University of Southern California Dissertations and Theses

The syntax of the Chinese BA-constructions and verb compounds: A morphosyntactic analysis

(USC Thesis Other)

The syntax of the Chinese BA-constructions and verb compounds: A morphosyntactic analysis

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. U M I films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter free, while others may be from any type o f computer printer. The quality o f this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margms, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UME a complete manuscript and there are missing pages, these w ill be noted. Also, if unauthorized copyright material had to be removed, a note w ill indicate the deletiorL Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the orignal, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back o f the book. Photographs included in the orignal manuscript have been reproduced xerographically in this copy. Efigher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact U M I directly to order. UMI A Bell & Howell Information Compaiy 300 North Zed) Road, Ann Aibor 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 SYNTAX OF THE CHINESE BA-CONSTRUCTIONS AND VERB COMPOUNDS: A MORPHO-SYNTACTIC ANALYSIS by Ke Zou 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 1995 Copyright 1995 Ke Zou Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9617013 Copyright 1995 by Zou, Ke All rights reserved. UMI Microform 9617013 Copyright 1996, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 90089 This dissertation, written by KE ZOU under the direction of his. 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 Dean Date DISSER TA TIO N COM M ITTEE Chairperson Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This dissertation is dedicated to my parents Yiu-Fong Chau and Shun-Gau Cheng and my wife Ying Liu for love, care, encouragement and support all these years 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS I have been looking forward to this opportunity to formally thank the people who have helped me complete my Ph.D. studies and to whom I own so much academically and personally, because it seems to me that there is no other better chance for me to do so. First of all, I am indebted to all four members of my dissertation committee, Jean-Roger Vergnaud, Audrey Li, Maria-Luisa Zubizarreta and Mario Saltarelli, for the continuous guidance on and support for my dissertation writing. Without their help and encouragement, this work could not have become a reality. Of these four members, I owe my deepest gratitude to Jean-Roger Vergnaud, who is both my advisor and the chair of my dissertation committee. It is not exaggerated to say that he is the best possible dissertation chair one could have chosen. His significant contribution to and scholarly expertise in syntax, phonology, morphology and other areas of linguistics have always been an important inspiration and a great source of linguistic knowledge to me, though, in his typical modesty, he has never tried to impress his views on me with anything but the pure force of his insights and wisdom. During my doctoral studies iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and my dissertation writing at the University of Southern California, he has always been there and available when I needed his advice and help, and he has always taken our discussions as the academic exchanges between colleagues without assuming any posture or attitude of a superior, which makes my student life not only tolerable but also relaxed and enjoyable. Due to his sharp and insightful comments, many of my primitive thoughts and half-baked ideas have successfully been converted into the published papers or the conference presentations. Especially with respect to my dissertation writing, his suggestions have helped me define a framework in which this project could grow and progress, and, at the same time, enabled me to improve the early drafts and produce a final version. He has been, and will always be a role-model to follow in my linguistics career. I am also extremely grateful to Audrey Li for her to introduce me into the field of modern Chinese linguistics and syntax, a privilege which few people have had so far. Her influential contribution to Chinese linguistics, her command of a wealth of Chinese data, her critical review of some simple and partial solutions and the thoroughness of her own linguistic approach and analysis have been and will always be an ideal for me. Besides, her insightful suggestions and comments on my papers and on the earlier iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. drafts of my dissertation have always kept me on the move to check the aspects and areas I have ignored and probe into the questions and problems I have left unanswered or unsolved. Scarcely behind them are Maria-Luisa Zubizarreta and Mario Saltarelli. Their knowledge of general linguistics and Romance linguistics have broadened my eyesight, made me aware of the similarities existing between Chinese and Romance languages and tremendously helped me to study the Chinese language from a cross-linguistic viewpoint and to account for the properties and characteristics of Chinese within the framework of universal grammar. Next, I would like to thank Joseph Aoun whose theory of generalized binding and other important contributions to generative linguistics attracted me to do my doctoral studies at the University of Southern California and has greatly influenced me and clarified my thoughts through my syntactic research. Besides, his sincere support and strong encouragement have helped me get through the most difficult time during my graduate studies. Without his help and advice, I can hardly have completed this crucial chapter of my academic life. My thankfulness also goes to Barry Schein, Bernard Comrie, John Hawkins, Hajime Hoji and Alicja Gorecka. My discussions with them not only passed me the knowledge of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. semantics, language universal & typology, Japanese syntcix and phonology respectively, but also helped me to examine and study languages from different perspectives, without ignoring important information and details. My appreciation extends to other faculty and staff members of the Linguistics Department at the University of Southern California, for the knowledge I have gained from every lecture, talk and discussion and for the help, support and encouragement I have received from them. I am especially grateful to Laura Reiter for her constant help, care and friendship. She called me from thousands of miles away to arrange my attendance at the University of Southern California, which calmed down my anxiety and offered the timely answers and solutions to my questions and problems regarding the student life in Los Angeles. Since I started my studies here, she has given me a lot of help, advice and support in every domain of life, both on and off campus. She is just a perfect graduate studies coordinator and the time she put to advising us graduate students has been more than we could ask for and beyond the call of duty. Thanks also to Kathy Stubaus and Linda Williams-Culver for their secretarial help and clerical support which have made my study and research activities very convenient and efficient. VI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To my fellow graduate students at the University of Southern California, I owned my progress and growth in the field of linguistics through discussing issues and exchanging ideas with them, especially those as follows: Ding-Xu Shi, Patricia Schneider-Ziogna, .Alfredo Arnaiz, Robert Belvin, Elabbas Benmamoun, Kevin Russell, Kaoru Horie, Dong-in Cho, Kyungseun Chang, Marcus Maia, Peter Petrucci, Paul Bruthiaux, Daeho Chung, Liliana Sanchez, Shin Watanabe, Elomari Abdesslam, Hsieh Miao-Ling, Shu- ing Shyu, Nancy Antrim, Jose Camacho, Elordieta GorJca, Xiu-Zhi Wu, Hiroshi Aoyagi, Yuki Matsuda, Keiko Miyagawa and Maki Watancibe. Special thanks go to James Huang and Lisa Cheng for their critical and valuable comments on my papers and on some parts of my dissertation. Their scholarly influence can be felt through many pages of my dissertation. I am also indebted to my teachers at the Ohio State University who have introduced me to the field of formal linguistics, laid a solid linguistics foundation for me, and made me academically ready and prepared to move into advanced linguistics studies. My gratefulness further goes to my colleagues and my friends working within the field of linguistics for their help, advice and cooperation in the past years, including Xiao-Ping Ao, Xiang-Ling Dai, Zheng-Sheng Zhang, Guo-He vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Zheng, Zhi-Ming Zhao, You-Wen Ye, Bao-Zhang He, Guo-Wen Huang, Di Shen, Guo-Ming Sung, Hong-Ming Zhang, Yang Gu, Rint Sijbesma, Dong-Ying Wu, Wen-Hui Zhong, etc. I would like to take this opportunity to express my sincere appreciation and special gratitude to the faculty and staff of the English Department of California State University at Dominguez Hills, especially Agnes Yamada, Burckhard Mohr, Vanessa Wenzell, Tom Giannotti, Ephriam Sando, Janice Kawaminami and Frances Nishikawa, who have provided me with the best teaching schedules and working environment in order for me to have enough time writing my dissertation while teaching full-time. I also want to thank Lyle Smith, Jackson Henry and Selase Williams for their tremendous support of my research activities, which made it possible for me to present my papers at a number of linguistics conferences. Back in P.R. China, I own Sun Zheng-Fang at Jiangxi Normal University for leading me to the field of general linguistics and encouraging me to pursue the studies of linguistics. I am also thankful to the faculty members of Guangzhou Foreign Language Institute for the lectures and courses on linguistics and applied linguistics, which highly motivated me to come to the United States for the doctoral studies. Vlll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. How could I have survived and completed ray doctoral studies without the help, moral support and comradeship from my personal friends within and outside the United States. My little achievement cannot be separated from them either. Because I have too many personal friends, it would be extremely hard for me to mention all of them without missing some of their names. So, I would rather skip the name list and express my thanks to all of them together. Sorry for doing this, my dear friends. Last but not least, I would like to thank my wife, Ying Liu, for her years of love, support, patience, care, and companionship, and for sharing with me all the joys, happiness and frustrations during these years, which have helped me get through this toughest period of time in my life. I also own a great deal to my parents, Yiu-Fong Chau and Shun-Gau Cheng, for their life-long care, advice and support which have made my education and achievement a reality. Besides, I thank my brother Kin Chow and my sister For Cheng for their support and assistance. In a word, this dissertation is as much as my family members' as mine. IX Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS ABSTRACT............................................ XV CHAPTER ONE: INTRODUCTION..... 1 1.1. The Alternations and Grammatical Function Changes. .2 1.1.1. Grammatical Function Changes in BA-constructions....................... 3 1.1.2. Grammatical Function Changes in Resultative Verb Compound Constructions. - 7 1.1.3. Grammatical Function Changes in Directional Verb Compound Constructions.. 9 1.2. Grammatical Function Changes and Morphology...... 11 1.3. Principles and Constraints......... 17 1.4. Conclusion..................................... 25 CHAPTER TWO: THE SYNTAX OF BA-CONSTRUCTIONS........... 29 2.1. Introduction................................... 29 2.2. Types of BA-Constructions........................35 2.2.1. The Objective BA-Construction............. 36 2.2.2. The Causative BA-construction............. 41 2.2.3. The Locative and Instrumental BA-Constructions......................... 44 2.3. The Properties and Constraints................... 46 2.3.1. The Aspectual Features....................46 2.3.2. The Situation Types of Verbs.............. 48 X Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3.3. The BA-NP................................55 2.3.4. The Postverbal NP......................... 61 2.4. The Status of BA.................................64 2.4.1. The Four Previous Analyses................ 64 2.4.2. BA as a Base-Generated Functional Category.................................69 2.5. A Morpho-Syntactic Analysis of BA-Constructions.. .78 2.5.1. Theoretical Background................... 78 2.5.2. The Ob j ective BA-Construction (1)........ 81 2.5.3. The Objective BA-Construction (2)........ 85 2.5.4. The Objective BA-Construction (3)......... 93 2.5.5. The Objective BA-Construction (4)........ 102 2.5.6. The Locative and Instrumental BA-Constructions........................ 106 2.5.7. The Causative BA-Construction ( 1 )........ 109 2.5.8. The Causative BA-Const ruction (2)....... Ill 2.5.9. The Causative BA-Const ruction (3)....... 114 2.6. The Consequences of the Morpho-Syntactic Analysis.......................................116 2.6.1. Negation and the BA-construction......... 116 2.6.1.1. The Function and Distribution of Negators...................... 116 2.6.1.2. Negation in the BA-construction.... 124 2.6.1.3. A Morpho-Syntactic Analysis of Negative BA-Constructions......128 2.6.2. The Selectional Restriction on the Situation Types of Verbs.......... 135 XI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6.3. The Definiteness and Specificity of the BA-NP............................137 2.6.4. The Definiteness Effect of the Postverbal NP....................140 2.7. Conclusion.....................................144 CHAPTER THREE: THE SYNTAX OF RESULTATIVE VERB COMPOUNDS............................ 146 3.1. Introduction.................................. 146 3.2. Types of Resultative Verb Compounds in Chinese... 150 3.2.1. The Agentive Verb Compound................151 3.2.2. The Agentive-Stative Verb Compound....... 153 3.2.3. The Agentive-Transitive Verb Compound.... 154 3.2.4. The Unergative-Transitive Verb Compound... 155 3.2.5. The Unergative-Stative Verb Compound..... 157 3.2.6. The Unergative Verb Compound............. 158 3.2.7. The Unaccusative Verb Compound............159 3.2.8. The Causative Verb Compound.............. 160 3.3. The Properties and Constraints...................163 3.3.1. The Aspectual features....................163 3.3.2. The Ref erentiality Effect.....'............165 3.3.3. The Thematic Relations....................167 3.4. Li's Lexical Analyses of Resultative Verb Compounds...................................... 170 3.4.1. Li's ( 1990) Analysis..................... 170 3.4.2. Li's (1993) Revised Analysis..............177 3.4.3. The Problems with Li's Lexical Analyses.. .180 xii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.5. A Morpho-Syntactic Analysis of Resultative Verb Compounds in Chinese....................... 193 3.5.1. Theoretical Background...................193 3.5.2. The Agentive Verb Compound Construction. ..198 3.5.3. The Agentive-Stative Verb Compound Construction....................204 3.5.4. The Agentive-Transitive Verb Compound Construction....................206 3.5.5. The Unergative-transitive Verb Compound Construction....................212 3.5.6. The Unergative-Stative Verb Compound Construction and the Unergative Verb Compound Construction....................215 3.5.7. The Unaccusative Verb Compound Construction....................219 3.5.8. The Causative Verb Compound Construction............................ 222 3.6. The Consequences of the Morpho-Syntactic Analysis of Resultative Verb Compounds. ...... 231 3.6.1. The Ref erentiality Effects ( 1) ....... 231 3.6.2. The Ref erentiality Effects (2)........... 239 3.7. Conclusion.....................................245 CHAPTER FOUR: THE SYNTAX OF DIRECTIONAL VERB COMPOUNDS............................. 248 4.1. Introduction ............................. 248 4.2. Types of Directional Verb Compounds in Chinese 251 4.3. Properties of Directional Verb Compounds in Chinese. ........ 255 4.3.1. Displacement Verbal Morphemes........... 255 Xlll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3.2. Directional Verbal Morphemes............. 257 4.3.3. Semantic Restrictions................... 259 4.3.4. Distribution of Objects..... 261 4.3.5. Aspectual Feature....................... 265 4.4. A Morpho-Syntactic Analysis of Directional Verb Compounds in Chinese....................... 266 4.4.1. Theoretical Background.................. 266 4.4.2. The Directional V-V Compound Construction (1)........................ 269 4.4.3. The Directional V-V Compound Construction (2)........................ 275 4.4.4. The Directional V—V-V Compound Construction............................280 4.5. The Consequences of the Morpho-Syntactic Analysis of Directional Verb Compounds.......... 285 4.5.1. Co-occurrence between the Verbal Morphemes of Directional Verb Compound....285 4.5.2. Directional V-V Compound and Distribution of Its Object........... 287 4.5.3. Directional V-V-V Compound and Distribution of Its Object........... 295 4.6. Conclusion.................................... 299 REFERENCES......................................... 302 XIV Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT The Chinese BA-constructions and resultative verb compounds are often-discussed topics in Chinese grammar and have attracted a great deal of attention in recent Chinese linguistic studies. The primary reason that the BA-construction has been so interesting to many Chinese linguists is partly due to its structural difference from the canonical SVO word order in Chinese (the word order of the BA-construction is SOV) and partly attributable to some syntactic and semantic constraints imposed upon its verb and other components. The major interest that many Chinese linguists take in the resultative verb compound, which consists of two lexical verbal morphemes and whose second verbal morpheme indicates the result of the action or process expressed by the first verbal morpheme, comes from the following three factors: a) there is a complex thematic relation between the resultative verb compound and its NP arguments; b) the formation of the resultative verb compound may involve a grammatical function change of its NP arguments; and c) the ref erentiality of its NP arguments may affect the interpretation or grammaticality of the resultative verb compound construction involved. XV Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. By contrast, much less serious attention has been paid to the Chinese directional verb compounds, which, like the Chinese resultative verb compounds, also have a complex thematic relation with its arguments and whose formation may involve grammatical function change of its arguments. But unlike the resultative verb compound, the directional verb compound is composed of two or three lexical verbal morphemes, and it may come before its object or have its object occur between its first verbal morpheme and second verbal morpheme or between its second verbal morpheme and third verbal morpheme. This dissertation attempts to discuss various types and essential properties of BA-constructions, resultative verb compounds and directional verb compounds in Chinese, to describe their semantic and syntactic constraints, and to propose a morpho-syntactic analysis of their formation within the minimalist theoretical framework: a) BA is a base-generated functional category and selects an aspect phrase or a resultative phrase as its complement; b) the aspect phrase and resultative phrase are also functional categories and their heads select a VP as complement; c) the verbal morphemes of the resultative and directional verb compounds project and head their own VPs; and d) the asymmetric semantic and "implication" relation between the events being associated with the verbal morphemes is xvi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. "canonically structurally realized" as a complementation relation between the superordinate verbal morpheme and subordinate VP. Under this analysis, the formation of the BA-constructions, the resultative verb compounds and the directional verb compounds simply becomes the side effects of verb-raising and NP-movement without requiring any lexical rules or stipulations, thus meeting with the Economy Principle in spirit. The motivation and argument for the verb-raising and NP-movement are: a) the verb- raising is driven by the morphological necessity that the inflectional features of the verb must be checked against the features of the aspect phrase and agreement phrase in the checking domain of the latter; or the derivation will crash at PF as the strong inflectional feature is not a legitimate object at PF; b) the NP-movement is forced by the Case Filter; and c) the verb-raising and NP-movement are restricted by such principles of universal grammar as the Minimal Link Condition, Shortest Movement Condition and Uniformity Condition. These motivation and argument clearly demonstrate that the proposed morpho-syntactic analysis not only has the ability to generate the well- formed BA-constructions, resultative verb compounds, and directional verb compounds, but also has the capability to rule out ill-formed ones. As a result of this morpho- syntactic analysis, the various semantic constraints and xvii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. syntactic properties of the BA-constructions, resultative verb compounds and directional verb compounds are nicely captured as the natural consequences of the principles of universal grammar, like the negation of BA-constructions, the selection restriction on the situation types of verbs in the BA-construction, the definiteness or specificity of the BA-NP, the definiteness effect from the postverbal NP in the BA-construction, the referentiality of the NP arguments on the interpretation or grammaticality of the resultative verb compounds and the alterable distribution of the objects of the directional verb compounds, etc. XVlll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER ONE INTRODUCTION This dissertation studies the formation of the BA- constructions, resultative verb compound constructions and directional verb compound constructions in Chinese. Through this study, I hope to uncover a deep formal unity behind the formation of the three types of constructions: namely, the alternations and grammatical function changes involved in the formation of these constructions are not attributed to any lexical rules or grammatical function changing rules; they are rather the side effects of verb- movement and NP-movement being driven by the properties of morphology. With this formal unity revealed, various properties of the BA-constructions, the resultative verb compound constructions and the directional verb compound constructions can be naturally explained and the semantic and syntactic constraints upon their component units can be nicely captured as the consequences of the principles of universal grammar without any further stipulations. To introduce this study, I will first present some canonical alternations and grammatical function changes involved in the formation of the Chinese BA-construction, resultative verb compound construction and directional 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. verb compound construction. Then I will characterize a set of important properties shared by these alternations and grammatical function changes. I will also define and motivate the deep formal unity behind these alternations and grammatical function changes: they are just the side effects of verb-raising and NP-movement. Finally, I will introduce an appropriate theoretical framework — which happens to be the Minimalist Theory of Chomsky (1993) — to embed and elaborate this formal unity. The remaining chapters of this dissertation will then go on to furnish this formal unity with a detailed discussion of various types of the BA-constructions, resultative verb compound constructions and directional verb compound constructions and to further develop its theoretical consequences for the analysis of the three types of constructions and for the explanation of the semantic properties and syntactic constraints involved. 1.1. THE ALTERNATIONS AND GRAMMATICAL FUNCTION CHANGES When we explore and examine these alternations and grammatical function changes involved in the formation of the Chinese BA-construction, resultative verb compound construction and directional verb compound construction, 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. we will find that they are highly restricted and that not every process of grammatical function changes is allowed, and the same grammatical function change may not apply to every NP argument of a verb which has such a grammatical function as subject, direct object, indirect object, etc. In the following subsections, I will show some canonical grammatical function changes observed in the formation of the three types of constructions. 1.1.1. Grammatical Function Changes in BA—constructions In one type of BA-constructions which consists of a subject, BA, a preverbal NP (i.e. BA-NP) and a transitive verb (cf. Section 2.1, Chapter 2), the BA-NP is formerly the object of the verb but it goes through a grammatical function change to become the object of BA, as shown by (1) (note: CL = classifier; ASP = aspect marker): (1) wo ba juzi bo—le. I BA orange peel-ASP 'I peeled the orange.' In (1), juzi 'orange' is the logical object of the verb bo 'peel' but it appears to be the surface object of BA. 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. By contrast/ wo 'I', which is the logical subject of the verb, cannot undergo the same grammatical function change to become a BÀ-NP. Neither can juzi be changed into the subject of the BA-construction: (2)*jjuzi ba wo bo-le. orange BA I peel-ASP In another type of BA-constructions which is made up of a subject, BA, a BA-NP, a transitive verb and a post verbal NP (cf. Section 2.2.1, Chapter 2), both the BA-NP and postverbal NP are the logical objects of the verb in the way that the BA-NP is what the verbal action affects, while the postverbal NP acts as the direct target of such an action. However, the BA-NP goes through a grammatical function change to become the object of BA, whereas the postverbal NP remains the object of the verb. Besides, there is an inalienably possessive or part-whole relation between the BA-NP and the postverbal NP: that is, either the BA-NP is an inalienable possessor and the postverbal NP is a possesses or the BA-NP denotes a whole entity and the postverbal NP refers to its part, as shown by (3) and (4): Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (3) wo ba juzi bo-le pi. I BA orange peel-ASP skin 'I peeled the skin of the orange.' (4) Lisi ba vifu reng-le vi iian. Lisi BA clothes throw-ASP one CL 'Lisi threw away one piece of clothes.' In (3), juzi, which is formerly an object of the verb bo and the possessor of the postverbal NP pi 'skin', changes its grammatical function category to be the object of BA. In (4), yifu 'clothes', which is also a former object of the verb reng 'throw' and designates a whole entity, goes through the same process of grammatical function change. Such a grammatical function change is restricted because the application of the same process to the postverbal NP pi in (3) and ji jian 'one piece' in (4) would yield two ungrammatical BA-sentences: (5)*wo ba pi bo-le iuzi. I BA skin peel-ASP orange (6)*Lisi ba vi iian reng-le yifu. Lisi BA one CL throw-ASP clothes In a third type of BA-constructions whose components are also a subject, BA, a BA-NP, a transitive verb and a 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. postverbal NP but whose BA-NP and postverbal NP are the former direct object and indirect object of the verb (cf. Section 2.2.1, Chapter 2), the BA-NP undergoes a process of grammatical function change to appear as the object of BA. However, such a grammatical function change does not apply to the postverbal NP that is the indirect object of the verb, as shown by the contrast between (7a) and (7b); (7) a. ta ba na ben shu gei—le pengyou. he BA that CL book give-ASP friend 'He gave his friend that book.' b.*ta ba pengyou gei-Ie na ben shu. he BA friend give-ASP that CL book Finally, in one type of Causative BA-constructions which consists of a subject, BA, a BA-NP, an intransitive verb and a resultative clause (cf. Section 2.2.2, Chapter 2), the BA-NP is formerly the subject of the resultative clause but it goes through a grammatical function change to be the object of BA (note: DE = resultative marker): (8) Lisi ba shoupai ku-de hen shi. Lisi BA handkerchief cry-DE very wet 'Lisi cried until the handkerchief got very wet.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (8), shoupai 'handkerchief' is originally the subject of the embedded resultative clause, but it appears as the object of BA. 1.1.2. Grammatical Function Changes in Resultative Verb Compound Constructions In one type of the Chinese resultative verb compound constructions (cf. Section 3.2.5, Chapter 3), the object of the verb compound is originally the subject of V2, as shown by (9): (9) Zhangsan xiao—xing—le Lisi. Zhangsan laugh-awake-ASP Lisi 'Zhangsan laughed so loudly that Lisi woke up.' In (9), Lisi is originally the subject of V2 xing 'awake' only, but it undergoes a grammatical function change to become the object of the whole verb compound. In a different type of the resultative verb compound constructions (cf. Section 3.2.7, Chapter 3), the subject of the whole verb compound is originally the object of Vi though it is also the subject of V2, as shown by (10): Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (10) na ke shu kan—dao-le. That CL tree cut-fall-ASP 'That tree was cut and as a result it fell.' In (10), na ke shu 'that tree' is originally the object of Vi kan 'cut' and the subject of V2 dao 'fall', but it goes through a grammatical function change to become the subject of the whole verb compound. In the third type of the resultative verb compound constructions (cf. Section 3.2.8, Chapter 3), the subject of the verb compound is originally the object of Vi while the object of the verb compound is originally the subject of both Vl and V2. There exists a Causer-Causee relation between the subject and the object of the verb compound, as shown by (11): (11) na ping jiu he-zui—le Lisi. that CL wine drink-drunk-ASP Lisi 'Lisi drank that bottle of wine and got drunk.' In (11), na ping jiu 'that bottle of wine' is originally the object of Vi he 'drink' but it becomes the subject of the verb compound through a grammatical function change. By contrast, Lisi, which is originally the subject of Vi he and V2 zui ' drunk ', goes through another greunmatical a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. function change to be the object of the whole compound. These two grammatical function changes are required, or the resulting resultative verb compound construction will be ungrammatical: (12)*Lisi he—zui-le na yi ping jiu. Lisi drink-drunk-ASP that one CL wine 1.1.3. Grammatical Function Changes in Directional Verb Compound Constructions In one type of the Chinese directional verb compound constructions (cf. Section 4.2, Chapter 4), the object of the V-V compound is originally the subject of V2 and the object of Vl, as shown by (13): (13) Lisi dai—shang—le tade maozi. Lisi wear-ascend-JiSP his hat 'Lisi put on his hat.' In (13), tade maozi 'his hat' is formerly the subject of V2 shang 'ascend' and the object of Vi dai 'wear', but it becomes the object of the V-V compound via a grammatical function change. However, the same grammatical function 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. change may not apply to Lisi, which is the former subject of Vl dai: (14)*tade maozi dai-shana-le Lisi. his hat wear-ascend-ASP Lisi In fact, the same grammatical function change also happens to the object of a V-V-V compound in the Chinese directional verb compound construction (cf. Section 4.2, Chapter 4), as shown by (15): (15) ta gian-chu—lai—le yi pi ma. he lead^exit-come-ASP one CL horse 'He brought out a horse here.' In (15), yi pi ma 'a horse' is originally the subject of both V2 chu 'exit' and V2 lai 'come' and the object of Vi qian 'lead', but it becomes the object of the whole V-V-V compound through such a grammatical function change. For the same reason behind the ungrammaticality of (14), this grammatical function change is not available to ta 'he', which is the logical subject of Vi qian 'lead': (16) *yi pi ma aian-chu-lai—le ta. one CL horse lead-exit-come-ASP he 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2. GRAMMATICAL FUNCTION CHANGES AND MORPHOLOGY If we examine the above examples and any other BA- constructions, resultative verb compound constructions and directional verb compound constructions in Chinese, we can find, at least, one morphological property shared by their verbs: that is, the verbs in these constructions all carry an aspect marker, either the perfective aspect morpheme —le, or the progressive aspect morpheme -zhe, or the experiential aspect morpheme -guo ( cf. Section 2.3.1, Chapter 2; Section 3.3.1, Chapter 3; and Section 4.3.5, Chapter 4), or even the resultative "aspect" morpheme -de (cf. Note 25, Chapter 2). In fact, the well-formedness of these three types of constructions is closely related to such aspect morphemes according to the observations of many Chinese linguists (e.g. Chao 1968; Hashimoto 1971; Lu 1973 & 1977; Mei 1978; Li and Thompson 1981; Lii 1984; Lu and Ma 1985; Liu 1987; Ma 1987; Cheng 1988; Fan 1988; Li 1990; Zhang 1991; Liu 1992; Sijbesma 1992; Zou 1993). However, such aspect morphemes are not obligatory on the verb of the corresponding non-BA sentences or sentences containing no verb compound. Moreover, for the Chinese resultative verb compound and directional verb compound, their verbal morpheme components are originally lexical verbs, and these lexical verbs express the same thematic 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relationships with their NP arguments as the whole verb compound does. But these lexical verbs do it in two or three separate clauses, rather than in one clause or one simple sentence, unlike the case of these verb compounds. Hence, the resultative verb compound construction and the directional verb compound construction are, in fact, the thematic paraphrases of the complex sentences which are composed of two or three separate clauses having two or three of these lexical verbs as the heads of their verb phrase (cf. Baker 1988). These two facts then raise the following question: what is the theoretical relationship between the morphological aspects of the BA-construction, resultative verb compound construction and directional verb compound construction and the grammatical function changes involved in their formation, given that the two appear to be associated? Since language is supposed to relate form to meaning systematically (according to the general consensus in the linguistic study) and since grammatical function changes seem to disrupt this association, there must be an overt signal in each of these constructions indicating that the grammatical function changes or other alternations have taken place, and this overt signal will serve as a cue to help to recover the association between form and meaning. This intuitive idea is not something new and has appeared 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in the traditional approaches to the grammatical function changes and other alternations observed in the formation of these constructions. In those traditional approaches, these grammatical function changes and other alternations are the results of the explicit transformation rules that perform the observed switches, such as the transformation rules for the BA-constructions (Wang 1963; Li 1970; Teng 1971; Wang 1973), and the transformation rules for the resultative and directional verb compound constructions (Lu 1977). Each of these transformation rules explicitly stipulates very specific grammatical function changes and alternations involved in the formation of each of these constructions: that is, the object of the verb is altered to be the object of BA in the BA-construction; one verb merges with another verb in the resultative verb compound construction or directional verb compound construction, etc. Though the transformation rules characterize these grammatical function changes and alternations observed, they merely achieve a superficial degree of explanatory adequacy. The real problem with the transformation rules is inherent with the notion of explicit rules themselves. Once we write an explicit rule for grammatical function changes and alternations in the formation of a specific construction in a specific language, we will immediately generate serious questions about the existence of such a 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rule: Why this particular rule, not other rules written for the same construction? How can a child acquire this particular rule in his or her language learning? How is this particular rule related to the general system of the language and to the principles of universal grammar? etc. To be more specific, if the grammatical function changes and the alternations involved in the formation of the BA- constructions, resultative verb compound constructions and directional verb compound constructions are nothing except the results of such explicit transformation rules, little progress could be made on the explanation of their association to the morphological properties of the three types of constructions. In order to capture the relationship between these grammatical function changes and alternations and those morphological aspects of the BA-construction, resultative verb compound construction and directional verb compound construction, I postulate that these grammatical function changes and alternations are not attributed to particular grammatical function changing rules or lexical rules, but they are simply the side effects of verb-raising and NP- movement driven by the morphological aspects of the three types of constructions. This is what I claimed above as a deep formal unity shared by the formation of the three types of constructions. To be more specific, all these 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. grammatical function changes and alternations are related to the morphological features reflected upon the pivotal verb which is the matrix verb in the BA-construction, the second verbal morpheme in the resultative verb compound construction, but the second or third verbal morpheme in the directional verb compound construction. This pivotal verb is generated as the lexical head of an independent verb phrase and then is raised to combine with the aspect morpheme in the BA-construction, assuming a very common practice of taking an aspect morpheme as an inflectional category that occurs higher than a VP in the hierarchical sentence structure. But in the resultative verb compound construction and directional verb compound construction, this pivotal verb is first raised to combine with one or two upper verbs and then moved to combine with the aspect morpheme. Such verb-raising will change the "c-command" relation and other syntactic relations in the structure, which will in turn yield the apparent effects of changing grammatical functions. Accordingly, all other aspects of the syntax of the three types of constructions, including the NP-movement, are expected to follow the principles of universal grammar. As a result, the property of all the grammatical function changes can be explained in terms of verb-raising plus independently motivated principles of syntax without referring to explicit and specific rules. 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In addition, such verb-raising presents two types of consequences for the structure of these constructions: i) it creates a complex X°-level category in the form of V- ASP, V-V-ASP or V-V-V-ASP, resulting in a morphological change; and ii) it builds a syntactic link between two, three or four positions in the phrase marker, creating a syntactic change. Thus, verb-raising makes it possible to explain, in a proper way, how and why the grammatical function changes involve the two. Because verb-raising is subject to the independently motivated constraints on movement (Chomsky 1986b, 1993 & 1994), it is restricted. This will then limit, in an explanatory way, the possible grammatical function changes involved in the formation of these three types of constructions. Hence, the burden of deciding the possibility of a grammatical function change or alternation will be shifted onto these independently motivated principles and constraints, linking the changes and alternations to the general system of Chinese and to the principles of universal grammar. Having articulated and motivated this formal unity behind the formation of the BA-constructions, resultative verb compound constructions and directional verb compound constructions, I would like to implant and elaborate this formal unity in a theoretical framework so as to develop its theoretical consequences for the analysis of these 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. three types of constructions and for the explanation of their semantic properties and syntactic constraints. The framework I adopt in this study is the Minimalist Theory of Chomsky (1993) and others, and it is introduced in the following section with focus upon some of its principles and constraints which are directly relevant to our study. 1.3. PRINCIPLES AND CONSTRAINTS In the framework of Minimalist Theory, the theory of a particular language is its grammar, while the theory of languages and the expressions it generates is Universal Grammar. Universal grammar designates the initial state of the relevant component of the language faculty and it also specifies a set of linguistic levels. Each of these linguistic levels provides the means for presenting some systematic information about linguistic expressions, and a linguistic expression is a sequence of representations at the linguistic levels. A very standard assumption in the Minimalist Theory is that a language consists of two components; namely, a lexicon and a computational system. The lexicon specifies the elements which enter into the computational system, and their thematic properties and idiosyncratic properties, while the computational system 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. uses these elements to produce derivations and linguistic expressions. As a result, the derivation of a particular linguistic expression involves a choice of elements from the lexicon and a computation which creates the sequence of representations at interface linguistic levels. Under this assumption, each language will determine a group of linguistic levels as its formal representations of sound and meaning. Some parts of the computational system will then be relevant only to the level representing meaning, which is called the level of Logical Form (LF), and other parts will be relevant solely to the level representing sound, which is called the level of Phonetic Form (PF). Besides, the parts of the computational system relevant to both levels are called overt syntax, which may include empty categories assigned no phonetic shape. With regard to the computational system, it is governed by a finite set of the invariant principles of universal grammar, and these invariant principles determine whether a derivation or linguistic expression is possible or not: that is, if a derivation creates a legitimate linguistic expression, the derivation converges; but if a derivation does not do so, it crashes. A typical operation of the computational system which links the linguistic levels and the sequence of representations is the movement operation called Move- Alpha, where "alpha" equals either a word-level category 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (i.e. X- or X°-level category) or a phrase-level category (i.e. XP- or X“*^-level category). The category moved by the operation of Move-Alpha is supposed to leave behind a phonetically null copy called trace. Put together, a moved category and its traces constitute a more abstract unit called chain. A chain may display the grammatical function change of a moved category and also preserve its thematic relations. With regard to the finite set of the invariant principles of universal grammar, they actually act as well-formedness conditions on the representations of linguistic expressions, and they consist of X'-Theory, Theta Theory, Case Theory, Binding Theory, Control Theory and Bounding Theory. These principles are introduced and outlined as follows. X-bar theory, which replaces the traditional phrase structure rules and constraints the set of phrase markers allowed (Chomsky 1981; Stowell 1981), decides how lexical elements selected from the lexicon are put together into phrases. In fact, both lexical and functional categories are subject to the following schemata that allows them to head their own projections (cf. Chomsky 1986b): (17) a. X' = X XP* b. X" = XP* X' 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (17), X ranges over the types of categories, the sign of asterisk means that any number of XP can occur cind the order between X, X' and XP are subject to cross-language parametrization. Besides, X in (17a) is called head, XP in (17a) is called complement, and XP in (17b) is called specifier. Two local relations are presented in the X- bar structure: the head-complement relation between X and XP, as in (17a); and the Specifier-head relation between XP and X, as in (17b). The head-complement relation is also related to thematic relations. In this way, X-bar theory defines the Maximal Projection of a category (i.e. jjinax Qj. XP) which is in turn used to define a fundamental structural relation called C-command (Aoun and Sportich 1983; Chomsky 1986b): (18) a C-commands p iff a does not dominate P and every maximal projection that dominates a dominates p. Theta theory is concerned with how semantic/thematic relations are represented in the linguistic structure and it classifies possible semantic relations into thematic roles or theta roles such as Agent, Experiencer, Theme, Goal, Source, Location, etc. The elementary principle of Theta theory is the Theta Criterion (Chomsky 1981) which 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. biuniquely requires that only one theta role be assigned to one argument position: (19) Each argument bears one and only one theta—role, and each theta—role is assigned to one and only one argument. The Theta Criterion also holds of a chain created by the movement of an argument. However, according to Hale and Keyser (1993), thematic roles are actually derivative of lexical syntactic relations and the argument structure is determined by lexical categories and their projections. This is why there are so few thematic roles, as thematic roles are nothing but the syntactic relations which are determined by lexical categories and their projections, and which are limited by the small inventory of lexical categories (namely, V, N, A, P) and by the principles of Unambiguous Projection and Full Interpretation (cf. Kayne 1984; Chomsky 1986a). Case theory involves the assignment of abstract Case into categories in Case-marking positions. There are two types of Case: structural Case and inherent Case (Chomsky 1986a). The basic principle of Case theory is Vergnaud's Case Filter (Rouveret and Vergnaud 1980; Chomsky 1981): 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (20) *NP, if NP has phonetic content but no Case. The Case Filter helps to identify how NPs are interpreted in the linguistic structure. So it also holds of a chain created by moving an NP, thus making the NP "visible" to the application of both the phonological rules and the LF rules that assign theta roles. This can be shown by the Visibility Condition (Aoun 1980; Chomsky 1986a); (21) An NP in the head position of a chain (i.e. the last position of a moved category) can only bear a theta role if it is assigned Case. In the Minimalist Theory, the structural Case is assigned via the Spec-head relation in terms of X-bar structure. Bounding theory specifies locality conditions on the operations of movement, and its basic principles are the Shortest Movement Condition (Chomsky 1993) in (22a) and the Minimal Link Condition (Chomsky 1994) in (22b): (22) a. If a and P are in the same minimal domain, they are equidistant from y. b. Move—a attaches a to the nearest target (in the already formed structure) that immediately C— commands a. 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (22a), a and p refer to two targets of movement while Y refers to a position occupied by a category which is a potential candidate for movement. Besides, the minimal domain refers to the domain created by the head-movement (i.e. X°-movement) which links the heads of two XPs, and the minimal domain consists of the specifiers of the two XPs and the complement of the lower XP. The operation of movement is also constrained by the Uniformity Condition on the links of a chain (Chomsky 1994): (23) A chain is uniform with regard to phrase structure status. This principle, in fact, states that only X° (X) can move to the head position, and only (XP) can move to the specifier position. In addition, the movement operation is governed by two economy principles: the Procrastinate Principle and the Greed Principle. The former principle states that non-overt LF-movement is less costly or much "cheaper" than overt movement, and the latter principle requires that an element go through movement only if its morphological properties cannot be otherwise satisfied. Binding theory deals with the relations of anaphors and pronominals to their antecedents. The fundamental 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. principles of Binding theory can be cast in the following interpretive manner (Chomsky 1993): (24) a. If a is an anaphor, interpret it as being co- ref erential with a C—commanding phrase in D. b. If a is a pronominal, interpret it as being disjoint from every C-commanding phrase in D. c. If a is an R-expression, interpret it as being disjoint from every C—commanding phrase. In (24) anaphor refers to elements such as reflexives and reciprocals, pronominal refers to pronouns, R-expression refers to referential expressions or denoting expressions and D refers to the relevant local domain. Control theory determines the choice of antecedents for empty pronominal elements such as PRO. Because empty pronominal elements are pronominals, they must always be C-commanded by their antecedents, as required by Binding theory (Chomsky 1981). The principles, constraints and rules outlined above will be referred to and used in the remaining chapters of the dissertation. Other principles and constraints which are related or relevant to the following discussions will be introduced when and where they are needed. 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.4. CONCLUSION In this chapter, I have discussed a set of canonical alternations and grammatical function changes involved in the formation of the Chinese BA-construction, resultative verb compound construction and directional verb compound construction. I have also characterized the relationship between the morphological properties of these three types of constructions and the grammatical function changes and alternations, which actually shows a formal unity shared by the formation of these three types of constructions; these alternations and grammatical function changes are not attributed to specific grammatical function changing rules or lexical rules, but are the side effects of verb- raising and NP-movement being driven by the morphological properties of the three types of constructions. In order to embed and elaborate this formal unity in a theoretical framework to develop its theoretical consequences for the analysis of these constructions, I have introduced the Minimalist Theory of Chomsky (1993) and others, and also outlined its basic concepts, assumptions and principles. In the remaining chapters of this dissertation, I will go on to furnish this formal unity by presenting a detailed study of the BA-constructions, resultative verb compound constructions and directional verb compound constructions 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and to apply these universal principles to the analysis of these constructions and to the explanation of their semantic and syntactic properties. The organization of the remaining chapters is as follows. Chapter 2 discusses types of the BA-constructions and their basic properties, describes the semantic and syntactic constraints imposed upon their verb and other components and explores the status of BA from synchronic, diachronic and cross-linguistic perspectives. Based on these discussions and investigations, a morpho-syntactic analysis is proposed for the BA-constructions: i) BA is a base-generated functional category and it selects an aspect phrase or a resultative phrase as its complement; and ii) the aspect phrase and the resultative phrase are also functional categories whose heads select VP as their complement. This morpho-syntactic analysis connects the morphological properties of the BA-constructions to the grammatical function changes involved in their formation, and it derives the BA-constructions as the side effects of verb-raising and NP-movement without resorting to any rules or stipulations. This analysis also accounts for other properties of the BA-constructions and the semantic and syntactic constraints upon their components, such as the negation of BA-constructions, the restriction on the situation types of verbs, the definiteness or specificity 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. requirement on the BA-NP and the definiteness effect from the postverbal NP. Chapter 3 studies various types of resultative verb compounds in Chinese and describes their properties and the semantic and syntactic constraints imposed on verbal morphemes and their NP arguments. It also reviews Li's (1990 & 1993) lexical analyses of these resultative verb compounds. Based on this study, an alternative morpho- syntactic analysis is suggested: i) two lexical verbal morphemes within the resultative verb compound project and head their own VPs; and ii) the asymmetric semantic and "implication" relation between the two events denoted by the two verbal morphemes is "canonically structurally realized" as a complementation relation between the first verbal morpheme and the VP headed by the second verbal morpheme. This morpho-syntactic analysis also links the morphological properties of the resultative verb compound construction to the grammatical function changes involved and derives the resultative verb compound construction as the side effects of verb-raising and NP-movement without referring to any specific rules or stipulations, exactly parallel to the analysis of the BA-constructions proposed in Chapter 2. Moreover, this analysis incorporates the complex thematic relations between the resultative verb compounds and their arguments, accounts for the semantic 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and syntactic constraints on their verbal morphemes and NP arguments, and explains the referentiality effect of the NP arguments on the interpretation or grammaticality of the resultative verb compound constructions. Finally, Chapter 4 examines types and properties of directional verb compounds in Chinese and it extends the morpho-syntactic analysis proposed in Chapter 3 to these directional verb compounds. Like the derivation of the Chinese resultative verb compounds, this morpho-syntactic analysis also derives these directional verb compounds as the side effects of verb-raising and NP-movement without resorting to any specific rules. It further incorporates the complex thematic relations between these directional verb compounds and their arguments and accounts for the semantic restriction on the co-occurrence of their verbal morphemes. As an important consequence of this analysis, the distribution of the objects of the directional verb compounds is captured in a principled way. 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER TWO THE SYNTAX OF BA-CONSTRUCTIONS 2.1. INTRODUCTION The BA-construction is a widely-discussed topic in Chinese grammar and has drawn a great deal of attention in previous and recent Chinese linguistic studies.^ The ^ One interesting fact about the previous studies of the BA-construction is that they have characterized the BA-construction with so many names in order to describe and capture its major semantic and syntactic properties. For example, Wang (1954), Lii (1955) and Chao (1968) call it a disposal construction: "The disposal construction states how a person is handled, manipulated, or dealt with; how something is disposed of, or how an affair is conducted. As it is specifically designed for disposing, the disposal sentence cannot be used unless the action possesses the quality of disposal." (Wang Li 1954:161) Frei (1956) refers to it as an ergative construction: "Syntactically, the BA-construction is, like any other prepositional complement in this position (preverbal), an adverbial determinant of the following verb, in spite of the translation in our western languages, by an object or a subject. This feature seems to be common, through many variations, to all ergative languages." (Frei 1956:46) Hashimoto (1971) regards it as an executive construction: "The 'executive' construction is marked by the formative BA, an object NP with definite reference, and, usually, a modified VP. Since in the surface structure the object NP occurs immediately after BA and before its verb, this sentence type has often been described as object permuta tion caused by BA where the verb is modified by certain elements." (Hashimoto 1971:65) Teng (1977) treats it as an accusative construction, in the sense that its object (the NP following BA) must be a Patient. However, in the recent studies, people take the BA-construction as it is, without using any other names (e.g. Li and Thompson 1981; 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reason that the BA-construction has been very interesting to many Chinese linguists is partly due to its structural difference from the canonical SVO word order in Chinese^ and is partly attributable to some syntactic and semantic constraints imposed upon its verb and other components. The canonical BA-constructions are exemplified as follows (note: ASP = aspect marker): (1) wo ba juzi bo-le. I BA orange peel-ASP 'I peeled the orange.' (2) wo ba juzi bo-le pi. I BA orange peel-ASP skin 'I peeled the skin of the orange.' In (1), the preverbal NP (i.e. BA-NP) juzi 'orange' is the logical object of the verb bo 'peel' but appears to be the surface object of BA. In (2), both the BA-NP juzi and the postverbal NP pi 'skin' are the logical objects of the verb bo, in the sense that juzi is what the verbal action affects or happens to, and pi is the direct target Huang 1982; Goodall 1987; Cheng 1986; Tsao 1987; Wang 1987; Li 1990; Sijbesma 1992). ^ The word order of the canonical BA-constructions is SOV, as shown by (1) and (2) above. 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of such a verbal action. But at the surface structure of (2), juzi appears to be the object of BA while pi remains the object of the verb. In recent generative linguistic studies, there are several noteible analyses of BA-constructions, as proposed by Huang (1982), Cheng (1986), Goodall (1987), Li (1990), and Sijbesma (1992) respectively. According to Huang's analysis, which is based on Thompson's (1973) original analysis, the BA-NP juzi in (1) and (2) is base-generated postverbally as an outer object of the verb, it receives a 6-role from V', and it moves to the preverbal position to get Case from BA; BA is an inserted dummy Case marker; and the postverbal NP pi in (2) is base-generated as the inner object of the verb and assigned a 6-role and Case by the verb (note: (I'a) and (2'a) = D-structure; (I'b) and (2'b) = S-structure): (1') a. [g wo [yp [y, [y bO-lC] jUZl]]] I peel-ASP orange b . [g wo [y p ba juzii [y, [ y bo-le] t ^ J ] ] I BA orange peel-ASP (2') a. [g wo [yp [y, ( y bo~le pi] juzi]]] I peel-ASP skin orange b. [g wo [y p ba juzii [y, [ y bo-le pi] t^]]] I BA orange peel-ASP skin 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Goodall^s analysis of (1) is somewhat similar to Huang's analysis, but his analysis of (2) differs from the latter in the way that the BA-NP juzi is treated as the specifier of the postverbal NP pi at D-structure (cf. (2"a) ) and it moves to the specifier position of VP at S- structure (cf. (2"b)): (2") a. Is w o [yp [y, bo-le juzi [„, p i]]]]] I peel-ASP orange skin b . Is w o [yp ba ju zii [y, bo-le [^ t ^ [^, p i]]]J] I BA orange peel-ASP skin In Li's analysis, the BA-NP ju zi in both (1) and (2) is base-generated in the preverbal position (cf. (1"'a) and (2"'a); the BA-phrase (i.e. BA and the BA-NP) in (1) receives a 0-role from the verb, and the BA-phrase in (2) receives a 8-role from V'; BA is treated as a preposition and assigns Case to the BA-NP; and the postverbal NP pi in (2) is base-generated to the left of the verb, gets a 0-role from the verb and moves to the postverbal position to get Case from the verb under the assumption that the Case assignment in Chinese is to the right (cf. (2"'b): (1”') a. & b. [g w o [yp [y, [pp ba juzi] [y bo— le ]]]] I BA orange peel-ASP 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (2"') a. [g wo [yp [pp ba juzi] [y, pi [y bo-le]]]] I BA orange skin peel-ASP b . [ g wo [ y p [ p p ba juzi] [ y , [ y bo-le] pi]]] I BA orange peel-ASP skin Cheng's analysis is almost the same as Li's analysis in treating BA as a preposition, but it differs from the latter by taking BA as a 0-role assigner which identifies its 6-role with the internal 8-role of the verb and then assigns the identified 8-role to the BA-NP. Departing from the analyses above, Sijbesma takes the BA-construction as a causative construction (CAUSP), the head of which (CAUS) is filled with the inserted BA; the subject of the BA-sentence is semantically dependent on the head of CAUSP in the sense that it causes the BA- NP to undergo the event denoted by the VP; the VP is just embedded under CAUS and is ergative without an external argument; and the BA-NP is originated in the postverbal position and it is moved to the preverbal position, as shown by (l""b) and (2""b) below: (l”"b) [CAUSP [CAUSP [CAUSE IvP IvP Iv I BA orange peel-ASP [» ti X]]]]] 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (2""b) wo [cAusp [CAUSE Ivp lyp Iv bo—le] I BA orange peel-ASP [%P ti [i, pi ]]]]]] skin The purpose of this chapter is tc discuss various types of BA-constructions and their basic properties, to describe the semantic and syntactic constraints imposed on the verb and other components of BA-constructions, to explore the status of BA, and to propose and argue for a morpho-syntactic analysis of BA-constructions as follows; i) BA is a base-generated functional category and selects an aspect phrase or resultative phrase as its complement; and ii) the aspect phrase and the resultative phrase are also functional categories whose heads select VP as their complement. Under this morpho-syntactic analysis, the formation of various BA-constructions simply becomes the side effects of verb-raising and NP-movement without any rules or stipulations. The verb-raising is driven by the morphological necessity that the aspectual feature or the resultative feature reflected on the verb must be checked against the head of the aspect phrase or the resultative phrase, and the NP-movement is forced by the Case Filter. They are both constrained by the principles of universal grammar. I will show how this morpho-syntactic analysis 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. captures the basic properties of BA-constructions and how it accounts for the semantic and syntactic constraints on the verb and other components of BA-constructions, the explanation of which may not be available under the other analyses. I will also provide empirical motivation and theoretical arguments for this morpho-syntactic analysis. Finally, I will present several important consequences of this morpho-syntactic analysis, concerning the negation of BA-constructions, the selectional restriction on the situation types of verbs, the requirement of definiteness or specificity on the BA-NP, and the definiteness effect demonstrated by the postverbal NP in the BA-construction. 2.2. TYPES OF BA-CONSTRUCTIONS In terms of the syntactic and semantic relations between the BA-NP and the verb and between the BA-NP and the postverbal NP, we may divide BA-constructions into three types: i) objective, ii) causative, iii) locative & instrumental (cf. Lii 1984; Lu and Ma 1985; Tsao 1987). Of the three types of BA-constructions, the objective and causative BA-constructions are the canonical ones, while the locative and instrumental BA-constructions are much less often used in the Modern Chinese (Wang 1987). These 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. three types of BA-constructions and their properties are going to be discussed in the following sections. 2.2.1. The Objective BA-Construction For the sake of convenience, we may also divide the Objective BA-construction into four subtypes, regarding the type of verb, the number and type of object, and the relation between the objects. One subtype of Objective BA-constructions consists of a subject, BA, a BA-NP, and a transitive verb. The BA-NP is the logical object of the verb but appears to be the surface object of BA, as shown by the following examples (note: CL = classifier): (3) wo ba juzi bo-le. J BA orange peel-ASP 'I peeled the orange.' (4) ta ba na jian chenyi xi—le. he BA that CL shirt wash-ASP 'He washed that shirt.' (5) Lisi ba tade che mai—le. Lisi BA his car sell-ASP 'Lisi sold his car.' 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The second subtype of Objective BA-constructions is composed of a subject, BA, a BA-NP, a transitive verb and a postverbal NP. Both the BA-NP and the postverbal NP are the logical objects of the verb in the sense that the BA-NP is what the verbal action affects or happens to and the postverbal NP is the direct target of such an action. But at the surface structure, the BA-NP appears to be the object of BA, while the postverbal NP remains the object of the verb. Besides, there is an inalienably possessive or part-whole relation between the BA-NP and postverbal NP: that is, either the BA-NP is an inalienably possessor and the postverbal NP is a possessee or the BA-NP denotes a whole entity and the postverbal NP refers to its part, as exemplified by (6b), (7b), (8b) and (9b) below:^ (6) a. wo bo-le iuzir-de) pi. I peel-ASP orange('s) skin 'I peeled the skin of the orange.' ^ By the inalienably possessive/part-whole relation between the BA-NP and the postverbal NP, we mean that the two NPs are inherently related to each other before the verbal action takes place. For example, in (6b) the skin is a part of the orange no matter whether one peels it or not. This inherent relation is also reflected in syntax, as demonstrated by the movement relation between (6a) and (i) below (Cheung 1973:379): (i) wo ba iuzif-de) pi bo-le. I BA orange('s) skin peel-ASP 'I peeled the skin of the orange.' 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. wo ba juzi bo-le pi. I BA orange peel-ASP skin 'I peeled the skin of the orange.' (7) a. ta bang-le Lisif—del liana zhi iiao. he tie-ASP Lisi('s) two CL foot b. ta ba Lisi bang—le liana zhi iiao. he BA Lisi tie-ASP two CL foot 'He tied up Lisi's two feet.' (8) a. Lisi reng-le yi iian yifu. Lisi throw-ASP one CL clothes b. Lisi ba vifu reng—le vi iian. Lisi BA clothes throw-ASP one CL 'Lisi threw away one piece of clothes.' (9) a. Mali fan-le ji ve zazhi. Mary turn-ASP several pages magazine b. Mali ba zazhi fan—le j_i_____ve. Mary BA magazine turn-ASP several pages 'Mary turned over several pages of the magazine. ' The third subtype of Objective BA-constructions is also composed of a subject, BA, a BA-NP, a transitive verb and a postverbal NP. But unlike the second subtype, the BA-NP and the postverbal NP in the third subtype are originally the direct object and the indirect object, the verb is a ditransitive verb, and there is no inalienably 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. possessive or part-whole relation between the BA-NP and the postverbal NP, as shown by (10), (11) and (12):* (10) a. ta gei-le pengyou na ben shu. he give-ASP friend that CL book b. ta ba na ben shu gel—le pengyou. he BA that CL book give-ASP friend 'He gave his friend that book.' (11) a. wo gaosu—le Lisi zhe jian shi. wo tell-ASP Lisi this CL matter b. wo ba zhe jian shi gaosu—le Lisi. I BA this CL matter tell-ASP Lisi 'I told Lisi this matter.' (12) a. wo song-le ta na ge qianbao. I give-ASP him that CL wallet b. wo ba na ge qianbao song-le ta. I BA that CL wallet give-ASP him 'I gave him that wallet.' Take (10b) for instance, in which the postverbal NP pengyou 'friend' was not the possessor of the BA-NP na ben shu 'that book' until it was given to him or her, and these two NPs cannot be moved together to the preverbal position, as shown by the contrast between (10a) and (i): (i) *ta ba pencryou-de na ben shu gei-le. he BA friend's that CL book give-ASP 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The fourth subtype of Objective BA-constructions has the same number of components as the second subtype, and its BA-NP remains what the verbal action happens to. But this subtype is different from the second subtype in two aspects: i) its postverbal NP indicates the result of the verbal action; and ii) no inalienably possessive or part- whole relation exists between the BA-NP and postverbal NP, as exemplified by (13), (14) and (15) below:^ (13) a.*ta xie-Ie na ge shif-de) vi pian baodao. he write-ASP the CL matter('s) one CL report b. ta ba na ge shi xie-le yi plan baodao. he BA the CL matter write-ASP one CL report 'He wrote a report about that matter.' (14) a.*wo bao-le vifuf-de) vi ae xiaobao. I pack-ASP clothes('s) one CL bundle b. wo ba yifu bao—le yi ge xiaobao. I BA clothes pack-ASP one CL bundle 'I packed the clothes into a bundle.' ^ The BA-NP and the postverbal NP in this subtype of Objective BA-constructions are not related to each other until the verbal action takes place: that is, it is the verbal action that brings the BA-NP into the state being indicated by the postverbal NP. Besides, these two NPs cannot appear together in the preverbal position either, as shown by the contrast between (13b) and (i) below: (i) *ta ba na ge shi(—de) yi pian baodao xie—le. he BA the CL matter('s) one CL report write-ASP 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (15) a.*Lisi hua—le beitunaf—de) liliana. Lisi change-ASP sadness('s) force b. Lisi ba beitung hua-le liliang. Lisi BA sadness change-ASP power 'Lisi changed his sadness into power.' 2.2.2. The Causative BA-construction The Causative BA-construction could also be divided into three subtypes in terms of the thematic relations between its subject, the BA-NP and the verb. The first subtype of Causative BA-constructions is composed of a subject, BA, a BA-NP and an unaccusative verb.® The BA- ® In Chinese, some verbs can be used as either un accusative verbs or intransitive verbs, depending on the meaning involved. For example, when pao and zou are used intransitively, pao means 'run' and zou means 'walk'. But when they are used unaccusatively, pao is interpreted as 'escape' and zou is interpreted as 'leave', as shown by the following two sets of examples from Gu (1992:78): (i) a. Zhangsan zai jieshang pao. Zhangsan at street run 'Zhangsan is running in the street.' b. Zhangsan zai lushang zou. Zhangsan at road walk 'Zhangsan is walking on the road.' (ii) a. laoli pao-le yi ge fanren. jail run-ASP one CL prisoner 'A prisoner escaped from the jail.' b. Zhangsan jia zou-le yi ge keren. Zhangsan home walk-ASP one CL guest 'A guest left Zhangsan's home.' 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. NP is the logical subject of the verb phrase, while the verb phrase does not have any thematic relation with the matrix subject. Besides, there exists a Causer-Causee relation between the matrix subject and the BA-NP, in the sense that the former causes something to happen to the latter, as shown below:^ (16) na ge kanshou ba yi ge zei pao-le. that CL warden BA one CL thief run-ASP 'That warden let a thief escape.' (17) na chang yundong ba yi ge da jie si—le. that CL campaign BA one CL old sister die-ASP 'That campaign made an old sister die.' (18) zhe jian shi ba ta pacheng-le na ge yang. this CL matter BA him fear-ASP that CL shape 'This matter make him fear like that.' The second subtype of Causative BA-constructions is made up of a subject, BA, a BA-NP, an unaccusative verb, and an embedded resultative clause. The BA-NP serves as the logical subject of both the matrix verb phrase and the resultative clause, while the matrix subject has no ^ This type of Causative BA-constructions is usually used to express some unexpected and unpleasant event(s), according to Wang (1954) and Lü (1984). 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. thematic relation with the matrix verb phrase. There is still a Causer-Causee relation between the matrix subject and the BA-NP, as displayed by (19), (20) and (21) below (note: DE = resultative marker): (19) zhe jian shi ba wo ji—de shui bu hao. this CL matter BA I hurry-DE sleep not well 'This matter made me so anxious that I couldn't sleep well.' (20) na chang xue ba ta dong-de zhi duosuo. that CL snow BA he chill-DE constantly shiver 'The snow made him so cold that he shivered constantly.' (21) zhe ping jiu ba Lisi zui—de zhan-bu-qilai. this CL wine BA Lisi drunk-DE cannot-stand-up 'This wine got Lisi so drunk that he couldn't stand up. ' The third subtype of Causative BA-constructions is composed of a subject, BA, a BA-NP, an intransitive verb, and a resultative clause. The BA-NP acts as the logical subject of the resultative clause, and the matrix subject is the logical subject of the matrix verb phrase. There still remains a Causer-Causee relation between the matrix subject and the BA-NP: 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (22) Lisi ba shoupai ku—de hen shi. Lisi BA handkerchief cry-DE very wet 'Lisi cried until the handkerchief got very wet.' (23) ta ba houlong han-de dou ya—le. he BA throat shout-DE even hoarse-ASP 'He shouted until his throat even got hoarse.' (24) Lisi ba women xiao—de mo—ming—qi—miao. Lisi BA we laugh-DE confused 'Lisi laughed and as a result we all got confused. ' 2.2.3. The Locative and Instrumental BA—Constructions The composition of the Locative and Instrumental BA- constructions is almost like that of the fourth subtype of Objective BA-constructions, except that the BA-NP of the former refers to either the location of the verbal action or the instrument with which the verbal action is performed or carried out, and its postverbal NP serves as the direct object of the verb. There is no inalienably possessive or part-whole relation between the BA-NP and the postverbal NP, and it is just the verbal action that 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relates the two NPs together, as shown by the following examples (cf. Cheung 1973);® (25) Lisi ba xinfeng tie—le youpiao. Lisi BA envelope paste-ASP stamp 'Lisi pasted stamps on the envelope.' (26) ta ba huaping cha—le yi ba hua. he BA vase insert-ASP one CL flower 'He inserted a bunch of flower into the vase.' (27) wo ba guo zhuang-le dian shui. I BA pot put-ASP little water 'I put some water in the pot.' (28) ta ba qiangkou dui—zhe ren. ta BA muzzle aim-ASP man 'he aimed at the people with his gun.' (29) wo ba shu fangman-le shujia. I BA book fill-ASP bookshelf 'I filled the bookshelf with books.' ® There might exist a conceptual possessive or part- whole relation between the BA-NP and the postverbal NP in the Locative and Instrumental BA-constructions (cf. Chu 1966). Besides, the use of the instrumental BA-construction, which shows the residue of the early Modern Chinese, is quite limited in the Modern Chinese (Wang 1987). Hence, some native speakers of Chinese may find this type of BA- constructions quite odd. 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3. THE PROPERTIES AND CONSTRAINTS 2.3.1. The Aspectual Features It has been noted by many linguists (e.g. Hashimoto 1971; Mei 1978; Lii 1984; Lu and Ma 1985: Cheng 1986 and 1988; Li 1990; Liu 1992; Sijbesma 1992) that the well- formedness of the BA-construction is closely related to the aspectual feature reflected on its verb. Normally^ the verb in the BA-construction needs to take either the perfective aspect marker -le or the progressive aspect marker -zhe (see the examples in Section 2.2. ), but such aspect markers are optional in the corresponding non-BA sentences:® (30) a. wo ba ta ma-le. I BA him scold-ASP 'I scolded him.' ® Notice that -le can be either a perfective aspect marker or an inchoative marker. Their major differences are: a) the perfective aspect marker is always attached to a verb/ while the inchoative marker always appears at the sentence-final position; and b) the perfective aspect marker indicates the completion of an action or presents a closed event, whereas the inchoative marker signifies a change of state or a currently relevant state (Li and Thompson 1981; Smith 1991). 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b.*wo ba ta ma. I BA him scold c. wo ma(-le) ta. I scold(-ASP) him (31) a. ni ba na feng xin dal—zhe! you BA that CL letter carry-ASP 'You carry this letter!' b.*ni ba na feng xin dai! you BA that CL letter carry c. ni dai(—zhe) na feng xin! you carry(-ASP) that CL letter The property shared by the perfective aspect —le and the progressive aspect marker -zhe is that they delimit the event or action conveyed by the verb and present specific or definite situation (Li and Thompson 1981; Cheng 1988). To be more specific, the perfective aspect indicates the completion of a specific action or closure of a definite event, while the progressive aspect signifies a change of state being resulted from the completion of a specific action or from the closure of a definite event. ^ Smith (1991) points out that in Chinese, a verb hosting a progressive marker can be ambiguous between a stative reading and an accomplishment reading. While the stative reading conveys a change of state which has taken place already, the accomplishment reading refers to a change of state that is just brought about. Of these two 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3.2. The Situation Types of Verbs The well-formedness of the BA-construction is also closely related to the situation type or situation aspect of its verb:^^ only verbs that present a situation with an initial point and a natural final point indicating a change of state, a completion of action or a closure of event can occur in the BA-construction (cf. Smith 1991; Liu 1992). Of the four situation types of verbs {stative verbs, activity verbs, accomplishment verbs, achievement verbs), stative verbs describe stable situations without dynamics or internal stages and do not present a natural final point indicating a change of state, a completion of action or a closure of event. Therefore, stative verbs cannot occur in the BA-construction, even if they are not prohibited from the corresponding non-BA sentences: readings, only the accomplishment reading is perfactively presented, which is exactly the reading when a verb hosts —zhe in the BA-construction. The situation aspect of verb is often classified into four types: i.e. state, activity, accomplishment and achievement. The verbal constellations of the four types are stative verbs, activity verbs, accomplishment verbs and achievement verbs (cf. Dowty 1979; Smith 1991). This is also noted by the other studies of BA- constructions under different terms, as in Wang (1954): that is, the disposal property of the BA-constructions requires their verbs to be dynamic. Notice that verbs like wang 'forget' are not stative verbs because they have a natural final point indicating 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (32) a.*wo ba na ben shu you—le. J BA that CL book own-ASP b. wo you na ben shu. I own that CL book 'I have that book.' (33) a.*ta ba bciba xiana-le. he BA father resewble-ASP b. ta xiang baba. He resemble father 'He resembles his father.' (34) a.*wo ba zhe ge wanju xihuan—le. I BA this CL toy like-ASP b. wo xihuan zhe ge wanju. I like this CL toy 'I like this toy.' (35) a.*wo ba zhe jian shi zhidao-le. I BA this CL matter know-ASP b. wo zhidao zhe jian shi. I know this CL matter 'I know this matter.' a change of state (for wang, the state of remembering is changed to the state of not remembering): (i) wo ba na jian shi wang-le. I BA that CL matter forget-ASP 'I forgot that matter.' 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stative verbs may also include psychological verbs and perception verbs which describe psychological states or perception states. Therefore, these psychological and perception verbs could not be used in the BA-construction either, though they are not barred from the corresponding non-BA sentences However, if a psychological or perception verb forms a V-V compound with a resultative or directional verb, then it can appear in the BA-construction: (i) ta ba wo hen-tou—le. he BA me hate-thorough-ASP 'He hated me thoroughly.' (ii) Lisi ba Mali ai-shang—le. Lisi BA Mary love-ascend-ASP 'Lisi fell in love with Mary.' (iii) Lisi ba na bu dianying kan-wan-le. Lisi BA that CL film see-finish-ASP 'Lisi finished that movie.' (iv) wo ba na shou ge ting-wan-le. I BA that CL song hear-finish-ASP 'I finished listening to that song.' This is due to the fact that compounding a psychological or perception verb with a resultative or directional verb changes the situation type of the former: that is, the V- V compound is not a stative verb any more, as it presents a situation with an initial point and a final point that signifies a change of state, a completion of action or a closure of event (Smith 1991; Liu 1992). However, as noted by Li (1990), the resultative verb tou 'thorough' can compound with hen 'hate' but not with ai 'love', though they are both psychological verbs: (V) *wo ba ta ai-tou-ASP. I BA him love-thorough-ASP (vi) *wo ai-tou-le ta. J love-thorough-ASP him Similarly, the directional verb shang 'ascend', which may form a compound with ai, cannot form a compound with hen, as shown by the following contrast: 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (36) a.*Llsl ba Mali ai-le. Lisi BA Mary love-ASP b. Lisi ai Mali. Lisi love Mary 'Lisi loves Mary.' (37) a.*wo ba Lisi hen—le. I BA Lisi hate-ASP b. wo hen Lisi. I hate Lisi 'I hate Lisi.' (38) a.*Lisi ba na bu dianying kan—le. Lisi BA that CL film see-ASP b. Lisi kan—le na bu dianying. Lisi see-ASP that CL film 'Lisi saw that movie.' (39) a.*wo ba na shou ge ting-le. I BA that CL song hear-ASP b. wo ting-le na shou ge. I hear-ASP that CL song 'I listened to that song.' (vii) Lisi ba Mali ai-shang-le. Lisi BA Mary love-ascend-ASP 'Lisi fell in love with Mary.' (viii)*Lisi ba Mali hen-shang-le. Lisi BA Mary hate-ascend-ASP (ix) *Lisi hen-shang-le Mali. Lisi hate-ascend-ASP Mary 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Besides stative verbs, pure activity verbs are not compatible with the BA-construction either, though they are not barred from the corresponding non-BA sentences. This is because activity verbs only depict atelic events which have homogenous successive stages but no natural final point indicating a change of state, a completion of action or a closure of event. In other words, activities may terminate or stop arbitrarily, but they do not finish naturally (Smith 1991; Liu 1992) (40) a.*wo ba lu zou—le. I BA road walk-ASP b. wo zou-le lu. I walk-ASP road 'I walked on the road.' Like psychological and perception verbs, activity verbs can appear in the BA-construction if they form V-V compounds with resultative or directional verbs: (i) wo ba lu zou-wan le. I BA road walk-finish-ASP 'I walked through the road.' (ii) Lisi ba fayu xue-hao-le. Lisi BA French learn-well-ASP 'Lisi has a good command of French.' (iii) ta ba che tui-dao-le. he BA cart push-fall-ASP 'He pushed a cart and as a result the cart fell.' This is also due to the fact that compounding an activity verb with a resultative or directional verb changes the situation type of the former and introduces a situation with an initial point and a natural final point. 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (41) a.*Iiisi ba fayu xue—le. Lisi BA French learn-ASP b. Lisi xue—le fayu. Lisi learn-ASP French 'Lisi learned French.' (42) a.*ta ba chezi tui. he BA cart push b. ta tui chezi. he push cart 'He is pushing a cart.' In contrast with stative verbs and activity verbs, accomplishment verbs and achievement verbs both describe telic events which have initial points and natural final points marking a change of state, a completion of process or a closure of event (Tai 1984; Smith 1991) . Hence, It has been observed that the difference between activity verbs and accomplishment and achievement verbs depends on the specificity of their object arguments (cf. Smith 1991) ; accomplishment and achievement verbs require their object arguments to be specific because the process and outcome of their action or event are finite; whereas activity verbs do not impose such a requirement, as shown by the contrast below: (i) chi san ge pingguo (accomplishment verb) eat three CL apple 'eat three apples' (ii) chi pingguo (activity verb) eat apples 'eat apples' 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. accomplishment and achievement verbs should be cible to occur in the BA-construction.^® This prediction is born out by the well-formedness of the BA-sentences below and those in Section 2.2. above, where the verbs are either accomplishment verbs or achievement verbs: (43) ta ba na ping jiu he—le. (accomplishment verb) he BA that CL wine drink-ASP 'He drank that bottle of wine.' The event in (i) has a natural final point, in the sense that once three apples are eaten, the event is finished. But the activity in (ii) has no natural final point since it could continue indefinitely. Though accomplishment and achievement verbs both describe telic events, they do differ from each other in terms of event duration: an accomplishment verb signifies a process of successive stages and an outcome, whereas an achievement verb describes an event having simultaneous initial and final points and introduces an instantaneous change of state (Smith 1991). There is an additional affectedness condition on the objects of achievement verbs in the BA-construction: that is, the logical object of an achievement verb in the BA- construction (i.e. BA-NP) must be affected by the action conveyed by the achievement verb (Cheng 1986); otherwise, the BA-construction would be ill-formed, as shown by the following contrast: (i) a.*tamen ba shanding daoda—le. they BA summit reach-ASP b. tamen daoda-le shanding. they reach-ASP summit 'They reached the summit.' (ii) a.*wo ba na ke shu kaniian—le. I BA that CL tree see-ASP b. wo kanjian—le na ke shu. I see-ASP that CL tree 'I saw that tree.' 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (44) wo ba na jian yifu xi—le. (accomplishment verb) 1 BA that CL coat wash-ASP 'I sold that car.' (45) ta ba yaoshi zbaodao—le. (achievement verb) he BA key find-ASP 'He found his key.' (46) wo ba na pan qi yinq—le. (achievement verb) I BA the CL chess win-ASP 'I won that chess game.' 2.3.3. The BA-NP It has been pointed out in almost all major work on the BA-construction that the BA-NP has to be definite or specific, and a nonspecific and indefinite noun phrase cannot serve as a BA—NP.^^ Before we examine the BA-NP For example, Hashiraoto (1971) says that the BA-NP must be an NP with definite reference, and Lü (1984) says that the BA-NP must refer to something specific or some thing known from the context or by implication. Besides, Sijbesma (1992) presents a good summary of the discussion on this very issue, suggesting that the BA-NP tends to be definite in terms of interpretation, but it might also be indefinite in form only if it is interpreted as specific, and that this constraint on the BA-NP might be translated into Barwise and Cooper's (1981) term: only strong NPs, which include both definite and specific indefinite NPs, may act as the BA-NP, but weak NPs, which are nonspecific indefinite NPs, may not. 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with respect to this constraint, we would like to take a look at the forms that a Chinese noun phrase may assume. It is well known in the literature that the Chinese noun phrase does not have plural ending and it may appear only in three forms: i) the bare form; ii) the overtly marked indefinite form; and iii) the overtly marked definite form, as shown by the examples below (Lü 1955; Wang 1987; Sijbesma 1992): (47) a. ma horse(s) b. yi pi ma a/one CL horse c. zhe pi ma this CL horse The difference between a definite NP and a specific indefinite NP, as pointed out by Li and Thompson (1981), is that a definite NP refers to an entity about which the speaker believes that the hearer knows, while a specific indefinite NP refers to a certain entity the speaker has in mind but about which the hearer does not necessarily know. Besides, a generic noun phrase might also serve as a BA-NP, as shown by the following two examples: (i) ta youshihou ba yan dang tang chi. he sometimes BA salt take:as sugar eat 'He sometimes eats salt as if it were sugar.' (ii) ta ba aian kan-de hen zhongyao. he BA money look-DE very important 'he regards money as being very important.' 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The bare NP may be interpreted as definite, indefinite, or generic, solely depending on the context, as shown by the examples from Huang (1987): (48) ren si—Ic. (definite) man die-ASP 'The person(s) died.' (49) you ren lai—le. (indefinite) have man come-ASP 'Someone came.' (50) wo xihuan shu. (generic) I like book 'I like books.' The overtly marked indefinite NP is ambiguous between a nonspecific reading and a specific reading, which also depends on the context, as shown by the examples below; (51) qin ni gei wo vi zhi bi. (nonspecific) Please you give me one CL pen 'Please give me a pen.' (52) ta cuoguo-le vi qe iihui. (specific) he miss-ASP one CL opportunity 'He missed an opportunity.' 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (51) yi zhi bi 'a pen' has a nonspecific reading as it can refer to any pen available to the listener; while in (52) yi ge jihui 'an opportunity' has a specific reading because it does not refer to just any opportunity but a particular one that he missed.^® Whether an overtly marked indefinite NP object is interpreted specifically or nonspecifically may also have something to do with the nature of the verb involved. Li (1974) suggests that transitive verbs can be divided into two classes in terms of the presence or absence of a pre supposition relationship between the verb and its object. One class of transitive verbs presupposes its object and describes an action made or happened to its object (e.g. ca 'erase', hui 'destroy', wang 'forget'), but the other class of transitive verbs does not presuppose its object (e.g. xie 'write', jian 'build', jizu 'remember'). Li calls the presupposition relation between a transitive verb and its object an anaphoric relation and treats it as a co-occurrent relation which needs to be stipulated in terms of the features of the verb and of its object. Consider the verb pair ca 'erase' and xie 'write'. Ca, as a member of the first class, presupposes that there is something specific to erase. By contrast, xie, being a member of the second class, does not presuppose something specific to write, because the result of writing may not exist before writing. Put it in another way, the object of transitive verbs in the first class must be specific or known to the speaker or the subject before the verbal action takes place, but the object of transitive verbs in the second class is only known to the speaker or subject by virtue of the verbal action, thus it can refer to any member of its entity before the verbal action is taking place. Based on this difference between the two classes of transitive verbs, we can come to the conclusion that if the object of verbs in the first class is an overtly marked indefinite NP, it is interpreted specifically; but if the same indefinite NP serves as the object of verbs in the second class, it is interpreted non-specifically. The object of transitive verbs in the first class is also referred to as the affected object whose referent exists or is known to the speaker or subject before the verbal action takes place; while the object of transitive verbs in the second class is often called the effected or 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Returning to the BA-NP, we can see from the previous observation that the definiteness/specificity constraint is indeed imposed upon the BA-NP, as explicitly evidenced by the examples in Section 2.2. and those below: i) The BA-NP is an overtly marked definite noun phrase : (53) ta ba na ge pingguo chi—le. he BA that CL apple eat-ASP 'He ate that apple. ' (54) wo ba zhe ping 1iu he-le. I BA this CL wine drink-ASP 'I drank this bottle of wine.' (55) Lisi ba tade shu gei—le Zhangsan. Lisi BA his book give-ASP Zhangsan 'Lisi gave his book to Zhangsan.' ii) The BA-NP is a bare noun phrase interpreted as definite: resultant object whose referent exists or is known to the speaker or subject only by virtue of the verbal action (cf. Quirk, Greenbaum, Leech, and Svartvik 1985). 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (56) ta ba pingguo chi—le. he BA apple eat-ASP 'He ate the apple.' (57) wo ba iiu he—le. I BA wine drink-ASP 'I drank the wine.' (58) Lisi ba shu gei—le Zhangsan. Lisi BA book give-ASP Zhangsan 'Lisi gave the book to Zhangsan.' iii) The BA-NP is an overtly marked indefinite noun phrase interpreted as specific: (59) ta ba yi ge iihui cuoguo-le. he BA one CL opportunity miss-ASP 'He missed an opportunity.' (60) wo ba yi bei cha dao—le. I BA one CL tea toss-ASP 'I emptied a cup of tea into the sink.' (61) Lisi ba vi ben shu dui—le. Lisi BA one CL book lose-ASP 'Lisi lost a book.' However, if a nonspecific indefinite noun phrase serves as a BA-NP, the BA-construction will be ill-formed: 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (62) a. qin ni gei wo vi zhi bi. Please you give me one CL pen 'Please give me a pen.' b.*qin ni. ba vi zhi bi gei wo. Please you BA one CL pen give me (63) a. ta mai—le hendo shu. he buy-ASP many book 'He bought many books.' b.*ta ba hendo shu mai—le. he BA many book buy-ASP (64) a. wo zhao—le san zhana xiana. I take-ASP three CL photo 'I took three photos.' b.*wo ba san zhana xiana zhao—le. I BA three CL photo take-ASP 2.3.4. The Postverbal NP As mentioned in Section 2.2., the postverbal NP in the BA-construction may or may not have an inalienably possessive or part-whole relation with the BA-NP. As a matter of fact, the difference between the inalienable relation and the alienable relation has long been a topic of interest to many linguists (e.g. Gueron 1985 and 1991; 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cheng and Ritter 1987; Tellier 1990; Yoon 1990; Vergnaud and Zubizarreta 1993), and the general consensus about this difference in the literature can be summarized and be adapted as follows: i) the inalienably possessed noun assigns an inherent 0-role to its possessor and contains predicate-like properties; ii) the inalienably possessed noun is not a R-expression and does not have independent reference; and iii) the inalienably possessed noun may not be modified by a descriptive adjective or by a non- restrictive modifier. The particular significance of the inalienable and alienable distinction to BA-constructions is that if there exists an inalienably possessive or part -whole relation between the BA-NP and the postverbal NP, then the postverbal NP cannot be definite: (65) a. ta ba Lisi bang-le vi tiao tui. he BA Lisi tie-ASP one CL leg 'He tied up a leg of Lisi's.' b.*ta ba Lisi bang—le na yi tiao tui. he BA Lisi tie-ASP that one CL leg (66) a. wo ba Luxun-de shu mai—le vi ben. I BA Luxun's book sell-ASP one CL 'I sold one copy of Luxun's books.'' b.*wo ba Luxun-de shu mai—le zhe ben. I BA Luxun's book sell-ASP this CL 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. But if there is no inalienably possessive or part- whole relation between the BA-NP and the postverbal NP, then the postverbal NP can be definite: (67) ta ba shu gei-le zhe wei penovou. (cf. (10b)) he BA book give-ASP this CL friend 'He gave the book to this friend.' (68) wo ba yifu bao—le zhe ge xiaobao. (cf. (14b)) I BA clothes pack-ASP the CL bundle 'I packed the clothes into the bundle.' (69) ta ba qianokou dui—zhe na qe ren. (cf. (28)) ta BA muzzle aim-ASP that CL man 'he aimed at that person with his gun.' The semantic properties and syntactic constraints of the BA-constructions discussed above clearly demonstrate that it is BA that determines the aspectual features, the situation types of verbs, and the other components of the BA-constructions. Any attempt at a satisfactory analysis of the BA-constructions must incorporate and capture this fact, which is too important to be ignored. With this in mind, I now proceed to explore the status of BA in the BA-construction, which is presented in the section below. 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.4. THE STATUS OF BA 2.4.1. The Four Previous Analyses The question about the status of BA has bugged many linguists for a long time, and there are basically four different kinds of analyses in the literature: i) BA is a lexical or matrix verb (Hashimoto 1971); ii) BA is an inserted Case marker (Huang 1982 & 1992; Koopman 1984; Goodall 1987); iii) BA is a preposition (Travis 1984; Cheng 1986; Li 1990); and iv) BA is an inserted dummy head of the causative phrase (Sijbesma 1992). Of the four different analyses of the status of BA, the first one, under which BA is claimed to be a lexical verb, faces some strong counter-evidence, as pointed out by Li (1990). That is, BA fails all three tests of the lexical verb identification: i) BA cannot occur in the V- not-V question; ii) BA cannot take an aspect marker; and iii) BA-phrase cannot function as a simple answer to a question, as illustrated below (cf. Li 1990): (70) a. Ta ba shui fang-(le) zai guo—li he BA water put-(ASP) at wok-in 'He put the water in the wok.' 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b.*Ta ba—bu-ba shui fang zai guo—li? he BA-not-BA water put at wok-in c.*Ta ba—le shui fang—zai guo—li. he BA-ASP water put-at wok-in d. A: Ta ba shenme fang—zai guo—li? he BA what put-at wok-in 'What did he put in the wok?' B; *ba shui. BA water The second analysis, which treats BA as an inserted Case marker, is also dubious. On the one hand, there is no syntactic evidence suggesting that BA is a dummy Case assigner (Li 1990). On the other hand, the selectional restrictions and the semantic and syntactic properties and constraints of the BA-constructions discussed in the above sections cannot be explained if BA were an inserted case marker. For instance, if BA were an inserted Case marker, it could not select the situation types of verbs which are allowed to appear in the BA-constructions (see Section 2.3.2.). This selectional restriction would have to be determined by the lexical properties of the verbs and would be realized at D-structure or before syntactic operations. If this were the case, there would probably be no explanation at all for the distinction between the 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. BA-constructions and the corresponding non-BA sentences regarding this selectional restriction (see the examples in Section 2.3.2.). Thus, the selectional restrictions, semantic properties and syntactic constraints of the BA- construction require BA to be base-generated. The third analysis, which takes BA as a preposition, is supported by a marginal case of coordination between a BA-phrase and a prepositional phrase (Li 1990:189); (71) ?ni you wei ta you ba ta qiang qian, shi you and for him and BA him rob money be shenme yisi? what meaning 'You forced away money for him and from him, what do you mean?' The question is how general the coordination argument is, since coordinating a BA-phrase with a different PP would make a sentence unacceptable: (72) a. wo ti Zhangsan bang—le liang zhi jiao. I for Zhangsan tie-ASP two CL foot 'I tied up two feet for Zhangsan.' 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. wo ba Lisi bang-le liang zhi jiao. I BA Lisi tie-ASP two CL foot 'I tied up Lisi's two feet.' c.*wo ti Zhangsan vou ba Lisi bang—le liang zhi I for Zhangsan and BA Lisi tie-ASP two CL jiao. foot Besides, the BA-NP only has a thematic relation with the verb, whereas the object of a preposition usually bears a thematic relation with the preposition (Cheng 1986): (73) a. wo wei ni zuo-le zhe jian shi. I for you do-ASP this CL matter 'I have done this thing for you.' b.*wo wei zhe iian shi zuo-le. I for this CL matter do-ASP (74) a.*wo ba ni zuo-le zhe jian shi. I BA you do-ASP this CL matter b. wo ^ zhe iian shi zuo-le. I BA this CL matter do-ASP 'I have done this thing.' Moreover, a preposition and its object form a constituent and can be moved together, while BA and the BA-NP cannot: 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (75) a. wo gen ta shuo—le hendo hua- I with him talk-ASP much speech b. gen ta wo shuo—le hendo hua with him I talk-ASP much speech 'I talked a lot with him.' (76) a. wo W iuzi bo-le pi. I BA orange peel-ASP skin 'I peeled the skin of the orange.' b.*ba iuzi wo bo—le pi. BA orange I peel-ASP skin As for the fourth analysis which regards BA as an inserted dummy head of the causative phrase, it may not be correct either, even if its approach is interesting. Like the second analysis which treats BA as an inserted Case marker, this analysis cannot explain the selectional restriction and the semantic and syntactic properties and constraints of the BA-constructions since it treats BA as an inserted element. Though it may be right to treat BA as the head of the Causative BA-construction, it sounds very dubious to take the causative BA as the head of the Objective, Locative or Instrumental BA-construction which do not involve a Causer-Causee relation or a causative meaning. 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.4.2. BA as a Base—Generated Functional Category If the four analyses of the status of BA presented above are not correct, what is really the status of BA? The question still remains, begging an answer. Before answering this question, let us first explore the origin of BA, which can be traced back to the Tang Dynasty (Zhu 1957; Wang 1958). According to Wang's (1958) study, BA was originally a lexical verb, meaning take, hold, use, as shown by his two examples from the ancient Chinese: (77) ba 3in kan take mirror look 'take mirror and look (at oneself)' (78) ba juan kan take book read 'take the book and read it' The residue of BA as a lexical verb is still reflected in a few idiomatic expressions of the Modern Chinese, and it is listed as a separate lexical entry in the dictionary (i.e. as a homophone or homograph of BA used in the BA- 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. construction; cf. Chao 1968; Hashimoto 1971; Cheung 1973; Lü 1955 and 1984) (79) Zhangsan zuotian ba—le men. Zhangsan yesterday keep-ASP door 'Zhangsan kept guard of the door yesterday.' However, in non-idiomatic expressions, BA cannot be used as a lexical verb any more in the Modern Chinese because it has fully grammaticalized into a functional category (LÜ 1955), as displayed by the contrast between (79) and (80):20 Besides, BA can also serve as a classifier in the Modern Chinese, as exemplified below: (i) yi ba dao one CL knife 'one knife' (ii) liang ^ yaoshi two CL key 'two keys' (iii) san ^ yizi three BA chair 'three chairs' The classifier BA is also treated as a homophone or a homograph of BA in the BA-construction and listed as a separate lexical entry in the dictionary (Lü 1984). BA also has a variant form in the ancient Chinese that is jiang, as depicted by Cheung (1973:345): "both ba and jiang are etymologically related to actions executed with the hands, and both were later used to introduce instrumental expression." Jiang was originally a lexical verb too and shared the Scune syntactic function with BA, as shown by the following example from Wang (1958:412): 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (80) a.*Zhangsan zuotian ba-le shu. Zhangsan yesterday keep-ASP book b.*Zhangsan zuotian ba-le wanju. Zhangsan yesterday keep-ASP toy The grammaticalization of BA into a functional category is also evidenced semantically, as it does not have any substantial meaning in the BA-construction (Chao 1968). In addition, there is a piece of phonetic evidence for the grammaticalization of BA into a functional category, as pointed out by Lü (1955); that is, the lexical verb BA, as in (114), has to be pronounced as [ba]; while the pronunciation of the functional category BA, as in the BA-construction, can be optionally changed from [ba] to [bai]. The grammaticalization of a lexical verb into a full functional category is not uncommon cross-linguistically. Several West African languages, which have constructions (i) ging jiang yu ban qiao hua pian. lightly take jade stick knock flower piece 'lightly strike the petals with a piece of jade' Later, jiang was also grammaticalized into a functional category and was coalesced with BA (cf. Lü 1955; Bennett 1981). Thus, it cannot be used as a lexical verb either in the Modern Chinese: (ii)*wo jiang—le shu. I take-ASP book 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sharing many similarities with the BA-construction, also exhibit the grammaticalization of a lexical verb meaning 'take' into a functional category (Stewart 1963; Bennett 1981; Wagner 1985; Déchaîne 1991; Da Cruz 1992, Sijbesma 1992 and 1994) . Consider the former lexical verb de 'take' in Twi, which has been grammaticalized into a full functional category that is similar to BA syntactically. The examples below are cited from Stewart (1963) (81) a. □ ^ ne p3nk3 no femm me he TAKE his horse that lend me 'He lent his horse to me.' b. 2 femm me ne p3nk3 no he lend me his horse that 'He lent me his horse.' These constructions of the West African languages are often known as a type of serial verb constructions. The sentences in (81a), (82) and (83) are exactly parallel to the following Chinese BA-sentences: (i) ta ba tade ma jie gei wo. He BA his horse lend to me 'He lent his horse to me.' (ii) ta ba qian gei-le wo. he BA money give-ASP me 'He gave me the money.' (iii) ta ba den fang—le zai zhuozi shang. he ba lamp put-ASP at table on 'He put a lamp on the table.' 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (82) □ ^ sika no maa me. he TAKE money that give me 'He gave me the money.' (83) 3 ^ kane a bisii pono no so. he TAKE lamp a put table that top 'He put a lamp on the table.' As shown by the interpretations of (81a), (82) and (83) and by the comparison between (81a) and (81b), the verb de in Twi has lost its independent meaning and lexical status and has become a functional category. Another example of the grammaticalization of the same lexical verb in West African languages comes from Fong (a Kwa language spoken in Benin), in which the verb so 'take' can be either a lexical transitive verb taking objects, as shown by (84) below, or a functional category without a substantial meaning, as shown by (85) (Da Cruz 1992) The sentences in (85) are also parallel to the following Chinese BA-sentences: (i) a. Lisi ba pangxie gei-le Zhangsan. Lisi BA crab give-ASP Zhangsan 'Lisi gave the crab to Zhangsan.' b. Lisi ba giche mai gei Zhangsan. Lisi BA car sell to Zhangsan 'Lisi sold the car to Zhangsan.' c. Lisi ba ji sa-le. Lisi BA chicken kill-ASP 'Lisi killed the chicken.' 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (84) a. Koku so ason. Koku take crab 'Koku took a crab.' b. Koku so awu yoyo o. Koku take dress new DET 'Koku took a new dress.' (85) a. Koku so ason na Asiba. Koku TAKE crab give Asiba 'Koku gave the crab to Asiba.' b. Koku so moto o sa nu Asiba Koku TAKE car DET sell to Asiba 'Koku sold the car to Asiba.' c. Koku so koklo o hu. Koku TAKE chicken DET kill 'Koku killed the chicken.' d. Koku so ahan o nu. Koku TAKE wine DET drink 'Koku drank the wine.' e. Koku so kofu o gba. Koku TAKE glass DET break 'Koku broke the glass.' d. Lisi ba jiu he—le. Lisi BA wine drink-ASP 'Lisi drank the wine.' e. Lisi ba boli dapo-le. Lisi BA glass break-ASP 'Lisi broke the glass.' 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In addition to these West African languages above, the similar phenomenon is also observed in English. It is well known that have can be either a lexical verb or an auxiliary verb. When have acts as a lexical verb, it takes an object and has the meanings like own, possess, hold, take, etc., as exemplified below; (86) a. John has a house in the country. b. Mary has blue eyes. c. John has a good chance of winning. d. She had a walk and a rest. But when have acts as an auxiliary verb, it usually forms the perfective aspectual construction with an -ed or -en participle: (87) a. I have finished my work. b. She has studied French for five years. c. He has left for New York. In addition to the two canonical uses of have above, have is also involved in a causative construction that has the same components as the perfective aspectual construction but differs from the latter in word order, as illustrated by the following examples:: 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (88) a. He had his car repaired. b. Mary had her bad tooth pulled off. c. John had his hair cut. This causative construction corresponds precisely to the following BA-sentences in terms of sentence-formation, even if the latter are ambiguous between a causative reading and an agentive reading, which depends on the context (Ma 1987); (89) a. ta ba tade giche xiuhao-le. he BA his car repair-ASP 'He had his car repaired.' 'He had repaired his car.' b. Mali ba tade huai yachi ba—diao-le. Mary BA her bad tooth pull-off-ASP 'Mary had her bad tooth pulled off.' 'Mary had pulled off her bad tooth.' c. Lisi ba tade toufa jian—le. Lisi BA his hair cut-ASP 'Lisi had his hair cut.' 'Lisi had cut his hair.' The relevant point here is that the causative have is not a lexical verb, because it cannot take an object alone 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. without an -ed or -en participle and it does not have a substantial meaning of its own, on a par with BA in the BA-construction. But the causative have is not a fully auxiliary verb either, as the interrogative form of this causative construction still requires do-support: (90) a. When did you have your hair cut? b.*When have you your hair cut? Hence, it appears plausible to assume that the causative have is a functional category which is derived from the grammaticalization of its lexical counterpart.2* Return to the status of BA in the BA-constructions. Considering the synchronic selectional restrictions, the semantic properties and syntactic constraints of the BA- construction, the diachronic evidence from the Ancient Chinese, and the cross-linguistic evidence from both the West African languages and English, we are now entitled to draw the following conclusion; that is, BA is a base- Although the grammaticalization of the lexical verbs meaning 'take' in the West African languages and the origin and grammaticalization of the causative have in English are both very interesting issues, pursuing them further will go far beyond the scope of the study here. Thus, I would like to leave these issues for the future study. 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. generated functional category which is derived from its lexical counterpart by means of grammaticalization. In the following sections, I am going to present a morpho-syntactic analysis of the three different types of BA-constructions introduced in Section 2.2., and to show how this morpho-syntactic analysis offers an account for the selectional restrictions, the semantic properties and the syntactic constraints of the BA-construction depicted in the above sections. 2.5. A MORPHO-SYNTACTIC ANALYSIS OF BA-CONSTRUCTIONS 2.5.1. Theoretical Background The discussion in Sections 2.2., 2.3. and 2.4. above suggest the following two generalizations about the BA- construction : (91) a. BA is a base-generated functional category and does not have a thematic relation with the BA- NP (cf. Section 2.4.). b. The BA-NP and the postverbal NP may or may not have an inalienably possessive or part-whole relation (cf. Sections 2.2. and 2.3.4.). 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In order to capture the two generalizations, I would like to propose the following two postulations: (92) a. BA projects its own maximal projection and it selects an aspect phrase (ASPP) or resultative phrase (DEP) (which are functional categories too) as its complement.25 b. The BA-NP and the postverbal NP are generated as components of a single noun phrase if they have an inalienably possessive or part-whole relation; otherwise, they are not. The postulation in (92a) incorporates the generalization in (91a) and captures the aspectual feature reflected on the verb as a selectional property of BA: that is, verbs in the BA-construction usually take the perfective aspect 25 That the aspect phrase and the resultative phrase are functional categories is evidenced by the fact that both aspect markers and the resultative marker -de are inflectional and bound morphemes and are required to be phonologically and morphologically attached to verbs (cf. Chao 1968; Lü 1984; Dai 1992). The argument for taking the resultative marker -de as a sort of "aspect" marker is from its overlapping distribution with aspect markers, as shown by the following examples from Dai (1992): (i) a. Lisi chi—le (fan), binqie chi—de hen bao. Lisi eat-ASP (rice), and eat-DE very full 'Lisi had the meal, and is full.' b.*Lisi chi—le—de /chi—de—le hen bao. Lisi eat-ASP-DE/eat-DE-ASP very full 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. marker -le, the progressive aspect marker -zhe, or the resultative marker -de ( of. Sections 2.2.2. and 2.3.1.). The postulation in (92b) captures the generalization in (91b) and it is motivated both semantically (see Section 2.3.4.; an inalienably possessed noun assigns an inherent 0-role to its possessor) and syntactically (see Section 2.2.: the inherently related BA-NP and postverbal NP can move as a single constituent, whereas the non-inherently related BA-NP and postverbal NP cannot). Based on the postulation in (92a), we can establish the basic structure of BA-constructions in the following, assuming the standard practice of taking VP as complement of a functional category in the literature:^® (93) BAP BA*"^ ASPP/DEP Spec ASP'/DE' ASP/DE ^VP The basic structure of BA-constructions given in (93) is also supported cross-linguistically in terms of the properties of functional categories: that is, BA, the aspect marker and the resultative marker -de, like any other functional categories across languages (Abney 1987; Ouhalla 1991), select only non-argument complements, have categorial-selectional properties specifying what kind of syntactic categories they select, and have morphological selectional properties concerning the categorial nature of an item they can attach to. 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (93), BAP corresponds to AGRPg, and ASP or DEP is on a par with AGRPq if we assume Chomsky's (1993) clausal structure. Treating BAP as a sort of AGRPg makes perfect sense if the structural Case assigned to or checked with the subject of the BA-construction is regarded as a sort of exhibition of the Spec-head agreement relation between the subject and the head of BAP. For the same reason, the treatment of an aspect phrase as a type of AGRPq is also consistent with the Case assignment or Case checking via the Spec-head agreement relation, which has been adopted in many current linguistic practices. With the two postulations in (92ab) and the basic structure of BA-constructions in (93), I now proceed to offer a morpho-syntactic analysis of the three types of BA-constructions in Section 2.2., as presented below. 2.5.2. The Objective BA-Constzniction (1) As we discussed in Section 2.2.1., the simplest sub- type of Objective BA-constructions consists of a subject, BA, a BA-NP and a transitive verb, and the BA-NP is the logical object of the verb, as shown by (3), (4) and (5) above. Take (3) for instance, which is repeated below; 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (3) wo ba juzi bo-le. I BA orange peel-ASP 'I peeled the orange.' Given the basic structure of BA-constructions in (93), an BA-construction like (3) would be expected to have the following structural representation in terms of thematic relation and X'-theory, assuming the VP-internal subject hypothesis (Koopman and Sportiche 1988; Kuroda 1988) (3') [gAP Iba Easpp [ASp] Ivp Enpi [y Iv bo—le] BA I peel-ASP Inp2 juzi]]]]] orange In (3'), the aspectual marker —le is attached to the verb bo 'peel' intrinsically, assuming that the verb carries its inflectional features in the lexicon as an intrinsic property (Chomsky 1993). In the derivation of (3) from (3'), the verb bo-le is first raised from V into ASP to check its aspectual feature against ASP in order for it to enter the PF-component under Spell-Out, and then the By thematic relation, we mean the relation that is determined by the small inventory of lexical catego ries and their unambiguous projection (Hale and Keyser 1993). 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. verb is nonovertly raised to BA at LF in order to satisfy the Principle of Full Interpretation (cf. Chomsky 1986 & 1993). NP2 juzi 'orange' receives a 6-role from the verb bo—le and moves into the Spec of ASPP to check its Case against ASP, assuming that Case, like the inflectional features taken by a verb, is inherently carried by an NP in the lexicon and needs to be checked in the appropriate position (Chomsky 1993). As for NPl wo 'I', it receives a 6-role from V by virtue of its internal subject status and moves into the Spec of BAP to check its Case against BA.^® Hence, (3) is simply derived by verb-raising and NP-movement, as shown below: (3") Ib a p [b a [a s PP [a s p I BA orange peel-ASP [yp [npi [y [y [np2 tj]]]]] The motivation and arguments for this derivation of (3) are presented as follows. First, raising the verb bo—le to ASP is morphologically driven as its aspectual Though the agreement between BA and its specifier is not overt in Chinese, it is overtly expressed in the counterpart of the BA-construction in a few West African languages. Take the Yoruba serial-verb construction for example, of which the first verb, which corresponds to BA in Chinese, is overtly marked by the subject high tone agreement element (cf. Déchaine 1991; Sijbesma 1992). 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. feature —le has to be checked against the feature of ASP in the checking domain of the latter; or the derivation will crash at PF because the strong aspectual feature of ASP is not a legitimate object at PF: that is, it is not a proper component of phonetic matrices (Chomsky 1993). This verb-raising is also licit under the Minimal Link Condition (Chomsky 1994), as it attaches the verb to the nearest head ASP which immediately c-commands VP, without skipping an already-filled head position. As for the non -overt verb-raising from ASP to BA at LF, it is forced by the Principle of Full Interpretation and meets with the Principle of Procrastinate, assuming that the agreement feature of BA is "weak" and is not able to attract overt verb-raising (cf. Chomsky 1993; and also Note 26 above). Second, the movement of NP2 juzi to the Spec of ASPP is forced by the Case Filter and is legitimate under the Shortest Movement Condition (Chomsky 1993), due to the verb-raising. That is, when the verb bo-le is raised to ASP from V to produce the chain (boj^-le, tj^), its minimal domain is {Spec of ASPP, Spec of VP, NP2}. Because the Spec of ASPP and the Spec of VP are in the same minimal domain, they are equidistant from NP2. Hence, NP2 could move to the Spec of ASPP by crossing the Spec of VP which is filled with NPl or its trace. 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Third, the movement of NPi wo to the Spec of BAP is also forced by the Case Filter, and it is legitimate by the Shortest Movement Condition, attributable to the non- overt verb-raising from ASP to BA at LF. 2.5.3. The Objective BA-Construction (2) The analysis presented above also accounts for the subtype of Objective BA-constructions which is composed of a subject, BA, a BA-NP, a transitive verb and a post verbal NP, and where there is an inalienably possessive/ part-whole relation between the BA-NP and the postverbal NP, as shown by (6b), (7b), (8b) and (9b) above. First, let us consider the BA-sentences in which there exists an inalienably possessive relation between the BA-NP and the postverbal NP. Take (6b) for example, which is rewritten below for the sake of convenience: (6b) wo ba juzi bo-le pi. I BA orange peel-ASP skin 'I peeled the skin of the orange.' Given the basic structure of BA-constructions in (93) and the postulation in (92) that the BA-NP and postverbal NP 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. are generated as the components of a single noun phrase if they have an inalienably possessive relation with each other, we will expect (6b) to be structurally represented in the following way: (6b') [gA ^^1 [aspp IaspI Ivp I NPl Iv' Iv BA I peel-ASP Inp2 Inp3 juzi] E x pi]]]]]] orange skin Like the derivation of (3) from (S') above, the verb bo le is first raised to ASP to check its aspectual feature and is then nonovertly moved to BA at LF to meet with the Principle of Full Interpretation, NPi wo receives a 0- role from V' and moves into the Spec of BAP to check its Case against BA, and NP2 juzi pi 'skin of orange' gets a 0-role from the verb bo-le. Besides, NP3 juzi receives an inherent possessional 0-role from N pi (Gueron 1985 & 1991; Tellier 1990; Yoon 1990; Sijbesma 1992; Vergnaud and Zubizarreta 1993) and it moves to the Spec of ASPP to check its Case against ASP. The specifierless NP2 gets an inherent Case from the verb bo-le, which is different from structural Case in several related ways according to 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Chomsky 1986a).Thus, (6b) is also derived by verb- raising and NP-movement; (6b") [sA [ a s p p j u z i j Iasp bo—le^^] [vp I BA orange peel-ASP InPI tv' Iv ^kl InP2 InP3 ^ j 3 Ik pi]]]]]] skin The motivation and arguments for raising the verb bo —le to ASP and for moving NPl wo into the Spec of BAP are the same as the ones presented in the section eibove. As for the movement of NP3 juzi to the Spec of ASPP, it is also driven by the Case Filter and is legitimate by the Shortest Movement Condition, due to the verb-raising: i) when the verb bo-le is raised from V to ASP to form the chain (bo-le%, tj^) with the minimal domain {Spec of ASPP, Spec of VP, NP2}, the Spec of ASPP and the Spec of VP are equidistant from NP2 or any element it contains; and ii) That is, inherent Case is generally associated with a particular thematic role (here theme), there is no adjacency requirement on its realization, etc. Thus, the marked Case property of the BA-construction shown in (6b) is due to the fact that the verb may assign this inherent Case when both the BA-NP and postverbal NP are available. In this respect, the BA-construction is exactly parallel to the "partial double object construction" in a Bantu language like Chimwiini and in English concerning the assignment of inherent Case, which has been discussed in detail by Baker (1988) and Larson (1988). 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. thus, NP3 juzi, which is a specifier of NP2, may move to the Spec of ASPP by crossing the Spec of VP which is now filled with NPi or its trace. One immediate consequence of this morpho-syntactic analysis is that it naturally captures the inalienably possessive relation between the BA-NP juzi and the post verbal NP pi in (6b) as a spec-head relation in a single noun phrase without any stipulation. Another consequence of this analysis is that it correctly rules out, in a principled way, the possibility of moving the possessee pi in (6b) to the Spec of ASPP with its possessor juzi left behind, which would yield an ungrammatical sentence: (6b"') *wo ba pi bo-le juzi. I BA skin peel-ASP orange * [ b a p Ib a [a s p p P ^ j Ia s p bo—le%] [^p I BA skin peel-ASP InPI Iv' Iv ^kl InP2 InP3 In bj]]]]]] orange That is, since pi is the head of NP2 and then an N°, the movement of N° pi out of NP2 into the Spec of ASPP would alter its "phrase structure status" from N° to N'"^*, thus resulting in a violation of the Uniformity Condition on the links of a chain (Chomsky 1994). 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Now consider the Objective BA-constructions in which there is an inherent part-whole relation between the BA- NP and the postverbal NP. Take (8b) for example, which is repeated below; (8b) Lisi ba yifu reng-le yi jian. Lisi BA clothes throw-ASP one CL 'Lisi threw away one piece of the clothes.' (8b) is almost the same as (6b), except for the internal structure of the postverbal NP: the BA-NP juzi in (6b) is originated as the specifier of the postverbal NP pi (i.e. juzi pi) but the BA-NP yifu in (8b) does not (i.e. *yifu yi jian). Given the basic structure of BA-constructions in (93) and the postulation in (92), we would expect (8b) to have the following structural representation, assuming Tang's (1990) DP/KP analysis of the Chinese noun phrases (note: KP = classifier phrase):^® The need for a DP/KP analysis of the Chinese noun phrase is argued and supported by the following facts: a) classifiers are obligatory in most Chinese noun phrases and there is a selectional relation between a classifier and an NP; b) the co-occurrence of a possessive NP and a determiner is allowed in the Chinese noun phrase; and c) more than one possessive NP may occur in a single Chinese noun phrase: (i) a. na san ben shu that three CL book 'those three books' 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (8b') [gap [ba [aspp [aspI [yp [kpi Lisi] (y, BA Lisi l y reng-le] [ ^ p [r yi jian] yifu]]]]]] throw-ASP one CL clothes Like the derivation of (6b) from (6b') above, the verb reng—le 'throw' is raised to ASP to check its aspectual feature and is then nonovertly moved to BA at LF, NPi Lisi receives a 0-role from V' and moves to the Spec of BAP to check its Case against BA, and KP yi jian yifu 'a piece of clothes' gets a 8-role from the verb rengle.^i b.*na san shu that three book (CL is missing) c.*na san ge shu that three CL book (CL is not correct) (ii) Lisi—de na san ben shu Lisi's that three CL book 'those three books of Lisi's' (iii) Lisi—de san ben Zhangsan-de shu Lisi's three CL Zhangsan's book 'Lisi's three copies of Zhangsan's books' For the treatment of numeral and classifier as components of the single K-node in (8b'), please see Tang (1990) for arguments. The evidence for the noun phrase status of KP is from the following two facts: a) any noun which denotes measure or quantity can be used as classifier (Chao 1968; Li and Thompson 1981); and b) when classifiers are used alone without being following by a noun, they behave like nouns rather than adjectives; (i) a. ta shihu hen xin. it seems very new 'it seems very new.' 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In addition, NP2 yifu 'clothes' receives a complement 8- role from K and moves into the Spec of ASPP to check its Case against ASP, and KP itself gets inherent Case from the verb reng-le. Thus, the derivation of (8b) also falls under verb-raising and NP-movement: (8b") [gAP Lisii [gA ba] [asp reng-le%] Lisi BA clothes throw-ASP [npi bil [v' [v b ] ( ] [gp [g yi jian] [gp2 t j ] ] ] ] ] ] one CL The motivation and argument for the verb-raising and the movement of NPl Lisi are the same as those presented above. As for the movement of NP2 yifu into the Spec of ASPP, it is driven by the Case Filter and is licit under the Shortest Movement Condition: i) as the raising of the verb reng-le from V to ASP forms the chain (reng-le^, t%) with the minimal domain {Spec of ASPP, Spec of VP, KP}, b.*ta shihu vi ben shu. it seems one CL book c.*ta shihu yi ben, it seems one CL (ii) a. you yi ben shu zai nali. have one CL book over there 'there is a book over there.' b. you yi ben zai nali. have one CL over there 'there is a copy over there.' c.*you hen xin zai nali. have very new over there 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the Spec of ASPP and Spec of VP are equidistant from KP or anything it contains. Thus, NP2 yifu, as a complement of K, is able to move to the Spec of ASPP by crossing the Spec of VP which is filled with NPi or its trace. As a consequence of this morpho-syntactic analysis, the inherent part-whole relation between the BA-NP yifu and postverbal NP yi—jian in (8b) is naturally captured by the head-complement relation in a classifier phrase without any stipulation. Another consequence of this analysis is that it correctly rules out the possibility of moving the numeral-classifier yi jian in (8b) to the Spec of ASPP (with its complement NP yifu left behind) as a violation of the Uniformity Condition on the links of a chain: that is, the movement of yi jian, which is the head of KP and then a K°, to the Spec of ASPP will change its "phrase structure status" from N® to N®^^: (8b"') *Lisi ba yi jian reng-le yifu. Lisi BA one CL throw-ASP clothes *Ibap Lisii ba] [ASPP ii**] Iasp reng-le*] Lisi BA one CL throw-ASP I VP I NPl Iv Iv IkP Ik ^3 3 InP2 yifu]]]]]] clothes 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5.4. The Objective BA-Construction (3) As displayed in Section 2.2.1., the third subtype of Objective BA-constructions is also composed of a subject, BA, a BA-NP, a transitive verb and a postverbal NP. But there is no inalienably possessive or part-whole relation between its BA-NP and the postverbal NP which are direct object and indirect object originally, as in (10), (11) and (12). Consider (10b), which is rewritten below: (10b) ta ba na ben shu gei—le pengyou. he BA that CL book give-ASP friend 'He gave his friend that book.' Given the postulation in (92b), the BA-NP na ben shu 'the book' and the postverbal NP pengyou 'friend' in (10b) can not be generated as components of a single noun phrase as pengyou is not the inalienable possessor of na ben shu (cf. Note 4). This then raises a serious question about the internal structure of VP in (10b): that is, where are the positions of the two NPs in VP if they do not have an inalienably possessive or part-whole relation? Before answering this question, we would like to review Cheng's (1989) study of the transitivity alternation in Chinese, which happens to be relevant to the issue here. 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cheng (1989) observes in Chinese that of the four situation types of verbs, only accomplishment verbs and achievement verbs allow the transitivity alternation to become intransitive verbs ( cf. Section 2.3.2.)* In other words, only the logical object of accomplishment verbs and achievement verbs can appear preverbally as subject, but this option is not available to the logical object of stative verbs and activity verbs. This is illustrated by the following contrast: (A) Stative Verbs: (94) a. wo you na ben shu. I own that CL book 'I have that book.' b.*na ben shu vou-l e . that CL book own-ASP (95) a. ta xiang baba. He resemble father 'He resembles his father.' b. *baba xiang— l e . father resemble-ASP (B) Activity Verbs: (96) a. w o zou-le lu. I walk-ASP road 'I have walked.' 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b.*lu zou-le. road walk~ASP (97) a. ta tui che. he push cart 'He is pushing a cart.' b.*che tui. cart push (C) Accomplishment Verbs: (98) a. wo bo-le na ge juzi. I peel-ASP that CL orange 'I peeled that orange.' b. na ge juzi bo-le. that CL orange peel-ASP 'That orange was peeled.' (99) a. ta xi—le zhe tiao maojin. he wash-ASP this CL towel 'he washed this towel.' b. zhe tiao maojin xi—le. this CL towel wash-ASP 'This towel was washed.' (D) Achievement Verbs: (100) a. ta zhaodao—le tade qianbao. he find-ASP his wallet 'He found his wallet.' 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. ta—de qianbao zhaodao—le. his wallet find-ASP 'His wallet was found.' (101) a. wo yino-le na pan qi. I win-ASP that CL chess ' I won that chess game.' b. na pan qi yino-le that CL chess win-ASP 'That chess game was won.' In fact, this transitivity alternation is also available to the accomplishment and achievement verbs that take two objects: that is, their direct object could appear in the preverbal position to act as subject while their indirect object remains in the postverbal position: (102) a. ta gei-le tade pengyou na ben shu. he give-ASP his friend that CL book 'He gave his friend that book.' b. na ben shu gei—le tade pengyou. that CL book give-ASP his friend 'That book was given to his friend.' Recall from Section 2.3.2. that only accomplishment verbs and achievement verbs, but not stative and activity 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. verbs, can occur in the BA-construction, and also notice that the achievement verb in (102) is exactly the same as that of the BA-construction in (10b), and the two objects in (102) are identical to those of (10b). The similarity between (102b) and (10b) naturally suggests that in (10b) the BA-NP na ben shu could be treated as a "subject" with respect to the verb (i.e. VP-internal subject), and the postverbal NP pengyou has to be treated as an object; or the sentence would be ungrammatical: (103) *ta ba pencvou gei—le na ben shu. he BA friend give-ASP that CL book Hence, the relation between the situation types of verbs and their transitivity alternation signals a right way in which the non-inherent relation between the BA-NP and the postverbal NP in (10b) could be captured. To capture the generalization that in (10b) the BA- NP na ben shu could be taken as a sort of subject or an "outer" object of the verb and the postverbal NP pengyou must be treated as an inner object, a natural approach is to take the BA-NP na ben shu as an "inner subject" of VP and to take the postverbal NP pengyou as complement of V before syntactic operations, assuming Larson's (1988) VP- shell hypothesis. The structural representation of this, 97 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. coupled with the basic structure of BA-constructions in (93), is shown below: (10b') BAP / ^ Spec BA' / "X BA ASPP I / ^ ba Spec ASP' / \ ASP VPl / \ NPl V'l I / \ ta VI VP2 he NP2 V'2 ! / " X na ben shu V2 NP3 'I I the CL book gei—le pengyou give-ASPfriend The internal structure of VP in (10b') can be intuitively understood as follows: i) V2 gei 'give' takes NP3 pengyou as complement, forming a small predicate V'2; ii) V'2 is predicated of the "inner subject" NP2 na ben shu, forming VP2; and iii) VP2 is, in turn, predicated of the "outer subject" NPl ta 'he' to yield a full sentence. Under this structural treatment of the non-inherent relation between the BA-NP na ben shu and the postverbal NP pengyou, the BA-construction in (10b) would also be derived by verb-raising and NP-movement: i) V2 gei—le is 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. raised to ASP via the "light" Vi to check its aspectual feature; ii) NPi ta receives a 0-role from V'l by viirtue of its "outer subject" status and moves into the Spec of BAP to check its Case against BA; iii) NP2 na ben shu gets a 8-role from V'2 by virtue of its "inner subject" status and moves into the Spec of ASPP to check its Case against ASP; and iv) NP3 pengyou receives a 8-role from V2 gei— le and gets inherent Case from the verb gei— le: (10b") [g A P ta^ ba] [^ ^ p p na ben shUj [^^p gei-le^] he BA that CL book give-ASP Ivpi tupi tv'l tk'l [vP2 In P 2 lv'2 Iv2 HI Inp3 pengyou]]]]]]] friend The motivation and arguments for raising the verb gei— le to ASP, the movement of NPi ta to the Spec of BAP and the movement of NP2 na ben shu into the Spec of ASPP are similar to those given in the above sections. That is, the verb-raising first attaches the verb to Vi which immediately c-commands VP2 and then it moves the verb to ASP which immediately c-commands VP2, without skipping an already-filled head position in between, thus satisfying the Minimal Link Condition. The movement of NP2 na ben shu is licit under the Shortest Movement Condition: i) as 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the raising of the verb gei— le from Vi to ASP forms the chain (gei— le%, t%,) with the minimal domain {Spec of ASPP, Spec of VPl, VP2}, the Spec of ASPP and the Spec of VPl become equidistant from VP2 or anything it contains; and ii) thus, NP2 na ben shu, as a specifier of NP2, may move to the Spec of ASPP by crossing the Spec of VPl that is filled with NPi or its trace. Besides, the movement of NPl ta into the Spec of BAP is also legitimate by the Shortest Movement Condition, attributable to the nonovert verb-raising from ASP to BA at LF. One consequence of this morpho-syntactic analysis is that it correctly rules out the ungrammatical sentence in (103) as a violation of the Shortest Movement Condition. The structure representation of (103) is the same as that of (10b') by the analysis above, and its derivation from (10b') is illustrated below: ( 103 ' ) * ba] [;^pp pengyoUj [;^p gei— Icj^j [ypi he BA friend give-ASP Inpi Iv'i 1^ 1 ^k'l Ivp2 Inp2 ben shu] [y»2 that CL book Iv2 InP3 (103') is the almost same as (10b”) except that NP3 peng— you rather than NP2 na ben shu moves to the Spec of ASPP. 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It is this movement of NP3 that violates the Shortest Movement Condition: i) as the raising of the verb gei—le from Vi to ASP forms the chain (gei-ie%, t%, ) with the minimal domain {Spec of ASPP, Spec of VPi, VP2}, the Spec of ASPP and the Spec of VPi are equidistant from VP2 or anything it contains; and ii) NP3 pengyou, though it is a complement of V2, c a n n o t move to the Spec of ASPP since such movement would cross NP2 na ben shu in the Spec of VP2, which is not a member of the minimal domain and is closer to the Spec of ASPP than NP3 (Chomsky 1993). Another consequence of this analysis is that it also correctly rules out the possibility of moving NP2 na ben shu to the Spec of BAP and moving NPi ta to the Spec of ASPP, which would yield another ungrammatical sentence: (104) *na ben shu ba ta gei—le pengyou. that CL book BA him give-ASP friend The derivation of (104) from (10b') is illustrated below: (104')*[3j^p na ben shUj [gA ba] [^pp ta^ [^gp gei-le%] that Cl book BA him give-ASP [vPl InPI Iv'l 1^1 t%,] Cvp2 [up2 ^jj Iv'2 Iv2 HI Inp3 pengyou]]]]]]] friend 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (104'), the minimal domain of the chain [gei—le%, t%,] is {Spec of ASPP, Spec of VPi, VP2}, and the Spec of ASPP is accessible to NP2 na ben shu which is the specifier of VP2. However, the Spec of BAP is not accessible to NP2 because it is not within the minimal domain of the chain (geijç—le, tj^) and it is closer to NPi ta than NP2 na ben shu. In other words, NPi blocks the movement of NP2 to any position higher than the Spec of ASPP. Hence, the movement of NP2 na ben shu to the Spec of BAP violates the Shortest Movement Condition. 2.5.5. The Objective BA-Construction (4) The morpho-syntactic analysis above also accounts for the fourth subtype of Objective BA-constructions that has the same number of components as the second subtype and the third subtype (i.e. the subject, BA, the BA-NP, the transitive verb, and the postverbal NP). Like the second subtype, the BA-NP is still what the verbal action happens to or affects. But like the third subtype, there exists no inalieneibly possessive or part-whole relation between the BA-NP and postverbal NP, as in (13b), (14b) and (15b) above (cf. Note 5). Consider (15b), which is repeated below: 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (15b) Lisi ba beitung hua-le liliang. Lisi BA sadness change-ASP power 'Lisi changed his sadness into power.' Given the basic structure of BA-constructions in (93) and the analysis of the non-inherent relation between the BA- NP and the postverbal NP in the above section, we would expect (15b) to be structurally represented in the same way as (10b) above: (15b') [ b j ^p [pA ^^1 [aspp [aspI [ypi [npx I*isi] [v»i BA Lisi 1^1 ] Evp2 Emp2 beitung] [■^•2 Ev2 bua—le] sadness change-ASP [pp3 liliang]]]]]]] power The argument for assigning the structural representation of (10b) to (15b) is from the fact that the transitivity alternation is also available to the accomplishment verb hua 'change' in (15b), in the sense that its outer object beitung 'sadness' could appear preverbally, serving as a subject in a corresponding non-BA sentence, and its inner object liliang 'power' remains as object in the same non- BA sentence (cf. (102)): 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (105) beitung hua— le lilia n a . sadness change-ASP power 'The sadness was changed into power.' The similarity between (105) and (15b) suggests that in (15b) the BA-NP beitung can be taken as an inner subject of VP and the postverbal NP liliang must be treated as a complement of V before syntactic operations apply. This prediction is born out, as the exchange of the functions of these two NPs would yield an ungrammatical sentence: (106) *Llsl ba lilia n g hua— le beitung. Lisi BA power change-ASP sadness Therefore, (15b) is derived from (15b') by the same kind of verb-raising and NP-movement that have derived (10b) from (10b') above: (15b") [gftp Lisii ba] [^sp? beitung^ [^p hua-le*] Lisi BA sadness change-ASP Ivpi InPI ïv'l [^1 t%,] [yp2 [np2 Iv'2 Iv2 I nP3 power 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Both the verb-raising and NP-movement are obligatory and legitimate by the motivation and arguments presented in the above sections. As a consequence of this analysis, the ungrammatical BA-construction in (106) is ruled out as a violation of the Shortest Movement Condition, as shown below (cf. the account of (103) in Section 2.5.4): (106')*[Bap Lisii ba] [^pp liliang^ [^sp hua-le^] Lisi BA power change-ASP Ivpi Inpi ^il Iv'i 1^1 t%,] [vp2 Inp2 beitung] [y/2 sadness Iv2 I nP3 1111111 This analysis will also rule out the ungreimmatical BA- construction below, where the movement of NP2 beitung to the Spec of BAP violates the Shortest Movement Condition (cf. the account of (104) in the above section): (107) *beitunq ba Lisi hua-le liliang. sadness BA Lisi change-ASP power ( 107 ' ) * beitungj [gj^ ba] [ j^pp Lisi^ Ij^p hua—le%] sadness BA Lisi change-ASP IvPl InPI ^il Iv'l t%,] [yp2 InP2 ^jl Iv'2 Iv2 ^kl [np3 lilia n g ]]]]]]] power 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5.6. The Locative and Instrumental BA—Constructions The morpho-syntactic analysis presented in the above sections further offers an account for the Locative and Instrumental BA-constructions. As described in Section 2.2.3., the Locative and Instrumental BA-constructions have the same components as the fourth type of Objective BA-constructions (i.e. the subject, BA, the BA-NP, the transitive verb and the postverbal NP), except that the BA-NP in the former refers either to the location of the verbal action or to the instrument with which the verbal action is performed. Besides, no inalienably possessive or part-whole relation exists between the BA-NP and the postverbal NP either. So the Locative and Instrumental BA-constructions would be expected to have the structural representation similar to that of (10b) according to the analysis of the non-inherent relation between the BA-NP and the postverbal NP presented above. This is actually evidenced by the fact that the transitivity alternation is available to the verb in the Locative and Instrumental BA-constructions, as displayed by the parallelism between the Locative emd Instrumental BA-constructions and the corresponding non-BA-sentences: 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (108) a. wo ba quo zhuang-le shui. (= (27)) I BA pot put-ASP water 'I put water in the pot.• b. quo zhuang-le shui. pot put-ASP water 'The pot was filled with water.' (109) a. ta ba aianqkou dui—zhe ren. (= (28)) ta BA muzzle aim-ASP man 'he aimed at the people with his gun.' b. qianqkou dui-zhe ren. muzzle aim-ASP man 'The people were aimed at with the gun.' Hence, the derivation of the Locative and Instrumental BA-constructions also falls under the same kind of verb- raising and NP-movement that have derived (10b) above. The following two examples exhibit the derivation of the Locative BA-construction in (25) and of the Instrumental BA-construction in (28): (25') [bap wOi Iba ba] [^gp? guOj Easp zhuang-le^] I BA pot put-ASP Ivpi ImPI bil [v'l [^1 t%,] [yp2 [up2 tj] [y,2 Iv2 bjçl [îip 3 shui] ]]]]]] water 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (28') [bap ba] [ggpp qiangkou^ [^gp dui-zhe%] he BA muzzle aim-ASP [ypi [mpi [y'l [^1 ^Jc'l [yp2 [np2 [v2 [v2 ^kl [np3 zen]]]]]]] man By the same arguments given in the above sections, the ungrammatical BA-sentences in (110) and (111) below are correctly ruled out as the violations of the Shortest Movement Condition (cf. the account of (103) and (104) in Section 2.5.4.): (110) a.*wo ba shui zhuang—le quo. J BA water put-ASP pot b.*shui ba wo zhuang—le guo. water BA I put-ASP pot (111) a.*ta ba ren dui—zhe qianqkou. ta BA man aim-ASP muzzle b.*qianqkou ba ta dui-zhe ren. muzzle BA he aim-ASP man 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5.7. The Causative BA-Constzniction (1) The morpho-syntactic analysis presented in the above sections further extends to account for the formation of the three subtypes of Causative BA-constructions depicted in Section 2.2.2. First, let us consider the subtype of Causative BA-constructions which consists of a subject, BA, a BA-NP, and an unaccusative verb, as in (16), (17) and (18). In this subtype of Causative BA-constructions, the BA-NP is both a Causee and the logical subject of the VP, whereas the matrix subject, which is a Causer, has no thematic relation with the VP. Take (16) for example, which is rewritten below: (16) na ge kanshou ba yi ge zei pao—le. that CL warden BA one CL thief run-ASP 'That warden let a thief escape.' Since the matrix subject and Causer na ge kanshou 'that warden' has no thematic relation with the unaccusative VP, it must be an additional argument being added to the predicate by causation (cf. Huang 1992; Sijbesma 1992). Assuming Hoekstra's (1990) analysis that the change of an unaccusative verb into the corresponding causative verb involves the incorporation of the unaccusative verb into 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a higher abstract causative predicate (CAUS), the Spec of which is occupied by the Causer (see Huang (1992) for the same idea), the matrix subject and Causer na ge kanshou in (16) can then be treated as the specifier of a higher VP-shell or CAUSP-shell along the lines of Larson (1988). If this sounds plausible, the Causative BA-construction in (16) will be expected to have the following structural representation, in accordance with the basic structure of BA-constructions proposed in (93); (16 ) [b a p [b a [aspp [aspI [c a u s p [npi 3® kanshou] BA that CL warden [cAUS' [cAus 1 [yp [np2 9® zei] [y, one CL thief [y pao-le]]]]]]] run-ASP With this structural representation, the Causative BA- construction in (16) can be derived by the same kind of verb-raising and NP-movement that derive the other types of BA-constructions above. That is, the unaccusative verb pao-le 'run' is raised to ASP via the empty CAUS to check its causative and aspectual features; NPl na ge kan— shou receives a 0-role from CAUS ' and moves to the Spec of BAP to check its Case against BA; and NP2 yi ge zei 'a thief' 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. receives a 0-role from V' and moves to the Spec of ASPP to check its Case against ASP. The derivation of (16) from (16') is illustrated below: (16") Ibap na ge kanshou^ ba] [^pp yi ge zeij that CL warden BA one CL thief Iasp pao”lejc] [causp Ikpi [caus' [caus t'kl run-ASP [yp [nP2 ^jl [v Iv ^kl 111111 Both the verb-raising and NP movement are obligatory and legitimate under the arguments in the above sections. As a consequence of this analysis, the logical subject-verb relation between the BA-NP yi ge zei and the verb pao—le is naturally captured by the spec-head relation of a VP without any stipulation. 2.5.8. The Causative BA-Construction (2) Now, let us consider the second subtype of Causative BA-constructions which is made up of a subject, BA, a BA- NP, an unaccusative verb and a resultative clause, as in (19), (20) and (21). Like the first subtype of Causative BA-constructions, the BA-NP is the logical subject of the 111 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. matrix VP, the matrix subject has no thematic relation with the matrix VP, and the relation between the matrix subject and the BA-NP remains a Causer-Causee relation. In addition, the BA-NP is also the logical subject of the resultative clause. Take the Causative BA-construction in (19) for example, which is repeated below: (19) zhe jian shi ba wo ji—de shui bu hao. this CL matter BA I hurry-DE sleep not well 'This matter made me so anxious that I couldn't sleep well.' Since the BA-NP wo 'I' is the logical subject of both the matrix VP and the following resultative clause and the resultative clause is the complement of the matrix verb ji 'hurry', the subject of the matrix VP will minimally C-command the subject of the resultative clause in (19) (Larson 1 9 9 0 ) Hence, the subject of the resultative clause can be treated as a Pro controlled by the subject of the matrix VP under a proposal made by Huang (1992). Assuming the analysis of the Causative predicate above, we would expect the Causative BA-sentence in (19) to have See Hoekstra (1988), Huang (1992) and Sijbesma (1992) for the arguments for taking the resultative clause as the complement of verb in sentences like (19). 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the following structural representation, in accordance with the basic structure of BA-constructions in (93): (19 ) [b a p [b a (d e p [d e ] [C A U S P [npi jian shl] BA this CL matter [cA U S ' [caus ] [yp [np2 “®il [y [y ji~de] Pro^ I hurry-DE shui bu hao]]]]]]] sleep not well With this structural representation, the derivation of (19) would fall under the same kind of verb-raising and NP-movement presented in the above sections: i) the verb j i — de 'hurry' is raised via the empty CAUS tc DE to check its causative and resultative features; ii) NPi zhe jian shi 'this matter' receives a 0-role from CAUS' and moves to the Spec of BAP to check its Case against BA; and iii) NP2 wo 'I' gets a 6-role from V' and moves to the Spec of DEP to check its Case against DE: (19") [b a p zhe jian shi^ ba] [j ,e p w O j [j ,e ji-de%] this CL matter BA I hurry-DE [C A U S P [npi [cA U S ' [cA U s [yp [n p2 ^j 1 [y [y ^kl [r c shui bu h ao]]]]]]] sleep not well 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The verb-raising and NP-movement in (19") are obligatory and legitimate under the motivation and argument given in the above sections. As a consequence of this analysis, the logical subject-verb relation among the BA-NP wo, the matrix verb ji and the embedded verb shui is captured in a natural manner by the spec-head relation of VP and by the controlled relation between the Spec of VP and the subject of the resultative clause without stipulation. 2.5.9. The Causative BA—Construction (3) Finally, consider the third subtype of Causative BA- constructions which consists of a subject, BA, a BA-NP, an intransitive verb and a resultative clause. Of this subtype of Causative BA-constructions, the BA-NP is only the logical subject of the resultative clause, the matrix subject is the logical subject of the matrix verb phrase, and the resultative clause is a complement of the matrix verb. The relation between the matrix subject and the BA-NP is remaining a Causer-Causee relation, as in (22), (23) and (24). Take (22) for example, which is repeated below and which would have the structural representation in (22') under the basic structure of BA-constructions proposed in (93): 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (22) Lisi ba shoupai ku-de hen shi. Lisi BA handkerchief cry-PAR very wet 'Lisi cried till the handkerchief got very wet.' (22') [qap Iba I DEP I DE ] I VP Inpi ^isi] [y, [y k u —de] BA Lisi cry-DE Irc Inp2 shoupai] hen shi]]]]] handkerchief very wet Hence, the derivation of (22) would fall under the same kind of verb-raising and NP-movement shown in the above sections; (22") Lisi^ I BA I DEP shoupaij [gg ku—dej^] [yp Lisi BA handkerchief cry-DE I NPl Iv Iv ^kl Irc Inp2 shi]]]]] very wet The motivation and arguments for the verb-raising and the NP-movement of NPi Lisi are the same as the ones presented in the above sections. As for the movement of NP2 shoupai 'handkerchief' to the Spec of DEP, it is also driven by the Case Filter and is licit under the Shortest Movement Condition: i) when the raising of the verb ku-de 'cry' from V to DE forms the chain (ku-de%, tj^) with the minimal domain {Spec of DEP, Spec of VP, RC}, the Spec of 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEP and the Spec of VP become equidistant from RC or any element it contains; and ii) thus, NP2 shoupai, as the initial NP of RC, can move to the Spec of DEP by crossing the Spec of VP which is filled with NPi or its trace. As a consequence of this morpho-syntactic analysis, the subject-verb relation between the matrix subject Lisi and the matrix verb ku-de is naturally captured by the spec-head relation of VP without any stipulation. This also applies to the subject-verb relation between the BA- NP shoupai and the embedded predicate hen shi. 2.6. THE CONSEQUENCES OF THE MORPHO-SYNTACTIC ANALYSIS 2.6.1. Negation and the BA-construction 2.6.1.1. The Function and Distribution of Negators Before we discuss the negation of BA-constructions, it is worth going over the properties of general negation in Chinese and finding out their relation to the negation of BA-constructions in particular. The following review is based on some previous studies of negation in Mandarin Chinese (Wang 1965; Chao 1968; Teng 1973 and 1974; Li and Thompson 1981; Lii 1984). In Mandarin Chinese, only two 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. negators are most commonly used: one is bu 'not' and the other is mei—you ' not-have '. You in mei—you is optional, especially in spoken Chinese. One difference between bu and mei—you is that bu can deny a state, but mei—you can not. Hence, it is possible for bu, but not for mei—you, to negate a predicate containing a stative verb or an adjective: (112) a. wo ^ zhidao zhe jian shi. J not know this CL matter 'I do/did not know this matter.' b.*wo mei-vou zhidao zhe jian shi. we not-have know this CL matter (113) a. ta ^ shi xiao zhang. she not be school chief 'she is/was not the principal.' b.*ta mei-vou shi xiao zhang. she not have be school chief (114) a. ta ^ conming. he not intelligent 'He is/was not intelligent.' b.*ta mei-vou conming. he not-have intelligent 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Although both bu and mei-you can negate an action or event predicate, bu negates the volition of performing an action or event while mei-you negates an actual action or event. Due to this difference, a negative sentence having an action or event predicate will change its meaning if its negator is changed from bu to mei-you or vice versa: (115) a. ta chi fan. he not eat rice 'He won't/wouldn't have a meal.' b. ta mei-you chi fan. he not-have eat rice 'He didn't have a meal.' (116) a. wo ^ mai na ge huaping. I not sell that CL vase 'I won't/wouldn't sell that vase.' b. wo mei-vou mai na ge huaping. J not-have sell that CL vase 'I didn't sell that vase.' (117) a. ta ^ gen wo jiang hua. he not with me speak speech 'He won't/wouldn't talk with me.' b. ta mei-vou gen wo jiang hua. he not-have with me speak speech 'He didn't talk with me.' 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As bu negates the volition of carrying out an action or an event, it will be semantically incompatible with a verb that hosts the perfective aspect marker -le, or the experiential aspect marker -guo or the progressive aspect marker -zhe, given that the action or event negated by bu does not happen or take place, while the verb hosting the perfective aspect marker -le indicates the completion of an action or an event, the verb hosting the experiential aspect marker -guo conveys the past experience of such an action or event, and the verb hosting progressive aspect marker -zhe refers to the ongoing process of an action or event. This semantic incompatibility is displayed by the ill-formedness of (118b), (119b) and (120b) below: (118) a. Mali bu chi na kuai dangao. Mary not eat that CL cake 'Mary won't/wouldn't eat that cake b.*Mali bu chi—le na kuai dangao. Mary not eat-ASP that CL cake (119) a. wo bu zuo feiji. I not sit airplane 'I won't sit on an airplane.' b.*wo bu zuo—quo feiji. I not sit-ASP airplane 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (120) a. ta bu na shu. he not hold book 'He won't/wouldn't hold the book.' b.*ta bu na—zhe shu. he not hold-ASP book In contrast with bu, mei-you can negate a verb that hosts the experiential aspect marker -guo or hosts the progressive aspect marker -zhe. This is because mei-you negates an actual action or event, so that it will not be incompatible with the two aspect markers semantically: (121) wo mei-you zuo—guo feiji. I not-have sit-ASP airplane 'I have not been on an airplane yet.' (122) ta mei-you na-zhe shu. he not-have hold-ASP book 'He is not holding the book.' However, mei-you cannot co-occur with a verb hosting the perfective aspect marker -le: (123)*Mali mei-you chi—le na kuai dangao. Mary not-have eat-ASP that CL cake 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (124)*Lisi mei-you mai—le na san zhi ji. Lisi not-have sell-ASP that three CL chicken (125)*wo mei-you xue-Ie yingwen. J not-have learn-ASP English According to Li and Thompson (1981), the ill-formedness of (123), (124) and (125) is also semantic. That is, the action or event conveyed by a verb hosting the perfective aspect marker -le is viewed in its entirety and bounded temporally, spatially or conceptually. In other words, the perfective aspect marker -le makes a verbal action or event bounded by, for example, overt phrases naming the extent to which the action or event occurred, or by the amount of time the action or event took, etc. Since an action or event that does not happen cannot be bounded, the negative meaning of mei-you will be incompatible with the bounded meaning of the perfective aspect marker -le. This semantic incompatibility is also evidenced by the complementary distribution between -le and mei-you in the following sentences: Some linguists even treat mei-you as the negative form of the perfective aspect marker -le, based on their complementary distribution (e.g. Wang 1965; Chao 1968). This treatment may not be correct as it wrongly predicts that the positive counterpart of every negative sentence with mei-you would contain the perfective aspect marker -le (Teng 1973): 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (126) a. Mali chi-le na kuai dangao. Mary eat-ASP that CL cake 'Mary ate that cake.' b. Mali mei—you chi na kuai dangao. Mary not-have eat that CL cake 'Mary didn't eat that cake.' (127) a. Lisi mai—le na san zhi ji. Lisi sell-ASP that three CL chicken 'Lisi sold those three chickens.' b. Lisi mei-you mai na san zhi ji. Lisi not-have sell that three CL chicken 'Lisi didn't sell those three chickens.' (128) a. wo xue-le yingwen. I learn-ASP English 'I learned English.' b. wo mei-you xue yingwen. I not-have learn English 'I didn't learn English.' (i) a. wo mei-you qu-guo Faguo. I not-have go-ASP France 'I have never been to France before.' b.*wo qu-guo—le Faguo. I go-ASP-ASP France (ii) a. ta mei-you dai—zhe maozi. he not-have wear-ASP hat 'He is not wearing a hat.' b.*ta dai—zhe—le maozi. he wear-ASP-ASP hat 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As for the distribution of bu and mei-you, they both normally stand between the subject and the predicate, as shown by the examples above. This is also the case when the subject and the predicate are complex: However, if a sentence contains a temporal or attitude adverb, the adverb must precede the negator: (i) a. wo iintian bu shufu. J today not comfortable 'Today I don't feel well.' b.*wo bu iintian shufu. I not today comfortable (ii) a. Lisi xianran mei-you qu xuexiao. Lisi obviously not-have go school 'Obviously, Lisi did not go to the school.' b.*Lisi mei-you xianran qu xuexiao. Lisi not-have obviously go school The distributional difference between temporal/attitude adverbs and manner adverbs (e.g. manmande 'slowly' in (130)) in the negative sentence is attributed to their inherent scope difference: temporal and attitude adverbs are sentential adverbs but manner adverbs are VP-adverbs. In other words, it is sentential adverbs, not VP-adverbs, that establish a semantic domain within which the action or event expressed by a predicate occurs. Due to this difference, a sentential adverb should not be expected to be within the scope of a negative predicate, whereas the negative predicate should be expected to be in the scope of a sentential adverb, given the fact that the scope of an element in Chinese is the portion of the sentence that follows the element in the surface form (Li and Thompson 1981). This is also evidenced by the fact that temporal and attitude adverbs, but not manner adverbs, can appear at the sentence-initial position in a negative sentence: (iii) a. iintian wo bu shufu. today I not comfortable 'Today I don't feel well.' b. xianran Lisi mei-you qu xuexiao. obviously Lisi not-have go school 'Obviously, Lisi did not go to the school.' (iv) *manmande Zhangsan bu/mei—you qi jiaotache. Slowly Zhangsan not/not-have ride bicycle 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (129) a. women zai zheli zuo shenovi bu rongyi we at here do business not easy 'It is not easy for us to do business here.' b. ta zban zai nali mei-you guanxi he stand at there not-have matter 'That he is standing there doesn't matter.' (130) a. Zhangsan bu/mei—you manmande ai 1iaotache. Zhangsan not/not-have slowly ride bicycle 'Zhangsan won't/didn't ride a bicycle slowly. ' b.*Zhangsan manmande bu/mei-you qi iiaotache. Zhangsan slowly not/not-have ride bicycle 2.6.1.2. Negation in the BA-construction The negation in the BA-construction is very similar to the negation in general in terms of the functions and distributions of the negators bu and mei—you.First, the change of bu to mei-you or vice versa in a negative BA-construction would also result in a meaning change: Because stative verbs cannot occur in the BA- construction (see Section 2.3.2.), it is trivially true that the negation of stative verbs is impossible in the BA-construction. 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (131) a. ta ^ ba ta-de went! suo—chulai. he not BA his problem speak-out 'He won't/wouldn't mention his problem(s).' b. ta mei-vou ba ta—de wenti suo—chulai. he not-have BA his problem speak-out 'He didn't mention his problem(s).' (132) a. Lisi ^ ba tade shu rengdiao. Lisi not BA his book throw 'Lisi won't/wouldn't throw away his book.' b. Lisi mei-you ba tade shu rengdiao. Lisi not-have BA his book throw 'Lisi didn't throw away his book.' Secondly, the same semantic constraints on the co occurrence of bu, mei-you and those three aspect markers are also observed in the negative BA-construction: i) bu cannot co-occur with a verb hosting the perfective aspect marker -le, or the progressive aspect marker -zhe or the experiential aspect marker -guo: (133) a. Mali ba na kuai dangao chi-le. Mary BA that CL cake eat-ASP 'Mary ate that cake.' b.*Mali ^ ba na kuai dangao chi-le. Mary not BA that CL cake eat-ASP 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (134) a. Lisi ba tade shu na—zhe. Lisi BA his book hold-ASP 'Lisi is holding his book.' b.*Lisi ^ ba tade shu na-zhe. Lisi not BA his book hold-ASP (135) a. wo ba tade yifu chuan-guo san ci. I BA his clothes wear-ASP three times 'I have worn his clothes three times.' b.*wo ^ ba tade yifu chuan-guo san ci. I not BA his clothes wear-ASP three times ii) by contrast, mei-you can co-occur with a verb hosting the progressive aspect marker -zhe or the experiential aspect marker -guo; (136) ta mei-you ba shu na-zhe. he not-have BA book hold-ASP 'He is not holding the book.' (137) wo mei-you ba tade yifu chuan-guo. I not-have BA his clothes wear-ASP 'I have not worn his clothes before.' iii) however, mei-you cannot co-occur with a verb hosting the perfective aspect marker -le: 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (138) a. Mali ba na kuai dangao chi—le. Mary BA that CL cake eat-ASP 'Mary ate that cake.' b.*Mali mei-you ba na kuai dangao chi-le. Mary not-have BA that CL cake eat-ASP Thirdly, as shown by the examples above, bu and mei- you usually follow the subject and precede the BA-phrase. However, they may also appear between the BA-NP and the verb in some idiomatic or set expressions, as observed by LÜ (1955 & 1984), Li and Thompson (1981) and Wang (1987): (139) Lisi ba wo bu dang ren. Lisi BA me not as man 'Lisi does not treat me as a human being.' (140) wo ba Lisi mei-you fang zai yuan li I BA Lisi not-have put at eye in 'I looked down upon Lisi.' (141) ta ba na jian shi yi zhi bu ti. he BA that CL matter one word not mention 'He didn't mention one word about that matter.' 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6.1.3. A Morpho—Syntactic Analysis of Negative BA- Constructions The discussion above suggests the following three generalizations about the negation of BA-constructions: (142) a. The negator bu cannot negate a BA-predicate which contains a verb hosting the perfective aspect marker -le, or the progressive aspect marker -zhe, or the experiential aspect marker -guo. b. The negator mei-you can negate a BA-predicate which contains a verb hosting the progressive aspect marker -zhe or the experiential aspect marker -guo, but not a BA-predicate containing the perfective aspect marker -le. c. The negators bu and mei-you usually follow the subject and precede BA. The question raised from these three generalizations is how we account for the selectional restriction which the negator imposes on the aspect markers when BA intervenes between them. To answer this question and capture these three generalizations above, I would like to propose the following two postulations, along the lines of Pollock 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (1989), Ouhalla (1990), Laka (1990) and Chomsky (1991 & 1993): (143) a. bu and mei—you head the negation phrase (NEGP) which is a functional category and is selected by BA in the negative BA-construction. b. bu selects an AGRPq as its complement, whereas mei-you selects a progressive aspect phrase, an experiential aspect phrase, or an AGRPq as its complement. The argument for an independent maximal projection of NEG in (143a) comes from the fact that both bu and mei—you can stand alone, serving as sentence-fragments, as shown by the following examples: 3? (144) SPEAKER A: rang ta kan dianying hao ma? let him see movie well Q 'Will you let him see the movie?' The aspect phrase is actually a type of AGRPq in the BA-construction, since it assigns Case to (or check the Case of) the verbal object which is moved to the Spec of the aspect phrase in the BA-construction, as discussed in the above sections. It is well known in the literature that only phrasal categories can serve as sentence fragments. As bu and mei—you serve as sentence-fragments in (149) and (150), they must head their own maximal projections. 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SPEAKER B: bu not (145) SPEAKER A: ni na-le wode shu ma? you take-ASP my book Q 'Did you take away my book?' SPEAKER B: mei—you not-have (143b) captures the aspectual feature on the verb as a selectional property of the negator: bu does not select an aspect phrase headed by the perfective aspect marker -le, the progressive aspect marker -zhe or the experien tial aspect marker -guo; and mei-you selects an aspect phrase headed by the progressive aspect marker -zhe or the experiential aspect marker guo, but not an aspect phrase headed by the perfective aspect marker -le, thus ruling out the ungrammatical negative BA-constructions in (133b), (134b), (135b) and (138b). In addition, there is a piece of independent evidence for placing NEGP between BAP and AGRPq/ASPP: that is, in some idiomatic and set expressions, the negator does occur between BA and the verb phrase, as in (139), (140) and (141). The two postulations in (143) lead to the following structure for the negative BA-construction, which is an 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. expansion of the basic structure of BA-constructions in (93): (146) ^ A P Spec BA' BA NEGP Spec NEG' NEG AGRPq/ASPP Spec AGRg/ASP' AGRg/ASP ^^VP With this negation structure, the derivation of negative BA-constructions would fall under the same kind of verb raising and NP-movement presented in the above sections. Let us first consider the negative BA-construction which contains the negator mei-you. Take the BA-construction in (136) for example, which is repeated below: (136) ta mei-you ba shu na—zhe. he not-have BA book hold-ASP 'He is not holding the book.' Given the negation structure in (146), we would expect (136) to be structurally represented in the following manner: 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (136') [ g a p [ b a [nEG P [n E G ® ® i~ y O U ] [ j ^ p p [ a s p ] BA not-have [vp [upi ^®] [v [v na—zhe] Înp2 shu] ] ] ] ] ] he hold-ASP book Like the derivation of the affirmative BA-constructions presented in the above sections, the verb na—zhe 'hold' is raised to ASP to check its aspectual feature, NP2 shu 'book' receives a 0-role from the verb na and moves into the Spec of ASPP to check its Case against ASP, and NPl ta 'he' receives a 8-role from V and moves into the Spec of BAP via the Spec of NEGP to check its Case against BA. Besides, the negator mei-you is raised from NEG to adjoin to the head of BAP. The derivation of (136) from (136') is illustrated below: (136") [bap [ba ®ei—youj^ ba] [begp [neg he not-have BA [ a s p p s h U j [A S P n a - z h e j j ] [ y p [ ^ p i t ^ ] [ y * [ v t % ] book hold-ASP [nP2 tj]]]]]] The motivation and arguments for the verb-raising and NP-movement in (136") are the same as those presented in the above sections. As for the negator-raising, which 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. makes the movement of NPi from the Spec of NEGP into the Spec of BAP possible, it is forced by the Principle of Full Interpretation in terms of negation scope. That is, in Chinese, the scope of a negator is a part or portion of the sentence that follows the negator at the surface structure (Li and Thompson 1981; Huang 1982). Thus, the raising of the negator to adjoin to BA will establish the whole BA-construction in the scope of the negator, which then matches the correct interpretation of the negative BA-construction accordingly. The negator-raising is also legitimate under the Minimal Link Condition because it attaches the negator to BA, which immediately C-commands NEGP without skipping any filled head position. Another argument for the legitimacy of the negator-raising comes from the very fact that nothing can intervene between the negator and ba in the negative BA-construction, as shown by the following contrast:^® There are two kinds of apparent counter-examples against the adjunction of the negator to ba in the BA- construction. First, some adverbs do occur between mei- you and ba, such as manner adverbs, as shown by (i): (i) Lisi mei—YOU manmande W wo jiaoxing Lisl not-have slowly BA me awaken 'Lisi did awaken me, but not slowly.' But in this case, mei-you negates only the manner adverb, not the BA-predicate, as shown by the interpretation of (i). Hence, such negation is a phrasal negation rather than the negation of the whole BA-construction, which is irrelevant to the adjunction of the sentential negator 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (147) a. ta jintian mei—you ba shu na-zhe. he today not-have BA book hold-ASP 'Today he is not holding the book.' b.*ta meiyou jintian ^ shu na—zhe. he not today BA book hold-ASP The same analysis also accounts for the negative BA- construction containing the negator bu, as illustrated by the derivation of (132a) below: (132a) Lisi bu ba tade shu rengdiao. Lisi not BA his book throw 'Lisi wouldn't throw away the book.' (132a') [sA Imegp Imeg ^“1 ^agrpo Eagro 1 ïvp BA not Inpi Sisi] ly, [ y rengdiao] [gpg tade shu] ]]]]] Lisi throw his book mei-you to ba. Second, a few 'co-verb' phrases may also appear between mei-you and ba, as exemplified below: (ii) Lisi mei-you yong wan ^ tang hediao. Lisi not-have use bowl BA soup drink 'Lisi ate the soup without using a bowl.' Like the negation of a manner adverb in (i), the negator mei-you negates only the co-verb phrase yong wan without affecting the whole BA-predicate, thus irrelevant to the adjunction of the sentential negator mei-you to ba too. As for the reason why in (147) the temporal adverb jintian 'today' cannot intervene between the negator and ba, please see Note 34. 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (132a”) Lisi^ bu^ ba] [jjegp Ineg ^ll Iagrpo Lisi not BA bade shu^ [^gro rendiao%] [ y p [y, [ y t^] his book throw InP2 ^jl 1 1 11 1 Hence, the morpho-syntactic analysis above captures the restrictions on the co-occurrence of negators and aspect markers in the BA-constructions and derives the negative BA-constructions as the side effects of verb-raising, NP- movement and negator-raising without further stipulation. 2.6.2. The Selectional Restriction on the Situation Types of Verbs As mentioned in Section 2.3.2., only accomplishment and achievement verbs but not stative and activity verbs can appear in the BA-construction, due to the fact that only accomplishment and achievement verbs, not stative and activity verbs, present a situation with an initial point and a natural final point indicating a change of state, a completion of action or a closure of event, thus meeting with the constraint on the BA-predicate. This constraint can now be captured by the morpho-syntactic 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. analysis proposed above as a selectional property of BA along the lines of Chomsky (1986a & 1993): that is, BA s- selects a predicate conveying a situation with an initial point and a natural final point indicating a change of state, a completion of action or a closure of event; and the predicate is then canonically structurally realized as an aspect phrase or a resultative phrase whose head contains these properties of situation aspect. When the verb of the BA-construction is raised into ASP or DE to check its aspectual feature or resultative feature, its property of situation aspect will also be checked at the same time against the situation aspectual property of ASP or DE. If the situation aspectual property of the verb matches that of ASP or DE, as in the case of achievement verbs and accomplishment verbs (cf. Section 2.3.2.), the derivation will converge at LF without any problem, thus ruling in both achievement and accomplishment verbs. But if the situation aspectual property of the verb conflicts that of ASP and DE, as in the case of stative verbs and activity verbs (cf. Section 2.3.2.), the derivation will crash at LF, thus ruling out both stative and activity verbs. 136 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.6.3. The Definiteness and Specificity of the BA—NP The discussion in Section 2.3.3. indicates that the BA-NP has to be definite or specific, and a nonspecific indefinite noun phrase cannot function as a BA-NP. This requirement 'triggers' the following effects on the bare noun phrase and the overtly-marked indefinite noun phrase which serve as the BA-NP: i) a bare noun phrase which may have a definite interpretation or an indefinite one, as in (48) and (49), has to be interpreted as a definite NP if it serves as a BA-NP, as in (56), (57) and (58); and ii) an overtly-marked indefinite noun phrase which may be ambiguous between a nonspecific reading and a specific reading, as in (51) and (52), must co-occur with a matrix verb that presupposes this overtly-marked indefinite noun phrase (i.e. a verb making this overtly-marked indefinite NP specific) if it acts as a BA-NP, as in (59), (60) and (61). According to Sijbesma (1992), the specificity or definiteness requirement on the BA-NP is solely related to the aspectual property of the BA-predicate, as argued by the relation between the perfectivity of a predicate and the definiteness, specificity and affectedness of its object (Verkuyl 1972; Dowty 1979; Li and Thompson 1981): namely, a predicate is perfective if its verb presupposes its object or suggests a change of state or a completion 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of action, or if it describes a bounded event. Once a predicate is perfective for these two reasons above, it can only co-occur with a definite or specific object, as evidenced by the two groups of examples from English and Chinese (Sijbesma 1992:176): (148) a. I finished the ccüces. b.*I finished cakes. c. I have eaten a cake, (a cake must be specific) (149) a. wo he-wan—le na ping jiu. I drink-finish-ASP that CL wine 'I finished that bottle of wine.' b. wo he—wan—le jiu. I drink-finish-ASP wine 'I finished the wine/*some wine.' c. Lisi chi—le yi kuai bing Lisi eat-ASP yi CL cake 'Lisi ate a particular cake/*some cake.' Since the predicate in the BA-construction possesses the perfective aspectual properties, as explicitly indicated by the discussion in Sections 2.3.1., 2.3.2, and 2.3.3., it will then require its BA-NP object to be definite or specific, thus forcing a definite or specific reading on 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a bare noun phrase or an overtly-marked indefinite noun phrase that serves as a BA-NP, as described above. It so happens that this definiteness or specificity requirement on the BA-NP can be naturally captured by the proposed morpho-syntactic analysis as the definiteness or specificity feature checking between the specifier and head of ASPP or DEP. As we argued in Sections 2.3.1. and 2.5.1., the perfective aspect marker -le, the progressive aspect marker -zhe and the resultative marker -de, which are the heads of ASPP and DEP, all express the meaning of definiteness or specificity: the perfective aspect marker -le signifies the completion of a specific action or the closure of a definite event; and the progressive aspect marker -zhe signals a change of state resulting from the completion of a specific action or closure of a definite event; and the resultative marker —de makes an action or event specific or bounded by means of an embedded clause that names the extent to which the action or event took place. This meaning of definiteness/specificity could be treated as a morphological property or feature of ASP and DE which imposes itself upon the Spec of ASPP and DEP by means of feature checking, thus forcing the BA-NP, which is moved into the Spec of ASPP and DEP for Case checking, to be definite or specific. 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. According to the discussion in Sections 2.3.2, 2.3.3 and 2.6.2., accomplishment verbs and achievement verbs, which presuppose their objects and are legitimate in the BA-constructions, must be raised into ASP and DE to check their aspectual features. So they should also be in the checking domain of ASP or DE and share this definiteness/ specificity feature with ASP and DE. Since the BA-NP is in the checking domain of ASP or DE after its movement to the Spec of ASPP or DEP for the reason of Case, it will also be checked against the accomplishment or achievement verb in terms of this definiteness or specificity feature via ASP or DE (cf. Chomsky 1993). Thus, the definiteness or specificity requirement on the BA-NP is very naturally linked, by the above morpho-syntactic analysis, to both the perfective and situation aspectual properties of the BA-predicate by means of the Spec-head relation without any further stipulation (cf. Note 15 in Section 2.3.2.). 2.6.4. The Definiteness Effect of the Postverbal NP It has been shown in Section 2.3.4. that if there exists an inalienably possessive or part-whole relation between the BA-NP and the postverbal NP, the postverbal NP cannot be definite; but if there exists no inalienably 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. possessive or part-whole relation between the BA-NP and postverbal NP, then the postverbal NP could be definite. This contrast is illustrated by (65) and (66) on the one hand, and by (67), (68) and (69) on the other hand, both of which are repeated below for the sake of convenience: (65) a. ta ba Lisi bang-le vi tiao tui. he BA Lisi tie-ASP one CL leg 'He tied up a leg of Lisi's.' b.*ta ba Lisi bang—le na yi tiao tui. he BA Lisi tie-ASP that one CL leg (66) a. wo ba Luxun-de shu mai—le yi ben. I BA Luxun's book sell-ASP one CL 'I sold one copy of Luxun's books.'' b.*wo ba Luxun-de shu mai—le zhe ben. I BA Luxun's book sell-ASP this CL (67) ta ba shu gei—le na wei penavou. (cf. (10b)) he BA book give-ASP that CL friend 'He gave the book to that friend.' (68) wo ba yifu bao-le zhe ae xiaobao. (cf. (14b)) I BA clothes pack-ASP the CL bundle 'I packed the clothes into the bundle.' (69) ta ba qianqkou dui-zhe na ae ren. (cf. (28)) ta BA muzzle aim-ASP that CL man 'he aimed at that person with his gun.' 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The definiteness effect exhibited by the postverbal NP in the BA-construction can be derived independently from the above morpho-syntactic analysis as a natural consequence of the Definiteness Condition (Fiengo and Higginbotham 1981:406): •'* — if x^ is free in a Specific NP. " [note: Xj^ is interpreted as an NP-trace or a variable]. Consider (65a) and (65b) first. Their derivation would be as follows by the proposed morpho-syntactic analysis: that is, the BA-NP and postverbal NP in (65a) and (65b) are generated as two components of a single noun phrase since they share an inalienably possessive relation (cf. the postulation in (92b) in Section 2.5.1.): (65a') ta^ [s& (aspp ^isij [j^p bang—le%] [^p he BA Lisi tie-ASP Inpi Iv Iv Ikp Inp2 ^jl Ik ' Ik one CL Inp3 tui]]]]]]] leg (65b')*[gj^ ta^ [g^ ba] [;^pp Lisij Ij^p bang—le%] [pp he BA Lisi tie-ASP INPl t^] ly. [y tjj] [gp [gp2 tj] [g, [g na] [jjp that Ix yi-tiao] Igp3 tui]]]]]]]] one CL leg 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (65a/), the NP2-trace tj is A-bound by its antecedent Lisij in the Spec of ASPP but is free in KP. As KP is a indefinite noun phrase, the Definiteness Condition won't apply, thus accounting for the grammaticality of (65a). However, in (65b') the NP2-trace tj, though also A-bound by Lisij in the Spec of ASPP, is free in DP which is a definite noun phrase, thus yielding a violation of the Specificity Condition.3* The same account also explains the grammaticality of (66a) and the ungrammaticality of (66b), in which the BA- NP and the postverbal NP are generated as two components of a single noun phrase since they have an inherent part- whole relation (cf. the postulation in (92b) in Section 2.5.1. and the derivation of (8b) in Section 2.5.3.): (66a/) woi [ba ba] I aspp I*«acan-de shUj [^p mai-le^] wo BA Luxun's book sell-ASP Ivp Impi ^i] Iv' Iv bjc] Ikp Ik ben] [^pz tj]]]]]] one CL The contrast between (65a) and (65b) is clearly reminiscent of the presence and absence of the definite ness effect in the following two English sentences which involve WH-movement out of an indefinite NP in (i) but out of a definite NP in (ii): (i) Who did you buy a picture of? (ii)*Who did you buy this picture of? 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (66b')*lBap woi ba] [^pp Luxun-de shUj [^p mai-loj,] wo BA Luxun's book sell-ASP Ivp I NPl Iv Iv Idp Id Ikp Ik ben] this CL InP2 bj]]]]]]] As for the BA-constructions in (67), (68) and (69), their BA-NP and postverbal NP are not generated as two components of a single noun phrase because they do not have an inalienably possessive or part-whole relation with each other (cf. the postulation in (92b)). This is clearly displayed by the derivation of (10b), (15b), (25) and (28) in Sections 2.5.4., 2.5.5. and 2.5.6. Hence, no definiteness effect would be expected to come from these BA-sentences even if their postverbal NP may be definite, since they involve no NP-movement out of their postverbal NPs. 2.7. CONCLUSION In this chapter, I have discussed the four different types of BA-constructions in Chinese and their essential properties, and I have also described the major semantic and syntactic constraints imposed upon the verb and other 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. components of the BA-constructions. In order to explain and account for these properties and constraints, I have proposed a morpho-syntactic analysis of BA-constructions and have provided both empirical evidence and theoretical motivation for such an analysis. As explicitly presented in this chapter, this morpho-syntactic analysis not only offers a uniform and principled account for the different types of BA-constructions but also captures their crucial properties and constraints as the natural consequences of general principles in universal grammar. As a result of this morpho-syntactic analysis, many important questions and issues concerning the BA-constructions may now have principled answers and solutions without referring to any particular stipulations. The impact and consequences of the morpho-syntactic analysis are not only limited to the study of the BA-constructions per se, they can be further extended to the study of the constructions of resultative verb compounds and directional verb compounds in Chinese, which are going to be discussed and presented in the next two chapters, as we proceed. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER THREE THE SYNTAX OF RESULTATIVE VERB COMPOUNDS 3.1. INTRODUCTION Like the BA-construction we dealt with in the last chapter, resultative verb compounds in Chinese are also an often-discussed topic and have drawn a lot of special attention in recent Chinese linguistic studies, as shown by several notable analyses of such compounds proposed by Huang (1988 & 1992), Li (1990 & 1993), Cheng (1992), and Gu (1992) respectively. According to Li and Thompson (1981), the Chinese resultative verb compound is always composed of two lexical verbs, of which the second verb signals the result of the action or process conveyed by the first verb. The fact that the Chinese resultative verb compound has been so interesting to many linguists is attributable to the following three factors: i) there is a complex thematic relation between the verb compound and its arguments; ii) the formation of the verb compound may involve grammatical function change of its arguments; and iii) the referentiality of its arguments might affect the interpretation or grammaticality of the resultative verb compound construction, as shown by (1), (2) and (3): 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (1) Zhangsan zhui—lei—le Lisi. (ambiguous) Zhangsan chase-tired-ASP Lisi a.'Zhangsan chased Lisi and as a result Lisi got tired.' b.'Zhangsan chased Lisi and as a result Zhangsan got tired.' c.'Lisi chased Zhangsan and as a result Lisi got tired.' (2) a. Lisi qi-lei-le ma. (ambiguous) Lisi ride-tired-ASP horse i. ' Lisi rode a horse and as a result the horse got tired.' ii.'Lisi did horse-riding and as a result he got tired.' b. Lisi qi—lei—le (na) yi pi ma. Lisi ride-tired-ASP (that) one CL horse 'Lisi rode that/one horse and as a result the horse got tired.' (3) a. Lisi he—zui—le iiu. Lisi drink-drunk-ASP wine 'Lisi drank wine and as a result he got drunk.' b.*Lisi he—zui-le fna) vi pina iiu. Lisi drink-drunk-ASP (that) one CL wine 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The three-way ambiguity of (1) shows the complex thematic relations between a resultative verb compound and its two NP arguments: i) in the reading of (la), Zhangsan is the Agent of Vi zhui ' chase ', while Lisi is both the Theme of Vi zhui and the Experiencer of V2 lei 'tired'; ii) in the reading of (lb) Zhangsan is both the Agent of Vi zhui and the Experiencer of V2 lei, whereas Lisi only remains the Theme of Vi zhui; and iii) in the reading of (Ic), Lisi becomes the Agent of Vi zhui and the Experiencer of V2 lei, while Zhangsan is turned into the Theme of Vi zhui. Thus, (Ic) involves a grammatical function change which affects the roles of Zhangsan and Lisi in a way that is exactly opposite to their interpretations in the reading of (lb). Moreover, the sentences in (2) and (3) exhibit the referentiality effects of their NP objects on their interpretations or grammaticality: that is, if the object of the verb compound is nonreferential, the sentence is ambiguous (as in (2a)) or grammatical (as in (3a)); but if the object of the verb compound is referential, then the sentence is unambiguous (as in (2b)) or ungrammatical (as in (3b)). The purpose of this chapter is to discuss various types of resultative verb compounds in Chinese and their basic properties, to describe the semantic and syntactic constraints imposed upon their verbal morphemes and their 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. arguments, to review Li's (1990 & 1993) lexical analyses of these resultative verb compounds, and to propose and argue for an alternative morpho-syntactic analysis of these resultative verb compounds: i) two lexical verbal morphemes in such a resultative verb compound project and head their own VPs; and ii) the asymmetric semantic and "implication" relation between the two events denoted by these two verbal morphemes is "canonically structurally realized" as a complementation relation between the first verbal morpheme and the VP headed by the second verbal morpheme (Vendler 1967; Dowty 1979; Chomsky 1986a; Hale and Keyser 1993). Under this morpho-syntactic analysis, the formation of resultative verb compounds in Chinese simply becomes the side effects of verb-raising and NP- movement without requiring any particular lexical rules, which is exactly on a par with the formation of the BA- constructions proposed in Chapter 2. The verb-raising is morphologically driven as the aspectual feature reflected on the second verbal morpheme must be checked against the head of the aspect phrase. The NP movement is forced by the Case Filter. They are both restricted or constrained by the principles of universal grammar. I will show how this morpho-syntactic analysis incorporates the complex thematic relations between the resultative verb compounds and their arguments, how it captures various grammatical 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. function changes involved, and how it accounts for both the semantic and syntactic constraints imposed upon the two verbal morphemes and their arguments, the explanation of which might not be available under the other analyses. I will also provide empirical motivation and theoretical arguments for this morpho-syntactic analysis. Finally, I will present some important consequences of the morpho- syntactic analysis concerning the referentiality effects of NP-arguments on the interpretation or grammaticality of the canonical resultative verb compound constructions. 3.2. TYPES OF RESULTATIVE VERB COMPOUNDS IN CHINESE As we mentioned above, the Chinese resultative verb compound is composed of two lexical verbal morphemes, of which the second verbal morpheme indicates the result of the activity conveyed by the first one. According to the syntactic and semantic relations between these two verbal morphemes and their arguments, we may divide the Chinese resultative verb compounds into eight types for the sake of convenience (their names are borrowed or adapted from Grimshaw (1990)): agentive, agentive-stative, agentive- transitive, unergative-transitive, unergative-stative, unergative, unaccusative, causative (cf. Chao 1968; Li 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and Thompson 1981; Ma 1987; Fan 1988; Zan 1989; Li 1990; Li 1990 & 1993). Of the eight types of resultative verb compounds, the agentive type is the most frequently used one, whereas the unergative-transitive type is much less often used. These resultative verb compounds and their basic properties are discussed in the following sections. 3.2.1. The Agentive Verb Compound In the agentive type of resultative verb compounds, Vi is a transitive verb, whereas V2 is a stative verb, an unaccusative verb or an unergative verb.^ This type of ^ According to Gu (1992), only unergative verbs that do not involve independent agentivity or have volitional involvement in an event or state can act as V2 to express the resultant state of the activity conveyed by Vi. Put it in another way, the logical subject of the unergative V2 should be construed as an Experiencer rather than an Agent, as shown by the contrast between (i) and (6): (i) a.*Zhangsan xia—tiao-le Lisi. (Gu 1992:78) Zhangsan frighten-jump-ASP Lisi b.*mama wei—zhang—le haizi. mother feed-grow-ASP child In (ia) and (ib), the actions expressed by the unergative verbs tiao 'jump' and zhang 'grow' do involve independent agentivity or volition, so they are not qualified as V2. In (6), however, the actions expressed by the unergative verbs ku 'cry', xiao 'laugh' and dong 'move' involve no independent agentivity or volition because they are just the results provoked by the activities conveyed by Vi. Besides, stative verbs and unaccusative verbs do not involve independent agentivity or volition, since stative 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. resultative verb compounds takes a subject and an object. The subject is the Agent of Vi and the object is both the Theme of Vi and the Experiencer of V2, as shown below: (4) a. ta ya-bian-le yi ge hezi. he press-flatten-ASP one CL box 'He pressed a box and as a result it got flat.' b. Zhangsan yao-xing—le Lisi. Zhangsan shake-awake-ASP Lisi 'Zhangsan shook Lisi and as a result Lisi awoke. ' c. Lisi zhan-bai-le Mali. Lisi fight-lose-ASP Mary 'Lisi fought Mary and as a result Mary lost.' (5) a. Lisi da—si-le na ge ren. Lisi hit-die-ASP that CL man 'Lisi hit the man and as a result the man died.' b. wo gan-pao-le yi zhi mao. I chase-flee-ASP one CL cat 'I chased a cat and as a result the cat fled.' c. ta kan-dao-le yi ke shu. he cut-fall-ASP one CL tree 'He cut a tree and as a result the tree fell.' verbs do not express action and unaccusative verbs are usually assumed not to have an external agent 6-role (cf. Perlmutter 1978; Burizo 1986). Therefore, they can serve as V2 in the resultative verb compound accordingly. 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (6) a. Lisi da-ku-le Mali. Lisi hit-cry-ASP Mary 'Lisi hit Mary and as a result Mary cried.' b. wo dou—xiao—le Lisi. I amuse-laugh-ASP Lisi 'I amused Lisi and as a result Lisi laughed.' c. women tui—dong-le yi bu qiche. we push-move-ASP one CL car 'we pushed a car and as a result it moved.' 3.2.2. The Agentive—Stative Verb Compound In the agentive-stative type of resultative verb compounds, Vi is still a transitive verb, while V2 is a stative or unaccusative verb. This type of resultative verb compounds also takes a subject and an object. The subject is both the Agent of Vi and the Experiencer of V2 while the object is only the Theme of Vi, as shown by the following examples: (7) Lisi he—zui—le jiu. Lisi drink-drunk-ASP wine 'Lisi drank wine and as a result he got drunk.' 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (8) Zhangsan chi—bao—le fan. Zhangsan eat-full-ASP rice 'Zhangsan ate rice and as a result he got full.' (9) wo chi-ni-le pingguo. I eat-bored-ASP apple 'I ate so many apples that I got tired of them.' (10) Lisi qi-lei-le jiaotache. Lisi ride-tired-ASP bike 'Lisi rode a bike and as a result he got tired.' 3.2.3. The Agentive-Transitive Verb Compound Of the agentive-transitive type of resultative verb compounds, both Vi and V2 are transitive verbs and share a subject and an object. The subject is both the Agent of Vi and the Experiencer of V2, while the object is the Theme of both Vi and V2, as exemplified below ^ According to Gu (1992), transitive verbs in V2 can not express an event which is totally independent of the event denoted by Vi. In other words, the transitive V2 cannot involve independent agentivity which is separate from that of Vi, as shown below: (i) *Lisi xie-si—le na feng xin. Lisi write-tear-ASP that CL letter (ii)*Zhangsan qie-chi-le yi kuai dangao. Zhangsan cut-eat-ASP one CL cake 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (11) ta xue—hui—le fayu. he study-understand-^P French 'He studied French and as a result he learned it.' (12) wo ting—dong—le zhe ju hua. I listen-understand-ASP this CL sentence 'I listened to this sentence and understood it.' (13) women da-ying—le na yi chang zhanzheng. we fight-win-ASP that one CL war 'We fought that war and as a result we won it.' (14) Lisi xia-shu-le yi pan qi. Lisi play-lose~ASP one CL chess 'Lisi played a chess game and lost it.' 3.2.4. The Unergative—Transitive Verb Compound In the unergative-transitive type of resultative verb compounds, Vi is an unergative verb while V2 is a In the ill-formed (i) and (ii), the events denoted by the transitive V2 si 'tear' and chi 'eat' are independent of the events denoted by Vi xie 'write' and qie 'cut' since the former require separate agentivity. However, in the well-formed (11), (12), (13) and (14), the events denoted by the transitive V2 hui 'understand', dong 'understand', ying 'win' and shu 'lose' do not involve any independent agentivity because they convey the results of the events denoted by Vi xue 'study', ting 'listen', da 'fight' and xia 'play' respectively. Therefore, their subject should be construed as an Experiencer rather than an Agent. 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. transitive verb. The subject of these resultative verb compounds is both the Agent of Vi and the Experiencer of V2 and their object is the Theme of V2, as shown below:^ (15) ta wan—wang-le zijide zhize. he play~forget-ASP own duty 'He played so much that he forgot his own duty.' (16) wo pao-diu—le yi zhi xie. I run-lose-ASP one CL shoe 'I ran and as a result I lost one shoe.' (17) Lisi zou-ying—le Zhangsan. Lisi walk-beat-ASP Zhangsan 'Lisi beat Zhangsan by walking.• (18) Zhangsan xiao—diao—le yi ke ya. Zhangsan laugh-lose-ASP one CL tooth 'Zhangsan laughed so much that he lost a tooth.' ^ That the logical subject of V2 in the unergative- transitive type of resultative verb compounds needs to be construed as an Experiencer rather than an Agent is evidenced by the fact that V2 in (15), (16), (17) and (18) cannot be modified by the Agent-oriented adverbial guyide 'deliberately' (Gu 1992:93): (i) ta (*guyide) wang-le zijide zhize. he (deliberately) forget-ASP own duty (ii) Lisi (*guyide) diu-le yi zhi xie. Lisi (deliberately) lose-ASP one CL shoe (iii) Lisi (*guyide) ying-le Zhangsan. Lisi (deliberately) beat-ASP Zhangsan (iv) Zhangsan (*guyide) diao—le yi ke ya. Zhangsan (deliberately) lose-ASP one CL tooth 156 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2.5. The Unergahive-Stative Verb Compound Of the unergative-stative type of resultative verb compounds, Vi is still an unergative verb, whereas V2 is a stative or unaccusative verb. The subject of this type of resultative verb compounds is only the Agent of Vi and their object is only the Experiencer of V2, as displayed by the examples below: (19) Zhangsan xiao-xing-le Lisi. Zhangsan laugh-awake~ASP Lisi 'Zhangsan laughed so loudly that Lisi woke up.' (20) ta ku—shi-le liang kuai shoujuan. He cry-wet-ASP two CL handkerchief 'He cried so much that two handkerchiefs got wet.' (21) Lisi chang—fan-le wo. Lisi sing-bored-ASP I 'Lisi sang so long that I became bored.' (22) Zhangsan ku—zou-le henduo keren. Zhangsan cry-leave-ASP many guest 'Zhangsan cried so much that many guests left.' (23) ta tiao—dao—le zhuo—shang—de huaping. he jump-fall-ASP desk-on's vase 'He jumped so hard that the vase on the desk fell. ' 157 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2.6. The Unergative Verb Compound In the unergative type of resultative verb compounds Vi is still an unergative verb and V2 is still a stative or unaccusative verb, like the unergative-stative type of verb compounds depicted in the above section. Unlike the unergative-stative type of verb compounds, the unergative type of verb compounds only takes a subject which is the Agent of Vl and the Experiencer of V2, as shown below; (24) L isi xiao-feng-le. Lisi laugh-insane-ASP 'Lisi laughed to the extent of becoming mad.' (25) w o zou— le i- le . I walk-tired-ASP 'I walked so much that I got tired.' (26) Laoshi j iang-fan-le. teacher talk-bored-ASP 'The teacher talked so long that he got bored.' (27) Mali chi-pang-le. Mary eat-fat-ASP 'Mary ate so much that she gained weight.' (28) Zhangsan shui— hutu— le. Zhangsan sleep-confuse-ASP 'Zhangsan slept so much that he got confused.' 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2.7. The Unaccusative Verb Compound The unaccusative type of resultative verb compounds can be divided into two subclasses, regarding the type of verbs and the relation between the subject and two verbs. In the first subclass, Vi is a transitive verb, V2 is an unaccusative or stative verb, and their subject is both the Theme of Vi and the Experiencer of V2; (29) na ke shu kan-dao-le. That CL tree cut-fall-ASP 'That tree was cut and as a result it fell.' (30) wode chenyi liang—gan—le. my shirt hang-dry-ASP 'My shirt was hung and as a result it became dry. ' (31) diren gan-pao-le. enemy chase-flee-ASP 'The enemy was chased and as a result they fled.' (32) zhe pi ma qi-lei—le. this CL horse ride-tired-ASP 'This horse was ridden so much that it got tired.' Although V2 remains an unaccusative or stative verb in the second subclass, Vl in the second subclass is an 159 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. unergative verb. Their subject is only the Experiencer of V2 and does not have any thematic relation with Vi; (33) duzi xiao—tong—le belly laugh-ache-ASP 'Someone laughed so much that his belly ached.' (34) shouj uan ku-shi—le. handkerchief cry-wet-ASP 'someone cried so much that his handkerchief got wet. ' (35) toufa chou-bai—le. hair depress-whiten-ASP 'Someone was so depressed that his hair became grey.' 3.2.8. The Causative Verb Compound The causative type of resultative verb compounds can also be divided into two subclasses in terms of the verb type and the relation between the subject, two verbs and object. In the first subclass, Vi is an unergative verb and V2 is an unaccusative or stative verb. The object is the Experiencer of both Vi and V2, while the subject does not have any thematic relation with either Vi or V2. In 160 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. addition, there exists a Causer-Causee relation between the subject and the object: that is, the former causes something to happen to the latter, as exemplified below: (36) na ping jiu zui-dao-le Lisi. that CL wine drunk^fall-ASP Lisi 'That bottle of wine made Lisi drunk and fall.' (37) zhe jian shengyi pao-lei-le Zhangsan. this CL business run-tired-ASP Zhangsan 'This business made Zhangsan run and tired.' (38) na chang jihuang e—si-le henduo ren. that CL famine starve-die-ASP many people 'That famine made many people starve and die.' But in the second subclass, Vi is a transitive verb though V2 remains an unaccusative or stative verb. The object is both the Agent of Vi and the Experiencer of V2, whereas the subject is only the Theme of Vi. There still exists a Causer-Causee relation between the subject and the object, as illustrated by the examples below: (39) na ping jiu he-zui-le Lisi. that CL wine drink-drunk-ASP Lisi 'Lisi drank that bottle of wine and as a result he got drunk.' 161 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (40) zhe pi ma qi—lei—le ta. this CL horse ride-tired-ASP he 'He rode the horse and as a result he got tired.' (41) na tang che deng-ji-le Zhangsan. That CL bus wait-irritate-ASP Zhangsan 'Zhangsan waited for that bus for so long that he became impatient.' To sum up, Vl in the eight types of resultative verb compounds presented above is either a transitive verb or an unergative verb, but cannot be an unaccusative verb or a stative verb. This is because only transitive verbs or unergative verbs, but not unaccusative or stative verbs, may denote a process or an action and involve agentivity. By contrast, V2 in these eight types of resultative verb compounds is usually an unaccusative or stative verb that indicates a change of state or a result of the process or action conveyed by Vi. Although V2 may be an unergative or transitive verb, it does not involve any agentivity or volition which is independent of Vi. Thus, the thematic subject of V2 must be construed as an Experiencer rather than an Agent (see Section 3.3.3 in the following for the discussion of some impossible thematic relations between a resultative verb compound and its arguments). 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3. THE PROPERTIES AND CONSTRAINTS 3.3.1. The Aspectual features According to Chao (1968), Li and Thompson (1981), Ma (1987) and Fan (1988), the resultative verb compound in Chinese normally takes the perfective aspect marker -le, as shown by the examples in the above sections. When the resultative verb compound takes no object, such an aspect marker will be obligatory (cf. Note 8 in Chapter 2): (42) a. Lisi xiao-feng-le. (= (24)) Lisi laugh-insane-ASP 'Lisi laughed to the extent of becoming mad.' b.*Lisi xiao-feng. Lisi laugh-insane (43) a. na ke shu kan-dao-le. (= (29)) That CL tree cut-fall-ASP 'That tree was cut and as a result it fell.' b.*na ke shu kan-dao. That CL tree cut-fall Sometimes the resultative verb compound may also take the progressive aspect marker -zhe or the experiential aspect marker -guo, as illustrated by the examples below: 163 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (44) Lisi zheng—kai—zhe yanjing bu shuo hua. Lisi stretch-open-ASP eye not say speech 'Lisi made and kept his eyes open without saying a word.' (45) Zhangsan die-duan—guo tui. Zhangsan fall-break-ASP leg 'Zhangsan once fell and broke his leg.' Although the focuses of the three aspects are different, what they share concerning the resultative verb compound is that they all indicate the completion of an action or the closure of an event denoted by the resultative verb compound: i) the perfective aspect marker —le indicates the completion of an action or the closure of an event; ii) the progressive aspect marker -zhe signals a change of state resulting from the completion of an action or the closure of an event; and iii) the experiential aspect marker -guo conveys that an event has been experienced at least once before (cf. Li and Thompson 1981; and Note 9 in Chapter 2). Therefore, they are all compatible with the resultant state indicated by V2 in the resultative verb compound. 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3.2. The Referentiality Effect As mentioned in Section 3.2.2., the object taken by the agentive-stative type of resultative verb compounds is the Theme of their Vi. In addition to its thematic role, this object is also required to be nonreferential, otherwise the resultative verb compound construction will be ungrammatical: (46) a. Lisi he—zui—le jiu. (= (7)) Lisi drink-drunk-ASP wine 'Lisi drank wine and as a result he got drunk.' b.*Lisi he—zui—le (na) yi ping jiu. Lisi drink-drunk-ASP (that) one CL wine (47) a. Zhangsan chi-bao—le fan. (= (8)) Zhangsan eat-full-ASP rice 'Zhangsan ate rice and as a result he got full. ' b.*Zhangsan chi-bao-le (na) yi wan fan. Zhangsan eat-full-ASP (that) one CL meal (48) a. wo chi-ni-le pingguo. (= (9)) I eat-bored-ASP apple ' I ate so many apples that I got tired of them. ' b.*wo chi-ni-le (zhe) yi ge pingguo. I eat-bored-ASP (this) one CL apple 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. But for the second subclass of the causative type of resultative verb compounds, their subject, which is also the Theme of Vi (cf. the discussion in Section 3.2.8.), is required to be referential, otherwise, the resultative verb compound construction involved will be ungrammatical too, as shown below: (49) a. na ping jiu he-zui—le Lisi. (= (39)) that CL wine drink-drunk-ASP Lisi 'Lisi drank that bottle of wine and as a result he got drunk.' b.*jiu he-zui-le Lisi. wine drink-drunk-ASP Lisi (50) a. zhe pi ma qi-lei-le ta. (= (40)) this CL horse ride-tired-ASP he 'he rode the horse and, as a result, got tired. ' b.*ma qi-lei-le ta. horse ride-tired-ASP he (51) a. na tang che deng-ji-le Zhangsan. That CL bus wait-irritate-ASP Zhangsan 'Zhangsan waited for that bus for so long that he became impatient.' (= (41)) b.*che deng-ji-le Zhangsan. bus wait-irritate-ASP Zhangsan 166 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This referentiality effect further involves the NP- arguments of such canonical resultative verb compounds as qi—lei 'ride-tired', of which Vi has both an Agent 0-role and a Theme 8-role and V2 has only an Experiencer 8-role. That is, if the object of qi-lei is nonreferential, the resultative verb compound construction will be ambiguous; but if the object of qi-lei is referential, then the verb compound construction will become unambiguous, as shown by (2) above, which is rewritten below as (52): (52) a. Lisi qi-lei-le ma. (ambiguous) Lisi ride-tired-ASP horse i.'Lisi rode a horse and as a result it got tired. ii.'Lisi did horse-riding and as a result he got tired.' b. Lisi qi-lei-le fna) yi pi ma. Lisi ride-tired-ASP (that) one CL horse 'Lisi rode that/one horse and as a result the horse got tired.' 3.3.3. The Thematic Relations Though there are various complex thematic relations between the resultative verb compound and its arguments, 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. as shown by Section 3.2.1. through Section 3.2.8., some thematic relations do not exist between them (Li 1990). Consider the agentive-transitive type of resultative verb compounds, which are described in Section 3.2.3., their subject cannot be the Theme of V2 (in contrast with its licit status as both the Agent of Vi and the Experiencer of V2), and their object cannot be the Agent of V2 (in contrast with its licit status as the Theme of both Vi and V2), as shown by the examples below (cf. Note 2): (53)*Lisi xue—shang-le wushu. Lisi learn-injure-ASP martial-art 'Lisi learned martial art and as a result it injured him.' (54) ta zhan—sheng—le san wei yingguo qishou. he fight-win-ASP three CL British chess-player *'He fought three British chess-players and as a result they beat him.' In addition, the thematic roles of their subject cannot be exchanged with those of their object: that is, their subject cannot be the Theme of both Vi and V2, and their object cannot be the Agent of Vi and the Experiencer of V2: 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (55)*zhe ju hua ting-dong-le wo. (cf. (12)) the CL words listen-understand-ASP I 'I listened to this sentence and understood it.' Now, let us consider the unergative-transitive type of resultative verb compounds (cf. Section 3.2.4.), whose subject cannot be the Theme of V2 (in contrast with its licit status as both the Agent of Vi and the Experiencer of V2), and whose object cannot be the Agent of V2 (in contrast with its licit status as the Theme of V2); (56) Lisi wan-wang—le zijide laopo. Lisi play-forget-ASP own wife *'Lisi played so much that his wife forgot him.' (57) ta pao-ying-le Mali. he run-beat-ASP Mary *'He ran against Mary and as a result she beat him. ' The thematic roles of their subject and object cannot be exchanged either; that is, their subject cannot be the Theme of V2, and their object cannot be the Agent of Vi or the Experiencer of V2: (58)*tade zhize wan-wang-le ta. (cf. (15)) his duty play-forget-ASP he 169 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.4. LI'S LEXICAL ANALYSES OF RESULTATIVE VERB COMPOUNDS 3.4.1. Li's (1990) Analysis Li (1990) argues that resultative verb compounds in Chinese are formed in lexicon and their formation can be accounted for by referring to Case theory and the three assumptions in the following; theta-role identification (Higginbotham 1985), theta-role prominency (Grirashaw 1990) and head-feature percolation (Lieber 1983): (59) a. THETA-ROLE IDENTIFICATION: Theta-roles from two verbs can be identified into one, which is then assigned to their argument NP. b. THETA-ROLE PROMINENCY: The assignment of theta- roles is in accordance with their prominence in the thematic hierarchy (Agent (Experiencer (Goal/ Source/Location(Theme)))) (Jackendoff 1972), which aligns with the structural prominence of the arguments of verbs. c. HEAD-FEATURE PERCOLATION: The Theta-role (or 6- role) prominency of the head must be strictly maintained in the theta-grid of the resultative verb compound. 170 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Li also suggests that there exists a causal relation between the two verbal morphemes of the resultative verb compound, and this causal relation can be represented by using a semantic operator CAUSE, as displayed below (Note that (...) stands for the theta-grids of the verbs): (60) V (...) <==> Vl(...) CAUSE V2(...) Li further assumes that Vi acts as the head of the resultative verb compound because the categorial feature of Vl determines that of the verb compound; that is, Vi is always a verb while V2 is what would be an adjective in many other languages. Now, let us see how resultative verb compounds in Chinese are analyzed by Li's lexical approach.* First, consider the resultative verb compound in (1), which is rewritten below; * For the convenience of illustration, I will follow Li (1990) to use <...> to mark a theta-grid. The 0-roles of Vl are represented as whole numbers 1, 2, ..., the 0- roles of V2 are represented as letters a, b , and marks 0-role identification. In addition, <1> stands for the most prominent 0-role of Vi, <2> for the second most prominent 0-role of Vi, etc., and <a> stands for the most prominent 0-role of V2, <b> for the second most prominent 0-role of V2, etc. 171 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (1) Zhangsan zhui—lei—le Lisi. (ambiguous) Zhangsan chase-tired-ASP Lisi a.'Zhangsan chased Lisi and as a result Lisi got tired.' b.'Zhangsan chased Lisi and as a result Zhangsan got tired.' c.'Lisi chased Zhangsan and as a result Lisi got tired.' In (1), Vl zhui 'chase' has two 0-roles represented by <1, 2>, and V2 lei 'tired' has one 6-role represented by <a>. According to Li (1990), the first two readings of (1) (i.e. the agentive reading in (la) and the agentive- stative reading in (lb); see Sections 3.2.1. and 3.2.2.) result from identifying the sole 8-role of V2 with either one of the two 8-roles of Vi and assigning the identified 8-role to one of the two arguments of the verb compound, as illustrated by his analyses below: (la') <1, 2-a> (lb') <l-a, 2> Zhangsan Lisi Zhangsan Lisi In (la'), the Agent 6-role of Vi is assigned to Zhangsan while its Theme 8-role is identified with the Experiencer 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6-role of V2. This identified 8-role is then assigned to Lisi. The two 8-role assignments are in accordance with the thematic hierarchy by realizing the Agent Zhangsan in the structurally more prominent subject position and the Experiencer/Theme Lisi in the structurally less prominent object position. Thus, Zhangsan is interpreted as the one who did chasing, while Lisi is interpreted as the one who was both chased and tired. However, in (lb'), the Theme 8-role of Vi is assigned to Lisi whereas its Agent 8-role is identified with the Experiencer 8-role of V2. This identified 8-role is then assigned to Zhangsan. The two 8-role assignments also observe the thematic hierarchy by realizing the Agent/Experiencer Zhangsan in the subject position and the Theme Lisi in the object position. Thus, it is Zhangsan who did chasing and got tired, while Lisi was only chased. Li's lexical analysis also rules out an impossible reading of (1) correctly, as shown by (61); (61) Zhangsan zhui-lei—le Lisi. *<2-a, 1> Zhangsan chase-tired-ASP Lisi { | *'Lisi chased Zhangsan and as a Zhangsan Lisi result Zhangsan got tired.' 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The reading in (61) results from assigning the Agent 0- role of Vl to Lisi, identifying the Theme 6-role of Vi with the Experiencer 0-role of V2, and assigning this identified 0-role to Zhangsan. With the most prominent Agent 0-role of Vi being assigned to the object Lisi and the least prominent Theme 0-role of Vi being assigned to the subject Zhangsan, the thematic hierarchy in (59b) is violated, thus making the reading in (61) unavailable. Li's lexical analysis further accounts for some other well-formed resultative verb compounds and excludes some ill-formed ones as well, as shown below: A) The Agentive-Transitive Type (cf. Sections 3.2.3. and 3.3.3.): Both Vi and V2 have two 0-roles; their subject is both the Agent of Vi and the Experiencer of V2; and their object is the Theme of both Vl and V2: (62) a. ta xia-shu-le na pan qi. <l-a, 2-b> He play-lose-ASP that CL chess | j 'He played that chess game and ta na-pan-qi lost it. ' b.*na pan qi xia-shu-le ta. *<2-b, l-a> the CL chess play-lose-ASP he | | na-pan-qi ta 174 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. c.*Lisi xue-shang-le wushu. *<l-b, 2-a> Lisi learn-injure-ASP martial-art { | 'Lisi learned martial art and as Lisi wushu a result it injured him.' d.*wushu xue-shang—le Lisi. *<2-a, l-b> martial-art learn-injure-ASP Lisi | | wushu Lisi B) The Unergative-Transitive type (cf. Section 3.2.4. and 3.3.3.); Vl has one 0-role and V2 has two 6-roles; their subject is both the Agent of Vi and the Experiencer of V2; and their object is the Theme of V2: (63) a. wo pao—diu-le yi zhi xie. <l-a, b> I run-lose-ASP one CL shoe | | 'I ran and as a result I lost wo yi-zhi-xie a shoe.' b.*yi zhi xie pao-diu-le wo. *<b, l-a> one CL shoe run-lose-ASP I I I I I yi-zhi-xie wo c.Lisi wan-wang-le wo. *<l-b, a> Lisi play-forget-ASP I | | *'Lisi played so much that I Lisi wo forgot him.' 175 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. d.Lisi wan-wang-le wo. *<a, l-b> Lisi play-forget-ASP I { j *'I played so much that Lisi Lisi wo forgot me.' C ) The Unergative-stative Type ( cf. Section 3.2.5.): Both Vl and V2 have one 0-role; their subject is the Agent of Vl; and their object is the Experiencer of V2: (64) a. Lisi ku-xing-le wo. <1, a> Lisi cry-awake-ASP I | | 'Lisi cried so loudly that Lisi wo I woke up.' b. Lisi ku-xing-le wo. *<a, 1> Lisi cry-awake-ASP I } j *'I cried so loudly that Lisi Lisi wo woke up.' D) The Unergative Type (cf. Section 3.2.6.): Both Vi and V2 have one 8-role and share a subject; and their subject is the Agent of Vi and the Experiencer of V2: (65) Lisi zou—lei—le. <l-a> Lisi walk-tired-ASP j 'Lisi walked so much that he got tired. ' Lisi 176 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. However, Li's (1990) lexical analysis cannot account for the third possible reading of (1) (i.e. the Causative reading in (Ic); also see (39), (40) and (41) in Section 2.3.8.), because the 6-role assignments involved in this reading violate the thematic hierarchy: the thematically less prominent Theme 8—role of Vi zhui is assigned to the subject Zhangsan, whereas its thematically most prominent Agent 8-role is assigned to the object Lisi: (Ic) Zhangsan zhui—lei—le Lisi. Zhangsan chase-tired-ASP Lisi 'Lisi chased Zhangsan and as a result Lisi got tired.' (lc')*<2, l-a> I I I I Zhangsan Lisi 3.4.2. Li's (1993) Revised Analysis In order to account for the causative reading of ( 1 ) in (Ic), Li (1993) assumes the existence of a causative hierarchy that determines both the Causer-Causee relation and the argument structure of resultative verb compounds in the following manner: 177 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (66) 0-roles could be assigned against the thematic hierarchy only if the arguments receiving them observe the causative hierarchy (Causer (Causee (...))) (Grimshaw 1990), which also aligns with the structural prominence of the arguments. The interpretation of Causer and Causee crucially depends on the two verb morphemes of a resultative verb compound, on the grammatical functions of their arguments, and on the ways their 8-roles are assigned, as formulated by Li (1993:12): (67) a. The subject receives the c-role Causer from a V-V compound if it receives a 8-role only from VI ; b. The object receives the c-role Causee from a V-V compound if it receives a 8-role at least from V2. Under Li's new approach, 8-roles are assigned in a random order, contrary to his former approach under which a less prominent 8-role must be assigned before a more prominent 8-role (cf. Li 1990). After the random 8-role assignment, there will be two checking procedures, of which the first one is the causative hierarchy and the second one is the 178 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. thematic hierarchy. If the subject and object arguments are interpreted respectively as Causer and Causee under the causative hierarchy, their 6-role assignment does not need to be in accordance with the thematic hierarchy; but if they lack the Causer and Causee readings, the thematic hierarchy will then determine how 8-roles are assigned to them, as in the former approach. The causative reading in (Ic) is now available under Li's new approach. That is, since Zhangsan receives a 8- role (i.e. <2>) only from Vi zhui and Lisi receives a 8- role (i.e. <a>) at least from V2 lei, Zhangsan and Lisi are interpreted as Causer and Causee respectively. As a result, the causative hierarchy 'overrides' the thematic hierarchy by having the aspectually more prominent Causer Zhangsan in the subject position and the aspectually less prominent Causee Lisi in the object position: (lc«) <l-a, 2> Zhangsan Lisi Causer Causee 'Lisi chased Zhangsan and as a result Lisi got tired.' To the very effect that the overriding force of the causative hierarchy is activated by the identification of Causer and Causee, Li's new approach makes the following 179 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. prediction: a Causer may be associated with the Theme of Vi, and a Causee may be related to the Agent of Vi. In addition, it still accounts for the well-formed and ill- formed resultative verb compounds given in Section 3.4.1. 3.4.3. The Problems with Li's Lexical Analyses Although Li's lexical approach is very appealing, it faces several serious problems. First of all, a Causer in the subject position does not necessarily receive a conceivable 6-role from Vi, contrary to Li's definition of Causer in (67a). This is evidenced by the nonthematic relation between a Causer and an unergative Vi in (36), (37) and (38), which are repeated below: (36) na ping jiu zui-dao—le Lisi. that CL wine drunk-fall-ASP Lisi 'That bottle of wine made Lisi drunk and fall.' (37) zhe jian shengyi pao—lei—le Zhangsan. this CL business run-tired-ASP Zhangsan 'This business made Zhangsan run and tired.' (38) na chang jihuang e-si-le henduo ren. that CL famine starve-die-ASP many people 'That famine made many people starve and die.' 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In these three sentences, the Causers na ping jiu 'that bottle of wine', zhe jian shengyi 'this business' and na chang jihuang 'that famine' in the subject position are not thematically associated with the unergative Vi zui 'drunk', pao 'run' and e 'starve' in any way. Put it in another way, in (36), it is Lisi, not na ping jiu, that got drunk; in (37), it is Zhangsan, not zhe jian shengyi, that did running; and in (38), it is henduo ren, not na chang jihuang, that were hungry. To accommodate this causative type of resultative verb compounds, Li (1993) argues that their Vi such as zui in (36) might be used as a transitive verb with the causative meaning of 'making ... drunk', as shown below: (68) zhe zhong jiu zui ren. this CL wine drunk people 'This kind of wine gets people drunk.' Under this analysis, a plausible account of (36) would be that it is the transitive verbal morpheme zui rather than its unergative counterpart that serves as Vi in (36), and that since the subject na ping jiu receives a 0-role only from Vi zui and the object Lisi gets one 8-role at least from V2 dao, na ping jiu and Lisi are licitly interpreted as Causer and Causee respectively, as displayed below: 181 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (36') <1, 2-a> I I I I na ping jiu Lisi Causer Causee This account, though very attractive, is problematic in several respects. First, although Vi zui in (36) might be used as a transitive verb, as in (68), Vi pao in (37) and VI e in (38) may not: (69)*zhe jian shengyi pao ren. this CL business run people (70)*na chang jihuang e ren. that CL famine starve people Secondly, Vi zui in (36) could hardly be construed as a full-fledged transitive verb or causative verb, as pointed out by Gu (1992). Consider (68), which Li uses to support his account of (36).^ in this sentence, the ^ Incidentally, Li's treatment of the single verb zui in (68) as a transitive verb which assigns a Causer role to the subject zhe zhong jiu 'this kind of wine' is not consistent with his fundamental assumptions "... the Causer-Causee readings crucially depends on two verbs forming a resultative V-V structure and on the specific ways their 6-roles are assigned." and "... I assume that C-role assignment is a property of the V-V compound." (Li 1993:12) 182 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. object ren 'people' is a generic noun and the verb zui is used as an individual-level predicate (cf. Carlson 1977; Kratzer 1988; Diesing 1992). If its object is replaced by a referential NP or its verb is used as a stage-level predicate, the sentence would become ungrammatical: (71) a.*zhe zhong jiu zui Lisi. this CL wine drunk Lisi b.*zhe zhong jiu zuotian zui—le ren. this CL wine yesterday drunk-ASP people The ungrammaticality of (7lab) strongly argues that Vi zui in (36) is not a truly causative verb or it does not assign a so-called Causer role to its subject zhe ping jiu. If it were, it would behave like a normal causative verb such as the English verb ' frighten ', which can take a referential NP object and can be used as a stage-level predicate, as shown below: (72) That storm frightened John. That the subject zhe ping jiu in (36) may not receive a 0-role from Vi zui is also supported by the fact that its 8-grid contains only an Experiencer 8-role which goes to 183 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the object Lisi, since Vi zui can be used unaccusatively with Lisi being its surface subject:® (73) Lisi zui—le. Lisi drunk-ASP 'Lisi got drunk.' Thus, it is not clear how the subjects zhe ping jiu, zhe jian shengyi and na chang jihuang in (36), (37) and (38) receive their Causer roles under Li's lexical analysis without assuming, against our intuition, that their Vi zui, pao and e are transitive verbs and have two 0-roles to discharge. The second problem with Li's lexical approach also concerns the relation between causativity and theta-role assignment. If a Causee, according to Li's definition in (67b), is an object receiving a 8-role at least from V2 of a verb compound, a sentence with such a Causee should also contain a Causer because the occurrence of a Causee entails the existence of a Causer intuitively. However, ® V2 dao 'fall' in (36) does not assign a 8-role to the subject zhe ping jiu either, as dao is only construed with the object Lisi and can be used unaccusatively with Lisi being its surface subject: (i) Lisi dao-le. Lisi fall-ASP 'Lisi fell.' 184 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the fact is that the occurrence of such a Causee does not guarantee the existence of a Causer or necessarily yield a causative structure, as evidenced by the unergative- transitive type of verb compounds in Section 3.2.4. and (63a) above. Consider (63a), which is rewritten below: (63a) wo pao—diu—le yi zhi xie. <l-a, b> I run-lose-ASP one CL shoe j | ' I ran and as a result I lost wo yi-zhi—xie a shoe.' In (63a) the object yi zhi xie 'a shoe' receives a 0-role from V2 diu 'lose' and should be taken as a Causee under Li's analysis. However, this sentence contains no Causer because the subject wo 'I', which receives an identified 8-role from Vl pao 'run' and V2 diu, does not qualify as a Causer by Li's definition in (67a). What is strongly suggested by (63a) here and by (36), (37) and (38) above is that causativity may not have any relation with 8-role assignment, and that the complex event structure involved in the resultative verb compound (that is, the process or action expressed by Vi and its resultant state conveyed by V2; cf. Grimshaw 1990) may not be the only criterion in determining causativity. 185 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The third problem with Li ' s lexical approach lies in its random 0-role assignment, which is too powerful or less constrained to the extent that any resultative verb compound in Chinese would be expected to be ambiguous (as the random 8-role assignment applies indiscriminately) if more than one result of its thematic operation meets with the causative hierarchy or thematic hierarchy. However, this prediction does not hold for such resultative verb compounds as the ones in (36), (37) and (38). Consider (36) again, which would acquire two unavailable readings under the random 6-role assignment (see (lb') and (Ic') for Li's analysis of the corresponding agentive-stative reading in (lb) and causative reading in (Ic)): (36") a. <l-a, 2> 1 I na ping jiu Lisi *'The bottle of wine got Lisi drunk and as a result it fell.' b. <l-a, 2> na ping jiu Lisi *'Lisi made that bottle of wine drunk and as a result Lisi fell.' 186 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The unavailable reading in (36"b) also suggests that the causative reading in (Ic) may not be derived in the way as described in Li's lexical approach. The fourth problem with Li's lexical approach lies in its 8-role identification. According to Li (1990 & 1993), the 6-role identification is obligatory when the two verbs of a verb compound have more 8-roles than the available structural Cases. This, in fact, will rule out the unaccusative type of resultative verb compounds (cf. Section 3.2.7. ) or assign them the wrong interpretations. Consider (33), which is repeated below; (33) duzi xiao-tong—le belly laugh-ache-ASP 'Someone laughed so much that his belly ached.' In (33), the subject argument duzi 'belly' can only be interpreted as the Experiencer of V2 tong 'ache'. As no separate overt argument is available to bear the Agent 8- role of the unergative Vi xiao 'laugh', this Agent 8-role must be either identified with the Experiencer 8-role of V2 and be then assigned to the subject argument duzi, or simply left unassigned: 187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (33') a. *<l-a> b. *<1, a> I I I I duzi If this Agent 0-role is not assigned, as in (33'b), the necessity of 8-role identification, which Li's analysis is crucially based on (i.e. 6-roles must be exhaustively identified and be assigned when the two verbs of a verb compound have more 8-roles than the available structural Cases), would be violated. But if this Agent 8-role is identified with the Experiencer 8-role of V2 and assigned to the subject duzi, as in (33'a), duzi would be taken as the one that both laughed and ached obviously a wrong interpretation. This is also shown by the incapability of Li's lexical analysis to account for another type of unaccusative verb compounds described in Section 3.2.7. Consider (29), the 8-role identification to whose verb compound would also assign it a wrong interpretation: (29) na ke shu kan-dao-le. That CL tree cut-fall-ASP 'That tree was cut and as a result it fell.' *<l-2-a> I I na ke shu 188 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hence, Li's lexical approach is too constrained to allow the unaccusative type of verb compounds unless he assumes a lexical rule to delete their Agent 0-role (Fagan 1988). Finally, Li's lexical approach fails to capture the referentiality effects on the grammaticality or meaning of some canonical resultative verb compound constructions (cf. Section 3.3.2.). To be more specific, it does not explain why the object of the resultative verb compound in (46), (47) and (48) above must be nonreferential: (46) a. Lisi he—zui—le jiu. Lisi drink~drunk-ASP wine 'Lisi drank wine and as a result he got drunk.' b.*Lisi he—zui-le (na) yi ping jiu. Lisi drink-drunk-ASP (that) one CL wine (47) a. Zhangsan chi—bao—le fan. Zhangsan eat-full-ASP rice ' Zhangsan ate rice and as a result he got full. ' b.*Zhangsan chi—bao—le (na) yi wan fan. Zhangsan eat-full-ASP (that) one CL meal (48) a. wo chi—ni—le pingguo. I eat-bored-ASP apple ' I ate so many apples that I got tired of them. ' b.*wo chi—ni—le (zhe) yi ge pingguo. I eat-bored-ASP (this) one CL apple 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Nor does it account for the unambiguity of (52b), where the object of the verb compound is referential: (52b) Lisi qi—lei-le fnal yi pi ma. Lisi ride-tired-ASP (that) one CL horse 'Lisi rode that/one horse and as a result the horse got tired.' Moreover, it has nothing to say cüDout the referentiality requirement on the subject of the verb compounds in (49), (50) and (51), which are repeated below: (49) a. na ping jiu he-zui—le Lisi. (= (39)) that CL wine drink-drunk-ASP Lisi 'Lisi drank that bottle of wine and as a result he got drunk.' b.*jiu he—zui—le Lisi. wine drink-drunk-ASP Lisi (50) a. zhe pi ma qi—lei—le ta. (= (40)) this CL horse ride-tired-ASP he 'he rode the horse and as a result he got tired.' b.*ma qi-lei—le ta. horse ride-tired-ASP he 190 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (51) a. na tang che deng-ji—le Zhangsan. That CL bus wait-irritate-ASP Zhangsan 'Zhangsan waited for that bus for so long that he became impatient.' (= (41)) b.*che deng-ji-le Zhangsan. bus wait-irritate^ASP Zhangsan In other words, Li's lexical approach would generate the ill-formed sentences in (46b), (47b), (48b), (49b), (50b) and (51b) and would make the sentence in (52b) ambiguous incorrectly. Besides, Li's lexical approach would also assign two unavailable readings to each of (46a), (47a) and (48a). Consider (47a), whose two unavailable readings below are totally legitimate under Li's lexical approach: (47a') i. <1, 2-a> I I I I Zhangsan fan *'Zhangsan ate rice and as a result it got full. ' ii. <l-a, 2> Zhangsan fan *'Rice ate Zhangsan and as a result it got full. ' Hence, it is not clear how these referentiality effects could be accommodated within Li's lexical approach. 191 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To summarize, Li's lexical approach, under which the formation of resultative verb compounds in Chinese should be lexically determined by the random 0-role assignment, 6-role identification, causative hierarchy and thematic hierarchy, runs into the problems of both overgeneration and undergeneration, as displayed by the examples above. The discussion in this section also indicates that the 6- grid of the two verbal morphemes of the resultative verb compound may not derive causativity, nor can the complex event structure involved in the resultative verb compound guarantee causativity. Therefore, causativity must have its independent status in grammar, and its effect on the organization of the argument structure of the resultative verb compounds needs to be independently implemented. In the following sections, I am going to propose and argue for a morpho-syntactic analysis of the eight types of resultative verb compounds examined above, to show how this analysis incorporates the thematic relations between these verb compounds and their arguments, how it captures the grammatical function change involved, how it accounts for the semantic and syntactic constraints upon the verb compounds and their arguments, and how it explains the referentiality effect on the grammaticality or meaning of some canonical resultative verb compound constructions. 192 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.5. A MORPHO-SYNTACTIC ANALYSIS OF RESULTATIVE VERB COMPOUNDS IN CHINESE 3.5.1. Theoretical Background As the Chinese resultative verb compound is always composed of two lexical verbal morphemes, of which the second one indicates the result of the action or process conveyed by the first one, the events associated with the two verbal morphemes can be nicely captured by Vendler's and Dowty's complex event structure for accomplishment verbs. According to Vendler (1967) and Dowty (1979), an accomplishment verb denotes a complex event that consists of an activity sub-event and a resultant state sub-event. Each of these two sub-events projects its own predicate structure. The combination and alignment of the two sub event structures then form a complex event structure for the entire sentence: (74) event / \ activity state The complex event structure in (74) directly applies to the events designated by the two verbal morphemes of a 193 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chinese resultative verb compound, because the Chinese resultative verb compound is a class of accomplishment verb (cf. Cheng 1989; Liu 1992; see Note 12 and 13 in Chapter Z) ; that is, the process or action expressed by the first verbal morpheme matches the activity part of the complex event, and the resultant state conveyed by the second verbal morpheme corresponds to the state part of the complex event. As the resultant state conveyed by the second verbal morpheme is activated or "triggered" by the activity denoted by the first verbal morpheme, a very natural way to capture the fact is to have the two events project their separate predicate structures with the two verbal morphemes as heads, and to render the asymmetric semantic or "implication" relation between the two events as a complementation relation between the first verbal morpheme and the VP-predicate headed by the second verbal morpheme (cf. Baker 1988; Hale and Keyser 1993); that is, the first verbal morpheme asymmetrically c-commands the second verbal morpheme because the activity denoted by the former "implicates" the state denoted by the latter: (75) a. eventi — > eventz b. VPi NP V'l VI VP2 194 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The complementation relation between Vi and VP2 is quite plausible as the syntactic embedding usually corresponds to a "semantic" composite in which the event denoted by the subordinate VP is a proper part of the event denoted by the matrix verb (Hale and Keyser 1993).^ The syntactic structure in (75b) and its associated semantic relation in (75a) nicely capture the so-called causal relation (which, according to Li (1990), exists between the two verbal morphemes of the resultative verb compound; see (60)) by virtue of the syntactic relation itself and by the event associated with the Vi category. In other words, the external argument of the matrix verb VI has an unambiguous syntactic relation with Vi, and its semantic connection to the syntactic structure (that is, agent) is also unambiguous. ^ According to Hale and Keyser (1993), the parallel semantic and syntactic characterizations of the relation between the events denoted by lexical categories suggest that thematic roles are derivative of lexical syntactic relations and that the argument structure is determined wholly by lexical categories and their projections. This naturally explains why there are so few thematic roles, as thematic roles are nothing but the syntactic relations which are determined by the lexical categories and their projections and which are limited by the small inventory of lexical categories (V, N, A, P) and by the principles of Unambiguous Projection and Full Interpretation. This also explains why 0-role assignment is in conformity with Baker's (1988) Uniformity of Theta Assignment Hypothesis, as the lexical-syntactic argument structure (e.g. (75b)) has biunique structural-thematic relationships for all lexical items. 195 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Moreover, the syntactic structure in (75b) further comprises what is assumed by Li's lexical analysis with respect to the headiness of Vi (cf. Section 3.4.1.); that is, Vl acts as the head of the resultative verb compound and determines its categorial status because Vi is always a verb and V2 is what would be an adjective in many other languages. Based on the syntactic structure and its associated semantic relation in (75ab), we can establish the basic structure of the resultative verb compound construction in Chinese as follows, assuming a version of Chomsky's (1993) clausal structure; (76) AGRP Spec AGR' AGR^ ^^ASPP Spec^^ ^^ASP ASP VPI Spec V'l VI VP2 In ( 7 6 ) , AGRP corresponds to AGRPg while A SP P is parallel to AGRPq (AGRPg and AGRP^ are present in Chomsky's ( 1 9 9 3 ) original clausal structure). The treatment of an aspect phrase as a type of AGRPg is consistent with our morpho- 1 9 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. syntactic analysis of the BA-constructions presented in Chapter 2: that is, the head of ASPP assigns Case to (or check the Case of) the object which is moved to the Spec of ASPP in the BA-construction (cf. Note 40 in Chapter 2).® The structure in (76) also captures the aspectual feature on the Chinese resultative verb compound as a selection property of AGR: that is, the resultative verb compound usually takes the perfective aspect marker -le, and it may sometimes take the progressive aspect marker -zhe or the experiential aspect marker -guo (cf. Section 3.3.1.). With the basic structure of the Chinese resultative verb compound construction established in (76), I would now proceed to offer a morpho-syntactic analysis of the eight types of resultative verb compounds introduced in Section 3.2., which is presented in the sections below. ® In fact, the syntactic structure proposed for the resultative verb compound construction in (76) is similar to and consistent with the syntactic structure proposed for the BA-constructions in Chapter 2 if the BA-phrase is treated as a type of A G R Pg. Treating the BA-phrase as a type of AGRPg makes perfect sense in our morpho-syntactic analysis, as the structural Case assigned to the subject of the BA-construction is regarded as an manifestation of the Spec-head agreement relation between the subject and the head of the BA-phrase (cf. Section 2.5 in Chapter 2). 197 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.5.2. The Agentive Verb Compound Construction As discussed in Section 3.2.1., the resultative verb compound of the agentive type consists of a transitive Vi and a stative, unaccusative or unergative V2. The subject of this type of verb compounds is the Agent of Vi and its object is both the Theme of Vi and the Experiencer of V2, as shown by (4), (5) and (6) above. Consider (4b), which is repeated below: (4b) Zhangsan yao—xing-le Lisi. Zhangsan shake-awake-ASP Lisi 'Zhangsan shook Lisi and as a result Lisi awoke.' Given the basic structure of the Chinese resultative verb compound construction in (76), (4b) would have the following structural representation in terms of thematic relation and X'-theory, assuming the VP-internal subject hypothesis (Koopman and Sportiche 1988; Kuroda 1988): (4b') [agrI Iaspp (aspI Ivpi Inpi Zhangsan] Zhangsan Ivi IvP2 ÏnP2 Iv'2 Iv2 xÎB9"le] ]]]]]] shake Lisi awake-ASP 198 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with this structural analysis, (4b) can be simply derived by verb-raising and NP-movement. V2 xing—le 'awake' is first raised to compound with Vi yao 'shake' by means of verb incorporation (Baker 1988).® The formation of such a verb compound follows the same morphological principle which applies to the combination of verbal morphemes in the lexicon; that is, the verbal morpheme with an aspect marker has to be at the right edge of the verb compound. Thus, the output of this verb compounding is yao—xing—le rather than the ill-formed *xing-le-yao. The resulting verb compound yao-xing-le is then raised to AGR via ASP to check its inflectional features (i.e. the aspectual and agreement features; cf. Chomsky 1993). NPi Zhangsan receives its 0-role from V'l and moves into the Spec of AGRP to check its Case feature against AGR, and NP2 Lisi Li (1990) argues that the Chinese resultative verb compound is unlikely the result of such an incorporation process because the first verbal morpheme in the compound does not 6-mark the maximal projection that is headed by the second verbal morpheme. Following his argument, Vl yao in (4b') could hardly be said to assign a 0-role to VP2, and VP2 would then be a barrier, making the raising of V2 xing-le impossible. However, under our morpho-syntactic analysis, the absence of the 0-marking relation between Vi and VP2 does not even arise, as the asymmetric semantic or implication relation between the activity denoted by the first verbal morpheme and the resultant state conveyed by the second verbal morpheme is "canonically structurally realized" as a complementation relation between Vi and VP2 (cf. Hale and Keyser 1993; also see Cheng's (1992) argument against Li's (1990) objection to such an incorporation approach to the Chinese resultative verb compound). 199 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. receives its 0-role from V'2 and moves into the Spec of ASPP to check its Case feature against ASP (cf. Chomsky 1993). The derivation of (4b) from (4b') is illustrated below: (4b") [^GRP ZhangsaUi t Iyao-(xing-le] Zhangsan shake-awake-ASP [asp Ivpi Inpi Ivi Ivi Ivpz Lisi I nP2 ^*1 Iv'2 Iv2 bj]]]]]]] The motivation and argument for the verb-raising and NP-movement in the derivation of (4b) above are presented as follows. First, the verb-raising from V2 into AGR via Vl and ASP is morphologically driven, as the inflectional features (especially the aspectual feature) reflected on V2 xing-le have to be checked against the features of ASP and AGR in order for the verb to enter the PF component under Spell-Out (cf. Chomsky 1993).^° This verb-raising Li (1993) uses the contrast between the following two sentences to argue against the V-to-I raising (i.e. V-to-ASP and V-to-AGR in our terms) in Chinese, assuming that adverbials like jingchang 'often' adjoins to VP: (i) Zhangsan jingchang zhui Lisi. Zhangsan often chase Lisi 'Zhangsan often chases Lisi.' (ii)*Zhangsan zhui jingchang Lisi. Zhangsan chase often Lisi 200 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is also licit under the Minimal Link Condition (Chomsky 1994), since each of its three steps (i.e. from V2 to Vi, from Vl to ASP, and from ASP to AGR) attaches the verb to the nearest head target that immediately c-commands the verb, without skipping an already-filled head position. However, there is an alternative way to account for the ungrammaticality of (ii), according to Chomsky (1994). Suppose that the adverbial jingchang occupies the Spec of VP. Then (ii) would have the following structure under our analysis: (ii') [agrp ^agr] Iaspp IaspI Ivpi Inpi Zhangsan] [y,i [yi] Zhangsan Ivp2 Ijingchang] Iv2 1*?% &isi]]]]]]] often chase Lisi In (ii'), VPi is a the maximal projection of the 'light verb' Vl, and VP2 is the maximal projection of the verb zhui ' chase '. To derive ( ii ) from (ii'), V2 zhui must be raised into AGR via Vi and ASP to check its inflectional features, NPi Zhangsan needs to move to the Spec of AGRP to check its Case, and NP2 Lisi has to move to the Spec of ASPP to check its Case. But the movement of NP2 Lisi to the Spec of ASPP is blocked by the intervening adverb jingchang in the Spec of VP2, thus making the derivation crash at PF. That is, the raising of V2 zhui to Vi forms the chain (zhuij, t^) with the minimal domain {Spec of VPi, Spec of VP2, NP2}, and the Spec of VPi and the Spec of VP2 are equidistant from NP2. However, NP2 Lisi cannot move to either of the two positions because they are both occupied by NPi Zhangsan and the adverbial jingchang. NP2 Lisi cannot move directly to the Spec of ASPP either even after the verb is raised from Vi to ASP. This is because this verb-raising would form a new chain with the minimal domain {Spec of ASP, Spec of VPi, VP2}, of which the Spec of VP2 is not a member. Hence, NP2 Lisi cannot cross the adverbial jingchang in the Spec of VP2 to reach the Spec of ASPP. (cf. Chomsky 1993). As for the grammaticality of ( i ), it is simply that the adverb jingchang is adjoined to AGRP (Chomsky 1994). 201 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Second, the movement of NPi Zhangsan to the Spec of AGRP and the movement of NP2 Lisi to the Spec of ASPP are forced by the Case Filter, and they are both legitimate under the Shortest Movement Condition (Chomsky 1993), due to the verb-raising. That is, as the verb compound yao- xing—le is raised from Vi to ASP to form the chain (yao— xing-lcjj, tj^) with the minimal domain {Spec of ASPP, Spec of VPi, VP2}, the Spec of ASPP and the Spec of VPi become equidistant from VP2 or any element it contains. Hence, NP2 Lisi, as a specifier of VP2, may move to the Spec of ASPP by crossing the Spec of VPi which is filled with NPi Zhangsan. When the verb compound yao—xing—le is further raised from ASP to AGR to form the new chain (yao—xing— lef, tj^) with the new minimal domain {Spec of AGRP, Spec of ASPP, VPl}, the Spec of AGRP and the Spec of ASPP are equidistant from VPi or any element it contains. Thus, NPi Zhangsan, as a specifier of VPi, may move to the Spec of AGRP by crossing the Spec of ASPP that is now occupied by NP2 Lisi. As a consequence of this analysis, the unavailable reading of (1), as in (61), is ruled out in a principled way. The structural representation of (61) is the same as that of (4b') under the above analysis. But in (61), the internal subject of VPi is moved to the Spec of ASPP, 202 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. while the internal subject of VP2 is moved to the Spec of AGRP, exactly opposite to the two NP-movements in (4b"): (61')*Iagrp ZhangsaUi [[zhui-[lei-le]t^spp Zhangsan chase-tired-ASP Z'isij, [asp [vpi [npi ^«1 [v'l (vi [yp2 Lisi [nP2 [v'2 [v2 tj]]]]]]] 'Lisi chased Zhangsan and Zhangsan got tired.' The movement of NPi Lisi and the movement of NP2 Zhangsan in (61') violate the Shortest Movement Condition, as the latter illicitly crosses the trace of the former. That is, as NPi Lisi moves to the Spec of ASPP, NP2 Zhangsan, which is an element of VP2, cannot move to the same Spec position which is now occupied by NPi Lisi. In order to check its Case, NP2 Zhangsan must move directly into the Spec of AGRP. However, this movement is impossible even after the verb compound is raised from ASP to AGR. This is because the verb-raising from ASP to AGR will form a new chain with the new minimal domain {Spec of AGRP, Spec of ASPP, VPl}, of which the Spec of VPi is not a member. Hence, NP2 Zhangsan cannot cross the NPi-trace t^ in the Spec of VPl to reach the Spec of AGRP. In other words, the movement of NPl Lisi to the Spec of ASPP blocks any 203 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. kind of Case assignment to NP2 Zhangsan, thus making the reading in (61) unavailable. 3.5.3. The Agentive—Stative Verb Compound Construction The morpho-syntactic analysis presented above also accounts for the agentive-stative type of resultative verb compounds, which is made up of a transitive Vi and an unaccusative or stative V2 and has a subject (both the Agent of Vi and the Experiencer of V2) and an object (the Theme of Vi), as in (7), (8), (9) and (10). In addition, the object of such resultative verb compounds has to be nonreferential, as displayed by the grammatical contrast between (46ab), (47ab) and (48ab). Consider (7), which is rewritten below: (7) Lisi he-zui—le jiu. Lisi drink-drunk-ASP wine 'Lisi drank wine and as a result he got drunk.' Given the basic structure of the Chinese resultative verb compound construction in (76), we will expect (7) to have the structural representation below, assuming Larson's (1988) analysis of double objects: 204 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (7 ) (agrp Iagr 1 (aspp Iasp 1 Ivpi Inpi [y,i Lisi [via [vi [mP2 [r2 [vP2 [nP3 [y'2 drink wine [v2 2ui-le]1]1]]1 drunk-ASP Like the derivation of (4b) in the above section, V2 tul le ' drunk ' in ( 7 ' ) is first raised to compound with Vi he 'drink', and NPi Lisi, which receives a 0-role from V'l, moves to the Spec of AGRP to check its Case against AGR. But unlike the derivation of (4b), the nonreferential N2 jiu 'wine' in (?') is moved to adjoin to Vi by means of incorporation (Baker 1988; Chomsky 1994; Hale and Keyser 1993). The resulting verb-noun compound he—zui—le—jiu is further raised into AGR via ASP to check its inflectional features. Besides, NP3 Pro in (7') is properly controlled by NPi Lisi or its trace which is the closest nominal C- commanding Pro (Huang 1992). Thus, (7) is simply derived by verb-raising, NP-movement and noun-verb incorporation: ( ) [agrp ^^sii Iagr [ [he-[zui—le] j—jiu;^] ilml [aspp Lisi drink-drunk-ASP-wine [asp [vPl [npI Ivi [via [vi ^lî [nP2 [n2 t*]]] [yp2 [nP3 [y'2 [v2 ^jllllîll 205 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The motivation and arguments for the verb-raising to AGR from V2 and for the movement of NPi Lisi to the Spec of AGRP are the same as the ones presented in the above section. But for the incorporation of N2 jiu to Vi, it is forced by the Case Filter and conforms to the Minimal Link Condition as it attaches the noun jiu to the nearest target Vl that immediately c-commands N2 without skipping an already-filled head position. 3.5.4. The Agentive—Transitive Verb Compound Construction The morpho-syntactic analysis proposed above further accounts for the agentive-transitive type of resultative verb compounds, of which Vi and V2 are transitive verbs and share a subject (the Agent of Vi and the Experiencer of V2) and an object (the Theme of Vi and V2), as shown by (11), (12), (13) and (14). Consider (11), which is rewritten below and whose structural representation would be in the form of (11') under the basic structure of the Chinese resultative verb compound construction in (76): (11) ta xue—hui—le fayu. he study-understand-ASP French 'He studied French and as a result he learned it.' 206 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (11') [agrp [agr] [aspp [asp! [vpi [npi [v'l [vx 3cue] he learn [vP2 [nP2 1*^®] [v'2 [v2 hui—le] [mP3 ^uyu] ] j j ] j ] understand-ASP French Like the derivation of (4b) in Section 3.5.2., V2 hui—le 'understand' in (11') is raised to compound with Vi xue 'study', and the resulting compound xue-hui—le is further raised to AGR via ASP to check its inflectional features. NPi ta 'he', which receives a 0-role from V'l, moves into the Spec of AGRP to check its Case against AGR. NP3 fayu 'French', which receives a 8-role from V2, moves into the Spec of ASPP to check its Case against ASP. And NP2 Pro, which is the specifier of VP2, is properly controlled by NPi ta or its trace that is the closest NP C-commanding Pro. Thus, the derivation of (11) also falls under verb- raising and NP-movement, as illustrated below: (11") [a g r p ta^ [^GR [[*ue-[hui-le]j]^]i] [ ^ p p fayu. he learn-understand-ASP French [asp [ypi [npi [v'l [vi [yp2 [np2 Pf®] [v'2 [v2 ] [nP3 t.]]]]]]] The motivation and arguments for the verb-raising to AGR from V2 and for the movement of NPi ta to the Spec of 207 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. AGRP are the same as the ones presented in Section 3-5.2. As for the movement of NP3 fayu into the Spec of ASPP, it meets with the Shortest Movement Condition, even with NP2 Pro intervened in between. That is, as the verb compound xue-hui—le is raised from Vi into ASP to form the chain (xue-hui-le%, tj^) with the minimal domain {Spec of ASPP, Spec of VPl, VP2}, the Spec of ASPP and the Spec of VPi are equidistant from VP2 or anything it contains. Thus, NP3 fayu, as a complement or V2, may move to the Spec of ASPP by crossing the Spec of VPi which is filled with NPi ta or its trace. Why doesn't NP2 Pro, which is a specifier of VP2 and is closer to the Spec of ASPP than NP3 fayu, block the movement of NP3 as a lexical element or an NP-trace does (see Section 3.5.2 and Note 13 above)? The answer to the question, I believe, lies in the nonvisibility of Pro and the visibility of a lexical element and a NP-trace at PF. According to Aoun (1980) and Chomsky (1981), an element is invisible at PF if it has no phonological features or Case. A lexical element is certainly visible at PF since it always has phonological features (and perhaps Case). An NP-trace is also visible at PF as it acquires its Case feature from the moved NP. According to Chomsky (1986a), the chain formed by a moved NP and its trace is headed by a Case-marked position (the position occupied by the NP) 208 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and terminates in a 0-position (the position occupied by its trace). Case is transferred from the head position to the terminal position of the chain, thus making the NP -trace visible for 8-marking. Then, the 6-role received by the NP-trace is transferred back to the head position of the chain which is occupied by the NP. As for Pro, it has neither phonological features nor Case, so it must be invisible at PF.^^ Hence, NP2 Pro, which is unlike the lexical element or NP-trace, will not block the movement of NP3 into the Spec of ASPP, assuming that the linkage between the (non)visibility and the blocking effect (i.e. minimality effect (Rizzi 1990)) is correct. One consequence of the analysis is that it correctly rules out the ungrammatical verb compound constructions in (55) and (62b), whose structural representations would have the same form as that of (11') under (76). Consider (55), whose derivation is shown below: Although Chomsky and Lasnik (1991) assume that PRO has a "null Case", such a null Case differs from the familiar ones like nominative, accusative, genitive, etc. in the sense that it will not make PRO visible at PF, as independently evidenced by the case of wanna-contraction shown below: (i) a- They want [gp (jp PRO to visit New York]], b. They wanna visit New York. In (ia), PRO, which intervenes between want and to, does not block the wanna-contraction, as shown by (ib). Thus, PRO is invisible to a PF rule like the wanna-contraction, even if it may bear a null Case. 209 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ( 5 5 ' ) * [ A G R P z h e j u h u a ^ [ I t i n g - [ d o n g - l e ] j ] „ ] i ] I a s p p the CL speech listen-understand-ASP EaSP IvPl InPI *"#] I V'l IVI ^kl IvP2 I Inp2 PfO] ïv'2 Iv2 ^jl ÎnP3 t;]]]]]]] It is the movement of NP3 zhe ju hua 'the speech' to the Spec of AGRP in (55') that violates the Shortest Movement Condition, as it illegitimately crosses the trace left by the movement of NPi wo 'I' to the Spec of ASPP. That is, though NP3 zhe ju hua can cross the invisible Pro in the Spec of VP2, it cannot move to the Spec of ASPP, as that position is occupied by NPl wo. In order to check its Case, NP3 zhe ju hua must move directly into the Spec of AGRP. But that NP-movement is impossible even after the verb compound is raised to AGR: namely, this verb-raising will form a new chain with the new minimal domain {Spec of AGRP, Spec of ASPP, VPl}, of which the Spec of VPi is not a member; thus, NP3 zhe ju hua cannot cross the NPi- trace t^ in the Spec of VPi to reach the Spec of AGRP. In other words, the movement of NPi wo into the Spec of ASPP would block the Case assignment to NP3 zhe ju hua. Another consequence of this analysis is that it also rules out the ungrammatical verb compound construction in (53) (= (62c)) and the unavailable reading of (54). Take 210 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (53) for example, which is repeated below and which would have the structural representation in (53') in accordance with the basic structure of the Chinese resultative verb compound construction in (76): (53) *Lisi xue-shang-le wushu. Lisi learn-injure-ASP martial-art 'Lisi learned martial art and it injured him.' (53')*[^qj{P [aspp (aSpI Ivpl IkPI [v'l Ivl *'^®1 Lisi learn [vP2 [ifP2 wushu] [y»2 [v2 shang—le] &isi] ]]]]]] martial-art injure-ASP Lisi The structure in (53') is ill-formed because NP3 Lisi can not be interpreted as disjoint in reference from NPi Lisi which c-commands NP3, thus violating Principle C of the binding theory (cf. Chomsky 1982 & 1993) This morpho-syntactic analysis further rules out the ungrammatical resultative verb compound construction in (62d) (which is repeated below), because its structural Even if we replace NPi Lisi and NP3 Lisi in (53) with the corresponding pronoun ta 'he', the verb compound construction and its structural representation in (53') are still ill-formed because NP3 ta cannot be interpreted as disjoint referentially from NPi ta that c-commands NP3 in the relevant minimal domain {Spec of VPi, Spec of VP2, NP2} established by the verb-compounding, thus violating Principle B of the binding theory (cf. Chomsky 1993). 211 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. representation would be the same as that of (53') and its derivation would violate the Shortest Movement Condition in the same manner as the derivation of (55) (cf. (55') above). Hence, (62d) is even worse than (55) in terms of unacceptability; (62d)*wushu xue—shang—le Lisi. martial-art learn-injure-ASP Lisi 'Lisi learned martial art and it injured him.' 3.5.5. The Unergative—transitive Verb Compound Construction As discussed in Section 3.2.4., the resultative verb compound of the unergative-transitive type consists of an unergative Vi and a transitive V2, and it takes a subject (which is both the Agent of Vi and the Experiencer of V2) and an object (which is the Theme of V2), as displayed by (15), (16), (17) and (18) above. Consider (16), which is rewritten below: (16) wo pao-diu-le yi zhi xie. I run-lose-ASP one CL shoe 'I ran and as a result I lost a shoe.' 212 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Following the basic structure of the Chinese resultative verb compound construction in (76), we would expect (16) to be structurally represented in the following way: (16') [agrp [agrI [aspp [aspJ [vpi [rpi [v'l [yi I run [yp2 [rp2 [v'2 [v2 [np3 *hi xie]]]]]]] lose-ASP one CL shoe (16') is on a par with the structural representation of (11) given in the above section. Thus, the derivation of (16) from (16') would be parallel to the derivation of (11) from (11'), and the motivation and arguments for the derivation of (16) would also be similar to those for the derivation of (11) (cf. Section 3.5.4.): (16") I AGRP wOi [AGR[[P®0“ [<li“’l®ljlkliHASpp zhi xie, I run-lose-ASP one CL shoe [asp [ypi [npi [v'l [vi ^kl [yp2 [nP2 PZO] [v'2 [v2 [nP3 t*]]]]]]] As a consequence of this analysis, the ungrammatical resultative verb compound constructions in (58) and (63b) are ruled out as the violations of the Shortest Movement Condition, as displayed by the derivation of (63b) below: 213 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (63b')*[^jjiip yi zhi xie^ [^ C R [[pao-[diu-le]j] [^pp one CL shoe run-lose"ASP I ASP I VPl [ffPl I V'l tvi ^kl IvP2 I InP2 Iv '2 Ïv2 InP3 ^il1111 1 1 That is, the movement of NP3 yi zhi xie 'a shoe' to the Spec of AGRP would illicitly cross the NPi-trace left in the Spec of VPi by the movement of NPi wo 'I' to the Spec of ASPP (cf. the analysis of (55) in Section 3.5.4). Another consequence of this analysis is that it also excludes the unavailable readings of the resultative verb compound constructions in (63c) and (63d), because their structural representations would violate Principle C of the binding theory, as shown below: (63c) Lisi wan-wang—le wo. Lisi play-forget-ASP I *'Lisi played so much that I forgot him.' (63c') *[^Qjjp IrgrH aspp ^Aspl îvpi Ifipi Lisi] wan] Lisi play IvP 2 Ihp2 Iv'2 Iv2 wang— le] (jjp3 l*isi] ]]]]]] I forget-ASP Lisi 214 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (63d) Lisi wan—wang-le wo. Lisi play-forget-ASP I *'I played so much that Lisi forgot me.' (63d') * [agrH aspp [asp! Ivpi Ihpi Iv'i Ivi wan) I play Iyp2 InP2 Iv'2 tv2 wang—le] [up3 wo]]]]]]] Lisi forget-ASP I In (63c') NP3 Lisi cannot be interpreted as referentially disjoint from NPi Lisi, because the latter c-commands the former. For the same reason, NP3 wo In (63d') cannot be interpreted as referentially disjoint from NPl wo (cf. the analysis of (53) and (62d) in Section 3.5.4.). 3.5.6. The Unergative-Stative Verb Compound Construction and the Unergative Verb Compound Construction Both the unergative-stative type and the unergative type of resultative verb compounds are no exceptions to the morpho-syntactic analysis presented above. Consider the unergative-stative type of resultative verb compounds first, which consists of an unergative Vi and a stative or unaccusative V2 and take a subject (the Agent of Vi) and an object (the Experiencer of V2), as in (19), (20), 215 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (21), (22) and (23) above. Take (20) for example, whose structural representation would be in the form of (20') under the basic structure of the Chinese resultative verb compound construction in (76): (20) ta ku-shi-le liang kuai shoujuan. He cry-wet-ASP two CL handkerchief 'He cried so much that two handkerchiefs got wet. ' (20') IagrH aspp I asp 1 Ivpi Inpi Ivi Ivi ^“1 Ivp2 he cry Ixpz liang kuai shoujuan] [y,; [yg shi-le]]]]]]] two CL handkerchief wet-ASP (20') is parallel to the structural representation given in (4b'). Hence, the derivation of (20) from (20') would be analogous to the derivation of (4b) from (4'b) and the motivation and arguments for the derivation of (20) would also be similar to those for the derivation of (4b) (cf. Section 3.5.2.): (20") [^grp I AGR I [ku-[shi—le] j]^]^] Ij^pp he cry-wet-ASP liang kuai shoujuan. [^p t^] [^pj [mpi t^] two CL handkerchief Iv'l Ivi IvP2 InP2 ^*1 IV'2 Iv2 ^jl 111111 216 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As a consequence of this analysis, the unavailable reading of (64b), which is repeated below, is ruled out as a violation of the Shortest Movement Condition: (64b) Lisi ku-xing-le wo. Lisi cry-awake-ASP I *'I cried so loudly that Lisi woke up.' (64b')*[^gjip Lisii I[ku-[xing-le]j]Ji] [^spp wo, Lisi cry-awake-ASP I Iasp Ivpi Inpi Ivi Ivi Ivpz Inp2 Iv'2 Iv2 1111111 That is, the movement of NP2 Lisi to the Spec of AGRP in (64b') would illicitly cross the NPi-trace t, left in the Spec of VPl by the movement of NPl wo 'I' to the Spec of ASPP (cf. the analysis of (61) in Section 3.5.2.). Now, consider the resultative verb compounds of the unergative type, which consist of an unergative Vi and a stative or unaccusative V2 and whose subject is both the Agent of Vl and the Experiencer of V2, as in (24), (25), (26), (27) and (28) above. The derivation of this type of resultative verb compound construction also becomes the side effect of verb-raising and NP-movement under the morpho-syntactic analysis proposed above, as exemplified by the derivation of (24) below: 217 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (24) Lisi xiao—feng—le. Lisi laugh-insane-ASP 'Lisi laughed to the extent of becoming mad.' (24') [xgrp [agrI (aspp ^Aspl Ivpi Inpi ^^^i] [yi Lisi Ivi xi&o] Iyp2 Inp2 Iv'2 Iv2 feng— le ]] ]]]] ] laugh insane-ASP (24") [j^GRP ^isi^ I;^ gr I [xia®"[ feng—le] j ] il Iaspp Lisi laugh-insane-ASP I ASP I VPl I NPl ^il I V'l IVI IvP2 InP2 P^^®1 Iv'2 Iv2 tj]]]]]]] The motivation and arguments for the verb-raising and NP- movement in the derivation of (24) are almost the same as those presented in the above sections. However, there is a difference between the unergative type of resultative verb compound construction and those we dealt with above: that is, the former contains one overt NP but the latter contains two. Since the only overt NP in the unergative verb compound construction is the logical subject of both Vl and V2, its ASP (AGRo) (cf. Section 3.5.1) will become inert, failing to assign Case or check the Case of an NP moved into its Spec position according to Chomsky (1993). 218 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This is why NPl Lisi in (24") moves directly to the Spec of AGRP to check its Case against AGR. This NP-movement meets with the Shortest Movement Condition as the Spec of AGRP and the Spec of ASPP are equidistant from NPl Lisi, due to the verb-raising from ASP to AGR; i) the raising of the verb compound xiao—feng—le from ASP to AGR forms a chain (xiao-feng-lcj^, t^) with the minimal domain {Spec of AGRP, Spec of ASPP, VPl}, so the Spec of AGRP and the Spec of ASPP are equidistant from VPl or any element it contains; and ii) hence, NPi Lisi, as a specifier of VPi, may move to the Spec of AGRP by crossing the Spec of ASP. 3.5.7. The Unaccusative Verb Compound Construction The morpho-syntactic analysis proposed above further extends itself to the unaccusative type of resultative verb compounds. Like the unergative type of resultative verb compounds, the unaccusative type of resultative verb compounds takes one overt NP. But unlike the unergative type of verb compounds, the only overt NP carried by the unaccusative type of verb compounds is either the logical object of Vl and the Experiencer of V2, as in (29), (30), (31) and (32), or the Experiencer of V2 only, as in (33), (34) and (35). Consider (29) first, which is rewritten 219 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. below and which would be expected to have the structural representation in (29') under the basic structure of the Chinese resultative verb compound construction in (76): (29) na ke shu kan-dao— Is. That CL tree cut-fall-ASP 'That tree was cut and as a result it fell.' (29') [agr] Iaspp IaspI Ivpi Inpi Ivi Ivi cat IvP 2 Inp2 shu] Iv'2 Iv2 dao— le ]] ]]]] ] that CL tree fall-ASP Since the overt NP na ke shu 'that tree' in (29') is the logical object of Vl and the Experiencer of V2, its AGRP will be missing (Chomsky 1993), rendering (29') to (29"): (29") [j^spp IaspI I VPl I NPl Pfo] I V'l Ivi han] [ypj cut Inp2 he shu] [y.j Iv2 le ]]]]]] that CL tree fall-ASP The derivation of (29) from (29"), as displayed by (29"') below, falls under the same type of verb-raising and NP- movement shown above, and their motivation and arguments are also the same as those given in the above sections: 220 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (29"') [aspp na ke shu^ [^p [kan-[dao-le]j]J [^pi that CL tree cut-fall-ASP Inpi [y,j^ [yi [ypj [^ps ^il Ev'2 IV2 tj]]]]]] The same morpho-syntactic analysis also accounts for the second subclass of unaccusative verb compounds whose only overt NP argument is the Experiencer of V2 and does not have any thematic relation with Vl, as illustrated by the derivation of (33) below: (33) duzi xiao—tong-le belly laugh-ache-ASP 'Someone laughed so much that his belly ached.' (33”) [aspp duzi^ [asp [%i&0"[tong—le] [yp^ Inpi belly laugh-ache-ASP Iv'l Ivi ^kl IvP2 InP2 Iv'2 Iv2 j 1 1 1 1 1 1 As a consequence of this morpho-syntactic analysis, the distinction between the unaccusative verb compound construction and unergative verb compound construction is now reduced to a trivial question of morphology — a very welcome result as we expect. In other words, the terms like unergative and unaccusative will have no substantive meaning in grammar any longer, as they can now be totally 221 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. determined by the choice of active versus inert/missing AGR or ASP (cf. Chomsky 1993). 3.5.8. Zhe Causative Verb Compound Construction Finally, the proposed morpho-syntactic analysis can offer an account for the resultative verb compounds which are of the causative type. According to the discussion in Section 3.2.8., one class of this type of resultative verb compounds consists of an unergative Vi and a stative or unaccusative V2. Their object is the Experiencer of both Vl and V2, whereas their subject does not have any thematic relation with either Vi or V2. Moreover, there is a Causer-Causee relation between the subject and the object: that is, the former causes something to happen to the latter, as in (36), (37) and (38). Consider (36), which is repeated below: (36) na ping jiu zui-dao—le Lisi. that CL wine drunk-fall-ASP Lisi 'That bottle of wine made Lisi drunk and fall.' Since there is no semantic relation between the subject na ping jiu 'that bottle of wine' and Vi zui 'drunk' or 222 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V2 dao 'fall', the causative relation between na ping jiu and the object Lisi must be established independently of the complex events denoted by Vi and V2. In other words, the subject na ping jiu must be treated as an additional argument added to the predicate of the resultative verb compound construction by causation (Bowers 1988; Huang 1992; Sijbesma 1992). Still assuming Hoekstra's (1990) analysis that the change of an unaccusative verb into the corresponding causative one involves the incorporation of the unaccusative verb to a higher abstract causative verb (CAUS), the Spec of which is occupied by the Causer (cf. Section 2.5.7. in Chapter 2), the subject and Causer na ping jiu in (36) would become the specifier of a higher CAUSP-shell (cf. Larson 1988). Following this analysis, the causative verb compound construction in (36) would be expected to have the structural representation in (36i) in accordance with the basic structure of the Chinese resultative verb compound construction in (76); (36i) IagrI Iaspp IaspI Icausp Inpi Piog jiu] that CL wine I CAUS' IcAUsI I VPl InP2 I*isi] (yi Ivi zui] [yp2 Lisi drunk Inp3 Iv'2 Iv2 dao—le]]]]]]]]] fall-ASP 223 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Given this structural representation, the derivation of (36) would fall under the same type of verb-raising and NP-movement that derive those resultative verb compound constructions above. That is, V2 dao—le 'fall' is first raised to compound with Vi zui ' drunk ', and the resulting verb compound zui—dao—le is further raised to AGR through CAUS and ASP to check its inflectional features (namely, causative, aspectual and agreement features). NPl na ping j iu ' that wine ', which gets a causative role from CAUS ', moves to the Spec of AGRP to check its Case. NP2 Lisi, which receives a 0-role from V'l, moves to the Spec of ASPP to check its Case. As for NP3 Pro, it is properly controlled by NP2 Lisi or its trace: (36i') na ping jiu^ [ [ [zui-[dao-le] that CL wine drunk-fall-ASP (aspp (asp (cAUSP [«PI (cAUS' (cAUS Lisi Ivpi Imp2 Ivi Ivi Ivp2 Inp2 Iv2 Iv2 ty]]]]]]]]] The motivation and argument for the verb-raising in (36i) are the same as the ones presented in the above sections. As for NPl-movement to the Spec of AGRP and NP2-movement into the Spec of ASPP, they are not only permissible but 224 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. also obligatory under the Shortest Movement Condition: i) when the verb compound zui-dao-le is raised from CAUS to ASP to form the chain ( zui-dao-le, t^) with the minimal domain {Spec of ASPP, Spec of CAUSP, VPi}, the Spec of ASPP and the Spec of CAUSP become equidistant from VPi or any element it contains. Hence, NP2 Lisi, as a specifier of VPl, may move to the Spec of ASP by crossing the Spec of CAUSP which is filled by NPl na ping jiu. As the verb compound zui—dao—le is further raised from ASP to AGR to form the new chain (zui-dao—le^, t^) with the new minimal domain {Spec of AGRP, Spec of ASPP, CAUSP}, the Spec of AGRP and the Spec of ASPP become equidistant from CAUSP or any element it contains. Thus, NPl na ping jiu, as a specifier of CAUSP, may then move to the Spec of AGRP by crossing the Spec of ASPP which is now occupied by NP2 Lisi. One consequence of this analysis is that the ill- formed sentence in (77) below, which is exactly the same as (36) except that the subject and object are exchanged with each other in position, is excluded as a violation of the Shortest Movement Condition: (77) *Lisi zui-dao-le na ping jiu. Lisi drunk-fall-ASP that CL wine 225 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. <77')*Iagrp U gr [[[zui-[dao-le]j]k]i]J Lisi drunk-fall-ASP Iaspp pzng I^sp t^l Icausp Inpi Icaus' that CL wine I CAUS IVPl InP2 IV'l Ivi ^kl IvP2 Inp2 Iv'2 Iv2 tjllllllll] That is, as NPi na ping jiu moves to the Spec of ASPP and leaves its trace in the Spec of CAUSP, NP2 Lisi then can not move to the same Spec position now occupied by NPi na ping jiu. In order to check its Case, NP2 Lisi must move directly to the Spec of AGRP. However, this movement is not possible even after the verb compound is raised from ASP to AGR. This is because the verb-raising from ASP to AGR will form a new chain with the minimal domain {Spec of AGRP, Spec of ASPP, CAUSP}, of which the Spec of CAUSP is not a member. Hence, NP2 Lisi cannot cross the NPi- trace t^ in the Spec of CAUSP to reach the Spec of AGRP. In other words, the movement of NPi na ping jiu into the Spec of ASPP would block any kind of Case assignment to NP2 Lisi, thus making the sentence in (77) ungrammatical. The same morpho-syntactic analysis also applies to the second subclass of causative verb compounds which is made up of a transitive Vi and an unaccusative or stative V2. The object of this class of causative verb compounds 226 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is both the Agent of Vi and the Experiencer of V2 and the subject is the Theme of Vi. There also exists a Causer- Causee relation between the subject and the object, as in (39), (40) and (41). Consider (39), which is rewritten below and whose structural representation would be in the form of (39') under the basic structure of the Chinese resultative verb compound construction in (76) and the assumption of CAUSP: (39) na ping jiu he—zui—le Lisi. that CL wine drink-drunk-ASP Lisi 'Lisi drank that bottle of wine and as a result he got drunk.' (39') [agrI Iaspp IaspI Icausp IcausI Ivpi Inpi Lisi Iv'i I via IVI Inp2 pi^g jiu]] [yp2 Iup3 P^^o] drink that CL wine I V'2 Iv2 ZUi-le]]]]]]]] drunk-ASP Like the derivation of (36) above, V2 zui—le 'drunk' in (39') is raised to compound with Vi he 'drink', and the resulting verb compound he—zui—le is further raised to AGR via CAUS and ASP to check its inflectional features. But unlike the derivation of (36), NPi Lisi in (39') is 227 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. moved to the Spec of ASPP to check its Case, and NP2 na ping jiu 'that bottle of wine' first moves to the Spec of CAUSP to check its Causer's feature and then moves to the Spec of AGRP to check its Case. And NP3 Pro is properly controlled by NPl Lisi or its trace which is the closest nominal c-commanding Pro; (39") Iagrp “ ping jiUi [agr I[[he-[zui-le]j]k]i]J that CL wine drink-drunk-ASP (aspp i*i®in Iasp ^*1 ^causp ^caus Ivpi Lisi ImPI Iv'l Iv'la tvi IuP2 1 ïvP2 ÏmP3 P^o J I V'2 Iv2 '^j 11111111 The motivation and argument for the verb-raising in (39”) are the same as the ones presented in the above sections, and NPl-movement to the Spec of ASPP and NP2-movement to the Spec of AGRP are also legitimate and obligatory under the Shortest Movement Condition. That is, when the verb compound he-zui-le is raised from Vi to CAUS to form the chain (he-zui—le%, tj^) with the minimal domain {Spec of CAUSP, Spec of VPl, NP2}, the Spec of CAUSP and the Spec of VPl are equidistant from NP2. Thus, NP2 na ping jiu can move to the Spec of CAUSP by crossing the Spec of VPi which is filled with NPi Lisi. As the verb compound he— 228 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. zui-le is further raised from CAUS to ASP to form the new chain (he-zui-lej^, t^) with the new minimal domain {Spec of ASPP, Spec of CAUSP, VPi}, the Spec of ASPP and the Spec of CAUSP become equidistant from VPi or anything it contains. Thus, NPi Lisi, as the specifier of VPi, may move into the Spec of ASPP by crossing the Spec of CAUSP which is now occupied by NP2 na ping jiu. When the verb compound he-zui—le is ultimately moved from ASP to AGR to form another new chain (he-zui—le*, t^) with the minimal domain {Spec of AGRP, Spec of ASPP, CAUSP}, the Spec of AGRP and the Spec of ASPP are equidistant from CAUSP or any element it contains. Hence, NP2 na ping jiu, as the new specifier of CAUSP, may move to the Spec of AGRP by crossing the Spec of ASPP which is occupied by NPi Lisi. There is an independent argument for the derivation of (39) above, which involves anaphoric binding (cf. Wu 1993 & 1994). According to Wu (1994), the causative verb compound construction in (78) is not ambiguous; (78) ziji-de mama zhui—lei—le Lisi. self's mother chase-tired-ASP Lisi 'Chasing his own mother made Lisi tired.' That is, the Causer and subject ziji-de mama 'oneself's mother' can only be interpreted as the theme of Vi, and 229 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the anaphor ziji-de, which is embedded within the Causer and subject, can only take the Causes and object Lisi as its antecedent. The unambiguity of (78) and the binding relation between Lisi and ziji-de are naturally captured by the morpho-syntactic analysis that derives (39) above, without stipulation. Given the analysis of (39) above, the structural representation of (78) would be parallel to that of (39): (78') [^GUP [j^GR ] [rspp [rsP 1 [ CAUSP I CAUS 1 I VPI lupi Lisi] [v'l Iv'ia Ivi zhui] Inp2 ziji-d e m am a]] Lisi chase self's mother lyp2 În p3 P zo] Iv» 2 Iv2 i® i-le ] ] ] ] ] ] ] ] tired-ASP In (78'), NP2 ziji-d e m am a is only the Theme of Vi zhui, and the anaphor ziji-d e, which is embedded with NP2, can only be interpreted as being coreferential with NPi L isi because L isi c-commands ziji-d e in the local domain VPi, thus satisfying Principle A of the binding theory before the derivation starts (Chomsky 1993).The derivation This, as a matter of fact, supports Belletti and Rizzi's (1988) suggestion that Principle A of the binding theory is an anywhere principle and can be satisfied at any stage of derivation to guarantee the well-formedness of its output. 230 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of (78) from (78'), which is displayed below, also falls under the verb-raising and NP-movement that derive (39), and their motivation and arguments are also the same as those presented for the derivation of (39): (78") ziji-de mama^ [ ( [zhui-[lei-le] self's mother chase-tired-ASP IaSPP IaSP [CAUSP IcAUS IvPl Lisi I NPI I V'l Iv'la IVI ÏnP2 1 lvP2 InP3 Iv'2 Iv2 1111111] 3.6. THE CONSEQUENCES OF THE MORPHO-SYNTACTIC ANALYSIS OF RESULTATIVE VERB COMPOUNDS IN CHINESE 3.6.1. The Referentiality Effects (1) As discussed in Section 3.3.2., the object of the agentive-stative verb compound is the Theme of Vi and is required to be nonreferential, otherwise the resultative verb compound construction involved would be ill-formed, as displayed by the grammatical contrast between (46ab), (47ab) and (48ab) above. Take (46ab) for example, which are repeated below: 231 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (46) a. Lisi he-zui-lc jiu. (= (7)) Lisi drink-drunk-ASP wine 'Lisi drank wine and as a result he got drunk.' b.*Lisi he-zui—le (na) yi ping jiu. Lisi drink-drunk-ASP (that) one CL wine Opposite to the object of the agentive-stative type of verb compounds, the subject for the second subclass of causative verb compounds is required to be referential even though it is also the Theme of Vi. Otherwise, the causative verb compound construction involved would be ill-formed too, as displayed by the grammatical contrast between (49ab), (SOab) and (Slab) (cf. the discussion in Section 3.2.8.). Take (49ab) for example, which are also repeated below: (49) a. na ping jiu he-zui—le Lisi. (= (39)) that CL wine drink-drunk-ASP Lisi 'Lisi drank that bottle of wine and as a result he got drunk.' b.*jiu he-zui-le Lisi. wine drink-drunk-ASP Lisi If we carefully compare and examine the four sentences in (46ab) and (49ab), we will find that (46a) and (49b) are 232 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. exactly the same except that their subject and object are exchanged with each other. So are (46b) and (49a). If we group (46a) and (49b) together and assign (46b) and (49a) into another group, we will end up with the following two pairs in (79) and (80): (79) a. Lisi he—zui—le jiu. (= (46a) or (7)) Lisi drink-drunk-ASP wine 'Lisi drank wine and as a result he got drunk. ' b.*jiu he-zui-le Lisi. (= (49b)) wine drink-drunk-ASP Lisi (80) a. na ping jiu he-zui-le Lisi. (= (49a) that CL wine drink-drunk-ASP Lisi or (39)) 'Lisi drank that bottle of wine and as a result he got drunk.' b.*Lisi he-zui-le (na) yi ping jiu. (= (46b) Lisi drink-drunk-ASP that one CL wine Now the questions raised by (79ab) and (80ab) are: i) why can the nonreferential jiu 'wine' be the object in (79a) but not the subject in (79b)? ii) why can the referential (na) yi ping jiu 'that/one bottle of wine' be the subject in (80a) but not the object in (80b)? The two questions, as a matter of fact, could find a natural and principled answer from the morpho-syntactic analysis proposed above. 233 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Put it in another way, both the graimaticality of (79a) and (80a) and the ungrammaticality of (79b) and (80b) can be captured by the proposed morpho-syntactic analysis as side effects of verb—raising and NP-movement without any stipulation. Since (79a) (= (7)) and (80a) (= (39)) have been explored and explained in Section 3.5.3. and Section 3.5.8. respectively, we only concern ourselves with (79b) and (80b) here. First, consider (79b), which would have the same structural representation as that of (79a) under the morpho-syntactic analysis above (cf. the analysis of (7) in Section 3.5.3.): (79b') [agrp (agrI Caspp IaspI Ivpi Inpi [y'l (via Lisi Ivi InP2 In2 Ivp2 InP3 Iv'2 drink wine [v2 zui-le]]]]]]] drunk-ASP As in the derivation of (7) above, V2 zui-le 'drunk' in (79b') is raised to compound with Vi he 'drink', the non referential N2 jiu 'wine' is moved to adjoin to Vi, and the resulting verb-noun compound he—zui—le-jiu is further raised to AGR via ASP. However, instead that NPl Lisi is moved into the Spec of AGRP, as in the derivation of (7), 234 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the nonreferential N2 jiu is moved out of the verb-noun compound he— zui— le — j iu to the Spec of AGRP, and NPi L isi is moved to the Spec of ASPP: (79b”)* [j^ Q U P jiu^ [^ G R [ [he— [zui— le] j— [^pp wine drink-drunk-ASP ^isiu I ASP Ivpi Inpi ^ * 1 Ivi I v i a Ivi Lisi InP2 In2 11 lvP2 InP3 Iv'2 Iv2 ^j 1 1 1 1 1 1 1 Although the movement of NPi L isi to the Spec of ASPP is legitimate under the Shortest Movement Condition, due to the verb-raising, the movement of N2 jiu out of the verb- noun compound he— zui— le — jiu into the Spec of AGRP would violate the Uniformity Condition on the links of a chain (Chomsky 1994): that is, jiu is a nonreferential N° and its movement to the Spec of AGRP would change its "phrase structure status" to an N"*^. This derivation will also be ruled out by morphology theory, as part of a compound word or single word cannot generally be moved out of the word to some other places. This is due to the fact that a trace can never be nonexhaustively dominated by an X° category a part of the old lexicalist hypothesis that still seems true (cf. Baker 1988). 235 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Even if (79b) may be derived in a different way, its alternative derivation is barred too: that is, instead of adjoining the nonreferential N2 jiu to Vi, NP2 jiu moves into the Spec of AGRP to check its Case against AGR, NPi Lisi moves to the Spec of ASFP to check its Case against ASP, and the verb compound he—zui—le is raised to AGR: <79b"')*EAGRP [agr I[he-[zui-le]j]k]i] [^spp Lisi. wine drink-drunk-ASP Lisi Iasp I VPI Inpi ^*1 Ivi I via Ivi ' • J c l Inp2 ^1]] IvP2 Ihp3 Iv '2 Iv2 tj]]]]]]] Although the movement of NPl Lisi to the Spec of ASPP is licit under the Shortest Movement Condition, the movement of NP2 jiu to the Spec of AGRP is not. That is, when NPl Lisi moves to the Spec of ASPP, NP2 jiu must move to the Spec of AGRP to check its Case. But the latter movement is impossible even after the verb compound is raised from ASP to AGR: i) the verb-raising from ASP to AGR forms a new chain with the minimal domain {Spec of AGRP, Spec of ASPP, VPi}, of which the Spec of VPi is not a member; and ii) hence, NP2 jiu cannot cross the NPi-trace t. in the Spec of VPi to reach the Spec of AGRP. As a result, the movement of NPi Lisi to the Spec of ASPP blocks any kind of Case assignment to NP2 jiu, making (79b) ill-formed. 236 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Now, consider (80b), whose structural representation would be the same as that of (80a) (= (39)) according to the morpho-syntactic analysis above (cf. the analysis of (39) in Section 3.5.8.): (80b') (x G R p IagrHaspp [aspI [C A U S P [cAusl Ivpi Inpi Lisi I V'l Iv'ia IVI he] [mp2 (na) yi ping jiu]] drink that one CL wine Inp3 Iv'2 Iv2 zni—ie] ]]]]]]] drunk-ASP Like the derivation of (39), V2 zui—le in (80b') raises to compound with Vi he, and the resulting verb compound he-zui-le is further raised to AGR through CAUS and ASP. But unlike the derivation of (39), NPl Lisi in (80b') is moved into the Spec of AGRP via the Spec of CAUSP, while NP2 (na) yi ping jiu moves directly to the Spec of ASPP: (80b")*I^cjiP Lisii I A G R [[[he-[zui-le]j]k]i].] Lisi drink-drunk-ASP I ASPP P^ng jiUjj [^p t^] [cA^sp Icaus ^i] that one CL wine I V P I I N P l hi] Iv'l Iv'ia Ivi hj(] [hp2 h^l 1 IvP 2 InP3 Iv'2 Iv2 hj]]]]]]]] 237 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It is the movement of NP2 (na) yi ping jiu to the Spec of ASPP that violates the Shortest Movement Condition, as it illicitly crosses the NP-trace t^ left in the Spec of VPi by the movement of NPi Lisi. That is, as NPi Lisi moves into the Spec of AGRP via the Spec of CAUSP, NP2 (na) yi ping jiu cannot move to the Spec of CAUSP occupied by the trace of NPl Lisi. In order to check its Case, NP2 (na) yi ping jiu must move directly to the Spec of ASPP. But this movement is impossible even after the verb compound is raised from CAUS to ASP; i) the verb-raising from CAUS to ASP forms a new chain with the minimal domain {Spec of ASPP, Spec of CAUS, VPI}, of which the Spec of VPl is not a member; and ii) thus, NP2 (na) yi ping jiu cannot cross the NPi-trace t^ in the Spec of VPi to reach the Spec of ASPP, since the latter is closer to the Spec of ASPP than the former. As a result, the Case checking of NP2 (na) yi ping jiu is blocked by the movement of NPi Lisi to the Spec of AGRP through the Spec of CAUSP, thus making (80b) ungrammatical. Therefore, both the nonreferential requirement upon the object of the agentive-stative verb compounds and the referential requirement on the subject of the second sub class of causative verb compounds, for which Li's lexical analysis fails to provide an account (cf. Section 3.4.3), are captured by the proposed morpho-syntactic analysis as 238 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. natural consequences of the Uniformity Condition on the links of a chain or of the Shortest Movement Condition on the targets of movement, without making any stipulation. 3.6.2. The Referentiality Effects (2) Besides affecting the grammaticality of the agentive -stative verb compound construction and the second class of the causative verb compound construction, the effects of referentiality further entangle the interpretation of several other resultative verb compound constructions, as discussed in Section 3.3.2. above. Consider (52), which is repeated below: (52) a. Lisi qi-lei-le ma. (ambiguous) Lisi ride-tired-ASP horse i.'Lisi rode a horse and as a result it got tired, ii.'Lisi did horse-riding and as a result he got tired.' b. Lisi qi-lei-le (na) yi pi ma. Lisi ride-tired-ASP (that) one CL horse 'Lisi rode that/one horse and as a result the horse got tired.' 239 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (52a), the object of the resultative verb compound qi- lei-le ' ride-tired ' is the nonreferential ma ' horse ', and the sentence has two different interpretations. One is the interpretation of an agentive verb compound, and the other is the interpretation of an agentive-stative verb compound. However, in (52b), the object of the same verb compound is the referential (na) yi pi ma 'the/one horse' and the sentence is unambiguous with the interpretation of an agentive verb compound only. The ambiguity of (52a) and the unambiguity of (52b), which are generated by the (non)referentiality of the NP object, could also be captured by the morpho-syntactic analysis proposed above. Consider (52a) first, which is assigned the following two structural representations by the proposed morpho-syntactic analysis to accommodate its two interpretations in (52ai) and (52aii) respectively (cf. the analysis of (4b) in Section 3.5.2. and also the analysis of (7) in Section 3.5.3.): (52ai') Iagr] Iaspp (aspI Ivpi Impi Iv'i Ivi Lisi ride IvP2 ImP2 Im2 J Iv'2 Iv2 11 1 11 ] horse tired-ASP 240 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (52aii') [^CRP (agr! Iaspp IaspI Ivpi Inpi [y,i Iv'i* Lisi Ivi 9^] InP2 Ih2 1 1 IvP2 In P 3 Iv'2 ride horse Iv2 lei-le]]]]]]]] tired-ASP Thus, the ambiguity of (52a) can be derived by the same type of verb-raising, verb-noun incorporation and NP- movement presented above. That is, in both (52ai') and (52aii'), V2 lei—le 'tired' is raised to form a compound with VI qi ' ride ' ; the nonreferential N2 ma ' horse ' is adjoined to Vi; the resulting verb-noun compound gi—lei— le-ma is further raised to AGR through ASP to check its inflectional features; NPl Lisi moves to the Spec of AGRP to check its Case against AGR; and NP3 Pro in (52aii) is properly controlled by NPi Lisi or its trace which is the closest nominal c-commanding Pro: ( 52ai" ) I AGRP I AGR I l î ^ — I ^ ® ^ ” ^ ® ] ll«I I ASPP Lisi ride-tired-ASP-horse I ASP I VPI I NPl I V'l IVI ^ l l IvP2 In P 2 In2 lv'2 Iv2 tj]]]]]]] 241 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (52aii") [j^ qrp Iagr I i1*1 Iaspp Lisi ride-tired-ASP-horse [asp ^«1 [ypi [npI tv'l [via [vi ^ll [nP2 [n2 t%]]] [yp2 [%fP3 [v'2 [v2 ^j 1 1 1 1 1 1 1 As for the unambiguous verb compound construction in (52b), it would have the same structural representation as that of (52ai) by the morpho-syntactic analysis above: (52b') [ a g r ] [a s p p [aspI [ypi [ r p i [ y i [ y ^ qi] Lisi ride [vP2 [n?2 (na) yi pi ma] [y.g [yg lei-le]]]]]]] (that) one CL horse tired-ASP However, the derivation of (52b) would differ from that of (52ai) above as its NP2 (na) yi pi ma 'that/one horse' is a referential NP, then an N“*^, whose incorporation to the verb compound would then change its "phrase structure status" to an N°, violating the Uniformity Condition on the links of a chain. So NP2 (na) yi pi ma in (52b') has to move to the Spec of ASPP to check its Case, as in The incorporation of (na) yi pi ma into the verb compound will also be ruled out by the morphology theory, as an X^^-level category cannot generally be embedded in an Xg-level category in natural languages (Baker 1988). 242 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the derivation of other agentive verb compounds (cf. the analysis of (4b) above); (52b") IjiGjip Lisi^ Iagr I jlklil Iaspp Lisi ride~tired-ASP [ (na) yi pi ma]^ Iasp Ivpi Inpi Ivi that one CL horse Ivi ^kl IvP2 InP3 ^«1 Iv'2 Iv2 ^j 1 1 1 1 Î 1 1 Now, there arises a very interesting question about the unambiguity of (52b): why couldn't (52b) be also assigned a structural representation similar to that of (52aii') (or why doesn't (52b) also have the reading of (52aii) ' Lisi rode the horse and as a result he got tired ' ) ? In fact, the reason for the unambiguity of (52b) is not that it cannot be assigned a structural representation similar to that of (52aii'), but that its NP object (na) yi pi ma has to move to the sentence beginning in its derivation from that structural representation. This is exactly the case we see from the derivation of the second subclass of causative verb compound constructions in Section 3.5.8., as shown by the derivation of (39) above. In other words, the structural representation for (52b), which would have the same reading as that of (52aii), is on a par with the 243 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. structural representation for the causative verb compound construction in (39), as illustrated by (81): (81) I AGRP IagrI (aspp IaspI [ CAUSP (cAusl Ivpi Inpi Lisi] Lisi Iv'l Iv'la Ivi ^ij IkP2 yi pi 1 IvP2 IhP3 ride that one CL horse Iv'2 Iv2 lei-le]]]]]]]] tired-ASP The derivation of (52b) from (81) is also parallel to the derivation of (39) (it is the side effect of verb-raising and NP-movement), and the motivation and argument for the verb-raising and NP-movement involved are also the same as those presented in the Section 3.5.8.: (81') I AGRP na yi Pi 11 [qi-[lei-le]^]J^].] that one CL horse ride-tired-ASP I ASPP Lisi* t^] I CAUSP I CAUS I VPI Lisi I NPl ^nl Iv'l I via IVI ^kl InP2 1 IvP2 Inp3 P^^®1 I V'2 Iv2 1111311] 'Lisi rode that/one horse and as a result he got tired.' 244 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thus, (52b), if assigned a structural representation with the same reading as that of (52aii), would change into a causative verb compound construction, and (na) yi pi ma, which is the Theme of its Vi, would be forced to move to the sentence-beginning position by the Causer's feature checking and the Case checking (cf. Section 3.5.8.). 3.7. CONCLUSION In this chapter, I have discussed the eight types of resultative verb compounds in Chinese and their essential properties. I have also described both the semantic and syntactic constraints on their verbal morpheme components and arguments. The review of Li's (1990 & 1993) lexical analyses of the Chinese resultative verb compounds in the early part of this chapter shows that a lexical approach based on such assumptions as random 0-role assignment, 6- role identification, head-feature percolation, causative hierarchy and thematic hierarchy will not only run into the problems of both overgeneration and undergeneration, but also fail to offer an account for the causative verb compound construction where there is no thematic relation between the verb compound and its subject, or to capture the (non)referentiality effects on the grammaticality or 245 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. interpretation of many typical resultative verb compound constructions. It also indicates that causativity is not derived by the 6-grid of the two verbal morphemes in the verb compound or directly by the complex event structure involved in the resultative verb compound, but instead it needs to have its independent status or to be implemented independently in syntax. In order to account for the eight different types of resultative verb compounds and to capture their syntactic and semantic properties and constraints in a principled manner, I have proposed a morpho-syntactic analysis and offered empirical evidence and theoretical motivation for it. Under this morpho-syntactic analysis, various issues involved in the formation of resultative verb compounds in Chinese, especially the grammatical function change of their arguments and the referentiality effects of their NP arguments on the grammaticality or interpretations of many canonical resultative verb compound constructions, all become side effects of verb-raising and NP-movement without assuming any particular lexical rules, which is exactly parallel to the morpho-syntactic analysis of the BA-constructions proposed in Chapter 2. The verb-raising and NP-movement are both morphologically motivated, since the inflectional features of a verb compound and the Case feature of its NP arguments have to be checked (otherwise 246 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. they cannot enter PF under Spell-out, and the derivation would crash at PF). The verb-raising and NP-movement are also constrained by the principles of universal grammar. In fact, the morpho-syntactic analysis presented in this chapter further applies to the study of directional verb compounds in Chinese, which are different from the resultative verb compounds in terms of composition and the distribution of their NP arguments. The discussion and analysis of the Chinese directional verb compounds will be presented in the next chapter, as we proceed. 247 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER FOUR THE SYNTAX OF DIRECTIONAL VERB COMPOUNDS 4.1. INTRODUCTION As mentioned in Chapter 3, the Chinese resultative verb compounds have attracted a great deal of discussion in recent Chinese linguistic studies. By contrast, much less attention is paid to the directional verb compounds of the same language or people treat the directional verb compounds simply as a type of resultative verb compounds. Like the Chinese resultative verb compound, the Chinese directional verb compound also reveals a complex thematic relation with its arguments and might involve grammatical function change of its arguments in its formation. But unlike the Chinese resultative verb compound, which is composed of two lexical verbal morphemes and precedes its object argument (cf. the examples offered in Chapter 3), the directional verb compound is composed of two or three lexical verbal morphemes and it may precede its object or have its object appear between its first verbal morpheme and second verbal morpheme, or between its second verbal morpheme and third verbal morpheme, as exemplified below: 248 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (1) a. Lisi ii-Qu-le yi feng xin. Lisi send-go-ASP one CL letter b. Lisi ii—le yi feng xin qu. Lisi send-ASP one CL letter go 'Lisi send a letter there.' (2) a. Lisi reng-chu—lai—le yi ben shu. Lisi throw-exit-come-ASP one CL book b. Lisi reng-chu—le yi ben shu lai. Lisi throw-exit-ASP one CL book come c. Lisi reng-le yi ben shu chu-lai. Lisi throw-ASP one CL book exit-come 'Lisi threw out a book.' In (1), the directional verb compound is composed of two lexical verbal morphemes, of which the first morpheme ji 'send' implies a displacement and the second morpheme qu 'go' indicates the direction in which the object yi feng xin 'a letter' moves as the result of this displacement. Besides, the object yi feng xin could appear either after the whole directional verb compound, as in (la), or right between its two morphemes, as in (lb). In (2), however, the directional verb compound is made up of three lexical verb morphemes, of which the first morpheme reng 'throw' signifies a displacement action, the second morpheme chu 'exit' signals the general direction of the displacement 249 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of the object yi ben shu 'a book', and the third morpheme lai 'come' indicates the direction of such a displacement with respect to the position of the speaker: that is, the object yi ben shu moves toward the speaker. Besides, the object yi ben shu can occur in three different positions: i) after the whole directional verb compound, as in (2a); ii) between its second morpheme and its third morpheme, as in (2b); and iii) between its first morpheme and its second morpheme, as in (2c). The purpose of this chapter is to discuss the types and properties of directional verb compounds in Chinese, and to extend the morpho-syntactic analysis proposed in Chapter 3 to the Chinese directional verb compound. Like the derivation of the Chinese resultative verb compounds illustrated in Chapter 3, this morpho-syntactic analysis can derive the Chinese directional verb compounds as side effects of verb-raising and NP-movement without requiring any particular lexical rule. It can also incorporate the complex thematic relations between these directional verb compounds and their arguments and can further capture the semantic restriction on the co-occurrence of their verbal morphemes. To support this morpho-syntactic analysis of the Chinese directional verb compound, both the empirical motivation and theoretical arguments will be given, which will not only show the capability of the morpho-syntactic 250 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. analysis to yield well-formed directional verb compounds but exhibit its constraint to rule out ill-formed ones as well. -Finally, I will present a crucial consequence of this morpho-syntactic analysis regarding the distribution of the objects of the Chinese directional verb compounds. 4.2. TYPES OF DIRECTIONAL VERB COMPOUNDS IN CHINESE As illustrated by (1) and (2) above, a directional verb compound in Chinese is composed of either two or three lexical verbal morphemes. Of such a directional verb compound, the first verbal morpheme designates a displacement like fei 'fly', ju 'lift', pao 'run', reng 'throw', tiao 'jump' and zou 'walk', and the second and third verbal morphemes indicate the direction of such a displacement (Li and Thompson 1981). To be more precise, if a directional verb compound is composed of two verbal morphemes, the second verbal morpheme may indicate the general direction of the displacement designated by the first morpheme (i.e. shang 'ascend', xia 'descend' jin 'enter', chu 'exit', qi 'rise', hui 'return', kai 'open' and guo 'cross'), as exemplified below: 251 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (3) Vi (displacement) - V2 (general direction) a. Lisi dai—shang—le tade maozi. Lisi wear-ascend-ASP his hat 'Lisi put on his hat.' b. wo fang—xia-le shubao. I put-descend-ASP schoolbag 'I put my schoolbag down.' c. Lisi zou— in—le (na zuo) wuzi Lisi walk^enter-ASP that CL house 'Lisi walked in the house.' d. Zhangsan na-chu—le yi ba qiang. Zhangsan take-exit~ASP one CL gun 'Zhangsan took out a gun.' Or the second verbal morpheme may indicate the direction of the displacement with respect to the position of the speaker; that is, whether the displacement is towards the speaker (lai 'come') or it is away from the speaker (qu 'go' ) : ^ ^ However, when the speaker is the subject of the directional verb compound, the direction indicated by V2 lai 'come' or qu 'go' is addressee-oriented rather than speaker-oriented (Lu 1973): (i) I pao-lai-le. J run-come-ASP 'I ran over here (towards the addressee).' 252 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (4) Vi (displacement) - V2 (speaker-oriented direction) a. ta ii—lai-le yi ben shu he send-come-ASP one CL book 'He sent a book over here (towards the speaker).' b. Lisi pao—lai—le. Lisi run-come-ASP 'Lisi ran over here (towards the speaker).' c. ta na—qu—le wode shouyinji. he take-go-ASP my radio 'He took away my radio (away from the speaker).' d. Lisi zou—qu-le. Lisi walk~go-ASP 'Lisi went away (from the speaker).' But if the directional verb compound is composed of three verbal morphemes, the second verbal morpheme will signal the general direction of the displacement denoted by the first verbal morpheme, and the third verbal morpheme will indicate the speaker-oriented direction, as shown below: (ii) I zou—qu—le. I walk-go-ASP 'I went away (away from the addressee).' 253 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (5) Vi (displacement) - V2 (general direction) - V3 (speaker-oriented direction) a. ta cha-shano-qu-le yi ge biaozhi. he insert-ascend-go-ASP one CL marker 'He inserted a marker up there.' b. Lisl ba—xia—lai-le yi ke dingzi. Lisi pull-descend-come-ASP one CL nail 'Lisi pulled down a nail.' c. wo ban—iin—ou—le liang ba yizi. I move-enter^go-ASP two CL chair 'I moved the two chairs inside.' d. ta qian—chu—lai—le yi pi ma. he lead-exit-come-ASP one CL horse 'He brought out a horse here.' e. Lisi linq-gi-lai-le yi ge shoutidai. Lisi lift-rise-come-ASP one CL handbag 'Lisi lifted a handbag up.' f. Zhangsan ban—hui—lai-le tade dianshiji. Zhangsan move-return-come-ASP his TV-set 'Zhangsan moved back his TV-set.' g. I tiao-quo-qu—le. I jump-cross-go-ASP 'I jumped over.' 254 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3. PROPERTIES OF DIRECTIONAI. VERB COMPOUNDS IN CHINESE 4.3.1. Displacement Verbal Morphemes According to Li and Thompson (1981:58), one type of displacement verbal morphemes is a verb signaling motion, such as fei 'fly', gun 'roll', liu 'flow', pa 'climb', pao 'run', tiao 'jump' and zou 'walk': (6) a. ta pa—lai—le. he climb-come-ASP 'He climbed here.' b. shitou gun—xia—qu—le. stone roll-descend-go-ASP 'The stone rolled down.' c. tade xue liu-chu-lai—le. his blood flow-exit-come-ASP 'His blood flowed out.' Another type of displacement verbal morphemes is an action verb inherently implying that its direct object undergoes a change of location, such as ban 'remove', ji 'mail', jie 'lend', ju 'lift', la 'poll', ling 'lead', reng 'throw', song 'send' and tui 'push': 255 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (7) a. Lisi ban-lai—le yi zhang shuzhuo. Lisi move-come-ASP one CL desk 'Lisi moved a desk here.' b. ta jie—qu—le wode jiaokeshu. he borrow-go-ASP my textbook 'He borrowed my textbook.' c. ta tui—kai—le na shan men. he push-open-ASP that CL door 'he pushed that door open.' A third type of displacement verbal morphemes is an action verb too, but it is the action expressed by the verb that causes its object to undergo a displacement or change of location, like bang 'tie', chi 'eat', da 'hit', gao 'make', jiao 'call' and shao 'burn': (8) a. ta da—chu-qu—le yi zhi laoshu. He hlt-exit-go-ASP one CL mouse 'He got a mouse out of the house by hitting.' b. ta bang—jin—qu-le san ben shu. he tie~enter-go-ASP three CL book 'he tied three books in.' c. Lisi jiao-1ai—le san wei pengyou. Lisi call-come-ASP three CL 'riend 'Lisi called three of his friends over.' 256 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3.2. Directional Verbal Morphemes According to Chao (1968), Li and Thompson (1981) and LÜ (1984), the general directional verbal morphemes only refer to the following eight verbs: i.e. shang 'ascend', xia 'descend', jin 'enter', chu 'exit', qi 'rise', hui 'return', guo 'cross' and kai 'open': (9) a. ta dai-shana-le maozi. he wear-ascend-ASP hat 'He put on the hat.' b. wo fang— xia— le shubao. I put-descend-ASP schoolbag 'I put down the schoolbag.' c. ta ban-j in—le na zuo xin fangzi. he move-enter-ASP that CL new house 'He moved into that new house.' d. ta reng-chu—lai—le yi ge pingzi. he throw-exit^come-ASP one CL bottle 'He threw out a bottle.' e. L isi i u-q i-le y i zhi shou. Lisi lift-rise-ASP one CL hand 'Lisi raised one of his hands* up.' 257 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. f. Lisi qu—hui—le tade zhaopian. Lisi fetch-return-ASP his photos ' Lisi brought back his photos.' g. qiche shi—guo—le na zuo qiao. ' car drive~cross-ASP that CL bridge 'The car crossed that bridge.' h. wo tui-kai—le men. J push-open-ASP door 'I pushed the door open.' And the speaker-oriented directional verbal morphemes are only lai 'come' and qu 'go': (10) a. tamen pai—lai-le yi ge daibiao. they send-come-ASP one CL delegate 'They sent a delegate over here.' b. na ge xiaohai pa-shang-lai—le. that CL child cliwb-ascend-come-ASP 'That child climbed up here.' c. tamen pai—qu-le yi ge daibiao. they send-go-ASP one CL delegate 'They sent a delegate over there.' d. na ge xiaohai pa-shang-qu—le. that CL child climh-ascend-go-ASP 'That child climbed up there.' 258 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3.3. Semantic Restrictions In the directional verb compound there is a semantic restriction on the co-occurrence of a displacement verbal morpheme and a general directional verbal morpheme: the directionality signaled by the general directional verbal morpheme must be compatible with the directionality that is inherently marked by the displacement verbal morpheme; or the directional verb compound would be ill-formed: (11) a. hei qi manmande iiana—xia—lai—le. black flag slowly fall-descend-come-ASP 'The black flag slowly fell down.' b.*hei qi manmande ^iana—shang—lai—le. black flag slowly fall-ascend-come-ASP (12) a. hong qi manmande shena-shana-qu—le. red flag slowly rise-ascend-go~ASP 'The red flag slowly rose up.' b.*hong qi manmande shena-xia-qu—le. red flag slowly rise-descend-go-ASP (13) a. ta chi—iin-qu—le san wan fan. he eat-enter-go-ASP three CL rice 'He gulped down three bowls of rice.' b.*ta chi—chu-qu—le san wan fan. he eat-exit-go-ASP three CL rice 259 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A similar semantic restriction is also observed on the co-occurrence of a displacement verbal morpheme and a speaker-oriented directional verbal morpheme: (14) a. wo pai—QU—le shenshangde chentu. I tap-go-ASP body's dust 'I tapped the dust off my body.' b.*wo pai—lai—le shenshangde chentu. I tap-come~ASP body's dust (15) a. Lisi iian-qu-le xiaode shuzhi. Lisi cut-go-ASP little tree-branch 'Lisi cut off the little tree-branches.' b.*Lisi 1ian—lai-le xiaode shuzhi. Lisi cut-come-ASP little tree-branch Besides, this semantic restriction further governs the co-occurrence of a general directional verbal morpheme and a speaker-oriented directional verbal morpheme: (16) a. ta iian-qi-lai—le yi ben shu. he pick-up-come-ASP one CL book 'He picked up a book.' b.*ta iian-qi-qu-le yi ben shu. he pick-up-go-ASP one CL book 260 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (17) a. ta tui-kai—lai—le na shan chuangzi. he push-open-come-ASP that CL window 'he pushed that window open.' b.*ta tui—kai—qu-le na shan chuangzi. he push-open-go-ASP that CL window 4.3.4. Distribution of Objects If the first verbal morpheme of a directional V-V compound is an action verb and its second verbal morpheme is a speaker-oriented directional verb, the object of the directional V-V compound can either follow the whole V-V compound, as in (4a), (4c), (7a), (7b), (8c), (10a) and (10c), or appear right between its two verbal morphemes: (18) a. ta ji-le yi ben shu lai. (cf. (4a)) he send-ASP one CL book come 'He sent a book over here.' b. ta na—le wode shouyinji gu. (cf. (4c)) he take-ASP my radio go 'He took away my tape-recorder.' c. ta ban—le yi zhang shuzhuo lai. (cf. (7a)) he move-ASP one CL desk come 'Lisi moved a desk here.' 261 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. d. ta jie-le wode jiaokeshu gu. (cf. (7b)) he borrow~ASP my -textbook go 'He borrowed my textbook.' e. ta iiao-le san wei pengyou lai. (cf. (8c)) he call-ASP three CL friend come 'He called three of his friends over.' f. tamen pai-le yi ge daibiao lai. (cf. (10a)) they send-ASP one CL delegate come 'They sent a delegate over here.' g. tamen pai-le yi ge daibiao gu. (cf. (10c)) they send-ASP one CL delegate go 'They sent a delegate over there.' But if the second verbal morpheme of this directional V-V compound is a general directional verb, the object of the V-V compound has to follow the whole V-V compound, as in (9), or the directional verb compound will be ill-formed: (19) a.*ta dai-le maozi shang. (cf. (9a)) he wear-ASP hat ascend b.*wo fang-le shubao xia. (cf. (9b)) I put-ASP bag descend c.*ta ban—le na zuo fangzi jin. (cf. (9c)) he move-ASP the CL house enter 262 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. d.*Lisi ju-le yi zhi shou c[i. (cf. (9e)) Lisi lift-ASP one CL hand rise e.*Lisi qu—le tade zhaopian hui. (cf. (9f)) Lisi fetch-ASP his photos back Besides, if the first verbal morpheme of a directional V-V compound is a motion verb, then the location object of the V-V compound can only appear after the compound, not between its two verbal morphemes, as shown below: (20) a. women pa-shano-le na zuo shan. we cliwb-ascend-ASP that CL mountain 'we climbed up the mountain.' b.*women pa-le na zuo shan shang. we climb-ASP that CL mountain ascend (21) a. Lisi zou-xia-le louti. Lisi walk-descend-ASP stairs 'Lisi walk down the stairs.' b.*Lisi zou-le louti xia. Lisi walk-ASP stairs descend (22) a. Zhangsan chuang-1in—le wuzi. Zhangsan rush-enter-ASP house 'Zhangsan rush into the house.' b.*Zhangsan chuang-le wuzi jin. Zhangsan rush-ASP house enter 263 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As for the object of directional V-V-V compounds, it could appear either after the whole V-V-V compound, as in (5a), (5b), (5c), (5d), (5e) and (5f), or right between its first verbal morpheme and its second verbal morpheme, as in (23a), (24a), and (25a) below, or even between its second verbal morpheme and its third verbal morpheme, as in (23b), (24b) and (25b) below:^ (23) a. ta cha—le yi ge biaozhi shang-qu. (cf. (5a)) he put-ASP one CL marker ascend-go b. ta cha—shang—le yi ge biaozhi qu. he put-ascend-ASP one CL marker go 'Lisi inserted a marker up there.' (24) a. ta ba—le yi ke dingzi xia—lai. (cf. (5b)) he pu 11-ASP one CL nail descend-come b. Lisi ba-xia-le yi ke dingzi lai. Lisi pull-descend-ASP one CL nail come 'Lisi pulled down a nail.' (25) a. Lisi qian—le yi pi ma chu-lai. (cf. (5d)) Lisi lead-ASP one CL horse exit-come b. Lisi gian-chu—le yi pi ma lai. Lisi lead-exit-ASP one CL horse come 'Lisi brought out a horse here.' ^ The first verbal morpheme of such directional V-V- V compounds needs to be an action verb. 264 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3.5. Aspectual Feature According to Chao (1968), Li and Thompson (1981), Liu (1987), LÜ (1984), Lu (1973 & 1977) and Zhang (1991), the directional verb compound in Chinese normally takes the perfective aspect marker -le, as demonstrated by the examples in the sections above. When the first verbal morpheme of a directional verb compound is a motion verb and does not take an object, such an aspect marker will be obligatory.3 This is exactly parallel to the Chinese resultative verb compound (cf. Section 3.3.1, Chapter 3 ) : (26) a. na bu qiche zuotian kai—lai-le. that CL car yesterday drive-come-ASP 'That car got here yesterday.' b.*na bu qiche zuotian kai—lai. that CL car yesterday drive-come (27) a. Lisi pao-kai-le. Lisi run-away-ASP 'Lisi ran away.' b.*Lisi pao-kai. Lisi run-away ^ Sometimes, the perfective aspect marker attached to the directional verb compound might be optional. But such optionality would suggest a sense of imperativeness (cf. Zhang 1991). 265 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4. A MORPHO-SXNTACTIC ANALYSIS OF DIRECTIONAL VERB COMPOUNDS IN CHINESE 4.4.1. Theoretical Background Since the Chinese directional verb compound consists of either two or three lexical verbal morphemes, of which the second and third verbal morphemes always indicate the direction of the displacement action or motion denoted by the first verbal morpheme, the events associated with the two or three verbal morphemes and their relation could be captured by the semantic and syntactic relations proposed in Section 3.5.1. of Chapter 3 for the events associated with the verbal morphemes of the Chinese resultative verb compound. That is, each of the two or three events being associated with the verbal morphemes of the directional verb compound projects its own predicate structure with the corresponding morpheme as head. If the directional verb compound is composed of two verbal morphemes, the relation between the two events associated with the two verbal morphemes is rendered as a verb-complementation relation between the first verbal morpheme and the VP- predicate headed by the second verbal morpheme. But if the directional verb compound is made up of three verbal morphemes, the relations between the three events being 266 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. associated with the three verbal morphemes are rendered as two verb-complementation relations, one of which is between the first verbal morpheme and the VP-predicate headed by the second verbal morpheme, and the other of which is between the second verbal morpheme and the VP- predicate headed by the third verbal morpheme. On a par with these syntactic relations, there are also similarly asymmetric semantic relations between these two or three events designated by the directional verb compound: the displacement action or motion denoted by its first verbal morpheme "implicates" the directional movement denoted by its second verbal morpheme; and the directional movement denoted by its second verbal morpheme "implicates" the speaker-oriented directional movement signified by its third verbal morpheme (cf. Baker 1988; Hale and Keyser 1993) : (28) a. eventl — > eventz — > (events) b. VPl VI ^ VP2 NP ^V'2 V2 (VP3) 267 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The verb-complementation relations between Vi and VP2 and between V2 and VP3 are quite plausible, as the syntactic embedding usually corresponds to a semantic composite in which the event denoted by the subordinate VP is a proper part of the event denoted by the superordinate verb (cf. Section 3.5.1., Chapter 3). Based on the syntactic structure and its associated semantic relation in (28ab), we can establish the basic structure of the directional verb compound construction in Chinese as follows, assuming a version of Chomsky's (1993) clausal structure; (29) AGRP / X Spec AGR' / X, AGR ASP? / X Spec ASP' ASP VPl Spec V'l / \ VI VP2 / \ HP V'2 / \ V2 (VP3) This basic structure is parallel to the basic structure of the Chinese resultative verb compound construction we proposed in Chapter 3, except for the possible inclusion 268 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of VP3 (cf. Section 3.5.1., Chapter 3 for the arguments and motivation for such basic structures ). Besides, this basic structure captures the aspectual feature reflected on the Chinese directional verb compound as a selectional property of AGR: that is, the directional verb compound normally carries the perfective aspect marker -le with it (cf. Section 4.3.5.). With the basic structure of the Chinese directional verb compound construction established in (29), I would now proceed to offer a morpho-syntactic analysis of the various types of directional verb compounds introduced in Sections 4.2. and 4.3. above, which is presented in the following sections. 4.4.2. The Directional V—V Compound Construction (1) As described in Sections 4.2. and 4.3., one type of directional V-V compounds is composed of a Vi conveying a displacement action and a V2 demonstrating the general direction of the displacement action. The subject of this type of directional V-V compounds is the Agent of Vi, and their object is both the Theme of Vi and the Experiencer of V2, as in (3a), (3b), (3d), (9d), (9e), (9f) and (9h). Consider (3a), which is repeated below: 269 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (3a) I i i s i d a i —s h a n g —l e t a d e m a o z i . Lisi wear-ascend-ASP his hat 'Lisi put on his hat.' Given the basic structure of the Chinese directional verb compound construction in (29), (3a) would have the following structural representation in terms of thematic relation and X'-theory, still assuming the hypothesis of VP-internal subject (Koopman and Sportiche 1988; Kuroda 1988): (3a') [j^GRp [ a g r ] I a s p p I a s p I I v p i I n p i [y i [yi d a i j Lisi wear Ivp2 I n p2 tade maozi] l y j shang-le]]]]]]] his hat ascend-ASP (3a') is on a par with the structural representation of (4b') displayed in Section 3.5.2 of Chapter 3. Thus, the derivation of (3a) from (3a'), which is exhibited below, would be similar to the derivation of (4b) from (4b') in Chapter 3. That is, V2 shang-le 'ascend' is first raised to compound with Vi dai 'wear' by verb incorporation, and the resulting V-V compound dai—shang—le is then raised to AGR via ASP to check its inflectional features. NPl Lisi receives its 0-role from V'l, and it moves to the Spec of 270 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. AGRP to check its Case against AGR. NP2 tade maozi 'his hat' receives its 6-role from V'2 and moves into the Spec of ASPP to check its Case against ASP. The derivation of (3a) from (3a') is illustrated below: (3a") [ftcRP Cagr [ [dai— [shang—le] [aspp Lisi wear-ascend-ASP tade maozi^ Iasp Ivpi Inpi Ivi Ivi ^kl his hat IvP2 InP2 ïv'2 Iv2 tjl]]]]]] The motivation and arguments for the derivation of (3a) would also be similar to those for the derivation of (4b) in Chapter 3. First, the verb-raising from V2 into AGR through Vi and ASP is morphologically driven, as the inflectional features reflected on V2 shang-le have to be checked against the features of ASP and AGR in order for the verb to enter the PF component under Spell-Out. This verb-raising is also licit by the Minimal Link Condition, because each of its three steps (i.e. from V2 to Vi, from Vi to ASP, and from ASP to AGR) attaches the verb to the nearest head target that immediately c-commands the verb. Second, the movement of NPl Lisi to the Spec of AGRP and the movement of NP2 tade maozi into the Spec of ASPP are forced by the Case Filter, and they are both licit by 271 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the Shortest Movement Condition, due to the verb-raising. That is, as the verb compound dai-shang—le is raised from VI to ASP to form the chain (dai-shang—le%, tj^) with the minimal domain {Spec of ASPP, Spec of VPi, VP2}, the Spec of ASPP and the Spec of VPi become equidistant from VP2 or any element it contains. Hence, NP2 tade maozi, as a specifier of VP2, may move to the Spec of ASP by crossing the Spec of VPi which is filled with NPl Lisi. When the verb compound is further raised from ASP to AGR to form the new chain ( dai-shang-le^, t^) with the minimal domain {Spec of AGRP, Spec of ASPP, VPl}, the Spec of AGRP and the Spec of ASPP are equidistant from VPi or anything it contains. Thus, NPi Lisi, as a specifier of VPi, is able to move to the Spec of AGRP by crossing the Spec of ASPP which is now occupied by NP2 Lisi. As a consequence of this morpho-syntactic analysis, the ill-formed directional verb compound construction in (30) below is correctly ruled out as a violation of the Shortest Movement Condition: (30)*tade maozi dai—shang-le Lisi. his hat wear-ascend-ASP Lisi The structural representation of (30) would be the same as that of (3a) by the above analysis (cf. (3a')). But 272 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in (30), the internal subject of VPl, namely Lisi, moves to the Spec of ASPP whereas the internal subject of VP2, namely tade maozi, moves to the Spec of AGRP. These two NP-movements are exactly opposite to the two NP-movements in (3a"), as shown by (30') below: (30')*I^grp tade maozii [ [dai-[shang-le] j] 1] his hat wear-ascend-ASP Iaspp ^isi^ [ASP til Ivpi Impi t,] Iv i Ivi t*] Lisi IvP2 IhP2 til IV'2 lv2 tjlllllll It is the movement of NP2 tade maozi to the Spec of AGRP in (30') that violates the Shortest Movement Condition, as it illicitly crosses the trace left by the movement of NPl Lisi to the Spec of ASPP. To be more specific, when NPl Lisi moves to the Spec of ASPP, NP2 tade maozi cannot move into the same Spec position. In order to check its Case, NP2 tade maozi has to move directly to the Spec of AGRP. But this movement is not possible, even after the verb compound is raised from ASP to AGR. This is because the verb-raising from ASP to AGR would create a new chain with the new minimal domain {Spec of AGRP, Spec of ASPP, VPl}, of which the Spec of VPi is not a member. Hence, 273 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. NP2 tade maozi cannot cross the NPi-trace t^ in the Spec of VPl to reach the Spec of AGRP. This morpho-syntactic analysis also derives another type of directional V-V compounds as the side effects of verb-raising and NP-movement. Of this type of directional V-V compounds, Vi is still a displacement action verb but V2 marks a speaker-oriented direction. Like (3a) above, the subject of this type of directional V-V compounds is the Agent of Vi, and their object is both the Theme of Vi and the Experiencer of V2, as in (4a), (4c), (7a), (7b), (10a) and (10c). Thus, their structural representation would be exactly parallel to that of (3a), and so would their derivation, as displayed by the derivation of (4a): (4a) ta i i - l a i — le yi ben shu he send-come-ASP one CL book 'He sent a book over here.' (4a') [ ; ^ G R p [agrI Iaspp IaspI Ivpi Inpi I v i Ivi 3^1 he send Iyp2 Inp2 ben shu] [y» 2 Iv2 111111 one CL book come-ASP (4a") [A G R P tUi [A G R I [ j i- I la i- le ] j ] J il [^pp yi ben shu, he send-come-ASP one CL book I A S P ^ll I V P l I N P l I V'l IV I IvP 2 In P 2 Iv'2 Iv2 tj]]]]]]] 274 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.3. The Directional V-V Compound Construction (2) The morpho-syntactic analysis above accounts for the formation of a third type of directional V-V compounds, of which Vi denotes a displacement motion and V2 signals the general direction of such a motion. The subject of this type of directional V-V compounds is both the Agent of Vi and the Experiencer of V2, and their object is only the Location of V2, as in (3c), (9c), (9g), (20a), (21a) and (22a). Consider (3c), which is rewritten below: (3c) Lisi zou-iin-le (na zuo) wuzi Lisi walk-enter-ASP (that CL) house 'Lisi walked in the house.' Given the basic structure of the Chinese directional verb compound construction in (29), we would expect (3c) to be structurally represented in the following way: (3c') [agrI I aspp IaspI Ivpi Inpi ^ isi] [y,i [yi zou] Lisi walk IvP2 In P 2 I V'2 Iv2 jio — le] [up3 w uzi]]]]]]] enter-ASP house 275 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (3c') is on a par with the structural representation of (16') displayed in Section 3.5.5 of Chapter 3. Thus, the derivation of (3c) from (3c'), which is exhibited below, would be similar to the derivation of (16) from (16') in Chapter 3. That is, V2 jin—le 'enter' in (3c') is raised to combine with Vi zou 'walk', and the resulting compound zou-jin-le is further raised to AGR via ASP to check its inflectional features. NPi Lisi, which receives a 6-role from V'l, is moved to the Spec of AGRP to check its Case against AGR. NP3 wuzi 'house', which receives a 6-role from V2, is moved to the Spec of ASPP to check its Case against ASP. As for NP2 Pro, which is the specifier of VP2, is properly controlled by NPi Lisi or its trace that is the closest NP C-commanding Pro. Thus, the derivation of (3c) also falls under verb-raising and NP-movement, as illustrated below; (3c") Lisi^ Iagr [ [zou—[jin—le] j ] 1) ^aspp Lisi walk-enter-ASP house Iasp Evpi Ihpi ^±1 Iv'i Ivi Ivp2 Inp2 Iv'2 Iv2 Imp3 t*]]]]]]] The motivation and arguments for the verb-raising to AGR from V2 and for the movement of NPl Lisi to the Spec of AGRP are the same as those presented in Section 4.4.2. 276 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As for the movement of NP3 wuzi into the Spec of ASPP, it is forced by the Case Filter and meets with the Shortest Movement Condition, even if NP2 Pro intervenes in between (see Section 3.5.4., Chapter 3 for the invisible status of Pro). That is, as the verb compound zou—j in—le raises itself from Vl to ASP to form the chain (zou-jin-le%, t%) with the minimal domain {Spec of ASPP, Spec of VPi, VP2}, the Spec of ASPP and the Spec of VPi become equidistant from VP2 or anything it contains. Hence, NP3 wuzi, as a complement of V2, may move to the Spec of ASP by crossing the Spec of VPi that is filled with NPl Lisi. As a consequence of this morpho-syntactic analysis, the ill-formed directional verb compound construction in (31) below, whose structural representation would be the same as that of (3c'), is ruled out as a violation of the Shortest Movement Condition correctly: (31) *na zuo wuzi zou—i in-le Lisi that CL house walk-enter-ASP Lisi (31')*Iagrp na zuo wuzi^ [ [zou-I jin-le] that CL house walk-enter-ASP I ASPP ^^ai„ (ASP I VPl I npi ^*1 Ivi Ivi Lisi IvP2 InP2 PfO] I V'2 Iv2 ^jl InP3 111 1 1 1 277 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. That is, the movement of NP3 na zuo wuzi 'that house' to the Spec of AGRP in (31') illicitly crosses the NP-trace t^, which is left in the Spec of VPi by the movement of NPl Lisi to the Spec of ASPP (cf. the account of the ill- formed directional V-V compound construction in (30) in Section 4.4.2.). This morpho-syntactic analysis also gives an account for the formation of the fourth type of directional V-V compounds which are the same as (3c) except that they do not take a location object, as shown by the derivation of (32) below: (32) wo tang-xia—le. I lie-descend-ASP 'I lay down.' (32') Ij^GRp Iagr] 'aspp ^aspI Ivpi Inpi Ivi Ivi tang] J lie IvP2 InP2 I V '2 Iv2 Î 1 1 Î ] Î descend-ASP (32") [agrp I AGR I [tang—[xia—le] j ] il laspp Iasp t^] I lie-descend-ASP I V P l I N P l tj.] Iv'l IV I tjç] Ivp2 InP2 Pf®] Iv'2 Iv2 t j ] ]]]]]] 278 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It is the lack of the location object in (32) that makes its ASP (or AGRo) (cf. Section 3.5.1., Chapter 3) inert, failing to assign Case or check the Case of an NP moved into its Spec position (cf. Section 3.5.6., Chapter 3). This is why NPl wo 'I' in (32") moves directly into the Spec of AGRP without passing ASP. This NP-movement also satisfies the Shortest Movement Condition since the Spec of AGRP and the Spec of ASPP are equidistant from NPl wo, due to the verb-raising from ASP to AGR: i) the raising of the verb compound tang—xia-le 'lie-descend' from ASP to AGR forms a chain (tang-xia—le^, t^) with the minimal domain {Spec of AGRP, Spec of ASPP, VPl}, so the Spec of AGRP and the Spec of ASPP are equidistant from VPi or any element it contains; and ii) thus, NPi wo, as a specifier of VPl, may move to the Spec of AGRP by crossing the Spec of ASP. The same analysis further accounts for the formation of the fifth type of directional V-V compounds which are the same as (32) except that their V2 signals a speaker- oriented direction, as in (4b), (4d) and (6a) above. The derivation of these directional V-V compounds is similar to the derivation of (32) from (32'), and its motivation and arguments are also the same as those given above, as illustrated by the derivation of (4b) below: 279 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (4b) Lis! pao— la i— l e ♦ Lisi run-come-ASP 'Lisi ran over here.' (4b') Lisi^ I [P^o— [la i— le] [j^pp I^sp ^lî Lisi run-come~ASP fypi InPI ^il ïv'l Ivi ^kl IvP2 InP2 Îv '2 IV 2 tjlllllll 4.4.4. The Directional V—V—V Compound Construction The morpho-syntactic analysis proposed above extends itself to account for the formation of directional V-V-V compounds in Chinese. As described in Sections 4.2. and 4.3., one type of directional V-V-V compounds is composed of a Vi signifying a displacement action, a V2 conveying a general direction, and a V3 denoting a speaker-oriented direction. The subject of this type of directional V-V-V compounds is the Agent of Vi, and their direct object is the Theme of Vi and the Experiencer of both V2 and V3, as displayed by (5a), (5b), (5c), (5d), (5e), (5f), (8a) and (8b). Consider (5d), which is repeated below: 280 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (5d) ta qian—chu—lai—le yi pi ma. he lead-exit-come-ASP one CL horse 'He brought out a horse here.' Given the basic structure of the Chinese directional verb compound construction in (29), we would expect (5d) to have the following structural representation: (5d') ÏagrÎ Iaspp IaspI Ivpi Inpi Iv'i Ivi 9^®“! he lead IvP2 IuP2 Iv'2 Iv2 chu] tvP3 IhP3 P^®1 one CL horse exit [v'3 Iv3 lai-le]]]]]]]]] come-ASP Like the derivation of the five types of directional V-V compound constructions presented above, the derivation of the directional V-V-V compound construction in (5d) also falls under verb-raising and NP-movement without needing any specific lexical rules. That is, V3 lai-le 'come' in ( 5d ' ) is first raised to compound with V2 chu ' exit ', and the resulting V-V compound chu—lai-le is then raised to combine with Vi qian 'lead', yielding the V-V-V compound qian—chu—lai-le. This V-V-V compound is further raised to AGR via ASP to check its inflectional features. NPl 281 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ta 'he', which receives a 0-role from V'l, moves to the Spec of AGRP to check its Case against AGR. NP2 yi pi m a 'a horse', which receives a 8-role from V'2, moves to the Spec of ASPP to check its Case against ASP. And NP3 Pro, which is the specifier of VP3, is properly controlled by NP2 y i pi m a or by its trace which is the closest NP C- commanding Pro. The derivation of (5d) is shown below: (5d") [AGRP [agr [[qian-lchu-llai-lejjl^lil,] [aspp he lead-exit-come-ASP y i pi m a^ [asp t^J [yp^ [xpi ^il [v i (vi (vp2 one CL horse IhP2 ^nl Iv2 [vP3 ÎHP3 [y'3 Iv3 ^j 1 1 1 1 1 1 1 1 1 The motivation and arguments for the movement of NPi ta into the Spec of AGRP and the movement of NP2 yi pi m a to the Spec of ASPP are the same as those provided in the above sections. As for the verb-raising from V3 into AGR via V2, VI and ASP, it is legitimate by the Minimal Link Condition, because each of its four steps (i.e. from V3 to V2, from V2 to Vl, from Vi to ASP and from ASP to AGR) attaches the verb to the nearest head target C-commanding the verb immediately, without skipping a head position that is already-filled. 282 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As expected, this morpho-syntactic analysis excludes the ill-formed directional V-V-V compouiW construction in (33) as a violation of the Shortest Movement Condition. The structural representation of (33) would be the same as that of (5d/) by the above analysis: (33) *yi pi ma gian-chu-lai—le ta. one CL horse lead-exit-come-ASP he (33')*Iagrp yi pi Iagr [ [qian-[chu-[lai-le] j] , , 1 il.l one CL horse lead-exit-come-ASP EasPP IaSP ^*1 Evpi ÎnpI Iv'l Ivi IvP2 he InP2 Iv2 ^kl lvP3 IkP3 PI’ O] [ ^ , 3 [ ^ 3 1 1 1 1 1 1 1 1 1 That is, it is the movement of NP2 yi pi ma 'a horse' to the Spec of AGRP in (33') that illicitly crosses the NP- trace t^ left in the Spec of VPi by the movement of NPi ta 'he' to the Spec of ASPP (cf. the account of the ill- formed directional V-V compound constructions in (30) and (31) presented in Sections 4.4.2. and 4.4.3.). The morpho-syntactic analysis proposed above further accounts for the formation of another type of directional V-V-V compounds whose Vi signifies a motion, and whose V2 and V3 still indicate a general direction and a speaker- 283 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. oriented direction respectively. The only argument that these directional V-V-V compounds take is their subject, which is the Agent of Vi and the Experiencer of both V2 and V3, as in (5g), (6b), (6c), (10b), (lOd), (11a) and (12a). Consider (5g), which is repeated below and whose structural representation would be in the following form under the basic structure of the Chinese directional verb compound construction in (29): (5g) I tiao—guo-qu-le. I jump-cross-go-ASP 'I jumped over.' (5g') [agrp IagrI Iaspp IaspI Ivpi Inpi Iv'i Ivi I jump IvP2 InP2 Iv'2 Iv2 IvP3 IhP3 Iv'3 cross Iv3 q«-ieiiii]jiii go-ASP The derivation of (5g) is analogous to that of (32) in Section 4.4.3. That is, V3 qu—le 'go' in (5g') is first raised to compound with V2 guo ' cross ', the resulting V-V compound guo-qu-le is then raised to combine with Vi tiao 'jump', yielding the V-V-V compound tiao—guo-qu-le. This V-V-V compound is further raised to AGR via ASP to check 284 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. its inflectional features. NPi wo 'I', which receives a 0-role from V'l, moves into the Spec of AGRP to check its Case. NP2 Pro, which is the specifier of VP2, is properly controlled by NPl wo or its trace. And NP3 Pro, which is the specifier of VP3, is properly controlled by NP2 Pro. Since NP2 is controlled by NPi and NP3 is controlled by NP2, Pro in NP3 would be indirectly controlled by NPi wo by means of control transition. The derivation of (5g) is illustrated below and its motivation and arguments are the same as those presented above: (5g") Cagrp Iagr [tiao-[guo-[gu-le]j]kIi]J Iaspp I jump-cross-go-ASP Ïasp ^*1 Ivpi Inpi Ivi Ivi ^il Ivp2 Ihp2 Iv'2 Iv2 IvP3 InP3 Iv'3 Iv3 ^j Î 1 Î 1 1 1 1 1 1 4.5. THE CONSEQUENCES OF THE MORPHO-SYNTACTIC ANALYSIS 4.5.1. Co-occurrence between the Verbal Morphemes of Directional Verb Compound As discussed in Section 4.3.3., there is a semantic restriction on the co-occurrence of a displacement verbal 285 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. morpheme and a general directional verbal morpheme in the directional verb compound: that is, the directionalities signified by the two verbal morphemes must be compatible with each other. The same restrictions are also found on the co-occurrence of a displacement verbal morpheme and a speaker-oriented directional verbal morpheme and on the co-occurrence of a general directional verbal morpheme and a speaker-oriented directional verbal morpheme, as in (11) through (17) above. These semantic restrictions, in fact, could be captured by the morpho-syntactic analysis proposed above as selectional properties of Vi and V2 of the directional V-V and V-V-V compounds (cf. Chomsky 1986 & 1993). That is to say, the displacement Vi s-selects a general directional VP-predicate whose directionality is compatible with the inherent directionality of Vi; and the general directional V2 s-selects a VP-predicate whose directionality is compatible with the directionality of V2. When V3 moves into V2 and V2 is raised to Vi in the process of deriving directional V-V-V and V-V compounds, the directionality of V3 will be checked against that of V2, and the directionality of V2 will be checked against that of VI. If the directionality of V3 matches that of V2 and the directionality of V2 matches that of Vi, then the derivation of directional V-V-V or V-V compounds will converge at LF, thus ruling in both the speaker-oriented 286 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. directional V3 or V2 and the general directional V2, as in (11a) through (17a). But if the directionality of V3 is incompatible with that of V2 or the directionality of V2 is incompatible with that of Vi, then the derivation of directional V-V-V or V-V compounds would crash at LF, thus ruling out the speaker-oriented directional V3 or V2 or the general directional V2, as in (11b) through (17b). 4.5.2. Directional V-V Compound and Distribution of Its Object As described in Section 4.3.4., the object of the directional V-V compound can either follow the whole V-V compound or occur between its two verbal morphemes if the first morpheme is an action verb and the second morpheme is a speaker-oriented directional verb, as shown by the correspondence between (4a) and (18a), (4c) and (18b), (7a) and (18c), (7b) and (18d), etc. But if the second morpheme is a general directional verb, then the object can only follow the whole V-V compound, as shown by the correspondence between (9a) and (19a), (9b) and (19b), (9c) and (19c), etc. Consider the contrast between the group of (4a) and (18a) and the group of (9a) and (19a), which are rewritten below as (34ab) and (35ab): 287 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (34) a. ta j i—lai—le yi ben shu (= (4a)) he send-come-ASP one CL book b. ta ji-le yi ben shu lai. (= (18a)) he send-ASP one CL book come 'He sent a book over here.' (35) a. ta dai—shang—le maozi. (= (9a)) he wear-ascend-ASP hat 'He put on the hat.' b.*ta dai-le maozi shang. (= (19a)) he wear-ASP hat ascend The contrast between these two groups of directional V-V compound constructions with respect to the distribution of the objects, I believe, lies in the lexical difference between the speaker-oriented directional verbal morpheme and the general directional verbal morpheme. Namely, the speaker-oriented directional verbal morpheme behaves like a free morpheme which could stand alone as an independent verb or be attached to another verbal morpheme, depending on whether it carries an aspect marker or not; while the general directional verbal morpheme behaves like a bound morpheme which needs to attach itself to a free verbal morpheme. This lexical difference is, in fact, evidenced by the formation of directional V-V-V compounds: that is, the general directional verbal morpheme must be attached 288 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. either to the displacement verbal morpheme, as in (23b), (24b) and (25b), or to the speaker-oriented directional verbal morpheme, as in (23a), (24a), (25a), or to both of them, as in (5a), (5b) and (5d); but the same requirement does not apply to the speaker-oriented directional verbal morpheme, as in (23b), (24b) and (25b). For the sake of convenience, (25a), (25b) and (5d) are rewritten below to illustrate the lexical difference between the two types of directional verbal :aorphemes : (36) a. Lisi qian-chu—le yi pi ma lai. (= (25b)) Lisi lead-exit-ASP one CL horse come 'Lisi brought out a horse here.' b.*Lisi chu-le yi pi ma lai. Lisi exit-ASP one CL horse come (37) a. Lisi gian-le yi pi ma chu-lai. (= (25a)) Lisi lead-ASP one CL horse exit-come 'Lisi brought out a horse here.' b.*Lisi gian-le yi pi ma chu. Lisi lead-ASP one CL horse exit (38) Lisi qian-chu-lai-le yi pi ma. (= (5d)) Lisi lead-exit-come-ASP one CL horse 'Lisi brought out a horse here.' 289 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thus, the different distribution of the objects of these two types of directional V-V compounds are nothing but a syntactic consequence generated by the lexical difference between the speaker-oriented directional verbal morpheme and general directional verbal morpheme. This syntactic consequence can also be captured by the morpho-syntactic analysis proposed above as a side effect of verb-raising. To be more specific, in the derivation of a directional V-V compound whose V2 is a speaker-oriented directional verbal morpheme, the raising of this directional verbal morpheme is optional, depending on whether it carries an aspect marker or not. This is attributable to the free morpheme status of the speaker-oriented directional verb, as shown by the derivations of (34a) and (34b) below (cf. the derivation of (4a) and its motivation and arguments in Section 4.4.2.): (34a') IagrI [ a s p p [aspI E v p i [ h p i I v i [vi 3^1 he send [ v p 2 [n p2 b e n s h u ] [ y ' 2 [ v 2 1 ] 1 1 11 one CL book come~ASP (34a") [ j^ g r p ^ ® i [ a g r I j l j c l i l [ a s p p ^ ® “ s h u ^ he send-come-ASP one CL book [ a s p [ y p i [ r p i [ v ' l [ v i '*'k l [ y p 2 [ r p 2 ^ « 1 [v'2 [v2 t j ] ]]]]]] 290 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (34b') [agrp IagrI Iaspp IaspI Ivpi Inpi Ivi Ivi he send-ASP IvP2 Inp2 ben sbu] lv'2 Iv2 one CL book come (34b" ) [j^GRp [agr I l 3 ^ ” ^ ® l j l k l I ASPP ^®“ shu^ he send-ASP one CL book I ASP ^ k l I VPl I NPl ^ i l I V'l IVI ^ j l IvP2 In P 2 *"ml Iv'2 Iv2 come In (34a), the perfective aspect marker —le is attached to the speaker-oriented directional V2 lai 'come', implying that V2 lai presumably carries -le in the lexicon as an intrinsic property before Spell-out (cf. Chomsky 1993). Hence, V2 lai must be raised to AGR via ASP to check its aspectual and agreement features, as shown by (34a') and (34a"). By contrast, in (34b) the same perfective aspect marker is attached to the displacement Vi ji 'send', thus making the speaker-oriented directional V2 lai a bare and uninflected morpheme.* As a result, Vi ji must move to * Although V2 in (34b) appears to be able to carry -le, that —le is not an aspect marker but an inchoative marker, because it occurs at the sentence-final position (see Note 8 in Chapter 2 for the differences between the perfective aspect marker -le and the inchoative marker —le in Chinese ). This is evidenced by the co-occurrence of the two types of —le in the directional verb compound construction, as shown below: 291 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. AGR via ASP to check its inflectional features, and V2 lai, which is not forced to move elsewhere, is then left in-situ, as shown by (34b') and (34b"). As for the derivation of a directional V-V compound whose V2 is a general directional verbal morpheme, the raising of the directional verbal morpheme is obligatory, due to its bound morpheme status. This is shown by the derivations of (35a) and (35b) below (cf. the derivation of (3a) and its motivation and arguments in Section 4.4.2 above): (35a') [j^Qup [agr] Iaspp IaspI Ivpi Ikpi Ivi Ivi dai] he wear IvP2 Ikp2 =aozi] [y,2 (v2 shang-le]]]]]]] hat ascend-ASP (35a") I AGRP ta^ Iagr IIdai-[shang-le]j],^lil Iaspp ®aozi. he wear-ascend-ASP hat I ASP ^ll I VPl I NPl Iv'l Ivi IvP2 InP2 ^*1 IV2 IV2 tj] ]]]]]] (i) ta ii—le yi ben shu lai—le. he send-ASP one CL book come-INC 'He sent a book over here.' In fact, the inchoative marker -le behaves more like a X°-category occupying the head COMP of the matrix clause, comparable to such Chinese question markers as -ma and ne (cf. Aoun and Li 1993). 292 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (35b') [j^ grp IagrI Iaspp IaspH vpi Inpi Ivi Ivi he wear-ASP IvP2 In p 2 maozi; Iv2 shang] ]]]]]] hat ascend ( 35^*') * Iagrp I AGR I I^ai— le] j]jj] [j^ g p p maozij, Iasp ^kl he wear ~ ASP hat I V P l I N P l ^il I V'l Ivi ^jl IvP 2 In p2 ^ *1 I V '2 Iv2 shang]]]]]]] ascend In (35a), both the aspectual feature taken by the general directional V2 shang— le 'ascend' and its bound morpheme status force this directional verbal morpheme to move to compound with Vl dai 'wear' and also drive the resulting V-V compound dai-shang-le to move to AGR via ASP to check its inflectional features, as shown by (35a') and (35a"). But in (35b), though the aspectual feature reflected on the displacement Vi dai-le would force this displacement verbal morpheme to move to AGR through ASP to check its inflectional features, the general directional V2 shang, which is a bound and uninflected morpheme, is left in- situ without being attached to any free verbal morpheme, thus violating the morphological principle that a bound morpheme needs to be attached to a free morpheme. This is shown by (35b') and (35b"). 293 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The lexical difference between, the speaker-oriented directional verbal morpheme and the general directional verbal morpheme is also responsible for the distribution of the location object of a directional V-V compound, of which Vl is a motion verb and V2 is a general directional verb; that is, the location object of such a V-V compound could only appear after the compound, not between its two morphemes, because its V2 is a general directional verbal morpheme, thus a bound morpheme which has to be raised to attach to Vl, as in (20), (21) and (22). Consider (20), whose two derivations are shown below (cf. the derivation of (3c) and its motivation and arguments in Section 4.4.3 and also the account of the distribution of the object of the directional V-V compound in (35a) and (35b)): (20) a. ta pa-shang-le (na zuo) shan. he climb-ascend-ASP that CL mountain 'He climbed up the mountain.' b.*ta pa—le (na zuo) shan shang. he climb-ASP that CL mountain ascend (20a') [agrI (aspp IaspI Ivpi 1»pi Iv'i Ivi he climb [vP2 Ïnp2 Iv'2 Iv2 shang—le] (|jp3 shan]]]]]]] ascend-ASP mountain 294 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (20a") [^Qjtp Iagr [ [pa—[shang—le] j] 1] [aspp shan^ he cliwb-ascend~ASP mountain [asp [yPl [nPI ^iî [vi [vi ^kJ [vP2 [kP2 [v'2 [v2 ^jl [mP3 ^*1111311 (20b') [agrp [AGR3 [aspp [aspI [ypi [rpi [y'l [vi pa—la] he climb-ASP [yp2 [rp2 [v'2 [v2 shang] [up3 shan]]]]]]] ascend mountc^in (20b")*[^gjjp ta^ [AGR [ [pa*"lel jlkl [aspp s^an^ [ASP b^] he climb-ASP mountain [vpi [rPI ^i] [v'l [vi ^ji [yP2 [nP2 Pf°] [y'2 [V2 shang] [^3 tj]]]]]] ascend 4.5.3. Directional V—V—V Compound and Distribution of Its Object The lexical difference between the speaker-oriented directional verbal morpheme and the general directional verbal morpheme further decides the distribution of the object of directional V-V-V compounds. That is, in the derivation of directional V-V-V compounds whose Vi is an action verb, the incorporation of the speaker-oriented 295 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. directional V3 with the general directional V2 and with the displacement Vi is optional, depending on whether the speaker-oriented directional V3 carries an aspect marker or whether the general directional V2 is incorporated to the displacement Vl. If the speaker-oriented directional V3 carries an aspect marker, then it has to be raised to incorporate with both the general directional V2 and the displacement Vi in order for it to move into AGR through ASP to check its aspectual and agreement features, thus accounting for why the object of this directional V-V-V compound appears after the whole V-V-V compound. This is illustrated by the derivation of (5d), as repeated below (cf. its motivation and arguments in Section 4.4.4.): (5d) ta qian-chu—lai—le yi pi ma. he lead-exit-come-ASP one CL horse 'He brought out a horse here.' (5d') [j^QRp [agr] Iaspp IaspI Ivpi Ibjpi ^^3 Ivi Ivi he lead IvP2 InP2 Iv'2 Iv2 chu] IvP3 InP3 one CL horse exit Iv3 Iv3 lai-le]]]]]]]]] come-ASP 296 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (5d") taj^ [agr [ [lai—le] [j^pp he lead-exit-come-ASP yi pi m a ^ lASP ^al [vpi [rpi [v'l [vi ^.i3 [yp2 one CL horse [ rP2 [v2 ^ k l [vP3 [uP3 [v'3 [v3 j 1 1 1 1 1 1 Î Î But if it is the general directional V2, not the speaker- oriented directional V3, that carries an aspect marker, then V2 will be forced to incorporate with Vi, and the resulting V-V compound will be further raised to AGR via ASP to check its aspectual and agreement features. As a result, V3, being uninflected and a free morpheme, would be left in-situ since it is not forced to move elsewhere, thus placing the object of the directional V-V-V compound between V2 and V3, as shown by the derivation of (25b); (25b) ta qian-chu—le yi pi ma lai. he lead-exit-ASP one CL horse come 'he brought out a horse here.' (25b') [j^Gpp [a g r I [a s p p [a s p I [v p i [r p i [v'l [vi he lead [vP2 [rP2 [v'2 [v2 chu-le] [yp3 one CL horse exit-ASP [n p 3 Pro] [v3 lai]]]]]]]]] come 297 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (25b") [j^ grp ^^îIagr [ le] [^^pp yi pi ma^ he lead-exit-ASP a CL horse Iasp Ivpi Ikpi Ivi Ivi ^kl Ivp2 Inp2 ^nl Iv'2 Iv2 IvP3 IkP3 Iv'3 Iv3 ^^^311111111 come Moreover, if it is the displacement V i, rather than the general directional V2 or speaker-oriented directional V3, that carriers an aspect marker, then Vi has to move to AGR via ASP to check its inflectional features, thus leaving V2 and V3 behind. As the general directional V2 is a bound morpheme which needs to be attached to a free morpheme, the speaker-oriented V3 will then be forced to incorporate with V2 to support it. This explains why the object of this directional V-V-V compound appears between Vl and V2, as shown by the derivation of (25a) below: (25a) ta gian-le yi pi ma chu-lai. he lead-ASP one CL horse exit-come (25a') (j^oRp I AGR 3 Iaspp Iasp3 Ivpi Inpi ^^3 Ivi Ivi he lead-ASP IvP2 IhP2 ®^3 I V'2 Iv2 ®bu] lyp3 InP3 P^®3 one CL horse exit Iv'3 Iv3 lai333 333 333 come 298 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (25b") ta^ [qian-le]j]R] [^gpp yi pi ma^ he lead-ASP a CL horse fftsp Ivpi Ihpi ^îÎ Ivi Ivi Ivp2 Inp2 Iv2 lvP3 I MP3 IV3 lv3 **]]]]]]]] exit-come As demonstrated above, the proposed morpho-syntactic analysis nicely captures the variable distribution of the objects of both directional V-V and V-V-V compounds as a syntactic consequence provoked by the lexical difference between the speaker-oriented directional verbal morpheme and the general directional verbal morpheme without any further stipulation. 4.6. CONCLUSION In this chapter, I have discussed the three major types of directional verb compounds in Chinese and also described their basic and crucial properties such as the types of their component verbal morphemes, the semantic co-occurrence restrictions on these component morphemes, the distribution of their objects, and the aspect marker attached to the directional verb compound. In order to 299 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. account for the formation of the Chinese directional verb compounds, to incorporate the complex thematic relations between them and their arguments, and to further capture the semantic co-occurrence restrictions on the component morphemes of the verb compounds, I have adapted and also applied the morpho-syntactic analysis proposed in Chapter 3 to the Chinese directional verb compounds. Under this morpho-syntactic analysis, the formation of the Chinese directional verb compound construction simply becomes the side effect of verb-raising and NP-movement without using any particular lexical rules, which is exactly on a par and consistent with the morpho-syntactic analysis of the Chinese BA-construction and the Chinese resultative verb compound construction presented in Chapter 2 and Chapter 3. Like the derivation of the BA-constructions and the resultative verb compound constructions shown in Chapters 2 and 3, the verb-raising and NP-movement involved in the derivation of the directional verb compound construction are morphologically motivated, because the inflectional features of these directional verb compounds and the Case feature of their NP arguments have to be checked against AGR and ASP (AGRq). Besides, the verb-raising and NP- movement are constrained by the principles of universal grammar like the Minimal Link Condition and the Shortest Movement Condition. As an important consequence of this 300 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. morpho-syntactic analysis, the distribution of the object of the directional verb compounds is naturally captured as a side effect of verb-raising provoked by the lexical difference between the speaker-oriented directional verb morpheme and general directional verb morpheme without any further stipulation. 301 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES Abney, Steven (1987). The English Noun Phrase in its Sentential Aspect. Doctoral dissertation. Cambridge, MA: MIT. Aoun, Joseph (1980). Feature transportation and the Move- a convention. Ms., MIT, Cambridge, MA. Aoun, Joseph and D. Sportiche (1983). On the formal theory of government. Linguistic Review 2, 211-36. Aoun, Joseph, N. Hornstein, D. Lightfoot and A. Weinberg (1987). Two types of locality. Linguistic Inquiry 18, 537-578. Aoun, Joseph and A. Li (1993). Wh-elements in-situ: syntax or LF? Linguistic Inquiry 24, 199-238. Baker, Mark (1988). Incorporation: A Theory of Grammati cal Function Change. Chicago: University of Chicago Press. Barwise, J. and R. Cooper (1981). Generalized quantifiers and natural language. Linguistics and Philosophy 4, 159-219. Belletti, Adriana and L. Rizzi (1988). Psych-verbs and theta-theory. Natural Language and Linguistic Theory 6, 291-352. Bennett, Paul (1981). The evolution of passive and disposal sentences. Journal of Chinese Linguistics 9, 61-89. Bowers, John (1988). A structural theory of predication. Ms., Cornell University, Ithaca, NY. Burizo, L. (1986). Italian Syntax: A Government Binding Approach. Dordrecht: Reidel. Carlson, N. G. (1977). Reference to Kinds in English. Doctoral dissertation. Amherst, MA: University of Massachusetts. 302 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chao, Yuen-Ren (1968). A Grammar of Spoken Chinese. Berkeley: University of California Press. Cheng, Lisa (1986). Clause Structures in Mandarin Chinese. M.A. thesis. Toronto: University of Toronto. Cheng, Lisa (1988). Aspects of the BA-construction. Ms., MIT, Cambridge, MA. Cheng, Lisa (1989). Transitivity alternations in Mandarin Chinese. In M. Cheuig and T. Ernst (eds). Proceedings of the Third Ohio State University Conference on Chinese Linguistics, 81-94. Bloomington, IN: Indiana University Linguistics Club. Cheng, Lisa (1992). Resultative Compounds and Lexical Relational Structures. Proceedings of the Third International Symposium on Chinese Languages and Linguistics, 81-97. Taiwan: Tsing Hua University. Cheng, Lisa and E. Ritter (1987). A small clause analysis of inalienable possession in Mandarin and French. In J. Blevins and J. Carter (eds). Proceedings of North Eastern Linguistic Society 18, 65-78. Amherst, MA: University of Massachusetts. Cheung, Kung-Nin (1973). A comparative study in Chinese grammars: the BA construction. Journal of Chinese Linguistics 3, 343-382. Chomsky, Noam (1981). Lectures on Government and Binding. Dordrecht: Foris. Chomsky, Noam (1982) Some Concepts and Consequences of the Theory of Government and Binding. Cambridge: MIT Press. Chomsky, Noam (1986a). Knowledge of Language: Its Nature, Origin, and Use. New York: Praeger. Chomsky, Noam (1986b). Barriers. Cambridge: MIT Press. Chomsky, Noam (1991). Some notes on economy of derivation and representation. In R. Freidin (ed). Principles and Parameters in Comparative Grammar, 417-454. Cambridge: MIT Press. 303 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chomsky, Nocim (1993). A minimalist program for linguistic theory. In K. Hale and J. Keyser (eds). The View from Building 20, 1-52. Cambridge: MIT press. Chomsky, Noam (1994). Bare phrase structure. MIT àccasional Papers in Linguistics 5. Cambridge, MA. Chomsky, Noam and H. Lasnik (1991). Principles and para meters theory. To appear in J. Jacobs, A. Stechow, W. Sternefeld and T. Vennemann (eds). Syntax: An International Handbook of Contemporary Research. Berlin: Walter De Gruyter. Chu, Chauncey (1976). Conceptual wholeness and retained object. Journal of Chinese Linguistics 4, 14-23. Cinque, Guglielmo (1980). On extraction from NP in Italian. Journal of Italian Linguistics 5, 47-99. DaCruz, M. (1992). Restrictions sémantiques dans les constructions serielles avec so 'prendre' et hEn 'Tenir' en fongbe. Journal of West African Languages 22, 115-133. Dai, Xiang-Ling (1992). Inflectional Morphology in Chinese. Doctoral dissertation. Columbus, OH: Ohio State University. Déchaine, R.-M. (1991). Serial verb constructions. Ms., University of Massachusetts, Amherst, MA. Diesing, Molly (1992). Indefinites. Cambridge: MIT Press. Ding, Sheng-Shu et al. (1961). Hanyu Yufa Wenti Jianghua (Lectures on contemporary Chinese grammar). Beijing: Chinese Linguistics Press. Dowty, David (1979). Word Meaning and Montague Grammar: the Semantics of Verbs and Times in Generative Semantics and Montague's PTQ. Dordrecht: Reidel. Fagan, S. M. B. (1988). The English middle. Linguistic Inquiry 19, 181-203. Fan, Xiao (1988). On resultative verb compounds. In Hu, S.-B. (ed). Grammar Research and Inquiry, 60-77. Beijing: Beijing University Press. 304 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Fiengo, Robert and J. Higginbotham (1981). Opacity in NP. Linguistic Analysis 1, 395-421. Frei, Henry (1956). The ergative construction in Chinese: theory of Pekinese BA. GENGO KENKYU 31, 22-115. Goodall, Grant (1987). On the argument structure and L- marking with Mandarin Chinese ba. In J. McDonough and B. Plunkett (eds). Proceedings of North Eastern Linguistic Society 17, 232-242. Amherst, MA: University of Massachusetts. Grimshaw, Jane (1990). Argument Structure. Cambridge: MIT Press. Gu, Yang (1992). The Syntax of Resultative and Causative Compounds in Chinese. Doctoral dissertation. Ithaca, New York: Cornell University. Gueron, J. (1985). Inalienable possession, PRO-inclusion and lexical chains, in J. Gueron, H. Obenauer and J.-Y. Pollock (eds). Grammatical Representation, 43- 86. Dordrecht: Foris. Gueron, J. (1991). La possession inalienable et l'aspect locatif. Ms., Université Paris X, Paris. Hale, Kenneth and J. Keyser (1993). On argument structure and the lexical expression of syntactic relations. In K. Hale and J. Keyser (eds). The View from Building 20, 53-110. Cambridge: MIT press. Hashimoto, Anne (1971). Mandarin syntactic structures. Unicorn 5. Higginbotham, James (1985). On semantics. Linguistic Inquiry 16, 547-593. Hoekstra, T.-A. (1988). Small clause results. Lingua 74, 101-139. Hoekstra, T.-A. (1990). Small clause, complex predicates and binarity. Ms., University of Leiden, Leiden. Huang, James (1982). Logical Relations in Chinese and the Theory of Grammar. Doctoral dissertation. Cambridge, MA: MIT. 305 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Huang, James (1984). On the distribution and reference of empty pronouns. Linguistic Inquiry 15, 531-574. Huang, James (1987). Existential sentences in Chinese and (in)definiteness. In E. J. Reuland and G. B. Meulen (eds). The Representation of (in) de finiteness, 226- 253. Cambridge: MIT press. Huang, James (1988). WO PAO DE KUÂI and Chinese phrase structure. Language 64, 274-311. Huang, James (1989). Pro-drop in Chinese: a generalized control theory. In 0. Jaeggli and K. Safir (eds). The Null Subject Parameter, 185-214. Dordrecht: Kluwer. Huang, James (1992). Complex predicates in control. In J. Higginbotham and R. Larson (eds). Control and Grammar, 109-147. Dordrecht: Kluwer. Jackendoff, Ray (1972). Semantic Interpretation in Generative Grammar. Cambridge : MIT Press. Kayne, Richard (1984). Connectedness and Binary Branching. Dordrecht: Foris. Koopman, Hilda (1984). The Syntax of Verbs: from Verb Movement in the Kru Languages to Universal Grammar. Dordrecht: Foris. Koopman, Hilda and D. Sportiche (1988). Subjects. Ms., University of California, Los Angeles, CA. Kratzer, A. (1988). Stage-level and individual-level predicates. In E. Bach, A. Kratzer and B. Partee (eds). Papers on Quantification. Amherst, MA: University of Massachusetts. Kuroda, S.-Y. (1988). Whether we agree or not: A compara tive syntax of English and Japanese. Lingvisticae Investigationes 12, 1-47. Laka, M. I. (1990). Negation in Syntax: on the Nature of Functional Categories and Projections. Doctoral dissertation. Cambridge, MA: MIT. Larson, Richard (1988). On the double object construction. Linguistic Inquiry 19, 335-391. 306 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Larson.. Richard (1990). Promise and the theory of control. Linguistic Inquiry 22, 103-140. Li, Charles and S. Thompson (1981). Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Li, Ying-Che (1970). An Investigation of Case in Chinese Grammar. Doctoral dissertation. Ann Arbor, MI: University of Michigan. Li, Ying-Che (1974). What does 'disposal' mean? features of the verb and noun in Chinese. Journal of Chinese Linguistics 2, 200-218. Li, Ya-fei (1990). On V-V compounds in Chinese. Natural Language and Linguistic Theory 8, 177-207. Li, Ya-fei (1993). Thematic hierarchy and causativity. Ms., Cornell University, Ithaca, NY. Li, Audrey (1985). Abstract Case in Chinese. Doctoral dissertation. Los Angeles, CA: University of Southern California. Li, Audrey (1990). Order and Constituency in Mandarin Chinese. Dordrecht: Kluwer. Li, Zi-Yun (1990). The predication of resultative verbs. Zhongguo Yuwen (Chinese Philology) 5, 338-343. Lieber, R. (1983). Argument Linking and compounds in English. Linguistic Inquiry 14, 251-285. Liu, Feng-Hsi (1992). Aspect and the BA sentences in Chinese. Ms., University of Arizona, Tucson, AZ. Liu, Shu-Xin (1987). On the scope of directional verbs. In Hu, S.-B. (ed). Grammar Research and Inquiry, 203-213. Beijing: Beijing University Press. LÜ, Shu-Xiang (1955). The Papers on the Chinese Grammar. Beijing: Science Press. LÜ, Shu-Xiang et al. (1984). Xiandai Hanyu Babai Ci (Eight hundred words of contemporary Chinese). Beijing: Commercial Press. 307 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lu, Jian-Ming and Z. Ma (1985). Xiandax Hanyu Xu Ci Shianren (Papers on modern Chinese functional categories). Beijing: Beijing University Press, Lu, John (1973). The verb-verb construction with a directional complement in Mandarin. Journal of Chinese Linguistics 1, 239-255. Lu, John (1977). Resultative Verb Compounds vs. Directional Verb Compounds in Mandarin. Journal of Chinese Linguistics 5, 277-313. Ma, Xi-Wen (1987). Certain types of resultative verb compounds. Zhongguo Yuwen (Chinese Philology) 6, 424-441. Mei, Kuan (1978). The BA sentences in modern Chinese. Bulletin of the College of Fine Arts 27, 145-180. Taiwan: Taiwan University. Ouhalla, Jamal (1990). Sentential negation, relativised minimality and the aspectual status of auxiliaries. Linguistic Review 7, 183-231. Ouhalla, Jamal (1991). Functional categories and parame tric variation. Ms., Queen Mary & Westfield College, London University, London. Perlmutter, David (1978). Impersonal passives and the unaccusative hypothesis. In J. Jaeger et al. (eds). Proceedings of the Fourth Meeting of the Berkeley Linguistics Society. Berkeley, CA: University of California. Pollock, Jean-Yves (1989). Verb movement, universal grammar and the structure of IP. Linguistic Inquiry 20, 365-424. Quirk, Randolph, S. Greenbaum, G. Leech and J. Svartvik (1985). A Comprehensive grammar of the English language. London: Longman. Rizzi, Luigi (1990). Relativized Minimality. Cambridge: MIT Press. Rouveret A. and J.-R. Vergnaud (1980). Specifying Reference to the subject: French causatives and conditions on representations. Linguistic Inquiry 11, 97-202. 308 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sijbesma, Rint (1992). Causatives and accomplishments: The case of Chinese ba. Doctoral dissertation. Leiden: University of Leiden. Sijbesma, Rint (1994). Comparing Chinese ba and na to TAKE in Kwa. A talk at The Sixth North American Conference on Chinese Linguistics, University of Southern California, Los Angeles, CA. Smith, Carlota (1991). The parameter of aspect. Dordrecht: Kluwer. Stewart, J.-M. (1963). Some restrictions on objects in Twi. Journal of African Languages 2, 145-149. Stowell, Tim (1981). Origins of phrase structure. Doctoral dissertation. Cambridge, MA: MIT. Tai, James (1984). Verbs and times in Chinese: Vendler's four categories. In D. Testen, V. Mishra and J. Drogo (eds). Papers from the Parasession on Lexical Semantics. Chicago: Chicago Linguistic Society. Tang, Jane (1990). A note on the DP analysis of the Chinese noun phrase. Linguistics 28, 337-354. Tellier, C. (1990). Underived nominals and the projection principle: inherent possessors. In J. Carter, R.-M. Dechaine, B. Phillip & T. Sherer (eds). Proceedings of North Eastern Linguistic Society 20, 472-486. Amherst, MA: University of Massachusetts. Teng, Shou-Hsin (1971). A Semantic Study of Transitivity Relations in Chinese. Doctoral dissertation. Berkeley, CA: University of California. Teng, Shou-Hsin (1973). Negation and aspects in Chinese. Journal of Chinese Linguistics 1, 14-37. Teng, Shou-Hsin (1974). Negation in Chinese. Journal of Chinese Linguistics 2, 125-140. Thompson, Sandra (1973). Transitivity and the BA- construction in Mandarin Chinese. Journal of Chinese Linguistics 1, 208-221. Travis, Lisa (1984). Parameters and effects of word order variation. Cambridge, MA: MIT Doctoral dissertation. 309 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Tsao, Feng-Fu (1987). A topic-comment approach to the BA- construction. Journal of Chinese Linguistics 15, 1-52. Vendler, Z. (1967) Linguistics in Philosophy. Ithaca, NY: Cornell University Press. Vergnaud, Jean-Roger and M.-L. Zubizarreta (1993). The definite determiner & the inalienable constructions in French and in English. Linguistic Inquiry 23, 595-652. Verkuyl, K. (1972). On the Compositional Nature of Aspects. Dordrecht: Foris. Wagner, D. (1985). Objects in Gokana. Studies in African Linguistics, supplement 9, 304-308. Wang, Huan (1987). Ba zi ju he bei zi ju (BA- and BEI- constructions ). Shanghai: Shanghai Education Press. Wang, Mian-Qing (1987). Transitivity and the ba- construction in Mandarin. Doctoral dissertation. Boston, MA: Boston University. Wang, Li (1954). A Chinese Modern Grammar, Beijing: Zhonghua Press. Wang, Li (1958). Hanyu Shigao (2nd Vol). Beijing: Science Press. Wang, Peter (1970). A transformational approach to Chinese BA and BEI. Doctoral dissertation. Austin, TX: University of Texas. Wang, William (1963). Some syntactic rules for Mandarin. Proceedings of the Ninth international Congress of Linguistics, 192-202. Wang, William (1965). Two aspect markers in Mandarin. Language 41, 457-470. Wu, Xiu-Zhi (1993). Psychological predicates in Chinese. M.A. thesis. Taiwan: Taiwan Normal University. Wu, Xiu-Zhi (1994). Causative constructions in Mandarin Chinese. A talk given at The Sixth North American Conference on Chinese Linguistics, University of Southern California, Los Angeles, CA. 310 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Yoon, James (1990). Theta theory and the grammar of inalienable possession constructions- In J. Carter, R.-M. Dechaine, B. Phillip and T. Sherer (eds). Proceedings of North Eastern Linguistic Society 20, 502-516. Amherst, MA: University of Massachusetts. Zan, Ren-Fung (1989), The predication of resultative verb compounds. Zhongguo Yuwen (Chinese Philology) 2, 105-111. Zhang, Bai-Jiang (1991). On the Word-Order of Directional Verb Compound with Object. Zhongguo Yuwen (Chinese Philology) 3, 183-191. Zhu, Min-Che (1957). Lun chuqi chuzhishi. Yuyanxue Lun- cong (Papers on Linguistics) 1, 17-33. Zou, Ke (1992). Specificity effects in the Chinese NP- movement and WH-extraction. In J. Nevis and V. Samiian (eds). Proceedings of West Conference on Linguistics 5, 380-391. Fresno: California State University. Zou, Ke (1993). The syntax of the Chinese BA- construction. Linguistics 31, 715-736. 311 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The syntax of focus and topic in Mandarin Chinese

PDF

Syntactic structures in nominals: A comparative study of Spanish and Southern Quechua

PDF

On the nature of particles in Japanese and its theoretical implications

PDF

The phonemics and morphology of Hokkaido Ainu

PDF

Intra-lexical noun-verb dissociations: Evidence from Chinese aphasia

PDF

On the representation and licensing of Q and Q-dependents

PDF

The syntax of clitic doubling in modern Greek

PDF

In the event of focus

PDF

Language and context: toward a model of psycholinguistic competence

PDF

The accidence and syntax in john Wallis' 1653 "grammatica linguae anglicanae": a translation and a commentary on its alleged relationship to the 1660 port-royal "grammaire generale et raisonnee"

PDF

Verb Second, mood and operator licensing.

PDF

Correspondence and faithfulness constraints in optimality theory: A study of Korean consonantal phonology

PDF

Evaluating heaviness: Relative weight in the spoken production of heavy-NP shift

PDF

A comparative study of focus constructions

PDF

Representation of focus and presupposition in Japanese

PDF

Rhetorical abstraction as a facet of expected response: A structural equation modeling analysis

PDF

Second language acquisition of English double object construction by Korean speakers.

PDF

Metaphor and metonymy: A comparative study of Chinese and Western poetics

PDF

Preview of the vanishing hero: a study of the protagonists in Jacobean drama

PDF

Frequency in sentence comprehension

Asset Metadata

Creator Zou, Ke (author)

Core Title The syntax of the Chinese BA-constructions and verb compounds: A morphosyntactic analysis

Degree Doctor of Philosophy

Degree Program Linguistics

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

Tag language, linguistics,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-481164

Unique identifier UC11352123

Identifier 9617013.pdf (filename),usctheses-c17-481164 (legacy record id)

Legacy Identifier 9617013-0.pdf

Dmrecord 481164

Document Type Dissertation

Rights Zou, Ke

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