaport Hovav (1995) classify them as unergatives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 2.5.3. W Compounding Relevance of verbal classification to the grammaticality of W compounds in Korean is noted by Byong-Kwon Kim (1993). Specifically, for our concern, a distinction of the two subclasses of intransitives is important in this respect. First, observe the following examples. The W compounds in (186a) and (186b) consist of two unergatives and two unaccusatives respectively and they are both grammatical. (186) a. John-i kil-eyse anc-e swi-ess-ta. John-NO M street-on sit rest-PAST-DEC Jo h n sat/rested on the street.’ (B y o n g -K w o n K im 1993, (19b)) b . ew u cengpayngi-ka kil-eyse el-e cw u k -ess-ta. drunkard-N O M street-on freeze die-PAST-DEC 'A d ru n k ard froze/died o n the street." (B y o n g -K w o n K im 1993, (20b)) In contrast, the following three examples are all ungrammatical. (187) a. * J o h n -i M ary-lul John-NO M Mary-ACC Jo h n stru ck M ary dow n.' b. * J o h n -i M ary-lul John-NO M Mary-ACC Jo h n p u sh ed M ary fainted." c. * Jo h n -i yeses si-ey ttayli-e ssuleci-ess-ta. beat fall-PAST-DEC m il-e kicelha-ess-ta. p u sh faint-PAST-DEC (B yo n g -K w o n K im 1993, (3 1 a-b )) kkay-e taUi-ess-ta. John-NO M six o"clock-at w ake up run-PAST-DEC "John w o k e/ran at six o'clock." (B y o n g -K w o n K im 1993, (21b)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 In (I87a-b) the iefthand constituents are transitive verbs while the righthand ones are unaccusative verbs. In (187c), the Iefthand one is an unaccusative and the righthand one is an unergative. The examples of (187a and c), however, can easily be made grammatical without changing much of their meaning. (188) a. John-i Mary-lul ttayli-e ssulettuli-ess-ta. John-NOM M ary-ACC beat fell-PAST-DEC John stru ck M a ry d o w n .' (B yong-K w on K im 19 9 3 .-footnote 7, (iv)) b. John-i yeses si-ey Uena-e talli-ess-ta. John-NOM six o'clock-at get up run-PAST-DEC Jo h n g o t u p /ra n a t six o 'clo ck .' (B y o n g -K w o n K im 1993, (21a)'*®) In (188a) the second constituent of the W compound is changed to the transitive ssulettuli-ta 'fell' from the morphologically related unaccusative and in (188b) the unergative verb is substituted for the original unaccusative wake up'. These changes have created two grammatical compound verbs, each of which consists of verbs of the same class. Citing W compounds made of an unergative and a transitive such as twi-e- nem- run/get over', kwulu-e-nem- roll/get over' and anc-e-mek- sit/eat' in addition to the above examples, Byong-Kwon Kim (1993; 181) concludes that Korean unaccusative verbs cannot combine with unergatives or transitives. Therefore, as in the case of Japanese W compounds (section 2.4.4), verbal classification, especially the distinction between unergatives and unaccusatives, is crucial to the formation of Korean W compounds. ^ This example is originally attributed to Chung (1992). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 2.5.4. Case Marker Drop from a Numeral Quantifier Korean also shows a morphosyntactic phenomenon that appears very similar to what we have observed in Japanese in section 2.4.5. Chungmin Lee (1989:475- 480) focuses on the condition of Case Marker Drop which is applied to the remaining numeral quantifier after the associated NP has moved out of the base generated syntactic position into a higher position. First, Lee's basic assumption is that the unmarked order of NP expressions with numeral and classifier is: N + Numeral + Classifier + Case Marker, in which a numeral and classifier functions as a Numeral Quantifier (NQ). After showing that in Korean, unlike in Japanese, the remaining NQ can be Case marked (while the moved NP may keep the original Case or may be marked with Topic marker in a higher position), Lee discusses the situations in which NQs appear without an overt Case marker. The significant point is that an NQ without an overt Case marker is permitted only if the original NP expression is generated as the sister of V and the NQ remains in this position. Therefore, a "bare" NQ is grammatical if the moved NP is the internal argument of a transitive verb such as "drink" or the sole argument of an unaccusative verb such as "arrive" and "die." The Case marker cannot be dropped if the argument is the subject of a transitive verb or an unergative verb such as "laugh." Chungmin Lee's work is quite important for us because it shows that the condition of Case Marker Drop of NQs gives direct evidence that the argument of an unaccusative verb is syntactically projected in sister of V. 2.5.5. Resultative Constructions English has two types of resultative constructions which we named for convenience the transitive-based and unergative-based resultative constructions (see sections Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 2.1.6 and 2.1.7). The former type of resultative construction is also observed in Japanese (section 2.4.9). The following examples show that the transitive-based resultative construction is found in Korean as well. (189) a. Ku-nun soy-lul pyongpyongha-key chyessta. He-TOP metal-ACC flat-COMP p o u n d ed 'H e pounded th e m etal flat.' b. K u-nun h w apw un-ul yele cokak-ulo p w u sw u e ssta . He-TOP vase-ACC m any pieces-RES broke He broke the vase into pieces.' (Jong-Bok Kim 1993, (la-b)) c. Ku-nun cha-lul kkaykkusha-key asiessta. He-TOP car-ACC clean-COMP w ash ed He washed the car clean.' (Jong-Bok Kim 1993, (35a)) d. B ob-i soy-lul ttukep-key talk w u essta. Bob-NOM metal-ACC hot-COMP heated B ob heated the m etal hot.' e. Alan-i kwaca-lul pasakha-key kwuesta. Alan-NOM cookies-A CC crisp-COMP baked 'Alan baked the cookies crisp.' (Kim and Maling, to appear) Just as in English and Japanese, passive and inchoative verbs can appear in the transitive-based resultative construction. (190) a. Thakca-ka kkaykkusha-key ttak-i-ess-ta. table-NOM clean-COMP wipe-PASS-PAST-DEC The table was wiped clean.' (Jong-Bok Kim 1993, (2a)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 b . Cha-ka kkaykkusha-key ssiskie ci-ess-ta. car-NOM clean-COM P wash PASS-PAST-DEC The car was washed clean.’ (Jong-Bok Kim 1993, (35b)) c. Hoswu-ka tantanha-key el-ess-ta. Iake-NOM soIid-COMP freeze-PAST-DEC 'The lake froze solid.’ (Kim and Maling, to appear) (190a-b) involve passive verbs and when compared with sentences such as (189c), suggest that there is a trace involved in the object position. (190c) is a sentence with an inchoative verb and also conforms to the general pattern. In contrast, according to Jong-Bok Kim’ s observation, typical unergative verbs in Korean do not form a grammatical sentence when they appear in the transitive-based resultative construction. (191) a. * Ku-nun aphu-key kichimhayessta. He-TOP sick-COMP co u g h ed H e co u ghed (and got) sic k .’ b. * K u-nun phikonha-key w u se ssta . He-TOP tired-COMP laughed He laughed (and got) tired.’ (Jong-Bok Kim 1993, (3a-b)) Despite the plausibility of someone getting sick or tired as a result of too much coughing or laughing, the sentences in (19la-b) are ungrammatical. Therefore, the Korean data on the transitive-based resultative construction conforms to the familiar pattern observed in English and Japanese, in which transitives and unaccusatives behave similarly while unergatives do not. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 Let us now observe how the unergative-based resultative constructions work in Korean. We have observed that there are three types of this general construction in English. They involve the subject's reflexive pronoun, an NP inalienably possessed by the subject, or an independent NP, of which a resultative phrase is predicated. Among the three subtypes, Jong-Bok Kim (1993) notes that Korean lacks the first. (192) a. * Ku-nun caki-ku michi-key nolayhayessta. He-TOP self-NOM crazy-COMP sang (literally) He sang himself crazy.' (Jong-Bok Kim 1993, (5a)) The two other subtypes, however, seem to have counterparts in Korean. (193) a. Ku-nun mok-i swi-key oychessta. he-TOP neck-NOM hoarse-COMP shouted '(literally) He shouted (his) neck hoarse.' (Jong-Bok Kim 1993, (7a)) b. John-i paykkop-i ppaci-key wus-ess-ta. John-NOM belly-NOM come.out-COMP laughed 'John laughed his belly out.' (Kim and Maling to appear, (8a)) (194) a. Ku-nun (ku-uy) sinpal-i talh-tolok talliessta. he-TOP (he-GEN) shoes-NOM w om -C O M P ran '(literally) He ra n his shoes w orn.' (Jo n g -B o k K im 1993, (5b)) b. K u-nun (ku-uy) sonsw uken-i ces-key w ulessta. he-TOP (he-GEN) handkerchief-NOM soggy-COM P cried He cried h is h an d k erc h ief soggy.' (Jo n g -B o k K im 1993, (36b)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 Notice that there is a crucial difference between (193) and (194) on the one hand and their English counterparts on the other. Unlike their English counterparts, whose non-subcategorized NPs are marked with Accusative Case, the Korean examples contain NPs marked with Nominative Case. Apparently, Korean seems to have a mechanism of Nominative Case assignment independent of some functional category under Infl. This speculation is further supported by the fact that even inchoative verbs can form unergative-based resultative constructions in this language. (195) a. Hoswu-ka phyomyen-i tantanha-key elessta. lake-NOM surface-NOM solid-COMP froze T h e lake froze (so that) its surface (becam e) solid.’ (K im and M aling, to appear) b . K il-i cilphenha-key nw un-i n o k assta. road-NOM slushy-COM P snow-NOM m elted T h e snow m elted (so th at) the ro ad (becam e) slushy.' (Kim and Maling, to appear) Needless to say, a similar construction is impossible in English. ( 196) a. The lake froze solid. b . *The lake froze its surface solid. (Kim and Maling, to appear) The data such as (195a-b) indicate that the unergative-based resultative construction does not differentiate unaccusative-inchoatives from unergatives. However, the contrast (190c) and (191a, c) suggests that transitive-based Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 resultative construction may in fact help to distinguish the two subclasses of intransitive verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 3. Unaccusative and Inchoative Verbs in L2 English The current work is concerned with the acquisition of unaccusative-inchoative verbs in second language contexts. Unlike Sorace's work (e.g. Sorace 1993a, 1993b, 1995) mentioned in section 1.3, however, the focus of this thesis is on the structural variations in which unaccusative-inchoatives are used by nonnative speakers. Li this chapter, I first review previous research both directly and indirectly related to the L2 acquisition of the English unaccusative-inchoative verb class, and then critically evaluate the relevant research in order to highlight some important research issues. 3.1. S t r u c t u r a l V a r ia t io n s This section is concerned with the ways in which English unaccusative-inchoative verbs are used by normative speakers. I review several structural patterns and phenomena noticed and discussed by researchers to date. Reviewed are phenomena such as "passivized" unaccusative-inchoative verbs, avoidance of inchoative verbs in the NP-V word order, transitivization of unaccusative verbs, and post verbal NP structures. The causes for these often ungrammatical structures suggested by various researchers are also presented so that they can be critically evaluated in section 3.2.2. 3.1.1. "Passivized" Unaccusative-inchoatives The most conspicuous way in which unaccusative-inchoative verbs appear in the English of nonnative speakers is what I call the "passivized" unaccusative- inchoative structure. In this structure, the nominal argument of an unaccusative- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 inchoative verb appears in preverbal position followed by the auxiliary be and the past participial form of the verb. Superficially, therefore, the "passivized" unaccusative-inchoative structure appears to be exactly what its name suggests: the passivization of unaccusative-inchoative verbs. However, to pinpoint actual linguistic and psychological factors responsible for its appearance in learners' English is not an easy task. A number of proposals have been made in the literature, sometimes with more than one interpretation suggested by a single researcher. Yet, there has been no general consensus to date on the true linguistic and psychological meaning of this non-target "passivization. " In the rest of this thesis, therefore, I use the term "passivized" simply as a descriptive term to refer to the structure in question, deferring the determination of its true meaning until section 4.4.1.5. 3.1.1.1. Burt and Kiparsky 1972 and Richards 1973 There are two early studies that recorded the "passivized" unaccusative errors in learners' English. They are Burt and Kiparsky 1972 and Richards 1973. The examples in (1) come from the former, which is an ESL teachers' grammar manual that compiled learners' errors with pedagogical suggestions. ( 1 ) a. *I hope that the good will between U.S A. and Republic of Korea will be lasted long time. b. *Some people afraid that if the war ends a financial crisis will be arisen. c. *He was arrived early. (Burt and Kiparsky 1972:47) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 ( la-c) ate the only examples given by Burt and Kiparsky in a section entitled "Making Intransitive Verbs Passive," which is within a more general section of "Inappropriate Use of Passive." As for the cause of ungrammaticality in (la-c), the authors write: Verbs like last, arise, and arrive are intransitive and that means objects may not follow them. But the idea of passive sentences in English is to put the object of the verb or another noun phrase in subject position. If there is no object of the verb or of a preposition, there is no noun phrase to become the subject of the passive verb. (Burt and Kiparsky 1972:47) As both the headings and the above quote indicate, Burt and Kiparsky saw the "passivized" unaccusative structure as a result of misapplication of passivization to the general intransitive verb class, which cannot be passivized in the first place because they lack an object. Needless to say, the semantic similarity among the three verbs in (1), i.e. the non-agentive or non-causer status of the subject arguments, was not recognized by the authors. Consequently, further subclassification of these intransitives was not attempted. Richards (1973) also noted "passivized" unaccusative errors such as (2). (2) a. *He was died last year. b. *O'0& à3iy It was happened. (Richards 1973: appendix) However, he saw (2a-b) simply as instances of a more general type of error related to the acquisition of the tense and aspect systems in English. Consequently, in his analysis, (2a-b) are not treated any differently from errors such as (3a-b).^^ (3) a. *The teacher was told us. I have nothing to say in this thesis about the kind of error exemplified by (3a-b). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 b. *They are opened the door. (Richards I973:appendix) For Richards, (2) and (3) were the same kind of errors, which resulted from learners' incomplete mastery of the tense and auxiliary system in English. In his view, both (2a-b) and (3a-b) should have been either in simple past or present perfective. Nothing special about the structure in (2) was pointed out by Richards. In hindsight, it is quite interesting that "passivized" unaccusative errors were noticed by ESL researchers long before Perlmutter articulated the Unaccusative Hypothesis in 1978. In the absence of such a theoretical perspective, however, the ESL researchers thus lacked the means to capture an important generalization concerning this type of non-target construction. 3.1.1.2. Hubbard 1983, Hubbard and Hix 1988, and Hubbard 1994 It was Hubbard (1983, reported in Hubbard 1994:55-56) who shed much needed theoretical light on the "passivized" unaccusatives. He reported that the overwhelming majority (more than 90 percent) of errors of this type were concentrated in a particular subclass of intransitive verbs identified by Perlmutter (1978) as unaccusative verbs. By finding the errors in compositions written by native speakers of eight different languages, Hubbard also showed that the use of this non-target structure was widely observed among English learners from various LI backgrounds. In Hubbard and Hix (1988) and Hubbard (1994), the "passivized" unaccusative structure, or the "intrusive be" (Hubbard and Hix 1988:93), is exemplified by the following examples. (4) a. *The strange event was occurred last May. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 b . *The problem is exist today in many countries. (Hubbard and Hix 1988:94) c. ^Something strange was happened before I could open the door. d. *This problem ejcijfcd for many years. (Hubbard 1994:55) In looking for a cause for the "passivization," Hubbard and Hix give two types of possible explanations. First, they compare (4a-d) with (5), which contains passive adjectives reviewed in section 2.1.3. (5) The theater is located/situated (*by Fred) on Elm Street. (Hubbard and Hix 1988:94) Seen in the context of the Unaccusative Hypothesis, it is clear that the original object appears as the subject in both (4) and (5). Also, as indicated by an asterisk within parentheses, the passive adjective in the predicate cannot be modified by an agentive by phrase. This structural restriction, they observe, is the same with the "passivized" unaccusatives since most of them appear without an agentive phrase as in (4a-d). Based on this similarity, Hubbard and Hix suggest that the "passivized" unaccusative could be analyzed as analogous to the passive adjective. A second suggestion Hubbard and Hix (1988:97) presents is possible interference from the morphosyntactic properties of the learner's LI. They note that some languages mark "intransitives in this class [i.e. the unaccusative class— HO] with morphology that either is identical with or overlaps that used in passive construction." Clearly, here, they seem to have in mind structures such as Italian "passato prossimo" (section 2.2.4) and French "passé composé," in which Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 unaccusative verbs appear in past participle forms with a "BE"-type auxiliary verb, namely the Italian essere and the French être. (6) a. Tre ragazzi sono airivati alia stazione. (Italian) b . Trois garçons sont arrivés à la gare. (French) Three boys (have) arrived at the station.' The structure in these languages expresses the meaning that corresponds to the English past tense or perfective aspect, but on the surface is identical to the "passivized" unaccusatives in Interlanguage English such as (7). (7) *Three boys are arrived at the station. (IL English) Hubbard and Hix's point seems to be that despite the semantic differences, the structure of (6a-b) may be directly transferred to IL English by native speakers of these languages.**® 3.1.1.3. Zobl 1989 Zobl (1989) is probably the first study specifically devoted to the issue of the structural irregularity apparent in the L2 use of English unaccusative-inchoative verbs. He noticed and discussed three kinds of non-target structure: the postverbal subject, the (non-causative) preverbal subject, and the "passive" unaccusative- inchoative. He also showed quite convincingly that the three seemingly unrelated non-target structures could be given a unified account on the basis of the Unaccusative Hypothesis. In this section, however, I will focus exclusively on his Hubbard (personal communication 1997) informs me that he now suspects the overgeneration of passive adjectives may not be as strong a reason as LI transfer or a universally available interlanguage structure, or even the combination of the two. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 analysis of the "passivized" unaccusative-inchoative construction, deferring the discussion of the other two structures to sections 3.1.4.1 and 3.1.3.4. 2tobl's data come from compositions written by college-age L2 learners of English (specifically, 90 Japanese, 10 Arabic, 10 Spanish, 1 Chinese, 1 Turkish, 1 Thai and 1 Indonesian in ESL programs in Canada and the United States). The following examples are productions of LI Arabic, Japanese, and Thai speakers respectively. Their proficiency levels, according to Zobl, are "high intermediate" to "advanced." (8) a. *The most memorable experience of my life was happened 15 years ago. b . *Most of people are fallen in love and marry with somebody. c . *My mother was died when I was just a baby. (Zobl 1989:204, (l)-(3)) Zobl accords a dual significance to the "passive" unaccusative structure such as (8a-c). On the one hand, he believes that the structure is an overt marker of the syntactic movement which learners apply to the NP in the D-Structure object position. In other words, he regards this structure as evidence that the sole argument of the unaccusative verb is correctly projected as the sister of V and then moved to the surface subject position. The only difference between the native grammar and the learners' grammar is that only in the latter is the movement overtly (and incorrectly) marked by the passive verbal phrase be + Ven. On the other hand, he suggests another more basic meaning in the passive verbal phrase as a lead to the correct D-Structure position of the argument of an unaccusative verb: [0 [V NP]]. Zobl argues that in the learner's grammar at the relevant stage, the lack of logical Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 subject is canonically associated with the passive verbal phrase (Zobl 1989:217, footnote IS) and that once the grammatical passive construction is mastered, the internal argument of an unaccusative verb is incorrectly projected in this structural fram e.* * ® 3.1.1.4. Yip 1989, 1994, 1995 As part of her research on the Interlanguage English of native speakers of Chinese, Yip (1989, 1995) discusses the "passivized" unaccusative-inchoative structure. She observes that the non-target structure is fiequently produced and judged grammatical, surprisingly, often by advanced learners. Her observation is based on examples such as (9a-d). (9) a. *1 do not think that such abusive action should be happened to a twelve year old child. b . *Rush hour traffic can be vanished because working at home is a new version. c . *This kind of diglossic situation can only be appeared in society where the two different variations should not be too different and too similar. d. *For last 15 years computers have drastically affected our life and this will be continued in the future. (Yip 1995:130, (3)-(6)) The unaccusative verbs in (9a-c) would not be passivized in target English. As for (9d), although the passive morphosyntax itself would be perfectly well-formed in * * ® Zxihl argues that before this developmental stage, the argument of an unaccusative verb is not projected in the sister of V. For a more detailed review o f his developmental account, see section 3.2.3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 the causative sense of the verb continue, the passivization of its inchoative sense is clearly inappropriate. Yip shows that the errors in (9) could have been avoided if the Chinese learners who produced them had simply transferred the Chinese word order in (10) to the English unaccusative-inchoative sentence. (10) Xue ronghua-le. snow melt-PFV The snow melted.' (Yip 1995:135, (21b)) The surface NP-V word order with an unaccusative verb is perfectly grammatical in Chinese as in English. However, as the sentences in (9) and many others like them indicate, the simple option of LI transfer is not always adopted by native speakers of Chinese when it comes to dealing with English unaccusative-inchoative verbs. Furthermore, Yip points out that production of the ungrammatical "passivized" unaccusatives is even more puzzling for two reasons. First, the passive structure itself is lexically and semantically restricted in Chinese (Yip 1989:sec. 4.4.2). Second, the passivization of unaccusatives in particular is clearly ungrammatical in Chinese just as in English. Consider (11) and (12). (11) *S h en m e bei fasheng-le? w h a t PASS happen-PFV '* W h a t w as h ap p en ed ?’ (12) *Shuye bei diao-le xia lai. leaf PASS fall-PFV down come ■*The leaves were fallen down.' (Yip 1995:136, (24)-(25)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 The ungrammaticality of (11) and (12) is clearly due to the existence of the passive morpheme bei, because without it, the sentences would be perfectly grammatical. Based on these data. Yip concludes that Chinese learners’ production of "passivized " unaccusatives cannot be attributed to LI transfer of similar sentences since there is no equivalent "passivized " unaccusative in Chinese. Therefore, Yip’ s data and analysis clearly show that Chinese speakers of English are apparently using a structure acceptable neither in their first language nor in the target language, avoiding, at the same time, an easy option of direct transfer of the same grammatical word order from their LI. In addition. Yip (1994) reports that this tendency to produce and/or accept the ungrammatical ""passivized” unaccusative-inchoative structure is not limited to Chinese speakers of English. This conclusion is drawn based on a grammaticality judgment test on a number of LI speakers, specifically, 2 Korean, 2 Indonesian, 1 German, 1 Spanish, 1 Greek, and 1 Hebrew as well as 2 Chinese speakers. As for the interpretation of the ""passivized” unaccusative-inchoative structure. Yip (1989, 1994, 1995) suggests three possible explanations. First, as Zobl does. Yip argues that the overt morphosyntax of the English passive be + Ven serves as a marker of syntactic movement of an argument from object position to subject position. Second, she suggests that the ungrammatical ""passivization” may be explained if we assume that L2 learners generally interpret both unaccusatives and inchoatives as underlyingly transitive verbs. On this point. Yip speculates that L2 learners of English may be "reluctant to believe that any change of state occurs spontaneously, without external causation (Yip 1994:130)."" This interpretation allows one to see the "passivized” unaccusative-inchoatives as true passives because they are in fact based on the "transitive verbs" postulated by the learners. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 The third idea suggested by Yip (1995:138) is that some unaccusative verbs may be treated by learners as inherently passive. She argues for the plausibility of this interpretation by drawing attention to the existence of inherent passive verbs in languages like Latin. This third interpretation of the "passivized" unaccusative- inchoatives appears to be somewhat similar to Zobl's idea that the lack of logical subject in the verb's argument structure is associated with the D-Structure passive morphosyntax by L2 learners of English. 3.1.1.5. Hirakawa 1995 Hirakawa (1995) has investigated the Japanese speakers' acquisition of syntactic NP-movement in English with respect to constructions that involve unaccusative, inchoative, passive and middle verbs. In her study, 22 second-year college students from Japan attending a month-long intensive English program in Canada and a control group of 14 native English speakers were given production and grammaticality judgment tasks. Among many interesting results that she obtained, the following two findings are particularly relevant to the "passivization" of unaccusative-inchoative verbs. First, Hirakawa reports that unlike the native speaker subjects, the Japanese subjects inappropriately passivized inchoative verbs. Among the six inchoative verbs used in the production task (namely, melt, break, continue, dry, spill, and increase), three verbs, break, spill, and dry, were passivized 13 times out of a total of 33 occasions. Second, Hirakawa has found a statistically signiticant difference between the two groups with respect to the ability to correctly reject "passivized" unaccusatives. When faced with verbs such as appear, die, fall, arrive, happen, and disappear in the passive structure, the Japanese subjects were not very Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 successful in rejecting the sentences as ungrammatical. Hirakawa suggests that her subjects treated these verbs as ergatives, i.e. verbs with both causative and inchoative usage, because some of their translational equivalents are in fact ergative verbs in Japanese. Therefore, she reasons, it is possible that the subjects accepted the "passivized" unaccusatives as the grammatical passive counterparts of the causative sentences. 3.1.2. Avoidance of Inchoative Verbs in the NP-V Order A second syntactic phenomenon noticed about the use of intransitive verbs by L2 English learners is their tendency to avoid the inchoative verbs in the NP-V word order. 3.1.2.1. Kellerman 1978 Kellerman (1978) reported the results from two transferability judgment experiments. The first was conducted on 83 first-year and 26 second-year students of English at Nijmegen University and the second on 50 first-year and 31 third-year English majors at Utrecht University. In the study, he investigated the transferability of various senses of the Dutch verb breken to the English verb break. Among his findings, what is of immediate interest to us is the tendency of college-age Dutch speakers to reject the typical English inchoative sentence in (13), favoring the causative in (14) and agentless non-adjectival passive in (15). (13) The cup broke. ( 14) Someone broke the cup. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144 (15) The cup was broken. This finding was rather surprising to Kellerman because there appeared no a priori reason why his subjects were less inclined to consider (13) as English-like compared to (14) and (15). After all, not only are (13)-(14) perfectly grammatical, but so are their Dutch equivalents. Furthermore, even in comparison with the following sentences with a metaphorically extended meaning, (13) was judged less English-like by his subjects. (16) She broke his heart. ( 17) After the accident he has become a broken man. (18) He broke his word. This second finding was even more troubling to Kellerman because his research on the relationship between semantic markedness and transferability of lexical items otherwise showed a general tendency of learners to view marked senses of a verb as less transferable to a target language than its more literal or "core" meanings. Kellerman concluded that the inchoative breken was somehow perceived as semantically more marked than the causative breken by his Dutch-speaking subjects, and consequently considered less transferable to English. The nature of this markedness associated with the inchoative use of the verb, however, was left unaddressed in Kellerman's study, in which his main concern was the issue of more general semantic markedness and transferability. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 145 Kellerman's finding was replicated when he asked 32 first-year college students to translate the Dutch sentence in (19) to English (Kellerman 1979:38-40). (19) Het kopje brak. (19) is the perfectly grammatical Dutch counterpart of (13), but 25 of the 32 subjects avoided the direct translation, instead, giving their answers in the passive structure or choosing paraphrases with other verbs. A follow-up of the breken studies is presented in Kellerman 1983. This time, Kellerman asked 70 Dutch speaking students to complete the sentence The child hit the vase until.... with an appropriate form of break. Out of the 70 subjects, 66 correctly supplied it broke for the answer. Kellerman interprets this result as evidence that the tendency to avoid the inchoative break can be overridden by pragmatic considerations. What is significant for our concern here is that this additional finding that the avoidance of the inchoative break is far less absolute than the first two studies at first may suggest is an indication that L2 grammar needs to be seen as a highly permeable entity. 3.1.2.2. Yip 1989, 1994, 1995 The avoidance of inchoative verbs in the NP-V word order is also observed by Yip (1989, 1994, 1995). In her contextualized grammaticality judgment test. Yip asked her subjects to supply corrections when they regarded given experimental sentences as ungrammatical. Many subjects rejected grammatical sentences with inchoative verbs in (20a-c), providing "corrections " in the "passive" structure. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146 (20) a. The mirror shattered during the earthquake. ~> was shattered b. My car has broken down. — > has been broken/was broken c. We had some ice cream, but it has melted. ~> has been melted/was melted (Yip 1994:129, (12)-(13), (15)) 3.1.2.3. Hirakawa 1995 As reviewed in section 3.1.1.5, Hirakawa (1995) has shown that her subjects inappropriately "passivized" inchoative verbs such as break, spill and dry, when the discourse contexts clearly demanded the intransitive use of these verbs. This tendency of Japanese speakers of English to undergenerate sentences with inchoative verbs was also observed in the results she obtained from the judgment test given to the same group of subjects. Unlike the native speaker subjects, the Japanese subjects were found to have difSculty in accepting the inchoative verbs in the correct NP-V word order. Hirakawa's results from both production and judgment tasks, therefore, give further support to the reality of inchoative avoidance in the NP-V word order in L2 English.^o Ingham (1996) reports that a grammaticality judgment test he administered to a group of Cantonese speakers has also shown a high rejection rate (i.e. more than 50%) for grammatical English sentences with inchoative verbs. Unfortunately, the number of subjects and their proficiency levels are not presented in the manuscript version of his paper that I obtained. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147 3.1.3. Transitivization of Unaccusative Verbs Another type of structure observed in English learners' use of unaccusative verbs is the (causative) transitive structure. 3.1.3.1. Rutherford 1987 Drawing on the observation originally reported in Scarceila 1984, Rutherford (1987) focused on a creative, though non-target, transitivization apparent in sentences such as (21a-b). (21) a. *The shortage of fuels occurred the need for economical engine. b. *This construction will progress my country. (Rutherford 1987:89) These examples show that although verbs such as occur and progress have only intransitive use in the grammar of most native speakers, they may be used as transitives by English learners. Rutherford suggests that this phenomenon observed in the Interlanguage English of various LI speakers is a result of overgeneralization of a causativization rule, which accounts for the ergative alternation in English, i.e. the alternation between the inchoative and causative versions of verbs such as open, melt and move.^^ 3.1.3.2. Yip 1989, 1994, 1995 Rutherford's (1987) examples are quoted by Yip (1989, 1994, 1995) to support her argument that the "passive" unaccusatives are produced based on the learners' misclassification of unaccusatives as underlyingly transitive verbs. According to In Zubizarreta (1987) and Levin and Rappaport Hovav (1995), this correspondence is argued to be a consequence of a detransitivization rule rather than a transitivization rule. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 this idea, the "passivized" unaccusatives can be considered as genuine passive sentences which are derived from transitive verbs hidden in the learners' lexicon. Drawing on the LI acquisition literature such as Carey 1985 and Roeper 1987, Yip argues that such a misanaiysis is likely to be present in L2 acquisition as well, due to language learners' tendency to assume the existence of an external force for a change of state (Yip 1995:138-140). A strong interpretation of this idea may be that L2 learners do not differentiate general transitive verbs and change-of-state unaccusative-inchoative verbs at all. This interpretation, however, is clearly too strong since there are many instances in which unaccusative-inchoative verbs correctly appear in active sentences produced by English learners. Therefore, I interpret Yip's idea as a suggestion that a lexical transitivization process may be at work in learners' grammar when they produce "passivized" unaccusative- inchoatives. This transitivization may be schematized as follows. (22) unaccusative-incljioative argument structure: (0 (y)) < = transitivization transitive argument structure: (x (y)) The transitivization in (22) adds an external argument (x) to the original argument structure of an unaccusative-inchoative verb. The newly added external argument, however, may not necessarily show up as a syntactic constituent since the external argument is suppressed in the process of passivization and the suppressed argument need not be projected into syntax (Jaeggli 1986b, Grimshaw 1990) (cf. sections 2.1.3 and 2.2.6). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 3.1.3.3. Hirakawa 1995 Hirakawa (1995) reports an important finding concerning the transitivization of unaccusative-inchoative verbs by Japanese speakers of English. As already noted, Hirakawa has found that Japanese learners of English have a tendency to accept incorrect "passivization." Interestingly, this tendency was found to be stronger than an equally plausible bias to accept incorrect transitivization of unaccusative verbs. That is, her subjects were found to be more inclined to regard ungrammatical "passivization" of unaccusative verbs as grammatical than to accept non-target transitivization of them. Hirakawa summarizes this finding by stating "|T|t was not always the case that those who allowed incorrect passivized unaccusatives also allowed transitive counterparts (Hirakawa 1995:300)." This observation, therefore, cautions against a hasty, though reasonable, interpretation that "passivized" unaccusatives generally come from their hidden transitive counterparts. In other words, the "passivization" may not always be attributable to a lexical transitivization rule such as (22). Consequently, the "passivization" of unaccusatives has to be accounted for, at least in part, independently of a lexical reanalysis of unaccusatives as transitives. 3.1.3.4. Zobl 1989 Finally, a slightly different type of transitivization discussed in the literature should be mentioned. Zobl (1989) lists (23a-b) as examples of one of the three non-target structures he found in his data. (23) a. *I hope he's always light up his face. (in the sense of) I hope his face will always light up.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150 b. *I changed myself a lot in those years. (in the sense of) I changed a lot in those years.’ (Zobl 1989:216, (16), (18)) Here, the verbs in (23a-b) are assumed to belong to the unaccusative-inchoative class because both have a sense of the change of state but do not have a clear external (or internal) causer. What is interesting is that in dealing with these verbs, the learners inserted non-causative arguments in subject position and kept the internal arguments in object position. The insertion of additional arguments blocks the movement of the intemal arguments to subject position. This non-causative transitivization, therefore, semantically contrasts with the causative transitivization discussed in sections 3.1.3.1 to 3.1.3.3. 3.1.4. Postverbal NP Structure The last structure I review here with respect to the use of English unaccusative- inchoative verbs by nonnative speakers is the postverbal NP structure. This is the third type of non-target structure that Zobl (1989) discusses in his work. 3.1.4.1. Zobl 1989 The following three sentences come from Zobl's Japanese subjects, who are in "low" to "high intermediate" proficiency levels. (24) a. *Sometimes comes a good regular wave, b . *I was just patient until dried my clothes. (in the sense of) I was just patient until my clothes had dried.' c. *I think it continue of today condition forever. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 51 (in the sense of) "I think the condition of today/today’ s condition will continue forever.’ (Zobl 1989:204, (4)-(6)) What is common in (24a-c) is that the surface subjects, which would appear preverbally in target English, are all in postverbal position. This position, according to the Unaccusative Hypothesis, is the object position where the arguments are generated at D-Structure. With the ill-formed "NP" o f today condition aside, (24c) is especially interesting because an expletive it occupies the surface subject position. Accordingly, Zobl characterizes all three sentences as (Pro) V NP structure. 3.1.4.2. Rutherford 1989 Some examples of the postverbal NP structure are also presented and extensively discussed in Rutherford 1989. Rutherford investigated the ways in which the canonical word orders and the grammatical versus pragmatic word orders (GWO/PWO)5- of the learners' target and native languages conspire with each other in shaping their interlanguage. PWO languages, in contrast with GWO languages, allow relatively free permutations on their canonical word order according to discoursal contexts. Spanish and Arabic, for instance, are considered to be PWO languages. Rutherford found that the native speakers of these languages occasionally violate the canonical word order of English, SVG, in their compositions. Apparently their L is’ permissibility on word order permutation is Characterization of a language as GWO or PWO is not meant to be absolute, but a matter of degree. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 carried over to their Interlanguage English. For example, (25a-b) were produced by LI Spanish speakers and (26a-d) by LI Arabic speakers.^ (25) a. *On this particular place called G happened a story which now appears on all Mexican history books,... b . And then at last comes the great day. (Rutherford 1989:178-179) (26) a. *... but lately happen some extra things or littel changes on this custom because of the civilization. b . After that, they'll be lead to their house, and with that comes the end of the wedding. c. The first people to come into the church were the closest relatives of the bride and bridegroom and then came the rest of the invited people. d. The bride was very attractive, on her face appeared those two red cheeks and above them beautiful deep eyes. (Rutherford 1989:179) In (25) and (26), only the verbs that require a single nominal argument are taken out of Rutherford's more extensive list of examples, which also contains the verb be, the raising verb appear, adjectives with a prepositional argument, passive predicates, and so on. As Rutherford himself noticed, what is remarkable is that most of the postverbal NP sentences collected by him contain unaccusative verbs and other verbs that manifest unaccusativity. Considering the restrictedness of the verb class that emerged in non-canonlcal English word orders and also the fact that possible word order permutations permitted in Spanish and Arabic are far more Notice that some of these examples may not be unacceptable in target English, especially in stylistically sophisticated prose. What is an issue here is that these examples are produced by only LI Spanish and Arbic speakers who otherwise appear to be not so sophisticated in their written English style. See also examples (39)-(40) below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 wide-ranging (for instance, they also involve sentences that contain unergative and transitive verbs), I suspect that what is carried over to the L2 English of the Spanish and Arabic speakers is not the trait of PWO in general, but something more specific in its syntactic consequence. Later in section 4.4.2,1 argue that this specific syntactic property that is susceptible to transfer is the null expletive subject whose existence is argued for the so-called "pro-drop" languages based on theoretical assumptions. 3.1.5. Summary When noticed in the early 70s, the anomaly in nonnative speakers' use of certain English intransitive verbs received no special attention either from ESL scholars or from theoretical linguists. Perlmutter's (1978) Unaccusative Hypothesis, however, provided a much needed theoretical framework in which to analyze the phenomenon. Since then, various L2 studies have collectively established that the subclass of intransitive verbs, i.e. the unaccusative-inchoative class, present a special linguistic challenge to L2 learners of English. There is enough evidence now that English unaccusative-inchoative verbs often lead to L2 phenomena such as "passivization," transitivization and the postverbal NP structures and avoidance. Moreover, these phenomena are found in the Interlanguage English of speakers of various first languages and they appear despite the fact that they are ungrammatical not only in target English but also sometimes even in the learners' native languages. I have reviewed the accounts for the non-target phenomena suggested by L2 researchers. The accounts come in many different forms and with varying emphasis on different components of grammar. Some are syntactic, others are lexical, and still others may be seen as conceptual. Some researchers suggest Ll- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154 based transfer accounts while others argue for IL-specific explanations. Beyond some basic understanding brought about thanks to the Unaccusative Hypothesis, however, many issues still remain unresolved or even previously unaddressed. The second half of this chapter attempts to critically evaluate these proposals and address other related issues. 3.2. UNRESOLVED QUESTIONS 3.2.1. Range of Structural Variations So far in the literature, only a few structural patterns have been pointed out with respect to the way unaccusatives and inchoatives are used by nonnative speakers of English. Aside from the inconspicuous phenomenon of inchoative avoidance, these IL structures are the "passive" unaccusative-inchoative, the unaccusative-based transitive and the postverbal NP constructions, all of which are ungrammatical in target English. They, however, by no means, exhaust the logically possible range of structural variations in which unaccusative-inchoative verbs could appear. For instance, think of the postverbal NP structures that we have reviewed in section 3.1.4. Zobl (1989) discussed two types of such structures, namely VNP and it V NP. However, there is one more possibility here, i.e. there V NP, which actually results in grammatical English sentences with a certain semantically restricted subclass of unaccusative verbs: (27) a. There appeared a huge iceberg in front of the ship. (appearance) b . There exist some challenging problems for all of us. (existence) c. There arrived three boys at the station. (inherently directed motion) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 155 Similarly, we cannot ignore the potential existence of structures which combine the postverbal NP and the "passive" verbal phrase of be + Ven. That is, we need to investigate if L2 learners of English also employ structural patterns such as be + Ven NP, it be + Ven NP, and there be + Ven NP. Structures equivalent to them are in fact grammatical in some languages. Consider (28a-b) in Italian and French. (28) a. 0 sono arrivati tre ragazzi alia stazione. (Italian) b . n est arrivé trois garçons à la gare. (French) The question is whether their hypothetical counterpart such as (29) is in fact observed in Interlanguage English. (29) * 0/It/There is/are arrived three boys at the station. (hypothetical IL English) In this respect, the question of possible structural variations in which unaccusative- inchoatives may appear in IL English is also closely related to an issue of potential LI transfer. In short, one of the fundamental goals of the research on the L2 acquisition of English unaccusative-inchoative verbs is to determine the full spectrum of syntactic variations in which this class of verbs appear. The following discussion in this section should prove that this question in fact is a prerequisite to any attempt to determine the true meaning of the non-target structures discussed in the literature. Without first determining which structural patterns are actually used and which are not, we cannot possibly answer why those in use are chosen by L2 learners of English. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156 3.2.2. Questions Raised by Suggested Accounts Now let us return to the three commonly observed structures that we have reviewed in sections 3.1.1, 3.1.3 and 3.1.4. They are postverbal NP structures, transitivization and "passivization." Currently, they have not been given satisfactory accounts that most L2 researchers can agree on. In the following pages, I critically evaluate the proposed accounts, point out questions still unresolved or newly raised by the accounts themselves, and discuss what kind of evidence would be needed to support or refute them. 3.2.2.1. postverbal NP structures The major question here is why the canonical Subject-Verb word order of English is so blatantly violated by some learners. Rutherford's (1989) data clearly show that not everyone commits this kind of error. Specifically, in his study the postverbal NP structures were found in the compositions of LI Spanish and Arabic speakers, but not in those of LI Japanese speakers. His finding contrasted with Zobl 1989, in which the author cited three examples of this structure produced by LI Japanese speakers. Nevertheless, both studies have shown that it is not just any kind of intransitive verbs but specifically unaccusative-inchoative verbs that are susceptible to the postverbal NP structures. Why is such the case? Rutherford answered this question by referring to potential transfer of the LI trait of pragmatic word order. This, however, appears to fall short because it clearly does not apply to sentences with unergative and transitive verbs. Zobl argued that in the postverbal NP structure, the argument of an unaccusative-inchoative verb is in object position where it is projected at D-Structure. In other words, the IL violation of the English Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 canonical word order, in Zobl's account, reflects non-application of NP-movement on the argument. Why, then, do LI Japanese speakers not produce postverbal NP structures as often as LI Spanish and Arabic speakers? Or, do they also employ them as often, as suggested by Zobl's data? These are important questions to be investigated more thoroughly. 3.2.2.2. transitivization The non-target transitive structure also presents some intriguing questions. One fundamental question, as with any other potential IL structures, is how commonly transitivized unaccusatives are used by English learners. Furthermore, are non target transitives created on any semantic subtypes of English unaccusatives? For instance, are verbs such as remain and exist as likely to be transitivized as occur and progress? Also, if there is a lexical transitivization process at work behind the "passivization" of unaccusative verbs, why do most "passivized" unaccusatives appear without an agentive by phrase? Finally, what does this all mean in light of Hirakawa’ s (1995) finding that the learners who allow incorrect "passivized" unaccusatives do not necessarily accept incorrect transitive counterparts? These are some of the important questions to be asked with respect to the E L transitivization of unaccusative verbs. 3.2.2.3. "passivization" As reviewed in section 3.1.1, the "passivized" unaccusative-inchoative structure has generated various suggestions for its interpretations, which are repeated below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 (30) a. Direct transfer from similar LI structures such as Italian "passato prossimo" and French "passé composé" (section 3.1.1.2.) b. Overgeneralization of passive and perfective adjective formation (section 3.1.1.2.) c. Lexical reanalysis of unaccusatives as transitives (section 3.1.1.4.) d . Identification of the lack of logical subject and the passive morphosyntax (section 3.1.1.3.) e. An overt marker of syntactic NP-movement (sections 3.1.1.3 and 3.1.1.4.) With some serious thinking, most of these proposals, however, seem to present as many questions as they are intended to answer. The LI transfer account based on the morphosyntax of the learner's LI in (30a) is probably the most commonly heard interpretation^'* and actually works well in the case of learners with certain LI backgrounds. In addition to French and Italian, German and Dutch, for instance, employ the same structure with unaccusative verbs. The problem, however, is that this account does not work in the case of native speakers of other languages. As Yip noted, this structure in Chinese, for example, is not grammatical and therefore caimot be a source of the non-target "passivized" unaccusative-inchoatives. The situation is no different in Japanese and Korean. Neither of these languages allows the structure in question either. ^ For example, in the fall of 1996,1 posted a query on the LINGUIST List, asking for information about unaccusativity in Korean. To indicate my research interest in L2,1 included in my posting a couple of examples o f "passivized" unaccusatives from IL English. To my surprise, my query generated more messages on the interpretation of the examples than it did on the literature on Korean unaccusative verbs. Most messages I received suggested that the cause o f the "passivization" is LI interference. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159 (31) a. Nani-ga okitano? what-NOM happen-PAST Q "What h ap p en e d ?' b. *Nani-ga okiraretano? what-NOM happen-PASS-PAST Q "W hat w a s happened?' (32) a. K inoha-ga otita. leaves-NOM fall-PAST 'T he le av es fell.' b . * K in o h a-g a otirareta. leaves-NOM fall-PASS-PAST 'T he leav es w ere fallen.' (33) a. Mwusun il-i nass-ni? w hat k in d o f event-NOM happen-PAST-Q "W hat k in d o f ev en t happened?' b . *M w ues-i naciess-ni? what-NOM happen-PASS-PAST-Q "w hat w a s happened?' (34) a. N am w u iph-i tree leave-NOM T h e le av es fell.' tteleciess-ta fall-PAST-DEC Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 b. *Namwu iph-i ttelecieciess-ta tree leave-NOM falI-PASS-PAST-DEC 'The leaves w ere fallen .' The examples in (31)-(34) clearly show that the unaccusative verbs cannot be passivized in Japanese and Korean. However, the "passivized" unaccusatives are produced in the IL English of the speakers of these two languages. It has to be concluded, therefore, that the potential LI transfer cannot be the reason for the "passivization " of the unaccusative-inchoative verb class found in L2 English of Japanese, Korean and Chinese native speakers. Notice, however, that this does not necessarily mean that the LI transfer account is totally out of the question. This is because in theory there can be more than one source for this kind of IL structure. One source may be LI transfer available only to certain LI groups and another (or others?) may be accessible to all L2 learners of English. If such is the case, we may ask, being open to both mechanisms, do speakers of Italian, French, German, Dutch, etc. produce more "passivized" unaccusatives and inchoatives in their IL English than the native speakers of Chinese, Japanese, Korean, and so on? This is an interesting question to raise, but to answer it we need extensive crosslinguistic data. The overgeneralization of passive and perfective adjectives in (30b) is a lexical solution to the question of the "passivized" unaccusative-inchoatives. In this account, the rare appearance of by phrases in the "passivized" unaccusative- inchoative clauses does not present a problem because the passive and perfective adjective formation deletes the external argument of the base verb if there is one. If there is no external argument as in the A-Structure of an unaccusative-inchoative Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 verb, the deletion applies vacuously without making the derived adjective ungrammatical (section 2.1.3). Either way, there is nothing in the A-Structure requirement of the derived adjective that licenses a by phrase. One serious problem to the account in (30b), however, is the semantic change that results from the adjectivization. All adjectives basically have stative meaning. The passive and perfective adjectives are no exception to this generalization. Therefore, whether or not all or even many "passivized" unaccusative-inchoatives in L2 English have stative reading is a question that should be carefully investigated. The questions related to the lexical reanalysis of unaccusatives as transitives in (30c) are already discussed in section 3.2.2.2. In order to apply this lexical reanalysis to the "passivization," the questions raised there must be answered first. Finally, let us discuss the last two interpretations of the "passivization" of unaccusative-inchoatives in (30d) and (30e). On the one hand, (30d) says that L2 English learners tend to associate the lack of logical subject in the A-Structure of the unaccusative-inchoative verb with the passive morphosyntax of be + Ven. On the other hand, (30e) argues that the ungrammatical "passivization" serves as a marker of syntactic movement of the NP argument from object to subject position. The two interpretations are so closely related that Zobl (1989:210-211) appears to treat them as two phases of a single process. The clear separation of them, however, is essential to understand the true nature of the "passivized" unaccusative-inchoative structure. If the passive morphosyntax is solely related to the D-Structure position of the NP argument, we should wimess one of the patterns in (35a-c). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 (35) a. 0^^ be + Ven NP b. it be + Ven NP c. there be + Ven NP In (35a), the NP argument simply stays in situ and nothing overt exists in the surface subject position. In (35b-c), the NP is also in situ, but the subject position is occupied by one of the two expletive nouns, i.e. it or there. Notice that these structures have already been mentioned in section 3.2.1 in terms of the possible range of structural patterns and in relation to potential LI transfer. The issue here, however, is not simply whether structures similar to those in (35) are transferred to the L2 English of the speakers of certain L is such as French and Italian. If (30d) has any validity, it is predicted that IL English grammar should generally allow sentences in the structures of (35a-c). Then, native speakers of Chinese, Japanese, Korean, etc. which do not have such structures are also expected to produce IL English sentences in the form of (35a-c). What is at issue in this respect, therefore, is the meaning of the "passivization" in general, which is potentially relevant to the speakers of any Lis. In contrast, if the main function of the "passive" structure is overt marking of syntactic movement of an argument from object to subject position, the S- Structure representation of (36) should be generated. (36) NPi be + Ven ti t _____________ The symbol "0" indicates that there is no overt linguistic element in the normal surface subject position. To determine whether there is in fact nothing there or that there may be something without phonetic content in this position is one of the questions investigated later in this work. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 In (36), the NP appears in the surface subject position as a result of syntactic movement. Looking at the same picture firom a slightly different perspective, we can say that sentences in the pattern of (35a-c) are exactly the kind of data we need in order to refute (30e). Postverbal NPs that appear with the passive morphosyntax clearly pose a challenge to (30e) since the passive structure cannot be a marker of any movement in such cases. If a substantial number of examples of this kind are found in actual IL English, therefore, the account of (30e) will be in serious trouble. (What (30e) does not say, however, is that all preverbal NPs are marked this way in IL English. In other words, preverbal NPs without the passive morphosyntax are not a problem to the idea expressed in (30e), though the lack of the ungrammatical "passivization" in those cases still needs to be accounted for in some way.) So, is there any evidence that IL English actually generates sentences in the pattern of (35a-c)? Zobl suggests, quoting the example in (37), that English learners do generate the structures in (35). (37) *It was nearly killed all of us. (Chinese LI; high intermediate) All of us were nearly killed.' (Zobl 1989:215, (15)) The problem with (37), however, is that it is not a structure built on an unaccusative verb but a so-called "impersonal passive" observed in languages such as French. Compare (37) and (38). ^ This should not be equated with another structure also called "impersonal passive" that is formed based on unergative verbs in languages such as Dutch and German. (i) Er wordt door de kinderen op het ijs geschaatst. (Dutch) I t is skated by the children on the ice.' (Perlmutter 1978:157, (1)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164 (38) n a été tué un homme, it has been killed a man 'A man has been killed.' (Belletti 1988:6, (10a)) Although a passive sentence also manifests unaccusativity in the spirit of Burzio's Generalization presented in section 1.2, (37) reveals nothing about English learners' grammar with respect to the unaccusative-inchoative verb class. Therefore, I have searched for evidence for postverbal NP "passivized" unaccusative-inchoative sentences in the L2 literature and found (39a-b) and (40) produced by LI Spanish and Arabic speakers respectively. (39) a. *]nth&{dkcotM;aidtcalbo was discovered the oil. b . *In this one was placed the national school of engineering. (Rutherford 1989:179) (40) On the walls of this monument are written the names o f the victories of Napoleon's battles. (Rutherford 1989:179) Notice, however, even these are the same as Zobl's example in (37), namely impersonal passive sentences that are based on transitives, not on unaccusative- inchoative verbs. I have not found a single example in a published work that falls in the structural patterns of (35a-c) and contains an unaccusative-inchoative verb. A more extensive inquiry into this issue, therefore, is needed to give a definite answer to Zobl’ s dual interpretation of the "passive" unaccusative-inchoative structure. (ii) Hier wurde den ganzen Abend getanzL (German) I t was danced here all evening.' (Perlmutter 1978:158, (5)) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 165 3.2.3. Developmental Variations on IL Structures Very few studies to date have paid serious attention to the developmental aspect of the L2 acquisition of unaccusative-inchoatives. A few studies that have touched upon the issue look at a particular phase of the development rather than an overall picture. One work which presents a detailed view of an overall developmental sequence in the L2 acquisition of English unaccusative-inchoatives is Zobl 1989. Following Baker (1983), Zobl assumes that L2 learners arrive at the correct argument structure of unaccusative-inchoative verbs, i.e. (0 (y)), through their inherent lexical semantics. As he puts it, "the unaccusative group is well enough defined crosslinguistically to warrant the assumption that universal aspects of predicate-argument structure make the distinction available [to the learner— HO] (Zobl, personal communication 1993)." Once the correct A-Structure representation for unaccusative-inchoative verbs is acquired, the learner faces a syntactic challenge, namely, how to use the class of verb without an external argument when the syntax of English requires an overt subject in every finite clause. Zobl suggests that the first solution the learner adopts in the face of this challenge is lexical in nature. He argues that instead of projecting the NP argument of an unaccusative-inchoative verb directly to D-Structure, the learner first acquires lexical NP movement which externalizes an internal argument prior to syntax (à la Keyser and Roeper 1984). This process, presumably at work in the lexical component of the learner's grammar, may be illustrated as in (41). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 (41) A-Strpcture I: (0 (y)) < = = lexical extemalization A-Structure 2: (x(0)> [x = y] Due to this lexical rule, the sole argument of an unaccusative-inchoative verb is projected asthe sister of V at D-Structure. This way, the lexical extemalization in (41) is intended to account for the target-like usage of unaccusative-inchoative verbs in the NP-V word order found in his data. Zobl also suggests three syntactic solutions to the unaccusative-inchoative challenge. The first is to project the internal argument ((y)) in (41) directly into object position at D-structure and keep it in situ at S-Structure where it receives Nominative Case. This, he believes, leads to the V-NP word sequence observed in his data. In his view, this is equivalent to the V-NP sentences with unaccusative- inchoatives in Italian. The second syntactic solution, according to Zobl, is the postverbal NP structure with an expletive subject, which seems to indicate awareness on the part of learners that English requires an overt NP in subject position. Zobl considers that an inserted subject NP in the non-causative transitivization structure (section 3.1.3.4) functions much the same way as a regular expletive pronoun. Finally, the third syntactic solution becomes available to the learner when the passive structure is acquired. The learner projects the right D- Structure for unaccusative-inchoatives in analogy to that of the passive sentence, and then moves the underlying object to subject position. Zobl argues that the apparent non-target surface structures that result from these solutions are evidence of learners' attempts to cope with the syntactic challenge presented by the argument stmcture of unaccusative-inchoative verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 In his account, lexical movement is considered as developmentally most primitive. As for the three syntactic solutions, the grammatical development is supposed to proceed in the order of presentation above. However, he concludes that even the developmentally most advanced solution, namely the "passivization," never really supplants lexical movement, keeping the learner's IL grammar in an indeterminate state. Consequently, the target-like NP-V word order is likely to persist in the production of English of learners who frequently use other non-target structures. Zobl's detailed developmental account, however, presents some problems. First, his argument that lexical movement is acquired early but partly replaced by syntactic movement later is hardly consistent with his own view expressed in the same work that the lexical movement analysis cannot be the correct account for English unaccusative sentences. To adopt it solely to account for the otherwise puzzling Interlanguage data, therefore, is not very appealing. The second problem concerns the transition of the proposed developmental stages. If most learners adopt the lexical movement solution first, why do they opt for syntactic movement later? Since the lexical movement solution derives nothing but the observationally adequate surface sequence of NP-V, there appears to be no strong incentive for learners to reorganize their X L English grammar. The semantic similarity between the passive and the unaccusative-inchoative is well taken, but it does not appear to be a strong enough reason for them to move away from the observationally adequate grammar. Third, Zobl's lexical movement account opens a new question concerning the data obtained on the L2 acquisition of English middle verbs. Although this takes us away from the issues we have been concerned with, I discuss it here Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 6 8 because it is essential in order to critically evaluate Zobl's developmental account of the L2 acquisition of English unaccusative-inchoatives. Hirakawa (1995) found that grammatical English middle sentences such as The book sells well were generally rejected as ungrammatical by her Japanese subjects. This clearly contrasted with the general acceptance of well-formed passive and grammatical unaccusative sentences by the same subjects. Because Hirakawa assumed that the middle construction involved syntactic NP-movement as argued by Keyser and Roeper (1984), Hale and Keyser (1986), Roberts (1987), Levin and Rappaport Hovav (1995), etc. and also because she regarded the L2 acquisition of NP-movement as construction-sensitive, she concluded that learners did not acquire syntactic NP-movement in the middle constmction as easily as they did in passive and unaccusative sentences. The theoretical validity of this assumption aside,^^ Hirakawa's conclusion leaves unanswered the question as to why there is such a difference in the relative difficulty in acquiring the same syntactic process, i.e. NP-movement, depending on the constructions. However, as Hirakawa herself notes (Hirakawa 1995:301, foomote 9), the issue of the derivation of middles is still a controversial one in the theoretical literature (cf. Hale and Keyser 1987). Actually, there is a contrasting view that middles are formed by extemalization of the internal argument in the lexicon, which is defended in at least as many studies (e.g. Fellbaum 1986, Williams 1987, Fagan 1988, Levin and Rappaport 1988, Doron and Rappaport Hovav 1991, Oshita 1994, 1995, Ackema and Schoorlemmer 1995). In addition, Authier and Reed (1996) In the GB framework (e.g. Chomsky 1981), grammatical principles and parameters are generally considered construction-neutral. However, Hirakawa's assumption is not about the linguistic system itself but about the acquisition of it. It appears to me, therefore, that her research question is a valid one. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 169 recently investigated "middle constructions" of some languages and concluded that unlike Continental and Canadian French middles which are syntactic, English middles are in fact lexical. Now, if the lexical analysis of English middle formation is correct, we can account for the very clear contrast in Hirakawa's subjects' judgments on the passive and unaccusative sentences on the one hand and the middle sentences on the other, by attributing the difference to the relative difEculties in acquiring syntactic and lexical movements.^» This account, by the way, also explains why Yip's ( 1994:135) subjects had unexpected difficulty dealing with the sentence What cooks most quickly? in the contextualized grammaticality judgment test. Although this usage of cook was considered as an inchoative by Yip, its stative generic property reading suggests that it is actually a middle verb. Also consider that cook normally takes an agentive subject unlike the true inchoative verbs such as break and melt whose causative usage does not discriminate external arguments as long as they are causative entities. A sentence with a non-agentive subject like The sun cooked the eggs clearly has a figurative reading. This is why NPs such as cooked eggs have only the reading based on the transitive sense while broken glasses and melted icecream may have a reading based on either their transitive or inchoative sense. (Recall the discussion in section 2.1.3.) These facts strongly suggest that the reason why cook posed greater difficulty to Yip's subjects than other inchoatives was that the verb was in fact presented as a middle verb. If so, this case, too, can I suppose that the relative difficulty of acquiring the lexical middle formation over the syntactic passive and unaccusative constructions is at least partly due to the radical semantic change involved in the first. For one thing, the middle formation completely deletes an external argument from the A-Structure requirement in contrast to the suppression of it in the verbal passive formation. For another, the semantics o f a verb changes from the activity/process of its transitive use to the stative property o f its middle version. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170 be accounted for in the same way as Hirakawa's finding. That is, the middle construction is generally quite difQcult for L2 learners of English to acquire. In sum, I believe that the proposed reanalysis of Hirakawa's and Yip's data on middle verbs is a quite reasonable one. This analysis, however, brings up a puzzling question about Zobl's lexical movement account for unaccusative- inchoative verbs. That is, if Zobl's theory is correct, why is it so difficult for L2 learners to acquire English middles when they are supposedly using essentially the same rule of lexical movement with unaccusative-inchoative verbs?^’ This suggests that whatever mechanism is at work to produce the NP-V sequence with unaccusative-inchoatives in IL English, it may be fundamentally different from the lexical movement proposed by Zobl. 3.2.4. Summary This section has critically evaluated the existing L2 research on the acquisition of English unaccusative-inchoative verbs. It has focused on the structural and developmental aspects. In the process, it was also pointed out that the crosslinguistic aspect such as LI transfer of certain structures must be examined. As the discussion above has made clear, the structural, crosslinguistic and developmental aspects in the normative acquisition of English unaccusative- inchoative verbs are closely intertwined. To investigate their subtle and complex relations is in fact one of the main goals of this thesis. In the following chapter, I try to answer some of the many questions raised in the second half of this chapter. Notice that this is the same kind of question as we face in Hirakawa's explanation of her data. That is, when the acquisition of a linguistic process appears construction-sensitive, what is the nature of this sensitivity? The question seems unavoidable when construction-sensitivity is held central to an explanation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 by analyzing some data obtained from a large computerized corpus of L2 English learners. In this process, I pay as much attention to the observable normality of IL English as to its non-target characteristics such as those discussed so far. To date, most of the existing studies have focused on the non-target properties of a few IL structures. As a result, we have gained some important insights into the ways the learner English works. However, to really understand why it works the way it does, it is not enough to concentrate on the aberrations. We need to look at the whole picture more carefully, including the target-like NP-V sequence used by L2 learners. In other words, in order to account for the obvious aberrations, we cannot avoid looking into the mundane. In next chapter, I argue that the key to solving much of the mystery of the L2 unaccusative-inchoative acquisition is in the treatment of the most ordinary structure produced by the learners. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172 4. Corpus-Based Research As we have seen in section 3.1, previous studies on the acquisition of unaccusative- inchoative verbs in L2 contexts have either looked at a particular LI group (e.g. Dutch in Kellerman 1978, Chinese in Yip 1989, 1995, Japanese in Hirakawa 1995) or dealt with a mixed language group in which some or all Lis are represented by a relatively small number of native speakers (e.g. Zobl 1989, Yip 1994). Two negative consequences of this situation are the lack of a single study which, by itself, convincingly demonstrates the crosslinguistic pervasiveness of unusual structures involved in the acquisition of unaccusative-inchoative verbs and the general neglect of the possible influence of the learners’ first languages on the interlanguage English structures that they produce. Clearly what is needed in order to overcome the shortcomings of the previous research is a study based on a large database of actual L2 English produced by various first language groups. In the research reported below, the data were gathered from a large computerized corpus and the research procedure was designed in order to make it possible to see the structural variations built on unaccusative verbs in their entirety rather than solely according to their well- or ill- formedness in light of the target grammar. The research corpus and procedure allow us to investigate both similarities and differences in the ways in which unaccusative verbs are used by native speakers of various first languages. Analyzing the data, particular attention was given to three main topics concerning the L2 acquisition of English unaccusative verbs, i.e. its structural, crosslinguistic and developmental aspects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 4.1. M a t e r ia l s The data for the study reported here^ was gathered from the 1993 version of the Longman Learners Corpus (henceforth, LLQ. The LLC is a large computerized database containing written English produced by native speakers of various languages. Four first languages, i.e. Italian, Spanish, Japanese and Korean, were chosen for the study and a total of 3362 essays written by native speakers of these languages were separated as a research corpus from the entire LLC. The number of essays in each language group was: Italian (684), Spanish (1079), Japanese (1363), and Korean (236). 4.2. M e t h o d 4.2.1. Selection of Verbs Ten unergative and ten unaccusative verbs were selected for the study. They are: (1) Unergatives: cough, cry, dance, joke, laugh, shout, smile, speak, talk, walk Unaccusatives: appear, arise, arrive, die,^^ disappear, exist, fall, happen, occur, rise The study presented in this chapter is part of the research done in collaboration with Joe Allen in the Linguistics Department of USC. I am grateful to Joe who kindly agreed to help me work with the computerized database. As noted in section 2.4.10, Kageyama (1993) shows that a verb in the sense o f die' can be either unergative or unaccusative. He argues that the Japanese verb sinu is unergative and siboosuru and nakunaru are unaccusatives. A similar observation is made on two Hebrew verbs that denote die': ligsos is unergative while lamut is unaccusative (Levin and Rappaport Hovav 1992). In addition, Rosen (1984) presents an interesting crosslinguistic observation on verbs of "dying". The English verb die was included in the 10-member unaccusative class because o f its tendency to be misused ftequently by nonnative speakers, which has been widely observed in the L2 literature. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 The selection of verbs fiom the unaccusative-inchoative class was deliberately restricted to unaccusatives, excluding inchoative verbs, in order to facilitate the investigation of transitivization and “passivization” of this class of intransitive verbs in learners' grammars (sections 3.1.3 and 3.1.1). Being unaccusatives, the ten verbs in (1) are not normally used as transitives by most native English speakers. Therefore we can assume that any instance of a “passivized” or transitivized unaccusative that we may find in the research corpus could not have been encountered in the input and would thus be an interlanguage-particular innovation. The preselection of verbs contrasts with the methods used in some previous studies of learners’ production data. For example, Zobl (1989) took absence of volitionality as a criterion of unaccusative verbs when collecting his data. This decision, on the one hand, led him to include in his paper a wide range of verbs which manifested unaccusativity. Included in his data were, for instance, inchoatives (e.g. break and change), verbs that require adjectival complements (e.g. become), phrasal verbs (e.g. fall asleep) and idiomatic phrases (e.g. fall in love). On the other hand, his criterion would exclude many instances of transitivized unaccusative verbs since transitivization often (but not necessarily) involves the subject's volitionality (cf. section 3.1.3). The preselection of verbs has enabled us to avoid the various kinds of verbs which may otherwise complicate our analysis of data afterwards. Although this decision undoubtedly reduces the number of token sentences due to the exclusion of other legitimately unaccusative verbs and therefore may not be feasible when dealing with a small corpus, the fairly large size of the research corpus of the study has made it possible to adopt this rather rigorous methodology. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 4.2.2. Extraction of Token Sentences A computerized search program was developed to extract token sentences built on an unaccusative verb firom the research corpus.^^ This program was designed to extract both the preceding and following lines as well as the entire sentence in which an unaccusative verb was used. The additional lines were extracted so that the use of each verb intended by the learner could be specified when the token sentence alone was not enough to make an adequate interpretation. When excerpts alone were not helpful to determine the nature of the verb's usage, the original essay was checked as well. Such cases were rare but did occur, for example, when it was necessary to decide whether a particular instance of it was a referential pronoun or an expletive subject. In extracting token sentences firom the research corpus, only the unaccusative verbs (and their usage) which required only one NP as an argument were selected irrespective of existence of other constituents such as adverbs and prepositional phrases. Care was taken to exclude several types of verb forms and usage. Some of these are listed in (2). (2) a. verbs that require complements other than a sole NP e.g. appear (to be) happy, happen to be in the room, etc. b . idioms and metaphorical usage of verbs e.g. fall in love, fall ill, be dying to do something, etc. c. nonfinite verbs e.g. infinitives (with or without to), participle constructions, gerunds The program was developed by Joe Allen. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 d. phrases and sentences which are apparently quoted from some other source and therefore cannot be taken as a reflection of learners' competence e. indeterminable structures and unintelligible sentences The verbs in (2a) are used as raising verbs which require a prepositional complement The surface subject of these sentences is analyzed as derived from the subject position of the embedded clause. Since it is the postverbal clause, not the NP subject, that these raising verbs subcategorize, they were excluded from our data. Idiomatic and metaphorical use of unaccusative verbs in (2b) were also considered problematic. Because they are likely to be learned as phrasal chunks, they may resist any kind of reanalysis of the verbs in grammars of L2 English speakers. Another problem is that they tend to be used in an identical pattern repeatedly within a single essay. For instance, expressions such as I was fallen in love (cf. Zobl's example (8b) in section 3.1.1.3) was often noted in our research corpus. Including them in the tally of the data, however, was judged to pose a danger to skew the overall results. Consequently, both idioms and metaphors were excluded. Nonfinite verbs in (2c) were also screened out because it is extremely difficult to analyze their relation to an NP argument (see section 4.2.3). Theoretically speaking, when nonfinite verbs appear without an overt NP, the existence of PRO is commonly assumed. The position of PRO, however, is impossible to determine simply because it is not overt. For instance, when faced with a sentence such as (3), there is no way to decide whether PRO precedes or follows the present participle arriving. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ill (3) Arriving home, Karen immediately telephoned her firiend Gary. In theory, we can and in fact have to assume that PRO precedes the verb, leaving trace in postverbal position as: (4) PROi arriving ti home, Karen immediately telephoned her firiend Gary. However, this assumption is difficult to support in analyzing L2 data without some concrete evidence for it. Without such evidence, sentences such as (3) caimot be classified either as an NP-V or V-NP pattern. 4.2.3. Classification of Token Sentences A total of 941 token sentences were obtained in this way: Italian (240), Spanish (347), Japanese (304) and Korean (50). A detailed distribution of these sentences with respect to LI, IL structure, and verb is given in Appendix A. The token sentences, except those in the target-like NP-V word sequence, are also listed in Appendix B. In classifying the token sentences, two issues were considered to be important. One was to see L2 grammar in its own light in the spirit of Bley- Vroman (1983). As Cook (1994) notes, recent L2 research in the framework of generative grammar which sees normative speaker language as a testing ground of linguistic theory is often concerned with whether or not L2 grammar conforms to or differs from the target language in terms of parameter settings. Although I also have adopted the framework of generative grammar for the analysis of L2 English in this thesis, my ultimate goal here is to account for the way that unaccusative- inchoative verbs are acquired by normative speakers. The first step, therefore, is to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 describe the learner's language as closely as we can. Because a large corpus of actual learner language is an ideal data source in which to observe diverse L2 structures, care was taken that the classification should maximally reflect the actual IL structures as they were produced on the preselected verbs. The second important consideration was to list and count all the relevant data without overlaps. In order to achieve these two goals, three structural criteria in (5) were adopted. (5) a. the position of the NP argument vis-à-vis the verb (i.e. preverbal or postverbal) b. the type of verbal phrase (i.e. "active " or "passive") c. the existence and type of expletive subject (i.e. there, it or 0® ^ ) The measure based on (5) produces the 8 predictable word orders described in (6i- viii). In addition, two more classifications in (6 ix-x) were needed in order to classify the structures observed in the data. (6ix) is the transitive structure discussed in section 3.1.3. (6x) is a structure which Yip (1989, 1995) discusses with respect to the L2 English of native Chinese speakers. In all, therefore, the token sentences were classified into a total of 10 structural patterns.®^ ® 3 The lack of an overt expletive in the V-NP sequence does not necessarily mean that there is a null expletive (shown as 0 above) in the preverbal position. As we will see, however, our analysis strongly indicates that there is in fact a null expletive in this position for some L2 learners of English. ® ^ Some clarification is in order. In reality, all the above types can appear with additional elements such as tense-marking elements, verbal inflection appropriate to a particular number/person agreement as well as the aspectual have and/or negative word, modal auxiliary, and modifiers. However, these elements were considered generally irrelevant to the classification of token sentence types and consequently ignored in the table in Appendix. Secondly, in classifying examples as "passive" structures, the existence o f be in the veibal phrase was regarded as sufficient even if the verb was not properly inflected as long as it was clear that the missing inflection is not the "progressive" -ing (cf. Hubbard and Hix's (1988:93) characterization of this kind of error as the "intrusive be" reviewed in section 3.1.1.2). This, in fact, did not happen often, but the classification was considered justifiable since the appropriate use of inflectional morphology is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 (6) Surface Syntactic Patterns for Classification of Token Sentences Preverbal NP structures: i. NP-V ii. NP-be-Ven Postverbal NP without "passive" structures: iii. there-V-NP iv. it-V-NP V. 0-V-NP Postverbal NP with "passive" structures: vi. there-be-Ven-NP vii. it-be-Ven-NP viii. 0-be-Ven-NP Transitive structure: ix. NP1-V-NP2 Special: X. there-be-NP-V apparently a difficult area to master in SLA. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 (6 i) is an apparently target-like NP-V structure such as Many guests arrived at the party later in the evening. (6ii) is the "passivized" unaccusative structure discussed in section 3.1.1, for example. The strange event was occurred last May. (6iii-v) are postverbal NP structures with an active verbal phrase. They differ from each other in the types of preverbal element. For instance, if there is no overt preverbal NP that could function as a subject, we may get a sentence like Sometimes comes a good regular wave. This example may also appear with an overt expletive subject. So, Sometimes it comes a good regular wave or Sometimes there comes a good regular wave is also a logical possibility. (6vi-viii) are also postverbal NP structures but contain the passive verbal phrase. They were included purely as logical possibilities. 4.3 . Re s u l t s The overall results of the relationship between the structure and the LI are given in Table 4.1. Table 4.1 Overall Structural Distribution of Unaccusative Verbs Italian Spanish Korean Japanese Total i: NP-V 216 327 39 269 851 ii: NP-be-Ven 8 4 8 17 37 iii: there-V-NP 2 0 1 1 4 iv: it-V-NP 5 6 1 1 13 V: 0-V-NP 7 8 0 1 16 vi: there-be-Ven-NP 0 0 0 4 4 vii: it-be-Ven-NP 0 0 0 0 0 viii: 0-be-Ven-NP 0 0 0 0 0 ix: NPI-V-NP2 2 2 1 7 12 X: there-be-NP-V 0 0 0 4 4 Total 240 347 50 304 941 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 What is most striking about Table 4.1 is that the majority of token sentences appeared in the target-like NP-V word sequence of (i). 851 out of the total 941 tokens (i.e. 90.4%) were obtained in this surface order. The distribution of the remaining token sentences was far firom even. Among the non NP-V structures, the "passivized" unaccusative structure of NP-be-Ven in (ii) was the pattern most commonly used by all four LI groups combined, confirming the previous observations by researchers on the appearance of this structure in L2 English. The combination of the postverbal NP and the "passive" verbal phrase of be-Ven, on the other hand, was either extremely rare as in (vi) or nonexistent as in (vii-viii). The scarcity of the postverbal NP with the "passive" verbal phrase of (vi-viii) is striking particularly in contrast to the preverbal NP with the "passive" verbal phrase of (ii) and to the postverbal NP without the "passive" structure in (iii-v). As for the transitive structure of NP-V-NP in (ix), there were a few instances, but the number was not very large with only 12 out of 90 (i.e. 13.3%) of the non NP-V patterns appearing in this structure. There was also no example of "pseudotransitive" sentences as in Zobl's examples (23a-b) in section 3.1.3.4. A closer look reveals that among the postverbal NP structures without "passivization," the frequency of the there-V-NP pattern of (iii), in comparison to those of the it-V-NP of (iv) and the 0-V-NP of (v), was surprisingly low. Only 4 of the 33 postverbal NP stmctures without the "passivization" (12.1%) in our data were in this surface pattern. Since, as far as English unaccusative verbs are concerned, the structure of (iii) is the only potentially acceptable stmcture with an expletive subject in the target EngUsh grammar, it is quite interesting that this grammatical stmcture was outnumbered in the research corpus by the two Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 ungrammatical postverbal NP structures of (iv-v). Apparently, the learners preferred the non-target structures to the grammatical target structure. On the use of the postverbal NP structures without "passivization,"®^ a crosslinguistic difference was observed between the Italian and Spanish speakers on the one hand and the Korean and Japanese speakers on the other. This difference becomes particularly clear when the four target-like token sentences in the structure of (iii) are excluded from our data. That way, we can see the influence of the learners' Lis on their X L English more clearly.^ In Table 4.2, the two language groups, Italian and Spanish on the one hand and Korean and Japanese on the other, are compared. Table 4.2 Italian/Spanish versus Korean/Japanese Italian/Spanish Korean/Japanese Total ii: NP-be-Ven 12 25 37 iv/v; (it)-V-NP 26 3 29 Total 38 28 66 Table 4.2 clearly shows that while both Italian/Spanish and Korean/Japanese native speakers use the "passivized" unaccusative structure of NP- be-Ven, the postverbal NP structure with or without the expletive it is used predominantly by the Italian/Spanish speakers. In fact, Italian and Spanish use this structure more often than they do the "passivization" while Korean and Japanese Since the number of tokens is extremely small for the three postverbal NP structures with "passivization” in Table 4.1, (vi-viii), they are not included in the above analysis. ^ Because the sentences can be acquired based on the positive evidence available in the target language input, they do not necessarily reflect the learners' LI grammar. Despite often wrong number agreement, all four token sentences contain the correct semantic subclass of unaccusatives, i.e. verbs of "existence" and "appearance." As we will see later, with them excluded, the remaining data obtained on the structures of (iv- v) in Table 4 .1 become a useful window to investigate the grammatical system o f the learners’ Lis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 use the "passivized" unaccusative structure almost exclusively. The Chi-square value of the distribution pattern in Table 4.2 was found to be significant at p<.001. The results on the use of the 10 preselected unergative verbs in the same research corpus revealed that these verbs were used in very different manners in comparison to the 10 unaccusative verbs. First, virtually no "passivization" (i.e. 1 out of 640 token sentences) was found in the research corpus, showing virtually all the instances of "passivization" involved unaccusatives, not unergatives. Second, no postverbal NP structure with or without an overt expletive and with or without "passivization" was found in the use of the unergative verbs. 4 .4 . D is c u s s io n 4.4.1. "Passivization" The data obtained from the Longman Learners Corpus can shed light on the validity of different interpretations of the significance of the "passivized" unaccusative structure. As reviewed in section 3.2.2.3, there are five possible interpretations suggested in the literature. They are repeated here for convenience as (7). (7) a. Direct transfer from similar LI structures such as Italian "passato prossimo" and French "passé composé" (section 3.1.1.2.) b. Overgeneralization of passive and perfective adjective formation (section 3.1.1.2.) c. Lexical reanalysis of unaccusatives as transitives (section 3.1.1.4.) d. Identification of the lack of logical subject and the passive morphosyntax (section 3.1.1.3.) e . An overt marker of syntactic NP-movement (sections 3.1.1.3 and 3.1.1.4.) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184 I present discussions on each interpretation in (7) in the following subsections. In the end, the obtained data cast doubt on (7a-d) but support (7e) as the major factor for the "passivization" of unaccusative verbs in L2 English. 4.4.1.1. LI transfer As for the LI transfer account of (7a), this was a weak explanation in the first place since researchers have already reported that unaccusatives are "passivized" even by learners whose L is do not permit such a structure (sections 3.1.1.4 and 3.2.2.3). The results obtained in our study not only confirm these previous observations but also give further support to them. For a clear illustration, let us first put aside the tokens of the structures of (ix-x) in Table 4.1. (ix) is a lexical reanalysis of unaccusative verbs and (x), according to Yip (1989, 1995), is not limited to the use of unaccusative verbs. With these two structures set aside, the ratio of the "passivized" unaccusative structure of (ii) to the other non NP-V structures (iii-viii) in Table 4.1 is higher in the token sentences obtained from Korean and Japanese speakers than in those from Italian and Spanish speakers. Specifically, the ratio is 36.4% in Italian, 22.2% in Spanish, 80% in Korean and 70.8% in Japanese productions. In other words, the "passivized" unaccusatives appear not only in the interlanguage English of Korean and Japanese speakers but also constitute the maforitv of the non NP-V sentences produced by these LI groups. This fact poses a challenge to the LI transfer account of (7a). If the LI transfer account were the only possibility, Korean and Japanese speakers would not be expected to produce "passive" unaccusatives at all. Clearly this is not the case. Therefore, the strong version of the LI transfer account is untenable. So, let us Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185 assume that there is another source for this non-target structiue that is accessible to all the L2 English speakers, including Korean and Japanese learners. If so, the question is whether Italian speakers have access to the presumed LI transfer mechanism, that is, "passato prossimo," in addition to this more universal mechanism. This is a weaker interpretation of the LI transfer account. This second scenario, however, is also difficult to support. If Italian speakers have two routes open, so to speak, to the "passivization," they would be expected to produce as many, if not more, "passive" sentences as Korean and Japanese speakers who have, after all, only one avenue to the structure. This expectation is not borne out by the obtained results. In short, as far as our corpus data show, it is unlikely that the "passive" unaccusative stmcture is transferred to L2 English from a similar LI structure such as Italian "passato prossimo." 4.4.1.2. passive and perfective adjective The account of (7b) based on a possible analogy to passive and perfective adjectives in the target English turns out to be mostly untenable for a semantic reason. As discussed in section 3.1.1.2, the passive and perfective adjectives do not appear with a prepositional by phrase. In this respect, at first, the results obtained from the LLC appear to support (7b) because almost all the tokens of the extracted "passive" unaccusatives were without a by phrase. The crucial difference between the two, however, is that the data from the LLC all lack the stative meaning, which is a semantic characteristic of adjectives, including those lexically derived ones such as passive and perfective adjectives (cf. section 3.2.2.S). In the data, the lack of stative meaning was evident in most token sentences even without a larger context. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 186 but with additional linguistic cues such as time, frequency or manner adverbials it becomes very clean (8) a. *... they were happened a few days ago .... {7275in2.iit}®^ c. *. . . terrorism is happened very often .... {0871uil.kox} b. *... suddenly pale face was appeared out of the window .... {6896inI.kos} d. *UsuaIly ash is fall in ground about one or two years. {5653ui2.jap} (8a-d) are examples obtained from the research corpus. In these examples, it is clear that the intended meaning is not a stative one such as the result of an event. Rather, it is the events themselves that were expressed by the sentences in (8). 4.4.1.3. lexical transitivization The data from the LLC also present some evidence against the lexical reanalysis account of (7c) as a major cause of the "passive" unaccusative structure. First, if there is a creative non-target transitive verb behind "passivization," there should be a plausible passive sense in every instance of a "passivized" unaccusative sentence. Our tokens of "passivized" unaccusatives, however, were carefully collected, excluding those with the clearly passive meaning. (Sentences with a clear passive sense were classified as (ix) in Table 4.1 instead.) This point is particularly evident, for instance, when there is no plausible causer as in the following examples.^ The coding numbers refer to original essays in the LLC from which token sentences were extracted. The passive semantics is also extremely hard to find in the "passivization” of existential Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 187 (9) a. *I was nearly arrived to my office .... {4899prl.iit} b. *... the word, "the role of women," is appeared just several years ago. {0939prl.kox} c. *... to find out what would be happened in the next stories .... {6887inl.kos} Contextually speaking, it is very hard to assume the existence of a clear causer in the situations described in (9). Especially, the events in (9b-c) are beyond anyone's control and therefore are unlikely to be presented as ones caused by some specific entity. Second, if "passive " unaccusatives are in fact based on lexically derived transitives, we should find many active sentences in which the unaccusatives appear as transitives. Such transitive sentences do show up in our data as indicated by (ix) in Table 4.1, but their number is not very large in comparison to the supposedly passivized versions. This observation is not only true in general but also holds for the data obtained from each LI group. That is, the overall ratio of transitivized to "passivized" sentences is 12 to 37, with the LI-based breakdown being 2 to 8 in Italian, 2 to 4 in Spanish, 1 to 8 in Korean, and 7 to 17 in Japanese. The relatively low frequency of the surface transitive usage is rather surprising if there are in fact lexically reanalyzed transitive verbs behind instances of "passivized" unaccusatives. Finally, no transitive use of the verb exist was found in our corpus data. This result may be expected because, after all, this verb without any possible sense unaccusatives: (i) *Thls problem is existed for many years. (Hubbard 1994:55) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 188 of change, either in location or in state, may be conceptually incompatible with transitivization. Nonetheless, sentences such as *This problem is existed for many years (section 3.1.1.2) are sometimes observed in L2 English. This small fact, therefore, may cast doubt on the transitivization account (at least as far as pure "existence" verbs such as exist are concerned). However, the small number of transitivized unaccusatives in our data do not warrant any strong conclusion concerning the class of verbs which may not be commonly transitivized but still may appear in their "passive" form.®’ Clearly, this issue deserves further investigation. In sum, although there is some evidence that L2 learners in fact use unaccusatives as transitives, the corpus data indicate that the majority of "passivized" unaccusatives cannot be attributed to the lexical reanalysis of unaccusatives as transitives. Therefore, "passivization" needs to be accounted for independently of the existence of transitivized unaccusatives. This conclusion also matches Hirakawa's (1995) finding that the learners who allow incorrect "passive" unaccusatives do not necessarily allow their transitive counterparts (section 3.1.3.3). On the other hand, 3 out of the 12 transitivized unaccusatives were rise used as causatives. In these cases, it is not improbable to regard them as simple lexical errors resulting from the confusion of rise with the target raise. If so, the instances o f truly "transitivized" unaccusatives may be a bit smaller than 12. This, of course, further weakens the argument for transitivization as a major cause of "passivization". Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 189 4.4.1.4. identification of the "passive" morphosyntax with the lack of logical subject As for (7d), there is clear evidence that the passive morphosyntax of English is not merely identified with the lack of logical subject in learners’ grammar. As discussed in sections 3.1.1.3 and 3.2.2.3, the expletive-be-Ven-NP structures of (vi-viii) in Table 4.1 serve to test this interpretation of the "passive" structure suggested by Zobl. If Zobl's account is correct, we would expect to find sentences which conform to these surface syntactic patterns. However, these structures rarely showed up in the research corpus as indicated in (vi-viii) in Table 4.1. Aside firom four tokens produced by Japanese learners, the expletive-be-Ven-NP structures are simply nonexistent. Moreover, even the native speakers of Italian, which allows a structure similar to this, did not use unaccusative verbs in these patterns.Further evidence against (7d) is the 33 tokens found in the expletive-V-NP structures of (iii- v) in Table 4.1. If the lack of logical subject is really associated with the passive verbal phrase, these examples should not show up in L2 English. In sum, the interpretation that the "passive " morphosyntax is canonically identified with the lack of logical subject gains no support from the LLC data. 4.4.1.5. marking of syntactic NP-movement The rarity of the expletive-be-Ven-NP structures in our corpus has just been discussed. The other side of this observation is that almost every time the "passive" verbal phrase is produced, the NP argument appears preverbally. In fact, in 37 out of the 41 instances (i.e. 90%) where we find the passive verbal phrase in our data. This fact also argues that a direct transfer is not at work when it comes to the structure o f Table 4.1, (vi-viii). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 190 the NP arguments are in subject position. This strongly suggests that the most plausible account of "passivization" is (7e), that is, to see its grammatical function in interlanguage English as an overt marker of syntactic NP-movement. Notice that among the structures with English verbs that manifest unaccusativity below, the passive structme in (10b) is the only one that overtly indicates the movement of NP into the surface subject position by the combination of an auxiliary verb and a past participial form of a main verb. ( 10) a. All the students respect the old principal. b. The old principal! is respected t! by all the students. (Passive Verb) (11) a. There arrived three boys at the station. b. Three boysi arrived ti at the station. (Unaccusative Verb) (12) a. It seemed that the boy was very popular. b. The boy! seemed t! to be very popular. (Raising Verb) (13) a. It is likelv that John will oppose the plan. b. John! is likelv t! to oppose the plan. (Raising Adjective) Clearly, this morphosyntactic characteristic of the English passive stmcture is generalized to the derivation of unaccusative sentences by nonnative speakers. This interpretation is consistent with most of the "passivized" sentences produced by the four LI groups in the research corpus. Why, then, are L2 learners compelled to signal the movement of a postverbally generated NP argument into subject position by this kind of overgeneralization? An answer to this question may be found in Kellerman's Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 191 ( 1983:122) "Reasonable Entity Principle (REP). " Kellerman characterizes REP as learners’ tendency to "treat [an] L2 as if it were a reasonable entity (unless [they] have evidence to the contrary)" and says that "in the absence of specific knowledge about the L2, learners will strive to maximalize the systematic, the explicit, and the "logical" in their X L ." This description matches the way "passivized" unaccusatives are produced in L2 English. Learners appear to simply treat English as something a little more systematic, explicit and logical than it actually is. Are the learners at fault, then, because of their belief in the systematicity, explicitness and logic of the English language? The answer is clearly negative. Although the movement of an argument to surface subject position is not overtly marked in English except in the passive construction, a relatively consistent morphosyntactic marking of this sort can be found in some languages. For instance, Rosen (1984:58-59) notes that the morpheme -ya in Sanskrit and "middle voice" in Albanian serve as a marker of 2-1 Advancement (i.e. Relational Grammar terminology that corresponds to object to subject movement). Choctaw is another language discussed by Rosen which has a mechanism for this kind of surface marking. In Choctaw, case-marking of a free-standing nominal is sensitive to its highest Grammatical Relation while its lowest Grammatical Relation determines the verb agreement affix in the derivation of a sentence. Therefore, where there is a discrepancy between them, it indicates object to subject movement (see Rosen 1984:58, (64)). When these phenomena are considered, the "passive" unaccusative structure in L2 English appears to be nothing uimatural. It is a type of morphosyntactic marking already found in other natural languages. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 192 4.4.2. Postverbal NP Structures Table 4.1 shows that the majority of postverbal NP structures, i.e. (iii-viii) obtained from the research corpus (33 out of 37, i.e. 89.2%) did not contain the passive verbal phrase. One finding that deserves our attention is that the there-V-NP structure of (iii), the potentially grammatical postverbal NP structure, was actually very rare. Only 4 out of 37 (i.e. 10.8%) were found in this structure. Its rarity, especially in comparison with the relatively frequent appearance of the non-target structures of (iv-v), is rather surprising. Clearly, many learners have not incorporated this grammatical structure of the target English in their developing grammar although the structure of (iii) itself is presumably available in the input that they are exposed to (albeit in a very limited way). Although the relatively firequent use of the other two postverbal NP structures of (iv-v) gives some support to the observations made by Zobl (1989) and Rutherford (1989) (section 3.1.4), a closer look reveals that the obtained data from the LLC do not quite match some aspects of their observations and assumptions. Recall that Zobl's examples of postverbal NP structure were produced by native Japanese speakers (section 3.1.4.1). Table 4.2, however, clearly shows that the Korean and Japanese speakers used the postverbal NP structures far less often than the "passive " structure. In fact, most of the tokens of the postverbal NP structures were produced by the Italian and Spanish speakers, who used both "passive" and postverbal NP structures but used the latter twice as often as the former. Therefore, the postverbal NP structures of (iv-v) in Table 4.1 may be seen even as a characteristic of the L2 English of Italian and Spanish native speakers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 193 This result, at first, may appear to support Rutherford's contention that the trait of pragmatic word order (PWO), unlike that of grammatical word order (GWO), is transferable from LI to L2 (see section 3.1.4). Both Italian and Spanish exhibit fairly fiee surface word order and may be characterized as PWO languages. If so, their PWO trait may be claimed to be responsible for the pattern shown in Table 4.2, (iv-v). This account, however, does not quite work with our data. If the PWO trait of Italian and Spanish is actually transferred to L2 English, we should expect to find non-canonical word order in sentences other than those based on unaccusatives. However, the result of the analysis on the use of the 10 preselected unergative verbs in the same research corpus produced absolutely no token of the postverbal subject structure. The observed contrast between unaccusatives and unergatives concerning their compatibility with a postverbal NP structure in L2 English is likely to be due not to a pragmatic reason but to a grammatical factor. Specifically, I argue that the reason why we find postverbal NPs only with unaccusative verbs is that they are generated in this position at D-Structure. Unlike the "passive" structure, the NPs clearly have remained in situ at S-Structure instead of moving to the surface subject position. This explains why we did not find postverbal NPs for the 10 unergative verbs, whose arguments are not base-generated in object position. In other words, the assumption that the correct A-Structure and D-Structure representations for the unaccusative and unergative classes seem to have already been acquired by the L2 English learners who produce "passive " and/or postverbal NP structures, and furthermore, whatever linguistic operation is responsible for the postverbal subject in unergative sentences in Italian and Spanish does not appear to be relevant in L2 English by the speakers of these languages. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 194 In order to adopt this account, however, we need to answer one crucial question, that is, why is it that native speakers of Korean and Japanese rarely produce postverbal NP structures? More specifically, the question is why an NP argument base-generated in postverbal position does not remain in situ at S- Structure in the L2 English of Korean and Japanese speakers as is assumed in the case of the L2 English of Italian and Spanish native speakers. As a key to answering this question, I would like to suggest the existence of "null expletive subjects" in Italian and Spanish and the lack of such elements in Korean and Japanese. Before I elaborate on my analysis, however, theoretical justification for the assumption of expletive subjects in some languages in relation to Nominative Case assignment has to be reviewed first. 4.4.2.1. arguments for and against null expletives Chomsky's (1986b) analysis of overt expletives in English provides a foundation for subsequent hypothesizing of null expletives in some languages. Consider the sentence in (14). (14) a. There is/*are a man in the room. b. There are/*is two men in the room.^^ Chomsky accounts for the mechanism of Case assignment to the postverbal NPs in ( 14a-b) in terms of CHAIN and "Case transfer." According to him, the expletive there in the surface subject position and its associate NP a man or two men in the postverbal theta position form a CHAIN, which may be represented as (15). I ignore here the colloquial phrase there’ s which is used a presentational expression with both singular and plural noun phrases. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 195 (15) [IP [NP expletivei] [r I [vp V NPi]]] _ _ _ _ _ _ _ I Nominative Case transfer The Nominative Case assigned to the expletive by Infl is "transferred" to the associate NP through the CHAIN. Burzio (1986) argues that this mechanism of Case transfer is also at work in some constructions with unaccusative verbs such as (16). (16) a. There arrives/*arrive a man at the station. b . There arrive/*arrives two men at the station. As in (14a-b), the verbs agree with the postverbal NPs in (16a-b). If the requirement of the agreement is not properly satisfied, the sentences turn out to be ungrammatical. In this respect, the English sentences in (16) contrast with their French counterparts in (17a-b). (17) a. n arrive/*arrivent un homme à la gare. it arrive(SG)/arrive(PL) one man at the station b . H arrive/*arrivent deux hommes à la gare, it arrive(SG)/arrive(PL) two men at the station (17a-b) do not show agreement between the verb and the postverbal NP. Whether the postverbal NP is singular or plural, the verb remains singular. Davis (1984) accounts for this contrast between English and French as a reflection of the difference in the feature properties of the two expletives, there and il. That is, unlike English there^ whose agreement features are underspecified and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 196 determined by the associate NP which forms a CHAIN with it, French il is inherently specified as third person masculine singular. Consequently, the verb always agrees with the features of the expletive it irrespective of the features of the associate NP in the postverbal position.^z In similar unaccusative sentences in Italian in (18a-b), however, no overt expletives show up. (18) a. Arriva/*arrivano un uomo alia stazione. arrive(SG)/arrive(PL) one man at the station b . Arrivano/*arriva due uomini aUa stazione. arrive(SG)/arrive(PL) two men at the station Despite the lack of an overt expletive, (18a-b) clearly show the subject-verb agreement. Burzio argues that (18a-b) are exactly like their English counterparts in (16a-b) except that the expletive, which is phonetically realized as there in English, has no phonetic value in Italian. This phonetically null expletive is underspecified with respect to the feature properties as in the case of there and agrees with the features of the postverbal NP.73 This null expletive is argued to form a CHAIN In the minimalist framework of syntactic theory, the contrast between English there and French il is reformulated in terms of feature-checking mechanisms (e.g. Chomsky 1995, Groat 1995). Groat (1995) argues that in the case of rtere-type expletives, they have Case features to be checked overtly but do not have agreement features. This induces NP movement after Spell-Out so that the agreement features of the moved NP can be checked against Agr(s) at LF. On the other hand, (7-type expletives, including English it, have Case features and phi-features, both of which are checked overtly. Therefore, NP cannot move without violating Greed. Chomsky (1995) presents a similar analysis though in his framework it is not the associate NP, but its features, that moves at LF. In either account, the contrast between the two types of expletives with respect to binding possibility can be accounted fon (i) There arrived three men (last night) without identifying themselves. (ii) *It est entré trois hommes sans s'annoncer. (Chomsky 1995) The following examples from Chomsky (1995) also support the analysis to regard the Italian null expletive as a rAerg-type rather than an (7-type. Compare them to (i) in the previous Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 197 with the postverbal NP and to transfer Nominative Case assigned by Infl to the associate NP as shown in (19). (19) [n> [NP null expletive^] [r I [vp V NPi]]] ___________________ Nominative Case transfer In this way, Burzio can account for Italian sentences with an unaccusative verb and a postverbal NP by hypothesizing a phonetically null expletive subject which functions exactly like the English expletive there As for the Nominative Case assignment in Japanese and Korean, there are two contrastive views. For convenience's sake, I wUl call them here Infl- independent and Infl-dependent theories of Nominative Case assignment. What is important in the context of this thesis is that their contrast also leads us to conflicting assumptions concerning the existence of null expletives in these languages. First, let us review Infl-independent theories. Kuroda (1988) argues that Case assignment in some languages is an instance of feature-sharing (i.e. cospecification) between a base category and a maximal projection of N that it governs. According to him, this feature-sharing is parameterized across languages. footnote. (i) Sono entrati tre uomini senza identificarsi. Three men entered without identifying themselves.’ (ii) Ne sono entrati tre t senza dire una parola. O f them three entered without saying anything.' In the minimalist framework (Chomsky 1991), what is important for the question we are concerned with in this thesis is that in languages such as English, French, and Italian, Tense has strong N-features which have to be checked before Spell-Out. This necessitates a sentence having an (either overt or null) expletive subject with relevant features in conformity with Extended Projection Principle when the associate NP remains in the object position (Marantz 1995:368). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 198 For instance, in languages such as English the agreement is forced and in languages such as Japanese it is not. Kuroda argues that unlike English, in which the subject NP moves from Spec of VP to Spec of IP to satisfy the agreement requirement and receive Nominative Case in the X-bar-theoretic Spec-Head configuration, the subject NP in Japanese does not need to move to Spec of IP. This is because the agreement requirement does not exist in Japanese and Spec of VP as well as Spec of IP is a case position in this language. (Kuroda ( 1992:chap. 6) assumes that Case assignment in Japanese is primarily done by cyclic application of a linear order Case-marking rule.) Notice that in Kuroda's theory the lack of an overt expletive subject in Japanese is a natural consequence of the fundamental characteristics of this language. Being a nonforced-agreement language, Japanese does not need to fill the Spec of IP with an appropriately marked ("agreed with") maximal category as in English. In a series of works (Fukui and Speasl986, Fukui 1987, Fukui 1995, Fukui and Nishigauchi 1992), Fukui has developed a view on the agreement phenomenon similar to Kuroda's (1988). In his theory, Fukui also sees the agreement in terms of Spec-Head agreement of a functional category and the Specifier as a "closing " element of projection of a functional category. Fukui considers functional categories in Japanese as either nonexistent or "very defective" in the sense that they lack agreement-inducing properties. Either way, Japanese, unlike English, does not observe the "principle of agreement" which mandates that agreement relation must be satisfied at S-stmcture, Therefore, Japanese not only lacks the movement of NP to Spec of IP but also has no overt expletive. As for Korean Nominative Case assignment, a view similar to Kuroda's and Fukui's has been presented in the literature. Yang Soon Kim (1988), for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 199 instance, suggests that although Infl in Korean contains Tense and other inflectional categories, Infl is not a Case-assigner. Therefore, unlike in English, the subject in Korean stays in VP and receives Case from V. In her theory. Spec of IP in Korean is considered to be a Topic position. Young-Suk Lee (1990) argues that Infl does not exist as an independent category in Korean. This conclusion is based on observations of independence of Nominative Case assignment from Infl and "multiple subject construction." According to Young-Suk Lee, the Korean clause has a subject as a consequence of a requirement that the clause have a subject- predicate structure. Therefore, the Nominative Case morpheme ka in Korean is not assigned by Infl but considered as the marker for the subject of a predicate. According to these views of Yang Soon Kim (1988) and Young-Suk Lee (1990), Korean lacks the Case-theoretic NP movement from Spec of VP to Spec of IP. Essentially the same idea is also expressed in Suh (1990) and Young-Joo Kim (1990). These Infl-independent views of Nominative Case assignment also account for the lack of overt expletives in Korean (e.g. Chungmin Lee 1989:470). Since an expletive subject is required when an NP has not moved to Spec of Infl in a language that requires Spec-Head agreement, if Nominative Case assignment in Korean is not contingent on this position, there is no reason why the language should contain an expletive in its lexicon. However, there are proposals in the literature that Nominative Case assignment to subject NP in Japanese and Korean is Infl-dependent just as in English, French, Italian and so forth. As for Japanese, for example, Takezawa (1987) and Terada (1987) independendy argue that the Nominative Case is assigned under government of Infl. In her later work, Terada (1990) specifically claims that Nominative Case assigmnent is done by the movement of NP into Spec of IP. In Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 200 her theory, this inflectional element in Infl that Case-marks the NP in this position is considered to be Tense. Similarly, Ueda (1993) suggests that Nominative Case in Japanese is an abstract Case assigned to the subject by the [+ tense] feature of Tnfl under government at S-stmcture. Miyagawa (1990) also argues that Japanese Nominative Case is licensed by a functional category in Infl. Unlike Terada (1990) and Ueda (1993), however, in Miyagawa's theory the inflectional element responsible for Nominative Case assignment is considered to be Aspect. As for Korean, Han (1987) and Ahn and Yoon (1989) contend that the full- fledged Infl (i.e. Tense and Agr) exists in this language. These studies claim that in Korean the subject-honorific marker si is an instance of subject agreement and it is this functional category, a type of Agr(S), that assigns Nominative Case to the subject NP.^^ Now let us consider the implications of these two opposing views concerning Nominative Case assignment in Japanese and Korean on the way in which the sole NP argument of an unaccusative verb is assigned Case. In Infl- independent approaches, the NP argument within V can stay in situ and gets Nominative Case by an alternative mechanism such as Kuroda's "linear order Case- marking." According to these theories, the nonexistence of overt expletives in these languages is a natural consequence of the "weak" or "defective" Infl which does not require Spec-Head agreement (as in Kuroda 1988, Yang Soon Kim 1988) or of the complete lack of such a functional category (as in Speas and Fukui 1986, Fukui 1987). Notice, however, that the rationale behind these theories also argues Young-Suk Lee (1990) specifically argues against an analysis which regards the honorific as a Nominative Case assigner. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 201 strongly against any assumption for phonetically null expletives in these languages. That is, the lack of Spec-Head agreement requirement within IP makes null expletives as well as overt expletives theoretically superfluous. Proponents of Infl-dependent approaches, on the other hand, consider Nominative Case assignment as a reflection of Spec-Head feature cospecification of the Infl (or some functional category subsumed under Infl such as Agr, Tense, Aspect, etc.). When the NP argument of an unaccusative verb stays as the sister of V at S-structure, an expletive like English there and French il has to be assumed to exist in the S-structure subject position to satisfy the agreement requirement. Moreover, since neither Japanese nor Korean has an overt expletive, if these languages have expletives, it has to be assumed that they are phonetically null. There seem to be two ways to avoid the assumption of null expletives in an Infl-dependent theory and still maintain the basic tenet of an Infl-based Nominative Case assigmnent. One is to assume that Infl lowers to V so that the NP can be Case-assigned in situ. This idea is in fact suggested by Takezawa (1987) for the Nominative go-marking of an NP in the object position of a stative verb. The lowering of Infl, however, creates a problem, because it leaves trace in a position that is not c-commanded by its antecedent (Haegeman 1994:601). The other way to avoid the stipulation of a null expletive is to argue that the NP movement out of VP into the Spec position of a functional category is obligatory in these languages as Terada (1990) suggests for Japanese. If this is true, it clearly renders the stipulation of a null expletive unnecessary because the requirement for Spec-Head agreement can always be satisfied by the NP raised into surface subject position. So, the question is whether or not the sole NP argument of an unaccusative- inchoative verb always moves to the surface subject position in Japanese and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 202 Korean. Some evidence that this is not always the case has already been reviewed in Chapter 2. For Japanese, for example, consider phenomena such as Case Marker Drop (section 2.4.5), "floating" numeral quantifier (section 2.4.6), interpretive tendency for the quantifier takusan (section 2.4.7), indirect passive (section 2.4.10) and passive causative (section 2.4.11). In these phenomena, the difference between unergatives and unaccusadve-inchoatives are clearly accounted for by assuming that the sole NP argument of the latter class generally stays in base-generated position. As for Korean, I do not have such clear evidence as in the case of Japanese. However, Case Marker Drop firom a numeral quantifier may be seen as evidence that when the NP does not float out to a higher position, it simply stays in object position and receives Nominative Case in situ. As a whole, in Japanese and Korean, there is no convincing argument that the NP argument of an unaccusative-inchoative verb has to be in Spec of IP at S-Structure but there is some evidence that it can actually stay in situ. In sum, the available data effectively eliminates, at least in Japanese and possibly in Korean, the possibility of proposing Infl-dependent theories of Nominative Case assignment without simultaneously assuming phonetically null expletives in these languages. In other words, Infl-independent and Infl-dependent theories of Nominative Case assignment lead to conflicting views concerning the existence of null expletive subjects in these two languages. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 203 4.4.2.2. on the psychological reality of null expletives With the above theoretical discussions in mind, let us now consider the data obtained from the research corpus. As noted before, there is a clear (p<.001) difference between the Italian/Spanish group and the Korean/Japanese group in the way they used some non-target surface structures. Namely, while the "passivized" structure of (ii) in Table 4.1 was used by both LI groups, the postverbal NP structures of (iv-v) were almost exclusively used by Italian and Spanish speakers. Where does this striking contrast originate? Let us first agree that it is highly unlikely that the two groups have been exposed to totally different kinds of target language input. Especially, considering the fairly large size of the research corpus, it is simply unimaginable that the two groups of learners systematically came across qualitatively different input. Even if we assume, as a remote possibility, that the Romance language speakers and the Korean and Japanese speakers were in fact exposed to significantly different kinds of L2 input, we still cannot account for the Italian and Spanish speaker's use of the postverbal structures of (iv-v) in Table 4.2 since there is no grammatical English construction that corresponds to them. Consequently, the crosslinguistic contrast in question has to be investigated in terms of the properties of the speakers' Lis rather than of the type of English input they encounter in L2 contexts. Moreover, because both (iv) and (v) are unacceptable and presumably nonexistent in the input, and because the Italian/Spanish group utilized them whereas the Korean/Japanese group rarely did so, we should suspect that whatever LI property may be crucial for the production of (iv-v) in Table 4.2 exists in Italian and Spanish, but not in Korean and Japanese. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 204 This LI property, I argue, is a null expletive whose theoretical justifications have been reviewed above. Consider the examples in (20). (20) a. * . . . like a mirage appeared the large expanse of the sea.... {7222in2.iit} b. *There is no doubt that does exist a big difference between .... {0378adl.üt} c. *One day happened a revolution. {2408ui 1 .lit} d. * During my schooldays happened so many things that I can't talk about all them. {2523ad 1 .iit} e. *Once happened something no very good for her. {7240in2.iit} f. *... a wonderful blue sky where still appeared a round white spot.... {5127uil.sec} g. *... because in our century have appeared the car and the plane .... {6737uil.ssp} h. *In every country exist criminals. {6738uil.ssp} When Italian and Spanish speakers produce these sentences, it may appear that they are violating the condition that finite English clauses require a subject at S- Structure. However, this impression is probably superficial. If the existence of a null expletive subject is assumed in preverbal position, the requirement for a surface subject is fulfilled by this expletive. In this view, it can be argued that as far as the Italian and Spanish learners who have produced the examples such as (20a-h) are concerned, there is a preverbal subject which meets the structural requirement of English.^^ Recall that a simple rightward movement analysis does not work here. If the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 205 This interpretation gets support firom the examples in (21) which at first may appear to be unrelated cases. (21) a. *... it existed a lot of restrictions .... {1559inl.iit} b. *. .. it happened a tragic event.... {7242in2.iit} c. *... it arrived the day of his departure .... {4407inl.ssp} d. *. .. it will happen something exciting .... {3043inl.ssp} e. *Now that I have left school, and I am at University, it happens something funny . . . . {4408inl.ssp} The sentence pattern in (21) does not exist in English either but appeared predominantly in the interlanguage English of Italian and Spanish speakers in the research corpus. We can regard these token sentences as cases in which null expletives transferred firom the speakers' Lis are phonetically realized in their L2 English. Though a little more complicated structurally, the following examples in (22) also support this possibility. (22) a. *I'm writing to you in order to make you aware of what it has happened to me in your hotel. {2163in4.iit} b. *I would be very glad if we could have a meeting to speak about the school times and what it's happened after it. {2390ui 1 iit} c. *... asking me for what it happened to me .... {7546pil.ssp} d. *. .. talking about what would it happen with our relationship .... {4435inl.ssp} postverbal NPs have been moved to the current position from a preverbal position, we should expect to find similar sentences built on unergative verbs as well. As stated in section 4.3, however, there was absolutely dû case of V-NP order involving unergative verbs in the research corpus. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 206 In each of (22a-d), the argument what of an unaccusative verb happen appears preverbally. However, it cannot assume the ordinary subject position because the position is already occupied by the expletive it. Sentences such as (20)-(22) were very rarely produced by the Korean and Japanese speakers in the research corpus. The question is why. The answer I would like to suggest is that null expletives, which are potentially accessible to Italian and Spanish speakers, are not available to speakers of Korean and Japanese. Let us assume that Korean and Japanese speakers, too, are aware that finite English clauses have to have an S-Structure subject, that is, a Nominative Case-marked maximal category in preverbal position. If these learners have correctly base-generated the NP argument of an unaccusative verb in object position, how can they meet this structural requirement? They can move the NP argument to subject position (and may wrongly mark the movement with the passive morphosyntax), which they clearly do. So do the Romance language speakers. What Korean and Japanese speakers do not do is to use an expletive it or leave the surface subject position phonetically blank. This clear contrast found in the research corpus is most elegantly accounted for if we assume that there are no expletives, null or overt, either in Korean or in Japanese. Therefore, unlike native speakers of Italian and Spanish, the only way for Korean and Japanese speakers to satisfy the structural requirement is to move the NP into the surface subject position. If in fact there is no null expletive in either Korean or Japanese, as I suspect, there is no possibility of transfer. What does not exist in the first place cannot be transferred.^^ Notice that it cannot be argued that there are in fact psychologically real null expletives Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 207 If this analysis of the data in Table 4.2 is on the right track, it poses a serious empirical challenge to Infl-dependent, agreement-based accounts of Nominative Case assignment proposed for Korean and Japanese. This is because, although (to my knowledge) no proponents of Infl-dependent theories have argued for a null expletive either for Korean or Japanese, the inclusion of such a lexical element in their theories appears inevitable if they are to be consistent. Therefore, tabulations such as Table 4.2 which cast doubt over the psychological reality of null expletives in these languages may be taken as evidence against Infl-dependent theories of Nominative Case assignment in Korean and Japanese. Before closing this section, let us take up another question concerning the nature of postverbal NP structures in L2 English. The question is about the choice of overt expletives made by the Romance language speakers who seem to transfer the null expletives from their first languages. They clearly prefer it to there although, as mentioned before, only the latter is the grammatical choice in the relevant structure of target English. Why don't they choose there over it, especially since in both Italian and Spanish, the null expletives in question are likely to be counterparts of English therel As English there, for instance, they are clearly underspecified with respect to the phi-features of person and number and the postverbal NP argument of an unaccusative verb agrees with the verb in Italian and Spanish. If this characterization of Italian and Spanish null expletives is correct, then, the phonetically salient expletive preferred in IL English should be there rather than it. That is, if Italian and Spanish learners come to realize that English does not allow null subjects as in (20) and phonetically realize the expletives transferred from in Korean and Japanese, but for some reason they are not transferred. Such an explanation is quite ad hoc unless it is accompanied by a reason why Korean and Japanese speakers are hesitant to transfer a null expletive while Italian and Spanish speakers are not. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 0 8 their Lis, they should choose there, not it. Why is this not the case in the actual L2 data? I can think of two possible answers to this question. One is the possibility that the acquisition of the expletive subject there in target English is limited to the structure of the there construction with the verb be in learners' grammar. That is, there is a possibility that despite a common theoretical analysis such as (15) and ( 19), the "there" construction with be is learned by L2 learners as a separate structural pattern for presentational/existential sentences. In other words, L2 learners may learn this structure not as (23a) but simply as (23b). (23) a. [ip [N P expletive^] [r I [vp V NP^]]] ___________________ Nominative Case b. [s There + be NP] Since in actual input, learners are likely to encounter the "there" sentences with be far more frequently than those with unaccusative verbs, the chance that this expletive gets associated with the prefabricated pattern of (23b) appears to be very high.’* The second and maybe more plausible account for the overwhelming preference of it over there by the Italian and Spanish speakers may be that it is more NP-like than there in target English. Unlike there, which has both pronominal and adverbial use, it always appears as a pronoun. Whether it stands for a that-clause or a nominal to-infinitive clause, it always takes a subject or object position, which is normally occupied by an NP. If L2 learners see the need to have an NP in ’* This possibility appears to me quite real, considering the way I myself learned this structure as a nonnative speaker of English. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 209 subject position, then it is natural that they choose it rather than there due to the former's overwhelming NP-likeness. In other words, for the Italian and Spanish learners who produced sentences such as (21) and (22), the need to have an overt NP appears to be more immediate than whether or not the expletive's features are underspecified as the null expletives in their mother tongues. 4.4.3. NP-V Structure One last major issue which we need to address is the large number of token sentences in the target-like NP-V word order found in the research corpus. In fact, the majority of L2 sentences built on unaccusative verbs (i.e. 851 out of 941 tokens, approximately 90.4% of our data) appear in this canonical word order of English. Given such a high percentage, we come to the realization that the question concerning non-target structures with unaccusative verbs such as "passivization, " postverbal NP structure, "transitivization, " etc. is in fact a two-fold one: viz. (i) Why are they not produced more often?" and (ii) "Why are they produced at all? " So far in this chapter, we have been concerned solely with the latter question, postponing the former. It is time to take up this challenging question 4.4.3.1. previous accounts In the literature, there have been a few attempts to characterize the use of the target like NP-V word order built on unaccusative-inchoative verbs. Recall that Zobl ( 1989) suggested a lexical NP-movement analysis for the NP-V word sequence found in his data (section 3.2.3). According to him, L2 learners lexically externalize the internal argument in the A-Structure of an unaccusative-inchoative verb before it is syntactically projected to D-Structure. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 210 Hîrakawa (1995) found in her study that her subjects were accurate in accepting unaccusatives in the NP-V word order. Contrasting this result, for instance, with their rather poor performance on middle verbs, she has concluded that syntactic movement of the NP argument of unaccusative verbs is acquired relatively early in the development of L2 English. In other words, Hirakawa assumes that as far as the lexical and syntactic representations of unaccusative verbs are concerned, the learners' grammar approximates the native grammar relatively early. Rosen (1991) goes even further, arguing that the acquisition of syntactic movement with respect to the argument of the unaccusative-inchoative verb class should not pose any major challenge in L2 contexts. This conclusion of hers comes from a purely theoretical consideration of natural language structure. Working in the Relational Grammar framework, she asserts that the 2 to 1 advancement (i.e. the Relational Grammar term for the object to subject movement) is probably universal. If so, she reasons, the movement is given as a default at the initial stage of L2 acquisition. A couple of problems that result from Zobl’ s account have already been discussed in section 3.2.3 with a detailed review of his idea. One major problem with it is that it cannot account in a satisfactory manner for the existence of non- target structures found in the use of unaccusative-inchoative verbs in L2 English. The question is what motivates learners to abandon the observationally adequate grammar of an earlier stage and take up another later when the new grammar in fact generates structures which they do not find in the input of target English. The same question arises with respect to Hirakawa's and Rosen's ideas. If the correct lexical representation and syntactic derivation of unaccusative-inchoatives are mastered Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 211 early or if they simply come for free at the initial stage of acquisition, why do learners subsequently begin to produce non-target structures? It would appear that the more emphasis is given to the apparent ease of acquiring the NP-V word order, the more difficult it becomes to answer questions about this persistent anomaly. 4.4.3.2. clues to a new account In order to understand why the target-like NP-V structure appears so frequently and what its status is vis-à-vis other IL English structures built on unaccusative- inchoative verbs, we first need to answer one fundamental (but often overlooked) question: "What kind of learners produce apparently non-target structures?" Although no studies have specifically focused on this question, there are quite interesting observations expressed occasionally in passing remarks made by researchers. These observations are quite important for us because collectively, they seem to point to an answer to our question. Consider, first, the fact that non-target structures are found in the L2 English of relatively advanced learners. For example, Hubbard and Hix (1988:93- 94) point out that the great majority of "passivized" unaccusatives appear in the writings of advanced learners. Similarly, Yip (1989, 1994, 1995) repeatedly states that "passivized" unaccusative-inchoatives present problems even to advanced learners. In fact, "passivized" unaccusative-inchoatives are familiar errors to those who teach advanced ESL courses in college settings. In this respect, the fact that Zobl's (1989) data comes from college-age students may not be a coincidence. Despite the fact that these observations all indicate that unaccusative- inchoative verbs clearly present problems to advanced learners, it is not clear at all whether or not the problems are limited to advanced learners. Because many Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 212 studies on the L2 acquisition of unaccusative-inchoatives are based on data collected from college-age subjects, there are very little data on which we can evaluate the challenge that unaccusative-inchoatives present in the acquisition of L2 English in relation to the learners' proficiency levels. There are, however, two studies that seem to give clues to the proficiency- related question. One is Hirakawa's (personal communication 1994) observation that the subjects in her pilot experiment, who had TOEFL scores of 550 to 630, made more errors than those in the reported experiment in her work, where the subjects' TOEFL scores ranged from 530 to 550. What this observation seems to indicate is that the difficulty related to the acquisition of English unaccusative- inchoatives increases as the learners' proficiency improves. Kellerman (1978), in fact, gives fairly concrete support to this suspicion. His study investigated a wide range of subjects and found that it is the advanced rather than the lower-level learners who face more challenges related to unaccusative-inchoative verbs. In the following quotation, Kellerman refers to the results obtained from his transferability judgment tests, which among other sentences contained Hij brak zijn been He broke his leg' and Het kopje brak "The cup broke'. Kellerman w r i t e s A glance at acceptance scores shows that in groups A2, A3, A4 and A5 , 'leg' and cup' are equally acceptable, as one would expect. In A6 there is some difference as there is in KU2. In KUl, KUO, U3 and U1 the differences are marked. The higher the proficiency level the greater the difference in acceptabilitv between these two items, generallv speaking. [....] More sophisticated learners who have a better grasp of the finer grammatical details of their own language and whose English is excellent have rejected the possibility of In the quoted statement, A2 through A6 refer to 2nd through 6th year secondary school pupils. KU2 are 2nd year students of English at Nijmegen University while both KUl and KUO are 1st year students of English at the same university. U1 and U3 indicate 1st year and 3rd year students of English at Utrecht University respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 213 translating or accepting in translation the intransitive form in English . . .8 0 [emphasis mine] (Kellerman 1978:84) With younger (and presumably lower-level) learners, Kellerman found that both transitive and inchoative sentences were judged equally transferable to English. However, as the age (and proficiency level) of subjects went up, the acceptability of the inchoative "cup" sentence, in comparison to the transitive "leg" sentence, started to decrease. Kellerman's study, therefore, shows in a clear manner that the difficulty associated with a verb's unaccusativity becomes more apparent as the proficiency level of learners advances. The observations by Hirakawa (1995) and Kellerman (1978) are part of the motivation for an overall scenario of the L2 acquisition of English unaccusative- inchoative verbs that I present in the following section. 4.5. A N e w H y p o t h e s is 4.5.1. Background The data obtained from the research corpus clearly support previous observations made on the syntactic structures in which unaccusative verbs appear in the English of normative speakers. The "passive" structure, for example, was found to be the most common type of error committed on this class of verbs by the four LI groups that we focused on. Transitivized unaccusatives were also found, though in a small number, in the research corpus. In addition, an interesting contrast was discovered 80 Immediately following this quotation, Kellerman, quoting some of his subjects' words, suggests that what the subjects feel is lacking in a sentence like Het kopje brak T he cup broke' is agency. Though this is similar to what Yip (1994) suggests, I do not agree with this interpretation for the reasons presented in the section on the significance of "passivization. " Subjects' comments based on self-reflection may not always be helpful since they often lack necessary training and/or vocabulary to analyze linguistic phenomena such as movement. Case assignment, etc. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 214 between the Italian/Spanish and the Korean/Japanese groups regarding the use of postverbal NP structures, which I have accounted for as a consequence of contrastive LI properties. One remaining question, however, is the one on the prevalence of seemingly target-like NP-V structure. Clearly, unaccusative- inchoative verbs are not misused as often as we may expect. But why not? This is a very tricky question to answer, because, as the discussion in section 4.4.3.1 shows, if the ease of acquisition of this target structure is emphasized, as done by Rosen (1991) and Hirakawa (1995), it potentially undermines the tacit assumption behind the already established observations on the use of non-target structures, namely, the learning difficulty associated with this class of verbs. An answer that I want to suggest for this challenging question is a three- stage developmental scenario. Specifically, I believe that in L2 contexts, English unaccusative-inchoative verbs are acquired at first as unergatives as far as their A- Structure representation is concerned, and later at a more advanced stage reanalyzed correctly as unaccusative-inchoatives. I argue that it is this second stage where some L2 learners start producing and/or accepting "passivized" sentences and postverbal NP structures and begin avoiding the use of the unaccusative-inchoative verbs in the NP-V word order. The complete acquisition may be achieved when the learners move on to the third (final) stage where their grammar takes on the lexical and syntactic properties characteristic of the native speaker grammar. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 215 4.5.1.1, structure of natural language grammar In the theoretical framework of (24) concerning the structure of natural language grammar, the three stages can be characterized in terms of the state of the lexical and syntactic components of a developing interlanguage grammar. (24) Structure of Grammar Lexicon: LSp. (Lexico-Semantic Representation) < Linking Rules A-Stmcture (Lexico-Syntactic Representation) I I <— X ’ -theory, theta-criterion, linearity parameters Syntax: D-^tructure <— Move Alpha (overt movement) S-Structure (Phonological Form) (Logical Form) Particularly, I argue that linking rules in the lexicon and NP-movement in overt syntax play significant roles in shaping the interlanguage grammar with respect to unaccusative-inchoative verbs. The levels of representation beyond S-Structure, i.e. Phonological Form and Logical Form do not concern us in the following discussion and are therefore enclosed in parentheses in (24). 4.5.1.2. linking rules As mentioned in the introduction, many linguists believe that there are semantic and syntactic levels of lexical representation and these two are systematically connected by some general linking mles within the lexicon. Levin and Rappaport Hovav (1995), for example, propose the following four rules: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 216 (25) a. Directed Change Linking Rule b . Existence Linking Rule c. Default Linking Rule d. Immediate Cause Linking Rule The first three rules are responsible for the projection of A-Structure of unaccusative-inchoative verbs from their semantic representations. Among them, the Directed Change Linking Rule and the Existence Linking Rule are especially relevant with respect to the 10 unaccusative verbs used in the analysis of the research corpus taken from the LLC. To repeat, they are appear, arise, arrive, die, disappear, exist, fall, happen, occur, and rise. Specifically, Levin and Rappaport Hovav argue that the LSR and A- Structure of the unaccusative verbs arise, arrive, die, fall and rise are mapped by the Directed Change Linking Rule in (26). (26) Directed Change Linking Rule The argument of a verb that corresponds to the entity undergoing the directed change described by that verb is its direct internal argument. [Levin and Rappaport Hovav 1995:146] Correctly applied, the Directed Change Linking Rule projects the A-Structure of an unaccusative verb, for example,^// from its LSR as in (27).** * * Letters that represent variables are changed from Levin and Rappaport Hovav (1995) for clarity. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 217 (27) <Unaccusatives (directed change)> L^R: [[0 DO-SOMETHING] CAUSE [y FALL]] <— Directed Change Linking Rule A-Structure: (0 (y)) In the LSR of (27), y is the entity that undergoes the directed change. The Directed Change Linking Rule projects this argument as the internal argument ((y)) of the unaccusative verb. As for the existence/(dis)appearance verbs, let us represent the core lexico- semantic properties of exist as (28) and those of appear, disappear, happen and occur as (29).*^ (28) <Unaccusatives (existence)> LSR: [y BE (Pioc z)] (29) <Unaccusatives (appearance/disappearance)> LSR: [y COME TO (not) BE (Ploc z)] In either case, z is a non-obligatory argument which stands for a spatial or temporal location and normally appears with an appropriate preposition in syntactic structures (Levin and Rappaport Hovav 1995). The LSRs of these existence/(dis)appearance verbs are connected to the unaccusative A-Structure by the Existence Linking Rule in (30). (30) and (31) are not Levin and Rappaport Hovav’ s representations but my own creation based on their theory. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 218 (30) Existence Linking Rule The argument of a verb whose existence is asserted or denied is its direct internal argument. [Levin and Rappaport Hovav 1995:153] In the case of existence verbs, the Existence Linking Rule makes the entity y the internal argument ((y)) as shown in (31).*3 (31 ) <Unaccusatives (existence)> LSR: [y BE (Ploc z)] <— Existence Linking Rule A-Structure: (0 (y)) The A-Structure of appearance/disappearance verbs is also projected in the same manner. These ways of semantics-syntax connection found with unaccusative- inchoative verbs contrast with the way in which the lexical semantic representation of an unergative verb is projected to its A-Structure. In the latter's case. Levin and Rappaport Hovav propose that it is the Immediate Cause Linking Rule in (32) that mediates the two levels of representation. (32) Immediate Cause Linking Rule The argument of a verb that denotes the immediate cause of the eventuality described by that verb is its external argument. (Levin and Rappaport Hovav 1995:135) We ignore the non-obligatory argument "z" since the manner of its projection is not relevant in the current thesis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 219 When applied to a typical unergative verb such as laugh, the Immediate Cause Linking Rule projects the correct unergative A-Structure from its LSR as in (33). (33) <Unergatives> L^R: [ix LAUGH] CAUSE [_________]] <— Immediate Cause Linking Rule A-Structure: (x (0)) In the LSR of the unergative laugh, the entity x is the immediate cause of the expressed eventuality. The Immediate Cause Linking Rule, therefore, projects it as the external argument (x) of the verb. In this way. Levin and Rappaport Hovav's linking rules guarantee that unaccusative-inchoative verbs have the A-Structure: (0 (y)) while the A-Structure of unergative verbs is projected as (x (0)) despite the great semantic differences within each verb class. 4.5.2. Stages of Interlanguage Development 4.5.2.1. first stage: lexical confusion The common assumption adopted in the previous accounts of the NP-V structure built on unaccusative-inchoative verbs by L2 learners is that this class of verbs is correctly differentiated from the unergative class at A-Structure level. This assumption makes it necessary to explain how the intemal argument in the A- Structure of unaccusative-inchoatives comes to assume the surface syntactic position normally occupied by an external argument. Zobl (1989) answered this question by resorting to the early acquisition of lexical NP-movement while Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 220 Hirakawa (1995) and Rosen (1991) emphasized the relative ease in acquiring syntactic NP-movement. These, however, are not the only accounts possible for the NP-V word sequence in L2 English. In fact, the NP-V word order is also possible for unaccusative-inchoative verbs if they are misanalyzed as unergatives at the A- Structure level. What I am suggesting here is a possibility that most learners initially adopt a wrong linking rule which projects a single obligatory nominal argument as the extemal argument of the verb irrespective of the distinction between the two subclasses of intransitive verbs. Let us call this hypothetical linking rule the Single Argument Linking Rule and define it as in (34). (34) Single Argument Linking Rule The single obligatory nominal argument of a verb is its extemal argument. Notice the critical difference between the Immediate Cause Linking Rule in (32) and the Single Argument Linking Rule in (34). Unlike the former, the latter does not distinguish verbs based on the semantic characteristics of their arguments. Consequently, even the non-causative entity y in the LSR of an unaccusative- inchoative, which either undergoes the directed change or whose existence is asserted in the semantics of the verb, is projected as an external argument (y) in its A-Structure representation. In other words, if applied, the Single Argument Linking Rule projects virtually identical A-Structures for both unergatives and unaccusative-inchoatives. Let us see how this hypothetical non-target linking rule may work in the grammar of L2 English learners, using the unergative verb laugh and the unaccusative-inchoative verb fall as examples. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 221 (35) <Unergatives> L^R: [\x LAUGH] CAUSE [_________]] <— Single Argument Linking Rule A-Structure: (x (0)) (36) <Unaccusative-Inchoatives> L^R: [[0 DO-SOMETHING] CAUSE (y FALL]] <— Single Argument Linking Rule A-Structure: (y (0)) The A-Structure in (35) is the same as the one in (33) and perfectly normal for unergative verbs. In contrast, the A-Structure in (36) is non-target because the entity y that is normally projected as an internal argument is in the extemal argument position of the A-Structure representation. However, the syntactic component of grammar treats both A-Structure representations in (35) and (36) as unergative ones:^ (37) D-Structure: [rp [v p NP [v V]] <— NP-Movement (Le. Move Alpha) S-Structure: [IP N?i [vp ti [v V]]] ^ The idea that unaccusatives are initially represented as unergatives is also advanced by Borer and Wexler ( 1986) for first language acquisition data in connection with their theory of maturation of A-Chain. While their idea is intended to account for the rarity of passive sentences and the overgeneration o f causativized sentences based on intransitive verbs in early child language, the idea I am presenting here has nothing to do with the availability o f A-Chain. The availability of A-Chain must be independent of the production of nontarget-like unaccusative sentences such as "passivized" ones in adult SLA because regular passive sentences appear to be produced by learners whether or not they also produce nontarget-like structures on unaccusative verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 222 The S-Structure representation in (37) is the correct one for unergative verbs, but not for unaccusative-inchoative verbs because it lacks trace within V where an intemal argument is normally projected. The surface word order of NP-V in (37), however, superficially matches the expected word order for sentences built on unaccusative verbs. This initial stage of IL grammar with respect to unaccusative-inchoative verbs can be summarized as in (38). (38) First Stage: LSR.........................................Correct I <— * Single Argument Linking Rule A-Structure............................. Incorrect I I D-Structure............................. Incorrect I <— Move Alpha S-Structure...............................Incorrect (but superficially target-like) (38) shows that the fundamental problem at this stage of interlanguage development is lexico-syntactic in nature. That is, the incorrect A-Structure for unaccusative- inchoative verbs originates firom the non-target Single Argument Linking Rule. Paradoxically, however, the sequence of incorrect grammatical representations results in the superficially target-like NP-V word order, making the learner's grammar at this stage at least observationally adequate. 4.S.2.2. second stage: syntactic confusion Recall the observations made by Hirakawa (personal communication 1994) and Kellerman (1978) that advanced learners are more susceptible to problems associated with unaccusative-inchoative verbs (section 4.4.3.2). These observations strongly suggest that learners beyond the initial stage of interlanguage Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 223 grammar in (38) come to face some unique challenge in using unaccusatives and inchoatives. I argue that this challenge arises when these learners replace the Single Argument Linking Rule with other linking rules that are defîned by abstract semantics-syntax correspondences rather than the number of arguments that a verb takes. Consequently, the initial misrepresentation of the A-Structure of unaccusative-inchoatives is corrected, but this reorganization of the lexicon unfortunately creates a new syntactic problem, that is, how to use a verb without an extemal argument when English clearly demands a subject for finite verbs at S- Structure according to the Extended Projection Principle. In section 4.4.2.2,1 have discussed two possible solutions to this syntactic question. One is to transfer an expletive from the lexicon of LI and insert it in the surface subject position. This transferred expletive may be phonetically null or converted to an overt expletive. This solution, however, is possible only if the learner's LI contains such a lexical item. The other solution is to move the NP argument generated in object position to subject position. In doing so, however, some L2 learners unnecessarily mark this syntactic movement by overgeneralizing the passive verbal phrase of be-Ven to the derivation of unaccusative-inchoative sentences. This results in the non-target "passivized" structure. Moreover, the avoidance phenomena involving both unaccusatives and inchoatives in the NP-V word sequence can be thought of as a third kind of syntactic solution. In sum, this second stage of interlanguage grammar development can be schematized as (39). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 224 ff (39) Second Stage: LSR..............................................Correct I <— Correct Linking Rules A-Structure................................ Correct I I D-Structure................................ Correct <---- * expletive transfer^ *intrusive "be -Ven, I and/or avoidance S-Structure................................. Incorrect As (39) clearly illustrates, the fundamental problem in the learner’ s grammar at this stage is not lexical but syntactic in nature.*^ 4.5.2.3. third stage: out of confusion The third (and final) stage of grammatical development with respect to the unaccusative-inchoative verb class is achieved when learners have learned not to use the wrong expletives, that is, a null expletive and it, and/or have unlearned the overt marking of movement of an NP argument from object to subject position with the passive morphosyntax.*^ The tendency to avoid unaccusative-inchoative verbs in the NP-V word order should also disappear once the syntactic movement of an intemal argument is correctly acquired. The third stage is summarized in (40). Incidentally, this theory predicts that reaching this second stage of interlanguage development is a necessary, though not sufficient, prerequisite for the complete mastery of morphological processes, e.g. -er nominalization (section 2.1.2), passive and perfective adjective formation (section 2.1.3) and -ing (compound) adjective formation (section 2.1.4), which are sensitive to the distinction between the extemal and intemal arguments at A-Stmcture le v e l. 86 O f course, with certain subclasses of unaccusative verbs, learners can learn to use the expletive there in subject position without syntactic NP-movement. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 225 (40) Third Stage: LSR...............................................Correct X < — Correct Linking Rules A-Structure...................................Correct D-Structure...................................Correct X < — Move Alpha S-Structure...................................Correct As far as the derivation of unaccusative-inchoative verbs is concerned, the interlanguage grammar at this stage finally attains English target criteria. 4.5.3. Reorganization of Interlanguage Grammar The three-stage model of interlanguage development has the potential to account for an otherwise confusing picture of L2 acquisition of English unaccusative-inchoative verbs. The usefulness of the model becomes especially clear when actual transition from one stage to another is considered. In this section, therefore, I discuss developmental issues that must be dealt with in this model. 4.5.3.1. to the first stage The first question concerns the hypothetical Single Argument Linking Rule in (34). In section 4.5.1.2,1 have argued that at the initial stage of interlanguage development, L2 learners have virtually identical A-Structure representations for both unergatives and unaccusative-inchoatives due to the non-target linking rule. Why do L2 learners adopt it, avoiding target rules such as (25a-d), which are after all supposedly universal? Let us assume that Levin and Rappaport Hovav’ s (1995) linking rules (or their equivalents) are in fact part of the fundamental properties of all natural languages called Universal Grammar (UG). If UG, or at least this sub Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 226 component of UG, is still accessible in adult L2 acquisition, then, L2 learners of English should have direct access to natural linking rules such as (25a-d). Even if UG is not directly accessible, its manifestation as linking rules in learners' LI may help them to reach target linking rules in L2. Why, then, aren't natural linking rules initially adopted as a default option by learners? The answer, I believe, is that the Single Argument Linking Rule is in fact the most "natural " choice for learners of this stage to assume in L2 contexts. Since the majority of unergatives and unaccusative-inchoatives appear in the NP-V word order in target English, by adopting the Single Argument Linking Rule learners can generate, without complication, sentences that conform to the surface grammatical pattern of both subclasses of intransitive verbs. In short, the Single Argument Linking Rule, which is input-driven or externally motivated in the sense that it is more responsive to surface word orders than to presumably universal semantico- syntactic regularities, is the most efficient and practical rule to adopt for L2 learners. This kind of makeshift rule appears to be extremely useful for learners who have overwhelming needs to put their L2 competence in unmediate use for communicative tasks whether they are in spoken or written language. 4.5.3.2. to the second stage If the grammar based on the Single Argument Linking Rule is the most useful one for L2 learners' immediate communicative needs, then, what motivates the transition firom the first stage of interlanguage grammar in (38) to the second stage in (39)? This transition is in effect when learners begin to regard the nominal argument of an unaccusative-inchoative verb as an argument more like the object of a transitive verb and to see that of an unergative verb as an argument more like the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 227 subject. Clearly, this transition is a risky one for the learners because they now abandon an older grammar that generates only grammatical sentences, albeit superficially, for a new one which may not. Let us consider two possible motivations for the change. The first possibility is that the transition is effected by some regularities in the target language input that learners at the relevant stage come to "sense" and exploit in reorganizing their developing grammar. The relevant input should indicate to learners that the argument of the unaccusative-inchoative verb class is an intemal argument just like the object of the transitive verb class. Is there positive evidence of this kind in target English? In fact, there is plenty such evidence available to learners. Consider ergative verbs such as bend, break, change, close, decrease, drop, increase, melt, open, etc. that could provide learners with concrete examples in which thematically identical NPs appear either as the subject or the object in active sentences. These and other ergative verbs may play a significant role in the reorganization process of the developing interlanguage lexicon by dissociating the preverbal subject position from the external argument in an A-Structure representation. In the case of non-alternating unaccusative verbs, however, evidence of this kind appears to be extremely limited in the input. Although some unaccusatives such as exist and arrive do appear in the "there" construction with a postverbal NP argument and may potentially suggest to learners that their argument is also an intemal argument, the use of this constmction is very limited. Furthermore, as shown in section 4.3, there is some evidence that this stmctural possibility for unaccusative verbs is generally not utilized by L2 learners of English. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 228 In sum, if positive evidence has any impact on the transition from the first to the second stage, inchoative verbs are likely to undergo the lexical reorganization before unaccusative verbs do. Consequently, we may observe that L2 learners start producing and/or accepting "passivized" inchoatives before they do "passivized" unaccusatives. Similarly, they are also expected to exhibit the tendency for avoidance of the NP-V word order with inchoatives before they do with unaccusatives. The second possibility is that the transition in question is internally, rather than externally, motivated. In this scenario, what L2 learners exploit is not regularities available in the input, but some presumably universal semantics-syntax correspondences manifested in the lexicon of a natural language. In other words, it is the lexical component of UG, directly or indirectly accessible to learners, that drives the change. One potential candidate for the lexical UG is outlined in Sorace (1993a, 1993b, 1995). Based on diachronic and crosslinguistic studies of Italian, French and English as well as L2 acquisition research involving these languages, Sorace has proposed that there is a possibly universal unaccusative hierarchy, which is schematized in (41) with Italian and French examples. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 229 (41) The Unaccusative Hierarchy (based on Sorace 1993a, 1993b, 1995) Core CHANGE OF LOCATION andare/aller 'go' CHANGE OF CONDITION sparire / disparaitre disappear' CONTINUATION OF A CONDITION durare / durer 'last' EXISTENCE OF A CONDITION esistere / exister exist' TRANSITIVE ALTERNANT aumentare / augmenter 'increase' UNERGATIVE ALTERNANT correre / courir run' Peripheral The Unaccusative Hierarchy in (41) is ordered based on verbs' semantics. Verbs placed higher are considered to be closer to the core of unaccusativity while those at lower ranks are regarded to be more peripheral. What is significant for us is that non-alternating unaccusatives are clearly higher than alternating inchoatives in the proposed Unaccusative Hierarchy. If the transition from the first to the second stage is internally motivated by the lexical component of UG such as the Unaccusativity Hierarchy, unaccusatives are predicted to undergo the lexical reorganization before inchoatives do. Consequently, the prediction is that "passivized " unaccusatives emerge before "passivized " inchoatives and similarly unaccusative avoidance before inchoative avoidance. Notice that this is the exact opposite of what the input-based account above has predicted. Finally, let us ask what percentage of L2 learners of English actually come to the second stage of grammatical development? This question is difficult to answer. The number, however, may not be very large, considering that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 230 approximately 90% of interlanguage English sentences with unaccusative verbs were found in the NP-V word order in the research corpus (section 4.3). Why is this so? First of all, since the first-stage interlanguage grammar with the Single Argument Linking Rule generates nothing but observationally adequate sentences, learners may not have much incentive to replace the Single Argument Linking Rule with more abstract linking rules. For example, even if the lexical UG such as Sorace's Unaccusative Hierarchy has potential to put the developing interlanguage grammar into a reorganization process, available semantic and syntactic clues to the division between unergatives and unaccusative-inchoatives may be too subtle for many L2 learners and may not in fact have significant effect on them. In this respect, L2 learners of English seem to be at a disadvantage compared to those learning Italian, for instance. When Italian is a target language, learners may come to realize the existence of the two subclasses of intransitive verbs more easily because of highly noticeable and consistent syntactic phenomena such as auxiliary selection and ne-cliticization. In addition, the morphologically marked ergative alternation with the clitic si may make L2 learners of Italian sensitive to the regularities in semantics-syntax correspondences at early stages of acquisition. In English, however, for the learners who have misanalyzed all intransitive verbs as unergatives at A-Stmcture level, there are very few clearly observable syntactic or morphological clues that may help them to correct the misanalysis and lead them to the correct distinction of the two subclasses of intransitives (Snyder and Stromswold 1997:294, footnote 18). As a result, many L2 learners of English may overlook available subtle evidence and simply remain at the first stage of grammatical development for a long time or even forever. This rather pessimistic Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 231 scenario accounts for the very high frequency of NP-V word order with unaccusative-inchoatives found in actual L2 data. 4.5.3.S. to the third stage Finally, how does the transition from the second to the third stage of interlanguage development proceed? The critical aspects of this transition are (i) unlearning of the overgeneralization of the passive morphosyntax to every object-to-subject movement and (ii) discontinuation of the use of non-target expletive subjects. In the case of L2 English of native Italian/Spanish speakers, the latter may involve two steps: first, realization that there is no null expletive subject in English, and second, awareness that the overt expletive it is not used with unaccusative verbs. When (i) and (ii) are achieved, the avoidance phenomenon is expected to disappear as well. Let us consider first how the target input available to L2 learners may influence the transition from the second to the third stage. In this connection. Yip (1989, 1994, 1995) has made an important prediction, concerning the relative difficulty of expunging "passivized" unaccusatives and "passivized" inchoatives. Her idea depends critically on the contrast between unaccusatives and inchoatives that is apparent in the following pairs of examples. (42) a. The accident happened. (an unaccusative verb) b. *The accident was happened. (based on an unaccusative verb) (43) a. The window broke. (an inchoative verb) b . *P^ The window was broken. (* if based on an inchoative verb, but ok if based on a causative verb ) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 232 Yip insightfully points out that unlike (42b), which is always ungrammatical and presumably non-existent in target language input, (43b) is perfectly grammatical and potentially available. This means that learners who are moving away from the "passivization" may be held back by positive evidence of grammatical passivization based on causative use of ergative verbs. This scenario, therefore, predicts that "passivized" inchoatives are harder to expunge than "passivized" inchoatives.*'^ In contrast, the account based on the lexical UG such as Sorace's Unaccusativity Hierarchy in (41) appears to predict very little about the transition from the second to the third stage. This is because the transition to the third stage is fundamentally syntactic in nature. As described above, the process involves unlearning of non-target syntactic operations such as overgeneralization of the passive morphosyntax and insertion of non-target expletives. The lexical component of UG has nothing to say about this unlearning process. Therefore, if overgeneralization of the passive morphosyntax is correctly unlearned by a learner, both "passivized" unaccusatives and "passivized" inchoatives should start disappearing from his or her L2 English around the same time. In short, UG, or at least the lexical component of UG, predicts very little about the transition from the second to the third stage of interlanguage development. How, then, do learners at the second stage of development manage to move to the third (and final) stage? This process in fact may be just as difficult as the previous transition from the first to the second stage, if not more so. The target language input potentially available to them appears to be of very little help when the task is to unlearn non-target syntactic properties their interlanguage grammar has The direct effect of positive evidence in the input on the postverbal NP structures and the avoidance of NP-V word order is not very clear to me at this point. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 233 incorporated. Moreover, as discussed above, at least in the case of "passivized" inchoatives, the input actually may work against a smooth transition. Explicit information that structures such as "passive" unaccusative-inchoatives and the (it-) V-NP word order are not acceptable in target English would be certainly helpful, but it is generally assumed that negative evidence of this kind is normally unavailable to learners (Yip 1994). Also, when it comes to the avoidance of NP-V word order, negative evidence is not easy to provide because avoidance is not a readily observable phenomenon to begin with. Without explicit negative evidence, learners may need to exploit indirect negative evidence that some structures which their interlanguage grammar generates are not present in ±e target language input. Although the effectiveness of indirect negative evidence for language acquisition is not very clear, a combination of indirect negative evidence and observations that the meanings which they intend to convey with structures such as "passivized" unaccusative-inchoatives are conveyed in other structures in the target language may be ultimately the only way for L2 learners to move out of the second stage to the third (and final) stage. In any case, as Yip (1994) suggests, a total recovery from the "unaccusative trap" may not be easy for L2 learners without some external help such as consciousness-raising in granunatical instruction. 4.5.3.4. summary In this section, I have considered how the L2 acquisition of unaccusative-inchoative verbs may proceed along the line of development outlined by the three-stage model. I have discussed what may be involved in a transition from one stage to another and what may motivate learners for the reorganization of their interlanguage grammar. Both available target language input and presumably universal properties of a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 234 natural language lexicon have been considered for their potential contribution to the grammatical reorganization process. It has been noted that input-based and UG- based predictions are sometimes contradictory with respect to the relative difficulty to acquire the grammatical properties of unaccusatives and inchoatives. Although this topic has not attracted much attention in the literature (i.e. with the notable exception of Yip 1989, 1994, 1995), the issue clearly deserves further research, especially with respect to the contrastive predictions that are made by the input- based and the UG-based accounts. 4.5.4. The "Unaccusative Trap" and U-shaped Development In this chapter, I have presented a new hypothesis concerning the way English unaccusative-inchoative verbs are acquired by nonnative speakers. In sum, I regard the various L2 phenomena on English unaccusative-inchoatives as manifestations of a formidable linguistic challenge, which has its origin in ±e mismatch between the lexical properties and the dominant surface syntactic pattern of NP-V shared by all the verbs of this class. The mismatch leads learners to a Catch-22 in which a wrong A-Structure representation for unaccusative-inchoatives at the initial stage of grammatical development ensures a simple and straightforward mapping to a superficially correct surface syntactic pattern, whereas the correct A-Structure representation at the second stage often results in complex (and often wrong) non target syntactic patterns. At first, the interlanguage lexicon seems to be organized in response to dominant surface syntactic patterns found in target language input. Later, it appears to be reorganized on the principle of internal consistency with respect to the syntax-semantics correspondences manifested in natural languages although it is not clear at this point whether the reorganization process is externally Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 235 driven by some subtle regularities observable in input or internally motivated by the learners' access to the lexical component of Universal Grammar. We have seen how a correct reorganization of the interlanguage lexicon creates a new question of syntactic derivation and often leads to the generation of non-target sentence patterns. This situation may be called the "unaccusative trap." As a real trap that catches unsuspecting victims, the "unaccusative trap " is mostly hidden from L2 learners and it may be difficult to avoid because it waits for its victims in the normal course of their interlanguage development. Once "trapped," the victims' English becomes unacceptable in light of the target grammar and escape may not be easy. From a theoretical point of view, however, the "unaccusative trap" is in fact much more sinister than it first appears. Notice that in the three-stage model proposed above, the learners "trapped" in a learning theoretic maze are not only those at the second stage of interlanguage development. Those who are still at the first stage are in a sense already caught in this linguistic trap. It is true that as far as the use of unaccusative-inchoatives are concerned, their language is observationally adequate at the moment. Nonetheless, it is this group of learners, after all, who are likely to start producing non-target structures once they reach a more sophisticated level of linguistic awareness and accordingly reorganize their interlanguage grammar. Therefore, a more proper characterization of the "unaccusative trap" is as a two-stage phenomena. At the first stage, potential damaging effects remain unapparent while at the second, symptoms of the problem actually become visible. In this picture, only when learners go beyond the second stage and reach the final stage of interlanguage development are they considered to be out of the sinister grip of the "unaccusative trap. " Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 236 The "unaccusative trap" account offers an interesting prediction. If it is on the right track, most observed anomalies in learners’ use of unaccusative- inchoatives should be attributable to those learners at the second stage of interlanguage development. Those at the initial stage should generate observationally adequate surface structures thanks to the Single Argument Linking Rule while those who have reached the final stage, of course, are expected to generate legitimately acceptable sentences. This scenario predicts that there will be a U-shaped developmental pattern concerning the acquisition of English unaccusative-inchoative verbs by nonnative speakers. This prediction, in fact, was borne out in the case of Kellerman's research on the transferability of the Dutch verb breken to English (Kellerman 1979:38-40, Sharwood Smith and Kellerman 1989:227-230). In this research, the subjects’ English proficiency was approximated by arranging them into eight groups according to the number of years they had studied English. They were from 13 to 23 years in age and those subjects 18 years old or above were English majors at university. A very clear U-shaped pattern of development was found to emerge with respect to the inchoative avoidance phenomenon discussed in section 3.1.2 in contrast to a normal developmental curve found for the acceptance of the transitive break. The acceptability scores for both inchoative and transitive break went up as the age of the subjects (and, presumably, their English proficiency, too) increased. At age 17-18, however, the acceptance rate of the inchoative began dropping while that of the transitive remained at the same high level. The inchoative acceptance rate kept dropping to age 19-20, where it started going up again. In other words, the phenomenon of inchoative avoidance was most prominent between 17 and 20. This clear pattern of U-shaped development on the acceptance (or avoidance) of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 237 inchoative break in the NP-V word order is fully consistent with the idea that Dutch-speaking learners of English undergo the kind of reorganization of their interlanguage grammar that I have proposed with the three-stage model in this chapter. Developmental patterns, however, have not been researched for the "passivized" unaccusative structure and the postverbal NP structure. Nonetheless, the observation made on the inchoative avoidance phenomenon strongly suggests that similar U-shaped patterns of development, which are independently predicted by the "unaccusative trap" account, also exist with respect to the non-target structures produced in L2 English. This question is one of the issues which I address in the next chapter on the grammatical intuitions of L2 learners regarding English sentences that contain unaccusative-inchoative verbs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 238 5. Grammaticality Judgment Test 5.1. I n t r o d u c t io n The L2 acquisition of English unaccusative-inchoative verbs is clearly a very complicated topic to study. The existing L2 literature reviewed in Chapter 3 and the results of the corpus-based research reported in Chapter 4 both attest to the grammatical, psychological, and developmental complexities involved in various aspects of the acquisition process. Some factors that potentially influence and/or derive from the development of L2 grammar in this process are, for example, the syntax-semantics correspondences manifested in linking rules, classification and subclassifrcation of verbs, morpho-syntactic operations such as "passivization" and transitivization, overgeneralization and avoidance of an aspect of the target grammar, grammatical characteristics of the learner's LI, the learner's proficiency levels, and so forth. Among these, most research in the past focused on particular phenomena or grammatical anomalies and investigated them often independently of each other. The three-stage model that I proposed at the end of the previous chapter (section 4.5) is intended to capture most, if not all, of the previous observations on the use of unaccusative-inchoative verbs in L2 English. To my knowledge, this is the first attempt to synthesize the complex and sometimes seemingly contradictory facts into one general coherent picture. In this chapter, instead of taking the inductive approach that I have pursued in most of the previous chapters, I try to see deductively what the three-stage model predicts about potential interactions of the various factors that conspire to shape L2 English grammar with respect to unaccusative-inchoative verbs. A series of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 239 predictions were made with the three-stage model in mind and tested against data obtained from a grammaticaliQr judgment test conducted on native speakers of Italian, Japanese and English. The first two LI groups were chosen in order to do further investigation of the crosslinguistic differences found in the corpus study discussed in Chapter 4. It is hoped that the series of predictions presented in the form of research hypotheses in this chapter also serve as a first step toward a comprehensive understanding of the way in which English unaccusative-inchoative verbs are acquired by normative speakers. 5.1.1. Research Hypotheses The research hypotheses presented below were developed with regard to the possible interactions of first language, syntactic structure, verb class and proficiency level of normative speakers. 5.1.1.1. "passivization" As evidenced in sections 3.1.1 and 4.4.1, "passivized" unaccusative-inchoatives are produced by diverse groups of normative speakers and may be considered to be a common feature of L2 English. In judging the grammaticality of experimental sentences, therefore, Italian and Japanese speakers are expected to reject "passivized" unaccusative-inchoatives less reliably than native speakers do. This prediction is summarized in the following two sets of hypotheses. Hypothesis 1 (re: "passivized" unaccusatives) A Italian speakers correctly reject "passive" unaccusatives less often than native English speakers do. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 240 B: Japanese speakers correctly reject "passive" unaccusatives less often than native English speakers do. Hypothesis 2 (re: "passivized" inchoatives) A: Italian speakers correctly reject "passive" inchoatives less often than native English speakers do. B: Japanese speakers correctly reject "passive" inchoatives less often than native English speakers do. One unstated assumption behind the first two sets of hypotheses is that nonnative speakers are more sensitive to the ungrammaticality of "passivized" unergatives than they are to that of "passivized" unaccusative-inchoatives. In fact, in the corpus-based research reported in Chapter 4, there was virtually no instance of "passivized" unergatives. Therefore, normative speakers are expected to reject "passive" unergatives as ungrammatical English more consistently than they do "passive" unaccusatives or "passive" inchoatives. Therefore, Hypotheses 3 & 4 are: Hypothesis 3 (re: "passivized" unaccusatives) A: Italian speakers correctly reject "passive" unergatives more often than they do "passive" unaccusatives. B: Japanese speakers correctly reject "passive" unergatives more often than they do "passive" unaccusatives. Hypothesis 4 (re: "passivized" inchoatives) A: Italian speakers correctly reject "passive" unergatives more often than they do "passive" inchoatives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 241 B: Japanese speakers correctly reject "passive" unergatives more often than they do "passive" inchoatives. At this juncture, let us ask a question concerning the relative strength in the nonnative speakers' tendency to accept (or fail to reject) "passive" unaccusatives vis-à-vis "passive " inchoatives. As discussed in section 4.5.3, two contrastive predictions are made depending on whether we regard the input or the lexical UG as primary motivation for developmental change of interlanguage. On the one hand, if the input has the primary impact on the developing grammar, "passive" inchoatives are predicted to be more common than "passive" unaccusatives in L2 English. First, as I speculated in section 4.5.3.2, the existence of ergative verbs that alternate between inchoative and causative usage may help learners to dissociate the preverbal subject position from the external argument in an A-Structure representation. If so, inchoatives are expected to undergo the correct reorganization of their A-Structure representation, which is a prerequisite for the non-target "passivization," earlier than unaccusatives do. In other words, inchoative verbs" entry to the second stage of grammatical development is likely to precede unaccusative verbs". At the other end, that is, at the exit of the second stage, "passive" inchoative verbs are expected to be harder to expunge than "passive" unaccusatives because of the existence of grammatical passive sentences based on the causative verbs that share the same morphological form with their inchoative counterparts. Such grammatical passive sentences in the input may prove to be sources of difficulty for learners who need to unlearn the overgeneralization of the passive morphosyntax associated with syntactic NP movement in their grammar. In sum, the input-based account predicts that the non- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 242 target "passivization" may set in earlier and persist longer with inchoatives than with unaccusatives. On the other hand, if the lexical component of UG such as the Unaccusativity Hierarchy proposed by Sorace (1993a, 1993b, 1995) plays the primary role in the transition from the first to the second stage of the three-stage model, the prediction turns out to be exactly the opposite. Unaccusative verbs, which are placed higher than inchoative verbs in the Unaccusative Hierarchy, are predicted to undergo the lexical reorganization before inchoatives do, with their sole NP argument correctly projected as an internal argument at the A-Structure level. Naturally, in this case, one is led to expect that "passive" unaccusatives emerge before "passive" inchoatives in the learners' language. At the exit phase of the second stage, however, no prediction can be made based on the Unaccusative Hierarchy, which is lexical in nature and has nothing to say about unlearning of non-target syntactic structures such as the "passivization" of English unaccusative- inchoative verbs. In sum, these theoretical considerations lead to the following two sets of predictions concerning the relative magnitude of incorrect "passivization" of unaccusatives and inchoatives in the grammar of L2 English. Hypothesis 5 (re; magnitude of "passivization" [input-based prediction]) A Italian speakers correctly reject "passive" unaccusatives more often than they do "passive" inchoatives. B: Japanese speakers correctly reject "passive" unaccusatives more often than they do "passive" inchoatives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 243 Hypothesis 6 (re: magnitude of "passivization" [UG-based prediction]) A: Italian speakers correctly reject "passive" inchoatives more often than they do "passive" unaccusatives. B: Japanese speakers correctly reject "passive" inchoatives more often than they do "passive" unaccusatives. 5.1.1.2. NP-V word order The second group of hypotheses concern the canonical NP-V word order in which the majority of both unergatives and unaccusative-inchoatives appear in target English. Previous research has shown that among L2 learners of English there is a tendency to avoid inchoative verbs in the canonical word order (Kellerman 1978, Yip 1989, 1994, 1995, Hirakawa 1995). If inchoative avoidance is a general phenomenon, it is expected that both Italian and Japanese speakers perform in a significantly different marmer in comparison to native English speakers. Thus, the first hypothesis concerning the NP-V word order is: Hypothesis 7 (re: inchoative avoidance) A. Italian speakers incorrectly reject inchoatives in the NP-V word order more often than native English speakers do. B. Japanese speakers incorrectly reject inchoatives in the NP-V word order more often than native English speakers do. In section 4.5,1 have suggested that inchoative avoidance is a result of a reorganization of the developing L2 lexicon, namely, a consequence of reclassification of inchoative verbs as such after a period of misanalyzing them as unergatives. This account of inchoative avoidance should apply to unaccusative Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 244 verbs as well if my contention that the latter also undergo a similar reclassification is correct. Therefore, the second prediction is that there is a phenomenon of avoidance of the NP-V word order in the use of unaccusatives by normative speakers: Hypothesis 8 (re: unaccusative avoidance) A: Italian speakers incorrectly reject unaccusatives in the NP-V word order more often than native English speakers do. B: Japanese speakers incorrectly reject unaccusatives in the NP-V word order more often than native English speakers do. The assumption behind Hypotheses 7 & 8 is that normative speakers' avoidance of the NP-V word order containing inchoatives and unaccusatives is selective in the sense that they do not exhibit the same tendency with unergative verbs. This assumption is summarized below as Hypotheses 9 & 10. Hypothesis 9 (re: inchoative avoidance versus unergative avoidance) A: Italian speakers incorrectly reject inchoatives in the NP-V word order more often than they do unergatives. B: Japanese speakers incorrectly reject inchoatives in the NP-V word order more often than they do unergatives. Hypothesis 10 (re: unaccusative avoidance versus unergative avoidance) A: Italian speakers incorrectly reject unaccusatives in the NP-V word order more often than they do unergatives. B: Japanese speakers incorrectly reject unaccusatives in the NP-V word order more often than they do unergatives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 245 As in the case of "passivization," an important question to ask at this point concerns the relative strength of the normative speakers' tendency to avoid inchoatives and unaccusatives in the NP-V word order. In this case, however, predictions that can be made by the input-based and UG-based perspectives are limited to the transition of L2 grammar into the second stage of the three-stage model. Neither perspective appears to have much to say about the grammatical reorganization characteristic of the third stage. On the one hand, as has been discussed in section 5.1.1.1, it is possible that the existence of many ergative verbs in input helps learners to move to the second stage in correctly reanalyzing the lexical representation of inchoative verbs. In this scenario, ergative alternation is hypothesized to help learners to dissociate the surface subject position and the external argument at A-Structure level. In the case of unaccusative verbs, however, available evidence of this kind is extremely limited. Consequently, the association of the surface subject position and a "theme/patient" argument is expected to remain strong with unaccusatives longer than with inchoatives. In sum, the input-based prediction is that the transition from the first stage to the second would be easier with inchoative verbs than with unaccusatives. The UG-based account, on the other hand, predicts the exact opposite. In this scenario, it is unaccusatives, not inchoatives, that first undergo the lexical reorganization into the second stage. If the interlanguage development at this phase is internally driven by UG-properties such as the Unaccusative Hierarchy, this is a natural prediction because non-alternating unaccusatives are placed higher in the hierarchy and are presumably better examples of verbs that manifest unaccusativity. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 246 The two hypothesized timings of lexical reorganization lead to contrasting predictions on the relative strength of the learners' tendency for the NP-V word order avoidance involving unaccusatives and inchoatives. Since the avoidance phenomenon is presumed to be a result of the correct A-Structure representation of these verb classes achieved at the second stage, it is expected to set in earlier for the verb class that undergoes the transition to this stage first. The final predictions concerning the avoidance of NP-V word order in L2 English, therefore, are opposing statements summarized as Hypotheses 11 and 12. Hypothesis 11 (re: magnitude of avoidance [input-based prediction]) A Italian speakers incorrectly reject inchoatives in the NP-V word order more often than they do unaccusatives. B: Japanese speakers incorrectly reject inchoatives in the NP-V word order more often than they do unaccusatives. Hypothesis 12 (re: magnitude of avoidance [UG-based prediction]) A Italian speakers incorrectly reject unaccusatives in the NP-V word order more often than they do inchoatives. B: Japanese speakers incorrectly reject unaccusatives in the NP-V word order more often than they do inchoatives. 5.1.1.3. postverbal NP structures The data on L2 learners of English with LI backgrounds in Italian, Spanish, Japanese and Korean obtained from the Longman Learners Corpus (LLC) has revealed a number of interesting facts concerning the use of postverbal NP structures on English unaccusative verbs among these nonnative speakers (section Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 247 4.4.2). One of the findings was the rarity of the "there" construction in comparison to the other two postverbal NP structures. If this is a common phenomenon, we expect that both Italian and Japanese tend not to regard this construction with unaccusative verbs as correct English. This is the basis of Hypothesis 13. Hypothesis 13 (re: avoidance of the "there" construction) A: Italian speakers incorrectly reject the there-V-NP structure more often than native English speakers do. B: Japanese speakers incorrectly reject the there-V-NP structure more often than native English speakers do. A second fact we observed was the relatively high frequency of the it-V-NP structure in the Italian and Spanish data which contrasted with the lack of it in the Korean and Japanese data. In section 4.4.2,1 suggested that the reason behind the observed contrast between the Italian/Spanish data and the Korean/Japanese data is the existence of null expletives in the former pair of languages and the lack of such a lexical item in the latter pair. I speculated that the observed overt expletive it is simply a phonologically realized version of the null expletive. If this analysis is correct, there should be a significant difference concerning the grammaticality judgment of the it-V-NP structure between Italian speakers on the one hand and native English and Japanese speakers on the other hand. This prediction is stated as Hypothesis 14 below. Hypothesis 14 (re: crosslinguistic difference on acceptance of the expletive it) A Italian speakers incorrectly accept the it-V-NP structure more often than native English speakers do. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 248 B: Italian speakers incorrectly accept the it-V-NP structure more often than Japanese speakers do. The analysis of the it-V-NP structure was in fact motivated by a similar observation on the use of the 0 -V-NP structure between the Italian/Spanish speakers and the Japanese/Korean speakers. In the research corpus, this postverbal NP structure was also relatively frequent in the L2 English of Italian and Spanish speakers but was rarely used by Korean and Japanese speakers. So, naturally, the last prediction concerning postverbal NP structures is: Hypothesis 15 (re: crosslinguistic difference on acceptance of a null expletive) A: Italian speakers incorrectly accept the 0 -V-NP structure more often than native English speakers do. B: Italian speakers incorrectly accept the 0 -V-NP structure more often than Japanese speakers do. 5.1.1.4. transitivization In the target grammar of English, the transitivization of unaccusative verbs is ungrammatical. However, previous research (section 3.1.3) and the data based on the LLC (section 4.3) show evidence for this non-target use of the verbs in L2 English. Although I have already argued against it, it is even suggested that the reason for the ungrammatical "passivization" of unaccusatives is the existence of ungrammatical transitivization (sections 3.1.3.2 and 3.1.3.3). In any case, if this tendency to use unaccusatives in the transitive structure of NPi-V-NPi is prevalent among L2 learners of English in general, we should expect to see that both Italian and Japanese speakers incorrectly transitivize unaccusatives more often than native Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 249 English speakers do. Therefore, the first hypothesis concerning the L2 transitivization of English unaccusative verbs is: Hypothesis 16 (re: transitivization of unaccusatives) A: Italian speakers incorrectly transitivize unaccusatives more often than native English speakers do. B: Japanese speakers incorrectly transitivize unaccusatives more often than native English speakers do. Theoretically speaking, the transitivization of unaccusatives is assumed to be the result of awareness on the part of learners that the unaccusative class of verbs lacks an external argument at the A-Structure level of representation (cf. section 3.1.3.2, (22)). That is, the vacancy in the external argument position in the A- Structure representation of unaccusative verbs makes it easy to add a causative argument to their A-Structure. Since the unergative class of verbs already have an external argument, this should discourage L2 learners from adding another external argument to the unergative A-Structure. Therefore, it is predicted, as stated in Hypothesis 17, that there is a significant difference among both Italian and Japanese speakers in their tendency to transitivize unaccusatives and unergatives. Hypothesis 17 (re: transitivization of unaccusatives versus unergatives) A: Italian speakers incorrectly transitivize unaccusatives more often than they do unergatives. B: Japanese speakers incorrectly transitivize unaccusatives more often than they do unergatives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 250 One question not addressed previously in the literature is the type of unaccusative verbs that may be transitivized in L2 English. As I hinted in section 3.2.2.2, however, whether or not all the subclasses of unaccusatives are equally open to transitivization is an important question to ask in order to understand the real nature of transitivized unaccusatives in L2 English. Because all the tokens of transitivized unaccusatives obtained in the corpus-based research discussed in Chapter 4 are verbs of (dis)appearance and verbs of directed change with no instance of verbs of existence (see Appendix B), I hypothesize that not all unaccusative verbs are equally susceptible to transitivization in the grammar of nonnative speakers' English. A specific prediction is expressed as Hypothesis 18. Hypothesis 18 (re: transitivization of semantic subclasses of unaccusatives) A: Italian speakers incorrectly transitivize verbs of (dis)appearance and verbs of directed change more often than they do verbs of existence. B: Japanese speakers incorrectly transitivize verbs of (dis)appearance and verbs of directed change more often than they do verbs of existence. 5.1.1.5. raising verbs One issue that has not been discussed in the current thesis yet is the way that raising verbs such as appear and seem are dealt with by nonnative speakers. As briefly mentioned in section 4.4.1.5, raising verbs also manifest unaccusativity defined in the spirit of Burzio's Generalization (section 1.2) and could involve movement of an NP to the subject position of the matrix clause from a position within the subcategorized subordinate clause. The NP movement involved in the raising Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 251 construction, however, differs crucially from the object-to-subject movement because it takes place across, not within, a clause boundary. Do L2 learners overgeneralize the passive morphosyntax to the NP movement involved in the raising construction as they do to unaccusative-inchoative sentences? Examples cited by Yip (1995) show that such a possibility is clearly not a mere theoretical speculation: (1) a. One will find that English poems were seemed to be more personal, b . The Chinese poems are seemed to be more sophisticated and . . . . [based on Yip 1995:146, (50)]8S In these examples, as in the case of "passivized" unaccusative-inchoatives, the passive morphosyntax of be + Ven clearly serves to indicate that the surface subjects English poems and the Chinese poems are syntactically derived. Existence of L2 data such as (1) seems to indicate that there may be two types of nonnative speakers of English who overgeneralize the passive morphosyntax as a marker of NP-movement.*^ With some learners, the overgeneralization may be limited to the NP-movement within a clause boundary while others may overgeneralize it for movement across as well as within a clause boundary. The first kind of overgeneralization, which results in the "passive" unaccusative-inchoative structure of NP-be+Ven, may be called "weak (or limited) overgeneralization" while the second with the surface word order of NP-be+Ven-to infinitive clause, may be characterized as "strong (or extended) overgeneralization." Yip’ s original examples show that these "passivized" raising constructions are used with the target raising construction by the same learner, revealing that non-target structures are often in competition with target structures in the grammar of L2 learners. I owe the original insight for the idea suggested here to Joseph Aoun. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 252 Thus, the research question, formulated as Hypothesis 19 below, concerns the status of the "strong (or extended) overgeneralization" of the passive morphosyntax in the grammar of L2 English: Hypothesis 19 (re: the raising construction) A: Italian speakers correctly reject the preverbal NP "passive" raising structure less often than native English speakers do. B: Japanese speakers correctly reject the preverbal NP "passive" raising structure less often than native English speakers do. Finally, the logic behind the presumed difficulty for normative speakers to reject the preverbal NP "passive" raising structure also leads to a purely theoretical speculation that the ungrammatical "passivization" should be easier to detect if the subject of the matrix clause is not a moved NP but the expletive it, that is, it- be+Ven-that clause. In such a case, there is no need for L2 learners to overgeneralize the morphosyntax of English passive sentences to mark syntactic NP movement since there is no NP movement involved in the derivation of the sentence. So, the second prediction about raising structures is: Hypothesis 20 (re: the raising construction) A: Italian speakers correctly reject the preverbal NP "passive" raising structure less often than the expletive subject "passive" raising structure. B: Japanese speakers correctly reject the preverbal NP "passive" raising structure less often than the expletive subject "passive" raising structure. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 253 5.1.1.6. proficiency levels and U-shaped development The discussion in section 4.5 has led to the prediction of a U-shaped developmental pattern in the use of unaccusative-inchoative verbs by nonnative speakers of English. As discussed in section 4.5.3, this prediction is already borne out in the case of inchoative avoidance observed in the Dutch speakers' judgment of the English inchoative verb break presented in the NP-V word order (Kellerman 1979, Sharwood Smith and Kellerman 1989). If the three-stage account proposed in section 4.5 is on the right track, we should expect a similar U-shaped development in other non-target structures and phenomena as well. Therefore, a general prediction on the developmental aspect of the L2 acquisition of English unaccusative-inchoative verbs is Hypothesis 21. Hypothesis 21 (re: developmental patterns) Concerning the non-target structures and the avoidance of target structures, development by both Italian and Japanese speakers reveals a U-shaped. 5.2. METHOD 5.2.1. Subjects Native speakers of Italian, Japanese and English participated in this study. All the subjects were undergraduate or graduate students who were attending a university in Italy, Japan or the United States at the time of the study. The Italian and Japanese data were collected in Rome and Nagoya respectively in the summer and the fall of 1996. The data from the native English speakers were collected in Los Angeles in February, 1997. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 254 AU the native Italian subjects were students at Université degU Studi di Roma "La Sapienza." Although the initial number of students who participated in the study was slightly higher, in the end a total of 64 subjects were included in the analysis of the data.^° The majors and the class levels of the Italian subjects are given in Table 5.1.1. Table 5.1.1 Italian Subjects Majors (class levels); Languages and Literatures (including English) (2 first-year, 2 second-year, 11 third-year, 10 fourth-year, 13 beyond course work, 3 unknow n)....................................................................................................... 41 Medicine ( 11 second-year, 1 third-year, 6 fourth-year, 2 unknown 20 Unidentifîed major (2 fourth-year, 1 unknown)................................ 3 t O T A L ................................... 64 The Japanese subjects were students of Nanzan University and Aichi Prefectural University. A total of 168 of them participated in the study: 125 from Nanzan University and 43 from Aichi Prefectural University. The majors and class levels of the Japanese subjects are shown in Table 5.1.2. Some responses were not included in this study because the subjects turned out to be bilingual or failed to provide vital information such as the version of the judgment test that they answered. Some Japanese and English speakers were excluded from the study for similar reasons. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 255 Table 5.1.2 Japanese Subjects <Nanzan University (125)> Majors (class levels): English (including Linguistics and American History) (30 freshmen, 1 sophomore, 52 juniors, 18 seniors & I graduate) 102 Foreign Languages (unidentified) (8 juniors)................................... 8 Education, Japanese, Japanese Education (1 junior & 4 seniors).... 5 P hilosop hy (5 juniors)...................................................................... 5 Politics, Business, Economics (2 sophomores, 1 junior & 1 senior).. 4 S p an ish (1 senior).............................................................................. 1 <Aichi Prefectural University (43)> Majors (class levels): English (19 sophomores & 2 juniors)............................................... 21 Spanish (17 freshmen & 4 sophomores)........................................... 21 Foreign Languages (unidentified) (1 junior).................................... 1 T O T A L .................................................................................................... 168 The average length of English study for the Italian and Japanese subjects combined was approximately 8.9 years at the time of the study. The native English speakers who participated in the study were students in two freshman writing classes at the University of Southern California and a graduate teaching assistant who was in charge of one of the classes. The majors and class levels of a total of 24 native speaker subjects are shown in Table 5.1.3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 256 Table 5.1.3 Native English-Speaking Subjects Majors (class levels); U ndecided (all freshmen)...................................................................... 9 B usiness (all freshmen)......................................................................... 4 B iology (2 freshmen).............................................................................. 2 F ilm (2 freshmen) ................................................................................. 2 M usic (1 freshman & 1 senior)............................................................... 2 Computer Science (1 freshman ......................................................... E ducation (1 graduate).......................................................................... Jou rn alism (1 freshman)...................................................................... Pre-dental Science (I freshman)......................................................... T heater (1 freshman)............................................................................ T O T A L ........................................................ :................................... H In the three language groups combined, the number of male and female subjects were 67 and 188 respectively with one unidentified. The participation was all voluntary and no subject was compensated monetarily or otherwise. 5.2.2. Materials Two kinds of tests were developed for the study. The first is a cloze test for measurement of the subjects' English proficiency and the other is a grammaticality judgment test for assessment of their English grammar on experimental items. Cloze tests are commonly used to measure the overall integrative proficiency in foreign languages and their validity has been supported by a considerable body of literature (e.g., Hinofotis 1980, Hinofotis and Snow 1980, Hanania and Shikhani 1986, Sciarone and Schoorl 1989, Fotos 1991). Unlike discrete-point tests, cloze tests measure advanced integrative proficiency that draws on the subjects' overall lexical, syntactic, semantic and rhetorical knowledge of the target language, including awareness of discourse-level cohesion and organization of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 257 text (Hanania and Shikhani 1986, Fotos 1991). Studies in the past have indicated high correlation between cloze tests and established ESL proficiency measures such as TOEFL (Hinofotis 1980). Following Sciarone and Schoorl (1989), who claim that the minimum number of blanks in a cloze test should be 75 and that the word deletion rate should be between every 5 to 10 words, a 75-item cloze test was developed using the fixed-ratio (i.e. every 5th word) method. Three separate narratives were adapted firom an ESL reader.®* The original stories came from 1000-, 1400-, and 1800- word level readings respectively. In each story word deletion was made except in the first and last sentences which were kept intact to provide contextual information. (The complete cloze test is shown in Appendix C). The three stories, with 25 blanks each, were approximately the same length and were presented in the order of their original difficulty in order to avoid discouraging subjects by presenting a difficult narrative first. The grammaticality judgment test was developed to assess the subjects' perception of a few English verb classes presented in various structural patterns. The verb classes included unergatives, unaccusatives, inchoatives (i.e. intransitive versions of ergatives), and raising verbs. Two lists of verbs were created in order to test more than one or two verbs in each class. Irregular verbs such as break and fall were excluded from the lists in order to prevent experimental items fi-om being judged as incorrect due to the subjects' confusion on the verbs' conjugation patterns. Each list contained 4 unaccusatives, 2 unergatives, 2 inchoatives and 2 raising verbs. These verbs are shown in (2) ® * The book is Alexander and Cornelius 1978. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 258 (2) List A Unaccusatives: Inchoatives: Unergatives: Raising verbs: List B Unaccusatives: Inchoatives: Unergatives: Raising verbs: exist* melt walk appear die appear* decrease smile seem happen* remain* arrive* disappear shatter increase dance laugh appear seem occur* (Verbs with * have potential to appear in the "there" construction.) Each verb, except for the two raising verbs, was presented in six structural patterns, while each raising verb was given in four syntactic structures: (3) Structural Patterns in Which Non-Raising Verbs were Presented: a. N P-V b. NP - be 4 - Ven c. there - V - NP d. it - V - NP e. 0 - V-NP f. NPi - V -NP2 Structural Patterns in Which Raising Verbs were Presented: g. NP - V - to-infinitive clause h. NP - be + Ven - to-infinitive clause i. it - V - that-clause Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 259 j. it - be + Ven - that clause Two sets of judgment tests were developed based on the two verb lists. Each test contained a total of 56 items to be judged on their grammaticality: 6 for each of the 8 non-raising verbs and 4 for each of the two raising verbs. In order to provide contextual information, short introductory passages were created. All the experimental clauses, which were underlined, followed one introductory passage. This introductory passage was the same for each verb irrespective of the structural patterns in which it was presented. Grammatical and contextual correctness of each underlined clause as an English structure was judged on a 5-scale measure: 1 for "not correct English," 2 for "probably not correct English, " 3 for '"cannot decide, " 4 for "probably correct English," and 5 for "correct English." (The two forms of grammaticality judgment test are given in Appendix D.) 5.2.3. Procedures All the tests were administered in regular class hours. The Italian and Japanese data were collected by the teachers who regularly taught the classes of the subjects. The tests on the native English speakers were administered by the researcher. First, the subjects were given a sheet written in their native language that explained the purpose of the study, the voluntary nature of participation and the guaranteed anonymity of their responses. Then, the order of test administration was the cloze test followed by the grammaticality judgment test. In the cloze test, the subjects were told to write a contextually and grammatically appropriate word in each blank. In the grammaticality judgment test, the subjects were asked to record their judgment on each experimental item by filling the appropriate oval space on a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 260 computer-readable multiple-choice answer sheet. A detailed written explanation about how to use the computer sheet was given to the subjects in their native language. Personal information such as name, sex, birth date, native language, and length of English study were collected on the computer-readable answer sheet for the grammaticality judgment test. Academic major and class level were indicated on the cloze test sheet. Each test took approximately 30 minutes, but for the grammaticality test an additional 5 to 10 minutes were allowed if the subjects needed extra time to complete the test. In the case of native English speakers, completion of each test took considerably less time. 5.2.4. Scoring and Analyses For scoring the cloze test answers, the acceptable-word method was used. The results of acceptable-word and exact-word methods usually show high correlation with each other. The acceptable-word method, nonetheless, is credited with superior reliability because it discriminates among levels far better than the exact- word method (Hinofotis 1980:126-127, Fotos 1991). A list of acceptable answers was compiled based on responses from 23 mixed LI subjects who participated in a pilot study and in close consultation with two native speakers of English. In deciding acceptable answers, coherence of the text, pragmatic appropriateness and stylistic fit were taken into account. Minor spelling errors and the confusion of upper and lower case were tolerated, but grammatical errors in part of speech, tense, number, etc. were not. (The complete list of acceptable answers for the cloze test is found in Appendix C.) The scores were not weighted item by item. The possible maximum score, therefore, was 75. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 261 The scoring of the grammaticality judgment test was done based on the correspondence between the subjects' answers and predetermined expected responses. For example, if the expected response for a particular experimental item was " 1 " (not correct English), both " 1 " and "2" were given a point of I. If the expected response was "5" (correct English), both "5" and "4" were given a point. All the other answers were scored as 0. In each form of the grammaticality test, the following cases were considered as "correct English. " (4) The Structure and Verb Combination Considered as "Correct English" In the NP-V word order: 4 unaccusative verbs, 2 inchoative verbs, and 2 unergative verbs In the there-V-NP structure: 3 unaccusative verbs marked with * in (5) In the NPj - V -NP2 structure: 2 inchoative verbs In the NP -V - to-infinitive clause structure: 2 raising verbs In the it -V - that clause structure: 2 raising verbs In total, 17 out of the 56 experimental items in each judgment test were considered as "correct English" for scoring purposes. The results obtained from the two versions of the judgment test were matched on the structure and verb class combination. In comparing verb classes whose sizes were not equal in the test, for example, the unaccusatives and the unergatives, the score on the smaller verb class was doubled. A 3 x 3 analysis of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 262 variance (ANOVA) by language and verb class (or structure) was applied to the scores. When significant effects were observed in the interaction of two main effects, s Scheffé test was employed in order to locate exactly where the non- random distribution in the scores was. In comparing each language group's performance on different structures or verb classes, a paired sample t-test was used. The level of significance set for the three statistical measures was a<.05. 5.3. R e s u l t s /D is c u s s io n s The results of the cloze test are shown in Table 5.2. Table 5.2 Cloze Results Mean Score Standard Deviation Minimum Score Maximum Score English (24) 58.375 5.492 45 68 Italian (64) 49.953 11.918 17 67 Japanese (168) 47.47 9.356 19 66 Italian/Japanese (232) 48.155 10.162 17 67 As for the grammaticality judgment test, the data were obtained on 130 responses to Form A (13 English, 29 Italian, 88 Japanese) and 126 responses to Form B (11 English, 35 Italian, 80 Japanese). 5.3.1. "Passivization" The expected judgment on the structure of NP-be+Ven was "not correct EngUsh" with unaccusative, inchoative, and unergative verbs. In case of "passive" inchoatives, despite the existence of graimnatical passive causatives of identical forms, the experimental clauses were unacceptable for contextual reasons. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 263 Therefore, points were given only if the subjects answered "I" or "2." The highest possible score was 4. The descriptive statistics of the three language groups' scores on the "passive" structure are shown in Tables 5.3.1 to 5.3.3. Table 5.3.1 Unaccusatives: NP-be+Ven (expected = rejection, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.583 .504 24 Italian 2.500 1.321 64 Japanese 2.351 1.248 168 Table 5.3.2 Inchoatives: 7VP-6e+Ven (expected = rejection, max. score : = 4.0) Mean Standard Deviation Number of Subjects English 3.333 .963 24 Italian 2.687 1.344 64 Japanese 2.762 1.195 168 Table 5.3.3 Unergatives: NP-be+Ven (expected =: rejection, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.833 .565 24 Italian 3.125 1.279 64 Japanese 2.738 1.341 168 At a glance, it appears that the English group identified the ungrammaticality of the experimental sentences in all three verb classes more consistently than the Italian and the Japanese groups did. In order to see if the observed differences between the native speakers and the nonnative speakers were statistically significant, analysis of variance was employed. ANOVA revealed significant effects of language (F (2, 253) = 14.79, /?< 0005) and of verb class (F (2, 506) = 5.18, p<.006). However, the interaction of the two main effects was not significant (F (4, 506) = 1.73, p<. 142). Therefore, despite the seemingly large Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 264 differences between the native speakers and the nonnative speakers in their performance on each verb class. Hypothesis 1 and Hypothesis 2 were not statistically supported. Next, the two nonnative groups' performance on the three verb classes was analyzed by using t-tests for paired samples. As for the Italian speakers, there was a significant difference (t = 2.62, df=^ 63, p<.011) between their scores on "passivized" unergatives (M = 3.125) and on "passivized" unaccusatives (M = 2.500), but no significant difference (t = 1.72, df= 63,p<.090) between "passivized" unergatives (Af= 3.125) and "passivized" inchoatives (M = 2.687). Similarly, the Japanese subjects' scored significantly higher (t = 3.12, df=^ 167, /7<.002) on "passivized" unergatives (M = 2.738) than on "passivized" unaccusatives (Af = 2.351) but their scores on "passivized" unergatives (M = 2.738) and "passive" inchoatives (Af = 2.762) were not found to be significantly different (X = .11, df= 167, p<.863). Therefore, Hypotheses 3A and 3B were both supported, but Hypotheses 4A and 4B did not receive support. Recall the two contrastive predictions made in section 5.1.1.1 concerning the relative strength of L2 learners' propensity for "passivized" unaccusative- inchoatives. On the one hand. Hypothesis 5, based on the view that input is the primary factor for the relevant grammatical development, predicted that ungrammatical "passive" inchoatives would be less often detected than equally ungrammatical "passive" unaccusatives. On the other hand, the UG-based prediction of Hypothesis 6 foresaw that the exact opposite would be the case. The results of the Judgment test clearly showed that both Italian and Japanese subjects rejected the "passivization" of inchoative verbs more reliably than the "passivization" of unaccusative verbs (Af = 2.687 versus Af = 2.500 for Italians and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 265 M = 2.762 versus M = 2.351 for Japanese), which clearly favors the UG-based Hypothesis 6 over the input-based Hypothesis 5. In particular, in the case of the Japanese subjects, this difference between "passive" inchoatives and "passive" unaccusatives was found to be significant (t = 4.02, df= 167, p<-G0G5), giving statistical support to Hypothesis 6B. Thus, the obtained data strongly suggest that the relative difficulty for L2 learners of English to recognize ungrammatical "passivization" is greater with unaccusatives than with inchoatives. If the logic behind the two sets of hypotheses is in fact correct, this finding seems to indicate that the transition to the second stage of the three-stage developmental model (section 4.5) is driven primarily by learners' "awareness" of some universal lexical properties such as the Unaccusative Hierarchy proposed by Sorace. Conversely, the potential influence of input such as the existence of ergative altemation (see section 5.1.1.1) and grammatical passivization of causative verbs (e.g. Yip 1989, 1994, 1995) may not be as great during the relevant reorganization period of interlanguage grammar. 5.3.2. A/P-Y Word Order In contrast to the passive structure, the NP-V word order in which unaccusative, inchoative, and unergative verbs appeared was expected to be accepted as "correct English." Therefore, scores were given if the subjects answered "4" or "5." The highest possible score was 4. Tables 5.4.1 to 5.4.3 show the descriptive statistics of the three language groups' performance on the three verb classes presented in the NP-V word order. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 266 Table 5.4.1 Unaccusatives: NP-V (expected = acceptance, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.500 .659 24 Italian 2.641 .915 64 Japanese 2.708 .962 168 Table 5.4.2 Inchoatives: NP-V (expected = acceptance, max. score = 4.0) Mean Standard Deviation Number of Subjects English 2.417 1.316 24 Italian 3.125 1.228 64 Japanese 2.548 1.289 168 Table 5.4.3 Unergatives: NP-V (expected = : acceptance, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.917 .408 24 Italian 3.469 1.023 64 Japanese 3.119 1.193 168 In order to find out if there were any statistically significant differences among the three language groups' performance over the verb classes, analysis of variance was applied to the data. ANOVA revealed significant effects of language (F (2, 253) = 6.45, p<.0G2), of verb class (F (2, 506) = 24.35, p<.0005), and of the interaction of the two main effects (F (4, 506) = 6.54, p<.0005). In their performance on the unaccusative class, significant differences among the language groups were found: F (2, 253) = 8.4325, p<.0003. Table 5.5 Results on Unaccusatives Source D.F. SS MS F ratio F probability Between Groups Within Groups Total 2 253 255 14.4948 217.4427 231.9375 7.2474 .8595 8.4325 .0003 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 267 Scheffé's test revealed that the native English speakers (Af = 3.500) performed significantly better (p<.05) than both Italian (M= 2.641) and Japanese (Af = 2.708) groups, supporting Hypotheses 8A and 8B on unaccusative avoidance. In addition, ANOVA showed that there were significant differences among the language groups in their performance on the inchoatives: F (2, 253) = 5.3004, p<.0056. Table 5.6 Results on Inchoatives Source D.F. SS MS F ratio F probability Between Groups Within Groups Total 2 253 255 17.2820 412.4524 429.7344 8.6410 1.6302 5.3004 .0056 Scheffé's test revealed that the Italian subjects' performance (Af = 3.125) was significantly better (p<.05) than the Japanese subjects' (Af = 2.548). This result, however, did not support Hypotheses 7A and 7B on inchoative avoidance because the significant difference was not between the native English speakers and a nonnative subject group. In order to see the performance of each nonnative speaker group on the three verb classes, paired sample t-tests were employed. The results showed that the Italian subjects performed significantly better on unergatives (Af = 3.469) than on either unaccusatives (Af = 2.641): t = 5.63, df= 63, p<.0005, or inchoatives (Af = 3.125): t = 2.2, df= 63, p<.047. Similarly, the Japanese subjects scored significantly better on unergatives (Af = 3.119) than on unaccusatives (Af = 2.708): t = 3.72, df= 167, p<.0005, and on inchoatives (Af = 2.548): t = 4.83, df= 167, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 6 8 p<.0005. Therefore, Hypotheses 9A, 9B, 10A, and lOB on inchoative and unaccusative avoidance were ail supported by the judgment data. As for the relative strength of inchoative and unaccusative avoidance in the NP-V word order, although the Japanese subjects rejected the inchoatives (Af = 2.548) more often than the unaccusatives (Af = 2.708), the difference was not statistically significant: t = 1.49, df= 167, p<.138. This result, therefore, statistically supported neither the input-based Hypothesis 1 IB nor the UG-based Hypothesis 12B. On the other hand, the Italian subjects' score on the inchoatives (Af = 3.125) was significantly better than their score on the unaccusatives (Af = 2.641) (t = 2.60, df= 63, p<.012). This clearly rejected the input-based prediction of Hypothesis HA and supported the UG-based prediction of Hypothesis 12 A. In sum, the overall results from the judgment data have confirmed normative speakers’ tendency to avoid unaccusative verbs in the canonical NP-V word order. Both Italian and Japanese subjects rejected the unaccusative verbs more often than the native speaker subjects did. Moreover, they considered the unaccusative verbs appearing in the NP-V word order as ungrammatical more often than they did the unergative verbs in the same syntactic structure. However, inchoative avoidance received mixed results. On the one hand, both Italian and Japanese subjects rejected the inchoatives in the NP-V word order as ungrammatical more often than they did the unergative verbs. On the other hand, their responses on the inchoatives were not significantly different compared to the native English speakers'. Two possible may account for this. One is that the Italian subjects scored an unexpectedly high mean (Af = 3.125) on this verb class. The other is that the native speakers' score (Af = 2.417) was surprisingly low. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 269 reasons for these unexpected scores of the Italian and English groups, however, are not clear at present. Finally, the judgment data have shown that there is no statistically significant evidence that inchoative avoidance is stronger than unaccusative avoidance, casting doubt on the importance of input for the hypothesized grammatical reorganization from the first to the second stage of interlanguage development. In fact, the statistical analyses have indicated that the relative strength of avoidance is actually the exact opposite, at least, as far as the Italian subjects are concerned. They had more difficulty in correctly accepting unaccusatives than inchoatives in the canonical NP-V word order. This supports the speculation that it is not directly observable input but rather learners' awareness of UG properties such as the Unaccusative Hierarchy that drives the reorganization of the interlanguage lexicon, which is presumed to be a precursor of the avoidance phenomenon as discussed in section 4.5.3.2. 5.3.3. Postverbal NP Structures In the analyses of the three types of postverbal NP structure, only those unaccusative verbs which potentially appear in the "there" construction were included. These verbs indicated with * in (5) are: exists appear, happen, remain, arrive and occur. All of them, except arrive, are verbs of existence or appearance. Even arrive, which has been characterized as a verb of directed change (or specifically, a verb of directed motion), seems to have some sense of appearance and existence since the entity that "arrives" at a location is not previously in sight and assumes a presence after its arrival. So, for convenience's sake, I will refer in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 270 the present section to this subclass of unaccusative verbs, including arrive, as verbs of existence and appearance. The scoring of the answers to the postverbal NP structures was done strictly by grammatical possibility. That is, the experimental clauses that contained the there-V-NP structure were considered to be "correct English" for scoring purposes and given a point if the subjects answered "5" or "4." The it-V-NP and 0-V-NP structures were considered to be "incorrect English" and were assigned a point if the subjects' answers were "I" or "2." Since each judgment test form contained three relevant verbs, the highest possible score was 3. The descriptive statistics of the three language groups' performance on the grammaticality judgment of the three postverbal NP structures are shown in Tables 5.7.1 to 5.7.3. T able 5.7.1 Existence/Appearance: there-V-NP (expected = acceptance, max. score = 3.0) Mean Standard Deviation Number of Subjects English 1.333 .702 24 Italian .891 .911 64 Japanese 1.464 .966 168 T able 5.7.2 Existence/Appearance: it-V-NP (expected = rejection, max. score = 3.0) Mean Standard Deviation Number of Subjects English 2.958 .204 24 Italian 1.844 .912 64 Japanese 2.030 .872 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 271 Table 5.7.3 Existence/Appearance: 0-V-NP (expected = rejection, max. score = 3.0) Mean Standard Deviation Number of Subjects English 2.625 .495 24 Italian 1.672 1.024 64 Japanese 2.131 .893 168 ANOVA revealed significant effects of language (F (2,253) = 29.81, p<.0005), of structure (F (2, 506) = 54.99, /?<.0005), and of the interaction of the two main effects (F (4, 506) = 4.09, p<.003). As for the there-V-NP structure, there were statistically significant differences among the language groups: F (2, 253) = 8.8048, p<.0002. Table 5.8 Results on there-V-NP Structure Source D.F. SS MS F ratio F^robability Between Groups Within Groups Total 2 253 255 15.2677 219.3534 234.6211 7.6338 .8670 8.8048 .0002 Although the Italian subjects' mean score (Af = .891) was much lower than both the English subjects’ (Af = 1.333) and the Japanese subjects’ (Af = 1.464), Scheffé’ s test revealed that the only statistically significant difference was between the Japanese and the Italians. This finding, therefore, did not support Hypotheses 13 A and 13B on "there" construction avoidance because there was no statistically significant difference between the native speaker group and either of the normative subject groups. ANOVA revealed that there were significant differences among the language groups in their scores on the it-V-NP structure: F (2, 253) = 15.7820, p<.00005. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 272 Table 5.9 Results on it-V-NP Structure Source D.F. SS MS F ratio F probability Between Groups Within Groups Total 2 253 255 22.4874 180.2470 202.7344 11.2437 .7124 15.7820 .00005 Scheffé's test found that the mean of the English subjects (M - 2.958) was significantly better than both that of the Italians (M = 1.844) and the Japanese (Af = 2.030). The result clearly supported Hypothesis 14A, which predicted that Italian speakers would incorrectly accept this structure more often than native English speakers would. However, Hypothesis 14B did not gain support from the data because there was no significant difference between the two normative subject groups in their scores. In fact, the result of Scheffé's test indicated that the Japanese subjects' responses were more like the Italians' in that they accepted the structure in question significantly more often than the native speakers did. As for the 0-V-NP structure, significant differences among the language groups were found: F (2, 253) = 11.2058, p<.00005. Table 5.10 Results on 0-V-NP Structure Source D.F. SS MS F ratio F probability Between Groups Within Groups Total 2 253 255 18.1466 204.8534 223.0000 9.0733 .8097 11.2058 .00005 Scheffé's test revealed that the native speaker subjects scored significantly better (Af = 2.625) in rejecting the ungrammatical structure than the Italians (Af = 1.672). The Japanese subjects' score (Af = 2.131) was also found to be significantly better than the Italian subjects'. The findings, therefore, support both Hypotheses 15A Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 273 and 15B which predicted the Italian tendency to incorrectly accept this postverbal NP structure. However, the English subjects' score was found to be significantly better than the Japanese subjects' as well. This result indicates that the Japanese subjects' grammatical judgment on this postverbal structure is somewhere between that of the native English speakers and the Italian subjects', i.e. significantly different from either of them. As for the within-subjects effects, the Italian subjects correctly rejected the it-V-NP structure slightly more (M= 1.844) than they did the 0-V-NP structure (Af = 1.672) while the Japanese did better rejecting the 0-V-NP structure (M = 2.131) than they did the it-V-NP structure (M = 2.030). However, these differences were found to be not significant in either language group: t = 1.10, df= 63, p<.277 for the Italian scores, and t = 1.11, df= 167, p<.261 for the Japanese scores. The mean score of the native English subjects on the there-V-NP structure was rather low, indicating that many experimental clauses in this structure were rejected by them as incorrect. This result itself is not a surprise since the "there" construction is rather restricted in actual English usage due to discourse factors. Although efforts were made to make the experimental passages with this structure as natural sounding as possible, many of them may not have appeared acceptable for their unnatiuralness in their discourse context. Still, the relatively low score by the Italian subjects is noteworthy though it was found to be significant only in relation to the Japanese subjects' score. Italians' tendency to avoid the "there" construction definitely deserves further investigation. Despite the prediction made by Hypothesis 14B, the it-V-NP structure was accepted by the Japanese subjects significantly more often than by the native English speakers. In fact, the contrast among the language groups turned out to be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 274 between the two nonnative speaker groups and the native speakers, rather than between the Italian subjects on the one hand and the native speakers of English and Japanese on the other. However, reasons for accepting this structure as correct English may be different for the Italian subjects and for the Japanese subjects. In section 4.4.2,1 have argued that the reason behind the relatively frequent use of this structure by Italians and Spanish is the psychological reality of the null expletive subject in these languages. Accordingly, the expletive pronoun it used in this structure is argued to be simply a phonetically realized version of the null expletive in their LI. This account should not work for the Japanese subjects who accepted the postverbal NP structure with it in the judgment test if the Japanese language does not contain an expletive in its lexicon as I have argued. Why, then, did the Japanese subjects accept the experimental clauses? One possibility is that the Japanese subjects considered the expletive it as a referential pronoun that functioned as a causative subject. In other words, the it-V- NP may have been understood by them as a transitive structure. Needless to say, the transitivization of unaccusative verbs is not grammatical. That is, as our data will show later (section 5.3.4), unaccusative verbs do not transitivize in target English. Also, the experimental passages were carefully created so that the expletive it could not be interpreted as a referential pronoun. For example, potential antecedents were presented in plural forms so that they would not agree with it in number. Despite these grammatical constraints, however, it is quite plausible that the Japanese subjects interpreted it as a referential pronoun for a causative entity. First, as we will wimess later (section 5.3.4), the Japanese subjects did accept transitivized unaccusative verbs in L2 English. Second, Japanese learners often Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 275 make errors on number agreement in their use of English, which reflects the lack of number agreement in their LI. Finally, the data obtained on the judgment of the V-NP structure presents a complicated picture. As predicted, the Italian subjects accepted this structure significantly more often than both the native English speakers and the Japanese subjects. However, the Japanese group were also found to accept the structure significantly more often than the native English speakers. As for this phenomenon, I do not have any speculations to offer at present. The implications of this finding for the presumed psychological non-existence of a null expletive in Japanese are not very clear. 5.3.4. Transitivization Three verb classes were included in the analyses of transitivization: unaccusatives, inchoatives and unergatives. In scoring the subjects' responses, transitivized unaccusatives and unergatives were considered to be "incorrect English" and a point was given to only the answers " 1" and "2." In the case of inchoative verbs, their transitive versions, i.e. causative verbs, were considered to be "correct English " and a point was given to the answers "5" and "4." The highest possible score was 4. The descriptive statistics of the results are given in Tables 5.11.1 to 5.11.3. Table 5.11.1 Unaccusatives: NPi-V-NPi (expected = rejection, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.750 .442 24 Italian 2.609 .919 64 Japanese 2.643 .943 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 276 Table 5.11.2 Inchoatives: NP1-V-NP2 (expected = acceptance, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.333 1.274 24 Italian 3.187 1.220 64 Japanese 2.857 1.324 168 Table 5.11.3 Unergatives: NP1-V-NP2 (expected = rejection, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.917 .408 24 Italian 3.156 1.224 64 Japanese 3.357 1.079 168 ANOVA showed significant effects of language (F (2, 253) = 11.18, p<.0005) and of verb class (F (2, 506) = 7.92, p<.0005). The interaction of the two main effects was also found to be significant (F (4, 506) = 2.95, p<.020). On the scores on the unaccusative verbs in the NP1-V-NP2 structure, ANOVA revealed significant between subjects effects: F (2,253) = 16.6530, /?<.00005. Table 5.12 Results on NP1-V-NP2 Structure Source D.F. SS MS F ratio F probability Between Groups Within Groups Total 2 253 255 27.1590 206.3058 233.4648 13.5795 .8154 16.6530 .00005 Scheffé's test indicated that the English group's score (Af = 3.750) was significantly higher (p<.05) than both the Italian subjects' (Af = 2.609) and the Japanese subjects' (Af = 2.643). This finding, therefore, supports Hypotheses 16A and 166 that predicted the transitivization of unaccusative verbs in the L2 English of Italian and Japanese speakers. Incidentally, no significant efiects of language Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 277 were found (F (2, 253) = 2.4732, /?<.0864) in correctly accepting the transitivized inchoatives (i.e. grammatical causatives), indicating that the three language groups accepted the structure equally well. To find out about within-subjects effects for the Italian and Japanese subjects, t-tests for paired samples were employed. As for the Itaban subjects, they performed significantly worse in rejecting the transitivized unaccusatives (Af = 2.609) than the transitivized unergatives (Af = 3.156): t = -3.38, df= 63, p<.001. Similarly, the Japanese subjects' score on the transitivized unaccusatives (Af = 2.643) was significantly worse than their score on the transitivized unergatives (Af = 3.357): t = -7.30, df= 167, p<.0 0 0 5 .^^ These findings support Hypotheses 15A and 15B which predicted that both Italians and Japanese would incorrectly transitivize unaccusatives more often than they would unergatives. Next, to find out whether lexical semantics have any effects on the L2 transitivization of English unaccusative verbs, analysis of variance was applied to the scores on three subclasses of unaccusative verbs: appearance/disappearance verbs {appear, disappear, happen and occur), directed change verbs {die and arrive) and existence verbs {exist and remain). The expected responses for all types of transitivized unaccusatives were "not correct English." The highest possible score was 2. Tables 5.13.1. to 5.13.3 show the descriptive statistics of the results. Incidentally, the t-test revealed no statistically significant difference in the native English speakers' scores on unaccusatives (Af = 3.750) and on unergatives (Af = 3.917): t = -1.70, df= 23, p<.103. The result, therefore, seems to indicate that transitivization of unaccusative verbs Is an interlanguage phenomenon. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 278 T able 5.13.1 (Dis)appearance; NP1-V-NP2 (expected = rejection, max. score = 2.0) Mean Standard Deviation Number of Subjects English 1.917 .282 24 Italian .953 .575 64 Japanese 1.167 .672 168 Table 5.13.2 Directed change: NPi V-NPz (expected = rejection, max. score = 2.0) Mean Standard Deviation Number of Subjects English 1.667 .761 24 Italian 1.594 .811 64 Japanese 1.452 .894 168 Table 5.13.3 Existence: NP1-V-NP2 (expected = rejection, max. score ; = 2.0) Mean Standard Deviation Number of Subjects English 2.000 .000 24 Italian 1.719 .701 64 Japanese 1.500 .869 168 A t-test for paired samples revealed that the native Italian speakers performed significantly worse in rejecting the (dis)appearance verbs (Af = .953) than in rejecting the existence verbs (Af =1.719): t = -8.40, df= 63, p<.0005. They also performed significantly worse in rejecting (dis)appearance verbs than in rejecting directed change verbs (Af = 1.594): t = -5.22, df= 63, p<.0005. However, no significant difference was noted in their performance on the directed change verbs and on the existence verbs: t = -1.07, df= 63, p<.289. A paired samples t-test revealed the same pattern in the native Japanese speakers' responses on the grammaticality of the transitivization of the three types of unaccusatives. The Japanese subjects scored significantly worse on the transitivized appearance and disappearance verbs (Af = 1.167) than on the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 279 transitivized existence verbs (M = 1.500): t = -4.15,167, p<.0005. Their score on the transitivized appearance and disappearance verbs was also significantly worse than their score on the transitivized directed change verbs (M= 1.452): t = - 3.33, df= 167, p<.001. However, there was no statistically significant difference in their scores on the transitivized directed change verbs and the transitivized existence verbs: t = -.48, d f— 167, p<.629. These findings support Hypotheses ISA and 18B in part. Both the Italian and Japanese groups had more difficulty with the appearance and disappearance verbs than with existence verbs presented in the transitive s t r u c t u r e . ^ ^ Despite the prediction of Hypothesis 18, however, the transitivized directed change verbs were rejected as often as the transitivized existence verbs. This shows that the directed change verbs were treated more like the existence verbs than the appearance and disappearance verbs by both the Italian and Japanese groups. In sum, the predictions on the transitivization of unaccusative verbs were mostly supported by the judgment data. In comparison to the native group, both the Italian and Japanese groups were found to have a propensity for accepting non target transitivization of English unaccusatives. This difference between the native and normative speakers was in contrast to the lack of any statistically significant difference in their acceptance of grammatical causative verbs. Both the Italian and Japanese groups also showed a differential treatment of transitivized unaccusatives and transitivized unergatives, identifying the latter’ s ungrammaticality more easily. In addition, ANOVA showed significant effects o f language (F (2, 253) = 11.64, p<.0005), of lexical semantics (F (2, 506) = 10.05, p<.0005) and of the two main effects (F (4, 506) = 4.22, p<.002). After ANOVA found significant between subjects effects on the transitivized appearance/disappearance verbs (F(2, 253) = 21.0929, p<.00005), Scheffé's test was employed. The result revealed that the English group's score (Af = 1.917) was significantly better (p>.05) than the score o f either the Italians (A/ = .953) or the Japanese (A4 = 1.167). This finding indicates that the low rejection rate of the transitivized appearance/disappearance verbs is an L2 English phenomenon. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 280 It was also discovered that the propensity for ungrammatical transitivization is not uniform among semantic subtypes of unaccusative verbs. Clearly, the verbs of appearance and disappearance were more susceptible to transitivization than verbs of existence. Despite the prediction, however, verbs of directed change were found to be as resistant to transitivization as verbs of existence. One possible reason for this result may be the potential effect of the number and selection of verbs of directed change. There were only two of them included in the judgment test, namely, die and arrive, for which there are commonly used transitive counterparts, kill and bring. So, the selection of these two verbs may have inhibited the subjects from accepting the transitivization as correct English. If so, similar tests that contain other verbs of directed change may produce different kinds of result. Therefore, the status of directed change verbs needs further investigation with respect to the transitivization of unaccusative verbs in L2 English. 5.3.5. Raising Verbs In scoring the responses to the four structures in which raising verbs were presented, both the NP-V-to infinitive and it-V-that clause structures were considered to be "correct English" while the NP-be+Ven-to infinitive and it- be+Ven-that clause structures were considered to be "incorrect English." The first two structures were treated as one group and compared to the other two structures. Therefore, the highest possible score was 4. The descriptive statistics of the results are given in Tables 5.14.1 to 5.14.3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 281 T able 5.14.1 Raising: NP V to infJIt V that... (expected = acceptance, max. score = 4.0) Mean Standard Deviation Number of Subjects English 2.875 .850 24 Italian 2.594 1.151 64 Japanese 2.952 .972 168 Table 5.14.2 Raising: NP-be+Ven to inf. (expected= rejection, max. score = 4.0) Mean Standard Deviation Number of Subjects English 3.667 .761 24 Italian 3.094 1.377 64 Japanese 2.667 1.558 168 Table 5.14.3 Raising: It be+Ven that... (expected = rejection, max. score = 4.0) Mean Standard Deviation Number of Subjects English 2.083 1.613 24 Italian 2.656 1.472 64 Japanese 2.119 1.527 168 ANOVA showed no significant effect of language (F (2, 253) = 2.13, p<. 121). However, significant effects were found of structure (F (2, 506) = 16.30, p<.0005) and of the interaction of language and structure (F (4, 506) = 5.54, p<.0005). As for the scores on the two presumably correct English structures, no significant between subjects effects were revealed: F (2, 253) = 2.9312, p<.0551, indicating that, as tacitly assumed, the subjects'judgment on these two structures were not very different among the three language groups. By contrast, ANOVA revealed significant differences among the language groups on their scores of the NP-be+Ven-to infinitive structure: F (2, 253) = 5.9665, p<.0029. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 282 Table 5.15 Results on NP-be-^Ven-to infinitive Structure Source D.F. SS MS F ratio F probability Between Groups Within Groups Total 2 253 255 25.3802 538.1042 563.4844 12.6901 2.1269 5.9665 .0029 Scheffé's test found that the native English group's score (M = 3.667) was significantly better (p<.05) than that of the Japanese (M = 2.667). The result, therefore, supported Hypothesis 19B that Japanese speakers' rejection of this structure would be less reliable than native English speakers'. Although the Italians' score (Af = 3.094) was lower than the English speakers', the difference did not turn out to be significant. Therefore, Hypothesis 19A on Italian speakers' judgment was not statistically supported. Hypotheses 20A and 2GB were not supported simply because both Italian and Japanese subjects scored better on the NP-be+Ven-to infinitive structure than on the it -be+Ven-that clause structure. In the case of the Japanese subjects, the difference of the scores (Af = 2.667 for the former structure and Af = 2.119 for the latter) was even found to be statistically significant: t = 3.80, d f— 167, p<.0(X)5. In other words, the results of the comparison between the NP-be-^Ven-to infinitive structure and the it -be+Ven-that clause structure turned out to be totally opposite to what was predicted by Hypothesis 20. The reason for this unexpected result is not clear, but what is easily noticed is the generally low scores obtained by the three language groups in rejecting the ungrammatical it -be+Ven-that clause structure. Especially, the very low mean score of the native English speakers (Af = 2.083), which was in fact lower than Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 283 those of the Italians (M = 2.656) and the Japanese (Af = 2.119), is quite remarkable. The English group also had the largest standard deviation, suggesting wide disagreement on this structure. Recall that the logic behind Hypotheses 19 and 20 was the possibility for the nonnative speakers to overgeneralize the English passive morphosyntax as a marker of NP-movement beyond a clause boundary. Since the raising structure with the expletive it lacks NP-movement of this kind, the ungrammatical it - be+Ven-that clause structure was predicted to be easier to reject than the NP- be+Ven-to infinitive structure. Nevertheless, their low score on the it -be+Ven-that clause structure cannot be attributed to a factor such as the overgeneralization of the passive morphosyntax to indicate NP-movement among nonnative speakers because the fact that native English speakers do not make such an overgeneralization is evidenced by, for example, their very high score on the NP- be+Ven-to infinitive structure (Af = 3.667). Why, then, were the mean scores on the ungrammatical it -be+Ven-that clause structure so low, even including the one for the native speakers? The answer to this question is not clear at all, but it appears to me that this issue needs to be investigated in a larger framework which involves, among other factors, the structure of sentences such as: (22) a. It was believed that she was the best person for the job. b . It was expected that most guests would come to the party. In other words, possible confusion with grammatical passive sentences such as (22) and the ungrammatical it -be+Ven-that clause structure could be one potential cause for the unexpectedly poor performance on this structure. In any case, more Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 284 research needs to be done if the tendency to fail to reject the ungrammatical it - be+Ven-that clause structure is repeatedly shown to be real not only among normative speakers but even among native speakers. As for the predicted tendency to accept the ungrammatical NP-be+Ven-to infinitive structure, the judgment test statistically supported this phenomenon among the Japanese subjects in comparison to the native English speaker subjects. Its existence among the Italian subjects, however, was only suggested by their relatively low mean score. In other words, the "strong (or extended) overgeneralization" of the passive morphosyntax hypothesized in section 5.1.1.5 was statistically confirmed among the Japanese subjects. The question concerning the "strong (or extended) versus weak (or limited) overgeneralization" may be studied in future research in connection with LI characteristics as well as other factors. 5.3.6. Proficiency Levels and U-shaped Development The last important prediction made by the three-stage account concerned a potential U-shaped pattern of development with respect to the ungrammatical structures and avoidance phenomena involving unaccusative-inchoative verbs. The U-shaped developmental pattern was expected to emerge if nonnative speakers of English progress through the predicted stages of reorganization of the IL lexicon and syntax in the development of interlanguage grammar. The present study attempted to correlate the proficiency levels of the subjects estimated by their cloze test scores and scores on the non-target experimental items in the grammaticality judgment test. A preliminary analysis was conducted by plotting the cloze test scores on Abscissa and the mean scores on non-target structures on Ordinate. Graphs were Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 285 drawn for each language group, showing how the subjects performed on the NP- be+Ven structure with unaccusative verbs, the NP-V structure with unaccusatives, and on the NP-be+Ven-to infinitive structure. In addition, the Italian subjects’ responses to each of the three postverbal NP structures were plotted in this manner. However, no clear pattern of U-shape development emerged from the preliminary analyses. In general, non-target responses were found all along the proficiency scale measured by the cloze test. Follow-up analyses, therefore, were forfeited. I suspect two possible reasons for this rather disappointing result. One potential cause is the general characteristic of the experimental subject population. As stated in section 5.2.1, all the subjects in the current study were undergraduate or graduate students who had been studying English as a second language for some time. Therefore, at the time of testing, most subjects had a fairly long experience learning English as a second language, which was attested to by the 8.9 years of average length of study. University students were selected for the study because of the length of the judgment test, which appeared to require relatively high proficiency, particularly reading ability, on the part of experimental subjects. The use of the cloze test was intended to separate these long-time English learners along the proficiency scale. Despite the well-distributed scores of the cloze test, however, the experimental subject population may have represented a more or less homogeneous group in crucial respects. Recall that within the three-stage account, non-target structures and phenomena are expected to emerge when the learner's interlanguage grammar is at the second stage of development where the A-Structures of unaccusative-inchoative verbs are correctly represented but the proper syntactic derivations are not acquired Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 286 yet. Neither before nor after the second stage, are the non-target characteristics of the interlanguage grammar expected to be observable. Therefore, the crucial question is when learners come to and move out of the second stage. In section 4.5, particularly 4.5.4,1 have made an implicit association between the levels of the developmental model and the learner's proficiency. The assumption behind plotting cloze scores and subjects' responses was that within the range of proficiency measured by the cloze test we would wimess the critical transitions from one stage to another. It is very likely, however, that many subjects in the study, irrespective of their cloze scores, had already crossed the threshold for the second stage and were equally susceptible to the "unaccusative trap" phenomena. It is in this sense that the subject population as a whole might have been homogeneous enough to hide any U-shaped pattern in their individual grammatical development. In other words, the selection of subjects in the study may have been too narrowly focused for the purpose of investigating the developmental pattern in question. The idea of finding a clear U-shaped pattern within the all college student population of the subjects may have been doomed from the start. Another potential reason for the lack of U-shaped patterns in the judgment data is individual differences with respect to the onset timing and speed of transition from one stage to another in the three-stage developmental route. It is quite possible that the individual pace of development may naturally vary within some general range of proficiency. That is, as long as learners have already reached a certain level of familiarity with the target language, exactly when they move from the first to the second stage and how long they remain at the second stage before they finally advance to the third may be very different from one learner to another. What is important, then, may not be to associate a stage of development with a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 287 fixed proficiency level (which may be impossible to do after all), but to try to see if there is a certain sequence of developmental stages relevant to individual L2 learners. In order to study the predicted sequence of developmental stages, longitudinal research on a group of subjects over an extended period of their L2 acquisition may be the best method of investigation. However, further granunaticality judgment tests with a few changes in research design may also prove to be useful. First, only a selected number of phenomena such as "passivization" and avoidance should be included in one study. Second, instead of presenting an experimental item in a longer context, a translation task such as the one used in Kellerman's study may be utilized. Third, selection of subjects should be done over a wider range of L2 English learners including, for instance, those at lower and higher secondary schools, at college, and at the near-native stage. In sum, research on developmental patterns involving unaccusative-inchoative verbs need to be conducted on a heterogeneous group of subjects with various degrees of familiarity with and proficiency of English as a second language, by focusing on a few non-target structures or phenomena. The difficulties of language and experimental task should be conducive to the ability of lower-proficiency subjects as well. If these measures are taken, the predicted U-shaped developmental pattem may be found as in the case of Kellerman's studies. 5.4. C o n c l u d in g R e m a r k s This chapter presented analyses of a grammaticality judgment test conducted on native speakers of Italian, Japanese and English. The statistical analyses investigated the "unaccusative trap" account with respect to (a) the performance Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 288 variation on different verb classes or structures within a language group and (b) the contrast between nonnative and native judgments. It is not possible to summarize the results of the analyses in a nutshell because of the large number of topics dealt with in the study. Instead, I focus on a general picture of the L2 acquisition of English intransitive verbs that has emerged from the study and the issues that may prove important for future research. In general, unaccusative verbs turned out, as expected, to show a clear contrast with unergative verbs in the judgment of nonnative speakers. For instance, within both Italian and Japanese subject groups, the "passive" unaccusatives were less easily identified as incorrect than the "passivized" unergatives. Similarly, the avoidance phenomenon and transitivization were more common with unaccusative verbs than with unergatives in both nonnative speaker groups. Even when the contrasts between the native and nonnative judgment on unaccusatives did not turn out to be statistically significant, the disparities in the mean scores usually indicated that the suspected differences existed at least as a tendency. The study also found that appearance and disappearance verbs were more susceptible to transitivization than existence verbs, indicating that not all unaccusatives may be equally transitivized in interlanguage grammar. The predicted tendency that native Italian speakers accept postverbal NP structures with a covert or overt expletive subject also received support firom the judgment data. One finding that may prove very important for future research is about potential roles that unaccusative and inchoative verbs may play in reorganizing the interlanguage grammar at the two transitional phases of the three-stage model of development that I proposed in Chapter 4. The statistically significant results obtained from the judgment test indicate that the non-target structure of "passivized" Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 289 unaccusative and the avoidance of NP-V word order are more likely to occur with unaccusative verbs than with inchoative verbs. If the input-based and the UG- based predictions developed in the previous chapter are correct, the results of the judgment test clearly point to the lexical component of UG (possibly Sorace's Unaccusativity Hierarchy, which is not directly observable in the target language input) as the primary driving force behind the reorganization of the interlanguage lexicon at the transition from the first to the second stage of development. That is, the second stage may be appropriately characterized as "UG-driven" in contrast to the "input-driven" first stage. A quite surprising discovery was the Japanese subjects' responses to the 0- V-NP structure. The Japanese speakers were expected to behave similarly to the native English subjects in judging this structure as ungrammatical but were found to accept it significantly more often than the native speakers did. Although the Japanese score was still significantly higher than the Italians, indicating the strong Italian propensity for this postverbal NP structure, the reason for the unexpected Japanese responses is not clear at the moment. Another surprising finding was made concerning the transitivizability of directed change verbs. This subclass of unaccusative verbs was predicted to undergo transitivization as much as appearance and disappearance verbs in interlanguage grammar, but was found to resist transitivization, patterning more like existence verbs. Quite interesting as they are, however, the results of this study need more corroborating evidence, especially where there has been no previous research on particular topics, for example, the relative difficulty order of unaccusatives and inchoatives. The study also had some inherent weaknesses such as the small Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 290 number of native English speakers and the limited selection of verbs that supposedly represented verb classes or subclasses. Still, this grammaticality judgment study is the first systematic attempt to investigate the L2 acquisition of English intransitive verbs from a perspective which suggests that it is shaped by a complex interaction of factors such as LI property, structural pattem, verb class, L2 proficiency, developmental stage, etc. The most important aspect of the study, in a sense, is the very fact that this new perspective led to the formulation of clear testable hypotheses as well as to an understanding of how each hypothesis fared against the judgment data. Needless to say, more research is needed to test these and other hypotheses for a better understanding of the acquisition of intransitive verbs in L2 contexts. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 291 6. The "Unaccusative T rap" and Related Issues One of the major proposals I have made in this thesis is that the distinction between unergatives and unaccusative-inchoatives is not properly acquired at the beginning of L2 acquisition of English, or more specifically, that many L2 learners start out assuming that unaccusative verbs are the same as unergative verbs as far as their lexico-syntactic properties are concerned. I have also argued that for the majority of learners this state of affairs is likely to continue over a long period of time. This analysis, which I have called the "unaccusative trap" hypothesis, is intended to answer some fundamental questions concerning the acquisition of English intransitive verbs in L2 contexts in a coherent way. Among them, the two most important are: (1) a. Why do L2 learners of various LI backgrounds produce and/or accept non-target syntactic structures with English unaccusative-inchoative verbs while they do not show the same tendency with unergative verbs? b . Why are such non-target syntactic structures with unaccusative- inchoative verbs not produced by L2 learners of English more widely? I have suggested that these questions can be accounted for rather elegantly if we see the development of the relevant aspects of interlanguage grarrunar as a three- stage process. At the initial stage, L2 learners come to the wrong Argument Structure representation for unaccusative-inchoative verbs via a non-target (and possibly "unnatural") linking mle. The Single Argument Linking Rule ensures a straightforward derivation that matches the dominant surface word order of NP-V, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 292 in which most unaccusative-inchoative verbs (as well as unergative verbs) appear in the target language input. The "passive" unaccusative-inchoatives, the postverbal NP structures, etc. are most likely to be consequences of reorganization of the interlanguage grammar, especially of its lexical component, when the learners move out of the first stage to the second. This reorganization takes place when relatively advanced learners who become sensitive to the internal consistency of the target language lexicon abandon the Single Argument Linking Rule for more natural linking rules, possibly such as those suggested by Levin and Rappaport Hovav (1995). The irony of the "unaccusative trap" is that it is mostly these relatively advanced learners that produce and/or accept non-target sentences while the majority of learners who have not reached the advanced state of grammatical awareness do not show a sign of abnormality. Still, it is predicted that these less advanced learners will be susceptible to the same types of error when, or even if, their interlanguage grammar undergoes the reorganization process. Presumably, the learners overcome the "unaccusative trap" when they unlearn non-target syntactic operations such as "passivization" of unaccusative-inchoatives and move to the third (and final) stage. Considering the difficulties inherent in the transition from one stage to another, however, the learners who successfully achieve the native-like grammatical representations of unaccusative-inchoative sentences at both lexical and syntactic levels are likely to be a very small minority of all the learners of English who come to learn this language in adulthood. Although it is based solely on L2 English data, the "unaccusative trap" hypothesis appears to be relevant to issues beyond the L2 acquisition of English. In this chapter, I discuss two areas of potential significance for future research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 293 namely, the status of the "unaccusative trap" in L2 acquisition of languages other than English and potential contrasts between LI and L2 acquisition from the perspective of this hypothesis. 6.1. THE "U n a c c u s a t iv e T r a p " d s t L2 J a p a n e s e The first topic of our concern is the importance of the "unaccusative trap" hypothesis in acquisition of languages other than English in L2 contexts. I present here, as an example, the mastery of the Case-marking system in Japanese by nonnative speakers (Inoue 1991 and Kageyama 1993).^'* First, observe the examples in (2) where the Accusative particle o is omitted, and (3) where the Nominative particle ga is omitted. (2) a. Akihabara de teepurekoodaa (o) kau toki.. . Akihabara LOC tape recorder (ACC) buy when 'When (you) buy a tape recorder in Akihabara . .( 3 3 6 0 America) b. Yan-san wa .. . syasin (o) misete ageta. Ms. Yan TOP photo (ACC) showed 'Ms. Yan showed the photo (to someone).' (3733 Nigeria) c. Watasi wa puroguramu (o) sosiki simasita. I TOP program (ACC) organization did 'I organized a program.' (3741 Australia) The data are taken from Inoue 1991 and Kageyama 1993, but originally come from Teramura 1990, which I have not yet been able to obtain. The filing numbers and learners' countries o f origin in parentheses are Teramura’ s. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 294 (3) a. mosi jikan (ga) areba. . . if time (NOM) exist 'If (you) have time . . . ' (3287 Taiwan) b. Kono mati no soba-ni takusan mod (ga) arimasu this town GEN near a lot forests (NOM ) exist 'There are a lot of forests near this town.' (3367 Indonesia) c. Gogo wa okyakusama (ga) utie kite... afternoon TOP guests (NOM) house to come In the afternoon, the guests came to my house and .. .' (3376 Hong Kong) d. Syukudai (ga) takusan aru k a ra . . . homework (NOM) a lot exist because 'Because (we) have a lot of homework . . .' (3436 Malaysia) As discussed in section 2.4.5, in the target Japanese the omission of ga and o is often permitted when the NP argument is the sister of V. The typical situation of Case Marker Drop is observed when the NP is the object of a transitive verb or the subject of an unaccusative-inchoative verb as in (2) and (3) respectively. These examples by L2 learners of Japanese, therefore, cannot be characterized as "grammatical errors," although they may sound slightly odd depending on the level of formality, intonation, etc. Instead, (2) and (3) indicate that the learners who produced them were subconsciously aware of the structural condition of Case Marker Drop. Moreover, the examples of (3) in particular suggest that the learners 'knew' that the surface subject of an unaccusative-inchoative verb is an internal argument in its A-Structure, which is syntactically projected as the sister of V. In Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 295 fact, noting that omission of ga was not found with NP arguments of unergative verbs, Kageyama (1993:63) argues that the L2 learners of Japanese who produce sentences such as (3) and (4) correctly distinguish unaccusative-inchoatives from unergatives. As in L2 acquisition of English, however, the correct A-Structure representation for unaccusative-inchoatives appears to lead some learners to clear syntactic errors. Observe, for instance, the following examples in which the Accusative particle o is erroneously substituted for the Nominative particle ga : (4) a. Daigaku sotugyoosita kara taihen nagaku jikan *o tatte imasita college graduated since very long time ACC had passed 'A long time had passed since I graduated from college.' (3290 Taiwan) b . Nihonjin tyuusin no bunka *o okorimasita Japanese centered GEN culture ACC arose There arose a Japanese-centered culture.' (3413 Hong Kong) c. Hutari no seikatu no naka de iroirona mondai *o detekuru . . . two people GEN life GEN inside LOC many problems ACC come out 'Many problems arise in the life of the couple ....' (3486 Indonesia) d. Kaze no densenbyoo *o ryuukoo suru toki . . . flu GEN epidemic ACC spread out when "When influenza spreads . . .' (35C X ) Iran) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 296 e. Nihon wa . . . jinkoo *o syuutyuusiteiru kuni desu. Japan TOP population ACC concentrate country CPL Japan is a country where its population is concentrated (in big cities).' (3579 China) The verbs in (4) are all unaccusative-inchoative verbs. Kageyama (1993:64) observes that this type of error was limited to this subclass of intransitive verbs because no example was found in which the argument of an unergative verb was erroneously marked with o. Now, in theory, the errors of this kind should not be found in the target language input to which the learners are exposed. The question, therefore, is where they come from. The answer I suggest is that they are ultimately due to the correct A- Structure representation of unaccusative-inchoative verbs. When the argument of this subclass of intransitive verbs is correctly projected as a D-Structure object, the NP may be incorrectly Case-marked with the Accusative o in analogy with the object of a transitive verb. Of course, the Case marking of the NP in the sister of V position can be affected in another way as well. That is, the object of a transitive verb may be Case-marked with the Nominative particle ga, not with the Accusative particle o. The examples in (5) show that such errors are actually found in L2 Japanese. (5) a. Oiwai no tegami *ga mimasita. celebration GEN letter NOM saw 'I saw the congratulatory letter.' (3291 Taiwan) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 297 b. Yakyuu no siai *ga yoku okonaimasu baseball GEN game NOM often do 'We often play baseball.' c. Nihonjin wa en *ga tukaimasu Japanese TOP yen NOM use Japanese use yen.’ d. Watasi wa . . . gakkoo de I TOP school Loo I learn many things at school.' (3297 Taiwan) (3357 Brazil) takusan no mono *ga naraimasu alot GEN thing NOM learn (3417 Malaysia) e. Ima A-sensei ga . . . no mondai *ga kenkyuu siteiru. now Prof. A NOM GEN problem NOM study doing 'Right now. Prof. A is studying a problem o f . . . .' (3528 Finland) In (5), the Nominative Case marking of the argument of an unaccusative-inchoative verb within V seems to be wrongly extended to the objects of the transitive verbs. Unfortunately, neither Inoue (1991) nor Kageyama (1993) discusses the proficiency levels of the L2 leamers of Japanese whose examples they cite. The examples themselves, however, suggest that the leamers were at least beyond the elementary stages of acquisition and in some cases were quite advanced. According to the three-stage developmental model of intransitive verb acquisition in L2 contexts, therefore, I would predict that these errors are more commonly and systematically observed among relatively advanced L2 leamers of Japanese, who have correctly reanalyzed their initial misrepresentation of unaccusative-inchoatives and project the intemal argument as the sister of V. In other words, I believe that these leamers are at the second stage of the "unaccusative trap" where the correct A- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 298 Structure representation presents syntactic challenges. In L2 English, NP movement is in the core of such a challenge while in L2 Japanese the trouble seems to center around Case marking of arguments. This analysis for the data in (4) and (5) as well as (2) and (3) accounts for why the surface subjects of unergative verbs are not incorrectly Case-marked with the Accusative particle o nor do they appear with the Nominative particle ga incorrectly omitted. Unlike unaccusative-inchoative verbs, unergative verbs do not undergo the lexical and structural reanalysis and their arguments remain as external arguments throughout the interlanguage development. Consequently, they do not present any major c h a l l e n g e s . ^ ^ It goes without saying that much more intensive research is needed in order to understand the phenomena related to the L2 acquisition of Japanese intransitive verbs. The data that I have discussed in this section are very limited. As Inoue (1991) and Kageyama (1993) noticed, however, these errors are not random but quite systematic. The acquisition of Case-marking in Japanese, if investigated seriously, may serve as an indication that the "unaccusative trap" hypothesis is relevant to L2 acquisition of languages other than English. 95 Inoue (1991:36) presents examples o f sentences by patients of agrammatism. Their number is limited, but we still find types of Case marking errors that are not presented with unaccusative-inchoative verbs either in the same work by Inoue or in Kageyama (1993). Observe the three examples below. (i) a. Ojiisan *o yama *ga siba (o) katte .... old man ACC mountain NOM brushwood (ACC) cut T he old man cut brushwood in mountains and . . . .' b. Hutari kodomo *o otosita. two children ACC drop (causative) The two chidden dropped (something).' c. Neko *o inemuri . . . cat ACC sleep T he cat sleeps . ..' In (ia-b), the subjects of transitive verbs are wrongly marked with the Accusative o. In (ic), the subject of an unergative verb is marked with o instead of ga. In other words, the external arguments are erroneously marked with the Accusative particle in these examples. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 299 6.2. POTENTIAL CONTRASTS BETWEEN LI AND L2 ACQUISITION 6.2.1. Verbal Classification and Initial Syntactic Representation 6.2.1.1. LI acquisition Hyams (1986), citing Brown 1973, noted that children rarely make errors in the word order of grammatical elements such as subject, object and verb. Subsequent studies have found that even when they do make such errors, the word orders actually correspond to the underlying basic structures or some transitional structures that emerge as the children gradually acquire the abstract properties of the target language (e.g. Clahsen and Muysken 1986, 1989, Clahsen 1988, Radford 1990, 1996, Pierce 1992, Déprez and Pierce 1993, Koppe 1994). In fact, as soon as children move beyond a single word stage, they seem to be able to classify words into lexical categories and properly project each lexical category to a corresponding phrasal category by setting the values of linearity parameters such as head parameter and specifier parameter (Radford 1990:chap. 2). In English and French, for example, the head parameter and the specifier parameter are set to "head-final" and "specifier-first" respectively. Consequently, the intemal argument at A-Structure follows the verb as its complement at D- Structure while the external argument precedes it as its specifier. The result is the underlying word order of SV(0) in these languages. Remarkably, the majority of the declarative VS word order produced by English-speaking children, for example, 90% of such cases in Déprez and Pierce 1993:43, are unaccusative verbs with postverbal subjects. A similar fact concerning the acquisition of French word order is pointed out by Koppe ( 1994), who notes that child French data show that subjects of unaccusative verbs occur more often in a postverbal position than subjects of other verb classes. What is important in these studies is that although Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 300 the surface word order of VS at first appears to be an error in word order, when analyzed with respect to verbal classes, the observed VS order with unaccusative verbs in fact correctly reflects the underlying position of the argument of these verbs in both English and French. These observations of early child English and French, therefore, indicate that children are indeed able to differentiate unaccusatives from unergatives and correctly generate the sole NP argument of the former in a postverbal object position. Otsu (1994) and Miyata (1992 cited in Otsu) present evidence that children acquiring Japanese as their mother tongue also show similar sensitivity to verbal classification and initial syntactic representation. Japanese, an S(0)V language, exhibits relatively free surface word order due to scrambling. Despite these syntactic characteristics and the phenomenon of Case Marker Drop (CMD) discussed in section 2.4.5, LI Japanese children appear to acquire verbal classificaüon and initial syntactic representation quite successfully at an early stage of acquisition. Otsu (1994), for example, has investigated the production and comprehension of utterances by 3- and 4-year-olds with respect to CMD and concluded that children as young as 3 years of age show a clear sensitivity to the hierarchical structure of Japanese sentences by correctly producing and accepting NPs without a Case particle only when they are adjacent to and c-commanded by the head verbs (Saito 1983, Takezawa 1987, Kageyama 1993). Miyata (1992 cited in Otsu 1994) has studied CMD in early child Japanese by analyzing 326 utterances with subjects produced by 2- and 3-year-olds. She has found that the 47 utterances which lacked the Nominative particle ga mostly contained either stative or "ergative" verbs with "theme" subjects. Miyata's study, therefore, shows that in early child Japanese not only surface objects of transitive verbs but also subjects of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 301 unaccusative-inchoative verbs appear without a Case particle. This finding in turn indicates that the intemal arguments of both transitive and unaccusative-inchoative verbs are correctly projected in the sister of V position by Japanese children as young as 2 years old. In LI acquisition of German, an underlyingly S(0)V language, it is widely observed that children conform to the underlying word order at an early stage of acquisition, despite the fact that the underlying word order is not always obvious in the surface syntax (e.g. Haegeman 1991). Although estimates vary, approximately 60 to 70% of all the verbs are reported to appear in sentence-final position at an early stage of syntactic development in child German (Clahsen and Muysken 1986, 1989, Clahsen 1988, Mills in press, cited in Clahsen and Muysken 1986). In sum, in all the aforementioned cases of LI acquisition, young children exhibit a remarkable ability not only to properly classify verbs but also to project their arguments in the correct D-Structure positions even when the underlying syntactic order of grammatical elements is not identical to their surface order. 6.2.1.2. L2 acquisition In the acquisition of word order, adult L2 English shows a superficial similarity to early child English: the analysis of the LLC data shown in section 4.3 has revealed that the postverbal NP word order was limited only to unaccusative verbs and absolutely no instance of V-NP word order was found with the 10 preselected unergative verbs. This similarity, however, conceals a potentially significant difference between LI and L2 acquisition of English. For one, almost all the observed V-NP word order was produced by native speakers of Italian and Spanish and the sentences produced by Korean and Japanese speakers rarely contained Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 302 postverbal NP structures. Moreover, there are some indications that unlike early child English, the postverbal NP structure with unaccusative verbs appears in the language of relatively advanced leamers. Accordingly, it has been hypothesized that unaccusative-inchoative verbs are not syntactically differentiated from unergative verbs at early stages of L2 acquisition and the correct A-Structure and D- Structure representations for unaccusative verbs are acquired only at a relatively advanced stage. In acquisition of Japanese, the surface word order of intransitive sentences reveals nothing about the learner's grammar because it is invariably NP-V. However, the data on the "Case confusion " phenomena involving unaccusative- inchoative verbs (section 6.1) strongly suggest that the "unaccusative trap" account is also valid in the context of L2 Japanese. Here again, adult L2 acquisition seems to differ from child LI acquisition in the timing of the correct representation of the lexico-syntactic and syntactic properties of verbs. The acquisition of the basic word orders in German and Dutch (both underlyingly S(0) V languages) by L2 leamers also reveals similar contrasts between LI and L2 acquisition. Clahsen (1988) and Clahsen and Muysken (1986, 1989) report that both native speakers of Romance languages (with the underlying SV(0) stmcture) and those of Turkish (with the underlying S(0)V order) start out with the assumption that German is an SV(0) language. That adult Turkish speakers reveal the same tendency with respect to the acquisition of L2 Dutch is also observed by Dittmar (1981 cited in Clahsen and Muysken 1986). In short, the nature of acquisition of L2 syntax at its early stages appears to be closely affected or even defined by the dominant word orders in the target language input that the leamers receive. In this sense, adult L2 leamers reveal a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 303 clear contrast with children acquiring their Lis who are sensitive to the underlying basic structures of their mother tongue from the start of their grammatical development. 6.2.2. Subsequent Development of Word Order It is not just the initial assumption made about the word order that is different between the first and second language acquisition. The subsequent syntactic development also appears very different in the two contexts. 6.2.2.1. LI acquisition In the child LI acquisition of languages with subject-verb agreement, the subsequent development of word order appears to be a result of gradual acquisition of the properties of functional categories. For instance, Déprez and Pierce (1993) illustrate how the parametric value of V-raising is instantiated as the child learns finite verb forms and how this particular parameter may be set very early in child language. They show that in child French, V-to-I movement is acquired as the child masters finite verb forms, while children acquiring English never wrongly raise V to I. The ultimate position of the single argument of an unaccusative verb is also determined by the child's acquisition of functional categories such as I and C. Koppe (1994), for instance, observes that NP arguments of French unaccusative verbs move from a postverbal to a preverbal position, following the acquisition of finite verb forms. Koppe argues that the very reason for this movement is to satisfy an agreement requirement of some inflectional category above VP. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 304 Clahsen (1988) and Clahsen and Muysken (1986, 1989) have found children acquiring German, who start out with sentence-final verb position in the majority of their utterances, begin to produce verb-second utterances as they acquire finite verb forms. Clahsen and Muysken also note that in this process of word- order shift, the emergence of the inflectional formative st for second person singular marks the complete mastery of the agreement paradigm. Once the paradigm is mastered, the change of word order is swift. Clahsen and Muysken claim that within a month the frequency of verb-second patterns increases from 40% to 90% of the children’ s utterances. In other words, the acquisition of the verb-fronting rule clearly correlates with the attainment of the subject-verb agreement system. Moreover, finite verbs appear in the correct clause-final position as soon as the first embedded clauses are used by the children. Clahsen and Muysken found no errors on verb-final patterns in embedded clauses in their data. This result gives more support to the theory that the development of word order in German correlates with the acquisition of functional categories because German-speaking children, who have already internalized the crucial V-to-C movement and "know" that the (S)VO order in the main clause is a result of this movement, would also "know" that the landing site for an inflected verb, i.e. Comp position, is already occupied in an embedded clause. Mills' (in press, cited in Clahsen and Muysken 1986) survey of diary studies also corroborates Clahsen and Muysken’ s claim. She shows that the verb- final utterance, which is dominant for an extended period after the initial stage of syntactic development, is later replaced by the verb-second pattem around the age of three. As in Clahsen and Muysken (1986), Miller also observes that errors on the grammatical SOV word order in subordinate clauses are practically non-existent Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 305 in child German. Similarly, Déprez and Pierce (1993) argue that in child Swedish as well as in child German the initial OV word order loses its dominance when the child acquires V-to-C movement, following the acquisition of V-to-I movement. In sum, in the languages that exhibit subject-verb agreement, the development of word order in child language strongly correlates with the mastery of functional categories such as Infl and Comp. Ô.2.2.2. L2 acquisition Clahsen (1988) and Clahsen and Muysken (1986, 1989) observe that unlike children acquiring German as their mother tongue, L2 learners go through stages of more or less unrelated rules before they come to the terminal state of grammar. For example, following the initial SVO assumption, the learners appear to go through several phases of development at which they typically employ various ad hoc rules such as "adverbial preposing," "extraposition of subject," "adverb placement between the object and the finite verb," and so forth The notion of agreement, especially with the focus on the "number" rather than on the "person and number", appears to be present from the beginning but the full mastery of the verbal inflection poses a great challenge to many leamers. Rapid change in dominant word order and a particular inflectional formative that marks the complete mastery of the inflection paradigm in child German are not found. There is no apparent relation between the acquisition of the subject-verb agreement paradigm and the observed word order patterns. Also, in L2 German, ungrammatical word orders with a verb in a position other than clause-final are often produced in subordinate clauses. In L2 English, unaccusative-inchoative verbs do not seem to cause any formidable challenges at early stages of acquisition. Most of this class of verbs Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 306 appear in the NP-V word order as unergatives, and the declarative VS word order on unaccusatives is produced only by the leamers of particular LI backgrounds at an advanced stage of acquisition. At this stage of interlanguage development, the unaccusative-inchoatives may also be avoided in the canonical NP-V word order or appear in non-target surface structures such as the "passive" construction. In conclusion, unlike child LI acquisition, the developmental changes of word order in adult L2 acquisition seem unconnected to mastery of functional categories. 6.2.3. Status of the "Unaccusative Trap" in LI Acquisition One question that comes to mind at this juncture may be whether the "unaccusative trap" exists in LI as well as in L2 acquisition. Although I have yet to conduct research on this question, the answer is probably negative. To my knowledge, structures such as "passivized" unaccusative-inchoatives, for example, are not commonly observed in child LI English. In fact, when the differences between LI and L2 acquisition reviewed above are taken into consideration, the lack of "unaccusative trap" phenomena in LI acquisition is what is expected. First, studies of various LI acquisition (e.g. Radford 1990, Miyata 1992 cited in Otsu, Pierce 1992, Déprez and Pierce 1993, Koppe 1994 and Otsu 1994) strongly suggest that unergatives and unaccusative-inchoatives are correctly represented in the early LI grammar. In fact, a recent work by Snyder and Stroms wold (1997: footnote 18) argues specifically against a view that unaccusative-inchoatives are given an erroneous unergative analysis and only later correctly reanalyzed (see, e.g. Borer and Wexler 1986). Snyder and Stromswold Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 307 say that this alternative possibility raises substantial theoretical concerns.^® Instead, the child acquiring her first language somehow figures out natural linking rules to arrive at the correct A-Structure representations and sets the values of linearity parameters for the right initial D-Structure representations. Since the "unaccusative trap" begins with the misanalysis of unaccusative-inchoatives as unergatives, the child appears to be immune to the problem from the outset of acquisition. But how does she avoid erroneously adopting the Simple Argument Linking Rule, which after all is a very simple way to generate structures that superficially correspond to dominant surface word orders in the input? A clue to the answer may be found in the following statement by Chomsky on the critical relation between thought and language; The child approaches language with an intuitive understanding of such concepts as physical object, human intention, volition, causation, goal, and so on. These constitute a framework for thought and language and are common to the languages of the world [___ ] (Chomsky 1988:32) Chomsky's statement strongly suggests that LI acquisition of linking rules must be considered in its relation to the general cognitive development of the child. Since linking rules function as bridges between lexical semantics and lexical syntax, both the concepts such as "entity," "state," "change," "causation," etc. and the hierarchical nature of Argument Structure have to be in place when the formation of linking rules begins in the mind of the language acquirer. In other words, both the cognitive and the linguistic primitives must be already available for matching. For arguments against Borer and Wexler 1986, see Koppe 1994 and the studies cited there. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 308 because with either of them missing, it would be impossible for the child to acquire the linking rules as part of her linguistic system. These conceptual and linguistic primitives must also be available to the adult L2 learner since without them he could not have learned his first language. What, then, makes it so difficult for him to acquire the correct target language linking rules in L2 acquisition while the child acquires them effortlessly? I believe that the difference results from the relative impact of input on the formation of linking rules in LI and L2 acquisition. For L2 acquisition, I have already argued that the available input, especially, prominent surface word orders and the number of obligatory NP arguments that a verb requires, has a direct influence on the way verbs are categorized. In LI, however, the direct impact of input on acquisition of the linking rules appears to be far less significant. Chomsky continues on the acquisition of vocabulary: [...] though words may not match precisely across languages, the conceptual fimnework in which they find their place is a common human property, [...] it is beyond question that acquisition of vocabulary is guided by a rich and invariant conceptual system, which is prior to any experience. (Chomsky 1988:32) If Chomsky is correct in his claim that vocabulary acquisition is internally guided, the formation of linking rules may also be internally driven. That is, linking mles may naturally "grow", so to speak, in the mind of the child on the available conceptual and linguistic primitives. What the mind does for linking semantics and syntax is not so much to seek discernible patterns in the available input as to match the concepts with the limited number of linguistic primitives such as categorial features (e.g. [+/- N] and [+/- V]), structural notions (e.g. external versus internal arguments) and so on. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 309 Second, the typical L2 error of the overgeneraiization of the passive morphosyntax as an overt marker of NP movement is unlikely to be made in early LI acquisition. According to Snyder and Stromswold (1997) who studied the utterances of 12 English-speaking children, unaccusatives are acquired on average 3.7 months earlier than passives. This finding seems to make intuitive sense. If the movement of NP to surface subject position is motivated by the agreement requirement of the strong N-features of Tense, NP arguments of unaccusatives may move when the functional category of Tense is acquired by the child. On the other hand, passive sentences involve two additional processes. First, the child has to learn that the morphological process of passivization, that is sufiixation of the passive morpheme -en, "absorbs" or "suppresses" the theta role normally assigned to the external argument of the base verb. Next, she must learn that the auxiliary verb be needs to be inserted in syntax as a Tense-bearing element since the passive participle is nonfinite and does not carry Tense. Therefore, each component of the passive morphosyntax of English, be + Ven, is clearly associated with a distinct function. Since the movement of NP is internally motivated by the agreement requirement, the LI child, who acquires unaccusatives before passives, is not likely to treat the be + Ven sequence as an overt marker of NP movement in general. In sum, my prediction is that the child who acquires her mother tongue according to the internally motivated linguistic principles (i.e. both lexical and syntactic, is not going to fall in the "unaccusative trap" as does an adult learning a second language who is more directly affected by the discernible patterns in the input. I hope that future research will bring forth more direct evidence to support this prediction. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 310 7. Conclusion Due to the issues involved in their lexical categorization and syntactic derivation, the unaccusative-inchoative verbs present a particularly interesting ground for research both in theoretical and applied linguistics. The current thesis analyzed the acquisition of this subclass of intransitive verbs by nonnative speakers of English based on the data obtained from a large computerized corpus of L2 English and a grammaticality judgment test conducted on college-age native speakers of Italian and Japanese. I have shown that some non-target structures typical in the language of L2 English learners have their roots in the mismatch between the lexico-syntactic properties and the prominent surface word orders involving the unaccusative- inchoative verbs. In order to account for the diverse and sometimes contradictory phenomena observed in the use of unaccusative-inchoative verbs both in the literature and in the Longman Learners Corpus, I have proposed the "unaccusative trap" hypothesis, whereby adult learners adopt a simple linking rule at the beginning of their second language acquisition. This simple rule, namely the Single Argument Linking Rule, which may be an L2 iimovation, treats both unergatives and unaccusative-inchoatives in the same way and guarantees that the interlanguage grammar generates superficially target-like intransitive sentences. This initial misanalysis of the two subclasses of intransitive verbs may not be corrected until the learners become more sensitive to the internal consistency of the target language lexicon at a more advanced stage of acquisition. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 311 The "unaccusative trap" hypothesis, which is probably relevant to L2 acquisition of languages other than English as well, has enabled us to formulate a number of detailed predictions, many of which are supported by the grammaticality judgment data. The various structural phenomena such as "passivization," postverbal NP structures, transitivization, NP-V avoidance, etc. are all analyzed from the perspectives offered by the hypothesis. The picture of acquisition of unaccusative-inchoative verbs that has emerged is a rich and complex one. The "passivization" of unaccusative-inchoatives, for example, has turned out to be an interlanguage-particular phenomenon of overgeneraiization widely observed among L2 English learners from various LI backgrounds. In contrast, postverbal NP structures, particularly it-V-NP and 0-V- NP, are claimed to be reflections of the characteristics of LI lexicon, namely, the existence of null expletives. This latter claim may prove to be potentially significant to the question of the psychological reality of null expletives, which are after all mere theoretical constructs. Avoidance of the canonical word order of NP-V is also a notable finding, which illustrates that the most simple-looking structure in the target language such as Subject + Verb may be avoided by L2 learners when their grarrunar is more sensitive to hidden properties of natural languages than to obvious patterns in the available input. Although this thesis has adopted a general framework of Chomskyan generative grammar (mostly Chomsky 1981 and occasionally Chomsky 1991), it contrasts with most L2 studies in the same research tradition. This is because, if the "unaccusative trap" hypothesis is correct, the availability of Universal Grammar in adult L2 acquisition cannot be a mere yes-no question. Instead, the question should be when and how UG figures in L2 acquisition if it does and what its Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 312 consequences might be. As for the L2 acquisition of English unaccusative- inchoative verbs, I have argued that a part of the lexical component of UG, namely, the linking rules and the Unaccusative Hierarchy, may be bypassed or "kept on hold" at the early phase of acquisition. Furthermore, it is when the lexical UG is tapped in to that the L2 learners' language starts showing apparent non-target characteristics. The most exciting aspect of the "unaccusative hypothesis" to me, however, is its potential to shed light on some fundamental differences between LI and L2 acquisition. Chomsky, for example, stated the following about the essence of child language acquisition: Language learning is not really something that the child does; it is somethmg that happens to the child placed in an appropriate environment, much as the child’ s body grows and matures in a predetermined way when provided with appropriate nutrition and environmental stimulation. (Chomsky 1988:134) As discussed in sections 6.2.1.1. and 6.2.2.1, recent research on LI acquisition in the framework of generative grammar has shown that the surface word order of the target language appear as the result of complex interactions of lexical and functional categories of the language regulated by the principles and parameters of Universal Grammar. In other words, as Chomsky envisioned, the child does not appear to be trying to acquire the surface word order of the language but the grammar she acquires determines what those orders and structural representations are. In this respect, adult L2 acquisition is very different from child LI acquisition. In general, the assumptions that adult L2 learners make about the word order and structural representations of the target language are more directly influenced by prominent patterns in the input to which they are exposed. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 313 contrast between LI and L2 acquisition, therefore, is most striking where the target language's surface word order is not a direct reflection of the underlying structure. We have seen three such cases in the acquisition of German transitive verbs, English unaccusative-inchoative verbs and the Japanese Case marking system. In all of these, adult L2 learners apparently start out with wrong assumptions of the basic underlying structures and produce errors that are unknown in LI acquisition when they later subconsciously attempt to correct the initial misanalysis. The difference in the way that the input may affect the process of LI and L2 acquisition is most clearly illustrated by an observation that Koppe (1994) made concerning the language of French-German bilingual children. According to her, it is when the order of arguments in relation to the verb is different at the D- and S- Structures of the target language that the learners exhibit verb- and subject-raising earliest. If her observation is correct, it indicates that children who start with the correct underlying VP configurations take advantage of the difference between the D- and S-Structure word orders in the process of acquiring movement, while the mismatch between the two syntactic levels causes formidable learning problems for adult L2 learners. In conclusion, the picture that has emerged in this study of L2 acquisition of intransitive verbs strongly suggests that child LI and adult L2 acquisition are different in some fundamental ways even in such a very basic realm of language acquisition as classiflcation of verbs. If so, it clearly points to the conclusion expressed by a number of researchers (e.g. Clahsen 1988, Clahsen and Muysken 1986, 1989, Schachter 1988, Strozer 1994) that unlike child LI acquisition, which primarily follows from and is constrained by principles of UG and its parameters, adult L2 acquisition may follow from and be shaped by some other mechanism Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 314 such as general inductive learning strategies as well as UG principles and parameters. The current thesis has investigated this complex picture of L2 development by focusing on the use of two subclasses of intransitive verbs. Future research is expected to further uncover how the linguistic faculty and the general cognitive ability of the learner conspire in shaping the process and product of adult L2 acquisition. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 315 References Ackema, Peter, and Maaike Schooriemmer. 1995. Middles and nonmovement. Linguistic Inquiry 26:173-197. Ahn, Hee-Don, and Hang-Jin Yoon. 1989. Functional categories in Korean. Harvard Studies in Korean 3:79-88. Alexander, L. G., and Edwin T. Cornelius, Jr. 1978. Comp: Exercises in comprehension and composition. New York: Longman. Authier, J.-Marc, and Lisa Reed. 1996. On the Canadian French middle. Linguistic Inquiry 27:513-523. Baker, Mark. 1983. Objects, themes and lexical rules in Italian. In Papers in lexical-functional grammar, ed. Lori Levin, Malka Rappaport and Annie Zaenen, 1-45. Indiana University Linguistics Club, Indiana University, Bloomington. Belletti, Adriana. 1988. The case of unaccusatives. Linguistic Inquiry 19:1-34. Bever, Thomas G., and Brian McEIree. 1988. Empty categories access their antecedents during comprehension. Linguistic Inquiry 19:35-43. Bever, Thomas G., and Montserrat Sanz. 1997. Empty categories access their antecedents during comprehension: Unaccusatives in Spanish, linguistic Inquiry 28:69-91. Bley-Vroman, Robert. 1983. The comparative fallacy in interlanguage studies: The case of systematicity. Language Learning 33:1-17. Borer, Hagit, and Kenneth Wexler. 1986. The maturation of syntax. Studies in the cognitive sciences 39. School of Social Sciences, University of California, Irvine. Bresnan, Joan. 1982. The passive in lexical theory. In The mental representation of grammatical relations, ed. Joan Bresnan, 3-86. Cambridge, Mass.: MIT Press. Brown, Roger. 1973. A first language: The early stages. Cambridge, Mass.: MIT Press. Burt, Marina K., and Carol Kiparsky. 1972. The gooficon: A repair manual for English. Rowley, Mass.: Newbury House. Burzio, Luigi. 1986. Italian syntax. Dordrecht: D. Reidel. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 316 Carey, Susan. 1985. Conceptual change in childhood. Cambridge, Mass.: MTT Press. Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris. Chomsky, Noam. 1986a. Barriers. Cambridge, Mass.: MIT Press. Chomsky, Noam. 1986b. Knowledge of language: Its nature, origin, and use. New York: Praeger. Chomsky, Noam. 1988. Language and problems o f knowledge. Cambridge, Mass.: MIT. Chomsky, Noam. 1991. A minimalist program for linguistic theory. In The view from building 20: Essays in linguistics in honor of Sylvain Bromberger, ed. Kenneth Hale and Samuel Jay Keyser, 1-52. Cambridge, Mass.: MTT Press. Chomsky, Noam. 1995. Categories and transformations. Ms., MIT. Cambridge, Mass. Chung, Yoon-Suk. 1992. Linking and serial verbs in Korean. SICOL Proceeding, 917-930. Cinque, Guglielmo. 1990. Two classes of intransitive adjectives in Italian. In Scrambling and barriers, ed. Gunther Grewendorf and Wolfgang Stemefeld, 261-294. Amsterdam: John Benjamins. Clahsen, Harald. 1988. Parameterized grammatical theory and language acquisition: A study of the acquisition of verb placement and inflection by children and adults. In Linguistic theory in second language acquisition, ed. Suzanne Flynn and Wayne O’Neil, 47-75. Dordrecht: Kluwer. Clahsen, Harald and Pieter Muysken. 1986. The availability of universal grammar to adult and child learners— a study of the acquisition of German word order. Second Language Research 2:93-119. Clahsen, Harald and Pieter Muysken. 1989. The UG paradox in L2 acquisition. Second Language Research 5:1-29. Cook, V. J. 1994. The metaphor of access to universal grammar in L2 learning. In Implicit and explicit learning, ed. N.C. Ellis, 477-502. London: Academic Press. Davis, Lori Jo. 1984. Arguments and expletives: Thematic and nonthematic noun phrases. Doctoral dissertation. University of Connecticut. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 317 Déprez, Viviane, and Amy E. Pierce. 1993. Negation and functional projections in early grammar. Linguistic Inquiry 24:25-67. Dittmar, N. 1981. "Regen bifichen Pause geht"— more on the puzzle of interference. Paper presented at the First European-North American Workshop on Cross- Linguistic Second Language Acquisition Research. UCLA. August 1981. Doron, Edit, and Malka Rappaport Hovav. 1991. Affectedness and extemalization. In Proceedings ofNELS 21, 81-94. GLSA, University of Massachusetts, Amherst. Fagan, Sarah M. B. 1988. The English middle. Linguisticlnquiry 19:181-203. Fellbaum, C. 1986. On the middle construction in English. Indiana University Linguistics Club, Bloomington. Fotos, Sandra S. 1991. The cloze test as an integrative measure of EFL proficiency: A substitute for essays on college entrance examinations? Language Learning 41:313-336. Fukui, Naoki. 1987. Spec, agreement, and the comparative syntax of English and Japanese. In Proceedings ofNELS 17, 193-210. GLSA, University of Massachusetts, Amherst. Fukui, Naoki. 1995. Theory o f projection in syntax. Stanford, Calif.: CSLI Publications. Fukui, Naoki, and Margaret Speas. 1986. Specifiers and projection. In MIT working papers in linguistics 8, 128-172. Department of Linguistics and Philosophy, MIT, Cambridge, Mass. Fukui, Naoki, and Taisuke Nishigauchi. 1992. Head-movement and Case-marking in Japanese. Journal o f Japanese Linguistics 14:1-35. Gass, Susan M., and Jacquelyn Schachter. 1989. Linguistic perspectives on second language acquisition. Cambridge: Cambridge University Press. Gass, Susan M., and Larry Selinker. 1994. Second language acquisition: An introductory course. Hillsdale, N.J.: Lawrence Erlbaum Associates. Grimshaw, Jane. 1990. Argument structure. Cambridge, Mass.: MTT Press. Groat, Erich M. 1995. English expletives: A minimalist approach. Linguistic Inquiry 26:354-365. Haegeman, Liliane. 1991. Introduction to government and binding theory. Oxford: Basil Blackwell. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 318 Haegeman, Liliane. 1994. Introduction ta government and binding theory, 2nd ed. Oxford: Blackwell. Hale, Kenneth, and Samuel Jay Keyser. 1986. Some transitivity alternations in English. Lexicon project working papers 7. Center for Cognitive Science, NOT, Cambridge, Mass. Hale, Kenneth, and Samuel Jay Keyser. 1987. A view from the middle. Lexicon project working papers 10. Department of Linguistics and Philosophy, MIT, Cambridge, Mass. Han, Hak-Sung. 1987. The configurational structure of the Korean language. Doctoral dissertation. University of Texas, Austin. Hanania, Edith, and May Shikhani. 1986. Interrelationships among three tests of language proficiency: Standardized ESL, cloze, and writing. TESOL Quarterly 20:97-109. Hinofotis, Frances Butler. 1980. Cloze as an alternative method of ESL placement and proficiency testing. In Research in Language Testing, ed. John W. Oiler, Jr. and Kyle Perkins, 121-128. Rowley, Mass.: Newbury House. Hinofotis, Frances Butler, and Becky Gerlach Snow. 1980. An alternative cloze testing procedure: Multiple-choice format. In Research in Language Testing, ed. John W. Oiler, Jr. and Kyle Perkins, 129-133. Rowley, Mass.: Newbury House. Hirakawa, Makiko. 1995. L2 acquisition of English unaccusative constructions. In Proceedings of the 19th Boston University Conference on Language Development, 291-302. Hubbard, Philip L. 1983. Relational grammar and language teaching. Paper presented at the Midwest Regional TESOL Convention, Minneapolis, Minn. October 1983. Hubbard, Philip L. 1994. Non-transformational theories of grammar: Implications for language teaching. In Perspectives on pedagogical grammar, ed. Terence Odlin, 49-71. Cambridge: Cambridge University Press. Hubbard, Philip L., and Donna Hix. 1988. Where vocabulary meets grammar: Verb subcategorization errors in ESL writers. CATESOL Journal (November):89-100. Hyams, Nina M. 1986. Language acquisition and the theory of parameters. Dordrecht: D. Reidel. Ingham, Richard. 1996. The ergative alternation in L2 acquisition. Ms., Department of Linguistic Science, University of Reading. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 319 Inoue, Kazuko. 1991. Gengo-ao koozoo-to kinoo [The structures and functions of language]. Sitsugoshoo Kenkyuu [Studies of Agranunatism] 11:116-123. Jackendoff, Ray. 1990. Semantic structures. Cambridge, Mass.: MTT Press. Jacobsen, W. M. 1992. The transitive structure o f events in Japanese. Tokyo: Kurosio Publishers. Jaeggli, Osvaldo A. 1986a. Arbitrary plural pronominals. Natural Language and Linguistic Theory 4:43-76. Jaeggli, Osvaldo A. 1986b. Passive. Linguistic Inquiry 17:587-622. Juffs, Alan. 1996. Leamability and the lexicon: Theories and second language acquisition research. Amsterdam: John Benjamins. Kageyama, Taro. 1993. Bunpoo to gokeisei [Grammar and word formation]. Kasugabe, Japan: Hitsuji Shobo. Kellerman, Eric. 1978. Giving learners a break: native language intuitions as a source of predictions about transferability. Working Papers on Bilingualism 15:59-92. Kellerman, Eric. 1979. The problem with difficulty. Interlanguage Studies Bulletin 4:27-48. Kellerman, Eric. 1983. Now you see it, now you don't. In Language Transfer in Language Learning, ed. Susan M. Gass and Larry Selinker, 112-134. Rowley, Mass.: Newbury House. Keyser, Samuel Jay, and Thomas Roeper. 1984. On the middle and ergative constructions in English. Linguisticlnquiry 15:381-416. Kim, Byong-Kwon. 1993. The structure and the argument-linking convention of V-V compounds in Korean. Harvard Studies in Korean Linguistics 5: ISO- 192. Kim, Jong-Bok. 1993. Syntax and semantics of Korean resultative constructions. Harvard Studies in Korean Linguistics 5:471-482. Kim, Soowon, and Joan Maling. to appear. A crosslinguistic perspective on resultative formation: English vs. Korean. In Proceedings of the 7th Japanese/Korean Conference. Kim, Yang Soon. 1988. Licensing principles and phrase structure. Doctoral dissertation. University of Wisconsin, Madison. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 320 Kim, Young-Joo. 1990. The syntax and semantics of Korean case; The interaction between lexical and syntactic levels of representation. Doctoral dissertation. Harvard University, Cambridge, Mass. Kishimoto, Hideki. 1996. Split intransitivity in Japanese and the unaccusative hypothesis. Language 62:248-286. Koppe, Regina. 1994. NP-movement and subject raising. In Bilingual first language acquisition: French and German grammatical development, ed. Jürgen M. Meisel, 209-234. Amsterdam: John Benjamins. Kuroda, S.-Y. 1988. Whether we agree or not: A comparative syntax of English and Japanese. In Papers from the second international workshop in Japanese syntax, ed.WiUiam J. Posner, 103-143. Stanford, Calif.: CSLI. Kudoda, S.-Y. 1992. Japanese syntax and semantics. Dordrecht: Kluwer. Lee, Chungmin. 1989. (In)definites, Case markers, classifiers and quantifiers in Korean. Harvard Studies in Korean Linguistics 3:469-487. Lee, Young-Suk. 1990. Is Infl universal?: A case study of Korean. In Proceedings o f ESCOL '90, 204-214. Legendre, Géraldine. 1992. Split intransitivity: A reply to Van Valin (1990). Technical Report 92-05. Institute of Cognitive Science, Department of Linguistics, University of Colorado, Boulder. Levin, Beth. 1993. English verb classes and alternations: A preliminary investigation. Chicago: University of Chicago Press. Levin, Beth, and Malka Rappaport. 1986. The formation of adjectival passives. Linguisticlnquiry 17:623-661. Levin, Beth, and Malka Rappaport. 1988. Nonevent -er nominals: A probe into argument structure. Linguistics 26:1067-1083. Levin, Beth and Malka Rappaport. 1989. An approach to unaccusative mismatches. In Proceedings of the NELS 19, 314-329. GLSA, University of Massachusetts, Amherst. Levin, Beth and Malka Rappaport Hovav. 1992. The lexical semantics of verbs of motion: The perspective from unaccusativity. In Thematic structure: Its role in grammar, ed. I. M. Roca, 247-269. Berlin: Foris. Levin, Beth and Malka Rappaport Hovav. 1995. Unaccusativity: At the syntax- lexical semantics interface. Cambridge, MA: M TT Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 321 MacDonald, Maryellen C. 1988. Processing binding in passive sentences. In Proceedings of the Fourth Eastern States Conference on Linguistics (ESCOL '87), 187-198. Ohio State University, Columbus. Marantz, Alec. 1995. The minimalist program. In Government and binding theory and the minimalist program, ed. Gert Webelhuth, 349-382. Oxford: Blackwell. Mills, A. in press. The acquisition of German. In The cross-linguistic study o f language acquisition, ed. Dan Slobin. Hillsdale, N.J.: Erlbaum Press. Miyagawa, Shigem. 1988. Unaccusative verbs in Japanese. In Proceedings o f the Fourth Eastern States Conference on Linguistics (ESCOL ’ 87), 199-210. Ohio State University, Columbus. Miyagawa, Shigem. 1989. Structure and case marking in Japanese. Syntax and Semantics 22. New York: Academic Press. Miyagawa, Shigem. 1990. Case realization and scrambling. Ms., Ohio State University, Columbus. Miyata, H. 1992. A study of the acquisition of the case particles in Japanese with special reference to ga and wo. Master’ s thesis, Osaka University, Japan. Nakayama, Mineham, and Masatoshi Koizumi. 1991. Remarks on Japanese subjects. Lingua 85:303-319. Napoli, Donna Jo. 1988. Review of Italian syntax: A government-binding approach by Luigi Burzio. Language 64:130-142. Oshita, Hiroyuki. 1994. Argument stmcture template and formation of English de verbal adjectives. In Proceedings of the Sixth Annual Student Conference in Linguistics, 247-265. (MIT working papers in linguistics 23.) Department of Linguistics and Philosophy, MIT, Cambridge, Mass. Oshita, Hiroyuki. 1995. Compounds: A view from suffixation and a-structure alteration. In The yearbook o f morphology 1994 ed. Geert Booij and Jaap van Marie, 179-205. Dordrecht: Kluwer. Otsu, Yukio. 1994. Case-marking particles and phrase stmcture in early Japanese acquisition. In Syntactic theory andfirst language acquisition: Cross- linguistic perspectives (vol. I), ed. Barbara Lust, Margarita Suner and John Whitman, 159-169. Hillsdale, N.J.: Lawrence Earlbaum Associates. Perlmutter, David M. 1978. Impersonal passives and the unaccusative hypothesis. In Proceedings o f the Berkeley Linguistics Society 4, 157-189. Pierce, Amy E. 1992. Language acquisition and syntactic theory: A comparative analysis of French and English child grammars. Dordrecht: Kluwer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 322 Radford, Andrew. 1990. Syntactic theory and the acquisition o f English syntax: the nature o f early child grammars o f English. Oxford: Basil Blackwell. Radford, Andrew. 1996. Towards a structure-building model of acquisition. In Generative perspectives on language acquisition: Empirical findings, theoretical considerations and crosslinguistic comparisons, ed. Harald Clahsen, 43-89. Amsterdam: John Benjamins. Richards, Jack C. 1973. A noncontrastive approach to error analysis. In Focus on the learner: Pragmatic perspectives for the language teacher, ed. John W. Oiler, Jr. and Jack C. Richards, 96-113. Rowley, Mass.: Newbury House. Rispoli, Matthew. 1988. Encounters with Japanese verbs: Categorization into transitive and intransitive classes. In Papers and Reports on Child Language Development 27, 81-88. Roberts, Ian. 1987. The representation of implicit and dethematized subjects. Dordrecht: Foris. Roberts, Ian. 1997. Comparative syntax. London: Arnold. Roeper, Thomas. 1987. The acquisition of implicit arguments and the distinction between theory, process, and mechanism. In Parameter setting, ed. Thomas Roeper and Edwin Williams, 309-343. Dordrecht: Reidel. Rosen, Carol G. 1984. The interface between semantic roles and initial grammatical relations. In Studies in relational grammar 2, ed. David M. Perlmutter and Carol G. Rosen, 38-77. Chicago: University of Chicago Press. Rosen, Carol G. 1991. Relational grammar: L2 learning and the components of LI knowledge. In Crosscurrents in second language acquisition and linguistic theories, ed. Thom Huebner and Charles A. Ferguson, 123-142. Amsterdam: John Benjamins. Rutherford, William E. 1987. Second language grammar: Learning and teaching. London: Longman. Rutherford, William E. 1989. Interlanguage and pragmatic word order. In Linguistic perspectives on second language acquisition, ed. Susan M. Gass and Jacquelyn Schachter, 163-182. Cambridge: Cambridge University Press. Rutherford, William E. 1997. Workbook in the structure of English: Linguistic principles and language acquisition. Oxford: Blackwell. Saito, Mamoru. 1983. Case and government in Japanese. In Proceedings of West Coast Conference in Formal Linguisitcs 2, 247-259. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 323 Scarcella, Robin. 1984. Cohesion in the writing development of native and nonnative English speakers. Doctoral dissertation. University of Southern California, Los Angeles. Schachter, Jacquelyn. 1988. Second language acquisition and its relationship to universal grammar. Applied Linguistics 9:219-235. Schroten, Jan. 1986. Ergativity, raising and restructuring in the syntax of Spanish aspectual verbs. Lingvisticce Investigationes 10:439-465. Sciarone, A. G., and J. J. Schoorl. 1989. The cloze test: Or why small isn’ t always beautiful. Language Learning 39:415-438. Sharwood Smith, Michael, and Eric Kellerman. 1989. The interpretation of second language output. In Tranter in language production, ed. Hans W. Dechert and Manfred Paupach, 217-235. Norwood, N.J.: Ablex. Simpson, Jane. 1983. Resultatives. In Papers in lexical-functional grammar, ed. Lori Levin, Malka Rappaport and Annie Zaenen, 143-157. Indiana University Linguistics Club, Indiana University, Bloomington. Snyder, William, and Karin Stromswold. 1997. The structure and acquisition of English dative constructions. Linguistic Inquiry 28:281-317. Sorace, Antonella. 1993a. Incomplete vs. divergent representations of unaccusativity in nonnative grammars of Italian. Second Language Research 9:22-47. Sorace, Antonella. 1993b. Unaccusativity and auxiliary choice in nonnative grammars of Italian and French: Asymmetries and predictable indeterminacy. French Language Studies 3:71-93 Sorace, Antonella. 1995. Acquiring linking rules and argument structures in a second language. In The current state of interlanguage: Studies in honor of William E. Rutherford, ed. Lynn Eubank, Larry Selinker and Michael Sharwood Smith, 153-175. Amsterdam: John Benjamins. Strozer, Judith R. 1994. Language acquisition after puberty. Washington, DC.: Georgetown University Press. Suh, Jinhee. 1990. Scope phenomena and aspects of Korean syntax. Doctoral dissertation. University of Southern California, Los Angeles. Takezawa, Koichi. 1987. A configurational approach to Case-marking in Japanese. Doctoral dissertation. University of Washington, Seattle. Takezawa, Koichi. 1993. Secondary predication and locative/goal phrases. In Japanese syntax in comparative grammar, ed. Nobuko Hasegawa, 45-77. Tokyo: Kurosio Publishers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 324 Terada, Michiko. 1987. Unaccusativity in Japanese. In Proceedings ofNELS 17, 619-640. GLSA, University of Massachusetts, Amherst. Terada, Michiko. 1990. Incorporation and argument structure in Japanese. Doctoral dissertation. University of Massachusetts, Amherst. Teramura, Hideo. 1990. Gaikokujin gakushuusha-no nihongo goyoorei-shuu [Collection of errors in Japanese by foreign learners]. Osaka University. Tsujimura, Natsuko. 1990a. Ergativity of nouns and Case assignment. Linguistic Inquiry 21:277-287. Tsujimura, Natsuko. 1990b. The unaccusative hypothesis and noun classification. Linguistics 28:929-957. Tsujimura, Natsuko. 1991. On the semantic properties of unaccusativity. Journal of Japanese Linguistics 13:91-116. Tsujimura, Natsuko. 1996. An introduction to Japanese linguistics. Cambridge, Mass.: Blackwell. Ueda, Masanobu. 1993. On the phrase structure of Japanese and English clauses. In Japanese syntax in comparative grammar, ed. Nobuko Hasegawa, 9-44. Tokyo: Kurosio Publishers. Van Valin, Robert D. Jr. 1987. The unaccusative hypothesis vs. lexical semantics: Syntactic vs. semantic approaches to verb classification. In Proceedings of NELS 17, 641-661. GLSA, University of Massachusetts, Amherst. Van Valin, Robert D. Jr. 1990. Semantic parameters of split intransitivity. Language 66:221-260. Van Valin, Robert D. Jr. 1993. Advances in role and reference grammar. Amsterdam: John Benjamins. Yip, Virginia. 1989. Aspects of Chinese/English interlanguage: syntax, semantics and leamability. Doctoral dissertation. University of Southern California, Los Angeles. Yip, Virginia. 1994. Grammatical consciousness-raising and leamability. In Perspectives on pedagogical grammar, ed. Terence Odlin, 123-139. Cambridge: Cambridge University Press. Yip, Virginia. 1995. Interlanguage and leamability: From Chinese to English. Amsterdam: John Benjamins. Williams, Edwin. 1987. English as an ergative language: The theta structure of derived nouns. In Proceedings of Chicago Linguistic Society, 366-375. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 325 Zobl, Helmut. 1989. Canonical typological structures and ergativity in English L2 acquisition. In Linguistic perspectives on second language acquisition, ed. Susan M. Gass and Jacquelyn Schachter, 203-221. Cambridge: Cambridge University Press. Zubizarreta, Maria Luisa. 1987. Levels of representation in the lexicon and in the syntax. Dordrecht: Foris. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 326 Appendices APPENDIX A: DISTRIBUTION OF SENTENCES BUILT ON UNACCUSATT/E VERBS ________________________Italian Spanish K orean Japanese Total Structures i: NP V 216 327 39 269 851 appear 11 22 I 27 61 arise 2 0 2 6 arrive 80 134 19 74 307 die 42 7 56 119 disappear 2 7 1 11 21 exist - 8 ' 8 0 20 36 faU 20 21 1 10 52 happen 71 84 7 49 211 occur 6 7 3 15 31 rise 2 0 0 5 7 ii: NP-be-Ven 8 4 8 17 37 appear 0 0 3 0 3 arise 0 0 1 0 1 arrive 4 1 0 1 6 die 0 1 0 6 7 disappear 1 0 2 1 4 exist 1 0 0 0 1 faU 0 0 0 1 1 happen 2 2 2 7 13 occur 0 0 0 1 I rise 0 0 0 0 0 iii: there-V-NP 2 0 1 1 4 appear 0 0 0 0 0 arise 0 0 I 1 arrive 0 0 0 0 0 die 0 0 0 0 disappear 0 0 0 0 0 exist : ■ - r ■ ~ 0 1 0 3 faU 0 0 0 0 0 happen 0 0 0 0 0 occur d 0 0 0 0 rise 0 0 0 0 0 iv: it-V-NP 5 6 1 1 13 appear 0 I 0 0 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 327 anse w V , . . _ 0 0 0 arrive 0 1 0 0 1 ICQS::'.. 0 0 0 disappear 0 0 0 0 0 exist 0 0 I faU 0 0 0 0 0 happen 4 1 - 1 9 occur i 0 0 0 1 rise . : 0 0 0 0 v: 0-V-NP 7 8 0 1 16 appear anse 2 ' 4 0 0 _ 0 0 0 6 0 arrive 0 1 0 0 1 die 0 0 0 0 disappear exist 0 2 0 0 0 0 0 4 fall 0 0 0 0 0 happen 3 1 0 1 5 occur 0 0 0 0 0 rise 0 0 0 0 0 vi: there-be-Ven-NP 0 0 0 4 4 appear 0 0 0 3 3 arise 0 0 0 0 0 arrive 0 0 0 0 0 die 0 0 0 0 0 disappear 0 0 0 0 0 exist 0 0 0 0 0 faU 0 0 0 0 0 happen 0 0 0 1 1 occur 0 0 0 0 0 rise 0 0 0 0 0 vii: it-be-Ven-NP 0 0 0 0 0 appear................................................................................................ anse . 0 Ü V 0 0 _ 0 0 0 0 0 arrive Ô 0 0 0 0 die ; 0 0 0 Q 0 disappear 0 0 0 0 0 exist 0 0 0 0 fall 0 0 0 0 0 happen 0 0 0 0 0 occur 0 0 0 0 0 rise 0 0 0 0 0 viii: 0 -be-Ven-NP 0 0 0 0 0 appear 0 0 0 0 0 arise 0 .0 0 0 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 328 arrive 0 0 0 0 0 die 0 0 0 0 disappear 0 0 0 0 0 exist 0 0 0 0 faU 0 0 0 0 0 happen ■ 0 0 0 0 occur 0 0 0 Ô 0 0 0 0 ix: NP1-V-NP2 2 2 1 7 12 appear 0 0 0 1 1 arise 0 0 0 0 arrive 0 0 0 0 0 die Q ' 0 0 1 I disappear 0 0 0 0 0 exist 0 0 0 0 0 faU 0 1 1 0 2 happen 0 1 0 2 3 occur 0 0 0 2 2 rise -2 0 0 1 3 X : there-be-NP-V 0 0 0 4 4 appear arise 0 0 0 0 0 0 0 0 0 arrive 0 0 0 0 0 die 0 0 0 0 0 disappear 0 0 0 0 0 exist 0 0 0 1 1 fall 0 0 0 0 0 happen 0 0 0 3 3 occur 0 0 0 0 0 rise ■ 0 0 0 0 Total 240 347 50 304 941 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 329 APPENDIX B: SENTENCES WITH UNACCUSATIVE VERBS (i) NP-V: (Omitted due to a large number of instances.) (ii) NP-be-Ven: (The existence of be may be taken as sufficient evidence for this structure.) Italian arrive {2182in4.iit}=but I didn't know at what time you should be arrive so I decided to... {4899prI.iit}=I was nearly arrived to my office {7216in2.iit}=You are arrived in the eternity city {7238in2.iit}=so he was arrived disappear {7213in2.iit}=this city with her ruins will be disappeared exist {2426uil.iit}=I want to say that this kind of problem will be exist forever... happen {2359ui5.iit}=We will tell each other what is happened in these five years and remember the great moments lived together {7275in2.iit}=they were happened a few days ago Spanish arrive {0421zzz.ssp} =The family could be arrives in Saturday morning, and had lunch with him, üien they could go walk around the jail or go to the town to watch a film or to go for a drinks. die {0456inl.ssp}=I thought that maybe I was died happen {3021 in 1 .ssp} =No body would tell her what was happened {3042in 1 .ssp} =Nothing it was happened, nothign it was said to her Korean appear {0939prl.kox)=the word, "the role of women," is appeared just several years ago {6896inl.kos}=suddenly pale face was appeared out of the window {6933inl.kos}=He is also appeared on the list of investigations of gold smuggling arise {0892prl.kox}=education boom , was arised among high societies Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 330 disappear {0900prl.kox}=the dysfunctions of urbanization will be disappeared in LA. {0908adl.kox}=the cultural uniqueness is now disappeared happen {0871uil.kox)=terrorism is happened very often {6887inl.kos}=to find out what would be happened in the next stories Japanese arrive {4479el9.jap}= After that we were arrive at the station, die {5538inl.jap}=until they are died {5633ui2.jap}=After a week patient will be died {5740inl.jap}=Near future, some children will be die. {5795uil.jap}=If people throw away litter, trash garbage, and waste, the earth must be die. {5815uil.jap}=their parents were died by car accident {8271 ui 1 .jap }=first wife was died disappear {5608uil.jap}=a prejudice may be disappeared fall {5653ui2.jap}=Usually ash is fall in ground about one or two years, happen {5567el2.jap}=I don't know what was happened {5575uil.japi=It never should be happened {5583uil.jap}=many problems will be hap^ned {5645ui2.japi=this incident was happened in some countries {5853pil.Jap}=Whenever serious incidents was happened somewhere in the world... {5866pil.Jap}=... when the war was happened between Iraqi and Kuwaiti... {7942ui2.jap }=Two or three days ago, die important trouble was happened. occur {5654ui2.Jap}=If the "fighting" are occure in other place, this time Japanese government try to send a lot of soldiers in a terrible places. (iii) there-V-NP: Italian exist {2365ui 1 .iit} =there exist two kinds of jobs {2474ad 1 .iit} =there exist a large share of population Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 331 Korean exist {0870prl.kox}=there exist many kind of prejudice Japanese arise {5588uil.Jap}=there often arise the problem of political indiference (iv) It-V-NP: Italian exist {1559inl.iit}=it existed a lot of restrictions happen {2163in4.iit}=I'm writing to you in order to make you aware of what it has happened to me in your hotel. {2390ui l.iit}=I would be very glad if we could have a meeting to speak about the school times and what it's happened after it. {7242in2.iit}=it happened a tragic event occur {4922pr 1 .iit} =When it occurs a damage to them, for instance, we can hardly realize it and... Spanish appear {4385inl.ssp}=So, the bands of desperate non-smokers able to steal from the tobacco shops or the rich and happy smokers would increase in a very difficult-to-solve way. And it would probably appear a new kind of non-legal sellers, who would create bands, more corrupted police and so on. arrive {4407inl.ssp}=it arrived the day of his departure happen {3043inl.ssp}=it will happen something exciting {4408inl.ssp}=Now that I have left school, and I am at University, it happens something funny: when I get off in the morning to go to the faculty I still continue saying "Bye, mum. I’ m going to school. {4435in 1 .ssp} =talking about what would it happen with our relationship {7546pil.ssp}=asking me for what it happened to me Korean happen {6898inl.kos}=it happened very strange thing Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 332 Japanese happen {Ol58zz5.jap}=it would happen to me bad things (v) 0-V-NP: Italian appear {7201in2.iit}=...and appeared a little green man who... {7222in2.iit}=like a mirage appeared the large expanse of the sea exist {0378adl.iit}=There is no doubt that does exist a big difference between... {5516adl.iitj=it is said that exist the purest Italian language happen {2408ui 1 .lit} =One day happened a revolution. {2523adl.iit}=During my schooldays happened so many things that I can't talk about all 6em. (7240in2.iit}=Once happened something no very good for her Spanish appear (0299inl.ssp}=in most of the films, appear violence, drugs, and all that things that people seems to adore (5127uil.sec}=a wonderful blue sky where still appeared a round white spot {6737uil.ssp }=because in our ccentury have a p p e a l the car and the plane {6741uil.ssp}=..., so appeared the phenomena called, video arrive (3027inl.ssp}=When arrives the summer they want to forget all the problems exist (6738uil.ssp}=In every country exist criminals {6738uil.ssp}=and in i l them will exist criminals happen (3028inl.ssp}=never happen something new to yourself Japanese happen (5851pil.jap}=And happened a few case. (vi) there-be-Ven-NP: Japanese appear (5544inl.jap}=there are appear_ new dreams Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 333 {5759uil.jap}=If is continued smoking there are appear_ some bad diseases like lung cancer and chronic Obstructive Pulmonary Disease and so on. {5842ui2.jap }= After the war, there were appeared a lot of women who believed that women should go outside and should be equal in the society. happen {5652ui2.jap}=there was happen_ problem about American minister (vii) it-be-Ven-NP: (viii) 0-be-Ven-NP: (ix) NP1-V-NP2: (a): non-causative reading (experiencer as subject, see Zobl 1989) (b): non-causative reading (locative as subjectc) (c): causative reading Italian rise {2464adl.iit}=I rose my hand {2521ad.l.iit}=he rose his eyes up to me Spanish fall {6756uil.ssp}=It (=a wall) was failed down in order to get a bigger green house (passive causative) happen {7487pil.ssp}=I was explaining that (=what) I had happened Korean fall {6852inl.kos }=he falls a piece of note into dough by mistake Japanese appear {6095el2.jap}=The TV appear all kinds of animals, die {802lui2.jap }=[the Congress] may die out the ban on women in the battle happen {1471ell.jap}=I happened two incident {5620ui2.jap}=engine that don't happen enviommental pollution Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 334 occur {5684inl.jap}=concentrated population could occur these kind of problem {5798uil.jap}=they occur many kinds of disease rise {5790uil.jap}=I rise a example which is familiar to you (x) expletive be + NP + V Japanese exist {0577prl.jap}=because I know there are so many kind of peopl exist in the U.S. so called U.S. is a melting pot. happen {1463el 1 .jap }=there was two thig happened {3999ui 1 .jap} =because there was a land of strike so often happened in that country. {5647ui2.jap }=there are a lot of strange things happen who saw the UFO Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 335 APPENDIX C; C l o z e Te st Directions: Read the following three stories carefully and fill in each blank with a contextually and granunatically appropriate word. To get an idea about each story, first you can read it once or twice and then try to fill in the blanks. Do not worry too much about difficult blanks. Just try to fill in as many blanks as you can -- by guessing if necessary. You have approximately 15 minutes, i.e. about 5 minutes for each story. Remember that you can write only one word in each blank. Story A: Some people do not seem to have a mind of their own. They seldom make their 1 ______ decisions and never express 2 own opinions. My brother______3_____one of these people. ______ 4____ night, for example, he 5 planning to spend a 6 evening at home reading 7______ book. At about seven-thirty, _____8______, his friend Tom dropped 9 and said, "Let's watch 10 tonight." "Okay," my brother 11 . By ten o'clock m y 12______ was tired and sleepy, 13______ I am sure he 14 to go to bed. .15______ Tom was not tired. " _____ 16______ go out and get hamburger," Tom said. "Good 18 my brother 19______, he very often says 20 he does not order to please others. _____22 than that, he does tell others what he 24 wants to do, thinking might offend them. In any case, my brother did not come home until midnight and woke up very late this morning. Story B: We were about to gather up our picnic things and return to our car when a man showed up. He looked very annoyed 1 ______ asked us angrily if 2______ realized that we were______3_____private property. My father, 4 looked very confused a t_____ 5______ man’ s statement, said that _6______ did not. The man_____ 7______ pointed to a sign_____8______ 17 replied. Like mean 21 23 25 said that camping an d 9 were strictly forbidden in 10_ area where we were 11 . Poor father explained that_____ 12______ had not seen the 13______ until then and had_____14______realized that it was 15______ property. Despite my father's_____ 16______, the man did not 17 satisfied at all and 18 him for his name 19______address. All the way 20______ , we were so upset 21______hardly anyone said a 22______ . Everyone in the car 23______wondering if the angry 24 would report us to 25______police. Although he didn't after all, this unpleasant incident completely ruined the wonderful time we had had in the country that day. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 336 Story C: Hunting was originally a means of providing food, but it has now become a sport or a cultural tradition. Although even today in 1 ______ parts of the world 2______ are still people who_____3______ wild fish, birds and hunting 8 10 12 14 16 _ to provide themselves with_ _ 6______ now as much a _____ _ great many years ago, _____S used birds to catch_____11. _ , m many countries that they would dive. 13. 15 to the fishermen after of fishing is said___ 19______ in the country's mythological 17 . activity as anything else, in a small Afincan The birds were trained the water and come _ a few fish. This 18______ be at least a years old and is Today, however, fishing in . 22______ tradition, since those who 20 21 24 longer seriously interested in way has simply become 23______ in this manner are fish for food. Their 25 real concern is simply to maintain this old cultural tradition. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 337 Cloze Answers Acceptable — minor spelling errors, UPPER and lower cases ♦Unacceptable — errors in part of speech, tense, number, styles, etc. Story A: 1. ow n 2. their 3. is 4. Last ♦One 5. was ♦began 6. quiet, whole, pleasant, lazy, relaxing, routine, nice ♦special, ♦normal, ♦good, ♦free, ♦boring, ♦wonderful, ♦calm 7. a ♦his, ♦my, etc. 8. however, unexpectedly, suddenly ♦then 9. in, by 10. TV, movies, videos, baseball, tennis, etc. ♦games, ♦movie, ♦video 11. said, replied, answered, responded, agreed 12. brother 13. and, so 14. wanted, needed, had ♦decided 15. But ♦However 16. Let's 17. a ♦some 18. idea ♦thinking 19. this, others ♦always, ♦me, ♦that 20. what, things ♦something, ♦words 21. in 22. More, Worse ♦Other, ♦Rather 23. not 24. really, truly, actually, himself 25. it, that, this, he Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 338 Story B: 1. and 2. we 3. on, using *in, *invading, trespassing 4. who 5. the, this ♦that 6. we, he 7. then, angrily, simply, just, silently, immediately ♦also, ♦suddenly, ♦strongly 8. which, that, and 9. picnic(k)ing, picnics, trespassing, parking ♦picnic, ♦fishing, ♦eating, ♦swimming, ♦cooking, ♦playing, ♦staying, ♦hiking 10. the, this ♦that, ♦his 11. picnic(k)ing, sitting, standing ♦at, ♦in, ♦then, ♦now, ♦seated, ♦staying, ♦parking, ♦camping, ♦hiking 12. he, we 13. sign ♦paper, ♦notice 14. not ♦never, ♦just, ♦now, ♦suddenly, ♦finally 15. private, his 16. apology, explanation, pleading, excuse, reply, answer ♦plea(s), ♦statement, ♦opinion, ♦words, ♦insistence 17. look, seem, appear ♦get, ♦become, ♦feel 18. asked ♦demanded 19. and 20. home, back 21. that ♦and 22. word, thing 23. was, kept ♦were 24. man ♦person 25. the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 339 Story C: 1. some, many, several, various ♦ail, *any, *most, *few, *rural 2. there ♦hunters 3. hunt, catch, capture, kill ♦get, ♦use 4. animals, game ♦animal, ♦beasts, ♦deer, ♦others 5. food ♦foods, ♦money 6. is 7. sporting, sports, leisure, social, traditional, cultural, ritual, recreational ♦sport, ♦pleasurable, ♦hobby, ♦wonderful, ♦necessary, ♦popular, ♦historical 8. A 9. people, hunters, fîshermen, men ♦those, ♦women 10. country, nation, village ♦town, ♦city, ♦island, ♦tribe, ♦area 11. fish ♦fishes 12. so 13. into, in, under 14. back, up ♦out, ♦home, ♦around, ♦immediately 15. catching, swallowing, capturing ♦hunting, ♦getting 16. way, method, kind, style, type, manner ♦technique, ♦means, ♦process, ♦tradition, ♦custom 17. to 18. thousand ♦thousands, ♦1000, ♦hundred 19. recorded, reported, mentioned, included, found, written, alive ♦ingrained, ♦seen, ♦continued, ♦placed, ♦done, ♦used, ♦now, ♦essential, ♦famous, ♦told 20. tales, stories, accounts, records, books ♦tale, ♦story, ♦account, ♦record, ♦book, ♦events, ♦parts, ♦culture, ♦ceremony, ♦way, ♦tradition, ♦study 21. this ♦a, ♦that, ♦the 22. a, their ♦the, ♦its 23. fish, live ♦catch, ♦hunt, ♦indulge 24. no 25. catching, the, these ♦eating, ♦hunting, ♦getting, ♦providing, ♦wild, ♦fresh Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 340 a p p e n d ix D: Gr a m m a t ic a u t y J u d g m e n t T e s t Form A Direction: Read the following passages carefully and decide whether each underlined part is both contextually and grammatically correct English. For your answer, use the following numbers. If you feel sure that it is not correct English. circle "1." If you feel not so sure but think that it is probably not correct English. circle "2." If you cannot decide. circle "3." If you feel not so sure but think that it is probably correct English. circle "4." If you feel sure that it is correct English. circle "5." Giye your answers by marking the appropriate circles on the answer sheet. Remember that you judge only underlined parts. There are 56 questions in total and you baye approximately 30 minutes to answer. 1 <................... 2 .................... 3 4 > 5 Not Correct Cannot Decide Correct [1] One cold evening, Nancy found three baby birds in the garden and brought them inside the house. She placed them under a warm blanket, but the cold night died two of them before the next morning. [2] Sometimes, our faces can change the mood of other people. For example, many of us naturally feel happy when a little baby is smiled at us. [3] Before our wedding ceremony started, there wasn't a single problem. Both families were there. The guests were arriving on time. For a short time, it was seemed that everything was going well. .. [4] Although some people think that only their life is full of troubles, some problems exist in evervone’ s life. [5] The drivers wanted to go closer to the flying object that had landed in the woods. They were getting out of their cars, when out of the spaceship suddenly appeared a little green man. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 341 [6] The true excitement of traveling is in the planning. We all wish that during the trip it will happen something interesting. [7] When the bus broke down, passengers had to find another way to travel. Some people took a taxi, some changed to another bus, and there walked others all the wav home. [8] My dog Kelly got excited when we had the first snow of the season. She started playing in the garden right away, but by the early afternoon most of the snow melted. [9] In industrialized countries, many young people prefer to marry and have babies later in life. If these trends continue, the lower birth rate will decrease the average familv size in these nations. [10] My grandmother is now 97 years old, but she appears to be still healthy and active for her age. 1 <--------------- 2 3 .. 4 > 5 Not Correct Cannot Decide Correct [11] One cold evening, Nancy foimd three baby birds in the garden and brought them inside the house. She placed them under a warm blanket, but two of them died before the next morning. [12] Although some people think that only their life is full of troubles, some problems are existed in evervone’ s life. [13] When the bus broke down, passengers had to find another way to travel. Some people took a taxi, some changed to another bus, and it walked others all the wav home. [ 14] The true excitement of traveling is in the planning. We all wish that during the trip will happen something interesting. [15] My grandmother is now 97 years old, but it appears that she is still healthy and active for her age. [16] Sometimes, our faces can change the mood of other people. For example, many of us naturally feel happy when diere smiles a little baby at us. [17] In industrialized countries, many young people prefer to marry and have babies later in life. If these trends continue, the average family size will decrease in these nations. [18] My dog Kelly got excited when we had the first snow of the season. She started playing in the garden right away, but by the earlv afternoon most of the snow was melted. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 342 [19] The drivers wanted to go closer to the flying object that had landed in the woods. They were getting out of their cars, when the spaceship suddenly appeared a little green man. [20] The true excitement of traveling is in the planning. We all wish that during the trip our little adventure will happen something interesting. 1 <-------------- 2 3 - 4 > 5 Not Correct Cannot Decide Correct [21 ] When the bus broke down, passengers had to find another way to travel. Some people took a taxi, some changed to another bus, and walked ofliers all the way home. [22] The drivers wanted to go closer to the flying object that had landed in the woods. They were getting out of their cars, when out of the spaceship a little green man suddenly appeared. [23] Sometimes, our faces can change the mood of other people. For example, many of us naturally feel happy when it smiles a little babv at us. [24] Before our wedding ceremony started, there wasn't a single problem. Both farriilies were there. The guests were arriving on time. For a short time, everything was seemed to be going well ... [25] My dog Kelly got excited when we had the first snow of the season. She started playing in the garden right away, but by the earlv afternoon there melted most of the snow. [26] In industrialized countries, many young people prefer to marry and have babies later in life. If these trends continue, the average familv size will be decreased in these nations. [27] Although some people think that only their life is fiill of troubles, there exist some problems in evervone's life. [28] One cold evening, Nancy found three baby birds in the garden and brought them inside the house. She placed them under a warm blanket, but two of them were died before the next morning. [29] When the bus broke down, passengers had to find another way to travel. Some people took a taxi, some changed to another bus, and the accident walked others all die wav home. [30] My dog Kelly got excited when we had the first snow of the season. She started playing in the garden right away, but by the earlv afternoon it melted most of the snow. 1 <.................. 2 3 .................. 4 > 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 343 Not Correct Cannot Decide Correct [31 ] My grandmother is now 97 years old, but it is appeared that she is still healthy and active for her age. [32] Sometimes, our faces can change the mood of other people. For example, many of us naturally feel happy when smiles a little babv at us. [33] In industrialized countries, many young people prefer to marry and have babies later in life. If these trends continue, there will decrease the average familv size in these nations. [34] Although some people think that only their life is full of troubles, it exists some problems in evervone's life. [35] One cold evening, Nancy found three baby birds in the garden and brought them inside the house. She placed them under a warm blanket, but there died two of them before the next morning. [36] The drivers wanted to go closer to the flying object that had landed in the woods. They were getting out of their cars, when out of the spaceship a little green man was suddenly appeared. [37] The true excitement of traveling is in the planning. We all wish that during the trip something interesting will happen. [38] Before our wedding ceremony started, there wasn't a single problem. Both fanülies were there. The guests were arriving on time. For a short time, everything seemed to be going well.... [39] One cold evening, Nancy found three baby birds in the garden and brought them inside the house. She placed them under a warm blanket, but it died two of them before the next morning. [40] Although some people think that only their life is full of troubles, exist some problems in evervone's life. 1 <-------------- 2 3 .. 4 ...................> 5 Not Correct Cannot Decide Correct [41 ] The true excitement of traveling is in the planning. We all wish that during the trip something interesting will be happened. [42] The drivers wanted to go closer to the flying object that had landed in the woods. They were getting out of their cars, when out of the spaceship there suddenly appeared a little green man. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 344 [43] When the bus broke down, passengers had to find another way to travel. Some people took a taxi, some changed to another bus, and others walked all the wav home. [44] Before our wedding ceremony started, there wasn't a single problem. Both families were there. The guests were arriving on time. For a short time, it seemed that everything was going well.... [45] In industrialized countries, many young people prefer to marry and have babies later in life. If these trends continue, it will decrease the average family size in these nations. [46] Sometimes, our faces can change the mood of other people. For example, many of us naturally feel happy when our soft and tender touch smiles a little babv at us. [47] My dog Kelly got excited when we had the first snow of the season. She started playing in the garden right away, but bv the early afternoon melted most of the snow. [48] The true excitement of traveling is in the planning. We all wish that during the trip there will happen something interesting. [49] In industrialized countries, many young people prefer to marry and have babies later in life. If these trends continue, will decrease the average familv size in these nations. [50] One cold evening, Nancy found three baby birds in the garden and brought them inside the house. She placed them under a warm blanket, but died two of them before the next morning. 1 <................... 2 -------------- 3 ................... 4 > 5 Not Correct Cannot Decide Correct [51] Sometimes, our faces can change the mood of other people. For example, many of us naturally feel happy when a little babv smiles at us. [52] My grandmother is now 97 years old, but she is appeared to be still healthy and active for her age. [53] The drivers wanted to go closer to the flying object that had landed in the woods. They were getting out of their cars, when out of the spaceship it suddenlv appeared a little green man. [54] When the bus broke down, passengers had to find another way to travel. Some people took a taxi, some changed to another bus, and others were walked all the wav home. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 345 [55] My dog Kelly got excited when we had the first snow of the season. She started playing in the garden right away, but by the early afternoon the sun melted most of the snow. [56] Although some people think that only their life is full of troubles, our modern-day lifestyle exists some problems in evervone's life. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 346 Form B Direction: Read the following passages carefully and decide whether each underlined part is both contextually and grammatically correct English. For your answer, use the following numbers. If you feel sure that it is not correct English. circle "1." If you feel not so sure but think that it is probably not correct English. circle "2." If you cannot decide. circle "3." If you feel not so sure but think that it is probably correct English. circle "4." If you feel sure that it is correct English. circle "5." Give your answers by marking the appropriate circles on the answer sheet. Remember that you judge only underlined parts. There are 56 questions in total and you haye approximately 30 minutes to answer. 1 <................... 2 -................... 3 4 > 5 Not Correct Cannot Decide Correct [ 1 ] More and more babies are bom in the world every year. Because of this population increase, it mav occur many problems in the near future. [2] Once a year John's high school has a big dance party. It is an exciting and happy night for the students and there dance many students until morning. [3] John invited his classmates to a birthday party at his house. Although most of his friends came on time, a traffic accident on the highway arrived others much later. [4] Last night I had a very scary experience. We had a couple of big earthquakes around 1 a.m., and in mv apartment some windows shattered. [5] Before our wedding ceremony started, there wasn't a single problem. Both families were there. The guests were arriving on time. For a short time, everything seemed to be going well. .. [6] Last year a big fire destroyed houses in Jack's town. Luckily Jack's house was saved, but the next morning only a few houses remained around it. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 347 [7] During the rainy season, standing near rivers is dangerous because in a short time the heavy rain mav increase the water level rapidly. [8] As technological advancement continues and international communication becomes easier, may soon disappear cultural traditions in manv countries. [9] The comedian was having a hard time that night. On the stage he told many jokes, but only a few people in the audience were laughed. [ 10] My grandmother is now 97 years old, but it is appeared that she is still healthy and active for her age. 1 <................... 2 .................. 3 .................. 4 ....................> 5 Not Correct Cannot Decide Correct [11] Once a year John's high school has a big dance party. It is an exciting and happy night for the students and manv students dance until morning. [12] During the rainy season, standing near rivers is dangerous because in a short time it mav increase the water level rapidly. [13] Last year a big fire destroyed houses in Jack's town. Luckily Jack's house was saved, but the next morning remained only a few houses around it. [14] The comedian was having a hard time that night. On the stage he told many jokes, but his jokes laughed only a few people in the audience. [15] As technological advancement continues and international communication becomes easier, there mav soon disappear cultural traditions in many countries. [16] Last night I had a very scary experience. We had a couple of big earthquakes around 1 a.m., and in mv apartment shattered some windows. [17] John invited his classmates to a birthday party at his house. Although most of his friends came on time, it arrived others much later. [18] More and more babies are bom in the world every year. Because of this population increase, manv problems mav be occurred in the near future. [19] Before our wedding ceremony started, there wasn't a single problem. Both fanülies were there. The guests were arriving on time. For a short time, everything was seemed to be going well. .. [20] John invited his classmates to a birthday party at his house. Although most of his friends came on time, others were arrived much later. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 348 1 <--------------- 2................... 3 — ............. 4 > 5 Not Correct Cannot Decide Correct [21] As technological advancement continues and international communication becomes easier, cultural traditions in many countries may soon disappear. [22] During the rainy season, standing near rivers is dangerous because in a short time the water level may be increased rapidly. [23] Once a year John's high school has a big dance p a ^ . It is an exciting and happy night for the students and dance many students until morning. [24] Last year a big fire destroyed houses in Jack's town. Luckily Jack's house was saved, but the next morning diere remained onlv a few houses around it. [25] More and more babies are bora in the world every year. Because of this population increase, the shortage of food and space may occur many problems in the near future. [26] The comedian was having a hard time that night. On the stage he told many jokes, but it laughed onlv a few people in the audience. [27] My grandmother is now 97 years old, but it appears that she is still healthy and active for her age. [28] Last night I had a very scary experience. We had a couple of big earthquakes around 1 a.m., and in mv apartment there shattered some windows. [29] Last year a big fire destroyed houses in Jack's town. Luckily Jack's house was saved, but the next morning the fire remained onlv a few houses around it. [30] As technological advancement continues and international communication becomes easier, it mav soon disappear cultural traditions in many countries. 1 <.................... 2................... 3 ................... 4 > 5 Not Correct Cannot Decide Correct [31 ] The comedian was having a hard time that night. On the stage he told many jokes, but onlv a few people in the audience laughed. [32] John invited his classmates to a birthday party at his house. Although most of his friends came on time, arrived others much later. [33] Last night I had a very scary experience. We had a couple of big earthquakes around 1 a.m., and in mv apartment the violent shakes shattered some windows. [34] Once a year John's high school has a big dance party. It is an exciting and happy night for the students and manv students are danced until morning. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 349 [35] More and more babies are bom in the world every year. Because of this population increase, there may occur many problems in the near future. [36] Before our wedding ceremony started, there wasn't a single problem. Both families were there. The guests were arriving on time. For a short time, it seemed that everything was going well. .. [37] During the rainy season, standing near rivers is dangerous because in a short time mav increase the water level rapidly. [38] John invited his classmates to a birthday party at his house. Although most of his friends came on time, others arrived much later. [39] Last year a big fire destroyed houses in Jack's town. Luckily Jack's house was saved, but the next morning onlv a few houses were remained around it. [40] More and more babies are bora in the world every year. Because of this population increase, may occur manv problems in the near future. 1 <.................... 2 3 4 > 5 Not Correct Cannot Decide Correct [41 ] My grandmother is now 97 years old, but she is appeared to be still healthy and active for her age. [42] As technological advancement continues and international communication becomes easier, this great change mav soon disappear cultural traditions in many countries. [43] The comedian was having a hard time that night. On the stage he told many jokes, but there laughed onlv a few people in the audience. [44] Once a year John's high school has a big dance party. It is an exciting and happy night for the students and it dances many smdents until morning. [45] During the rainy season, standing near rivers is dangerous because in a short time the water level may increase rapidly. [46] Last night I had a very scary experience. We had a couple of big earthquakes around 1 a.m., and in mv apartment some windows were shattered. [47] John invited his classmates to a birthday party at his house. Although most of his friends came on time, there arrived others much later. [48] Before our wedding ceremony started, there wasn't a single problem. Both frirailies were there. The guests were arriving on time. For a short time, it was seemed that everything was going well.... Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 350 [49] Last year a big fire destroyed houses in Jack's town. Luckily Jack's house was saved, but the next morning it remained onlv a few houses around it. [50] As technological advancement continues and international communication becomes easier, cultural traditions in manv countries mav be soon disappeared. 1 <--------------- 2 3 4 > 5 Not Correct Cannot Decide Correct [51] Once a year John's high school has a big dance party. It is an exciting and happy night for the students and the excitement dances manv students until morning. [52] The comedian was having a hard time that night. On the stage he told many jokes, but laughed only a few people in the audience. [53] More and more babies are bom in the world every year. Because of this population increase, many problems may occur in the near future. [54] My grandmother is now 97 years old, but she appears to be still healthy and active for her age. [55] During the rainy season, standing near rivers is dangerous because in a short time there mav increase the water level rapidlv. [56] Last night I had a very scary experience. We had a couple of big earthquakes around 1 a.m., and in mv apartment it shattered some windows. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A comparative study of focus constructions

PDF

Beyond words and phrases: A unified theory of predicate composition

PDF

'Beowulf' and 'the hobbit': elegy into fantasy in j. R. R. Tolkien's creative technique

PDF

Asymmetry of scope taking in wh -questions

PDF

Automated postediting of documents

PDF

Arthur Machen’s treatment of the occult and a consideration of its reception in England and America

PDF

An anaphoric approach to clitic position in Spanish

PDF

Andrej Belyj's "Petersburg" and James Joyce's "Ulysses": A comparative study

PDF

A rhetoric of the short story: A study of the realistic narratives of Flaubert, Maupassant, Joyce, and Hyon Chin'gon

PDF

A machine learning approach to multilingual proper name recognition

PDF

A critical edition of the Escorial manuscript of 'historia de los indios de la neuva espana' of fray toribio de benavente (motolinia) (spanish text)

PDF

A study of the technology of color motion picture processes developed in the United States

PDF

"Why doesn't anybody tell them their own mothers have stories?": Representations of mother/daughter relationships in contemporary American fiction

PDF

Clouds of witnesses: A rhetorical analysis of narrated witness in the Gospels.

PDF

Barium 6snd state core excitation using short laser pulses

PDF

Anglo agonistes: English masculinities in British and American film

PDF

A study of the influence of the United States Catholic church on union organizing and community organizing: A historical review, Los Angeles in the 1990s, and future relations

PDF

A case grammar of the parker manuscript of the "Anglo-Saxon chronicle" from 734 to 891

PDF

Causal-integrative model: A management tool for the analysis of cost overruns in major systems acquisition

PDF

A study of the intonation patterns of black and Standard English speaking children in formal and informal situations

Asset Metadata

Creator Oshita, Hiroyuki (author)

Core Title "The unaccusative trap": L2 acquisition of English intransitive verbs.

Degree Doctor of Philosophy

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

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

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Rutherford, Bill (committee chair), Polinsky, Maria "Masha" (committee member), Zubizarreta, Maria Luisa (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-294334

Unique identifier UC11350120

Identifier 9816056.pdf (filename),usctheses-c17-294334 (legacy record id)

Legacy Identifier 9816056.pdf

Dmrecord 294334

Document Type Dissertation

Rights Oshita, Hiroyuki

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