Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Causativity in Korean: Syntactic causative, control, and purpose
(USC Thesis Other)
Causativity in Korean: Syntactic causative, control, and purpose
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UM3 films the tdirectly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send U M I a complete manuscript and there are m issing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., m ap s, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher 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. TT-a /sr'vr A Bell & Howell Information Company >00 N o rth Zeeb Road, Ann Arbor Ma 48106-1346 U S A 7 i t rt< i ^ 9 0 0 / * 7 1 A / V I T ( W A V U W Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CAUSAilVITY IK KOREAN: SYNTACTIC CAUSATIVE, CONTROL, AND PURPOSE by Kyungan Kins 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 / v - 5 December 1995 Copyright 1995 Kyungan Kim Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9625252 UMI Microform 9625252 Copyright 1996, by UMI Company, AH rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Hoad 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 ........ KYUNGAN.KIM................. 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 PHILO SO PHY Dean Date ^Qy- ember 2, 1995 DISSERTATION COMMITTEE 3 A IP .p j.-V L ;l^ v p 'V . ... . .................... Chairperson Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I dedicate this dissertation to our heavenly Father, God the Almighty Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Acknowledgments As I look back on the years of my stay at USC,- I realize that I have had so many valuable experiences, academically and personally. Even though I had some minimal linguistic training before I came to USC, I must say that my genuine interest in studying languages blossomed through the courses I have taken with my teachers in the graduate program of the linguistic department at USC, and particularly through my encounter with a great teacher, Bernard Comrie. I can always tell people without hesitation that I learned all the basics to become a language expert at USC. My teachers are Elinor Ochs, Joseph Aoun, Debby Svhlindwein, Douglas Biber, Jean-Roger Vergnaud, James Gee, Elaine Andersen, Bernard Comrie, Jack Hawkins, Ed Finegan, Mark Seidenberg, Hajime Hoji, Maria Polinsky, Roger Woodard, and Audrey Li. I also thank all my language teachers at University of Washington, USC, Universite de Dijon, and Casa de America. I first thank from the bottom of my heart Bernard Comrie, who is my thesis adviser, teacher, and friend. I must confess that I would not have come this far without his presence in my life. He is the one who has sustained me whenever I had doubts. He is the one who has always been patient with his "lazy" student, and encouraged him to hang on and go on. I will do my best to become a good language expert whom he can proudly tell people about. I am so much indebted to you. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Bernard. I will always remember your work ethic and your genuine care and understanding for your students. I will try to be like you in many ways. I know it is an awfully hard task, but I will try. Please watch over me, and be my guiding light. I found a true and special friend in you. I am also very grateful to my qualifying committee members: Bernard Comrie, Jack Hawkins, Roger Woodard, Maria Polinsky, and Namkil Kim. I also thank my dissertation committee members: Bernard Comrie, Roger Woodard, and Jack Hawkins. They thoroughly read my dissertation, and made so many valuable comments. Thank you again. I also thank Roger Woodard and Ed Finegan for their friendship and understanding. I had so much comfort from them. Thank you, Roger and Ed. I have so many friends and family to give credit to for helping me complete one of the most momentous journeys in my life. I will mention some of them here. First, I thank my mother for believing in me and waiting for me all these years. I know you have missed me so much, Mom. I have missed you, too. I am on my way! My mother's sacrifices and prayers are probably the most important factor that made me what 1 am today. I also thank my father and my brothers and sisters, and relatives, especially my friend and cousin, Jungwoo Shin, for being a special part of my life. Next I want to thank Jack and Maggie Dcdds for their love, friendship, and financial support. They have always treated me as their son. Without their financial support many of the wonderful experiences I iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. enjoyed could not have happened. They made my life here a blessed one. Thank you, Jack and Maggie. I also want to single out my dear friend, Rickard Shive, who has proofread this dissertation and its earlier versions several times, and was always there for me as a genuine friend whenever I was down and depressed. I cannot thank you enough, Richard. Thank you so much. I also thank you, Richard for being my main English language consultant. I will always cherish your friendship. I also acknowledge the financial support I received from Seoul Rotary club, and Olympic Lyons club. I particularly thank Walter W. Park and his wife for their care and support. The scholarship Seoul Rotary club provided me helped me for the entire first year of my study. I sincerely thank Namkil Kim for giving me the great opportunity to teach Korean language classes for four and a half years. I acknowledge that the teaching assistantship I was given was the biggest financial support I received for my study. I also want to thank and remember all my students in my Korean classes. I want them to know that they enriched my life, and hope for the best for each and every one of them. I also want to thank my friends, George Totten, Lilia Li, Doug Elliott, Bill Manson, John Swanson, Daein Kang, Joseph Sungho Choi, Alex Hyungkyu Kwon, David Sunghyun Kwon, Siegfrid and Margot Eisenmann, Kilhwan Ko, Wayne Whiteman, Hazel Bates, Laura Rreiter, Soonyoung and Soonja Kang, the late Rev. John Cole, Sharon Tool, all the friends I have had in Linguistics and East Asian Languages and v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cultures, and many others for sharing their lives with me. I also thank my teacher, Yonggol Lee, for being always there for me. I also thank all my teachers at Hankuk University of Foreign Studies. Finally I dedicate this first fruit of my intelligence to my heavenly Father, the God who has sustained me all through my life, and guided me all along. Father, I will always try to do my best in bringing glory to You! Please make me a person after Your heart and allow me to gain wisdom and knowledge to guide and council others not to lose love and hope for You and other beings. Thank You for this fruit, and I dedicate it to You. You are the only one who truly deserves it and more. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table of Contents DEDICATION...................................................... ii ACKNOWLEDGMENTS........................................... „ . iii ABSTRACT. .................. ........ .................. ix ABBREVIATIONS .................... ............................. xi Chapter 1: Introduction. .......................... 1 A. Objectives of this study.............................. 1 B. Main points of this study................. ........... 3 C. Structure of this study............................... 7 D. Significance of this study............................ 12 Chapter 2: Subjecthood tests and the verb 'hata (do)'...... 15 A. Subjecthood in Korean.................................. 15 A.I. Honorification. .................................. 15 A. 2. 'Kkeyse' marking.................................. 19 A. 3. 'Myense' construction............................. 21 A. 4. Coordination...................................... 27 B. The verb 'hata (do) '.................................. 28 B.l. Introduction....................................... 28 B.2. Types of 'hata'................................... 29 B.3. Light verb construction.......................... 33 B.4. 'Hata' in negation............................ . 43 B.5. 'Hata' as a main verb............................. 47 B.6. Discussion........................................ 53 B.7. Conclusion......................................... 56 Chapter 3: Syntactic causative in Korean.................. 57 A. Introduction........................................... 57 B. Literature review and a critique of the monoclausal analysis by Gerdts (1990)................ 62 B.l. Literature review................................. 62 B.2 - a critique of the monoclausal analysis by Gerdts (1990)........................ 80 C. Evidence for fciciausality .................. ......... SO C.l. NPs, pronouns, and reflexives.................. .. 90 C.1.1. NPs............................................ 90 C.l.2. Pronoun and reflexives........................ 105 C.2. Quantifiers. .................................. 113 C.3. Deletion of redundant elements.................... 125 C.4. Negation........................................... 132 C.5. Position of adverbials, PFs, and subordinate clauses..................... 139 D. Causative in Korean and Romance languages....... 152 E. Conclusion. .................... ..................... . 155 vi-i Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 4: Control in Korean 157 A. Subject control....................... 158 A.I. 'Nolyekhata (try)'-type........................... 158 A.2. 'Yaksokhata (promise)'—type...................... 173 A.3. 'Wenhata (want)'-type............................. 189 B. Object control.................... 196 B.l. Previous research on object control in Korean................ 198 B.2. Our analysis of obligatory object control............................ 205 B.2.1. Overt element in the embedded subject position...................... 206 B.2.2. Empty embedded subject......................... 211 B.2.3. Optionalitv of the main clause object.......................................... 215 C. Semantics of the verb and complementizer. , ........... 219 C.l. Interplay of the semantics of the verb and the complementizers 'tolok' and 'lyeko'...... 228 D. Change of controller.................................... 232 D.l. The English facts............................. 232 D.2. The Korean facts.................................... 238 E. Comparison of the syntactic causative and object control...................................... 242 Chapter 5: Syntax and semantics of 'cause' and 'effect'............................... 253 A. Semantics of 'cause' and 'effect'...................... 257 B. Syntax of 'cause' and 'effect'......................... 265 B.l. Korean................................ 266 B.2. English............................................. 267 B.3. Thai .............................. 270 B.4. Japanese........................... 273 C. Scale of match between syntax and semantics of 'cause' and 'effect'..................... 275 Chapter 6: Conclusion. = = . = = 278 Footnotes.................................. 282 References............................................... 291 viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT The main point of this study is to demonstrate that the syntactic causative, object control, and purpose clause sentences all semantically contain 'causation'. For this study we define 'causation' as in (1). (1) Definition of Causation (a) The causing event A encoded in the main clause occurred to bring about the desired event B encoded in the embedded clause. (b) The causer and the causee are not equivalent. With this definition, as long as the desired event encoded in the embedded clause is intended to occur by the causing event encoded in the main clause, whether the desired event is actually realized or not, the definition (1) is satisfied. In the Korean syntactic causative literature, one of the main issues has been biclausality vs. monoclausality of the syntactic causative. We argue that no matter what marker the causee is marked with the syntactic causative sentence is hielaUsal at every level of derivation. We argue that when the causee is marked with the accusative marker or the dative marker there is a zero anaphor in the embedded subject position, and the nominative causee is the embedded subject at every level of derivation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. We claim that in Korean the so-called control structure, both subject control and object control, is mainly pragmatic rather than syntactic in the sense that the referent for the zero anaphor in the embedded subject position is not always found within the main clause. Finally, we claim that the syntactic causati'/e, object control, and the purpose clause sentences all contain 'causation' in the semantic sense. But syntactically they may be expressed differently in different languages. In Korean and Thai, for instance, they all share the same syntactic structure. In Thai, however, the syntactic causative can be syntactically expressed in more than one way. In English, three syntactic structures, to-infinitival, 'so that' clause and bare-infinitive, are used, whereas in Japanese the syntactic causative uses a morphological process that forms a monoclausal sentence at the final level, while the object control sentence and the 'so that' sentence share a biclausal structure, using the complementizer 'vooni'. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABBREVIATIONS ACC Accusative CAUSE Causative morpheme CL Classifier COMP Complementizer CON Conjunctor DAT Dative DAT(HON) Dative (Honorific Form) DO Direct Object FUT Future Tense GEN Genitive HON Honorific Morpheme IMP Imperfactive IND Indicative 10 Indirect Object LOC Locative NEG Negation Morpheme NM Nominalizer NOM Nominative NOM(HON) Nominative (Honorific Form) PASS Passive PL Plural POL Polite Ending PRES Present Tense PRF Perfactive n n A t? J C I W i Progressive PST Past Tense QUES Question Particle SUG Suggestive TOP Topic VOL Volitional Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 1. Introduction (A). Objectives of this Study Causation has long been studied cross-linguistically as one of the moiii issues in Ixnguxstics, mainly focusing on lexical and syntactic causatives. In Korean, likewise, causativity— lexical and syntactic— has equally enjoyed attention from many linguists. The main issue in the syntactic causative in Korean has been whether the causative sentence is monoclausal or bi- clausal. It has been commonly assumed, according to proponents of the monoclausal analysis, that there are two clauses at one level, and in the process of derivation they merge into one clause and syntactically function as a single clause at the final level. We claim in this thesis, however, providing ample examples that syntactic causatives remain biclausal at every level of derivation, regardless of which marker the causee a* r* « ia o • The second objective is to claim that the sentences with verbs commonly known as obligatory control verbs like 'yaksok- hata (promise)/, and 'seltukhata (persuade)! do not behave as they do in many other languages. For instance, the embedded subject position does not have to obligatorily remain empty. It can be filled with an overt element. This is quite diffe rent from languages like English and Chinese which do not allow any overt element to occur in the embedded subject 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. position. We claim that in Korean when an overt element appears in the embedded subject position it does not have to be coreferential with either the subject or the object. We further claim that even when the embedded subject position stays empty, it may not necessarily be coreferential with a main clause argument, although there 13 a strong preference for subject control in the 'promise'-type, and for object control in the 'persuade'-type. The third objective and main goal for this study is to claim that the structure of the sentence with verbs tradi tionally known as obligatory object control verbs is equi valent to the structure of the syntactic causative sentence. This is due to the fact that both phenomena encode causation in the sense that the event described in the main clause is the causing event which occurs to bring about the desired event depicted in the embedded clause. This structural simi larity between the syntactic causative and the obligatory • * /N M 4* 1 ^ « • « « ^ « « n* < * « » A « I* < 9 ^ I * in •• a i 9 W J b u w « * u A . G j u i a j r u c c u n x u i i t - u c a u i u w L U i c which can best be translated as the 'so that' structure in English or the 'purpose' clause structure with different agents for the event in the main clause and the event in the embedded clause. We also show that the syntax and semantics of causative sentences with a different causer and causee may match as in the case of Korean and Thai, or mismatch as in English, or be somewhere in the middle of the continuum as in Japanese. 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (B). Main points of this Study Causation may be defined in many ways. In this study, we mainly focus on the causative relation where the causer and the causee are not equivalent. We define causation as following: (1) Definition of Causation (a) The causing event A encoded in the main clause occurred ho bring about the desired event B encoded in the embedded clause. (b) The causer and the causee are not equivalent. With this definition of causation, we argue that in Korean the syntactic causative, so-called object control, and the equi valent of the English 'so that' clause all contain the rela tion of 'cause' and 'effect' which we are interested in for this study. We may unite (a) and (b) as following: / n \ w ; i. c a . £ P 1/C A A U A W J.V U W J . W Q U aO L J . W 1 1 A initiated the event A' to make B perform the event B'. This definition shows that there are two events— the causxng one and the desired one— and there are different causer and causee. (3) Emeni—ka Mary—eykey [0 ku umshik—ul mek]—tolok mother-NOM -DAT that food-ACC eat-COMP hay-ess-ta. do-PST-IND 'Mother made Mary eat that food.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. According to the above definition of causation, 'emeni (mother)', who is the initiator of the event described in the main clause, initiated an event which was intended to make the event of ‘ "Mary's eating the food' depicted in the embedded clause happen. The causative verbs like 'make', 'cause' do not clearly tell us what the causing event is. whether the causing event is clearly expressed or not in the semantic sense, it is certain that there is an event which occurred to cause the de sired event expressed in the embedded clause. Unlike the English translation, the Korean counterpart in (1) does not require the entaiiment which says that the desired event actually happened as the result of the causing event caused by the causer 'emeni (mother)'. Thus both of the following sen tences can be said after the sentence (3). (4) (a) Kulayse Mary-ka ku umshik-ul mek-ess-ta. so -NOM that food—ACC eat-PST-IND 'So Mary ate that food.' (b) Kulena Mary-ka ku umshik-ul ani-mek-ess-ta. but —NOM that food—ACC NEQ—eat—PST—IND 'But Mary did not eat that food.' The English translation of (3) has the entaiiment of (4a) . Thus the translation may not be that compatible although it is the closest one. We may also use a translation like 'Mother did something to make Mary eat that food'. As the above exam ples show, and according to the definition we have given above, the desired event does not have to be realized as long Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. as the causing event occurred in order to bring about the desired event. The object control sentences with verbs like 'seltukhata (persuade)' and 'myenglyenghata (order)' can be construed to contain causation according to the definition we have in the above since the events expressed by these verbs may be the causing events which are used to bring about the events expressed in the embedded clause. The object control verbs give a little better idea of what the causing event might be than the causative verbs like 'make' or 'cause' do. (5) Sacang-i ku sawen-eykey [0 ku il-ul ha]-tolok president-NOM the employee-DAT the work-ACC do-COMP myenglyenghay-ess-ta. order-PST-IND 'The president ordered the employee to do the work.' The reason why the president ordered the employee to do the event described in the embedded clause is to bring about that event. In (5), the causer, the president, is different from the causee, the employee, and the causing event and the de sired event are expressed in two separate clauses. Thus this sentence meets all the conditions of the definition of causa tion which we will focus on in this study. As we claim that the syntactic causative sentence as in (3) and object control sentence as in (5) can be identified as causation in the se mantic sense, we will also compare the structural similarities of these two phenomena. We will show that they have basically the same syntactic structure. 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The nest sentence, which is similar to the English 'so that' sentence or 'purpose clause' sentence with the main clause subject and the embedded subject being different, also fits the conditions of the definition of causation we will look at in this study. (6) John-i Mary-eykey [0 ku kes-ul sa]-tolok ton-ul -NOM -DAT that-ACC buy-COMP money-ACC cwu-ess-ta. give-PST-IND 'John gave money to Mary so that (she/ .. ) could buy that.' In this sentence, the causing event in the main clause is 'John's giving money to Mary' which occurred to bring about the desired event of 'Mary's buying that'. This sentence sa tisfies all the conditions of the definition of causation. First, there is a causing event which happened to bring about the desired event. Second, there are separate causer and causee. Third, the causing event and the desired event are encoded in two separate clauses. The causing event in (6) is obvious, tho nno fho syntsctic G3izs3.vd.vc sentence (3) or the so-called object control sentence (5), with the desired event in ail these sentences being semantically clear. It is also noteworthy that it is not necessary for there to be an empty embedded subject slot in a 'purpose clause' sentence. It is a matter of the subcategorization of the main verb. When the main verb is intransitive like 'cata (sleep)', for instan ce, there is no object which will be coreferential with the 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. empty embedded subject. Thus the embedded subject position may be filled with an overt element. (7) John-i [Mary-ka ku chayk-ul ilk]-tolok ca-ass-ta. -KOM -NOM the book-ACC read-COMP sieep-PST-IND 'John slept so that Mary could read the book.' In this case, the c a u s i n g event is 'John's sleeping' which occurred to bring about the desired event, 'Mary's reading the book'. Since there is no argument which may be coreferential with the embedded subject, the embedded subject position is filled with a name 'Mary'. The embedded clauses in (3) and (5) are complements of the main verb, whereas in (6) and (7) they are adjuncts. The desired event is a complement in the case of the syntactic causative and object control, and an adjunct in the case of the 'purpose clause' sentence. Semantically speaking, however, in all these sentences the main clause encodes the causing event while the embedded clause encodes the desired event. (C). Structure of this Study We claim in this study that in Korean the syntactic struc ture of the syntactic causative and the object control struc ture, and the 'purpose clause' sentence with different causer and causee are equivalent. While there is no dispute about the fciclausality of the control structure, there have been, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. though, different opinions on whether the syntactic causative is monoclausal or biclausai. We will argue that it is biclau- sal at every level of derivation. By arguing for the biclausa- lity of the syntactic causative we can claim that the syn tactic causative and the object control structure, and the 'purpose clause' sentence with different causer and causee share the same syntactic structure with the same complemen tizer, 'tolok' or 'key'. In Chapter 2, we give several 'subjecthood' tests in order to demonstrate that in the syntactic causative in Korean the nominative causee is the subject of the embedded clause. This is important because those who argue for a monoclausal analy sis claim that the nominative causee is not the subject of the embedded clause, but a non-term. By demonstrating that the nominative causee is the final subject of the embedded clause while there is a separate main clause, we claim that the syntactic causative with nominative causee is biclausai at every level of derrvatzon. We also coiuiiient on tne dirferent usages of the verb 'hata (do)' which is used as the causative verb in Korean. 'Hata' may be used as a 'pro' verb, causative verb, light verb, etc. Then we put all the usages of the verb 'hata' into two, one as heavy verb and the other as light verb. In Chapter 3, we deal with the issue of monoclausality vs. biclausality of the syntactic causative. We will show that the embedded subject slot does not have to stay empty, and can be 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. filled with an overt element regardless of what marker the causee takes. In this case, the overt embedded subject does not necessarily have to be coreferential with the causee. There are several possibilities. For example, it could be coreferential with the causee, it could be the causee and a third party together, or it could be the causer and the causee together, etc. Even when the embedded subject position is empty, we claim that the empty embedded subject does not have to be coreferential when pragmatics is at work. We will first summarize the relevant literature in order to review what has been discussed. The subjecthood tests in Chapter 2 demonstrate that the nominative causee is the final subject of the embed ded clause, thus arguing for biclausality of the syntactic causative with nominative causee. We also argue that there are two clauses in the syntactic causative sentences with accusa tive or dative causee. By doing so, we show that no matter which marker the causee has, the syntactic causative sentences are all biclausai at every level of representation. We also look at the similarities and differences of syntactic causa tives in Korean and some Romance languages analyzed by Burzio (1986) and Zubizarreta (1985). In Chapter 4, we deal with the control sentences containing verbs like 'yaksokhata (promise)', and 'seltukhata (persu ade)'. We claim that the Korean examples are different from the ones in languages like English, Chinese, or Italian which have infinitive clauses in that the embedded subject position 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is not obligatorily empty, and can be filled with an overt element. Another difference we will point out in chapter 3 is that we claim that in Korean when the embedded subject posi tion is filled by an overt element, or even when it is empty, it does not have to be coreferential with the main clause subject when the main verb is the 'promise'-type, or with the main clause object when the main verb is the 'persuade'—type. At the end of this chapter, we will compare the similarities between the object control structure and the syntactic causa tive structure. We start this chapter with subject control of the 'nolyek- hata (try)', 'yaksokhata (promise)', and 'wenhata (want)'- types. We claim that the embedded subject position may be filled and when filled with an overt element, it does not have to be equivalent with the main clause subject unlike many other languages. In the next section, we review the relevant literature on object control in Korean and then we argue that there is no absolute constraint for the embedded subject posi tion, whether empty or filled, to be coreferential with the main clause object. In the following section, we show that due to a language specific structural reason, the main clause object does not have to be overt, leaving the controller choice to pragmatics. In section C of this chapter, we first show that in Korean it is not exclusively the semantics of the main verb which determines subject control or object control, but the choice of the complementizer also plays a crucial role 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in deciding it. Then as an example of that, we show that choo sing 'toiok' as the complementizer, the dominant tendency is that it is object control, and choosing 'lyeko' as the comple mentizer, it is likely to be subject control. Then we show how the choice of the controller is done in Korean and what the implication of it is. Finally we discuss the similarities of the syntactic causative and the object control sentences and argue that they both have the function of semantic causation. In Chapter 5, we claim that the Korean equivalent of the English 'so that' or purpose clause contains causation, that is the relation between the cause, the causing event and the effect, the desired event. We also show that the commonality of the syntactic causative and the object control structure, and the purpose structure is that the complementizer 'tolok' or 'key' is used in all three of these structures marking the desired event, showing that the main clause contains the cau sing event and the embedded clause the desired event. Our main point here is that in Korean, when the causing event and the desired event are encoded in two separate clauses and the cau ser and the causee are not equivalent, then there is a match between syntax and semantics of causation. In English three separate syntactic structures— one for the syntactic causative using 'bare-infinitive', one for the object control using 'to— infinitival', and one for purpose clause using 'so that' structure— may be used to encode semantic causation. In Japa nese, two syntactic structures are used, one for the syntactic 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. causative where the final or surface structure is monoclausal having the main verb and the embedded verb form a morphologi cal compound verb, and the shared one with the complementizer 'yooni' for the object control structure and the purpose clause structure. In Thai, the complementizer 'hay' is consis tently used for all of the three structures. There is, how ever, more than one type of syntactic causative structure in Thai, and one of them does not utilize the complementizer 'hay'. Thus we will have the following continuum of match- mismatch of syntax and semantics of causation defined as in (2). Mismatch Match English — > Japanese — > Thai — > Korean (D). Significance of this Study The main point we want to demonstrate in this study is to show that syntactically apparently unrelated phenomena in some languages may be proven to be related in some other languages when one starts from a dxfferent angle, anu sees tne similari ties of the phenomena. This is the case with the three pheno mena we look at here: the syntactic causative, object control, and the purpose clause structure with different main and em bedded subjects. Since these phenomena were initially studied in languages where there are not many apparent similarities among them, they are usually considered separate phenomena. We Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. started from the fact that the syntactic causative, the object control, and the purpose clause with different main and embed ded subjects share a lot in common. First, the main clause subject in ail these three phenomena is the initiator of the event described in the main clause which occurs to bring about the event in the embedded clause, thus ha*/ing a very close semantic link between the two events. Second, in all these sentences the one who carries out the event in the main clause is different from the one who carries out the event in the embedded clause whether the second one is expressed in the main clause or not. Third, all these phenomena use the same complementizers, 'tolok' or 'key'. Fourth, the realization of the desired event is not crucial for all these three phenomena as long as the causing event occurred to bring it about. These similarities encouraged us to attempt to discover the commona lity behind these phenomena normally treated separately. The commonality is that they all contain the relationship between 'cause' and 'effect'— the causing event in the main clause being the 'cause' and the desired event in the embedded clause being the 'effect'. This convinced us to claim that in Korean they are all semantic causatives sharing the basic syntactic structure. Our attempt to discover this commonality triggered the effort to see if it is the case in some other languages or confined to one specific language only. Although we need to do more thorough research on this, we found that it could be more than just language-specific. This is encouraging for any fur- 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ther research on this topic, and we will continue to work on this matter to have a better idea of the shared properties of these phenomena on the cross—linguistic level as well as lo cally in Korean. 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter I I . 'Subjecthood7 tests and the verb 'hata (do)7 (A). Subjecthood in Korean The notion of 'subject7 is important in order to see if the nominative causee is the final subject of the embedded clause or not. According to Gerdts7 (1990) claim the nominative causee is not a final subject due to the fact that the nomina tive causer is the sole subject in the final union clause when the sentence becomes monoclausal. As we prove that the nomina tive causee is indeed the final subject of the embedded clause, we can see that there is no causative clause union occurring. We will provide several "subjecthood*1 tests1, and apply these tests to support our biclausal argument. These tests apply to the final surface subject. / n t \ 11 * jp 5 • ? ( n . j . . ; n u t i u x . L i i U d C l u n I t h a s b e e n c l a i m e d t h a t t h e s u b j e c t h o n o r i f i c m a r k e r ' — s h i — 7 c a n b e a f f i x e d to t h e v e r b w h e n t h e s u b j e c t i s s o m e o n e w h o i s h o n o r a b l e ( E . Y . C h o , 1988; H.S. C h o e , 1988; H . S . H a n , 1987; Y.B. K i m , 1988; C. Y o o n , 1990). (1) a. John-i hakkyo-ey ka-ass-ta. -NOM school-to go-PST-IND 'John went to school.7 * b . J o h n - i h a k k y o - e y k a - s h i - e s s - t a . -HQN- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (2) a. Kim kyoswunim-i ku chayk-ul ilku-shi-ess-ta. professor-NOM the book-ACC read-HON-PST-IND •'Prof. Kim read the book. ' b. halapeci-ka Seoul-ey ecey ka-shi-ess-ta. grandfather-NOM -to yesterday go-HON-PST-IND 'Grandfather went to Seoul yesterday.' 'John' is in the subject position in (la&b), but it cannot trigger subject honorification because given names are not considered to be honorable in Korean. However, 'kyoswunim (professor)' and 'halapeci (grandfather)' in (2a&b) are honorable, and trigger subject honorification. (3) a. Mary-ka Kim kyoswunim-eykey ku namwu-lul ponay- -NOM professor-DAT the tree-ACC send— ess—ta. PST-Ind 'Mary sent the tree to Prof. Kim.' *b. Mary-ka Kim kyoswunim-eykey ku namwu-lul ponay- -NOM professor-DAT the tree-ACC send shi-ess-ta. HON-PST-IND 'Mary send the tree to Prof. Kim.' (4) a. Mary-ka halapeci-iul onul manna-ass-ta. -NOM grandfather-ACC today meet-PST-IND 'Mary met grandfather today. ' *b. Mary-ka halapeci-lul onul manna-shi—ess-ta. -NOM grandfather-ACC today meet-HON-PST-IND 'Mary met grandfather today.' 'Prof. Kim' and 'grandfather' in (3) and (4) are honorable, but they are not in the subject position, so cannot trigger subject honorification. Therefore, (3b) and (4b) are not gra mmatical. Even if an NP is in nominative case, it cannot tri gger honorification unless it is the subject. ■It3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (5) a. Mary-ka emeni-ka poko-shiph-ess—ta. -NOM mother-NOM see-want-PST-XNB 'Mary wanted to see mother.' *b. Mary-ka emeni-ka poko-shiphu-shi-ess-ta. -HON- 'Mary wanted to see mother.' c. Emeni-ka Mary-ka poko-shiphu-shi-ess-ta. -HON- 'Mother wanted to see Mary.' In (5a), the first nominal in nominative case, not the second one, is the subject of the sentence. Since 'Mary' is not hono rable, it cannot trigger subject honorification, thus (5b) is ungrammatical. But 'emeni (mother)' in (5c) is the subject, thus is able to trigger subject honorification. What we have seen so far tells us that only a subject NP which refers to an honorable person can trigger subject hono rification. But a little more careful observation may give us a somewhat different result. (6) a. John-uy apeci-ka incaha-shi-ta. —GEN father—riGM gentle—HON—IND 'John's father is gentle.' *b. John-uy yeca cninkwu-ka yeppu-shi-ta. girl friend-Nom pretty-HON-IND 'John's girlfriend is pretty.' (7) a. Kim kyoswunim-uy kanguy-ka hwulyungha-shi-ess-ta. professor-GEN lecture-NOM great-HON-PST-IND 'Prof. Kim's lecture was great.' *b. John-uy kanguy-ka nwuiyungna-shi-ess-ta. -HON- 'John's lecture was great.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'Apeci (father)' is the head noun of the noun phrase which is subject in (6a), and it triggers honorification. And 'yeca chinkwu (girl friend)' is the head noun of the noun phrase in (6b), and it does not trigger honorification. This agrees with what we have seen so far. However, the head noun in (7a) is not [+human], yet the honorific affix '-shi-' occurs in the predicate, when (7b) is not grammatical. Why is this possible? Can this be a counterexample against the argument that only subjects trigger subject honorification? It seems that the honorific triggering power of the dominating (/possessor) noun of the noun phrase percolates to the dominated (/possessed) noun when the referenta of the second noun (/head noun) are inalienable parts of the possessor such as his hand or nose. This is also true when the referents are mental abilities or individual skills such as speech ability2. This is possible only when the noun phrase with the honorable possessor is the subject of the sentence. If the noun phrase is not a subject, then it is not possible to trigger subject houorixication. (8) a. Mary-ka Kim kyoswunisn-uy kanguy-lul tul-ess-ta. -NOM professor-GEN lecture-ACC hear-PST-IND 'Mary Ixstened to riux. rvxm xeccure. *b. Mary-ka Kim sensayngnim-uy kanguy-lul tulu—shi- -HON-- ess—ta. 'Marv listened to Prof. Kim's lecture.' This shows that the honorable referent should be at least a part of a subject noun phrase in order to be able to trigger the subject honorification. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I n p a s s i v e c o n s t r u c t i o n s , t h e d e r i v e d s u b j e c t t r i g g e r s h o n o r i f i c a t i o n . (9) a. Kyengchal-i emeni-lul cap-ass—ta. poiiceman-NOM mother-ACC catch-PST-IND 'The policeman caught mother.' b. Emeni-ka kyengchal-eykey cap-hi-shi-ess-ta. mother-NQM policeman-DAT catch-Fass-KON-PST-IND 'Mother was caught by the policeman.' (10) a. Emeni-ka Mary-lul capu-shi-ess-ta. -NOM -ACC catch-HON-PST-IND 'Mother caught Mary.' *b. Mary-ka emeni-eykey cap-hi-shi-ess-ta. -NOM -DAT -Pass-HON- 'Mary was caught by mother.' 'Emeni (mother)' is the derived subject in a passive con struction (9b), and it triggers honorification. However, the underlying subject in (10b) does not trigger honorification. (A.2.) Honorific Nominative marker 'kkeyse' Honoring the subject is accomplished in two ways: firstly, the subject honorific marker '-shi-' can be affixed to the predicate; secondly, the honorific nominative marker 'kkeyse' can be substituted for the plain nominative case marker '—i/— ka' . In this section, we will look at how the honorific nominative marker 'kkeyse' behaves. 'Kkeyse' replaces the plain nominative marker only when the NP is the subject of the sentence. So in case of a double nominative construction, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'kkeyse' replaces only the noun phrase which is the sentential subject, (11) a. Mary-ka emeni-ka poko-shiph-ess-ta. -NOM mother-NOM see-want-PST-IND 'Mary wanted to see mother,' *b. Mary-kkeyse emeni-ka poko-shiph-ess-ta. -NOM(HON) 'Mary wanted to see mother.' *c. Mary-ka emeni-kkeyse poko-shiph-ess-ta. 'Mary wanted to see mother.' *d. Mary-kkeyse emeni-kkeyse poko-shiph-ess-ta. 'Mary wanted to see mother.' e. Emeni-kkeyse Mary-ka poko-shiph-ess-ta. -NOM (HON) 'Mother wanted to see Mary.' 'Mary', which is a given name, and thus considered not hono rable, is the subject in (lla-d). Therefore, the 'kkeyse' marker cannot be assigned to the subject 'Mary', nor to the second noun 'emeni (mother)'. 'Emeni (mother)', being hono rable in (lie), however, is the subject, thus 'kkeyse' can be assigned to it as the subject marker. 'Kkeyse' can also co occur with the subject honorific morpheme '-shi-' as in (12) since they are both used to honor the subject. -'-Shi-' can also appear alone without 'kkeyse' to honor the subject as we saw in (5c). (12) Emeni-kkeyse Mary-ka poko-shiphu-shi—ess—ta. —NOM(HON) —HON— 'Mother wanted to see Mary.' 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In the passive sentence, 'kkeyse' is added to the derived subject, not to the underlying subject as in (13). (13) Apeci-kkeyse kyengcnal-eykey cap-hi-shi-ess-ta. father—NOM(HON) poiiceman-BAT catch-Pass-HON-PST-IND 'Father was caught by a policeman.' (14) a. Apeci-kkeyse John-ul capu-shi-ess-ta. 'Father caught John.' b. John-i apeci-eykey cap-hi-ess-ta. 'John was caught by father.' *c. John-i apeci-kkeyse cap-hi-ess-ta. 'John was caught by father.' 'Apeci (father)' is the subject in an active sentence (14a), and 'kkeyse' is assigned to it. But it is no longer a subject in (14c), and cannot trigger subject honorification. (A.3.) '-mye(nse)' and '-taka' construction In Korean, the subject position of most of the subordinate clauses can be controlled either by the subject cf the main clause or by an object of the main clause3. (15) [°i/j ku i 1-11.1 ha-n hwu-ey], kycswunimi—kkeyse the work—ACC do-PST after-LOC -Nom(HON) Mary.-eykey ton-ul cwu-shi-ess-ta. -DAT money-ACC give-HON-PST-IND 'Johni gave money to Maryi after (he^she.) did the work.' The subject position of the subordinate clause is controlled by the subject of the main clause, 'John' or the recipient, 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'Mary'. Control of the zero anaphora in the subordinate clause by the subject of the main clause as well as the object occurs in other structure such as '— cilato (although)', — ki tta- mwuney (because)', '■— myen (if)', '— kkaci (until)', ' ■ — tusi (as if)', '— ttay (when)', etc. However, there are some cases in which only the subject of the main clause can control the zero anaphor which is the subject of the subordinate clause, '-myense (while)' con struction is one of them (C. Yoon, 1SS0). '—Myense' is a con— junctor whose contracted form is '—mye'. The unexpressed subject of the '-myense' subordinate clause is controlled only by the subject of the main clause. (16) Kim kyoswunin^—kkeyse [0i chayk-ul ilku—shi-myense] professor-NOM(HON) book-ACC read—HON-while nolay-lul pwulu-shi-ess-ta. song-ACC sing-HON-PST-IND 'Prof. Kim.j sang a song whle 0i reading.' The subject honorific marker '-shi-' in the subordinate clause shows that the covert subject of the subordinate clause is coreferential with the subject of the main clause. No argument other than the subject can be the controller of the zero anaphora in a '-myense' subordinate clause. (17) a. [0- j 7* chayk-ul ilku—myensej kyoswuniss^—kkeyse book-ACC read-while professor-NOM(HON) Mary^lul pwulu-shi-ess—ta. -ACC call—HON—PST-IND 'The professori called Mary while 0i reading a book. ' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. k. E°i/*j umak-ul tulu-myense] kyoswunim^-kkeyse music-ACC hear-while Mary--wa ku il-ul ha-shi-ess-ta. -with the work-ACC do-HON-PST-IND 'The professor^ did the work with Mary^ while listening to the music.' c- umak-ul tulu-myense] kyoswunim^kkeyse Mary^-lul wihaye ku kulim-ul kuli-shi-ess-ta. -ACC for the picture-ACC draw The professori drew the picture for Mary< while listening to the music. ' Another example of obligatory subject control for the co vert subject in the subordinate clause comes from the '-taka (while)' subordinate clause as in Y.S. Choi (1988) (The diffe rence between '-myense (while)', and '-taka (while)' is that the event in the '-myense' clause is simultaneously done, along with the event in the main clause, while the event in the '-taka' clause is stopped and the event in the matrix clause takes over). (18) (his 14a-b) a* tei/*j/* hakkyo-ey ka-taka] John-i chinkwu^-lul school—to go—while —NOM friend—ACC manna-ass-ta. meet-PST-IND 'While 0^*^ going to school, Johni met a friend^.' b. [*John-i/*ku-ka/*caki-ka hakkyo-ey ka-taka] John-i chinkwu—lul manna-ass-ta. 'While going to school, John met a friend.' The covert subject of the subordinate clause led by the '- taka' complementizer,- cannot be filled by an overt NP, pro noun, or a reflexive pronoun which is coreferential with the subject of the matrix clause, as in (18b). We can test, 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. though, the fact that the unexpressed subject of the subordi nate clause is controlled by the matrix subject by being able to assign the subject honorific marker '-shi-' to the verb when the subject of the matrix clause is honorable. (19) hakkyo-ey ka-shi-taka] kyoswunimi-kkeyse Tl^47 _ JT _ _ _ _ _ m * • m — — v —KGN— ^ruieSSuL-nuM(nuN) Mary.-lul manna-shi-ess-ta. -HON- 'While 0^*^ going to school, the professori met Mary^.' The subject of the matrix clause, 'kycswunim (professor) ' is honorable, therefore, the subject honorific marker '-shi-' can be affixed to the verb in the subordinate clause because the covert subject of that clause is coreferential with the subject of the main clause. If the subject of the matrix clause is not honorable, then '-shi-' cannot be assigned to the verb of the subordinate clause. (20) *a. [0i/*i hakkyo-ey ka-shi-taka] Mary^-ka kyoswuninu-ul -HON— manna-ass-ta. / w . , M . i . i . 1 . M _ . v . * 1 - A • — — » 2 t « u c j w i i X X C l j UX i l l j to school.' *b. [Ow*; hakkyo-ey ka-shi-taka] Mary^ka -HON- kyoswunimj=eykey ku chayk ul cwii—ess—ta. the book-Acc give-PST-IND 'Mary3 gave the book to the professor, while 0i/4j going to school.' The honorable referent, 'kyoswunim (professor) ' is patient in (20a), and recipient in (20b). Since the covert subject in the subordinate clause agrees only with the subject of the matrix 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. clause, the sentence becomes ungrammatical if '-shi-' is assigned to the verb of the subordinate clause in (2Qa-b) because the subject 'Mary' is not honorable. These data show that the unexpressed subject of a '—taka' clause is controlled only by a subject. A '-myense' clause whose subject is corefe— rential with the subject of the matrix clause can be embedded in the subordinate clause because the unexpressed subject in the '-taka' clause which is controlled by the matrix subject controls the subject of the '-myense' clause. (21) umak-ul tulu-shi-myense] [0i/4i hakkyo-ey ka- music-ACC hear-HON- shi-taka]] kyoswunim^kkeyse Mary.-lul manna-shi-ess-ta. HON-while 'The professori met Mary^ while 0i / J k j going to school, listening to the music. ' Finally, the derived subject of the passive construction controls the covert subject in the '-taka' clause. (22) a. [Oi/*j ku chayk-ul ilk-taka] Mary^ka kyoswuniKij- the book—ACC read— —NOM orofessor- eykey cap-hi-ess-ta. DAT catch-Pass-PST-IND 'While O^j reading the book, Mary^ was caught by the professor^. ' *b. [0 ku chayk-ul ilku-shi-taka] Mary-ka -HON- kyoswunim-eykey cap-hi-ess—ta. The ungrammaticality of (22b) shows that the covert subject in the '-taka' clause is controlled not by the underlying subject 'kyoswunim', but by the derived subject 'Mary'. This fact can 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. be demonstrated by having an honorable referent as the derived subject of the passive structure. (23) [Oi/*j ku chayk-ul ilku-shi-taka] kyoswunimi-kkeyse -HON- professor—NOM kyungchalj-eykey cap-hi-shi-ess-ta. policeman-DAT -HON- 'While reading the bock, the professors was caught hv the r>oJ. iceman . ' There are some other similar constructions in which the covert subject of the subordinate clause is controlled only by the subject of the matrix clause. (24) Mary-lul manna-shi-n-tus] kyoswunin^-kkeyse -ACC meet-HON-PST-as if professor-NOM(HON) Jane.-eykey malssumha-shi—ess-ta. -DAT say-HON-PST-IND 'The professor spoke to Jane as if (he) met Mary.' (25) chayk-ul tu-shi-n-chay-(lo) ] Kyoswunimi-kkeyse book-ACC hold—HON—PRES—as—(LOC) -NOM(HON) Mary.-eykey malssumha-shi—ess—ta. -DAT say-HON-PST-IND 'The professor spoke to Mary just as (he) was holding the book.' * ? h \ f ft T k h i i ? n n h a ~ c " h i - m t r o n c a f \ r l - n/*j f------------- — ------ - —1 — w w j M.^2 wo- tired-HON-even though Mary^-lul manna-shi-ess-ta. 'Even though (he) is tired, the professor met Mary.' The unexpressed subject in the subordinate clause led by 'tus (as if)' in (24), 'chay (just as)' in (25), and 'myenseto (even though)' in (26) is controlled exclusively by the sub ject of the main clause. 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (A.4.) coordination In coordination with the conjunction '-ko', if the subject is the same for all the conjuncts, then obligatory coordinate deletion occurs. (27) Joh^-i Mary.-eykey ku chayk-ul cwu-ko [0^*= ku il-ul -NOM -DAT the book—ACC give-CON the work—ACC hay-ess-ta. ' Johni gave the book to Mary^ and did the work. ' The covert subject of the second conjunct in (27) is only controlled by the subject of the first conjunct, 'John'. This can be shown by the appearance of the subject honorification marker '-shi-' in the two conjuncts. (28) Kyoswunin^-kkeyse Mary.-eykey chayk-ul cwu-shi-ko, professor-NOM(HON) -DAT book-ACC give-HON-CON [0i/4- ku il-ul ha-shi-ess-ta. the work—ACC do-HON-PST-IND 'The professori gave the book to Mary- and 0i / ( , ; did the work.7 J a '-myer.se' clause can be inserted somewhere in conjunct. (29) Kyoswunisi.j-kkeyse Mary,-cykcy chayk-ul cwu=shi-ko, ku il-ul [ umak-ul tulu-shi-myense] music—ACC hear—HON— ha-shi-ess-ta. 'The professori gave the book to Mary^, and 0i / ! k j did the work while 0, listenina to music.' i,-j If the subject of the second conjunct is different from the first one, then it must overtly occur in the sentence. on Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (30) John—i ku chayk—ul ilk-ko, Mary—ka/*0 talun chayk—ul -NOM the book-ACC read-CON iik-ess-ta. 'John read the book, and Mary read another book.' The subject of the first conjunct is 'John', and the subject of the second conjunct is 'Mary'. 'Mary' must appear in the sentence since the first subject does not control the second subject. If one of the conjuncts is about time or weather, the subject of that clause can be deleted, but '-ko' is not used as the conjunction, but another conjunction '-se' is used. (31) [shikan(/0)-i nute-se(/*-ko)] John-i thaykshi-lul time-NOM iate-because -NOM taxi-ACC tha-ass-ta. ride-PST-IND 'It was late, and John took a taxi.' (32) [nalssi(/0)-ka maywu chwuwe-se(/*—ko), Mary-ka weather-NOM very cold-because ttukewun wuyu-lul mashi—ess—ta. hot milk drink 'It was very cold, and Mary drank some hot milk.' The appearance of the first subject is optional in these examples where the first subject is time or weather. Only '— se', though, can be used as the con junctor in these cases. The use of '-ko' as the conjunction in (31)-(32) makes the sentence ungrammatical. (B) The verb 'hata (do)' (B.l) Introduction We claim in this study that the syntactic causatives are biclausal, regardless of which case the causee is in. This 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. means that the verb 'hata (do)' is the matrix verb, and it is not a part of a compound verb which may be claimed, in a monoclausal analysis, to be formed by compounding the main verb and the embedded verb. It is certain that causative sen tences have the basic interpretation that the causer caused the causee to do something. But with 'hata-' as the main verb it is not clear what kind of action is taken on the causer's side because 'hata' only assigns the interpretation that the causer did something to cause the event to happen. In this section, we will look at the different functions which 'hata' carries depending on its usage, and we will attempt to claim that all the functions of 'hata' may be grouped into two: one light 'hata', and the other heavy 'hata'. In section (B.2), we lay out the different usages of 'hata'. In section (B.3), we will discuss 'hata' as a light verb. In section (B.4), I will discuss 'hata' in long form negation and in a delimiter (/VP—focus) construction. In section (5* 5) I will deal wxth 'hata* eta tuc uidin vero m some biclausal structures. In section (B.6), I will have a dis cussion for generalizing and categorizing the functions of 'hata' into a minimal number of them. (B.2) Types of 'hata' 'Hata' is the most generic verb in Korean in the sense that it is used in so many constructions and in so many ways. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. First, -'hata' is used as a regular main verb in a monoclausal sentence as in the next example. (1) John-i kukes-ul hay-ess-ta. -NOM that—ACC do-PST-IND 'John did that.' Secondly, 'hata' is used as a light verb. A light verb is a verb which "subcategories and case-marks a direct object NP, without assigning it a theta-role" according to Grimshaw and Master's (1888) definition, whereas a regular verb assigns case as well as theta-role to arguments. (2) a. John-i Mary-wa senmwul-ul kyohwan-ul hay-ess-ta. -NOM -with gift-ACC exchange-ACC do-PST-IND 'John exchanged gifts with Mary.' b. John-i Mary-wa senmwul-ul kyohwanhay-ess-ta. -NOM -with gift-ACC exchange-PST-IND 'John exchanged gifts with Mary.' According to Grimshaw and Mester, 'kyohwan (exchange)' in (2a) is a direct object of the light verb 'hata', which does not have theta-role assigning capability. (2b) is the incorporated version of (2a), in the sense that the noun 'kyohwan' and the verb 'hata' are merged to form a verb. This kind of incorpora tion process occurs in nativizing a foreign word4. A loan word cannot be used as a verb unless it is combined with 'hata'. In the examples in (1), 'kyohwan' is of Chinese origin, and it must be combined with 'hata' to be used as a verb in Korean. We will explain the verbalizing process in more detail in section (C). n a ou Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thirdly, 'hata' is used as an auxiliary verb, as in long form negation, and the delimiter (/VP—focus) construction. (3) a. John-i ku chayk-ul ani-ilk-ess-ta. -MOM the book-ACC NEG-read-PST-IND 'John did not read the book.' b. John-i ku chayk-ul ilk-ci ani-hay-ess-ta. —NOM the book-ACC read-Cl NEG-do-PST-IND 'John did not read the book.' As mentioned in K.A. Kim (1993), (3a) is called short form negation as the NEG morpheme precedes the verb, and (3b) is called long form negation as the NEG follows the main verb. In the long form negation, the NEG morpheme 'ani' precedes and is attached to the auxiliary 'hata'. There are some particles such as '-to (also)', '-man (only)', which are referred as 'delimiters' by Korean linguists. (4) John-i ku umshik-ul po-ki-man hay-ess-ta. -NOM the food-ACC see-NM-only do-PST-IND 'John only saw the food (but he did not eat it).' If the main verb is nominaiized and has a delimiter attached to it, the auxiliary 'hata' must appear as in (4). Fourthly, 'hata' is used as the main verb in some biciausai structures. Take a syntactic causative sentence (5) as an example. (5) John-i Mary.j-eykey [0i ku chayk-ul ilk]-key hay- -NOM -DAT the book-ACC read-COMP do- ess-ta. PST-IND 'John caused Mary to read the bock.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'Hata' in (5) is the main verb which has 'John' and 'Mary1 - as its arguments. In (5) the embedded subject is not overt, but its position can be filled with an overt element as we argue in the next chapter. The structure of the syntactic causative sentence (5) is quite similar, if not equivalent with a typical so—called object control structure as in (6). (6) John-i Mary^eykey [0i ku chayk-ul ilk]-key seltukhay- -NOM -DAT the book-ACC read-COMP persuade ess-ta. PST-IND 'John persuaded Mary to read the book.' The main verb in (6) is a so-called object control verb 'seltukhata (persuade)'. We claim in this dissertation that the syntactic causative sentence as in (5) is biclausal like the so-called object control structure with verbs like 'sultukhata (persuade), thus 'hata' is used as a main verb while there is a separate embedded verb. E. Y. Cho (1987) claims that in the case in which the causee is marked accusative in a syntactic causative sentence, then subject-to-object raising has occurred. The following is his SXSEiplS - (7) (Cho's (2) on page 3) Nay-ka Inho-lul chayk-ul sa-key hay-ess-ta. I—NOM -ACC book-ACC buy-COMP do-PST-IND 'I caused Inho to buy a book.' He claims that the causee 'Inno' in (7) once was the embedded subject at some point of derivation, and became an object of the main sentence. In his analysis, the clause boundary is 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. maintained, so the structure of (7) for him will be as follow ing; ( 8 ) N a y - k a I n h o - l u l [ s c h a y k - u l s a ] - k e y h a y - e s s - t a . In his analysis, the subject position of the embedded clause does not exist since the embedded subject became an object of the main clause. Therefore, in his analysis a syntactic causative whose causee is marked accusative is a subject-to- object raising structure. However, we will demonstrate in section (E) that this is not correct. Instead, we claim that there is a covert subject position in the embedded clause when the causee is not only dative-marked but also accusative- marked5. I n n e x t s e c t i o n , w e w i l l d e a l w i t h l i g h t v e r b ' h a t a ' c o n s t r u c t i o n . (B.3) Light verb construction Grimshaw and Mester (1988) divide NPs into two categories; opaque NPs and theta—transparent NPs. An opaque NP is an NP none of whose inside elements "interact thematically with any element outside it". A theta-transparent NP means an NP which "assigns theta-roles outside its own maximal projection (which is the NP itself)". They claim that in the following Japanese examples the capitalized nouns which are, according to them, the direct object of the verb 'suru' are transparent NPs6. 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (9) (their 3a-c, p207) a. John-wa Mary-ni HANASHI-o shita. -TOP -to taik-ACC suru 'John talked to Mary.' b. John-wa Tookyoo-kara SHUPPATSU-o shita. -TOP Tokyo-from departure-ACC suru 'John departed from Tokyo.' c. John-wa murabito-ni [ookami—ga kuru—to] KElKOKU-o —TOP villager-to wolf—KOM come—COMF warn—ACC shita. suru 'John warned the villagers that the wolf was coming.' The selection of the sentential arguments in the sentences in (9) is decided by the capitalized noun, not by the verb 'suru'. They attempt to prove this point by the fact that these nouns can head the nominalization which includes all sentential arguments as follows: (10) (their 4a-c) a. John-no Mary-e-no HANASHI -GEN -to-GEN talk 'John's talk to Mary' b. John-no Tookyoo-kara-no SHUPPATSU nmt J j--- u u n u ~ x x uui~ \jcjxi u c y a i u u i e 'John's departure from Tokyo' c. John-no murabito-e—no [ookami-ka kuru-toj-no -GEN villager-to-GEN wolf-NOM come-COMP-GEN 1 * ^ < 5 n c x n u n u warning 'John's warning to the villagers that the wolf is coming' According to them, in Japanese light verbs like 'suru' cooccur with theta-transparent NPs, and verbs other than the light verb 'suru' are heavy verbs which "take only theta- opaque objects". They make a distinction between theta-opaque O A Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. NPs and theta-transparent NPs, which is that theta-opaque NPs can be topicalized, whereas tbeta-transparent NPs cannot. (11) (their 6a—b) a. John-ga [[ookami-ga kuru-to]-no HOQKOKU]-o -NOM wolf-NOM come-COMP-GEN report-ACC wasureteita. had forgotten 'John had forgotten the report that the wolf was coming.' b. [[Ookami-ga kuru-to]-no HOOKGKU]-wa John-ga wolf-NOM come-COMP-GEN report-TOP -NOM wasureteita. had forgotten 'John had forgotten the report that the wolf was coming.' (12) (their 7a-b) a. John-ga [ookami-ga kuru-to] HOOKOKU-wa shita. —NOM wolf-NOM come—COMP report—Top suru 'John reported that the wolf was coming.' b. *HOOKQKU-wa John-ga [ookami-ga kuru-to] shita. report-TOP -NOM wolf-NOM come-COMP suru 'John reported that the wolf was coming.' The bracketed NP in (11a) is a theta-opaque NP, and it is topicalized freely as in (lib). But the theta—transparent NP in (12a), which is 'HOOKQKU (report)' cannot be topicalized, resulting in ungrammaticality as in (12b). They claim that the theta—transparent NP has the ability to assign theta roles to the sentential arguments, and the light verb 'suru' is only capable of assigning accusative case, since it is considered transitive, to the transparent NP which is marked accusative '-o'. The light verb 'suru' simply functions, according to them, as "a bearer of verbal inflec tion for the clause and as a case assigner, allowing the 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. transparent noun in its direct object to assign theta-roles in a verbal context". The transparent head noun, by being com bined with 'suru', becomes the functional equivalent of a verb. They also claim that the theta-marking ability of the noun is absorbed by 'suru', and 'suru' assigns theta—roles to the sentential arguments. The process through which the light verb 'suru' acquires a theta-marking ability is called, by Grimshaw and Mester, 'Argument Transfer'. They also claim that "light 'suru' resembles in many ways the 'do' of English 'Do Support', which carries inflection but assigns no tneta-roies and imposes no seiectional restric tions. The key difference between English 'do' and Japanese 'suru' is", they argue, "that 'suru' is transitive, so it can combine with NPs for theta—role assignment, whereas 'do' must combine with another verb. Note that the intransitivity of 'do' supports the idea that the transitivity of 'suru' is a lexical property, since there is lexical variation among light verbs". H.D. Ahn (1990) claims that a light verb in Korean or Japanese does not have ability to assign either case or theta- role. He defines a light verb as "a verbal category that has incomplete (if not null) theta-grid but often combines with the theta-assigning verbal nouns (VN) and the complex becomes a full-fledged theta-assigning verb". However, he agrees with Grimshaw and Mester in the sense that the theta-role assigning capacity is transferred from the verbal noun to the light 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. verb7. He additionally argues that the case—assigning capacity is also transferred from the verbal noun to the light verb. The latter point is different from Grimshaw and Mester's claim that the light verb has its own capacity to assign an accusa tive case to the verbal noun. 'Suru' in the following example is treated as a heavy verb. (13) (Ahn's 4a) John-ga Seoul-ni toochaku-o shita. -NOM -to arrival-ACC did rJohn arrived in Seoul•f (14) (Ahn's 5a) John-ga Tokyo-kara shuppatsu-o shita. -NOM -from departure-ACC did 'John departed from Tokyo.' He generally treats 'suru' as a heavy verb if the verbal noun is marked accusative as in (13)-(14), and considers 'suru' as a light verb if it is incorporated with the verbal noun as in (15). /1C\ 1 ■ * ■ * • ’ / John-ga eigo-o benkyoo-shita. —NOM English-ACC study-did 'John studied English.' He claims, however, that the incorporated noun 'benRyoo (studying)' in (15) is not a verb, but a noun, in fact a phrasal NP8. (He maintains the view that a verbal noun is a noun categorically.) In the case of the incorporated verbal noun, which is the only true light verb construction in Ahn's sense, it cannot be extracted to head a relative clause or be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the focus in a cleft construction. He accordingly makes the following assumption. (16) (his 10) Verbal nouns cannot be extracted. He maintains this assumption throughout his paper. Because (15) is a light verb construction according to him, the verbal noun 'benkyoo (study)' cannot be extracted as expected. (17) (his 12b—c) a. *John-ga eigo~o t shita benkyoo -NOM English-ACC did study *'The study that John did English' b. *John-ga eigo-o t shita no—wa benkyoo-da. -NOM English-ACC did- study *'It is the study that John did English.' He claims that the following Korean example (18) is not a true light verb construction because the verbal noun is extractable as shown in (19). / 1 Q \ /U4 « = 1 \---/ V--------- • — f John—i Mary—eykey chwungko—lul hay—ess—ta. -NOM -DAT advice-ACC do-PST-IND 'John advised Mary.' / * l O V /Ui r. 1 CUV V • * ■ " / — ■» * — / John—i Mary-eykey t ha-n chwungko -NOM -DAT do advice 'The advice that John did to Mary' He claims that (19) is from the sentence (18), not from the incorporated version as in (20). 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (20) John-i Mary—eykey chwungko-hay-ess-ta. -NOM —DAT advice-ao—PST—IND 'John advised Mary-' He claims that NPs which have their own internal structure as in (21)-(22) can be formed if 'suru' in Japanese, or 'hata' in Korean is a heavy verb: the bracketed NP in (21) has the structure of 'Noun-genitive case + Noun', and the bracketed NP in (22) has the structure of 'S (/adjective) + Noun'. (21) (his la) John-ga [eigo-no benkyoo]-o shita. -NOM English-GEN study-ACC did 'John did a study of English.' (22) (his 16c) John-i Sewul-eyse [ttay nuc]-un chulpal-ul -NOM -from be late-COMP departure—ACC hay-ess-ta. do-PST-IND 'John departed from Seoul late.' '(lit.) John made a late departure from Seoul.' However, in a true light verb construction, Ahn says, an adjectival modification cannot be made to the verbal noun as the ungrammaticality of (23) demonstrates. (23) (his 17a) * John—i yenge-lul [slyswun ko»3pwu]—jj«y—sss—ts. -NOM English-ACC difficult study-do-PST-IND *'(lit.) John did a difficult study English.' K.B. Park (1992) claims that regardless of the occurrence of an accusative case assigned to the verbal noun, the follow ing sentence is a light verb construction. (In this respect, his claim is consistent with Grimshaw and Mester (1988).) 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (24) (his 18b) John-i ku saken-ul CQSA-(lul) nay-ess-ta. -NOM the affair-ACC investigate(-ACC) do-PST-IND 'John investigated the accident.' He defines a light verb as "a class of verbs that are seman tically empty or incomplete". He argues that the sentential internal argument (/object) is not an argument of the light verb 'hata', but of the verbal noun. According to his claim, 'ku saken (the affair)' in (24) is an argument of the verbal noun 'cosa (investigation)'. Unlike Grimshaw and Mester (1988) or H.D. Ahn (1990), he claims that the theta-roles and the case are directly assigned to the arguments by the verbal noun. Therefore, in his theory, there is no transfer of the theta-role assigning capacity from the verbal noun to the light verb. In Grimshaw and Mester (1988)'s theory, the verbal noun is an argument of the light verb, but Park claims that it is not the case. He claims that there is no syntactic rela tionship between the verbal noun and the light verb. He assigns the feature [-N, -V] to the verbal noun. He justifies his position on this by saying that a [— N] category such as a verb or a preposition may assign Case. He claims that the light verb 'hata' only occurs when the verbal noun requires an affix with the feature [+V] . Since the verbal noun cannot have an affix such as tense, or agreement marker attached to it self, a light verb is required to host an inflectional affix. Due to this necessity of attachment of an affix to an element with [+V] feature, 'hata' insertion is needed. If a light verb 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'hata' is not inserted, and an affix is directly attached to a verbal noun, then that sentence will be ungrammatical, as (25). (25) (his 24) *John-i ku saken-ul COSA-ass-ta. -NOM the affair—ACC investigate—PST—XND 'John investigated the accident.' Sentence (25) is ungrammatical because an affix such as a tense marker '-ass-' cannot be directly attached to the verbal noun. We may look at the combination 'a verbal noun + hata' from a slightly different angle. The verbal noun 'cosa (investiga tion) ' in (24) is of Chinese origin, which may be used as a verb as well as a noun in Chinese. (See K.A. Kim (1992) for more detail on nativization of loanwords in Korean.) In a general sense, 'hata' is used to verbalize a loan noun or verb, or a native noun. (26) a. John-i ku chayk-ul summarizu(-iui) hay-ess-ta. -NOM the book-ACC summarize(-ACC) do-PST-IND 'John summarized the book.' b. John-i ku chayk-ul summary(—lul) hay-ess-ta. -NOM the book-ACC summary^-ACC) do-PST-IND 'John summarized the book.' In order to nativize and verbalize an English verb like 'summarize' as in (25a), or a noun like 'summary' as in (26b), a light verb 'hata' is needed to host an inflectional affix 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. because a loan word itself cannot host an inflectional affix like tense. (27) *John-i ku chayk-ul summarize-ess-ta. A light 'hata' is also used to verbalize a native noun such as 'mal (language)', or 'il (work)' because a noun cannot h ncf S i Z l inflectional affix. (28) John-i Mary-eykey kulekhey mal(-ul) hay-ess-ta. -NOM -DAT so language(-ACC) do-PST—IND 'John told Mary so.' This means that a light 'hata' is obligatorily required to host an inflectional morpheme in verbalizing a loanword or a native noun. As we have seen above, Grimshaw and Mester (1988), H.D. Ahn (1990), and K.B. Park (1992) differ in treating which one is a genuine light verb construction, and which one is not. And they have different views on the role and function of the verbal noun and the light verb in the sentence. However, they all agree that a light verb behaves differently from a regular verb, which they call a heavy verb, in that a light verb lacks theta-role assigning capacity. And they all agree that there are two types of 'hata' or 'suru'— one is a light verb, the other one is a heavy verb. (29) (Park's 78) John-i ku saken-uy COSA-lul hay-ess-ta. -NOM the affair-GEN investigation-ACC do-PST-IND 'John investigated the affair.' '(lit. John did the investigation of the affair.' 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. They all agree that 'hata' in a sentence like (29) is a heavy verb which has the NP 'ku saken-uy cosa (investigation of that affair)' as its argument. (B.4) 'Hata' in Negation and Delimiter Construction As often mentioned in the literature (H.S. Han, 1987; S.C. Song, 1988; J.H. Suh, 1990, etc.), in Korean there are two types of negation; short form and long form. (30) a. John-i ku chayk-ul ilk-ess-ta. -NOM the book-ACC read-PST-IND 'John read the book.' SHORT FORM b. John-i ku chayk-ul ani-ilk-ess-ta. -NOM the book-ACC NEG-read-PST-IND 'John did not read the book.' LONG FORM c. John-i ku chayk-ul ilk-ci ani-hay-ess-ta. -NOM the book-ACC read-CI NEG-do-PST-IND 'John did not read the book.' Tn rnp S0H50 t nnr rnp truth of (3Oct) is the tWC types of negation in (30b) and (30c) share the same basic interpretation. The difference between these two types of negation is that in the short form the NEG morpheme 'ani' precedes the verb, while in the long form the NEG morpheme is affixed not to the content verb— this term is used by M.Y. Kang (1988), and K.B. Park (1992)— but to an auxiliary verb 'hata'. Another structure which uses an auxiliary 'hata' is 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the VP-focus construction— this term is used by M.Y. Kang (198S). (31) John-i ku chayk—ul ilk-ki-nun hay-ess-ta. -NOM the book-ACC read-KX-CON do-PST-IND 'John read the book, but...... ' (31), due to the presence of a contrastive marker '-nun', has a contrastive meaning. For instance 'John reads that book, but not this one.' or 'John reads the book, but did not write an essay.' We may expand this use of 'hata' in a broader sense. An auxiliary 'hata' is obligatorily used when the content verb such as 'ilkta (read)' in (31) is accompanied by one of the elements typically called delimiters in the Korean literature. I.S. Yang (1972:59) defines it as 'a semantic one which im plies that the meaning of the element to which a delimiter is attached is 'delimited and/or specified.' Delimiters are elements such as 'mata (every)', 'man (only)', 'pakke (only, in a negative sense)', 'to (also)', 'ya (at least; of course)'. M.Y. Kang (1983), following the claim made by S.C. Song / 1 Q * 7 * 3 \ u B T / 1 O T O K * 1 £ m \ _ „ ^ .jl jl_ * js .a, i - .. _ *i y 1 ^ / / ** • M • A iWV y A ^ I W M f X ^ / 4 * f f O D D uiuc a t»UCl U U11C J. U 11 k* L l U I l d i element '-ci' in a long form negation as in (30c) is histo rically derived from '-ki' which is commonly used as a complementizer (/nominalizer). He claims that long form negation and the VP-focus construction are syntactically in the same format, except that in long form negation there is a NEG morpheme 'ani' attached to 'hata'. Based on this, M.Y. 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kang claims that long form negation and the VP-focus con struction share the same structure as following; (32) John-i [vp [yp ku chayk-ul ilk-ki(/ci) ]-(nun) -NOM the book-ACC read-KI(/CI)-(CON) (ani)-hay-ess—ta. (NEG)-do-PST-IND He assumes that 'hata' can assign an accusative case 'lul' to the VP in a long form negation, for instance to [vp ku chayk-ul ilk-ci] in (30c). However, an accusative case cannot be assigned to the VP in a VP-focus sentence. (33) *John-i ku chayk-ul ilk-ki—lul hay-ess—ta. -NOM the book-ACC read-KI-ACC do-PST-IND 'John did read the book.' He explains that this is due to the fact that a nominative case 'ka' or an accusative case 'lul' cannot cooccur with one of the delimiters mentioned above9. 4 A \ / I - * i » * * » . * * » v y j - * / ( u x a / ^ a — j j r r / i ) a. *Cheiswu—ka—to hakkyo—e ka—ass—ta. —NOM—also school-LQC go-PST-IND 'Chelsu went to school also.' b. Chelswu-to hakkyo—e ka—ass—ta. -also school-LOC go-PST-IND 'Chelsu went to school also.' (35) (his 73a—b, P71) a. *Cheiswu-ka chayk-ul-man ilk—nun-ta. -NOM book-ACC-only read-IMP-IND 'Chelsu reads only books.' b. Chelswu-ka chayk-man ilk-nun—ta. —NOM book—also read—IMP—IND 'Chelsu reads only books.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The reason why (34a) and (35a) are ungrammatical is, Kang claims, that in (34a) a delimiter 'to (also)' is attached to a nominative case 'ka', and in (34b) another delimiter 'man (only)' follows an accusative case 'lul'. With this kind of empirical evidence he assumes that an overt accusative case 'lul' cannot be attached to the content verb with the noraina- lizer 'ki' in a VP-focus construction as the ungrammaticality of (33) shows. He argues that the reason why the nominalizer 'ki (/ci)' is required in the VP-focus construction and in long form negation is that the VP needs to become a nominal form to receive a case from the auxiliary 'hata'. K.B. Park (1992) claims that the VP-focus construction such as (31) is generated exactly in the same way as the light verb in the light verb construction, as we have seen in section (C), is generated. He claims that the affix 'ki' needs to be introduced to the content verb in order to attract a delimiter. This turns the content verb into an element with the feature [—V], thereby the content verb cannot host inflectional elements like tense. This process of nominalizing the content verb triggers insertion of the light verb 'hata', which will host the inflectional affix since it is a verb with the feature [+V]. He argues that the light verb 'hata' is needed in a long form negation sentence like (30c) for exactly the same reason: hosting an inflectional affix. Since the content verb led by '-ci' cannot host any inflectional affix, 'hata' is needed to carry out this role. His overall claim is 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. that "whenever a verb is prevented from attaching to an inflectional affix by an intervening element, the light verb 'ha'-support is needed". (B.5) 'Hata' as the main verb in biclausal structures In K.A. Kim (1993), we argued that the syntactic causatives in Korean are biclausal at every level, providing several pieces of empirical evidence no matter which case is assigned to the causee— nominative, accusative, or dative. We also argued that the causative sentence with the causee in dative or accusative case is structurally similar to, if not the same as, the object control structure, in that the unexpressed embedded subject is coreferential with the object which is the causee, not with the main clause subject. The fact that the syntactic causative is biclausal means that 'hata' is not merged with the embedded verb to be a compound verb, but remains the main verb. E.Y. Cho (1SS7) claims that when the causee is in the accusative case, subject=to==object raising has occurreci. We will repeat below the sentence (7), and the structure (8), which may be claimed by his analysis, for the sake of convenience. (36) (=7) Nay—ka Inho-lul chayk-ul sa-key hay-ess-ta. I—NOM -ACC book-ACC buy-COMP do-PST-IND 'I caused Inho to buy the book.' 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (37) (=8) Nay-ka Inho-lul [s chayk-ul sa]-key hay-ess-ta. -NOM -ACC ' book-ACC buy-COMP do-PST-IND 'I caused Inho to buy the book.’ He claims that the embedded subject moved to the object posi tion of the main clause, hence there is no subject position left in the embedded clause, although the biclausal structure of the sentence is maintained. We demonstrated in K.A. Kim (1993) that an overt pronoun or reflexive pronoun can occur in the embedded subject position as in the following sentences. (38) John-i Mary-lul [kunye-ka ku il-ul honcase ha]-key -NOM -ACC she-NOM the work-ACC alone do-COMP hay-ess-ta. do-PST-IND 'John instructed Maryi that shei do the work alone. ' (39) John-i Mary-lul [kunye cashin-i ku il—ul honcase -NOM -ACC she self-Nom the work-Acc alone ha]-key hay-ess-ta. do-COMP do-PST-IND 'John instructed Marvi that she^lit. herself) read the book.' If subject-to-object raising occurred in (38), the embedded subject position should not exist, but it does exist as we see in (38)-(35), which is filled by a pronoun as in (38), or by a reflexive pronoun as in (39). Therefore, the subject-to- object raising analysis for the syntactic causative whose causee is in the accusative case cannot be maintained. We will see that it is not an isolated phenomenon that 'hata' in a syntactic causative sentence, which is biclausal in my analysis, is used as a main verb. N.K. Kim (1984) 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. claims that one of the quotative sentences in Korean has 'hata' as its main verb. (40) (his 47a, on P126) John-i ka-ass-ta-ko ha-n-ta. -NOM go—PST—IND-COMP do-IMP-IND '(They) say that John went.' He notes that a syntactic peculiarity with this structure is that the main clause subject is deleted and not present in the surface. The deleted main clause subject is understood to be 'kutul (they)' or 'saramtul (people)'. The following is the structure with the main clause subject inserted in its position. (41) saramtul-i [John-i ka-ass-ta]-ko ha-n-ta. people—NOM -NOM go-PST-IND-COMP do-IMP-IND 'People say that John went.' 'Hata' in (40)-(41) can be interpreted as 'malhata (say)' or 'cenhata (report)' as indicated by N.K. Kim (1984). In the embedded clause led by the 'ko' complementizer all four major sentence types, which arc declarative, interrogative, impera tive, and propositive (/suggestive), can occur in indirect quotation. (See N.K. Kim (1984) for detailed arguments.) It needs to be noted that 'hata' can come as the main verb in each case, although usually a verb which is comparable with the embedded sentence tvne is used as the main verb. (42) a. John.j-i Mary-eykey [0i ku il-ul hay-ess-ta]-ko -NOM -DAT the work-ACC do-PST-IND-COMP malhay-ess-ta (/hay-ess-ta). teil-PST-IND (/do-PST-IND) ,Johni told Mary that (he^ did the work.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. John-i Marjr-eykey [0i ku il-ul hay-ess-nun- -NOM -DAT the work-ACC do-PST- ya]-ko mwul-ess-ta (/hay-ess-ta). INTR-COMP ask-PST-IND (/do-PST-IND) 'John asked Maryi if (she.,) did the work. ' 'John asked Maryi whether (she^ did the work.' c. John-i Mary.j-eykey [0i ku il-ul ha-la]-ko -NOM -DAT the work-ACC do-IMP-COMP myenglyenghay-ess-ta (/hay-ess-ta) . order-PST-IND (/do~PST~IND) 'John ordered Mary to do the work.' d. John-i Mary.-eykey [0i+- ku il-ul hamkkey -NOM -DAT the work-ACC together ha-ca]-ko ceyanhay-ess-ta (/hay-ess-ta). do-SUG-COMP suggest-PST-IND (/do-PST-IND) 'John suggested Mary to do the work together.' When 'hata' is used as the main verb in (42a) whose embedded sentence type is 'declarative', 'hata' may be interpreted as 'malhata (tell)', 'cwucanghata (claim)', 'taytaphata (answer)' etc. Since the embedded sentence is a quotative complement as argued by Kim (1984), in the usual sense, 'hata' means 'malha ta (say)' when it is used as the main verb in sentences like (42a). The embedded sentence in (42b) is an 'interrogative' sentence type, thus 'hata' used as the main verb is normally interpreted as 'mutta or cilmwunhata (ask)'. However, it may also be interpreted as 'malhata (tell) ' since it is again quotative. The embedded clause in (42c) is an 'imperative' sentence type, thus 'hata' as the main verb in such a sentence may be interpreted as 'myenglyenghata (order)', 'malhata (tell)', 'kangyohata (force)', 'yochenghata (request)', 'seltukhata (persuade)', etc. Finally, 'hata' in sentences like (42d) may mean 'ceyanhata (suggest)', 'malhata (tell)', Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'yochenghata (request)', etc. because the embedded clause is a 'propositive' type. What this means is that interpretation of 'hata' in these cases may be context-dependent, although the scope of interpretation is highly restricted because of the number of verbs which can substitute for 'hata'. The function of 'hata' as the main verb in each case is quite limited, and unless the previous context clearly suggests some specific interpretation, the most likely meaning in each case may be assigned to interpret 'hata'. 'Hata' is also used as the main verb in some other bi clausal structures. (43) John^un [0i ku il-ul ha]-ki-lo hay-ess-ta. -TOP the work-ACC do-COMP-LOC do-PST-IND 'John decided(/promised/planned) to do the work.' *'(lit. John did to do the work.') (44) John.j-un [0i ku chayk-ul sa]-lyeko hay-ess-ta. -TOP the book-ACC buy-COMP do-PST-IND 'John planned(/thought) to buy the book.' As shown in the English translation, 'hata (do)' in (43) may mean one of several things such as 'kyeishimhata (decide)', 'yaksokhata (promise)', 'kyeyhcykhata (plan)', etc. , even though it may normally be interpreted as 'kyeishimhata (decide)' in a usual situation. 'Hata' in (44) may be inter preted as 'kyehoykhata (plan)', or 'sayngkakhata (think)'. We may say, therefore, that 'hata' as the main verb in such biclausal sentences as we have seen in the above may not be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. semantically clear or unambiguous, and the interpretation of it may depend on the context to a certain degree. As mentioned before, Grimshaw and Mester (1988) , H.D. Ahn (1SS0), K.B. Park (1992) all acknowledged that there are two types of 'suru (do)' in Japanese, and 'hata (do)' in Korean; light verb and heavy verb. They all agree that a light verb 'suru', or 'hata' lacks theta-role assigning capacity, while a heavy verb 'suru', or 'hata' has both case- and theta—role assigning capacities, functioning as a regular main verb. In all of their analyses, a sentence like (29) includes a heavy 'hata'. (45) (=29) John-i [f , p ku saken-uy COSA]-lul hay-ess-ta. -NOM the affair-GEN investigation-ACC do-PST-IND 'John investigated the affair.' '(lit.)John did the investigation of the affair.' 'Hata' in (45) is a heavy verb, which assigns an accusative case and a theta-role to its argument NP 'ku saken—uy cosa (the investigation of the affair)'. K.B. Park (1992) claims, as we mentioned earlier, that a light verb occurs only when there is no element in the sentence to host an inflectional affix like a tense morpheme. 'Hata' in (42a—d) is a heavy verb in his definition and Grimshaw and Mester's, and Ahn's, since it is a case- and theta-role assigning regular verb, and there is no need for a light verb to host an inflectional affix. We argue in the next chapter that 'hata' in syntactic causative 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sentences, which we analyze as biclausal at all levels, is also a regular heavy verb which assigns case and theta-role to its own arguments. (B.6) Discussion We have seen so far the different roles the verb 'hata' plays: first, a main verb (/heavy verb) in a monoclausal sentence; secondly, a light verb; thirdly, an auxiliary verb in long form negation and the VP-focus construction (/delimi ter construction); and fourthly a main verb in biclausal sentences including the syntactic causative construction. K.B. Park (1992) claims, as we have seen above, that 'hata' in the long form negation and the VP-focus construction is needed to host an inflectional affix such as tense. This is the function which a light verb 'hata' has: hosting an inflectional affix. We think that his argument is on the right track, and in this paper we take his position on the light verb 'hata' in long form negation, VP-focus construction, and 'hata' with a verbal noun. Thus, we can categorize all the different functions of 'hata' into two: light 'hata' and heavy 'hata'. A heavy 'hata' is the one which is used as a main verb in monoclausal and biclausal sentences while a light verb 'hata' is the one which does not fully function as a regular verb as defined by Grim shaw and Mester (1988). Except for the heavy 'hata' used as r o Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the main verb in a monoclausal sentence, the interpretation of the heavy 'hata' used in various biclausal sentences is not really unambiguous, and is quite context dependent. The interpretation of a light 'hata' in the VP-focus construction, and in long form negation is rather dependent on the content verb. Take a VP—focus sentence for an example, which has an adjective as its predicate. (46) John-i onul aphu—ki-nun hay-ess-ta. -NOM t o d a y sick-NM-TOP do-PST-IND 'John is sick today, but....... ' In this case, the light verb 'hata' does not have the meaning 'do' which a heavy action verb 'hata' has, but rather indica tes 'state of John's welfare'. A similar aspect of a light 'hata' can be found in a long form negation. (47) John-un ku chayk—ul ilk-ci-lul ani—hay-ess-ta. -TOP the book-ACC read-CI-ACC NEG-do-PST-IND 'John did not read the book.' /Aft \ John—un h—yn^pckh2“Ci*"kH sni—hsy^sss—ts ■ -TOP is happy-CI-NOM NEG-do-PST-IND 'John was not happy.' If the predicate is a verb, then the case marker attached to the predicate is usually an accusative as in (47) . A nomina tive case is normally attached to the predicate, however, if the predicate is an adjective as in (48). In the case of the sentence (47), whose content predicate is a verb, a progres- 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sive aspectual morpheme '-ko iss—' or an imperfactive aspec tual morpheme '-n-' can be attached to the light verb 'hata'. (49) a. John-un ku chayk-ul ilk-ci-lul ani-ha-ko iss-ta. —TOP the book-ACC read-CI-ACC mEG—do—PROG—IND 'John is not reading the book.' b. John—un ku chayk-ul ilk-ci ani—ha—n-ta. -TOP the book-ACC read-CI NEG-do-IMP-IND 'John does not read the book.' The progressive aspectual morpheme '-ko iss-' is attached to the light 'hata' in (49a), and an imperfactive morpheme '-n-' to the light 'hata' in (49b). These sentences are perfectly grammatical. However, it does not seem possible with the light 'hata' in (48), because the content predicate is not a verb, but an adjective. (50) a. *John-un hayngpokha-ci—ka ani-ha—ko iss-ta. -TOP is happy-CI-NOM NEG-do-PROG-IND 'John is not being happy.' b. *John-un hayngpokha-ci-ka ani-ha-n-ta. -TOP is happy-CI-NOM NEG-do-IMP-IND Thus it seems that a heavy 'hata' as a main verb in a biclausal sentence and a light 'hata' in long form negation and the VP-focus construction may share one thing, which is that the interpretation of them is vague, and their inter pretation depends on something other than themselves. 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (B.7) Conclusion We have looked at several usages of the verb 'hata'. Following K.B. Park (1992)'s proposal, we may call 'hata' in 'verbal noun (taccusative) + hata', long form negation con struction, and the VP-focus construction light 'hata'. We call 'hata' used as the main verb in monoclausal and biclausal sentences heavy 'hata'. One common thing with a light 'hata' in long form negation and VP-focus construction and a heavy 'hata' as a main verb in biclausal sentences is that the interpretation of them is not unambiguous, and is influenced by something other than themselves. We argue that there are some evidence that there are two types of 'hata': one light 'hata', and the other heavy 'hata'. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 3. Syntactic Causativity in Korean (A) Introduction In many languages, the paradigm case for the causative sentence depends on the intransitivity/transitivity of the embedded clause. In causatives of intransitives, the embedded subject stands as direct object as in (1). In causatives of transitives, the embedded subject functions as indirect object in derived structure as in (2). In causatives of ditransi— tives, the embedded subject appears as one of the obligue cases other than subject, direct object, or indirect object as in (3) (Comrie, 1976). Take Turkish, for instance. (1) the embedded subject = direct object (Comrie*s (1976) (6)) Ali Hasan-i ol—dur —du. -DO die Cause PST ' A l i c a u s e d H a s a n t o d i e , k i l l e d H a s a n . ' (2) the embedded subject = indirect object (Comrie's (1976) (7)) Disci mektub-u mudur -e imzala-t -ti. dentist letter—DO director—10 sign—Cans —PST 'The dentist made the director sign the letter.' (3) the embedded subject = an oblique case (Comrie's (1976) (8)) Disci Hasan-a mektub—u mudur tarafindan dentist ~IO letter—DO director by goster-t-ti. show-Caus-PST 'The dentist made the director show the letter to Hasan.' 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Many Relational Grammarians claim that in many languages the causative sentence which was biclausal at one level becomes monoclausal at the final level, two clauses being collapsed into one clause, and they argue for the causative clause union analysis as follows (cf. Perlmutter and Postal (1974), Aissen and Perlmutter (1983): (4) a. The final 1 (Subj) of a finally intransitive complement heads a 2-arc (DO) in the matrix clause. b. The final 1 (Subj) of a finally transitive complement heads a 3 (ID)-arc in the matrix clause. This hypothesis basically says that the matrix clause and the embedded clause in a causative sentence merge at some level, and become a single union clause. According to this hypothe sis, the subject of the intransitive complement in the Spanish sentence (5), for instance, is a direct object in the final monoclausal sentence, thus is in the accusative 'la'. (5) (Aissen & Perlmutter (1983)'s (P2a)) La hice correr. her-ACC I-made run 'I made her run.' The subject of the transitive complement is an indirect object in the final monoclausal sentence in the final level, thus is in the dative case 'le'. (6) (Aissen & Perlmutter (1983)'s (P3a)) Le hice buscar las herramientas. her-DAT I-made seek the tools 'I made her look for the tools.' •— D O Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Following the union analysis, Gibson & Raposo (1986) claims that 'les garcons' in a French sentence (7) is a 2 (direct object) of the matrix clause, and 'son amie' in (8) is a 3 (indirect object) of the union clause. (7) (Their (la)) II laisse courir les garcons. he lets to-run the kids 'He lets the kids run.' (8) (Their (2a)) II laissera manger les gateaux a son amie. he will-let to-eat the cakes to his friend 'He will let his friend eat the cakes. ' Unlike arguments made by Relational Grammarians, in generative framework, the structure does not change in the derivational process. Burzio (1986) says, summing up the basic position of the generative framework for the causative sentence, that "In the current framework, in which semantic interpretation deri ves from S—structure, and in which Q (Theta)—roles are assign ed at all levels, one generally assumes that syntactic opera tions cause no 'loss' of structure. It is then natural and, strictly speaking, necessary to assume that the S complement and its relevant internal structure, [NP VP] , are preserved in derived structure". So he claims that the Italian sentence (9) has the structure like (10), maintaining the biclausal structure. (9) (his (lc), P228) Maria ha fatto riparare la macchina a Giovanni. has made repair the car to 'Maria had Giovanni repair the car. ' 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (10) (his (5), P230) Maria ha fatto [vp riparare la macchina] has made repair the car [s a Giovanni ] to He claims that the VP [riparare la macchina] has moved out of the embedded clause, and '---• ' in (10) is the trace of the moved VP. Zubizarreta (1985) claims, assuming the generative frame work, that French causatives such as (11)-(12) have dual structures; one is biclausal (i), and the other monoclausal (ii) . (11) (her (87)) Pierre a fait tracer les plans (par son associe) (i) [s NP [vp V [s[vp V NP (PP) NP]]] (ii) [s NP [vp V NP (PP)]] (12) (her (90a-b)) a. Pierre a fait lire passages a Jean (i) [s NP [vp V [s[vp V NP] NP]]] (ii) [s NP [vp V NP PP]] b. Pierre a fait travailler Jean (i) [s NP [vp V [s[vp V NP]]] (ii) [s NP [vp V NP] ] 1 In her analysis, the embedded subject position in (11) functions as a dummy element when the sentence is biclausal, having structure (i). She also claims that when the sentences (12a-b) are biclausal, 'Jean' serves as subject and object at the same time. We are not going to explain in this paper how 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'Jean' can serve as a subject and an object at the same time. Refer to Zubizarreta (1985) for a detailed explanation. One thing unusual about the syntactic causative in Korean is that the causee can be in nominative, accusative, or dative case, regardless of the intransitivitv/transitivity of the complement clause. (13) John-i Mary-ka/-eykey/-lul ca-tolok hay-ess-ta. -NOM —NOM/-DAT/—ACC sleep-COMP do-PST-IND 'John caused Mary to sleep.' (14) John-i Mary-ka/-eykey/-lul ku chayk-ul ilk-tolok -NOM —NOM/-DAT/—ACC the book-ACC read-COMP hay-ess-ta. do-PST-IND 'John caused Mary to read the book.' (15) John-i Mary-ka/-eykey/-lul Bill-eykey ku chayk—ul -NOM —NOM/-DAT/—ACC -DAT the book-ACC phal]-tolok hay-ess-ta. sel1-COMP do-PST-IND 'John caused Mary to sell the book to Bill.' The complement clause is intransitive in (13), transitive in (14), and ditransitive in (15) . The causee can be in nomi native, accusative, or dative case, regardless of the senten tial type of the embedded clause. In this dissertation, we claim that the syntactic causative sentences such as (13)-(15) are biclausal at all levels of representation. We claim that when the causee is in nomina tive case, it is the subject of the complement clause. When the causee is either in accusative or in dative, we argue, it belongs to the matrix clause at every level. In this case, the 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. subject position of the complement clause is controlled by the causee, and it is usually not expressed. The unspecified sub ject position of the complement clause, in these cases, can optionally be filled by a name (/NP), a pronoun or a reflexive pronoun which is coreferential with the causee, or some element which has a semantic and pragmatic connection with the causee when it is not coreferential with it. In section (B), we review several different views on the structure of the syntactic causative in Korean and give a critique of a monoclausal analysis argued by Gerdts (1990). In section (C), we attempt to prove the biclausality of the so- called syntactic causative in Korean, providing several pieces of evidence for our analysis. Finally, in section (D), we summarize our claim made in this chapter, and draw our conclusion. (B). Literature Review and a Critique of the Monoclausal Analysis by Gerdts (1990) (B.l.) Literature Review As in many other languages, Korean has both lexical causatives and syntactic causatives. (1) Kanhowen-i hwanca-evkey umshik-ul mek-i—ess-ta= nurse-NOM patient-DAT* food-ACC eat-CAU-PST-IND fThe nurse fed the patient food.' (2) Kanhowen—i hwanca-eykey umshik—ul mek-tolok hay-ess-ta. nurse-NOM patient-DAT food-ACC eat-COMP do-PST-IND fThe nurse caused the patient to eat food.' CO Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A lexical causative like (1) has a causative morpheme '-i~'1 affixed to the verb, and a syntactic causative as in (2) uses a causative verb 'hata (do)', which is separated from the embedded verb, 'mek (eat)', by the complementizer 'tolok'. I.S. Yang (1972), following the generative sematicists' view on causati^/es, claims that in Korean a lexical causative and its syntactic causative counterpart are synonymous, thus they share the same deep structure. He chooses the syntactic causa tive structure as the underlying structure for both the lexi cal causative form and the syntactic causative form. He attempts to show that the underlying structure which is equi valent with the syntactic causative form can be reduced to the lexical causative form by going through several rules such as modality deletion, predicate raising, etc. However, Shibatani (1973) argues that the lexical causative and the syntactic causative in Korean do not share the same underlying structure, arguing against not only I. S. Yang (1972) , but 21SO C^GXlGITGl ^CncrowlVc S3mauulClStS CjLalul Oil causatives (cf. Lakoff, 1965; McCawley, 1968). (Shiatani's argument, however, does not have anything to do with the mono— clausal/biclausal structure of the syntactic causative at the surface level.) He first points out that there are semantic distinctions between the lexical causative and the syntactic causative. (3) (Shibatani(1973)'s (3a-b) a. Emeni-nun ai-eykey os-ul ip-hi-ess-ta. mother-TO? child—DAT clothes-ACC wear-C&U-PST-INB 'The mother put the clothes on the child.' 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. Emeni-nun ai-eykey os-ul ip-key hay-ess-ta. motner-TGP child—DAT clothes-ACC wear—CMP do-PST-IND 'The mother had the child put on the clothes.' He claims that "the periphrastic (/syntactic) causative often involves action on the part of the causee while this is not generally the case with the lexical causative”. Shibatani claims that adverbial modification should create ambiguity as to which event an adverbial modifies if there are two underlying events with two separate agents. (4) (Shibatani's (7b&8b)) a. Pak ssi-nun ku salam-ul yel shi-ey cwuk-key Mr.-TOP the person-Acc ten o'clock-at die-COMP hay-ess-ta. do-PST-IND 'Mr. Pak made the man die at ten o'clock. ' b. Pak ssi-nun ku salam-ul yel shi-ey Mr.-Top the person-Acc ten o'clock-at cwuk-i—ess-ta. die-CAU-PST-IND 'Mr. Pak killed the man at ten o'clock.' He claiss that an example like (4a) is ambiguous in meaning because it can mean that Mr Pak's killing the person occurred at ten or Mr Pak arranged things so that the man die at ten. With this he argues that it is better to posit two underlying sentences for the syntactic causative to account for the ambiguity in meaning. (5) (Shibatani's (11a—b) a. (Pak ssi (ku salam cwuk) yelsi-ey hay-ess-ta) b. (Pak ssi (ku salam yelsi-ey cwuk) nay-ess-ta) 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The time adverbial belongs to the matrix clause in the underlying structure (5a), and to the embedded clause in (5b). However, the time adverbial in a lexical causative like (4b) modifies the entire event, meaning there is only one underlying clause for the lexical causative sentence. Next he points out that reflexivization in Korean can occur across clause boundaries, and ambiguity arises about which argument the reflexive pronoun 'caki' refers to when the underlying structure of a sentence has two or more subjects. This is the case with the syntactic causative as in (6). (6) (Shibatani's (19b)) Emeni-nun atul-eykey caki-uy os-ul ip-key mother-TOP son-DAT self's clothes-ACC wear-COMP hay-ess-ta. do-PST-IND 'The mother had the son put on his/her own clothes.' 'Caki' can refer either to 'emeni (mother)' or 'atul (son)' in (6). However, 'caki' in a lexical causative sentence like (7) refers only to one antecedent 'eiiieiix ^motner) ' . (7) (Shibatani's (20b)) Emeni-nun atul—eykey caki-uy os—ul ip-hi-ess-ta. motherTOF son—DAT self's ciocues-AUJ wear—cAU—fST—IND 'The mother dressed the son with her/*his own clothes.' He claims that this distinction between the lexical and the syntactic causative in reflexivization shows that they are not from the same underlying structure. 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. He points out another difference between the lexical causative and the syntactic causative with regard to subject honorification. The subject honorific marker '—shi—'2 can be added to the verb if the subject of the sentence is someone honorable. (8) Kyoswunim-i ku chayk-ul ilku-shi—ess-ta. professor-NOM the book-ACC read-HON-PST-IND 'The professor read the book.' In a syntactic causative sentence, '-shi-' can appear in the embedded verb. (9) (Shibatani's (25a)) Haksayng—un kyoswu-eykey wusu-shi-key hay-ess-ta. student-TOP professor-DAT smiie-HON-COMP do-PST-IND 'The student had the professor smile (hon).' (10) (Shibatani's (26a)) *Haksayng-un kyoswu-lul wusu-shi—ki-ess-ta. student-TOP professor-ACC smile-hon-PST-IND 'The student made the professor smile(hon).' Noting the fact that '-shi--' appears in the embedded verb because 'kyoswu (professor)' is honorable, he claims that in the underlying structure, 'kyoswu' is the subject of the embedded clause in a syntactic causative sentence like (9). But (10) is ungrammatical because the deference is toward the student, not toward the professor, proving that 'kyoswu' is not a subject in the underlying structure for the lexical causative sentence like (10). 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Y.S. Kang (1986) assumes that in case the causee is in the accusative case as in (lib), verb raising has occurred, attaching the embedded verb to the matrix verb3. (11) (his 96a-b) a. John—i [s,[s Mary-ka ha]-key] ha-ess—ta. -NOM -NOM go-COMP cause-PST-IND '•John caused Mary[4-NOM] to go. ' b. John-i Mary-lul ka-key ha-ess—ta. -NOM -ACC go-COMP cause-Past-IND 'John caused Mary[+ACC] to go.' Therefore, the biclausal sentence at the initial level becomes monoclausal at the final level, and the causee in accusative case is the direct object of the newly created complex verb in the final monoclausal sentence. If verb raising does not occur, then the sentence stays biclausal, having the causee in nominative case as in (11a) . His analysis says that in case the causee is in the nominative case, then it is biclausal at the final level, and in case the causee is in the accusative case, then it is monoclausal. Ke will show that this is not the case, and the syntactic causatives in Korean are biclausal, no matter which case marker the causee is in, including when it is in accusative case. H.S. Choe (1988) argues that no matter which case marker a causee is in, the causee is the subject of the embedded clause. She claims that a reciprocal 'selo (each other)' is bound by the closest subject binder. an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (12) (her (23)) Wuli^-ka kutul^i selo^-lul salangha-n-ta~ko we-NOM they-NOM each other-ACC love-PRES-IND-comp sayngkakha-n-ta. think-PRES-IND 'We, think they, love each other ,,.' -j -i/] In her analysis, the reciprocal 'selo (each other)' is only bound by the subject of the embedded clause 'kutul (they)', not by the subject of the matrix clause 'wuli (we)'. Applying these binding facts to the '-key ha-' syntactic causative, she argues that the causee is the subject of the embedded clause because the causee is the only available binder of the reci procal 'selo', regardless of which case the causee is in. (13) (her (24) wuli.j-ka kutuli-{ka/lul/eykey} selo^-lul salangha-key we-NOM they-NOM/ACC/DAT each other-ACC love-COMP hay-ess-ta. do-PST-IND 'We- made them, love each other-,*,.' 1 J J / * l 'Selo' in (13) can be bound only by the causee 'kutul (they)', not by the causer 'wuli (we)'. Another piece of evidence which she briefly mentions is the restriction on the distribution of pronouns, which is binding condition (B). (14) (her (25)) Chelswu.j-ka Yengswu--eykey (/-lul/-ka) ku^.-lul -NOM -DAT/ACC/NOM he-ACC piphanha—key hay—ess—ta. criticize-COMP do-PST-IND 'Chelswui caused Yengswu, to criticize him-^. ' Only the causer 'Chelswu', not the causee 'Yengswu' can be the antecedent which is coreferential with the pronoun 'ku (he) ' . 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Her claim that the syntactic causative is biclausal at every level is consistent with our biclausal analysis. One major difference between ours and hers is that we do not claim that the causee in dative or accusative case is the subject of the embedded clause. We claim in section (C) that the usually empty subject position of the embedded clause is coreferential with the causee in dative or accusative, but it can optionally be filled by an overt NP (/name), a pronoun, or a reflexive pronoun 'casin'. Cheong Youn (1990) makes a quite similar claim as does H.S. Choe (1988) about the embedded subject. In a passive sentence, the final subject controls subject honorification4. (15) (Cheong Youn (1990)'s (19) & (22)) a. Apenim-kkeyse sunkyeng-eykey cap-hi-shi-ess-ta. father—NOM(HON) policeman-by catch-PAS-HON-PST-IND 'Father was caught by a policeman.' b. Yengswu-ka sensayngnim-eykey cap-hi-(*shi)-ess-ta. -NOM teacher-by catch-PAS-(HON)-PST-IND 'Youngsoo was caught by the teacher.' 'Apenim' in (15a) is the final subject of the passive sentence, and it triggers subject honorification. So the subject honorification marker '-shi-' is affixed to the verb. 'Sensayngim (teacher)' in (15b) which is the underlying subject, however, cannot trigger subject honorification because it is no longer the subject. Based on this fact, Cheong Youn (1990) formulates the following condition. 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (16) (his (33)) A final 1(subject) can control Subject Honorification. He argues that the causee is the final subject of the embedded clause because it triggers subject honorification, and the embedded verb can have the subject honorification marker ' — shi-' as in the following example. (17) (his (115)) Ai-tul-i sensayngnim-i/-ul/-eykey ku chayk-ul chile-PL-NOM teacher-Nom/-Acc/-Dat the book-Acc ilku-shi-key hay-ess-ta. read-HON-COMP do-PST-IND 'The children made the teacher read the book.' He claims that 'sensayngnim (teacher)' in (17) is the final subject of the embedded clause, and triggers the subject honorification. He also argues that if the subject of the embedded clause is not honorable, then it cannot trigger subject honorification. / i o \ / k - i / 1 n \ \ Sensayngnim-i ai-tul-i/-ul/—eykey chayk-ul teacher-NOM child-PL-NOM/-ACC/-DAT book-ACC ilku=(*shi)-key hay—ess—ta. read-(HON)-COMP do-PST-IND The matrix subject in (18) is honorable, so triggers subject honorification as in (19). (19) (his (119)) Sensayngnim-i ai—tui-i/-ul/—eykey chayk-ul teacher-NOM chile-PL-NOM/-ACC/-DAT book-ACC iiku-(*shi)-key ha-shi-ess-ta. read—(HON)—COMP do—HON—PST—IND 'The teacher made the children read books.' 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. With this fact, he claims that this subject honorification test proves that not only are the syntactic causative bi clausal in their initial structure, but also the causee is the final subject in the embedded clause. He also claims that the antecedent of the reflexive pronoun 'casin' can be coreferentxal wxth an argument wxthxn or out of the same clause as long as the antecedent is a subject. Due to this fact, an ambiguity arises. (20) (his (120)) Swuni-nun [Chelswu-ka casin-ul haktaehay-ess-ta]-ko -TOM -NOM self-ACC abuse-PST-IND-COMP mit—nun—ta. believe—PRES-IND ,Soonii believes that ChulsoOj has abused self^.' 'Casin' in (20) can refer either 'Suni' which is the subject of the matrix clause or 'Chelswu' which is the subject of the embedded clause. (21) (his (29a)) Chelswu-ka Suux—lul Yengswu—eykey casin—uy -NOM -ACC -DAT self—GEN samwushil-eyse sokayhay-ess-ta. office—in introduce—FST—IND 'Chulsooi introduced Sooni^ to Youngsook in self's^*^^ office. • ' 'Casin' in this monoclausal sentence is coreferential only with the sentential subject 'Chelswu'. If this is correct, he argues, and if the syntactic causative is monoclausal, then there should not be any ambi guity in coreferentiality of 'casin' and its antecedent. How- *7 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ever, ambiguity does arise. The reflexive is coreferential with the causer in one reading, and with the causee in another reading. (22) (his (121)) Chelsu-ka Suni-ka/-lul/-eykev casin-uy kohyang-ulo -NOM -NOM/-ACC/-DAT ~ self-GErf hometown-to tolaka-key hay—ess—ta. go back-COMP do-PST-IND ,Chulsooi made Sooni^ go back to self's^.' The interpretation for (22) is ambiguous because 'casin-' can refer to either the causer 'Chelsu' or the causee 'Suni'. He claims that these facts about the reflexive 'casin' proves the 'subjecthood' of the causee, saying that only the final sub ject of the embedded clause can occur in three different cases — nominative, accusative, and dative. (Again his claim is different from ours in that we argue that the causee in dative or accusative case is not the subject of the embedded clause.) O'Grady (1991) takes the position that in a causative with the causee xn the nominative case, the causer which bears the actor role is the subject, and the embedded clause is the direct object of the causative verb 'hata'. (23) (his (1), P173) John-i [Sue-ka chayk-ul ilk]-key hay-ess-ta. -NOM -NOM book—ACC read-COMP do-PST-IND 'John made Sue read the book.' First, he claims that case in the embedded clause is assigned according to the familiar norm, the accusative case 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is assigned to the NP 'chayk (book)' which is a part of the VP, and the nominative case is assigned to the NP which combines with a VP to create a sentence. Secondly, each clause in (23) has its own subject term that triggers a reflexive pronoun 'caki' interpretation and subject honorific agreement. (24) (his (3), pl74) [John-i^ [Sue-kai caki-uyi/i chayk-ul ilk-key] -NOM -NOM self—GEN book-ACC read-COMP hay-ess-ta] do-PST-IND 'John- made Sue ^ read self's.^ book.' (25) (his (4)) [Kyoswunim-i [Sue-ka chayk-ul ilk-key] ha-shi-ess-ta. professor—NOM —NOM book-ACC read-COMP do-HON—PST-IND 'The professor made Sue read the book.' (26) (his (5)) [Sue-ka [kyoswunim-i chayk-ul ilku-shi-key] hay-ess-ta. -NOM professor-NOM book-ACC read-HON-COMP do-PST-IND 'Sue made the professor read the book.' 0'Grady assumes the following structure for the syntactic causative with the causee in dative case. / O T \ / h 4 o / "i \ n 1 1 C \ \ — • / \--- \ / jr- ■ ~ / [s John-i Sue . j —eykey [s proi chayk-ul ilk]-key -NOM -DAT book—ACC read-COMP hay-ess-ta. do-PST-IND / ,7aV« Ciia *•/> « ■ » t W V MM «UUW W UV ^ VUVI Uil'O M • ( 2 8 ) ( h i s ( 4 ) , p l 7 7 ) [s John-i Sue^eykey [s proi ilha-key] hay-ess-ta. -NOM -DAT work-COMP do-PST-IND 'John made Sue work.' The covert subjects of the embedded clause in (27)-(28) are controlled by the causee in the dative case. 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, he proposes that in the case of the syntactic causative with the causee m the accusative case, the causative 'hata' has the following lexical property. (29) (his (2), pl81) hata: -NPa IVPt (Intransitive Verb Phrase) In this analysis, the causee in the accusative case is a part of the intransitive verb phrase. (30) (his (1), pl81) John-i Sue—lul chayk—ul ilk-key hay—ess—ta. -NOM -ACC book—ACC read-COMP do-PST-IND 'John made Sue read the book.' According to him, the first step is to form an IVP (intransi tive verb phrase) by combining the transitive verb 'ilkta (read)' with its theme term 'chayk (book)'. And this IVP and the causative verb 'hata' form a transitive verb phrase. The next step is that the NP ' Sue' is incorporated into syntactic e < 4 > a a T T 7D / “ i » % + ■ v* i ^ i m U **•« < - • a \ H J Z t ^ U a W > i . M W M M J . W M W W ^ W U M W U l i A V * y M 4 . U M l 3 U M W M . V V # V aoc y • n J U V l U l t c final step is that this IVP combines with the actor term which is the causer 'John' related to the causative verb 'hata', forming an S. O'Grady's main point is that the syntactic causative with the causee in either nominative or dative case is biclausal, but the one with the causee in the accusative case is mono clausal. But we claim, in this paper, that even the latter is biclausal. 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gerdts (1990) claims that the syntactic causative in Korean is monoclausal at the final level, even though it is biclausal at some level. (31) (Gerdt's (9a)) Sensayngnim-i na-lul tolaka-key hay-ess-ta. teacher-NOM I-ACC return-COMP5 do-PST-IND 'Teacher made me return.' (32) (Gerdt's (10a)) John-i na-eykey ttek-ul mek-key hay-ess—ta. -NOM I—DAT rice cake-ACC eat-COMP do-PST-IND 'John made me eat the rice cake.' She argues that if the embedded clause is an intransitive clause, tken the subject of the embedded clause would become the direct object as in (31). If the embedded clause is a transitive clause, she argues, then the subject of the embed ded clause becomes the indirect object of the final clause and the direct object of the embedded clause keeps its original grammatical relation in the final clause, inheriting its relation as in (32) (cf. Gibson and Raposo, 1986). She gives two types of evidence for her monoclausality. One of them is predicate fusion, which says that the two predicates in the syntactic causative are fused, and become one complex predicate at the final level. She says that about 60% of the 30 Korean native speakers she surveyed accept sentences like (33}-(35). (33) (Gerdts' (17)) Sensavngnim-ul ttena-kev haksavng-i hay-ess-ta. teacher-ACC leave-COMP student-NOM do-PST-IND 'The student made the teacher leave.' 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (34) (Gerdts' (18)) John-eykey ttek-ul mek-key nay—ka hay-ess-ta. -DAT rice cake-ACC eat-COMP I-NOM do-PST-IND 'I made John eat the rice cake.' (35) (Gerdts' (19c)) Haksayng-i sensayngnim-ul ttena-key onul hay-ess-ta. student-NOM teacher-ACC leave-COMP today do-PST-IND 'The student made the teacher leave today.' She speculates that whether some elements can appear between the two predicates or not, predicate fusion has occurred. With this reasoning, she claims that "for some speakers, the causa tive predicate 'ha' is morphologically realized as a verbal suffix. Since the two predicates form a single word, lexical material, including scrambled subjects as in (33)-(34) and adverbials as in (35), cannot intervene between them. For other Korean speakers, she says these causatives involve causative clause union, but the two predicates are separate words, and there is no ban on intervening lexical item (this second claim is, we will argue in the following subsection (B.2.) and section (C), a speculation which is not supported by empirical evidence because no element can come between the two parts of a genuine compound verb). The second evidence which she suggests is 'plain topicali- zation'. She says that only the terms (subject, direct object, and indirect object) can become the plain topic. She claims that when the causee is in accusative case in a causa tive with a transitive embedded clause as in (36), the first nominal in the accusative case is the direct object and the second nominal in the accusative case is chomeurized in the 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. final clause, losing its direct objecthood. Therefore, only the first nominal in the accusative case can become the plain topic, as in the following examples (38a-b). (She claims that the causee in dative case as in (37) is an indirect object in the union clause, thus causing no problem for the direct object 'ttek (rice cake)' to be a plain topic.) (36) (Gerdts' (25)) John—i na-lul ttek-ul mek-key hay-ess-ta. -NOM I—ACC rice case-ACC eat-COMP do-PST-IND •'John made me eat the rice cake. ' (37) (Gerdts' (32)) Ttek—un John-i na-eykey mek-key hay-ess-ta. rice cake-TOP —NOM I-DAT eat-COMP do-PST-IND 'John made me eat rice cake.' (38) (Gerdts' (33)) *Ttek-un John-i na-lul mek-key hay-ess-ta. rice cake-TOP -NOM I-ACC eat-COMP do-PST-IND 'John made me eat rice cake.' I'John made the rice cake eat me.' She argues that there is only one final subject in the union stratum. She also claims that when the causer, the upstairs initial subject, and the causee, the pivot nominal, both take the nominative case, the causee in the nominative case is no longer a subject in the final single union clause because there is already a nominal which is the subject. (39) Mary-ka John-i cip-ey ka-key hay-ess-ta. -NOM -NOM house-to go-COMP do-PST-IND 'Mary made John go home.' In her clause union analysis, 'Mary' is the final subject, and 'John' is not a subject any more in (39), and is a chomeur in 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the final union level since in her theory there is only one final subject within a clause. (40) (Gerdts' (47)) Sensayngnim-i haksayng-i ttena-key ha-shi-ess-ta. teacher-NOM student-NOM leave-COMP do-HON6-PST~IND 'The teacher made the student leave.' (41) (Gerdts' (49) Haksayng-i sensayngnim-i ttena-key ha-(*shi)-ess-ta. student-NOM teacher-NOM leave-COMP do-(HON)-PST-IND 'The student made the teacher leave.' According to Gerdts, the reason why the subject honorification marker '-shi-' can be affixed to the verb 'hata (do)' in (40) is because the causer is the final subject, and in (41), '- shi—' is not allowed because the causee in the nominative case is not the final subject. She claims that when a syntactic causative is embedded under a control verb, the PRO corresponds to 'causer', as in (42), not the 'causee', as in (43). (42) (Gerdts' (53)) Nay—ka John^eykey [PROi Mary—ka ttena-key ha-tolok] I—NOM -DAT -NOM leave-COMP do-COMP pwuthakhay-ess-ta. ask-PST-lND 'I asked John to make Mary leave.' (43) (Gerdts' (54)) *Nay-ka Johr^-eykey [Mary-ka PROi ttena-key ha-tolok] I—NOM -DAT -NOM leave-COMP do-COMP pwuthakhay-ess-ta7. ask-PST-IND 'I asked John that Mary make him leave.' On the basis of this fact, she claims that PROs are limited to final subjects, and argues that the causer in the nominative 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. case, not the causee in the nominative case, is the final sub ject. Finally, she uses the 'plain topicalization' test as a third argument for her claim that the first nominal in the nominative case is the final subject.' (44) (Gerdts' (57)) Sensayngnim-un haksayng-i ttena-key hay-ess-ta. teacher-TOP student-NOM leave-COMP do-PST-IND a. 'The teacher made the student leave.' b. 'The student made the teacher leave.' (45) (Gerdts' (58)) Emeni-nun ai-ka ttek-ul mek-key hay-ess—ta. Mother-TOP child—NOM rice cake-ACC eat-COMP do-PST—IND a. 'The mother had the child eat rice cake.' ?b. 'The child made the mother eat rice cake.' She claims that the non-topicalized counterpart sentence for the (b) reading in (44) is not (46), but (47). (46) Haksayng-i sensayngnim—i ttena-key hay-ess-ta. student-NOM teacher-NOM leave-COMP do-PST-IND 'The student made the teacher leave.' (47) Haksayng-i sensayngnim-ui ttena-key hay-ess-ta. student-NOM teacher-ACC leave-COMP do-PST-IND 'The student made the teacher leave.' Since the causee in the accusative case as in (47), in her analysis, is a term, direct object, so it can be a plain topic. She concludes that the first nominal in the nominative case is a final term because it can be a plain topic, and the second nominal in the nominative case is not a final term, but a chomeur, since it cannot be a plain topic. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (B.2.) A Critique of the Monoclausal Analysis by Gerdts (1990) In this subsection, we will give a critique of the monoclausal analysis by Gerdts (1990), which is completely opposite from ours. While we argue that there is no gramma tical relation change in the syntactic causative because the syntactic causative is biclausal at every level of represen tation, Gerdts claims that the matrix clause and the embedded clause merge, and as a result, some change occurs in the grammatical relation which some of the arguments held in the original biclausal structure. Gerdts says that for those who accept (33)-(35) as gramma tical, there is no ban on intervening lexical material between the predicate of the embedded clause (/in Gerdts', the predi cate of the underlying downstairs) and the matrix verb. This means that any element in the sentence can be inserted between the two verbs. However, this does not seem to be true. As we will see, the elements which can intervene between the two verbs are only those which belong to the main clause. We will take (34) as an example. (34) John-eykey ttek-ul mek-key nay—ka hay—ess—ta. -DAT rice case-ACC eat-COMP I-NOM do-PST-IND 'I made John eat the rice cake.' The basic word order for (34) is assumed to be the next sentence. 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (48) Nay-ka John-eykey ttek-ul mek-key hay-ess-ta. -NOM -DAT rice cake-ACC eat-COMP do-PST-IND 'I made John eat the rice cake.' If there is no ban in inserting any element interverbally as argued in Gerdts (1990), then 'ttek (rice cake)' should be able to occur between the two verbs. This is, however, not possible at all. *(49) Nay-ka John—eykey mek-key ttek-ul hay-ess-ta. I—NOM -DAT eat-COMP rice cake-ACC do-PST-IND 'I made John eat the rice cake.' And the following orders are not possible, either. *(50) Nay-ka mek-key John-eykey ttek-ul hay-ess-ta. I—NOM eat-COMP -DAT rice cake-ACC do-PST-IND ' I made John eat the rice cake.' *(51) John-eykey mek-key Nay-ka ttek—ul hay-ess-ta. -DAT eat-COMP I-NOM rice cake-ACC do-PST-IND 'I made John eat the rice cake.' It cannot be claimed, therefore, that there is no ban on intervening elements between the two predicates. The cons traint here is, we argue, that only the elements which belong to the main clause can be inserted between the verbs. The scrambling of the elements in the same clause is quite flexible except when there are two elements in the same case. (52) John-i Mary-eykeyse ku chayk-ul ecey pat-ass-ta. -NOM -from the book-ACC yesterday receive-PST-IND 'John received the book from Mary yesterday.' 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. There is no case conflict among the elements in (52). There fore, the elements in the sentence are scrambled around freely without losing the basic meaning, as long as the predicate is the final element. (53) John-i ku chayk-ul ecey Mary-eykeyse pat-ass-ta. (54) John-i ecey Mary-eykeyse ku chayk-ul pat-ass-ta. (55) Mary-eykeyse ecey John-i ku chayk-ul pat-ass-ta. (56) Ecey ku chayk-ul John-i Mary-eykeyse pat-ass—ta. 'John received the book from Mary yesterday.' If (34) is really monoclausal at the final level, and there is no ban for any element to occur interverbally, then there is no reason why 'ttek (rice cake)' in (49), which is the theme argument of the embedded clause, is not allowed to appear interverbally. The constraint here, again we argue, is that only the elements which belong to the main clause can occur between the embedded verb and the matrix verb. (57) Nay-ka ttek-ul mek-key John-eykey hay-ess-ta. I—NOM rice cake-ACC eat-COMP -DAT do-PST-IND 'I made John eat the rice cake.' (58) John—evkey ttek—ul mek-key nay—ka hay-ess-ta. -DAT rice cake-ACC eat-COMP I-NOM do-PST-IND 'I made John eat the rice cake.' (59) Ttek-ul mek-key nay-ka John-eykey hay-ess-ta. rice cake-ACC eat-COMP I-NOM -DAT do-PST-IND 'I made John eat the rice cake.' The following sentence has a ditransitive embedded clause. (60) John-i Mary-eykey [0 ku chayk-ul Bill-eykey phal]-key -NOM -DAT the book-ACC -DAT sel1-COMP hay-ess-ta. do-PST-IND 'John made Mary sell the book to Bill.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In Gerdts' union analysis, the causee in the dative case is the indirect object, and the recipient of the embedded clause 'Bill' is a chomeur. We claim that when two elements are in the same case, they are not allowed to cross over each other. However, if there is no ban on the insertion of elements interverbally as Gerdts claims, then 'Bill' should be able to be placed interverbally without changing the basic meaning. (61) John-i Mary-eykey ku chayk-ul phal-key Bill-eykey -NOM -DAT the book-ACC sell-COMP -DAT hay-ess-ta. do-PST-IND ?(a) 'John made Bill sell the book to Mary.' *(b) 'John made Mary sell the book to Bill.' The only interpretation for (61v is (a), not (b). If there is no ban on inserting elements interverbally, the (b) meaning is expected for the sentence (61). Therefore, the structure of (61) is not (62a), but (62b), which is different from (60) in that 'Mary' is a part of the matrix clause, and 'Bill' is a part of the embedded clause in (60), but 'Mary' is a part of the embedded clause and 'Bill' a part of the matrix clause in (62b). (62) a. John-i Mary-eykey [ku chayk-ul phal—key Bill—eykey] hay-ess—ta. b. John-i [0 Mary-eykey ku chayk—ul phal-key] Bill—eykey hav-ess-ta. This again shows that there is a restriction on inserting elements between the two verbs. The theme argument in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. embedded clause, even in this case, cannot be placed inter verbally just as it is not possible in case of the sentence (49), in which the embedded clause is transitive. Thus the sentence (63) is ungrammatical. *(63) John-i Mary-eykey Bill-eykey phal-key ku chavk-ul -NOM -DAT -DAT sell-COMP the book-ACC hay—ess—ta. do-PST-IND 'John made Mary sell the book to Bill.' It cannot be argued, therefore, that there is no ban on intervening elements between the two predicates. The const raint, as we have mentioned elsewhere, is that only the ele ments which belong to the main clause can come between the verbs. (64) John-i [0 Bill-eykey ku chayk-ul phal]-key Mary-eykey -NOM -DAT the book-ACC sell-COMP -DAT hay-ess—ta. do-PST-IND 'John made Mary sell the book to Bill.-' (65) Mary-eykey [0 ku chayk-ul Bill-eykey phall-key John-i -DAT the book-ACC -DAT sell-COMP " -NOM hay-ess-ta. do-PST-IND 'John made Mary sell the book to Bill.' (66) [0 Bill-eykey ku chayk-ul phal]-key John-i Mary-eykey -DAT the book-Acc sell-COMP -NOM -DAT hay-ess-ta. do-PST-IND 'John made Mary sell the book Bill.' In (64), the recipient of the matrix clause is placed inter verbally, in (65), the subject of the matrix clause, and in (66), the subject and the recipient of the matrix clause. 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The second thing which Gerdts uses to argue for the clause union analysis is 'plain topicalization'. She claims that only terms— subject, direct object, and indirect object— -can become the plain topic. So when there are two nominals which are both in accusative case, then only the genuine direct object can be the plain topic as in (38a-b). (67) (=36) Double Accusative John-i na-lul ttek-ul mek-key hay-ess-ta. -NOM I-ACC rice cake-ACC eat-COMP do-PST-IND 'John made me eat the rice cake.r (68) (=38a-b) *a. Ttek-un John-i na—lul mek-key hay-ess-ta. rice case-TOP -NOM I-ACC eat-COMP do-PST-IND 'John made me eat the rice cake.' b. Na-nun John-i ttek-ul mek-key hay-ess-ta. I—TOP -NOM rice cake-ACC eat-COMP do-PST-IND 'John made me eat the rice cake.' In her analysis of sentences like (67), 'na (I)' is the genuine direct object, and 'ttek (rice cake)' is not. So 'ttek' cannot be a plain topic as in (68a), and 'na' can become a plain topic as in (68b). She expands this test to the sentences with causee in nominative case, and causee in dative case. (69) (70) 85 Sensayngnim-un haksayng-i ttena-key hay-ess-ta. teacher-TOP student-NOM leave-COMP do-PST-IND (i) 'The teacher made the student leave.' (ii) 'The student made the teacher leave.' Emeni-nun ai-ka ttek-ul mek-key hay-ess-ta. mother-TOP chile-NOM rice cake-ACC eat-COMP do-PST-IND (i) 'The mother had the child eat rice cake.' (ii) 'The child made the mother eat rice cake.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In her analysis, 'sensayngnim (teacher)' in (69), and 'emeni (mother)-' in (70) are the subject in each clause. So they can become plain topics. Gerdts claims that interpre tation (ii) of (69) is not from (71), but from (72). (71) (Gerdts' (59)) Haksayng—i sensayngnim—i ttena—key hay—ess—ta. student-NOM teacher-NOM leave-COMP do-PST-IND 'The student made the teacher leave.' (72) (Gerdts' (60)) Haksayng-i sensayngnim-ul ttena-key hay-ess-ta. student-NOM teacher-ACC leave-COMP do-PST-IND 'The student made the teacher leave.' In her analysis, 'sensayngnim (teacher)' in (71) is a non- term, that is a chomeur, but 'sensayngnim' in the accusative in (72) is a direct object of the clause. Therefore, the unscrambled version of the interpretation (ii) in (69) is (72), which allows the direct object 'sensayngnim' to be scrambled to the initial position to be a plain topic. (73) (Gerdts' (75)) «a=eykey—nun Mary—ka John—eykey kii-ul mwut—key I—DAT—TOP -NOM -DAT way-ACC ask-COMP hay-ess-ta. do-PST-IND (i) 'Mary had me ask John the way.' *(ii) 'Mary had J o h n ask me the way.' (74) (Gerdts' (76)) John-eykey-nun Mary-ka na-eykey kil-ul mwut-key -DAT-TOP -NOM I—DAT way-ACC ask-COMP hav—ess-ta. do-PST-IND *(i) 'Mary had me ask John the way.' (ii) 'Mary had John ask me the way.' In Gerdts' analysis, 'na (I)' is a term, indirect object, but 'John', which is also in the dative case, is a non-term. 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Therefore, 'na' can be a plain topic as in (73), but 'John' cannot as the interpretation (73ii) shows. So 'John' in (74), in her analysis, is the indirect object, not 'na (I)'. However, if we look at the relevant examples carefully, then we can see that the problematic sentences like (38a), and the problematic interpretations like (73ii) and (74i) all have violations of the following double case constraints (Refer to K.A. Kim (1992) for more detail (cf. Jo, 1986). (75) Constraint for Double Nominatives in Causatives The second nominal in the nominative case cannot cross the first nominal in the nominative case unless the whole embedded clause which the second nominal in the nominative case belongs to moves to the left of the first nominal in the nominative case. (76) Constraint for Double Datives in Causatives The second nominal in the dative case cannot cross the first nominal in the dative case unless the whole embedded clause which the second nominal in the dative case belongs to moves to the left of the first nominal in the dative case. (77) Constraint for Double Accusatives in Causatives The two nominais in the accusative case cannot cross each other. (exception: If the verb is 'cwuta (give' or contains 'cwuta (give)' as a part of a compound verb, then the crossing between the two nominais in the accusative case is marginally allowed for some speakers.) If we change, for example, the order between the two nominais in the accusative case as in (36), then we get the following ungrammatical sentence. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (78) *a. John-i ttek-ul na—lul mek—key hay-ess-ta. -NOM rice cake-ACC I-ACC eat-COMP do-PST-IND 'John made me eat the rice cake.' *b. Ttek-ul John-i na-lul mek-key hay-ess—ta. rice cake-ACC -NOM I-ACC eat-COMP do-PST-IND 'John made me eat the rice cake.' In the above sentences (78a—b), the order between the two nominais in the accusative case is reversed, violating the double accusative case constraint (77). Additionally, the reason why the interpre-tation (74i) is not possible is because the two nominais in the dative case cannot cross each other individually as this violates the restriction (76). Therefore, the first one is always the causee, and the second one is the recipient argument of the embedded clause. In conclusion, we might say 'predicate fusion', and 'plain topi- calization' cannot be taken as sufficient evidence to claim that the syntactic causative is monoclausal. Finally, it can also be shown that 'subject honorification' is not a legitimate test for the monoclausal analysis, either. For the sake of convenience, examples (40)-(41) are repeated here. (79) (=40) Sensayngnim-i haksayng—i ttena—key ha—shi-ess—ta. teacher-NQM student-NOM leave-COMP do-HON-PST-IND 'The teacher made the student leave.' (80) (=41) Haksayng-i sensayngnim-i ttena-key ha-(*shi)-ess-ta. student-NOM teacher-NOM leave-COMP do-(HQN)-PST-IND 'The student made the teacher leave.' 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gerdts' claim is that the subject honorification marker '—shi— ' can be assigned to the verb 'hata (do)' in (79) because the causer 'sensayngnim (teacher)' is the final subject, and hono rable, but subject honorification is not possible in (80) because 'haksayng (student)' which is the subject is not honorable. However, this argument cannot be maintained, since the subject honorification marker '-shi-' can also be affixed to the first predicate as in (81). (81) Haksayng-i sensayngnim-i ttena-shi-key hay-ess—ta. student-NOM teacher-NOM leave-HON-COMP do-PST-IND 'The student made the teacher leave.' If the causee 'sensayngnim' is not the subject, then '-shi-' should not be affixed to the first predicate. If we deal only with the surface structure, then '-shi-' should not be affixed to the predicate of the embedded clause, since the causee is not a subject in the Gerdts' monoclausal analysis. Nevertheless, this might not constitute sufficient evidence for the biclausal analysis because it may be assumed that '— shi-' morpheme affixation to the embedded predicate was done in the underlying structure, then shi-' appears in the surface structure (Comrie, personal communication). So we may say that the 'subject honorif ication' test is at best neutral. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C. Evidence for the biclausalitv (C.l) NPs, pronouns, and a reflexive 'casin' in the embedded subject position (C.1.1.) NPs In English, a name (/NP) cannot be bound by any antecedent (binding condition (C) in Chomsky, 1981). This restriction causes the next sentence to be ungrammatical. *(1) John^ thinks that John., is happy. The second 'John' which is the subject of the embedded clause is bound by the first 'John', violating binding condition (C). This restriction, however, does not seem to hold universally. For instance, in Thai or in Vietnamese, this restriction might not be operable. (2) (Lasnik (1986);s (20)) coon khit waa coon chalaat *Johni thinks that Johni is smart. (3) (Lasnik (1986)'s (21)) John tin John se thang *Johni think Johni will win. The second 'John' in the Thai data (2) and in the Vietnamese data (3) is bound by the first 'John*'. But they are not un grammatical unlike their English counterparts. Japanese shares with Thai and Vietnamese the fact that a name can be bound. 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (4) (Oshima (1879, p. 431)) (a ) John^-ga [s, Mary-ga John^o nikunde~iru to] omot-te- John—NOM —NOM —AOM hates that thinks iru (koto) is *,Johni thinks that Mary hates John^ ' (b) John^-ga John^no hon—o mottekita (koto) -NOM -GEN book-ACC brought *' John^ brought Johni ■ ' s book. ' The second 'John' is bound by the first 'John' in the above Japanese data. Therefore, the sentence are expected to be ungrammatical, violating binding condition (C). But they are acceptable in Japanese. This is also what we find in Korean. (5) a. Mary^nun [Mary^uy sensayngnim-i cal sayngkiess-ta]- -TOP -GEN teacher-NOM well look-IND- ko sayngkahan-ta. COMP think-IND */Maryi thinks that Mary^s teacher is handsome.' ?b. John^un [John^i ceypep ttokttokha-ta]—ko -TOP —NOM fairly smart-IND-COMP ^ ^ believe-IND *'Johni believes that Johni is quite smart.' The second 'Mary' in (5a) is a part of the subject NP of the embedded clause, and is bound by the first 'Mary'. In (5b), the second 'John' is the subject of the embedded clause, and is bound by the first 'John'. The latter sentence is not very natural, but not ungrammatical. (This may be due to the fact that the second 'John' is occupying the embedded subject posi tion.) Both of these examples are interclausal. Even in a 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. single clause, a name can be bound as in the Japanese example (4b) . (6) John.j—un cikum John^uy pang-eyse ca-koiss-ta. —TOP now —GEN room—in sleep—PROG—IND *'Johni is sleeping in John/s room.' The second 'John' is bound by the first one. However, there is some restriction on binding a name (/NP). The local domain for the name (/NP) to be free is an NP, or S (cf. Hong, 1985). (7) A name (/R-expression) in Korean must be free in its local domain; the local domain for a name is an NP or S. This is equivalent to binding condition (B) , and Japanese also has the same restriction for names according to Oshima (1979). This means that unless a name is inside a noun phrase (as the possessor), it must not be bound by any argument in a clause. The next examples are all ungrammatical because they violate this restriction. (8) *a. Johni-i Joh^-eykey chayk-ul cwu-ess-ta. -NOM -DAT book-ACC give-PST-IND ♦'John, gave a book to John^.' *b. John,-i John,-ul miwehan-ta. -NOM -ACC hate-IND ♦'John- hates John^' ♦c. John^—i John.j-eykeyse ku chayk-ul pat—ass-ta. -NOM -from the book-ACC receive-PST-IND ♦'Johni received the book from Johni. ' With these facts in mind, the following restrictions may be formulated for names (/NPs). 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (9) a name (/R-expression) may, in a single clause, not be bound, except when it appears as the possessor noun in an NP. (10) a name cannot appear more than once in a single clause, except as the possessor noun in an NP. These restrictions should apply to the so-called syntactic causatives if the sentence were to be monoclausal. If the causee in dative or accusative case was the subject of the embedded clause at one level and becomes an argument other than subject in the final level, having the two clauses collapsed info one, then there should not be a subject position in the embedded clause. The next example with the causee in the dative case shows that the subject position of the embedded clause still exists and can be filled by a proper name which is coreferential with the causee. (11) Kyoswunnim^-i Mary.-eykey [Mary*i„/*k-ka ku il-ul professor-NOM J -DAT -N0MJthe work-ACC nayil-kkaci kkutnayj-key imi ha—shi—ess-eyo. tomorrow-until finish-COMP already do-HON-PST-POL 'The professor already brought it about to Mary that she (lit. Mary) finish the work by tomorrow.’ If there is only one clause in the sentence (11), then it should be ungrammatical because ic violates the restriction (10). The second 'Mary' must be within a noun phrase as the possessor if the sentence is monoclausal in order not to vio late the restriction (11). The sentence, however, is grammati cal even though the second 'Mary' is not within a noun phrase. The fact that this sentence is grammatical shows that there Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. must be two clauses in the sentence; the first 'Mary' belongs to the matrix clause, and the second one belongs to the embedded clause, functioning as the subject of the clause, thus not violating the single clause restriction (10). The "subjecthood" tests which are used in section (C) show that the second 'Mary' is the subject of the embedded clause. (12) Joh^-i kyoswuninu-kkey [kyoswunim, ,*k—kkeyse ku il-ul -NOM professor-DAT(HON) professor-Nom(HON) the work-ACC [Qj umak-ul tulu-shi-myense] honcase ha-shi]-key mus ic-A.CC hear-HON-which alone do-KON-COMP hay-ess-ta. ,Johni brought it about to the professor^ that he^ (lit.professor) do the work alone while 0j listening to the music.' The second 'kyoswunim (professor)' has the subject honorific marker 'kkevse', and it controls the covert subject of the '- myense (while)' clause. It also triggers the appearance of the subject honorific marker '-shi-' in the embedded clause as well as in the '-myense' clause whose subject is obligatorily ccr.trcllcd by a subject. 'YaksoKuaua ^iruiu^&cj * xs one ox tne verbs which allow only subject control of the zero anaphora in the subject position of the embedded clause. This kind of subject control structure can be embedded in the embedded clause. (13) John-i kyoswunim--kkey [ kyoswunim./ik-kkeyse [0-y*^ ku -NOM prof essor-DAT (HON) professor—NOM (HON) the chayk-ul ilu-shi-keyess-ta]-ko yaksokha-shi]-key book-ACC read—HON-IMPERF—IND-COMP promise-HON-COMP hay-ess-ta. /Johni brought it about to the professor that he (lit. professor) promise to read the book.' 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The second 'kyoswunim'’ controls the non-overt subject of the clause led by the complementizer '—ko1, also triggering the subject honorification for the verbs 'ilkta (read)', and 'yaksokhata (promise)'. These tests show that the second 'kyoswunim' is indeed the subject of the embedded clause. The nest sentence has the causee in the nominative case. (14) Kyoswunim-i [Mary-ka ku il-ul ha]-key hay-ess-ta. professor-NOM —NOM the work-ACC do-COMP do-PST-IND 'The professor instructed that Mary do the work.' The bracketed string in the Korean is the content of the pro fessor's instruction, as the embedded clause in its English counterpart is. Therefore, 'Mary' does not belong to the ma trix clause, but to the embedded clause, being the embedded subject. Another name 'Mary', which is coreferential with 'Mary' in the nominative case can optionally occur in the interverbal position. The fact that 'Mary' in (14) is the embedded subject is proven by all the "subjecthood" tests in section (C). (15) Kyoswunim-i [Maryi~ka ku il-ul nayil-kkaci professor=NOM NGn tine wurK—a w tomorrow—untxi kkutnay]-key imi Mary.-eykey ha-shi-ess-eyo. finish-COMP already -DAT -HGN-PST-POL 'The professor already brought it about to Mary that she (lit. Mary) finish the work by tomorrow.' This sentence is a scrambled version of (11), in that 'Mary' in the dative case may be scrambled to the interverbal posi tion from the position immediately preceding the embedded 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. clause. Thus we claim that in the case of (14), the recipient in the matrix clause which is coreferential with the embedded subject may not be overtly expressed in order to avoid redun dancy. The causee in the following sentence is in the accusative case, but still there can be an overt embedded subject noun phrase which is coreferential with the causee. (16) Kyoswunim-i Mary^man-ul [0i amwu-uy towum-to pat-ci professor-NOM -only-ACC anybody-GEN help-also receive ani-ha-ko, Maryi/)t--ka honcase ku il-ul kkutnay ]-key -CON -NOM alone the work-ACC finish-COMP ha-shi-ess-ta. do-HON-PST-IND 'The professor instructed Mary to finish the work by herself without getting any help from anybody else. ' This means that the first 'Mary' belongs to the main clause, and the second 'Mary' belongs to the embedded clause, being the subject of the clause in order not to violate the restric tion (10). The above examples such as (11), (12), and (16) with two coreferential NPs allowed to be in the sentence show that the syntactic causative in Korean, indeed, is biclausal. An additional piece of evidence is provided by the fact that if the causee in either dative or accusative case is more than one person, then it can be split into two or more, each being the subject of a separate embedded clause. If the causee in dative or accusative case was the subject of the embedded clause, and became either direct object or indirect object in 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the final union clause as claimed by Gerdts (1990) within the framework of Relational Grammar, there should not be any empty subject position in the embedded clause, and it is not possi ble to have a part of the causee as the embedded subject in each embedded clauses. (17) Kyoswunim-i Mary-wa Bill-eykey [Mary-man-i yenge-lul professor-NOM -CON -DAT -only-NOM English-ACC kongpwuha-ko, Bill-un swuhak-ul kongpwuha]-key study-CON -TOP mathematics-ACC study-COMP hay-ess-ta. 'The professor instructed Mary and Bill that only Mary study English, and Bill study mathematics.' There are two embedded clauses in the sentence (17) connected by the morphological conjunction '-ko'. 'Mary' is the subject of the first conjunct, and 'Bill' the subject of the second. These two subject NPs are parts of the causee, and they together form the causee which is the whole set 'Mary-wa John (Mary and John)'. This means that the causee could not have been the subject of either of the two embedded clauses at all levels of representations because the independent subject of each conjunct which is the embedded clauses exists at all levels, and neither of the embedded subjects xs corexerentiai with the causee. Therefore, the monoclausal claim that the causee was once the embedded subject, and becomes seme other argument such as direct object or indirect object at the union level cannot be maintained because the causee which is the recipient argument of the matrix clause and the subject of each embedded clause are both present at the same level. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Especially in this example the causee and the subject of the each conjunct are not coreferential. The same point can be made even when the causee is in the accusative case. (18) Kyoswunim-i Mary-wa Bill-ul [Mary-man-i yenge-lul professor—NOM —CON -ACC --only—NOM English-ACC kongpwuha—ko, Bill-un swuhak—ul kongpwuha]-key study-CON -TOP mathematics-ACC study-COMP hay-ess—ta. 'The professor instructed Mary and Bill that only Mary study English, and Bill study mathematics.' The subjects of the two conjuncts in the embedded clause are parts of the causee. This means that the causee in the accusative case does belongs to the matrix clause at every level just like the causee in the dative case as in (17), and there is a separate subject for each conjunct in the embedded clause. The examples (17) , (18) have a coordination structure in the embedded clause, and it is also possible to have a subordina tion structure in the embedded clause, and split the causcc into more than one subject. (19) Kyoswunim—i Mary—wa Bill—eykey [Mary—ka ku chayk—ul professor-NOM -CON -DAT -NOM the book-ACC ilk-nun tongan, Bill—an talun chayk-ul ilk]-key read-while -TOP another book-ACC read-COMP ha—shi-ess-ta. 'The professor instructed Mary and Bill that Bill read another book while Mary is reading the book.' The subject of the main clause within the bracketed string is 'Bill' which is a part of set, forming the causee, and 'Mary', 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. which also is a part of the set which forms the causee, is the subject of the subordinate clause led by 'tcngan (while)' . The causee can be also in accusative case, maintaining the rest of the sentence same. (20) Kyoswunim-i Mary-wa Biii-ul [Mary-ka ku chayk—ul professor-NOM -CON -ACC -NOM the book-ACC ilk-nun tongan, Bill-un talun chayk-ul ilk]-key read while -TOP other book-ACC read-COMP ha—shi-ess—ta. 'The professor instructed Mary and Bill that Bill read another book while Mary is reading the book.' The fact that in the embedded clause, 'Mary' and 'Bill' are the subject of each clause, and the causee 'Mary-wa Bill (Mary and Bill)' is a part of the matrix clause shows that the sentence is biclausal at all levels because again the causee cannot have been the subject of the matrix clause or the subordinate clause in the bracketed embedded clause. (21) Kyoswunim—i Mary—wa Bili-eykey(/—ul) [Mary-ka professor-NOM -CON -DAT(/-ACC) -NOM Bill-eykey ton-ul ewun-hwu-ey, Bill-i Mary-eykey ku -DAT money-ACC give—after -NOM -DAT the chayk-ul cwu]-key hay-ess-ta. book-ACC give-COMP 'The professor instructed Mary and Bill that Bill giT /e the book to Mary after she give him money. ' Again in (21) 'Bill' is the subject of the matrix clause within the bracketed clause and 'Mary' is the subject of the subordinate clause which is led by 'hwuey (after)', and 'Bill' is the subject of the matrix clause in the embedded clauses. We, thus, claim that there was no change in grammatical 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relation for the causee, and no collapsing of clauses into one. (22) Kyoswunim-i Mary-wa Bill-evkey(/-ul) [Mary-ka chayk-ul professor-NOM -CON -DAT(/-ACC) -NOM book-ACC twu kwen sa-myen, Bill-un sey kwen sa]-key hay-ess-ta. two CL buy-if —TOP three CL buy-COMP 'The professor instructed Mary and Bill that Bill buy three books if Mary buys two books.' (23) Kyoswunim-i Mary-wa Bill-eykey(/-ul) [Mary-ka Bill-ul professor-NOM -CON -DAT(/-ACC) -NOM -ACC ttayli-kena, Bill-i Mary-lul ttayli]-key hay-ess-ta. hit-or -NOM -ACC hit-COMP •'The professor instructed Mary and Bill that Mary hit Bill or Bill hit Mary.' The fact that 'Mary' and 'Bill7 occupy the subject position in each clause in the embedded clauses, in (22) with an if conditional clause and in (23) with a coordination structure, having a conjunction '-kena (or)', and they are not coreferen tial with the causee leaves no room for the causee to have been the subject of any of the two clauses in the bracketed part. Thus we claim that the causee belongs to the matrix clause, and the syntactic causative in Korean is biclausal at every level. The causee can be a part of the whole set which is used as the subject of the embedded clause. (24) Kyoswunim-i Maryi-eykey(/-lul) [kunye casi^-kwa Bili-i professor-NOM -DAT(/—ACC) she self—CON —NOM ku nolay-lul hamkkey pwulu]-kev hay-ess-ta. the song-ACC together sing 'The professor instructed Mary that she(lit. herself) and Bill sing the song together.' 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'Mary' the causee is one of the two people who form the subject of the embedded clause together. If one claims that 'Mary7 was once the subject of the embedded clause, then the structure of that clause should have been as following, which is ungrammatical. *(25) Mary-ka Mary casin-kwa Bill-i ku nolay-lul hamkkey -NOM self-CON -NOM the song—ACC together pwulle-ess-ta. s ing-PST-IND This sentence shows again that the causee belongs to the main clause at all levels. There is the subject of the embedded clause which is not exactly coreferential with the causee, so the causee 'Mary7 could not be the subject of the embedded clause. Again in case the causee 'Mary' is one of the two who form the subject of the embedded clause, the recipient 'Mary' come between the two verbs as in (26). (26) Kyoswunim-i [Mary-wa Bill-i ku il—ul ha]-key professor-NOM -CON =«0M the work-ACC do-CGMP Mary-eykey ha-shi-ess-ta. -DAT 'The professor instructed Mary that Mary and Bill do the work together.' In (26), 'Mary' in the dative case is a part of the subject of the embedded clause, which is 'Mary-wa Bill (Mary and Bill)' . If the recipient argument, 'Mary-eykey' in the case of (26), is not overtly expressed, then it is assumed to be coreferential with the subject of the embedded clause. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The examples (24) and (26) show that as long as the causee is a part of the subject of the embedded clause, the sentence is grammatical. This is true with the sentences in which the causee is a larger set than the subject of the embedded clause. (27) Kyoswunim-i Mary-wa Bill-eykey(/-ul) [Mary-man-i ku professor-NOM -CON -DAT -only-NOM the chayk-ul ilk]-key ha-shi-ess-ta. book-ACC read-COMP do-HON-PST-IND 'The professor instructed Mary and Bill that only Mary read the book.' The subject of the embedded clause is a part of the larger set which is the causee, and this sentence is grammatical. Therefore, the claim that the causee was the subject of the embedded clause at one point, and an argument of the single Union clause cannot be supported because at the final level there are the recipient 'Mary-wa Bill (Mary and Bill)' and the subject of the embedded clause 'Mary—man (only Mary)', co existing at the same time. The causee 'Mary—wa Bill(Mary and Bill)' is a part of the matrix clause at all levels, and 'Mary-man-i (only Mary)' is the subject of the embedded clause. In this case just like (26), the causee can again come between the two verbs. (28) Kyoswunim-i [Mary-man-i ku chayk-ul ilkj-key professor-NOM -only-NOM the book-ACC read-COMP Mary-wa Bill-eykey imi ha-shi-ess-ta. -CON -DAT already do-HON-PST-IND 'The professor instructed Mary and Bill that only Mary read the book.' 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. If the subject of the embedded clause is totally irrelevant from the causee, the acceptability of the sentence becomes rather low. ?(29) Kyoswunim-i Mary-eykey(/-ul) [Bill-i professor-NOM —DAT/—ACC -NOM ku chayk-ul ilk]-key hay-ess-ta. the book-ACC read-COMP do-PST-IND 'The professor instructed Mary that Bill read the book.' The causee 'Mary' is not coreferential with the embedded subject 'Bill', nor does it hold a part—whole relationship with it, thus lowering the acceptability. However, if there is certain relationship between the causee and the embedded clause where the causee can make the embedded subject initiate the embedded event, then there is no such problem. Suppose the causee is the 'class president' and the embedded subject is the students in the class. The teacher wants the class presi dent to tell the students to read books. With this kind of contest, the next sentence has no problem with acceptability. (30) Sensayngnim-i pancang-eykey(/-ul) [haksayng—tul—i teacher-NOM class president-DAT(/-ACC) student-FIr-NCM chayk-tul-ul ilk]-key hay-ess-ta. book—PL—ACC read-COMP do-PST—IND 'The teacher instructed the class president that the students read books.' The embedded clause is not coreferential with the causee, but this sentence is grammatical. This is possible not only when the embedded subject is filled with an overt element, but also when the embedded subject is empty. 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (31) Sensayngnim-i pancang-eykey(/-ul) [0 sakwa-lul han apple-ACC one *ku-ka (he-NCM) haksayngtul-i (students-NOM) saram-ey han kay sshik mek]-key hay-ess-ta. person—each one CL each eat-COMP 'The teacher instructed the class president that (the students) eat one apple per-person.' The phrase 'han saram-ey han kay sshik (one per-perscn)' shows that the embedded subject cannot be singular. Thus the covert embedded subject is not coreferential with the causee. When it is overt, a pronoun 'ku (he)' cannot appear, but 'haksayngtul (students)' can be a good candidate as shown. The same point will be made in the next chapter when we look at control phenomenon in Korean. In this subsection, the following are what we have claimed. (32) a. Regardless of which case the causee is in, the syntactic causative in Korean is biclausal at any level. b. If the subject of the embedded clause in a syntactic causative is not expressed, then it is implicitly understood to be coreferential with the causee in dative or accusative case, unless expressed differently. c. if the causee is either in dative or accusative case, and the subject position of the embedded clause is filled by a noun phrase, then it should either be coreferential with the causee, or hold a part-whole relationship with the causee. Otherwise, the acceptability of the sentence suffers unless the context makes it clear. d. In case that the causee is in nominative, then it is, we claim, the embedded subject (proven by ail the "subjeethood" tests in section (C)). 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (C.1.2.) Pronouns and a Reflexive pronoun 'casin' Pronouns in Korean seem to obey the binding condition (B) (Chomsky, 1981). (33) Binding condition (B) (Chomsky, 1981) A pronoun must be free in its local domain: the local domain for a pronoun is an NP or S. According to this restriction, a pronoun cannot be bound by an antecedent in a single clause, unless it is within a noun phrase. (34) a. John^un kui-uy cip-eyse chayk-ul ilk-koiss-ta. -TOP he-GEN house-at book-ACC read-PRQG-IND ' Johni is reading a book at his^ house. ' *b. John^i ku.j-eykey chayk-ul phal-ass-ta. -NOM he-DAT book-ACC sell-PST-IND *,johni sold the book to him.,. ' *c. John.j-i ku.j-lul coahan-ta. -NOM he-ACC iike-IND *,johni likes him^' *d. John-i ku.j-wa hamkkey ku il-ul hay-ess-ta. -NOM he-with the work-ACC do-PST-IND *' John^ did the work with himi. ' A pronoun 'ku (he)' in (34a) is within a noun phrase, and free in it, which is the local domain. The sentence, therefore, is grammatical. However, the pronoun in (34b—d) is not within a noun phrase, so it should be free within the clause (/S) according to binding condition (B). But it is bound by 'John' in each sentence, and thus creates an ungrammatical sentence. Just like NPs, a ronoun cannot appear if there is a potential 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. antecedent within the same clause, unless the pronoun is within a noun phrase. So we may formulate the restriction for the behavior of the pronouns in a single clause as following: (35) A pronoun may not be bound within a clause, unless it appears as the possessor noun within an NP. (36) A pronoun may not occur in a clause when there is a potential antecedent, except as the possessor within an NP. We assume that binding condition (B) in (33), along with the above restriction, helps to confirm the biclausality of the syntactic causative in Korean (cf. H.S. Choe, 1988). If the syntactic causative sentence (37) is monoclausal, then a pronoun is expected to be free in the sentence, unless the pronoun is within a noun phrase. (37) John.j-i Mary,-eykey/-lul [0- ku^lul nayil manna]-key -NOM —DAT/-ACC he—ACC tomorrow meet-COMP hay-ess—ta. do-PST-IND 'Johni instructed Mary that (she) meet himi tomorrow.' Suppose the syntactic causative sentence (37) is monoclausal. Then 'ku (he) ' is expected to be free within the sentence, but it is bound by the antecedent 'John'. If it is monoclausal, the sentence is expected to be ungrammatical. However, the sentence is not ungrammatical. Suppose the sentence is biclausal, then there is no violation of the binding condition (B), because the pronoun is free within its local domain which 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is the embedded clause ( /S ) . This can also be applied even when the causee is in the nominative case. (38) Jokn.j-i [Mary-ka ku^lul ttayli]-key hay-ess-ta. -NOM -NOM he-ACC hit-COMP do-PST-IND ,Johni instructed that Mary hit hin^.' If this sentence is monoclausal as claimed by Gerdts (1990), then it should be ungrammatical because the pronoun is bound by an antecedent 'John' which is in its local domain, S. However this sentence is not ungrammatical. Again there is no violation of binding condition (B) if the sentence is bi- clausal as we claimed because the pronoun 'ku (he)' is free in its local domain which is the embedded clause. The next sentence violates the restriction (36), if it is analyzed as monoclausal. (39) John.j-i Mary-eykey(/-lul) [kunye^ka ku chayk—ul ilk] — -NOM —DAT/—ACC she-NOM the book-ACC read- l/atr hay— a ( COMP do-PST-IND 'John instructed Maryi that shei read the book. ' 'Mary' is the potential antecedent for the pronoun 'kunye (she)' in (39), and the pronoun is not within a noun phrase as the possessor. So according to the restriction (36), this sentence should be ungrammatical if it is monoclausal. But it is grammatical. Suppose, however, that this sentence is biciausal as we claim. Then there is no problem because the 1Q7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. antecedent 'Mary' belongs to the matrix clause, and the pronoun belongs to the embedded clause, not violating the restriction in (36). A reflexive pronoun 'casin (self) ' in Korean can be bound across the clausal boundary as well as within the clause in which the antecedent is also located8. (40) a. John.j-i casin^ul salanghan-ta. -NOM self-ACC love-IND 'Johni loves himself^ ' b. John.j-i casin.j-uy chinkwu-eykey chayk-ul cwu-ess-ta. -NOM self—GEN friend-DAT book-ACC give-PST-IND 'Johni gave a book to hisi (lit. self's) friend.' In the above sentences, the reflexive 'casin' and its antecedent are both in the same clause, and the former being bound by the latter, which is considered the typical case of reflexive pronouns crosslinguistically. (41) a. John.j-i [casin^i chencay-la]—ko sayngkakhan-ta. -NOM self-NOM genius—IHD-CQMP think-IND 'John^ thinks that he (lit. self) is a genius.' b. Kyoswunim-i Mary-eykey [casin^i ku il-ul professor-NOM -DAT self—KOM the work-ACC ha—tolok malhay-ess—ta. do-COMP say—PST—IND 'The professor told Mary that she (lit. self) do the work.' The reflexive pronoun in (41a) is in the subject position of the embedded clause, and bound by 'John' which is the matrix subject. The antecedent of 'casin' in (41b) is the recipient 'Mary' which belongs to the matrix clause because this is an object control structure, thus the embedded subject position 1 no Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is exclusively controlled by 'Mary'. In order to clarify the number and gender of the antecedent, a pronoun can precede 'casin'. (42) [John-kwa Mary^-nun [kutul casin^-ul salanghan-ta. —and —TOP they self—ACC love—IND 'John and Mary loves themselves.' (43) John^un [[ku casin]^-i chencay-la]-ko sayngkakhan-ta. -TOP he self-NOM genius-IND-COMP think-IND 'John thinks that he (lit. himself) is a genius.' (44) John-i Mary-eykey [[kunye casin],-i ku il-ul -NOM -DAT she self-NOM the work-ACC ha-tolok malhay-ess-ta. do-COMP do-PST-IND 'John told Mary that she (lit. herself) do the work.' The antecedent in (42) is plural, so the reflexive pronoun has 'kutul (they)' as a part of it. The antecedent in (43) is masculine, thus the reflexive pronoun has 'ku (he)', and the antecedent in (44) is feminine, thus the reflexive pronoun gets 'kunye (she)'. What is relevant for this study is that a reflexive pronoun 'casin', or 'ku casin (himself)/kunye casin (herself)' can occur as the subject of the embedded clause, while its antecedent, for example, is the subject of the matrix clause as in (43), or the recipient of the matrix clause as in (44). As a pronoun which is coreferential with the causee can occur in the subject position of the embedded clause as in syntactic causative sentences like (39), a reflexive pronoun 'casin' or 'kunye casin (ku casin/kutul casin)' can also occur as the subject of the embedded clause. 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (45) John-i Mary-eykey(/-lul) [kunye casin^i ku chayk-ul -NOM -Eat(/-ACC) she self-NOM the book-ACC ilk]-key hay-ess-ta. read-COMP do-PST-IND 'John instructed Maryi that shei (lit. herself) read the book.' It is well-known that the controlled subject position of the embedded clause can be filled with either a pronoun or a reflexive pronoun also in Japanese. (46) (Sakaguchi (1990)'s lb) Tarooi-ga Jiroo--ni [jibun*-,--ga kuruma-o naosu] -NOM -DAT ~ self-NOM car-ACC fix k a r e * i/ j/ . | c *Goroof c -ga yooni itta. COMP said 'Taroo- said to Jiroo^ for self*-.; him-,;.*., to 1 J * l / j 1 / J / * K fix the car. (47) (his lc) Taroo^ga Jiroo.-ni [jibnn^-ga kuruma-o naosu] to -NOM -DAT self-NOM car-ACC fix COMP kare^^-ga Le J i/*3/*k *Goroof c -ga yakusokusita. promised 'Taroo- promised Jiroo- for self ^ / h i m ^ ^ to *Gorook fix the car.' 'iu (say)' is an object control verb in Japanese, thus the unexpressed subject of the embedded clause in (46) must be coreferential with the object, 'Jiroo' of the matrix clause, not with the matrix subject, 'Taroo'. This is why pronoun 'kare (he)', or reflexive 'jibun (self)' which optionally fills the embedded subject position is obligatorily corefe- 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rential with the recipient argument 'Jiroo', not with the matrix subject 'Taroo' . • ' Yakusoku—suru (promise)' is a subject control verb in Japanese, and the embedded subject position is usually unexpressed, being assumed to be core ferent ial with the matrix subject. When a pronoun or a re flexive pronoun, however, optionally occupies this position as in (47), it is obligatorily controlled by the matrix subject. It is similar also in Korean that the controlled subject position of the embedded clause can either stay empty or be filled by a pronoun, or a reflexive pronoun. (48) Sajangnim-i Mary-eykey [0 ku il-ul ha]-tolok president -DAT the work-ACC do-COMP cishiha-shi-ess-ta. direct-HON-PST-IND 'The president directed Mary to do the work.' (49) Sajangniir^-i Mary.-eykey [kunyefti/- /?k-ka ku il-ul president-NOM -DAT she-NOM the work-ACC ha—tolok cishi-shi-ess-ta. do-COMP direct-HON-PST-IND 'The president directed Mary that she do the work.' (50) Sajangnim^-i Mary.-eykey [kunye casin^.^^-i < 1 AW 4- AWW 2 *“ ^ — — I n v n i / n x 9 U C b C i 1 n u n ku il-ul ha]-tolok cishiha-shi-ess-ta. 'The president directed Mary that she (lit. herself) do the work.' The subject position of the embedded clause in (48) is controlled saliently by the object 'Mary', not by the subject of the matrix clause, 'sajangnim (president)' since the matrix verb is a object-oriented control verb. Therefore, if a pro noun 'kunye (she) ' as in (49) or a reflexive pronoun as in (50) occupies the position, it is coreferential with the 111 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. object 'Mary'. This position can also be filled by a noun phrase which is coreferential with the object 'Mary'. (51) Sajangnim^-i Maryj-eykey [Mary^^-ka ku il-ul president-NOM -DAT -NOM the work-ACC ha]-tolok cishiha-shi-ess-ta. do-COMP direct-HON-PST-IND 'The president directed Mary that she (lit. Mary) do the work.' This phenomenon is equally applied to the so-called syntactic causative as seen before. Finally, if the causee is in the Qnominative case, then a pronoun, or a reflexive which is coreferential with it can occur between the two verbs. (52) Kyoswunim-i [Mary^ka ku chayk-ul ilk-key) professor-NOM -NOM the book-ACC read-COMP kunye ^-eykey-man ha-shi-ess-ta. she-DAT-only do-HON-PST-IND 'The professor instructed Mary that she read the book.' (53) Kyoswunxin—x [Mai'yi-ka ku chayk—ui ilk—key] professor-NOM -NOM the book-ACC read-COMP kunye casin^eykey ha-shi-ess-ta. she self-DAT do-HON-PST-IND 'The professor instructed Mary that she (Lit. herself) read the book.' In this subsection, we showed that binding condition (B) on pronouns is not violated if the syntactic causative are bi- clausal as we assumed, and also showed that a pronoun or a reflexive 'casin (self)', or 'ku casin (himself)', along with NPs can optionally fill the embedded subject position in a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. syntactic causative, demonstrating that there is a subject position of the embedded clause due to the biclausal structure. (C = 2.) Quantifiers and Quantifier Floating In this section, we will deal with quantifiers— numeral quantifiers on one hand, and universal quantifier 'raotwu (all)', and existential quantifier 'ilpwu (some)' on the other — to show that the syntactic causative is indeed biclausal at all levels. Numeral quantifiers in Korean mostly appear with a classi fier which shows the semantic category of the noun with which the quantifier is related. C.M. Lee (1989) believes that the unmarked order is 'Noun + Numeral + classifier'. (1) (his 9) a c o n n i h a w n t . m n — - — r - --- r”' * * * scholar one CL: person(HON) b. nolay twu kok song two CL: music piece The numeral and its classifier can be located either before the noun or after the noun. We will call them a prenominal numeral quantifier and postnominal numeral quantifier respectively, following &lan Hyen-Oak Kim (1990)'s terms. The prenominal numeral quantifier must be accompanied by a genitive case marker. 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (2) a. sey cang-uy congi three CL-GEN paper *b. sey cang congi The postnominal numeral quantifier can either stay inside the noun phrase as in (3a) or float out of the noun phrase as in (3b) . (3) a. John-i [chayk sey kwon uol-ul sa-ass-ta. book three CL NP ■John bought three books.' b. John-i chayk-ul sey kwon sa-ass-ta. 'John bought three books.' The floated numeral quantifier can be optionally assigned the case which the noun phrase it floated out of is in. If it is marked for case, it must be the same case which the original NP is marked. Otherwise, the sentence becomes ungrammatical. The numeral quantifier in (4s) is in the accusative just like the noun phrase, thus the sentence is grammatical. The numeral quantifier in (4b) is marked nominative case, whereas the noun phrase is marked accusative. This inconsistency of the case marker assigned to the noun phrase and the floated quantifier leaves the sentence ungrammatical. (4) a. John—i chayk—ul sey kwon—ul sa—ass—ta. -NOM book-ACC three CL-ACC buv-PST-IND 'John bought three books.' *b. John-i chayk-ul sey kwon-i sa-ass-ta. -NOM book-ACC three CL-NOM buy-PST-IND 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Based on this empirical fact, we have the following restric tion on the relationship between the noun phrase and its floated numeral quantifier. (5) In a single clause, the NP and the floated numeral quantifier (with its classifier) must share the same case marker9. Keeping this point in mind, we will try to use numeral quantifiers and quantifier floating to test the clausal relations in a causative sentences. (6) Kyoswunim-i haksayng-tul-eykey(/-ul) [kutul cwung-eyse professor-NOM student-PL-DAT(/-ACC) they among sey myeng-man-i ku il-ul ha-key] hay-ess—ta. three CL-only-NOM the work-ACC do-COMP do-PST—IND 'The professor instructed the students that only three of them among them do the work.' In (6), the causee is in the dative case, and the numeral quantifier is in the nominative case. The causee is a larger set of people, of which the numeral quantifier is a subset. The numeral quantifier should be assigned the case marker which the noun phrase is in if it floated out of the noun phrase. The noun phrase also should not be larger than the numeral quantifier in quantity if the quantifier is floated out of the noun phrase, but the noun phrase which is the causee is a larger set than the numeral quantifier. These empirical facts, therefore, demonstrate that the quantifier 'sey myeng (three man)' did not float out cf the NP, but it is ■tin 1J.D Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. an independent noun phrase in its own right, and it is the embedded subject at every level10. (7) Kyoswunim-i haksayng-tul-eykey(/-ul) [kutul cwung-eyse professor-NOM student-PL-DAT(/-ACC) they among [kacang ttokttokhan haksavng sey myengHP]-man-i ku most smart student three CL-only-NOM the il-ul ha]-key ha-shi-ess—ta. work-ACC do-COMP do-HON-PST-IND 'The professor instructed the students that among them only three of the smartest students do the work.' The numeral quantifier in (7) is a part of the embedded subject NP, which is not coreferential with the causee, but a subset of it. A monoclausal analysis such as Gerdts' (1990), which claims that the causee is the embedded subject at some level, and becomes an argument of the single union clause, cannot explain the empirical fact that a causee and an embedded subject which are two separate NPs exist at the same level. (S) Kyoswuniiu—i ku sey myeng—uy naksayng—eykey(/—ul) [nan professor-NOM the three CL-GEN student-DAT(/-ACC) one myeng—man—i yenge-lul kongpwuha—ko, twu myeng-un CL-oniy-NOM Engiish-ACC stuay-CON two CL-TOP swuhak-ul kongpwuha]-key ha-shi-ess-ta. math—ACC study—COMP do—HON—PST— IND 'The professor instructed the three students that only one of them study English, and two of them study Mathematics.' The causee in (8) is the larger set which covers the two subsets which are split into the two subject of the embedded conjuncts. In (8), the subject of the first conjunct and the second one are both numeral quantifiers. This also proves that 1 1 r 1. 4.0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the quantifier did not float out of the causee noun phrase, but they are individual NPs, functioning as the subject of each clause. (9) Kyoswunim~i [ku haksayngtul-i ku il—ul ha]—key professor-NOM the student-NOM the work-ACC do-COMP kutul-cwung sey myeng-eykey imi ha-shi-ess-eyo. they-among three CL already -HON-PST-POL 'The professor already instructed three of the students among them that they (lit. the students) do the work.' In sentence (9), 'sey myeng (three people)' in the dative case is a subset of 'ku haksayngtul (the students)'. The numeral quantifier which is located between the two verbs is not scrambled from the position immediately following the subject of the matrix clause, 'kyoswunim (professor)', because the sentence with it in that position is ungrammatical. *(10) Kyoswunim-i kutul-cwung sey meyng-eykey [ku they—among three CL-DAT haksayngtul-i ku il—ul ha]-key imi hay-ess-ta. 'The professor instructed three of them among them that the students do the work.' This is consistent with the fact that a pronoun or a reflexive 'casin' which is located between the two verbs cannot be scrambled to the position immediately following the matrix subject when there is a coreferentral noun phrase preceding it as in the examples (50)—(51) in section (D). 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (11) ((50) in section (D)) a. Kyoswunim-i [Mary^ka ku chayk-ul ilk]-key professor-NOM —NOM the work-ACC do-COMP kunye ^-eykey-man ha-shi-ess-ta. she-DAT-only do-HON-PST-IND 'The professor instructed Marvi that she^ read the book.' *b. Kyoswunim-i kunye^eykey [Mary^ka ku il-ul ha]-key professor-NOM she-DAT -NOM the work-ACC do-COMP imi ha—shi—ess—ta. already do-HON-PST-IND *'The professor instructed shei that Maryi read the book.' (12) ((51) in section (D)) a. Kyoswunim-i [Mary^ka ku il-ul ha]-key professor-NOM -NOM the work-ACC do-COMP kunye cashin^eykey imi ha-shi-ess-ta. she self-DAT already do-HON-PST-IND 'The professor already instructed Maryi that she^lit. herself) do the work.' *b. Kyoswunim-i kunye cashin^eykey [Mary^ka ku il-ul professor-NOM she self-DAT -NOM the work-ACC ha]-key imi ha-shi-ess-ta. do-COMP already do-HON-PST-IND 'The professor already instructed heri (lit. herself) that Maryi do the work.' When a pronoun as in (11a) or a reflexive pronoun as in (12a) follows the noun phrase which is coreferential, then the sentence is grammatical. However when either of them precedes the more referential noun phrase, the sentence becomes ungrammatical as in (lib) and (12b). This shows that the quantifier, pronoun, or reflexive pronoun in sentences (9), (11a), (12a) are not moved out of the embedded clause, but they are a part of the matrix clause at all levels, being positioned between the two verbs. The interverbal position is the clause boundary. No part of the embedded clause in a 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. biclausal sentence, therefore, can be placed in this position, and only elements belonging to the main clause may be placed there. (13) Kyoswunim-i haksayngtul-eykey(/-ul) [kutul cwung-eyse professor-NOM students-DAT(/-ACC) they among Mary-wa Bil 1-man-i ku chayk-ul ilk]-key ha-shi-ess-ta. -CON -only-NOM the book-ACC read-COMP do-HON-PST-IND 'The professor instructed the students that among them, only Mary and Bill read the book.' In (13), 'Mary-wa Bill (Mary and Bill)' is the subject of the embedded clause, and is a subset of the causee 'haksayngtul (students)'. It is an independent noun phrase which is a part of the embedded clause at all levels, and the causee in dative or accusative is a part of the matrix clause. Secondly, the universal quantifier 'motwu (all/every)', and the existential quantifier 'ilpwu (some)' behave quite simi- In a single clause, the floated universal quantifier and existential quantifier must get the same case marker as the noun phrase they floated out of. However, in the next causative sentence, the noun phrase and the floated universal quantifier have two different case markers, still being grammatical. larly with the numeral quantifiers we have seen xn tne acove. (14) Kyoswunim-i haksayngtul—ul [han salam-uy yeywoy—to professor-NOM student-ACC one person-GEN exception- epshi motwu-ka ku il-ul ha]-key hay-ess-ta. without all-NOM the work-ACC do-COMP do-PST-IND 'The professor instructed the students that without any exception, all of them do the work.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. If the sentence (14) is monoclausal, then it should be ungra mmatical, since the noun phrase 'haksayngtul (students)' and the universal quantifier 'motwu' are in two different cases; namely accusative and nominative each. But the sentence is grammatical. One might attempt to analyze examples like (14) along the lines of subject-to-object raising. (We do not consider it plausible as we demonstrate shortly.) One might claim that the noun phrase which was the embedded subject moved out of the embedded clause, and became a direct object of the matrix clause, leaving the floated quantifier in the embedded clause (cf. Eui-yon Cho, 1987; O'Grady, 1991). Suppose that the first underlying structure for (14) was (15) in which the quantifier is inside the subject noun phrase of the embedded clause. (15) Kyoswunim-i [[haksayngtul motwuNP]-ka han salam-uy professor-NOM students all-NOM one CL-GEN yeywoy-to epshi ku il-ul ha]-key hay-ess-ta. exception without the work-ACC do-COMP do-PST-IND 'The professor instructed that without any exception, all of the students do the work.' (16) Kyoswunim-i [haksayngtul-i han salam-uy yeywoyto professor-NOM students-NOM one person-GEN exception epshi motwu-ka ku il-ul ha]-key hay-ess-ta. without all-NOM the work-ACC do-COMP do-PST-IND 'The professor instructed that without any exception all of the students do the work.' As the next step, the quantifier is floated out of the subject NP as in (16), and the subject NP alone moves out of the embedded clause, and becomes the direct object of the matrix clause as in (14). This raising analysis may explain why the 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. noun phrase is in the accusative case, and the quantifier is a nominative, stiii maintaining the biclausal structure of the sentence. The raising analysis may so far seem to be a plausible way to explain examples like (14) . But examples such as the following one force us to forsake the raising analysis, and to find a more reasonable explanation. (17) Kyoswunim-i haksayngtuli-ul [han salam-uy yeywoy-to professor-NOM students—ACC one CL—GEN exception epshi [kutuli motwuNP]-ka ku il-ul ha]-key hay-ess-ta. without they all-NOM the work-ACC do-COMP do-PST-IND 'The professor instructed the students that without any exception, all of them do the work.' If the universal quantifier 'motwu (all)' floated out of the noun phrase 'haksayngtul (students)', and the noun phrase moved out of the embedded clause to become an argument of the main clause, then the universal quantifier should be alone. However, the universal quantifier is a part of a noun phrase 'kutul motwu (all of them)' which is the embedded subject. This shows that the causee belongs to the main clause at all levels, and the numeral quantifier in (14) is an independent noun phrase and is the subject of the embedded clause, as the noun phrase 'kutul motwu' in (17) is the subject of the embedded clause at all levels. Finally, we will look at the behavior of the existential quantifier 'ilpwu (some)'. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (18) Kyoswunim-i haksayngtul-ul [kutul-cwung [ttokttokhan professor-NOM students-ACC they-among smart haksayngtul ilpwuHp]-man-i ku elyewun mwuncey-lul students sorae-only-NOM the difficult question-ACC phwul]-key hay-ess-ta. solve-COMP do-PST-IND 'The professor instructed the students that among them, only some smart students solve the difficult questions.' In the above sentence, the existential quantifier 'ilpwu' is a part of the noun phrase 'ttokttokhan haksayngtul ilpwu (some of the smart students)' which forms the subject of the embed ded clause. This again shows that the existential quantifier was not a part of the causee. Thus it may be claimed that the causee did not move out of the embedded clause, but is an argument of the matrix clause at all levels. (19) Kyoswunim-i haksayngtul-eykey/-ul [ilpwu-nun professor-NOM student-DAT/-ACC some-TOP chayk-ul ilk-ko, ilpwu-nun nolay-lul pwulu]-key book-ACC read-CON some-TOP song-ACC sing-COMP hav-ess-ta. do-PST-IND 'The professor instructed the students that some of them read books, and some of them sing songs.• ' Again the two subjects of the embedded conjuncts 'ilpwu (some)-* each are subparts of the larger set 'haksayngtul (students)'. This shows that the quantifier floating analysis and the raising analysis of the embedded subject must be forsaken. Just like the numeral quantifiers, the universal quantifier 'motwu' and the existential quantifier 'ilpwu' can appear between the embedded verb and the main verb. 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (20) Kyoswunim-i [ku haksayngtul-i ku il-ul ha]-kev professor-NOM the students-NOM the work-ACC do-COMP kutul-cwung iipwu-eykey-man mence ha-shi-ess—ta. they-among some-DAT-only first do-HON-PST—IND 'The professor first instructed some of the students that they do the work.' (21) Kyoswunim-i [ku haksayngtul-i ku il-ul ha]-key professor-NOM the students-NOM the work-ACC do-COMP motwu-eykey irni ha-shi-ess-ta. all-DAT already do—HON—FSx—IND 'The professor already instructed all of the students that they do the work.' (22) Kyoswunim-i [ku haksayngtul-i ku il-ul ha]-key professor-NOM the students-NOM the work-ACC do-COMP [kutul motwuNpj-eykey imi ha-shi-ess-ta. they all-DAT already do-HON-PST—IND 'The professor already instructed all of the students that they do the work.' An existential quantifier is located interverbally in (20), the universal quantifier which is coreferential with the subject of the embedded clause is in that position in (21), and a noun phrase which contains a pronoun, being coreferen tial with the embedded subject noun phrase and a universal quantifier is placed in that slot in (22). A quantifier or a noun phrase in this position cannot be scrambled from the preembedded clause position because scrambling them to that position would make the sentence ungrammatical. *(23) Kyoswunim-i kutul-cwung ilpwu-eykey [ku haksayngtul—i professor-NOM they-among some-DAT the students-NOM ku il-ul ha]-key ha-shi-ess-ta. the work-ACC do-COMP do-HON-PST-IND 'The professor instructed some of the students that they do the work.' *(24) Kyoswunim-i motwu-eykey [ku haksayngtul-i ku il-ul professor-NOM all-DAT the students-NOM the work-ACC ha]—key ha-shi-ess-ta. do-COMP do-HON-PST-IND 'The professor instructed all of the students that they do the work.' 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. *(25) Kyoswunim-i [kutul motwuNP]-eykey [ku haksayngtui-i professor-NOM they all-DAT the students-NOM ku il-ul ha]-key ha-shi-ess-ta. the work-ACC do-COMP do-HON-PST-IND 'The professor instructed all of the students that they do the work.' The fact that the examples (23)-(25) are ungrammatical means that the quantifiers in (20)—(22), and the noun phrase with a quantifier in it as in (22) are placed obligatorily between the two verbs at every level. This means that there is a natural clause boundary between the embedded verb and the matrix verb. This is consistent with the fact that a numeral quantifier, a pronoun, or a reflexive pronoun in the inter verbal position cannot be scrambled out of that position as we saw in sentences (9), (11), and (12). In summary, we argued in this section that the subject-to- object raising and the quantifier floating analyses cannot be maintained for the syntactic causative. We also claimed that the causee, being a numeral quantifier, a universal quanti fier, or an existential quantifier, belongs to the embedded clause at every level, and can appear even when there is an independent causee. We also showed that an independent noun phrase which is a quantifier or which contains a quantifier in it, can appear in the subject position. We argued again that the causee could not be the embedded subject at every level, unlike the claim by Gerdts (1990), but is a part of the matrix clause at all levels. 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (C.3) Deletion of the redundant elements In Korean, there is a tendency to delete elements which were mentioned in the previous clause or discourse context in order to avoid the repetition of the redundant information. (1) [Mary^ka ku il-ul cal hay-ess-ki ttaymwuney] emeni.-ka -NOM the work-ACC well do-PST-NM because mother-NOM 0i senmwul-ul cwu-shi-ess-ta. gift-ACC give-HON-PST-IND 'Because Mary did the work well, Mother gave a present (to her)11.' (2) Mary-ka [Bill-eykey 0i cwu-lyeko] ku chavk^ul -NOM -DAT give-in order to the book-ACC sa-ass-ta. buy-PST-IND 'Mary bought the book in order to give (it) to Bill.' The recipient of the matrix clause in (1) is coreferential with the subject of the subordinate clause, and is not shown overtly, being unambiguously understood in the context. The theme of the center-embedded clause in (2) is coreferential with the theme of the main clause, and is not shown overtly. This kind of deletion process is very productive in Korean. (3) [Jonn.j-i Maryj-eykey cnayk-to phai]-ko 0i Oj kongcnayk-to -NOM -DAT book-also sell-CON notebook-also phal-ass—ta. sell-PST—IND 'John sold a book to Mary, and also sold a notebook (to her).' (4) [John--i chayk--ul Mary-eykey-to phal]-ko, -NOM boox-ACC -DAT-also sell-CON 0i 0i Bill-eykey-to phal-ass-ta. -DAT-also sell—PST—IND 'John sold a book to Mary, and also sold (one) to Bill. ' n r Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (5) Mary.j-eykey chayk--ul John-to pha.l-kof Bill-to 0i 0 - -DAT book-ACC -also sell-CON -also phal-ass—ta. sell—PST—IND 'John sold a book to Mary, and Bill also sold (one) (to Mary).' The recipient (and the subject) in the second conjunct in (3) is omitted, being coreferential with the recipient (and the subject) of the first conjunct, and the theme (and the subject) is omitted in the second conjunct of (4) because it is traceable from the theme (and subject) of the first conjunct. Both the recipient and the theme of the first conjunct are missing in the second conjunct in (5) because again they are recoverable in the previous context. This sort of deletion process only works with a full word. For instance, if the verb is a compound verb, then the whole compound verb, not a part of the verb, must be repeated. (6) a. John^i ku chayk-—ul Mary-eykey-to ilk-e cwu-ko, 0i 0- -NOM the book-ACC -DAT-also read give-CON 1 1 — — 1 J £11- — ------------------J OJbXJL“ C y A C )T “ t U l l h “ c UWU“ C a » “ U d • -DAT-also read give-PST-IND 'John read the book to Mary, and also read (it) to Bill.' *b. John^—i ku chayk;—ul Mary—eykey—to ilk—e cwu—ko, —NOM the book-ACC —DAT—also read give-CON 0i 0; Biii-eykey-to cwu-ess-ta. -DAT-also give-PST-IND 'John read the book to Mary, and read (it) to Bill.' *c. John.j“i ku chayk^-ul Mary-eykey-to ilk-e cwu-ko, -NOM the book-ACC -DAT-also read give-CON 0i 0; Bill-eykey-to ilk-ess-ta. -DAT-also read-PST-IND 'John read the book to Mary, and read (it) to Bill.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (6b) Is ungrammatical because only the second half of the compound verb is repeated in the second conjunct, and (6c) is ungrammatical because only the first half of the compound verb is repeated. 'ilk-e cwuta (read (for someone)' is a compound verb, and the second half of it 'cwuta (give) ' gives the interpretaion that the subject does some favor for the reci pient. Since it is a compound verb, no argument can intervene between the two parts of the verb. *(7) John-i ku chayk-ul ilk-e Mary-eykey-to cwu-ko, -NOM the book-ACC read -DAT-also give-CON Bill-eykey-to cwu-ess-ta. -DAT-also give-PST-IND 'John showed the book to Mary, and also showed (it) to Bill.' This means that the full verb form must be repeated if the verb is a compound verb. When a sentence has an embedded clause, then the main verb can be repeated in the process of deleting the redundant information. / o \ __4 rrv " ii-___j— i i.~. ^ w w a a a a - 4 . n u i . j r ■ w j r L w ' i / * i nu uuajriv-'ux — uaj fvv -NOM -DAT the book-ACC read-PST-IND-COMP malhay-ess-ta. say-PST-IND 'John told Mary that (he) read the book.' In this example, 'malhata (say/tell)' is the matrix verb, so it can be repeated in the deletion process. (9) John-i [ku chayk-ul ilk-ess-ta]^ko Mary-eykey-to -NOM the book-ACC read-PST-IND-COMP -DAT-also malha-ko, 0i Bill-eykey-to 0, malhay-ess-ta. say-CON -DAT-also say-PST-IND 'John said to Mary that (he) read the book, and also said (so) to Bill.' 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The complement clause of the first conjunct is omitted in the second conjunct because the two conjuncts share the same complement clause, and it can be understood that the omitted part in the second conjunct is the complement of the first conjunct. So far, we have dealt only with coordination structure, but this deletion (/omission) phenomenon is not limited to coordination. (10) [John--i [ku chayk-ul ilk—ess-ta]--ko Mary-eykey -NOM the book-ACC read-PST-IND-COMP -DAT malha-n hwu-ey], 0^ Bill-eykey-to 0j malhay-ess-ta. say—PST after -DAT-also say-PST-IND 'After John said to Mary that (he) read the book, (he) also said (so) to Bill.' The complement clause in the subordinate clause, in (10), is shared in the matrix clause, so it is not repeated in the matrix clause because it is redundant. Just like the example (5), only the subject and the matrix verb can be repeated without being ungrammatical when the complement clause is the shared information. (11) Mary-eykey [0j ku chayk-ul ilk-ess-ta]-ko John^to -DAT tne book-ACC read-FST-IND-COMF -also malha-ko Bill-to malhay-ess-ta. / i A M _ 1 _ _ t m a s a j r — u v n ~ a x a u s a y r o i " i « u 'John said to Mary that (he) read the book, Bill also said (so) to Mary.' Any constituent in a clause can be repeated with everything else deleted as long as the verb in its complete form is repeated. In a syntactic causative, this test can be applied TOO Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. to see if the two verbs are fused together and form a single compound verb or if each verb maintains its independence. (12) Kyoswuniu^-i Mary^-eykey [0*i/fj/*k ku nolay-lul pwulu]-key professor-NOM -DAT the song-ACC sing-COMP hay-ess-ta. do-PST-IND 'The professor instructed Mary to sing the song.' If the two verbs in (12) are fused and function as a compound verb then, the reduction process cannot take place in the interverbal position, but this is not what we find. (13) Kyoswunin^-i [0 ku nolay-lul pwulu]j-key Mary-eykey-to professor-NOM the song-ACC sing-COMP -DAT-also ha-ko, Bill-eykey-to 0- hay-ess-ta. do-CON -DAT-also do-PST-IND 'The professor instructed Mary to sing the song, and (he) instructed Bill (to sing) (the song).' The example (13) is grammatical. This shows that the verb 'hata' is not a part of the fused compound verb, but is an independent verb which is the matrix verb. The fact that the embedded clause is the shared information means that it is an independent constituent. This process is also applied when it is not a coordination structure. (14) KyoswuninVj-i [0 ku nolay-lul pwulu]j-key Mary-eykey professor-HOM the song-ACC sing-COMP -DAT ecey hay-ess-ki ttaymwun-ey 0i Bill-eykey-to 0j onul yesterday do-PST-NM because -DAT-also today hay-ess-ta. do-PST-IND 'Because the professor instructed Mary to sing the song yesterday, (he) also instructed Bill (to sing) (the song) today.' 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (15) Kyoswuni.in.j-i [0 ku nolay-lul pwulu] ^-key Mary-eykey professor-NOM the song-ACC sing-COMP —DAT ha~ki cen=ey 0^ Bill—eykey 0- mence hay—ess—ta. do-NM before —DAT first do-PST-IND 'Before the professor instructed Mary to sing the song, first (he) instructed Bill (to sing) (the song).' These two sentences have subordinate clauses, the first one led by '— ttaymwuey (because)', and the second one by '—ceney (before)'. The complement clause is the shared information, and it is omitted in the matrix clause. If the embedded verb is & part of a coip.pound V0rbt this omission procsss must create an ungrammatical sentence. But these sentences are grammatical, proving that the complement clause is a complete clause at every level. (16) [0 ku nolay-lul pwulu] .j-key Kim kyoswunim-i Mary-eykey the song-ACC sing-COMP professor-NOM -DAT ha-ki cen-ey, Lee kyoswunim-i Jane-eykey 0^ mence do-NM before professor-NOM -DAT first hay-ess-ta. 'Before Prof. Kim instructed Mary to sing the song, Prof. Lee instructed Jane (to sing) (the song).' The only shared information between the main clause and the subordinate clause is the complement clause, so it is deleted in the matrix clause, without making the sentence ungrnmmnti— 1 T w a c e n K i o n ^ /*\<F n 1 *anr/5 U J. • A U W M V *4 V A W f J k/ M M J W W W W U W M W 4. U U U W ^ along with the matrix verb 'hata (do)', is repeated. (17) Mary^evkey [0 ku nolay-lul pwulu] j-kev Kim kyoswunim-to -DAT the song-ACC sing-COMP professor—also ha-shi-ko, 0^ 0. Lee kyoswunim—to ha—shi—ess-ta. do-HON-CON professor-aiso do-HON-PST-IND 'Prof. Kim instructed Mary to sing the song, and Prof. Lee also instructed (her) (to sing the song).' 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (18) Mary-eykey [0 ku nolay-lul pwulu]j-key Kim kyoswunim-i -DAT the song-ACC sing-COMP professor-NOM ha-shi-ki cen-ey, 0^ 0- Lee kyoswunim-i mence ha-shi- -HON-NM before professor-NOM first do-HON- ess-ta, 'Before Prof. Kim instructed Mary to sing the song, Prof. Lee first instructed (her) (to sing) (the song).' (17) is a coordination structure with the conjunction '-ko', and (18) is a subordination structure with the subordinate clause introduced by '-ceney (before)'. The two conjuncts in (17) and the subordinate and the matrix clause in (18) share the recipient and the main verb, thus only the subject and the main verb 'hata' are repeated in the matrix clause. The data we have seen in this section show that the embed ded clause in the so-called syntactic causative sentences in Korean functions as a fully independent constituent of the sentence, and the verb 'hata' also functions as a single independent verb, not as a part of a fused compound verb. In extreme cases, only adverbials, clauses, or PPs with the matrix verb can occur in the second clause. (19) Kyoswunimi-i Mary^-eykey [ku chayk-ul ilk]f c -key professor-NOM -DAT the book-ACC read-COMP ecey-to ha-shi-ko, 0i 0- Qk onul-to ha-shi-ess-ta. yesterday-also do-HGN-CON toaay-also do-HON-PST-IND 'Yesterday the professor instructed Mary to read the book, and today (he) also instructed (her) (to read) (the book).' (20) Kyoswunim^-i Marypeykey [ku chayk-ul ilk]-—key yenge professor-NOM -DAT the book-ACC read-COMP English swuep-i shicakha-ki cen-ey-to ha-shi-ko, 0i 0^ 0k class-NOM start-NM before-also do-HON-CON yeksa swuep-i shicakha-ki cen-ey-to ha-shi-ess-ta. history class-NOM start-NM before do-HON- 'The professor instructed Mary to read the book before the English class started, and (he) instructed (her) (to read) (the book) also before the history class started. 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (21) Kyoswunimi-i Maryj-eykey [ku chayk-ul ilk]k-key hakkyo- professor-NOM -DAT the book-ACC read-COMP school- eyse-to ha-shi-ko, 0i 0, 0k secem-eyse-to ha-shi-ess-ta. at-also do-HON-CON bookstore-at-also do-HON-PST 'The professor instructed Mary to read the book at school, and (he) instructed (her) (to read) (the book) in the bookstore.' All the shared information, including the subject of the matrix clause is omitted in the second conjunct in (19)-(21), except the new information, an adverbial 'onul (today)' in (19), a subordinate clause 'yeksa swuep-i shicakha-ki cen-ey (before the history class started)' in (20), a PP 'secern ap- eyse (at the bookstore)' in (21) each is added, proving that the verb 'hata' is the matrix verb of each sentence. (C.4) Negation In Korean, there are two types of negation which share the same basic meaning. The short form is the one with prefixed negation morpheme to the verb as in (la), and the long form is the one with the negation morpheme prefixed to an auxiliary verb 'hata' which follows the matrix verb (cf. Han, 1987; Song, 1988). (I) a. John-i Mary-lul anki-manna-ass-ta. -NOM -ACC NEG-meet-PST-IND 'John did not meet Mary.' b. John-i Mary-lul manna-ci ani-ass~ta(=ani-hay-ess-ta). -NOM -ACC meet NEG 'John did not meet Mary.' The two forms of negation share the same interpretation also in compound verbs. 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (2) a. John-i ku mwulken-ul caki cwumeni-ey —NOM the thing-ACC self pocket-into cip-e ne-ess-ta. pick put in-PST-IND 'John put the thing into his pocket.' b. John-i ku mwulken-ul caki cwumeni-ey -NOM the thing-ACC self pocket-into an-cip-e ne-ess-ta. NEG-pick put-PST-IND f John c l not pnt it into his pockst r c. John-i ku mwulken-ul caki cwumeni-ey -NOM the thing-ACC self pocket-into cip-e net-chi an-ass-ta(/ani-hay-ess-ta). pick put- NEG- 'John did not put it into his pocket.' At some level of representation, it is the case that the two verbs 'cipta (pick)' and 'netha (put in)' are two independent verbs. But at the final level, they are merged and became a compound verb which functions as one word. This is why the short form and the long form negation of the compound verb share the same interpretation. Another example is in order. (3) John-i ku saca-lul chong—ulo sso-a cwuki-ess—ta. -NOM the lion-ACC rifle-with shoot kill-PST-IND ^ John choi^ anj killsd f ho 3_2.on t'Jith h rifls.f There are two verbs 'ssota (shoot)', and 'cwukita (kill)' in (3), and they are merged to become a single compound verb. The fact that these two verbs are merged, and function as a single word can be proved by the "negation" test. (4) a. John-i ku saca-lul chong-ulo ani-sso-a cwuki-ess-ta. —NOM the lion-ACC rifle—with NEG—shoot kill—PST—IND 'John did not shoot and kill the lion with a rifle.' b. John-i ku saca-lul chong-ulo sso-a cwuki—ci an-ass— -NOM the lion-ACC rifle-with shoot kill NEG-PST ta (=aiii—hay—ess—ta). 'John did not shoot and kill the lion with a rifle.' 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The two sentences in (4) share the same basic meaning, al though there are two verbs involved here . This proves that these two verbs are fused, and became a single compound verb. This shows that the verbs which get merged together and became a single verb must show with the "negation" test that they behave like one verb at the final level. If two verbs are not fused to be a compound verb, then they will behave like two independent verbs. (5) John.j-i Mary^-eykey [0^*- ku chayk-ul ilk-ess-ta]- -NOM -MT the book-ACC read-PST-IND ko cenhay-ess-ta. COMP report-PST-IND 'John reported to Mary that (he) already read the book.' 'ilkta (read)' is the embedded verb, and 'malhata (tell)' is the matrix verb, and the fact that the two clauses did not collapse into a single clause can be shown by the "negation" (6) a. Joh^-i Mary^-eykey [0^*, ku chayk-ul an(i)-ilk-ess- -NGM -DAT the book-ACC HEG-read-PST ta]-ko cenhay-ess-ta. IND-COMP report-PST-IND 'John reported to Mary that (he) did not read the book.' b. John-i Mary.-eykey [0^*, ku chayk-ul ilk-ess-ta]-ko -NOM -DAT the book-ACC read-PST-IND-CMP cenha-ci an-ass-ta (=ani-hay-ess-ta). report NEG-PST-IND 'John did not report to Mary that he read the book.' If the two T /erbs were fused and became a single compound verb, then (6a) would have the short form negation of the compound « o « • » A Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. verb, and (6b) the long form negation of it. But their inter pretations are not the same at all. The scope of the negation in (6a) is the embedded clause, and the scope of the negation in (6b) is the matrix clause. This shows that each clause maintains its own clausality. (6a) has its own long form «na/*n4* * 5 a m A A i t m 4- a *»mai 9m/l / C K V r* i +■ c* A«.*m £ «•» w >x v u s^vw m ^ uuu y w w | ju I C * O ub vvtll > 3 U V ^ i. L J L U J L ill counterpart. (7) John^i Mary.-eykey [0i/tj ku chayk-ul ilk-ci an-ass —NOM —DAT the book-ACC read NEC—PST (=ani-hay-ess)-ta]-ko cenhay-ess-ta. -IND-COMP report-PST-IND 'John reported to Mary that (he) did not read the book.' (8) John^i Mary^-eykey [0i > r *j ku chayk-ul ilk-ess-ta]-ko -NOM -DAT the book-ACC read-PST-IND-COMP an-cenhay-ess-ta. NEG-report-PST-IND 'John did not report to Mary that (he) read the book.' (7) is the long form negation counterpart of (6a), and shares the same basic meaning, and (8) is the short form negation counterpart of (6b), sharing the same basic meaning. If the +> r . i A t » a e * « « « * a w i a < » * a a i 4 « \ m «4 ^ P u m a 4 > 4 A m •» r> a w a m U i . l > a w W T C W V W^*/ei3 Ub V > ^ U4AVI AMMWWAWM U«9 U WVIU^/WUUU V < C J L JJ ^ UUCU the first part is not supposed to have its own long form counterpart, and the second part is also not allowed to have its own short form counterpart. *(9) John-i ku saca-lul chong-ulo sso-ci an-a (=ani-hay) —NOM the lion-ACC rifle-with shoot NEG cwuki-ess—ta. kill—PST-IND 'John did not shoot and kill the lion with a rifle.' *(10) John—i ku saca-lul chong-ulo sso-a an-cwki—ess-ta. —NOM the lion—ACC rifle-with shoot NEG—kill—PST—IND 'John did not shoot and kill the lion with a rifle.' 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In (9), only the first part of the compound verb has the long form negation, and in (10) the second part, having the short form negation, thus both being ungrammatical. The fact that (S)-(10) are ungrammatical shows that the two verbs do function as a compound verb. There is another negation morpheme 'mos' which implies some reason which prevented the action from being accomplished. Just like 'ani' negation, there are also two types of negation with this NEG morpheme 'mos': the short form as in (11a), and the long form as in (11b). (11) a. John-i ku chayk-ul mos-ilk-ess-ta. -NOM the book—ACC NEG-read-PST-IND 'John was not able to read the book.' b. John-i ku chayk-ul ilk-ci mos-hay-ess-ta. -NOM the book—ACC read NEG-do-PST-IND 'John was not able to read the book.' This NEG morpheme also behaves just like 'ani' with compound verbs. (12) a. John-i ku saca-lui chong-ulo mos-sso-a cwuki-ess-ta. -NOM the lion-ACC rifle-with NEG-shoot kill-PST-IND 'John could not shoot and kill the lion with a rifle.' b. John-i ku saca-lul chong-ulo sso-a cwuki-ci —NOM the lion-ACC rifle—with shoot kill mos-hay-ess-ta. NEG-do-PST-IND 'John could not shoot and kill the lion with a rifle.' (12a) is the short form negation and (12b) is the long form negation for the compound verb 'sso-a cwukita (shoot and 136 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. kill). insertion of the NEG morpheme 'mos' between the two parts of the compound verb will make the sentence ungramma tical . (13) *a. John—i ku saca-lul chong-ulo sso-ci mos-hay -NOM the lion-ACC rifle-with shoot NEG-do cwuki-ess-ta. kill—PST—IND 'John could not shoot and kill the lion with a rifle.' *b. John-i ku saca-lul chong-ulo sso-a mos-cwuki-ess-ta. -NOM the lion-ACC rifle-with shoot NEG-kill-PST 'John could not shoot and kill the lion with a rifle.' The sentences (13a-b) are ungrammatical because an attempt was made to negate each part of the compound verb. The claim that the two clauses in the syntactic causatives are merged, and function as a single clause means that the two verbs also behave like one compound verb(cf. Gerdts (1990)). This means that the newly created compound verb should show that it behaves as one unit in negating the sentences. (14) Kyoswunim-i Mary^eykey (/-lul) [0i ku chayk-ul mos-ilk]- professor-NOM -DAT(/-ACC) the book-Acc NEG-read key ka-shi=ess-ta. C O M P d o — H O N — P S T — I N D 'The professor instructed Mary not to read the book.' (15) Kyoswunim-i Mary--eykey(/-lul) [0i ku chayk-ul ilk]-key professor-NOM -Dat(/-ACC) the book-ACC read-COMP ha—ci mos-hay-ess-ta. do NEG-do-PST-IND 'The professor could not instruct Mary to read the book.' If the two verbs are fused, and function as a compound verb in the above sentences, then (14) should be the short form and 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (15) the long form, sharing the basic meaning. But the NEG morpheme 'mos' in (14) negates the embedded verb only, with the given interpretation. 'Mos' in (15) only affects the main verb 'hata', giving the assigned interpretation. This shows that the two verbs are not merged to become a compound verb, but they behave as two independent verbs. This " ’negation" test shows that the so-called syntactic causatives in Korean behave exactly like other biclausal sentences such as (6a-b). If the two verbs in (14)-(15) are in fact independent verbs, (14) must have its own long form counterpart, and (15) its short form counterpart. This is exactly what we find. (16) Kyoswunim-i Mary-eykey [0 ku chayk-ul ilk-ci mos-ha]- professor-NOM -DAT the book-ACC read NEG-do key ha-shi-ess—ta. COMP do-HON-PST-IND 'The professor instructed Mary not to read the book.' (17) Kyoswunim-i Mary-eykey [0 ku chayk-ul ilk]-key professor-NOM -DAT the book-ACC read-COMP mos-ha-shi-ess-ta. NEG-do—HON—PST—IND 'The professor could not instruct Mary to read the book.' (16) is the long form counterpart of (14), and (17) the short form counterpart of (15) . Thus (14) and (16) share the same basic meaning as (15) and (17) do. The same point can be made with other biclausal object control sentences. (18) Kyoswunim-i Mary-eykey [0 ku chayk-ul mos-ilk]-key professor—NOM —DAT the book—ACC NEG—read—COMP cishiha-shi-ess—ta. direct-HON-PST-IND 'The professor instructed Mary not to read the book.' 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (19) Kyoswunim-i Mary-eykey [0 ku chayk-ul ilk]-key professor-NOM -DAT the book-ACC read-COMP cishiha-shi-ci snos-hay-ess—ta. direct-HON- NEG-do-PST-IND 'The professor could not instruct Mary to read the book.' The NEG morpheme in (IS) negates only the embedded verb, and the NEG morpheme in (19) negates only the matrix verb, con firming their biclausality. Each sentence has it own counter part in negation; namely, (18) has its own long form negation counterpart, (19) its short form negation counterpart. The "negation" test which we have seen in the above shows that the so-called syntactic causatives are in fact biclausal at all levels. (D.5) The position of adverbials, PPs, and subordinate clauses In Korean, scrambling elements belonging to the same clause around the clause does not affect the basic meaning. (1) a. John-i Mary-lul ecey manna-ass-ta. -NOM -ACC yesterday meet-PST-IND 'John met Mary yesterday.' b. John-i ecey Marv—lul manna-ass—ta. -NOM yesterday -ACC meet-PST-IND 'John met Mary yesterday.' c. Ecey John-i Mary-lul manna-ass-ta. yesterday -NOM -ACC meet-PST-IND 'John met Mary yesterday.' The time adverbial 'ecey (yesterday)' is in the preverbal position in (la), and immediately before and after the subject 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in (Ib)-(lc) each, not changing the basic meaning. If a sen tence is biclausal, the adverbial belonging to the embedded clause cannot be positioned interverbally because the comple mentizer which is suffixed to the embedded verb is the clause boundary between the matrix clause and the embedded clause. *(2) John-i Mary-eykey [ku chayk-ul ilk-keyess-ta]-ko -NOM -DAT the book-ACC read-VOL-IND-COMP nayil yaksokhay-ess—ta. tomorrow promise-PST-lND 'John promised Mary that (he) would read the book tomorrow.' This shows that the adverbial belonging to the embedded clause cannot intervene between the two verbs, but must be placed inside the embedded clause. If there is an adverbial which belongs to the matrix clause, however, it can be positioned interverbally. (3) John-i Mary-eykey [ku chayk-ul nayil ilk-keyess-ta]- -NOM -DAT the book-ACC tomorrow read-VOL-IND ko onul achim-ey yaksokhay-ess-ta. COMP today morning-on promise-PST-IND ’ this morn inn John nrnmi bpJ M a m that (het would read ---- ^ — ------- f — — ----------------.£ \ the book tomorrow.' 'Nayil' belongs to the embedded clause, and 'onul achim (this morning)' to the matrix clause. What is to be emphasized here is that the adverbial which is a part of the embedded clause never occurs between the two verbs when the sentence is bi clausal. This can be contrasted with the fact that an adver bial can move around quite freely as long as the sentence is monoclausal as in (la-c). We claim, firstly, that if a sen- 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tence is monoclausal, adverbials should be able to move around as flexibly as other elements can at every level. Secondly, in a biclausal sentence the adverbial which belongs to the embed ded clause cannot occur interverbally since only the elements which belong to the matrix clause are allowed to be placed at this position. Thirdly, another important thing we need to note is that adverbials cannot intervene between the two parts of a com pound verb. (4) John-i ku chayk-ul ecey cap-a ccit-ess-ta. -NOM the book-ACC yesterday grab tear-PST-IND 'John grabbed and tore up the book yesterday.' At some level of representation, it may be assumed that 'capta (grab)' and 'ccitta (tear)' were two independent verbs which had their own argument. But when they merged to become a com pound verb at another level, then the time adverbial cannot be placed between these two. * i 5' i John—i ku ehavk—u! cap—a ecey ccit—ess—ta. -NOM the book-ACC grab yesterday tear-PST-IND 'John grabbed and tore up the book yesterday.' (6) a. John-i ecey ku efaayk—ul cap-a ccit—ess—ta. —NOM yesterday the book—ACC grab tear—PST—IND 'John grabbed and tore up the book yesterday.' b. Ecey John-i ku chayk-ul cap-a ccit-ess-ta. yesterday -NOM the book-ACC grab tear-PST-IND 'John grabbed and tore up the book yesterday.' Except for this one restriction that the adverbial cannot come between the two parts of the verb, it can move around the 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sentence as in <6a—b) because the sentence is monoclausai. We claim, therefore, that no adverbial can intervene between the two parts of the compound verb. The fourth claim we will make here is that an adverbial which is not compatible with the tense encoded in the verb cannot occur in the single clausal sentence• *(7) John—i ku chayk-ul nayil cap-a ccit-ess-ta. —NOM the book-ACC tomorrow grab tear-PST-IND *John grabbed and tore up the book tomorrow.' 'Nayil (tomorrow)' encodes a future time reference, while the morpheme '-ess—' in the verb denotes a past time reference. Therefore, the sentence is not grammatical. In a monoclausai sentence, more than one time adverbial can appear as long as they are compatible. (8) Onul-un John-i Mary-eykey ku chayk-ul achim-ey today—TOP -NOM —DAT the book-ACC morning—at cwu-ess-ta. /v i v ro -D C < T MT U n 3-*- * * « ■ ' — 'Today John gave the book to Mary in the morning.' 'Onul (today)' and 'achim (morning)' are compatible, and as long as it is past 'this morning' already, that agrees with the past time reference encoded by the past tense morpheme '- ess—' in the verb. However, pairs of time adverbials such as 'onul (today)' and 'ecey (yesterday)' 'ecey (yesterday)' and 'nayil (tomorrow)', 'ecey (yesterday)' and 'sam il hwu-ey (three days later)', or 'onul (today)' and 'il cwuil hwu-ey 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (one week later)' are not compatible and cannot appear toge ther in a single clause. Therefore, any attempt to have any pair of them together in a monoclausai sentence will leave the sentence ungrammatical. *(S) Ecey John-i Mary-eykey chayk-ul onul cwu-ess-ta. yesterday -NOM —DAT book—ACC today give—PST-IND *'Yesterday John gave the book to Mary today.' *(10) Ecey John-i Mary-eykeyse ku chayk-ul senmwul—lo yesterday -NOM -from the book-ACC gift-as sam il-hwu-ey pat-ass-ta. three day—after-at receive-PST—IND *'Yesterday John received the book as a gift from Mary three days later.' However, if a sentence is biclausal, these pairs of adverbials which cannot appear together in a single clause can occur as long as each of the pair belongs to a different clause. (11) Ecey John-i Mary^-eykey [O^j ku il-ul sam yesterday -NOM -DAT the work—ACC three il-hwu-ey ha-keyess-ta]-ko malhay-ess-ta. day-after-at do-VOL-1NQ--C0MP say-PST-IND 'Yesterday John told Mary that he would do the work three days later.' 'Ecey (yesterday)' in (11) belongs to the matrix clause, and 'sam il hwu-ey (three days later)' the embedded clause. Since the two time adverbials belong to two different clauses, no conflict arises. If the syntactic causatives in Korean were monoclausai, and the two verbs were fused, thus becoming a compound verb at the final level, then it is expected that the claims we made in the above should be true with the causatives. 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The first claim we made in the above is that at all levels, an adverbial should be able to move around in the clause which it belongs to. This means that an adverbial would be allowed to move around in the causative sentence if it were monoclau- sal. (12) Kyoswunim-i Mary-eykey [0 ku chayk-ul nayil professor-NOM -DAT the book-ACC tomorrow ilk]-key ha-shi-ess-ta. read-COMP do-HON-PST-IND 'The professor instructed Mary that (she) read the book tomorrow.' If the sentence (12) were monoclausai, the adverbial should be able to move around in the sentence. The adverbial 'nayil', however, can only move around inside the bracketed string (which, we are claiming, is a clause). (13) a. Kyoswunim-i Mary-eykey [0 nayil ku chayk-ul professor-NOM -DAT tomorrow the book-ACC ilk]-key ha-shi-ess-ta. read-COMP do-HON-PST-IND 'The professor instructed Mary that (she) read the book tomorrow.' *b. Kvoswun i m — l navi 1 Mary-oyjrgw r Q b.i "hnvk-ul a - - ----------------- — — —j , j , * 2 2 L w X * * * professor-NOM tomorrow -DAT the book-ACC ilk]-key ha-shi-ess-ta. read-COMP do-HON-PST-IND 'The professor instructed Mary that (she) read the book tomorrow =' *c. Nayil Kyoswunim-i Mary-eykey [0 ku ehayk-ui tomorrow professor-NOM -DAT the book—ACC ilk]-key ha-shi-ess-ta. read-COMP do-HON-PST-IND 'The professor instructed Mary that (she) read the book tomorrow.' If the syntactic sentence in (12) is monoclausai, the time adverbial 'nayil' should be able to move around the sentence. 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. But this is not the case. The time adverbial is only allowed to be scrambled within the bracketed string in (13a). Thus if the time adverbial is scrambled out of the bracketed string (a clause in our analysis) as in (13b)-(I3c), the sentence becomes ungrammatical. Secondly, we clarm that m the syntactic causative (12), the adverbial 'nayil' encoding FUTURE time reference should not be allowed from the beginning if the sentence were monoclausai because it is in conflict with the fact that the morpheme '--ess-' in the matrix verb encodes PAST time refe rence. If they are biclausal, and the adverbial 'nayil' belongs to the embedded clause, however, there is no conflict in time reference, and this is the case with (12). Thirdly, the time adverbial which belongs to the embedded clause cannot intervene between the two verbs if they have merged and become a compound verb as in (5). *(14) Kyoswunim-i Mary-eykey [0 ku chayk-ul ilk]—key professor—NCM —DAT the book—ACC read—viuni* nayil ha-sni—ess—ta. tomorrow do-HON-PST-IND 'The professor instructed Mary that (she) read the book tomorrow.' However, If the sentence is biclausal, a time adverbial which belongs to the matrix clause, and is compatile with the time reference encoded by the morpheme '-ess-' should be able to occur interverbally as in (4). 1 A C Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (15) Kyoswunim-i Mary-eykey [0 ku chayk-ul nayil professor-NOM -DAT the book-ACC tomorrow ilk]-key ecey ha-shi-ess-ta. read-COMP yesterday do-HON-PST-IND 'Yesterday the professor instructed Mary that (she) read the book tomorrow.' 'Ecey' belongs to the matrix clause, and agrees with the time reference made by the morpheme '-ess-', so the sentence is grammatical. The fact that this adverbial can intervene between the two verbs means that the two verbs are not fused, indicating that they are two independent verbs. Additionally, the two adverbials 'nayil (tomorrow)' and 'ecey (yesterday)' are not compatible, so they should not occur together if the sentence (15) were monoclausai. However, the sentence (15) is grammatical. This shows that there are two clauses in sentences like (15). The same point can be expanded to include the general fact about the scrambling of other elements belonging to the same clause. Any NPs, PPs, or subordinate clauses can scramble rather freely inside the clause they belong to. All the elements which belong to the matrix clause in (15) can be placed immediately before the matrix verb 'hata'. (15) [0 ku chayk-ul nayil ilk]-key kyoswunnim-i the book-ACC tomorrow read-COMP professor-NOM Mary-eykey ecey ha-shi-ess-ta. -DAT yesterday do-HON-PST-IND 'Yesterday the professor instructed Mary to read the book tomorrow.' However, the argument 'ku chayk (the book)' cannot occur in that position because it does not belong to the matrix clause. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. *(17) Ecey kyoswunim-i Mary-eykey [0 nayil yesterday professor-NOM -DAT tomorrow ilk]-key ku chayk-ul hay-ess-ta. read-COMP the book-ACC do-PST-IND 'Yesterday the professor instructed Mary to read the book tomorrow.' This shows that the elements belonging to the embedded clause cannot intervene between the two verbs. (18) Kim kyoswunim-i Mary-eykey [0 ku chayk-ul Bill-kwa professor—NOM —DAT the book—ACC -with hamkkey ilk]-key John-kwa hamkkey ha-shi-ess-ta. together read-COMP -with together do-HON-PST-IND 'Prof. Kim., along with John instructed Mary to read the book with Bill.' There are two identical PPs in the sentence (18), which is not allowed in a monoclausai sentence as in (19). *(19) John-i Bill-kwa hamkkey Mary-eykey Jane-kwa hamkkey -NOM —with together —DAT —with together ku chayk-ul phal-ass-ta. the book-Acc sell-PST-IND *'With Bill John sold the book to Mary with Jane.' The same restriction is applied to cooccurence of certain subordinate clauses. More than one '—myense (while)' or '-taka (while)' cannot occur with a monoclausai sentence as in / ■ i n v *(20) John-i [Q umak-ul tul-u myense] ku chayk-ul -NOM music-ACC hear while the book-ACC [0 theyrepi-lul po-myense] ilk-ess-ta. T.V.—ACC see-while read-PST-IND *'While listening to music, John read the book while watching T.V.' However, this is not a problem if the sentence is biclausal. 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (22) John^i Mary.-eykey [0- umak-ul tul-u myense —NOM -DAT music-ACC hear while ku chayk-ul ilk-u-la]-ko [0i theyrepi-lul po-myense] the book-ACC read-IMP-COMP T.V.-ACC see-while malhay-ess-ta. say-PST-IND •'While 0i watching T.V., Johni told Maryj to read the book while 0j listening to music.' If the causative were monoclausai, then two '-myense' subcrdi— nate clauses should not occur in the sentence. (23) [0i Theyrepi-lul po-myense] Kyoswunim^i Mary.-eykey [(k T.V.-ACC see-while professor-NOM -DAT [0^ umak-ul tul-u myense] ku chayk-ul ilk]-key music-ACC hear while the book-ACC read-COMP ha-shi-ess-ta. do-HON-PST-IND 'While 0i watching T.V., The professori instructed Mary^ to read the book while 0^ listening to music. ' This syntactic causative sentence with two '—myense (while)' clauses is grammatical, having the first '-myense' clause belonging to the matrix clause, and the second the embedded clause. Likewise, two 'taka (while)' subordinate clauses cannot occur in a monoclausai sentence. (24) [Q.j cemshim-ul mek-taka] kyoswunim^-i Maryj-eykey lunch—ACC eat-whiie professor-NOM -DAT [0- ku chayk-ul [0^ ku mwuncsy-lul pwu-n hwu-ey] the book-ACC the question—ACC solve after ilk]-key ha-shi-ess-ta. read-COMP do-HON-PST-IND 'while 0^ eating lunch, the professor instructed Mary^ to read the book after 0, solving the problem. ' The first '-taka (while)' clause belongs to the matrix clause, and the covert subject of this clause is controlled by the 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. matrix subject, and the second '-taka' clause belongs to the embedded clause, having its implicit subject controlled by the covert subject of the embedded clause, which is subsequently controlled by the causee in the dative case. One more thing to note about the position of adverbials, PPs, or subordinate clauses in a sentence comes from the fact that in a monoclausai sentence, the basic meaning does not change wherever in the sentence they are located. (25) a. John-i Mary-lul hakkyo-eyse manna-ass-ta. -NOM -ACC school-at meet-PST-IND 'John met Mary at school.' b. John-i hakkyo-eyse Mary-lul manna-ass-ta. -NOM school—at —ACC meet—PST—IND 'John met Mary at school.' c. Hakkyo-eyse John-i Mary-lul manna-ass-ta. school-at -Nom -Acc meet-PST-IND 'John met Mary at school.' This fact should apply to the syntactic causative sentence if it is monoclausai. (26) a. Kyoswunim-i Mary-eykey [0 ku chayk-ul hakkyo-eyse professor-NOM —DAT the book-ACC school-at ilk-key ha-shi-ess-ta. read-COMP do-HON-PST-IND 'The professor instructed that Mary read the book at school.' b. Kyoswunim-i Mary-eykey [0 ku chayk-ul ilk-key] professor-NOM —DAT the book-ACC read-COMP hakkyo-eyse ha-shi-ess-ta. school-at do-HON—PST—IND •'At school the professor instructed that Mary read the book at school.' 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (26a) means that the professor arranged things so that the place where Mary reads the book be 'hakkyo (school)'. However, the meaning of (26b) is quite different from that of (26a) in the sense that it only means that the professor's instructing Mary to read the book was done at school, and it does not necessarily mean that Mary should read or will read f jjg book at school. (27) Kyoswunim-i Mary-eykey [0 ku chayk-un cip-eyse-man professor-NOM -DAT the book-TOP house-at—only ilk]-key hakkyo-eyse ha-shi-ess-ta. read-COMP school-at do-HON-PST-IND 'At school the professor instructed Mary to read the book at home.' The professor's instruction was done at 'school', and the actual place where Mary should or will read the book is at 'home'. This shows that the causative sentences (26a-b) are biclausal, and the interpretation changes, depending on which clause the adverbial belongs to. In this section, we first claimed that the scrambling of a time adverbial is only allowed within the clause boundary. Thus the sentence becomes ungrammatical if the time adverbial scrambles across the clause boundary= Secondly, in a biclausal sentence, a time adverbial which belongs to the embedded clause cannot intervene between the two verbs. Thirdly, we noted that a time adverbial is not allowed to intervene between the two parts of a compound verb. Fourthly, a time adverbial which conflicts with the time reference encoded by the tense morpheme such as the past tense morpheme '-ess' 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cannot occur in a single clause. Finally, we claimed that two incompatible time adverbials such as 'ecey (yesterday) ■ ' and 'nayil (tomorrow)' cannot occur in a monoclausai sentence. We showed, in this section, that time adverbials, PPs, and subordinate clauses behave in the syntactic causative in Korean just as they do in any other biclausal sentences, proving that the syntactic causative are biclausal at every level. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (D) Syntactic Causatives in Korean and Romance We argued in the previous section that the Korean syntactic causatives are consistently biclausal,and provided several pieces of empirical evidence. Burzio (1986) assumes a bi clausal structure for the Italian 'fare-infinitive-' causative construction, and Zubizarreta (1985) claims that the 'fare- object (/fare-infinitive)' construction is monoclausai, and the French/Spanish 'faire-object' construction is simulta neously monoclausai and biclausal. There are several diffe rences between my consistent biclausal analysis and their analysis of Romance causatives. First of all, the Case assignment to the causee in my analysis of Korean causatives does not depend on the transi tivity of the embedded clause, while in their analysis of Romance languages the Case assignment mainly depends on the transitivity of the embedded clause. In my analysis, the causee can be in the nominative, accusative, or dative, regardless of the transitivity of the embedded clause. / C 9 \ / T in p o n w c T C T iT trx ? Ts*\roT?r\T\T?T% y p y 1U111 U 4 . A A « u Uru^UL/l/Ui/ / a. John-i [s Mary-ka cip-ey ka]-key hay-ess-ta. -NOM -NOM house-to go-CGMP do-PST-IND 'John caused Mary to go home.' b. John-i Marv^eykev [s Ct cip-ey ka]-key hay-ess-ta. -NOM -DAT house-to go-COMP do-PST-IND 'John caused Mary to go home.' c. John-i Mary-lul [s 0i cip-ey ka]—key hay-ess-ta. -NOM -ACC house-to go-COMP do-PST-IND ■'John caused Mary to go home.' 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The embedded clause in (53a-c) is intransitive, but the causee can be in the nominative as in (53a), the dative as in (53b), or the accusative as in (53c) . This kind of flexibility of case of the causee is shown even when the embedded clause is transitive as in (54) . (54) (TRANSITIVE EMBEDDED CLAUSE) a. John-i [s Mary-ka ku chayk-ul ilk]-key hay-ess-ta. -NOM -NOM the book—ACC read-COMP do-PST-IND 'John caused Mary to read the book.' b. John-i Mary^evkev [s 0i ku chayk-ul ilk]-key -NOM *-DAT * " the book-ACC read-COMP hay-ess-ta. do-PST-IND ' John caused Mary to read the book.' c. John-i Mary^lul [s 0i ku chayk-ul ilk]-key -NOM -ACC the book—ACC read-COMP hay-ess-ta. do-PST-IND 'John caused Mary to read the book.' Secondly, Burzio (1986) has to assume that some sort of syntactic change (/movement) occurrs in the Italian causative because the following sentence; which is supposed to be the D— structure for the 'fare-infinitive' construction, is not (55) *Maria ha fatto [s Giovanni riparare la macehinaj. Maria had made Giovanni repair the car Therefore, he claims that VP-movement occurs in order to generate the right surface order as in (56) for (55). 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (56) Maria ha fatto riparare la macchina a Giovanni. Maria has made repair the car to Giovanni 'Maria had Giovanni repair the car.' Zubizarreta (1985) claims that the internalization of the external argument of the embedded verb gives the 'faire- object' construction. She is assuming that during the derivation the external argument of the embedded verb is internalized, and becomes an object of the new complex predicate, thus creating sentences like (57a—b). (57) a. Pierre a fait lire ces passages a Jean. 'Pierre made Jean read the passages.' b. Pierre a fait travailler Jean. 'Pierre made Jean work.' However, in my consistent biclausal analysis there is no subject-to—object conversion (/internalization of an external argument) or a syntactic movement such as VP-movement. There fore, in my analysis of Korean causatives there is no change in the biclausal structure. In Burzio's analysis 'Giovanni', in (56), which is in the dative case, is the embedded subject, and when the embedded clause is intransitive as in (58), the subject is assigned accusative case. (58) Maria fa [vp lavorare] [s Giovanni ] . 'Maria makes Giovanni work.' This is the same in Zubizarreta's biclausal analysis of (57a- b) . In the biclausal structure, 'Jean' in each sentence, (57a) with a transitive embedded clause and (57b) with an intran- 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sitive embedded clause, is the subject. However, in Korean causati\'es as in (53)-{54), according to my analysis, the embedded subject is in the nominative case, and the causee in dative or accusative case is not the embedded subject, but an object of the main verb 'hata' at every level. In the analysis of Burzio (1986) and Zubizarreta (1985) of Romance causatives there is no empty subject position in the embedded clause. In Burzio's analysis the embedded subject is present as an NP in the dative as in (56) or in the accusative as in (58), and this is the same in Zubizarreta's analysis of sentences like (57a-b). However, in my analysis of Korean causatives, there is an empty subject position in the embedded clause if the causee is in dative or accusative case as in (53b-c) and (54b-c). (E) Conclusion In section (C) , we showed that the syntactic causative in Korean is biclausal no matter which case marker the causee has, providing several pieces of evidence. We claimed that when the causee is in the nominative case, it is th.e subject of the embedded clause at any level. As a matter of fact, our analysis does not consider any derivational process for the syntactic causative in Korean. We also claimed that when the causee is in axther datxve or accusative case, the causee 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. belongs to the matrix clause, and the subject of the embedded clause can optionally occur. When the embedded subject is overt, we showed that it does not need to be coreferential with the causee. The overt embedded subject, however, has some semantic and pragmatic connection with the causee even if the two are not coreferential. This is also true when the covert embedded subject is not coreferential with the causee. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 4. Control in Korean Much research has been done on the topic of subject and object control. In English, for instance, if the covert sub ject of the infinitival clause is coreferential with the sub ject of the main clause, this is called subject control. If it is coreferential with the object of the main clause, this is called object control. The following sentences show the typi cal cases of the subject control and the object control. (1) a. John promised Mary to read the book, b. John persuaded Mary to read the book. The agent for the event expressed in the infinitival comple ment in (la) is the main clause subject 'John', not the object 'Mary'. Thus it is a subject control structure. On the other hand, the agent for the activity shown in the infinitival complement in (lb) is not the subject 'John', but the object 'Mary', this being an object control structure. Many linguists have approached the control phenomena from a syntactic basis (Bach (1979, 1960) Bresnan (1982) Chomsky (1980, 1981); Huang (1989); Koster (1984); Larson (1988); Manzini (1983); Mohanan (1983)) while some people have approached it from a semantic basis (Chierchia (1983, 1984, 1989); Dowty and Jacobson (1989); Jackendoff (1972, 1974)). Comrie (1984, 1986) emphasizes the importance of the interaction of syntax, seman tics and pragmatics, and Higgins (1973) and Dowty (1989) deal 157 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with inter-sentential control and control in discourse. Our approach to the control phenomenon in Korean is similar to Comrie's approach emphasizing the role of pragmatics. In this chapter, we will look at the Korean data with the type of control verbs used in other languages in the literature. We will argue that there is no strict restriction of coreferentiality between the main clause subject and the embedded subject when the main verb is the 'promise-type' . This also applies to the 'persuade-type', which is typically object control, thus we claim that there is no strict restric tion of coreferentiality between the main clause object and the overt embedded subject1. We also argue that it is not absolute that the covert position of the embedded subject in both of these cases is obligatorily controlled by the subject in the former case and by the object in the latter case when semantics and pragmatics allow that. This leads us to conclude that there is no subject control or object control in Korean i_n rnp S0HS6 ot E^^lish 02T cess cthsr vjith the typical type of infinitival complement. (A) Subject Control (A.I.) 'nolyekhata (try)'-type In the generative literature, a difference is drawn between two types of zero anaphors; one in the null subject position m the Romance type of languages where the maxn clause subject 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. can be overt, and the other in the covert embedded subject position in control structure calling the former 'pro' and the latter 'PRO'. Borer (1989) claims that in Korean there is no 'PRO', but only 'pro'. She uses the following examples to show that the covert embedded subject in these examples is not (2) (her 32a-c) (a) John^ka [CP ei ttena-lye~ko] nolyek hay-ess-ta. -NOM leave-wi11-COMP try do-PST-IND 'John trxed to leave•' (b) John^-ka [CP kui ttena-lye-ko] nolyek hay-ess-ta. -NOM he leave-wi11-COMP try do-PST-IND 'John tried to leave.' (c) John.j-ka [CP cakii ttena-lye-ko] nolyek hay-ess-ta. 'John tried to leave.' (3) (her 33a-b) (a) *John-ka [CP Bill-ka ttena-lye-ko] nolyek -NOM -NOM leave-wiil-COMP try hay-ess-ta. do-PST-IND (b) *John--ka [CP ku- ttena-lye-ko] nolyek hay-ess-ta. -NOM he leave-wiil-COMP try do-PST-IND She argues that when the embedded subject position is empty, as in (2a) it must be coreferential with the main clause subject, and when it is filled with an overt element as a pronoun in (2b) and an anaphor in (2c), it should be coreferential with the main clause subject. Filling the embedded subject position with either an NP as in (3a) or a pronoun not coreferential with the main clause subject as in (3b), she argues, is not allowed. She claims that ”in certain 159 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. kinds of Korean infinitives, the INFL node, which we will tentatively characterize as [—tense, +AGR], can assign nominative Case to its I-Subject". She summarizes her finding, based on the above set of examples, as follows: (4) (her 34) 'try' with a [~tense/+AGRj sentential complement: [ e ] NP try [ [ pronoun ] nom. AGR] [ anaphor ] [ *NP ] With this conclusion, she argues that the embedded subject in the sentences in (2)-(3) •'must be coindexed with the matrix subject". She also says "any other possibility would lead to ungrammaticality." Her generalization for the above data is that "the embedded subject may not be assigned independent reference". S.H. Kim (1994) also, following Borer's claim, argues that regardless of whether it is overt or null, the embedded subject of an infinitive clause with a 'nolyekhata (try)'-type verb must be coindexed with the matrix subject; i.e. the embedded subject KF is obligatorily controlled. Her relevant examples, wkxck are m e xoxxowxng ones, are almost identical to Borer's. (5) (her (14a&b)> (a) Joh^-un [ei ttena-lyej-ko nolyehay-ess-ta. -TOP leave-will-COMP try-PST-IND 'John tried to leave.' (b) John^-un [caki^ka ttena-lye]-ko nolyekhay-ess-ta. -TOP self-NOM leave-will-COMP try-PST—INB 'John himself tried to leave. ' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Her examples are the same as Borer's (32a&c), which are (2a&c) above. She has just changed the English translation for (5b). One immediate problem with Borer's and Kim's claim, that the embedded clause is [-tense/+AGR], is that they marked '- lye-' as 'will'. If the embedded clause is [-tense], then there should not he an overt future morpheme. We claim that ' — lye-' is not a future morpheme, but a part of complementizer '—lyeko'. The second thing which is not clear is that they did not show what the subject agreement feature is. It is well accepted that the honorific morpheme '—shi-' indicates subject honorification2. (6) Sensayngnim-i ku il-ul ha-shi-ess-ta. teacher-NOM the work-ACC do-HON-PST-IND 'The teacher did the work.' The subject 'sensayngnim (teacher)' is someone honorable in Korean culture, thus creating the subject honorification with the morpheme '-shi-' in the verb. It is true that if the sub ject is 'sensayngnim' in the place of 'John' in the examples (2), (3), and (5), the main verb, and the embedded verb can be affixed v?itb ths subject honorific Hisrksr / —shi—f r rsgsrdlsss of whether the embedded subject slot is empty or filled with a pronoun 'ku (he)' or an anaphor 'caki (self). (7) Sensayngnim^i. [e/kui/cakii ku ii-ul na-shi3]—lyeko teacher-NOM he/self the work-ACC do-HON-COMP nolyekha-shi-ess-ta. try-HON-PST-IND 'The teacher tried to do the work.' 161 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The presence of the subject honorific morpheme '-shi—' is optional. Even though the subject is honorable, it is not obligatory to have '-shi-'. The following examples are both grammatical. (8) (a) Sensayngnin^-i [e/kui/cakii ku ii-ui ha]-lyeko teacher-NOM he/self the work-ACC do-COMP nolyekha-shi-ess—ta. try-HON-PST-IND 'The teacher tried to do the work.' (b) Sensayngnimi -i fe/kui/cakii ku il-ul ha-shi]-lyeko teacher-NOM he/self the work-ACC do-HQN-COMP nolyekhay-ess-ta. try-PST-IND 'The teacher tried to do the work.' In (8a) the subject honorific marker '-shi-' is attached only to the main verb, and only to the embedded verb in (8b). It may be the case that the embedded clause in (7) —(8) is at best optionally [+AGR]. The optionality of the agreement feature is again well demonstrated in the following example. (9) Sensayngnim^-i [e/kui/cakii ku il-ul ha]-lyeko teacher-NOM he/self the work-ACC do-COMP nolyekhav-ess-ta. try-PST-IND 'The teacher tried to do the work.' In (9) '-shi—' is attached neither to the main verb nor to the embedded verb. This does not worsen the grammaticality of (9) . The sentence (9) is as grammatical as the sentences in (7) and (8) . As in English, it may be argued that in (9) the main clause subject is the only possible controller of the embedded subject slot because there are no other referents in the sen- 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tence to be the antecedent or the controller if the controller must be looked for within the sentence. There is no reason, however, that the controller must be within the sentence. It may be possible that the controller may not be in the sen tence. The next example demonstrates this point. (10) Sensayngnim.j-i [caki-wa Mary-ka- ku il-ul ha]-lyeko teacher-NOM self-and -NOM the work-ACC do-COMP nolyekhay—ess—ta. try-PST-IND 'The teacher tried (for himself and Mary) to do it.' The subject slot of the embedded clause in this example is filled by the NP 'caki—wa Mary (self and Mary) '. The referent of 'caki' is the main clause subject 'sensayngnim (teacher)'. But the trouble is that 'Mary' does not have a referent within the main clause. The exact referent for the embedded subject NP, therefore, cannot be found within the sentence because the embedded subject is partially coreferential with the main clause subject, but not fully coreferential. It is, there- 4 . U juu -i « - —a.------ J I V !• 'C5 f U V J k G A C lG U V X a X W J. I . U u i t c u t a j : u uxauoc 3 uuj x i u t controlled by it. This example goes against Borer's (1989) finding in (4) because an NP, a part of it being an anaphor, is in the embedded subject position, without making the sentence ungrammatical. Her argument that the embedded subject must be coindexed with the main clause subject does not stand because in the sentence (10) the embedded subject 'caki-wa Mary (self and Mary) is not coindexed with the main clause subject 'sensayngnim (teacher)'. 163 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. When the embedded subject slot is empty, the primary interpretation is that it is coreferential with the main clause subject, thus controlled by it. Even in that case, however, other interpretations are possible when an appropriate context is given. (11) a. kulena pwucang^un [0^ ilkop shi-kkaci il—ul ha]- but raanager-TOP seven o'clock till work-ACC do- lyeko nolyekhay-ess—ta. COMP try-PST-IND 'But the manager tried to work until seven.' Suppose the previous context for this sentence is that all the employees in the department wanted to finish work at six and go home. There was, however, a great deal to do. Thus the manager wanted everyone to work until seven, and (s)he tried to make it happen against everybody else's will. The covert subject of the embedded clause in (11) may not be the manager, the main clause subject, alone, but all the employees with the manager included or not. But when this kind of clear context is not given, the most likely interpretation would be the covert embedded subject is coreferential with the main clause subject4. Unlike the situation in (10), where the main clause subject is a part of the embedded subject, it is also possible that the embedded subject may be a part of the main clause subject. (12) John-kwa Mary--ka [kutul cwungeyse Mary-mair—i ku -and —NOM they among —only—NOM the il-ul ha]-lyeko nolyekhay-ess-ta. work-ACC do]-COMP try-PST-IND 'John and Mary tried (just Mary between them) to do the work•' 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The embedded subject ''Mary' is a part of the main clause subject 'John, and Mary', The agent of the event described in the embedded clause is only 'Mary'. The subject of the embedded clause in this case is also not controlled by the subject of the main clause. Some may claim that 'kutul cwungeyse Mary-man-i' actually belongs to the main clause because the following sentence is possible. (13) John-kwa Mary-ka kutul cwungeyse Mary-man-i ku il-ul -and -NOM they among -onlv-NOM the work-ACC hay-ess-ta. do-PST-IND 'As for John and Mary, among them only Mary did the work. (13) is a simplex sentence, but there are two nominative NPs. We claim that 'John-kwa Mary (John and Mary)' is a topic, and the subject is 'Mary-man-i (only Mary)'5. With this sentence possible, one may say that in (12), 'John-kwa Mary (John and Mary)' and 'Mary-man (Mary only)' both belong to the main clause. This means that there may be a zero anaphora in the embedded subject slot, which may be coreferential with the subject 'Mary'. However, this line of argument does not go through. The subject in (13) 'Mary—man (Mary only)' eannot move rightward crossing over the nominative-marked topic. (14) *(a) Kutul cwungeyse Mary-man-i John-kwa Mary-ka they among —only-NOM -and -NOM ku il-ul hay-ess-ta. the work-ACC do-PST-IND 'As for John and Mary, among them only Mary did the work.' 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. *(b) Mary-man-i John-kwa Mary-ka kutul cwungeyse -only-NQM -and -NOM they among ku il-ul hay-ess-ta. the work-ACC do-PST-IND 'As for John and Mary, among them only Mary did the work.' In (14a) the subject 'Mary' and a postpositional phrase 'kutul cwungeyse (among them)' have moved leftward crossing the nominative topic 'John-kwa Mary (John and Mary)', while in (14b) only the subject has moved. The result is that both of them are ungrammatical. Actually the ungrammaticality of the examples in (14) may not be caused by the subject's crossing over the topic. It is due to the restriction that nominative marked NPs within the same clause cannot cross over each other, and these sentences violate this restriction. The usual topic marker in Korean is (n)un. (15) John-un Mary-lul saranghay-ess-ta. -TOP -ACC iove-FSx-IND 'As for John, (he) loves Mary.' The direct object 'Mary' can cross over the topic 'John' without causing any problem of grammaticality. (16) Mary-lul John-un saranghay-ess-ta. -ACC 'As for John, (he) loves Mary.' This is true also when the direct object is topic-marked, and there is a nominative subject. 166 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (17) Ku chayk-un Mary-ka ilk-ess-ta. the book-TOP -NOM read-PST-IND 'As for the book. Mary read (that).' The object 'ku chayk (that book)' is topicalized, being marked with the topic marker '-un', and there is a nominative subject 'Mary' in the sentence. If it is the case that the nominative subject cannot cross over the topic rightward, then the next sentence should be ungrammatical. (18) Mary-ka ku chayk-un ilk-ess-ta. -NOM the book-TOP read-PST-IND 'As for the book, Mary read the book.' This sentence is, however, not ungrammatical. This means that sentences (14a-b) are ungrammatical not because the nominative subject has crossed over the topic. We claim that the real reason why they are ungrammatical is because the topic is also nominative-marked. There is a restriction in Korean that nominative NPs within the same clause cannot cross over each other, as mentioned in chapter 3. The sentences (14a—b) violate this restriction. In Korean there are several structures which allow more than one nominative NP in the sentence. (19) (a) Mary—ka son—i khuta. -NOM hand—NOM big 'As for Mary, (her) hand is big. ' *(b) Son-i Mary-ka khuta. hand-NOM -NOM big 'As for Mary, (her) hand is big. ' 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (20) (a) Haksayng-i yel myeng-i wa-ass-ta. student—NOM ten CL—NOM eome-PST-IHD 'As for students, ten persons came.' *(b) Yel myeng-i haksayng-i wa-ass-ta. (19a) has a so-called possessor ascension structure, and (20a), a floating numeral quantifier. In a simplex sentence, nominative NPs cannot cross each other, thus the examples (19b) and (20b) are ungrammatical. The next examples also show the same effect. (21) (a) Sayngsen-i yene-ka ceyil khuta fish-NOM salmon—NOM most big 'As for fish, salmon is biggest.' *(b) Yene-ka sayngsen-i ceyil khuta. salmon-NOM fish-NOM most big (22) (a) John-i ton-i philyohata. -NOM money—NOM need 'John needs money.' *(b) Ton—i John—i philyohata. money—NOM —NOM need (21a) has a "membership” relationship between the two nominative NPs. The first nominative NP represents a 'whole set' and the second a 'member of the set'. Both of (21b) and (22b) are ungrammatical because the order of the two nominative NPs has changed. All of the double nominative sentences show that nominative NPs within the same clause cannot cross each other. Thus the real reason why (14a) and (14b) are ungrammatical is because the two nominative NPs 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. within the same clause have crossed each other. However, if the two nominative NPs are in two separate clauses, then the restriction is relaxed. (23) John~i [Bill-i ku il-ul hay-ess-ta]-ko malhay-ess-ta. -NOM -NOM the work-ACC do-PST-IND-COMP say-PST-IND 'John said that Bill did the work.' There are two nominative NPs, 'John' and 'Bill'. 'John' is the main clause subject, and 'Bill', the embedded subject. The embedded subject cannot move leftward, crossing the main clause subject because of the restriction, otherwise the main clause subject will be interpreted as the embedded subject, and the embedded subject, the main clause subject. However, 'John' can be placed right before the main verb. (24) [Bill-i ku il-ul hay-ess-ta]-ko John-i malhay-ess-ta. -NOM the work-ACC do-PST-IND-COMP -NOM say-PST-IND 'John said that Bill did the work.' This is because the embedded subject is not moved alone out of the embedded clause. The clause structure is maintained just as in (23), and the main clause subject is placed next to the main verb. This tells us that in a complex sentence the nomi native embedded subject cannot move leftward crossing the nominative main clause subject, but the main NP can be preverbally positioned. This can be a gccd test for whether the example (12) is a complex or a simplex sentence and also may be a good test for whether 'kutul cwungeyse Mary-man-i 16S Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (only Mary among them) ' belongs to the main clause or the embedded clause. For the sake of convenience, (12) is repeated here. (25)(=12) John-kwa Mary-ka [kutul cwungeyse Mary-man^-i -and -NOM they among -only-NOM ku il-ul ha]-lyeko nolvekhay-ess-ta. the work-ACC do-COMP try-PST-IND 'John and Mary tried (among them, only Mary) to do the work.' If this sentence is complex, the leftward movement of the embedded subject only should be thwarted, but placing the main clause subject in the preverbal position should be allowed. That is what we find. (26) [Kutul cwungeyse Mary-man.-i ku il-ul ha]-lyeko they among -only-NOM the work-ACC do-COMP John-kwa Mary^ka nolyekhay-ess-ta. -and -NOM try-PST-IND 'John and Mary tried (among them, only Mary) to do the work.' The main clause subject 'John-kwa Mary (John and Mary)' is preverbally placed, causing no grammaticality problem. If the embedded subject 'Mary' is a part of the main clause, then as we saw above with examples in (14), the sentence should be ungrammatical because the restriction says that within a simplex sentence, nominative NPs cannot cross each other. The fact that (26) is grammatical shows that it is a complex sentence, and 'Mary' is indeed the embedded subject. The same argument can be made when the embedded subject is a numeral. 170 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (27) Haksayngtuli-i [kutul cwungeyse twu myeng-man—i student-NQM them among two CL—only-N(JM ku il-ul ha]-lyeko nolyekhay-ess-ta. the work-ACC do-COMP try-PST-IND 'The students tried (among them, only two) to do the work.' As the ungrammaticality of (20b) shows, the main clause subject 'haksayngtul (students)' cannot be placed immediately before the main verb 'nolyekhata (try)' if the nominative numeral quantifier 'twu myeng (two persons) ' is a part of the main clause. The main clause subject can, however, be placed in that position as the following example shows. (28) [kutul cwungeyse twu myeng-man,-i ku il-ul they among two CL-only-NOM the work-ACC ha]-lyeko haksayngtul^i nolyekhay-ess-ta. do-COMP students-NOM try-PST-IND 'The students tried (among them, only two) to do the work.' This is possible because the sentence is complex, and the numeral quantifier is not a part of the main clause, but the embedded subject. The next example also shows that the embedded subject may not need to be coreferential with the main clause subject, not even partially. (29) Kim sajangnin^-i [caki hoysa^-ka ku phurojeykthu-lul president-NOM self company-NQM the project-ACC ha]-lyeko nolyekhay-ess-ta. do-COMP try-PST-IND 'President Kim tried (self's company) to do the project.' 171 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The main clause subject 'president Kim' is the one who manages 'his company', the embedded subject. But they are not corefe— rential, thus the main clause subject does not control the embedded subject6. The next example illustrates a similar situation. (30) John^i [caki kacok.-i ku il-ul ha]-lyeko -NOM self family-NOM the work-ACC do-COMP nolyekhay-ess-ta. try-PST-IND 'John tried (self's family) to do the work.' 'John' is a part of 'self's family', but not coreferential with it. 'John' may or may not be a part of the family which is the agent of the action involved in the embedded clause. Again the embedded subject is not controlled by the main clause subject since they are not coreferential. All the examples we have dealt with so far show that in Korean 'try'-type verbs do not require the embedded subject— covert or overt— to be exclusively controlled by the main clause subject. When the embedded subject is empty, it is more prone to pragmatics than when it is filled by an overt element. It is true that when the embedded subject slot is empty, the primary interpretation given to the sentence is that the embedded subject is coreferential with the main clause subject, thus controlled by the main clause subject. One of the most frequently used verbs in the control literature is 'promise'. This is because there can be an object within the main clause even though it shows 'subject control'. In English-type languages, even though there is an 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. object, the covert subject of the infinitival complement has been assumed to be exclusively controlled by the subject when the infinitival complement is not a passive structure. In the next subsection, we will look at the 'promise'-type control structure in Korean. We will argue that just as in the 'try'— type ’ /erb construction, the covert embedded subject is not always controlled by the subject, but can be controlled by the subject, the subject and some other person(s) combined, the subject and the object combined, or a third party. (A.2.) 'Yaksokhata (promise)'-type verb 'Promise'-type subject control has puzzled many linguists because there is an object which could be a controller of the covert subject of the infinitival clause, but the controller of the infinitival covert subject is allegedly always the main clause subject. This, as some people have indicated, violates Rosenbaum's (1967) generalization, which is based on 'adjacen cy' . (33.) jnKn promissd Mssry to 2TwuCi the Jscclc* The covert subject of the infinitival clause in the English sentence (31) is controlled by the main clause subject 'John', not by the main clause object 'Mary' . This has been accepted by almost all linguists, if not all. But we argue that it might not be entirely true. It is true that the main clause 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. subject 'John' is making the promise of 'reading the book'. But it does not mean that the main clause subject will necessarily perform the action of 'reading the book' alone. It could be the case that John's promise was that he and Mary read the book together. This point can be explained better if there is the word 'together' in the infinitival clause. (32) John promised Mary to read the book together. The person who promised is certainly 'John', but the agent of the action involved in the infinitival complement is not exclusively 'John'. It is most likely that 'John' and 'Mary' are reading the book together. So the agent of the event involved in the infini-tival clause is 'John' and 'Mary' together, not 'John' alone. This means that the covert subject of the infinitival clause in (32) is not the main clause subject. If this is true, then it is not possible to claim that the covert subject of the infinitival clause is corefe— rential with the main clause subject. We may say that in the case of (32) the controller of the covert subject of the infinitival clause is not the main clause subject alone, but is the combination of the main clause subject and the main clause object. A similar kind of claim can be made in Korean with 'vaksokhata (promise)' as the main verb. (33) Jonn-i Mary-eykey [0 ku il-ul ha]-ki—lo -NOM -DAT the work-ACC do-COMP-Loc yaksokhay-ess-ta. promise-PST-IND 'John promised Mary to do it.1 174 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The subject position of the embedded clause can stay empty as in (33), or be filled with an overt element as in (34a) and (34b). (34) (a) Joh^-i Mary-eykey [caki^-ka ku il-ul ha]-ki-lo -NOM -DAT self-NOM the work-ACC do-COMP-LOC yaksokhay-ess-ta. prcmise-PST~INB 'John promised Mary (self) to do the work.' (b) John^i Mary-eykey [ku^ka ku il-ul ha]-ki-lo -NOM -DAT he-NOM the work-ACC do-COMP-LOC yaksokhay-ess-ta. promise—PST—IND 'John promised Mary (him) to do the work.' The embedded subject position is filled with an anaphor in (34a), and with a pronoun in (34b), both of which are corefe rential with the main clause subject. This possibility of having an overt element fill the embedded subject position is different from English since in English the embedded subject position must stay empty. If having a covert embedded subject as in (33) or embedded subject coreferential with the main clause subject as in (34) are the only options, Borer (1989) and S.K. Kim (1994) correctly predict that the covert or overt embedded subject must be coreferential with the main clause subject. Just like the 'nolyekhata (try)'-type, there are other possibilities where the common claim that the embedded subject is controlled by the main clause subject does not stand. Just like the English sentence (32), for instance, the embedded subject could be the combination of the main clause subject and the main clause object. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (35) John.j-i Mary-eykey [caki-wa kunye--ka ku il-ul kachi -NOM —DAT self—and she—NOM the work-ACC together ha]-ki—lo yaksokhay-ess-ta. do-COMP-LOC promise-PST-IND 'John promised Mary (himself and her) to do the work together.' The embedded subject slot is occupied by the NP 'caki—wa kunye (self and she)'. 'Caki (self)' refers to the main clause subject 'John', and 'kunye (she)' refers to the main clause object. Just like the English sentence (32), the embedded subject is the combination of the main clause subject and the main clause object. What 'John' promised is that someone will do the work. It is not absolute that what John promised is that it is only he himself who will do the work although he is certainly included among the potential agents of the action in (35). This point is clearly illustrated in the following example. (36) John.:—i Mary—eykev [caki-wa Bill--i ku il-ul -NOM -DAT ” self-TOP -NOM the work-ACC haj-ki-lo yaksokhay-ess-ta. John promised Mary (himself and Bill) to do the work.' John promised that he and Bill will do the work, not ke alone. The verb 'yaksokhata (promise) ' does not require that the embedded subject must be coreferential with the main clause subject. It does not require that the referent must be within the sentence. Therefore, in the case of Korean, it is not correct to exclusively call it 'subject control' structure when the main verb is 'yaksokhata (promise)' if 'subject control' is defined as the covert subject (the optionally 176 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. overt subject also in the case of Korean) being coreferential (/identical) with the main clause subject. The next example also shows that the embedded subject is not controlled by the main clause subject. (37) John.j-i Mary-eykey [caki-wa kunye-wa Bil^-i ku il-ul -NOM -DAT self-TOP she-TOP -NOM the work-ACC ha]-ki-lo yaksokhay-ess-ta. do-COMP-LOC promise-PST-IND 'John promised Mary (himself, her, and Bill) to do the work.' Again if the embedded subject, must be controlled by the main clause subject, this sentence should not be grammatical since the embedded subject not only includes the main clause subject and the main clause object, but also includes a third party who is not introduced in the sentence. But (37) is a perfect ly grammatical sentence. Thus the above examples (35)-(37) d e m o n s t r a t e t h a t i n K o r e a n t h e e m b e d d e d s u b j e c t d e e s n o t n e e d to be identical with the main clause subject. It is also not necessary that the main clause subject should be a part of the embedded subject. (38) Kim pwucang.j-i sajang-eykey [caki-uy pwuse manager-NOM president-DAT self—GEN department sawentul- —i ku il-ul ha]—ki-lo yaksokhay-ess-ta. emploee-NQM the work-ACC do-COMP-LOC promise-PST-IND 'Manager Kim promised the company president (his department employees) to do the work.' In (38), the agent for the action involved in the embedded clause may not need to include the main clause subject in 177 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. order to be grammatical. The manager has enough ''authority' and 'power' to promise something on behalf of the whole group of employees who belong to his department, and will actually do the work. When there is no such pragmatic relationship between the main clause subject and the embedded subject, the sentence becomes little less acceptable. ?(39) John-i Mary-eykey [Bill-i ku il-ul ha]-ki-lo -NOM -DAT -NOM the work-ACC do-COMP-LOC yaksokhay-ess-ta. promise-PST-IND 'John promised Mary (Bill) to do it.' A part of the reason why this sentence is less than perfect is that the relationship between the main clause subject 'John' and the embedded subject 'Bill' is not clear. One cannot know without further context if John has enough authority to pro mise on behalf of Bill. If one knows, and the context makes it clear that for example, 'Bill' is an actor and 'John' is his agent, it is not hard to imagine that John can represent Bill and make promises on his behalf. This kind of clear context helps one to rather easily accept the sentence as grammatical. Another interesting thing is that when the main clause subject represents more than one person, it can split to be the embedded subjects of two or more embedded clauses. (40) John-kwa Mary-ka Bill-eykey [John-i sakwa-lul mek-ko, -and -NOM —DAT -NOM apple-ACC eat-and Mary-ka pay-lul mek]-ki-lo yaksokhay-ess-ta. -NOM pear-ACC eat-COMP-LOC promise-PST-IND 'John and Mary promised Bill that John would eat an apple, and Mary would eat a pear.' 17S Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In this example, there are two embedded clauses and each embedded subject is a part of the main clause subject. This kind of phenomenon is not restricted only to when the main clause subject represents more than one person. The second embedded subject could be either the main clause object, a third party who is not introduced in the sentence, or a part of the main clause object. (41) John-i Mary-eykey [caki-ka ku il-ul ha-ko, kunye-ka -NOM —DAT -NOM the work-ACC do-and -NOM talun il-ul ha]-ki-lo yaksokhay-ess-ta. another work-ACC do-COMP-LOC promise-PST-IND 'John promised Mary that he (lit. self) do the work, and she do a different work.' (42) John-i Mary-eykey [caki-ka ku il-ul ha-ko, Bill-i -NOM -DAT self—NOM the work-ACC do-and -NOM ku il-ul ha]-ki-lo yaksokhay-ess-ta. the work-ACC do-COMP-LOC promise-PST-IND 'John promised Mary that he (lit. self) do the work, and Bill do a different work.' (43) John-i Mary—wa Bill-eykey [caki-ka ku il-ul ha-ko, —NOM —and —DAT self—NOM the work-ACC do—and Mary-ka ku il-ul ha]-ki-lo yaksokhay-ess-ta. -NOM the work-ACC do-COMP-LOC promise-PST-IND 'John promised Mary and Bill that he (lit. self) do the work, and Mary do a different work.' As the English translation for (40) indicates, the Korean sentence may be similar to the English structure with an embedded clause led by the 'that' complementizer, rather than one with an infinitival complement. In English, the infiniti val subject is restricted either to be coreferential with the main clause subject, or at least to include the main clause subject as a part of it. But when it is a 'that'-led embedded 179 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. clause, there is no such restriction. However, one distinction between the English sentences with a 'that'-led embedded clause and the above Korean examples with a 'ki-lo'-led embedded clause is that in English, tense can appear rather freely with the embedded verb, but in Korean, tense is suppressed. Thus any attempt to assign tense to the embedded verb in a Korean sentence ends up creating an ungrammatical sentence. This is because when the combination of the complementizer 'hi' and the locative 'io' marks the embedded clause, the tense is suppressed meaning this combination requires the embedded clause to be non-finite. (44) John-i Mary-eykey [ku il-ul *ha-l]-ki-lo -NOM -DAT the work-ACC do-FUT-COMP-LOC *hay-ess- do-PST yaksokhay-ess-ta. 'John promised Mary that (he) will do the work.' 'John promised Mary that (he) did the work.' In this sense, the Korean examples are similar to the English examples with an infinitival clause because tense is suppressed in the English infinitival clause. Thus the Korean embedded clause with 'yaksokhata (promise)' behaves like the English infinitive in tense restriction and behaves like the 'that'—led clause in that the embedded subject position is filled by an overt element. In the latter case in English, however, it is required that the embedded subject must be present. 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Just like the case of 'nolyekhata (try)', a numeral quantifier can be used as an embedded subject in 'yaksokhata (promise)'-type in the place of an NP. (45) Haksayng-tul^i Sensayngnim-eykey [caki-tul cwungeyse student—PL—NOM teacher—DAT self—PL among yel myengj-i ku ii-ui ha]-ki-lo yaksokhay-ess-ta. ten CL-NGw the work-ACC do-COMP-LOC promise-P5T-INB 'The students promised the teacher that ten of them will do the work.' As shown in subsection (A.I.), if the sentence is monociausal, the independent numeral quantifier cannot precede the NP. (46) *(a) caki-tul cwungeyse yel myeng-i Haksayngtul-i self-PL among ten CL-NOM students-NOM ku il-ul hay-ess-ta. the work-ACC do-PST-IND 'As for the students, ten of them did the work.' *(b) yel myeng-i haksayngtul—i caki-tul cwungeyse ten CL—NOM students—NOM self—PL among ku il-ul hay-ess-ta. the work-ACC do-PST-IND 'As for the students, ten of them did the work.' In (46a), the phrase 'caki-tul cwungeyse (among themselves)' and the numeral quantifier 'yel myeng (ten persons)' precede the NP, and in (46b) only the numeral quantifier precedes the NP, resulting in both of them being ungrammatical. This shows that the NP must precede the numeral quantifier in a simplex sentence. If the numeral quantifier is actually within the main clause, not within the embedded clause, the next sentence should be ungrammatical. 181 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (47) Sensayngnim-eykey [caki—tui cwungeyse yel mveng -i teacher-DAT self-PL among ten CL-NOMJ ku il-ul ha]-ki-lo haksayngtul^i yaksokhay-ess-ta. the work-ACC do-COMP-LOC students-NOM promise-PST-IND 'The students promised the teacher that ten of them will do the work.' The sentence (47) is grammatical even though the numeral quantifier precedes the NP. This shows that the numeral quantifier is within the embedded clause, functioning as the embedded subject. As we have seen in the 'nolyekhata (try)'—type example (11), the covert embedded subject is not always coreferential with the main clause subject. (48) Kim pwucang^-i sacang-eykey [ 0 - hyuka-lul han saram-ey manager—NOM president-DAT leave-ACC one person-per twu cwuil—sshik kat]—ki-lo yaksokhay-ess-ta. two week-each take-COMP-LOC promise-PST-IND 'Manager Kim promised the president that the employees will each take two weeks' leave.' The main clause subject 'Kim pwucang (manager Kim)' is singu lar, whereas the covert embedded subject must be plural due to the fact that the phrase 'han saram-ey twu cwuil-ssik (two weeks-per-person)' requires a plural subject. If a singular pronoun 'ku (he)' is used in the embedded subject position, the sentence becomes grammatical. *(49) Kim pwucang^-i sacang—eykev [k^—ka han saram-ey "he—NOM twu cwuil-sshik hat]-ki-lo yaksokhay-ess-ta. 182 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The embedded subject must not be coreferential with the main clause subject since it makes it ungrammatical as in (49) . This example shows that even when the embedded subject is empty, it is not always coreferential with the main clause subject. So far we have demonstrated that in the case of the so- called 'subject control' structure in Korean, regardless of whether it is 'nolyekhata (try)', which has only one main clause argument, or it is 'yaksokhata (promise)', which has two main clause arguments, it does not behave like the English structure with an infinitival clause, but a structure with a 'that'-led clause. This similarity is due to the fact that the embedded subject, whether covert or overt, does not necessari ly need to be identical with the main clause subject. But one difference between them is that in the case of the Korean structure, the tense of the embedded verb is supressed, while this is not the case with the English structure. There is ano ther structure in Korean which is similar to the 'that'—led clause structure. Just as a 'that'—led clause can come as a complement for many verbs, a 'ko'-ied clause in Korean can be used as a complement for many verbs, 'ko' can be used as the complementizer instead of the combination of 'ki' complementi zer and '-lc' marker for the verb 'yaksokhata (promise)'. For the relationship between various complementizers and the verbs, refer to N.K. Kim (1984) and J.D. Kim (1988). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (50) John-i Mary-eykey [ku il-ul ha-keyess-ta]-ko -NOM -DAT the work-ACC do-FUT-IND-COMP yaksokhay-ess-ta. promise-PST-IND 'John promised Mary that (he) will do the work.' In this example, the future tense morpheme '—keyess—' is attached to the embedded verb 'hata (do)'. This is similar to the English translation in the sense that the embedded verb in English is also tensed. This shows that the Korean examples up to (47) with 'yaksokhata-' as the main verb are rather diffe rent from the English 'that'-led complement structure, and (50) is similar to the English one. It is not possible to have a past tense attached to the embedded clause when the main verb is 'yaksokhata (promise)' whether the complementizer is 'ki' or 'ko' because the event described by the embedded clause is non-factual, or irrealis. Due to this fact, the embedded clause cannot be a past event. 'Mayngseyhata (swear)', however, can have a complement which describes a past event or a future event because the complement could be either factual or non-factual. (51) (a) John swore to Mary that he would read the book. (b) John swore to Mary that he read the book. The event described by the complement clause (51a) is non- factual, or unrealized, and the event described by the complement clause (51b) is factual, or realized, which has 184 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. happened, or at least claimed to have happened. When the embedded clause in Korean is led by the 'ko' complementizer, both of them are possible. (52) (a) John-i Mary-eykey [0 ku chayk-ul ilk-keyess-ta]-ko -NOM -DAT the book—ACC read—FUT—IND—CGMP mayngseyhay-ess-ta. swear-PST-IND 'John swore to Mary that he would read the book.' (b) John-i Mary-eykey [0 ku chayk-ul imi ilk-ess— -NOM -DAT the book—ACC already read—PST— -ta]-ko mayngseyhay-ess-ta. -IND-COMP swear-PST-IND 'John swore to Mary that he had already read the book.' The embedded clause in (52a) is non-factual, while the one in (52b) is factual, reporting a past event. Thus we may conclude that the Korean 'ko'-led complement clause is quite similar to the English 'that'-led complement clause. When the complemen tizer is 'ki' and the locative marker '-io' follows it, again the tense of the embedded verb is suppressed. (53) John-i Mary-eykey [0 ku chayk-ul ilk]-ki-lo -NOM -DAT the book-ACC read-COMP-LOC mayngseyhav-ess—ta. swear-PST-IND 'John swore to Mary to read the book.' Another problem with identifying 'ki'+'-lo' with the 'that'- led complement is that in the latter case, the embedded verb may be marked with future tense, while it is not allowed with the former. This situation arises not only with 'swear', but also with 'promise'. In English, with either of these verbs as 185 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the main verb, the embedded verb should be future tensed when the embedded clause is marked by 'that', except when the complement clause for 'swear' denotes a past event. This is because the event described by the complement clause is something which has not happened, but will happen. This shows that as far as the tense restriction for the embedded clause is concerned, the Korean examples with 'yaksokhata (promise)' or 'mayngseyhata (swear)', when led by the 'ki'+'lo' combination behave like the English ones with an infinitival complement, rather than the ones with a 'that'-led complement clause. As for the fact that the embedded subject can be something other than the main clause subject, however, the Korean examples with 'ki'+'-lo' behave like the English examples with a 'that'-led complement clause because the ones with an infinitival clause do not allow something other than the main clause subject to be the controller of the embedded subject, except when it is pragmatically or semantically clear that the embedded subject can be a larger entity including the main clause subject as shown in (32) . The choice of the embedded subject in a sentence with a 'that'-led complement is rather flexible in the sense that it does not need to be always identical or corefential with the main clause subject. One last thing to mention is that when the main verb is 'yaksokhata (promise)', or 'mayngseyhata (swear)', the argument in the main clause, which is not the main clause 186 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. subject, does not need to be an indirect object. It could be an oblique argument with a postpositional particle wa (with)'. (54) John-i Mary-wa [0 ku il-ul ha]-ki-lo -NOM -with the work-ACC do-COMP-LOC yaksokhay-ess-ta. promise—PST-IND 'John promised with Mary to do the work.' One may argue that 'Mary-wa (with Mary)' belongs to the embedded clause, giving the interpretation of 'John promised someone to do the work with Mary'. This sentence is actually ambiguous because there is no overt embedded subject, and the position of 'Mary-wa (with Mary)' can be either in the main clause or in the embedded clause. If it is within the embedded clause, then the interpretation would be 'John promised (someone) to do the work with Mary'. If it belongs to the embedded clause, however, then it must not be in the position which immediately precedes the main verb because in Korean any element which belongs to the embedded clause cannot be placed at the right side of the embedded clause. (55) John-i [Bill-i Mary-wa ku il-ul hay—ess-ta]-ko -NOM -NOM -with the work-ACC do-PST-IND-COMP malhay-ess-ta. say-PST-IND 'John said that Bill did the work with Mary.' The English interpretation clearly shows that the oblique 'Mary-wa (with Mary)' in (55) belongs to the embedded clause. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. If it is placed immediately before the main verb, the interpretation changes. (56) John-i [Bill-i ku il-ul hav-ess-tal-ko Mary-wa -and malhay-ess-ta. 'John spoke with Mary that Bill did the work.' (56) does not share the same interpretation with (55) because the elements which come between the embedded verb and the main verb must belong to the main clause. This can be used as a test for the sentence (54). If 'Mary-wa (with Mary)' in (54) is an element which belongs to the embedded clause, then it cannot be placed between 'ha-ki-lo', and the main verb 'yak sokhata (promise)' because only main clause element can be placed there. (57) John-i [0 ku il-ul ha]-ki-lo Mary-wa -NOM the work-ACC do-COMP-LOC -with vaksokhav—ess—ta. promise-PST-IND 'John promised (to) Mary to do the work7. ' This proves that the oblique 'Mary-wa (with Mary)' belongs to the main clause. This gives the interpretation 'John promised Mary to do the work', not 'John and Mary promised (someone) to do the work.' In this section, we argued that the Korean analog of the English subject control structure is quite different from the English one. Regardless of whether the main verb is 'nolyek- hata (try)'-type, which only has one main argument, or 'yak- 188 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sokhata (promise)' with two main arguments, the embedded subject position does not need to be empty, and the covert (/overt) embedded subject does not need to be identical with the main clause subject. In this sense, they behave quite similarly to the English examples with a 'that'—led complement clause because the English examples with a 'that'-led comple ment clause do not require the embedded subject to be identi cal with the main clause subject, but rather free. This is also true in that the embedded subject can be overt. (A.3.) 'Wenhata (want)'-type Subject Control In the previous subsections, we took a look at the 'nolyekhata (try)', and 'yaksokhata (promise)'-type saliently subject control structure. In both cases, we claimed that 'subject' control of the covert embedded subject position is not guaranteed, and when the pragmatics is at work, we saw it is the case that the embedded subject, when made overt, can be the combination of the main clause subject and the main clause object, or the main clause subject and a third party who is not mentioned within the sentence. In this section, we will deal with the 'wenhata (want)'-type saliently subject control structure. We claim, as in the case of 'nolyekhata (try) ' and 'yaksokhata', that the covert embedded subject is saliently controlled by the main clause subject, but when pragmatics is at work, it does not need to be the main clause subject which 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. solely controls the covert, or the overt, embedded subject position. When the embedded subject position is filled with an overt element, it can be coreferential with the main clause subject, or it can be someone else, or the combination of the main clause subject and another person not mentioned within the sentence. The embedded clause is assigned accusative case, acting like a direct object of the verb 'wenhata (want)'. There may occur an accusative NP which is not a part of the embedded clause, but it is quite restricted as a constituent of the main clause. For instance, it cannot be scrambled to the immediately preverbal position. This is quite similar to the fact that in syntactic causatives we dealt with in chapter 2, the accusative-marked causee cannot be placed in the imme diate preverbal position. This is also consistent with the fact that the controlee in 'myenglyenghata (order) ' and 'pwuthakhata (ask)' can be assigned accusative, but in this case it cannot be placed immediately before the verb. This is also the case with so-called subject-to-object raising (/ECM) where the accusative NP cannot be placed in the immediate preverbal position. We deal with these constructions in connection with the 'wenhata (want)'-type saliently subject control verb. The complementizer choice for 'nolyekhata (try)' and 'yaksokhata (promise)'-type subject control was different. '— lyeko' is used for 'nolyekhata', and 'ki' for 'yaksokhata'. For 'wenhata (want)-'-type the complementizer 'ki' is used. 19(3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (58) John-i [0 ku il-ul ha]-ki-lul wenhay-ess-ta. -NOM the work-ACC do-COMP-ACC want-PST-IND 'John wanted to do the work.' One distinction in using the complementizer 'ki' in the case of 'yaksokhata' and 'wenhata' is that for 'yaksokhata', the locative marker 'lo' follows the complementizer, but the accusative marker 'ul' is used with 'wenhata'. The embedded clause in (58) functions as the direct object of the verb 'wenhata', thus structurally the example (58) is similar to the following sentence. (59) John-i ku kwail-ul wenhay-ess-ta. -NOM the fruit-ACC want-PST-IND 'John wanted the fruit.' 'ki' is also used as a nominalizer as follows: (60) ilk-ki (reading); nol-ki (playing); mek-ki (eating) With everything neutral, in (58) the main clause subject con trols the covert embedded subject, which may optionally be present. (61) John.j-i [k^-ka ku il-ul ha]-ki-lul wenhay-ess-ta. -NOM he-NOM the work-ACC do-COMP-ACC want-PST-IND 'John wanted to do the work.' Considering that the embedded subject can be overtly express ed, this structure is different from the English infinitival structure. The overt embedded subject 'ku' may be preferen tial with the main clause subject, or the two may be disjoint Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in reference. The reflexive 'caki' may be used to make sure that the embedded subject is coreferential with the main clause subject. (62) John^i [caki^-ka ku il-ul ha]-ki-lul wenhay-ess-ta. -NOM self-NOM the work-ACC do-COMP-ACC want-PST-IND 'John wanted to do the work.' When the embedded subject is overt as in (61)-(62), it may not necessarily be coreferential with the main clause subject as one of the interpretations of (58) shows. (63) John-i [Mary-ka ku il-ul ha]-ki-lul wenhay-ess-ta. -NOM -NOM the work-ACC do-COMP-ACC want-PST-IND 'John wanted Mary's doing the work.' The overt embedded subject may also be the combination of the main clause subject and another entity which is not present within the sentence. (64) John.j-i [caki-wa Mary ^-k a ku il-ul ha]-ki-lul wenhay- -NOM self-and -N'OM the work-ACC do-COMP-ACC wanfc- ess—ta. PST-IND 'John wanted that he (lit. self) and Mary do the work.' Another characteristic of the 'wenhata'-type is that even when the embedded subject is not overtly present, it does not always need to be coreferential with the main clause subject. (65) John-i [0 kekiev ka]-ki-lul wenhay-ess-ta. -NOM there “go-COMP-ACC want-PST-IND 'John wanted to go there.' 192 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The salient interpretation for (65) is that the covert embed ded subject is coreferential with the main clause subject, but this is not the only interpretation possible. It is also possible to interpret the embedded subject as being 'John' and some other entity together. With 'kathi (together) ' within the embedded clause the latter interpretation is dominant. (66) John.j-i [0- ku umshik-ul kathi mek]-ki-lul wenhay- -NOM the food-ACC together eat-COMP-ACC want- ess-ta. PST-INB 'John wanted to eat the food together.' The agent of eating must be more than one person due to the presence of 'kathi (together)', thus the covert embedded subject is not coreferential with the main clause subject, but it includes the main clause subject. Even when the embedded subject is not overt, it could be something which is complete ly different from the main clause subject. (67) John.j-i [0^ ppaiii kkutna]-ki-iui wenhay-ess-ta. -NOM quickly end-COM-ACC want-PST-IND 'John wanted that (something) end quickly.' The intransitive verb 'kkutnata (end)' does not allow a [+human] subject, thus the main clause subject 'John' does not qualify to be coreferential with its subject. It could be the case that John wants 'his class', 'his work', or 'the movie' to end quickly. (68) JohUj-i [il--i ppali kkutna]—ki—lul wenhay—ess—ta. -NOM work-NOM quickly end-COMP-ACC want-PST-IND 'John wanted that the work end quickly.' 193 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The embedded subject position can also be occupied by an overt noun ' il (work)'. The last point we want to touch upon is that there can be an NP marked accusative which is not a part of the embedded clause or a regular part of the main clause. (69) John-i Mary-lul [0 yeki-ey o]-ki-lul wenhay-ess-ta. -NOM -ACC here-to come-COMP-ACC want-PST-IND 'As for Mary, John wanted her to come here.' The embedded verb 'kata (go)' does not require a direct object, thus the accusative NP 'Mary' is not a part of the embedded clause. This NP is coreferential with the covert embedded subject. (70) John-i Mary^lul [kunye^-ka yeki-ey o]-ki-lul wenhay- she-NOM ess-ta. PST-IND 'As for Mary, John wanted her to come here.' That 'Mary' in (69) and (70) is not a part of the main clause is shown by the fact that it cannot be placed right before the main verb. *(71) John-i [0 yeki-ey o]-ki-lul Mary-lul wenhay-ess-ta. -ACC What John wanted was 'Mary's coming here', not 'Mary' herself. Thus the direct object of 'wenhata (want)' is the embedded clause, not 'Mary'. The syntactic causative has the same structure when the causee is marked accusative. 1 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (72) John-i Mary-lul [0 miswul chayk-ul ilk]-key hay-ess-ta. -NOM -ACC art book-ACC read-COMP do-PST-IND '"As for Mary, John did something so that she would read an art book.' The direct object of the verb 'hata (do) ' is not present in the sentence, and the accusative NP 'Mary' cannot be the direct object because the direct object should be an 'inanimate being' or an event. Since 'Mary' is not the direct object of the verb, it cannot be placed right before the verb. *(73) John~i [0 miswul chayk-ul ilk]-key Mary-lul hay- -ACC ess-ta. PST-IND The subject-to-object raising (/ECM) also has the same structure. The next example shows that the accusative NP is supposed to move from the embedded subject, position. (74) John-i Mary-lul [ chencay-la]-ko sayngkakhay-ess-ta. -NOM -ACC genius-COP-COMP thimk-PST-IND 'John thought Mary to be a genius.' l£ 'Mary - ' is indeed the direct object of the verb 'sayngka— khata (think)', then it should be able to be right before the main verb. But this is not possible as the next example shows. *(75) John-i [0 chencay-la]-ko Mary-lul sayngkakhay-ess-ta. -ACC The reason why (71), (73), (75) are all ungrammatical is because they are not real direct objects of the main verb. We will tentatatively call the accusative NPs in these examples 195 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'accusative topic'. This is another big research topic, and we will work on this issue in future research since it is not directly related to what we try to establish in this study. In section (A), we showed that no matter whether we have the 'nolyekhata (try)'-type, 'yaksokhata (promise)'-type, or 'wenhata (want)'-type of the so-called subject control, the overt or covert embedded subject does not need to be corefe rential with the main clause subject. This shows that the sentences with these verbs in Korean are not the traditional ly-called 'subject control' structure which requires that the embedded subject be covert and be coreferential with the main clause subject. (B) Object Control In the previous section, we argued that when the embedded subject in a sentence with 'nolyekhata (try)', and 'yaksokhata (promise)' is filled with an overt element, it does not necessarily need to be coreferential with the main clause subject. It could be a part of the main clause subject, the combination of the main clause subject anci mam ciause object, or a third party (normally semantically or pragmatically related to the main clause subject). In this section, we will view the construction with 'seltukhata (persuade)', 'kangyohata (force)', 'myenglyenghata (order)', etc.. as the main verb, which is usually dubbed i r > / r J L > U Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'obligatory object control'. This is due to the fact that in many languages the unexpressed embedded subject is coreferen tial with the main clause object. (76) John persuaded Mary to go home. The agent of the action described in the above English infinitival clause is 'Mary', which is the main clause object, not the main clause subject 'John'. In the case of English there is no possibility for a third person, who is not intro duced in the sentence, to be the subject of the infinitival clause. The infinitival subject is, thus, coreferential with the main clause object. We argue in this section, however, that in Korean there is no 'obligatoriness' requirement for the embedded subject to be identical or corefential with the main clause object. There is also no constraint for the embedded subject position to stay ftm nf.v . — j - _ . (77) John-i Mary-lul [0 kukes-ul kongpwuha]-tolok -NOM -ACC that-ACC studv-COMP seltulhay—ess—ta. persuade-PST-IND 'John persuaded Mary to study that.' The salient interpretation for this sentence is that the covert embedded subject is identical with the main clause object 'Mary'. As we will see, however, that is not the only 197 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. option. We will first look at what has been said about object control in Korean, and then argue for the analysis advanced here. (B.l.) Previous Research on Object Control in Korean N.K. Kim (1978) argues that the underlined embedded subject position in the next example is empty due to the rule of 'equi-itF deletion' . (78) (his la) Nay-ka kim kwun-eykey [_____ ttena]-tolok I—NOM Mr.-DAT leave-COMP myenglyenghay-ess—ta. order-PST-IND 'I ordered Mr. Kim to leave.' His argument is that the covert embedded subject is equivalent to the main clause object 'Kim kwun (Mr. Kim)', and gets de leted by 'Equi-NP' deletion. He shows another set of examples to prove this point. (79) (his 32a-d) *a. Kim sensayng-un John--eykey [John.;—i i kos-ev Mr.-TOP -DAT —NOM this place-to oj-toiok myengiyenghay-ess-ta. come-COMP order-PST-IND *'Mr. Kim ordered John that John will come here.' ?b. Kim sensayng-un Johr^-eykey [ku^ka i kos-ey Mr.-TOP -DAT he-NOM this place-to o]—tolok rayenglyenghay-ess—ta. *'Mr. Kim ordered John that he will come here.' *c. kim sensayng-un John^-eykey [caki^ka i kos-ey Mr.-TOP -DAT self—NOM this place-to o]—tolok myenglyenghay-ess—ta. *'Mr. Kim ordered John that himself come here.' 13 S Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. d. Kim sensayng-un Johnc-eykey [__________i i kos-ey o]-tolok myenglyenghay-ess-ta. 'Mr. Kim ordered John to come here.' He claims, without much explanation, that this set of sen tences shows that the embedded subject must be identical with the main clause indirect object, not the main clause subject. We assume that what he says is that the pronominal 'ku' in (79b) is coreferential with the indirect object 'John', and the anaphor 'caki (self)' in (79c) is coreferen-tial with the main clause subject 'Kim sensayng (Mr. Kim)'. (79c) is out because according to him, the embedded subject must be coreferential with the main indirect object. N.K. Kim (1986) says that in traditional transformational grammar, (80a) is derived from (80b), by the process of 'equi NP deletion'. (80) (his 29a-b) /a \ .Tnhn«iir) M artr— ovVav T n in — ow 1/al— f a 1 a !/ ^ U p wv^tttttttt Mtttt J ~ jr ^ -TOP -DAT house-to go-COMP myengiyengnay-ess—ta. 'John ordered Mary to go home.' (b) John-un Mary-eykey [Mary-ka cip-ey ka]-tolok -NOM myenglyenghay-ess—ta. N.K. Kim (1986) adopts Williams's (1980) obligatory control theory, which argues that obligatory controls have the following properties. (81) (Williams's (26)) OC 1. Lexical NP cannot appear in the position of PRO. OC 2. The antecedent precedes the controlled PRO. OC 3. The antecedent c-commands the controlled PRO. OC 4. The antecedent is thematically or grammatically uniquely determined. OC 5. There must be an antecedent. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kim does not include 'OC 4' because he thinks that its deletion will not affect his argument seeing the obligatory control as 'predication' in Korean. He argues that the following Korean sentence meets all the requirements of 'OC 1— 5', with exception of not considering 'OC 4' for the Korean sentence. (82) (his (30a) John-un Mary-eykey [PRO cip-ey ka]-tolok -TOP -DAT house—to go-COMP myenglyenghay-ess-ta. order-PST-IND 'John ordered Mary to go home.' He claims that a lexical NP as required by 'OC 1' cannot appear in the 'PRO' position. If there is an overt NP in the place 'PRO', according to him, the sentence becomes ungramma tical . (83) (his (33)) *John-un Mary-eykey [Bill-i cip-ey ka]-tolok —NOM myenglyenghay-ess-ta. *'Jonn ordered Mary for Bill to go home.' (80b), however, is grammatical to us as mentioned and in (80b) there is an overt NP in the 'PRO' position which is coreferen tial with the main clause object. He claims that the Korean example meets 'OC 2' and 'OC 5' also because there is an ante cedent in the sentence, which is required by 'OC 5'. He also says that it precedes the PRO element, which is required by 'OC 2'. But the main clause object can be placed in the imme— 200 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. diate preverbal position following the embedded clause as shown in (84) without making the sentence ungrammatical. (84) John-un [0 cip-ey ka]—tolok Mary-eykey -TOP house-to go-COMP -DAT myenglyenghay-ess-ta. order-PST-IND 'John ordered Mary to go home.' Thus 'OC 2' of Williams is not required in Korean, either. The examples also meet 'OC 3', he claims, in that the antecedent 'Mary' c-commands the PRO. Having these conditions met, Kim agrees with Williams (1980) in reducing obligatory control to 'predication'. As we have seen in the above several points Kim made can be argued against. H.B. Lee (1987) claims that the following two sentences in (85) are different in their behaviors. (85) (his 51a-b) a. John-i Bill^eykey [0^ Mary-lul manna-tolok] -NOM -DAT -ACC meet-CGMF seltukhay-ess-ta. persuade—FST—IND 'John persuaded Bill to meet Mary.' b. John-i [0 Mary-lul manna-ass-ta]-ko -NOM -ACC meet—PST—IND—COMP maihay—ess—ta. say-PST-IND 'John said that (he) met Mary.' Ke claims that (85a) is a typical case of obligatory control structure, and the covert embedded subject must be bound by the main indirect object. (85b), on the other hand, is not obligatorily controlled by the main clause subject, so it can 201 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. refer to someone definite in the discourse. He claims that the fundamental difference between these two sentences is that the antecedent of the covert embedded subject in (85a) should be within some kind of local domain as an anaphor has the same 'antecedent'-requirement, whereas the empty embedded subject in (85fc) does not have this requirement just as a pronoun does not. This distinction makes him claim that the empty subject in (85a) is 'PRO', and the one in (85b), 'pro'. D.W. Yang (1982) shows that the empty embedded subject position in a typical object control structure with 'seltukhata (persuade)' can be readily filled by a pronoun which is still controlled by a main clause object. (86) (his (20)) John-i Bill^eykey [0i Mary-lul manna-tolok] -NOM -DAT -ACC meet-COMP seltukhay-ess-ta. persuade-PST-IND 'John persuaded Bill to meet Mary.' The fact that the empty embedded subject position can be filled by a pronoun shows that the embedded clause is a clause, not a VF. (87) (his (21)) John—i Biii--eykey [ku^ka Mary-lul manna-tolok] he-NOM seltukhay-ess-ta. 'John persuaded Bill that he meet Mary.' He claims that even when the embedded subject is realized by a pronoun, this pronoun is only controlled by the dative 202 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. object 'Bill', and says that the reason why the pronoun 'ku (he)' in (87) is only controlled by 'Bill' is because of the semantics of the main verb 'seltukhata (persuade)'. D.W. Yang (1984) argues that the embedded subject can be covert, or overt being filled by a pronoun or a name which is equivalent with the main clause object. (88) (his 3a) Chelswu-ka Yengswu^eykey8 [0i ttana]-tolok -NOM -DAT leave-COMP ku^ka he-NOM Yengswu . j —ka -NOM *cakij-ka self—NOM seltulhay-ess-ta. persuade-PST-IND 'Chelswu persuaded Yengswu to leave.' He claims that even if the embedded subject position is filled by a pronoun 'ku (he)', or a name 'Yengswu', it must refer to the main dative object 'Yengswu', thus it is an object control construction. The same point is made in D.W. Yang (1985). We will show, however, that this is not true. H.S. Lee (1986) claims that in Korean there is a difference between the behaviors of the embedded empty subjects depending on whether the embedded clause is finite or non-finite. He argues that when the embedded clause is non-finite, given the main verb is of the 'sultukhata (persuade)'-type, the covert embedded subject must be coreferential with the main clause object while when the embedded clause is finite the covert 203 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. embedded subject does not necessarily have to be coreferential with the main clause object. (89) (his (15)) John-un [ ei ilccik ttena]-tolok Biil^ui seltukhay-ess-ta. -TOP early leave-COMP -ACC persuade-PST-IND 'John persuaded Bill [PRO to leave early]. ' The embedded clause led by the ' tolok ■ ' complementizer is a non-finite clause, meaning [-tense]. He claims that the covert embedded subject must be coindexed with the main clause object 'Bill'. He claims that when the embedded clause is finite, the covert embedded subject could be coindexed with some other elements as well as the main clause object. (90) (his (17)) John-un Bill-eykey [ e ttena-ya ha-n-tal-ko -TOP -DAT leave-shouid—IMP-IND-COMP seltukhay-ess-ta. persuade-PST-IND 'John persuaded Bill that (he) should leave.' He says that the empty subject of the finite embedded clause in (90) can be coindexed with the main clause object 'Bill', the topic 'John', or an unexpressed discourse topic. The following examples given by him show that an overt lexical item can fill the embedded subject position in (89) . (51) (his (19)) John.j-un Bill^-eykey [kui . x-ka ttena]-tolok seltukhay- -TQP —DAT' he—NO?' leave—COMP persuade— ess-ta. PST-IND 'John persuaded Bill for him to leave.' (92) (his (20)) John-un Bill-eykey [Tom-i ttena]-tolok seltukhay-ess-ta. -TOP -DAT —NOM leave-COMP persuade-FST-IRD 'John persuaded Bill for Tom to leave.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In English when the embedded clause is an infinitival one, there cannot be a lexical item in the place of the empty subject in a sentence like (89). (93) (his (21)) *John- persuaded Bill. [him. . to leave]. 1 “ J I i J i X / A V / t i \ \ *34; *ulS \**)) *John persuaded Bill [Tom to leave]. Thus he claims that the empty embedded subject, PRO can occur in a position where an overt lexical item can appear. (B.2.) Our analysis of so-called obligatory object control If D.W. Yang (1982, 1984) is correct, the embedded subject should be identical with the main clause object whether it is overt or covert. This is against what H=S= Lee (1986) claims because Lee argues that there can be an overt lexical item in the embedded subject position as in examples (31)-(92) which is not identical with the main clause object. In (B.2.1) we agree with Lee in that an overt element which is not identical with the main clause object can appear in the embedded subject position. In (B.2.2.) we will show that we do not agree, though, with either N.K. Kins, D.W. Yang, or U.S. Lee in that when the embedded subject position is empty, it should always be identical with the main clause object. We will argue that the empty embedded subject does not need to be always identi- 205 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cal with the main clause object, and it can refez- to some entity which may or may not include the main clause object. Of course the salient interpretation will be the one which includes the main clause object. In (B.2.3), we will argue that it is not necessary for the controller to be present in the sentence. This will prove that in Korean the so-called control phenomenon is not syntactic, but largely pragmatic. (B.2.1.) Overt elements in the embedded subject position In English obligatory control sentences, the embedded subject position cannot be filled with an overt lexical item. (95) John persuaded Mary *her to go home. This is due to the fact that the infinitival clause structure has a restriction for the subject to be suppressed when there is a main clause object which is coreferential with it, pre sent as in (95) . But in Korean, thez^e can be an overt embedded subject even if the embedded clause is non-finite. Yang (1982, 1984) claims that the overt embedded subject must be identical with the main clause object. (96) (=87) John-i Bill^evkey [ku^ka Mary-lul manna-tolok] -NOM -DAT he-NOM -ACC meet-COMP seltukhay-ess-ta. persuade-PST-IND 'John persuaded Bill that he meet Mary.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. According to him 'ku (he)' in (96) must be coreferential with 'Bill'. But Lee (1987) claims that this is not the case. (97) (=91) John.j-un Bill.-eykey [kui/i/k-ka ttenaj-tolok -TOP -DAT he-NOM leave-COMP seltukhay-ess-ta. persuade-PST-IND 'John persuaded Bill for him to leave.' Lee claims that 'ku (he)' in (97) can be coreferential not only with the dative object 'Bill', but also with the main clause subject 'John', or a third party. We do not agree with Lee's judgment that the overt embedded subject can be corefe rential with the main clause subject, but we agree with him that an overt subject which is not coreferential with the dative object can occur in the embedded subject position9. In (97) there is no clear context to identify the pronoun 'ku (he)' to be a third person, thus it is sscst likely to be coreferential with the object 'Bill'. We will show that in a sentence with a verb like 'seltukhata (persuade)' the overt embedded subject is not necessarily coreferential with the object when there is a clear context inside or outside the sentence. It is true that the embedded clause in (97) is non-finite. Unlike an English infinitival clause, however, there can be an overt subject of the embedded verb as shown in (96) —(97) . We claim that the overt embedded subject does not need to be identical with the dative main clause object. The next sentence proves our point. 207 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (98) John-i Bill-eykey [ku-wa Mary-ka hamkkev ttena]-tolok —NOM —DAT he—and —NOM together leave-COMP seltukhay-ess-ta. persuade-PST-IND 'John persuaded Bill that he and Mary leave together.' The embedded subject 'ku-wa Mary (he and Mary)' is not identical to the dative object 'Bill'. This inconsistency, however, does not make the sentence ungrammatical, 'ku (he)' which is a part of the embedded subject is interpreted to be 'Bill'10, but it cannot refer to the main clause subject 'John' since the semantics of the verb 'seltukhata (persuade) ' prohibits it. If one wants to make sure that the dative object is a part of the embedded subject, then the name 'Bill' can be used. (99) John-i Bill-eykey [Bill-kwa Mary--ka hamkkey -NOM -DAT -and -ISOM together ttena]—tolok seltukhay—ess—ta. leave-COMP persuade-PST-IND 'John persuaded Bill that Bill and Mary leave together.' These examples clearly show that the Korean structure is not identical to the English structure with an infinitival clause. As shown by the English translations (98)-(99), the Korean structure is similar to the English one with a 'that'-led complement clause. If this is right, it is not surprising that an overt embedded subject can be different from the dative object. In English, the next sentence is possible. (100) John persuaded the singer's manager that the singer (should) perform at the party. 208 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The embedded subject 'the singer' in (100) is different from the main clause object, 'the singer's manager'. In a structure like (100), the main clause subject and the main clause object can form the embedded subject together. (101) John persuaded Mary that they should do the work together. 'They' in (101) can be the combination of 'John and Mary' or some other people. This means that when the complement clause is a 'that'-led one, then the embedded subject does not need to be identical to the main clause object. The Korean facts we see in the above also show that the overt embedded subject and the main clause object do not always need to be identical. In this sense, we support H.S. Lee's (1986) claim that the overt embedded subject can be disjointed from the main clause ob ject. One difference, though, between an English sentence like (101) and a Korean sentence like (99) with 'tolok' as the complementizer is that in the English sentence, tense can appear with the embedded verb, whereas it is not possible with the Korean one, showing that there is finite/non-finite difference. (102) John persuaded Mary that they will have to do the work together. In this English sentence, the future tense auxiliary 'will' can appear in the embedded clause, and the past tense also can be expressed. 209 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (103) John persuaded Mary that Bill already died. The complementizer 'tolok' in the Korean sentence suppresses expression of any tense. Thus any attempt to express a tense in the embedded clause results in ungrammaticality. (104) John-i Mary-eykey [kunyei-ka ku il-ul -NOM -DAT she-NOM the work-ACC ha-n]-tolok seltukhay-ess-ta. do-*PRE-COMP hay-ess-tolok do-*PST-C0MP 'John persuaded Mary that she do the work.' 'John persuaded Mary that she did the work.' This shows that the Korean structure is similar to the English structure with a ' that'-complement clause in that the embedded subject can be overtly expressed. They are different, however, in that in Korean, tense is suppressed in the embedded clause, being a non-finite one, while in English, tense can be ex pressed, thus behaving like the English to-infinitive as far as tense is concerned. In this subsection, we argued that it is not the case that the embedded subject, must be coreferential with the main clause object. When the embedded subject is overt, it can be coreferential with the main clause object, or the main clause object and a third person together, or someone (/something) that is not present in the sentence as shown in H.S. Lee(1986) . Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (B.2.2.) The empty embedded subject All the works reviewed in (B.l.) claim that the empty embedded subject must be coreferential with the main clause object in a sentence like the next one. (105) John-i Mary-eykey [0 talun il-ul ha]-tolok -NOM -DAT other work-ACC do-COMP kangyohay-e s s-ta. force-PST-IND 'John forced Mary to do another work'. According to their claim, the empty embedded subject in (105) must be identical with the dative main clause object 'Mary'. We have seen in (B.2.1.) that when the embedded subject is overt, it can be something other than the main clause object. If it is true that the covert embedded subject must be core ferential with the main clause object, why is it so? Is it due to the constraint for 'Equi-NP' deletion? Is the embedded subject deleted because it is equivalent to the main clause object? We argue that it is not the case that the covert embedded subject must be coreferential with main clause object. The next examples will prove our point. ^ku^-ka (he-Nom) (106) Sensayngnim-i pancang.-eykey [ 0j sakwa-lul han teacher-NOM class chief-DAT apple-ACC one saram-ev han key-sshik mek]-tolok myenglyenhay-ess-ta. peson-each piece-each eat“COMP order-PST-IND 'The teacher ordered the class president that (the students) eat one apple each.' What the teacher ordered the class president is that each student in the class eat one apple each. Unless there is more m i Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. than one class president in the class, the agent of the event in the embedded clause is the students in the class. Thus the covert embedded subject cannot be coreferential with the main clause object 'pancang (class president)'. This is why the pronoun 'ku (he)' cannot occur in the embedded subject posi tion. This becomes clear when the covert subject position is filled with an overt NP. (107) Sensayngnim-i pancang^eykey [ku pan-uy haksayngtul^-i teacher-NOM class chief—DAT the class-GEN students—KCM sakwa-lul han saram-ey han key-sshik mekj-tolok apple-ACC one person-at one piece-each eat-COMP myenglyenghay-ess-ta. order-PST-IND 'The teacher ordered the class president that the students in the class eat one apple each.' The class president is someone who represents all the students in the class. So when the teacher orders the class president to do something, in Korean, the order may be one for the entire class, not only for the class president. This example shows that even when the embedded subject is covert, it may not be coreferential with the object, thus disagreeing with N.K. Kim (1986), D.W. Yang (1982; 1984), or K.S. Lee (1586). This is, however, probably not syntactic, but pragmatic (Bernard Comrie, personal communication). The point we are trying to make with this example is that there is no obliga toriness for the covert embedded subject to be coreferential with the main clause object whether the reason is syntactic or pragmatic. The exact same point can be made with the example below: 9 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (108) John-i Mary^eykey [0, ku nolay-lul kathi -NOM -DAT the song-ACC together pwulu]-tolok seltukhay-ess-ta. sing-COMP persuade—PST-IND 'John persuaded Mary to sing the song together.' The presence of the phrase 'kathi (together)' shows that the actor (/agent) for the situation described in the embedded clause is more than one person. In the example (108), the covert embedded subject can be the combination of the main clause subject 'John' and the main clause object 'Mary', or the combination of the main clause object 'Mary', and somebody else who is not expressed in the sentence. Again the covert embedded subject is not identical to the main clause object. The English translation for (108) is also pragmatically ambiguous11. The people who actually sing can be 'John' and 'Mary' together, or 'Mary' and someone else (/a group of people) not expressed in the sentence, and mentioned in the previous discourse. What we have to consider is that it is not absolutely obligatory for the implicit embedded subject to be coreferential with the main clause object either in English or in Korean. When there is no clear pragmatic context involved, however, it is likely that the covert embedded subject will be coreferential with the main clause object in both languages. This is because the main clause object is the one present in the sentence, making it a strongly salient choice. If prag matics clearly indicates otherwise, then the possible candi date may be found outside the sentence. 213 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Suppose that the previous context for the following sentence is that the obstetrician thinks that it is the best for him to deliver the baby a little earlier than scheduled for the sake of the pregnant woman's safety. (109) The obstetrician persuaded the pregnant woman's family to deliver the baby three weeks earlier than scheduled. The person who is actually delivering the baby is 'the obste trician', not 'the pregnant woman's family'. There is always the possibility that the situation is that the pregnant woman's family will deliver the baby themselves. But it is not likely in reality especially if the obstetrician is around. The more likely situation is that the obstetrician will deli ver the baby. The verb 'persuade' in English is an object control verb, thus the main clause subject is not supposed to be the controller according to the syntactic constraint. But the pragmatics in this case overrides the syntax and the main clause subject controls the covert embedded subject. In the same vein, pragmatics works when the main verb is 'ask'. Depending on who are involved, this verb can function as a subject control verb, an object controx vem, or be ambiguous (Comrie, 1984; 1986), (cf: Chomsky, 1381; Franka, 1988). (110) (a) The teacher asked the child to go to the bathroom, (b) The child asked the teacher to go to the bathroom. In (110a), the person who is going to the bathroom will most certainly be 'the child', thus object control. In (110b), 214 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. either 'the child' or 'the teacher' can go to the bathroom, depending on the situation. If 'the child' is asking 'the teacher' permission to go to the bathroom, then it is the child who wants to go to the bathroom. This will be a case of subject control. If 'the teacher' signalled in the middle of the class that (s)he needs to go to the bathroom, then 'the child', who is one of the students in the classroom, can ask the teacher to go to the bathroom. In this case it would be 'object control'. (110b) is, therefore, ambiguous, and the pragmatics (/situation) decides on whether it would be a subject control case or an object control case. In this subsection, we argued that in Korean it is not absolutely obligatory for the covert embedded subject to be coreferential with the main clause object when the main verb is 'myenglyenghata (order)', 'seltukhata (persuade)' or 'kangyohata (force)', which are normally considered to be obligatory object control verbs. We saw that even in English, l r. nnr. 8bSOJ.Ut0 rnar rnp COV0rt SUbj9Ct of ths infinitival clause be coreferential with the main clause object when the main verb is 'persuade', or 'ask'. Pragmatics loosens the syntactic requirement of 'obligatory object control'. (B.2.3.) Optionality of the main clause object If the control relationship is between an overt main argument and a covert embedded subject, then the main argument 215 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. should be present in the sentence as Williams' (1980) 'OC 5' indicates, which says that "there must be an antecedent". This is what we see in English. In the case of the obligatory object control structure, the main clause object is present, and in the case of the subject control, the subject must be present12. In Korean, this condition is not met due to language-specific reasons. The fact that Korean is a pragma tically-geared language is shown through the phenomenon of zero anaphora. When they are clear in the discourse, almost all of the arguments can be deleted13. (111) John: Cwu-ess-ni? give-PST-QUES 'Did (you/he) give (something) (to someone)?' Mary: Cwu-ess-e. give-PST-IND '(I/he) gave (something) (to someone). In the dialogue (111), the speakers (A) and (B) assume that they do not need to mention the 'agent', the 'theme', or the 'goal' arguments due to the fact that they are clear to them, and they mutually agree that both of them share the knowledge about the unexpressed information. Otherwise, 'Mary' will ask 'John' to clarify whatever information John assumed Mary to share with him, but not understood by her. (112) John: Cwu-ess-ni? give-PST-QUES Mary: Mwues? What o i r X . J.O Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'Mary' in (112) does not know what was supposed to be the 'Theme' although she knows who the 'Agent' and the 'Goal' are, thus she has to ask 'John' what he was referring to. This shows that when there is any missing argument in the sentence, it is assumed to be understood unless challenged. Thus the interpretation of the zero anaphoras heavily depends on pragmatics of context and discourse. This also applies to the object control structure. As long as it is clear in the context, only the main verb can occur in the surface. (113) John: Seltukhay-ess-ni? persuade-PST-QUES 'Did (someone) persuade (someone else) (to do something)?' Mary: Eung Yes. Obviously 'Mary' knows what 'John' is talking about. Or else she may ask him to clarify. This means that the subject, the object, and the embedded clause all depend on the context. It l e n c * 4 k 1 » n 4 m kk n4> kr» / /■*/%*%'!- 1 f — 1 •**. 4 ■ •» » « W M V W u w W A U A M I W M M W W M W W W M b X W X i . <C b X V U O U X J L / JL 9 purely syntactic because in the John's question in (113), even the embedded clause is not present. This means that without help from pragmatics or context building up the control relation for this sentence cannot be done. The embedded clause can be added as follows: (114) [ 0 phyenci-lul ssu]-tolok seltukhay-ess-ni? letter-ACC write-COMP persuade-PST-QUES 'Did (someone) persuade (someone else) to write a letter?' 217 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. There are no main arguments overt in the above sentence. The covert embedded subject may be coreferential with the main clause object which is not expressed, thus should be drawn from the context. The main clause object may or may not be coreferential with the covert embedded subject as we have seen with examples (105} and (1C8). As we can see, m the case ox Korean, the control relation between a main argument and an embedded one depends on pragmatics much more than in English because in English the main arguments are obligatorily expressed in the sentence, and there is not much need to depend on the context for choosing the controller of the unexpressed embedded subject. In this subsection, we reviewed the literature on object control in Korean, and argued that when there is an overt embedded subject, it does not need to be coreferential with the main clause object, but is quite flexible as far as its indexation is concerned. This is quite similar to the English structure with a 'that'—led complement clause. We also argued that the covert embedded subject is normally coreferential with the main clause object, although this preference may change when pragmatics is taken into consideration. This pragmatic intervention also applies to a certain extent to English. We also looked at the fact that Korean is pragmatics- oriented especially with the interpretation of zero anaphors. When the main clause object is not present in an object control structure, it may not be easy to assign the relation- 218 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ship between the covert main clause object (/something else) and the covert embedded subject unless pragmatics clarifies it. In the next section, we will argue that in Korean the semantics of the main verb is not solely responsible for the embedded subject being controlled either by the main clause subject or the main clause object. We claim that the choice of the complementizer also plays a crucial role at times. (C) Semantics of the verb and complementizers What makes certain structures be 'subject' control and certain structures be 'object' control? In answer to this question, both Foley and Van Valin (1984) and Comrie (1986) introduce Searle's (1969; 1975) approach to speech acts. They claim that verbs like 'persuade' belong to 'directives' speech act category, and verbs like 'promise' belong to 'commissives' speech acts category. Searle (1975) defines 'directives' as ' attempts . . . by the speaker to get the hearer to do some thing. They may be very modest "attempts," as when I invite you to do it or suggest that you do it, or they may be very fierce attempts as when I insist that you do it. He also defines 'commissives' as " those illocutionary acts whose point is to commit the speaker ... to some future course of oig Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. action" (Searle 1975). Sag and Pollard (1991) use the semantic terms INFLUENCE TYPE ', ' COMMITMENT' , and ' EXPERIENCE' . Their 'INFLUENCE TYPE', which includes verbs like 'order', and 'permit', is similar to Searle's 'directives', although they claim it to be more general. Their 'COMMITMENT' type is meant to be more extensive then 'commissives■', in the sense that they include 'attempts' and 'promises'. The 'EXPERIENCE' type includes verbs like 'want', and 'expect'. Sag and Pollard claim that 'INFLUENCE TYPE' verbs trigger 'object control', and 'COMMITMENT' verbs and 'EXPERIENCE' verbs trigger 'subject control'. As we have seen earlier, in Korean the embedded subject can be optionally overt. When it is overt, it does not need to be coreferential either with the subject or with the object. Even when it is not overt, it is not obligatory for the embedded subject to be exclusively controlled by the subject or the object. Pragmatics plays a crucial role in creating this exception to an otherwise seemingly quite restricted constraint. In this section we will argue that in Korean it is not only the semantics of the verbs, but also the choice of the complementizer that helps decide the controller for the covert embedded subject. When the main verb has ' hat a (do) ' as a part of it, in some cases the verbal noun part can be deleted, leaving 'hata (do) ' as the main verb. The main verb in the next sentence is 'hata (do)'. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (115) John-i Mary-eykey [0 ku il-ul hay-ess-taj-ko hay- -NOM -DAT the work-ACC do-PST-IND-COMP do- ess-ta• PST-IND ♦'John did Mary that (someone) did the work.' The main verb 'hata' does not give a clear idea of what kind of speech act it contains. Thus we have to search for a clue for what kind of speech act the main verb might contain. The first hint may come from the complementizer. The complementi zer in this example is '-ko', and it is the one which can come with almost any kind of main verb. In this case, therefore, the complementizer may not give much help. The next possibili ty is the sentence type of the embedded clause. The embedded clause has an indicative ending '-ta'. This helps fix the interpretation of the main verb '—hata' because it eliminates verbs like 'mwutta (ask)', which requires a interrogative sentence type as the embedded clause, 'myenglyenghata (order)', which requires a directive sentence type as the embedded clause, or 'ceyanhata (suggest)', which requires a suggestive sentence type. (116) a.John-i Mary-eykey [0 ku il-ul hay-ess-nya]-ko mam_ _ _ _ _ _ _ _ _ _ _ nnm aa nom _ aitaa nvrA i/nx uac nvi. n .— n w v* uu— r a x— —wvnr rawul-ess-ta. ask—PST-IND 'John asked Mary if (she) did the work.' b. John—i Marv—evkey [0 ku il—ul ha—la]-ko -NOM —DAT* * the work-ACC do-DIR-COMP myenglyenghay—ess-ta. order-PST-lND 'John order Mary that (she) do the work.' 9 9 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. c. John-i Mary-eykey [0 ku il-ul ha-ca]-ko -NOM -DAT the work-ACC do-SUG-COMP ceyanhay-ess—ta. suggest-PST-IND 'John suggested to Mary that (they) do the work.' When '-ko' is the complementizer, the embedded clause takes a specific sentence type ending depending on the semantics of the main verb. Since the main verb in (116a) is 'mwutta (ask)', the embedded verb has the interrogative ending '-nya', and the main verb in (116b) is 'myenglyenghata (order)', the embedded verb has the 'directive' ending '-la'. Since the main verb in (116c) is 'ceyanhata (suggest)', the embedded verb has the suggestive ending '-ca'M . Thus the sentence type of the embedded clause helps reduce the possible options of the interpretation of the main verb 'hata (do)' in (115). The last hint is the subcategorization of the verb. There are two main arguments in (115), the subject and the dative object. The verbs which can replace 'hata' must be the ones which allow these two arguments in their subcategorization schema. Some of the possible ones are the following: (117) malhata (say), cwucanghata (claim), thongpohata (notify), kwangkohata (advertize) The next example also has 'hata' as the main verb. The complementizer is, however, not '-ko', but '~ki'. This choice of the complementizer shortens the list of the verbs which can replace 'hata'. For instance, the verbs which can have an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. embedded clause led by a '-tolok' complementizer may not replace 'hata' in this case. Secondly, there is a 'locative' marker '-lo' following the complementizer. (118) John-i Mary-eykey [0 ku il-ul ha]-ki-lo hay-ess-ta. -NOM -DAT the work-ACC do-COMP-LOC do-PST-IND *'John did to Mary to do the work.' This eliminates any verbs which need an accusative marker assigned to the embedded clause. (119) John-i [Mary-ka ku il-ul ha]-ki-lul wen-ess-ta. -NOM -NOM the work-ACC do-COMP-ACC want-PST-IND 'John wanted Mary to do the work.' Verbs like 'hwimanghata (hope)', 'kitayhata (expect)', 'wen- hata (want)' require the '—ki' complementizer—led embedded clause to be marked with an accusative case. The next thing to help eliminate the verbs which cannot replace 'hata' in (118) is again 'subcategorization' of the verb. There are several verbs which can take '-ki' complementizer-led embedded clause which is assigned the 'locative' marker. But some of them do net allow a dative object. (120) John-i [0 ku il-ul ha]~ki-io kyelshimhay-ess-ta. -NOM the work-ACC do-COMP-LOC decide-PST-IND 'John decided to do the work.' 'Kyelshimhata (decide)' does not subcategorize for a dative argument. Thus the next example with a dative argument is not grammatical. 223 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. *(121) John-i Mary-eykey [0 ku il-ul ha]-ki-lo -NOM -DAT the work-ACC do-COMP-Loc kyelshimhay-ess-ta. decide-PST-IND 'John decided to Mary to do the work.' All the above-mentioned factors considered, the verbs which can replace 'hata' in (118) are the ones like ’yaksokhata (promise)', 'mayngseyhata (swear)'. If the semantics of the verb is the only thing which is to be considered in deciding whether it is 'subject control' or 'object control', then the examples with 'hata (do)' we consi dered in the above may not be clear. Thus we propose that in Korean, and possibly in some other languages, the semantics of the verb may not be enough to decide whether it is 'subject control' or 'object control'. In the case of Korean at least the choice of the complementizer should also be considered as one of the deciding factors. This point may be made clear when we consider a verb like 'malhata (say/ tell)'. Many of the verbs which hrrgger erther 'subject controx* or 'object control' are the ones which have speech acts which are verbally done. For example, the 'subject control'-triggering verb 'yaksokhata (promise)', and the 'object control'- triggering verbs like 'myenglyenghata (order)', ’seltukhata (persuade)' contain speech acts which are verbally done in most situations. This is true with verbs like 'taytaphata (answer)', 'mwutta (ask)', 'ceyanhata (suggest)' since the acts expressed by these verbs are normally done verbally. The 224 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. shared property of all the speech acts conveyed by these verbs, thus, is that they can all be verbally achieved. The verb 'malhata (say/ tell)' has this shared property as its main speech act. Due to this property, this verb can appear in various structures. First it can appear with an indicative embedded clause. (122) John-i Mary-eykey [0 wul-ess-ta]-ko malhay-ess-ta. -NOM -DAT cry—PST—IND-COMP tell-PST-IND 'John told Mary that (someone) cried. / The embedded clause is the content of what John told Mary, the embedded clause being a statement sentence type. In Korean, this verb can also take an interrogative sentence as the embedded clause. (123) John-i Mary-eykey [0 wul-ess-nya]-ko malhay-ess-ta. -NOM -DAT cry-PST-OUE-COMP say-PST-IND *'John told Mary if (someone) cried.' The most natural verb in this case is 'mwutta (ask)' because the embedded clause is a question. Since the speech act of 'mwutta (ask)' can be verbally done, Korean grammar allows 'malhata (say/ tell) to take a question as the embedded clause. In the same way, this verb can take a suggestive embedded clause as in (124), and a command embedded clause as in (125). (124) John-i Mary-eykey [0 ca-ca]-ko malhay-ess—ta. -NOM -DAT sleep-SUG-COMP say-PST-IND 'John told Mary that (they) sleep.’ 225 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (125) John-i Mary-eykey [0 ca—la]-ko malhay-ess-ta. -NOM -DAT sleep-COM-COMP say-PST-IND ' John told Mary that (she) sleep.' As we have seen, 'malhata (say/ tell)' can have all of the four sentence types as the embedded clause. What is relevant for this section is that this verb can work as a 'subject control' verb as well as an 'object control' verb. (126) John-i Mary-eykey [0 ku chayk-ul ilk]-ki-lo -NOM -DAT the book-ACC read-COMP-Loc malhay-ess-ta. say-PST-IND 'John told Mary that (he (/someone)) read the book. (127) John-i Mary-eykey [0 ku chayk-ul ilk]-tolok -NOM -DAT * the work-ACC read-COMP malhay-ess-ta. say-PST-IND 'John told Mary to read the book.' The covert embedded subject in (126) is not always corefe rential with the main clause subject as we have seen, espe cially when pragmatics is at work. The salient interprtation, however, is that it is identical with the main clause subject. The same point can be made for the example (127) where the covert embedded subject is saliently coreferential with the main clause object. If it is the semantics of the verb which solely decides whether it is going to be 'subject control' or 'object control', the distinction between the examples (126) and (127) may be problematic because the same verb is used for both sentences. It may be hard to argue that the semantics of the verb 'malhata (say/ tell)' is different in these sentences 226 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. since the meaning of this verb in these examples is the same. This means that we have to seek some other explanation for why (126) is prominently 'subject control'—oriented, whereas (127) is mainly 'object'-oriented. The only difference between (126) and (127) is that two different complementizers are used. In (126), the complementizer '-ki' is used along with the locative marker '-lo', and the '-tolok' complementizer is used in (127). The combination of '-ki' complementizer and the locative marker '-io' is used for the mainly subject control- oriented verbs like 'yaksokhata (promise)', and 'mayngseyhata (swear)', and '-tolok' is used for verbs like 'myenglyenghata (order)', 'seltukhata (persuade)', 'kangyohata (force)', which are normally 'object control' verbs. This means that when the semantics of the verb cannot solely decide, the choice of the complementizer helps make the decision. The 'tolok' complementizer, however, is not exclusively used for 'object control'-oriented verbs. In subsection (A.i.) we dealt with a subject control-oriented verb 'nolyekhata (try)'. (128) John-i [0 ttenaj-lyeko nolyekhay-ess-ta. -NOM leave-COMP try-PST-IND 'John tried to leave.' The complementizer which is usually used for this verb is 'lyeko' as in (128), but 'tolok' can be also used as the complementizer (cf. N.K. Kim; 1978). 227 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (129) John-i [0 ku shihem-ev pwutl-tolok yelshimhi -NOM the exam-at pass-COMP hard nolyekhay—ess-ta. try-PST-IND 'John tried hard to pass the exam.' When the embedded subject is not overt as in (129), it is usually coreferential with the main clause subject. This is also due to the fact that 'nolyekhata (try)' does not subca— tegorize for a dative object. Even though the complementizer in (125) is 'tolok'', thus, the unexpressed subject is promi nently coreferential with the main subject. This shows that the choice of the complementizer may not always be a deciding factor for 'subject control' or 'object control'. In some cases like 'malhata (say/ tell)', the choice of the comple mentizer is crucial to determine whether it is 'subject con trol' or 'object control' as we saw, but in some cases, the semantics of the verb may play a crucial role as in the case of 'nolyekhata (try)', regardless of the complementizer used. (C.l.) Interplay of the semantics of the verb and the complementizers 'tolok' and 'lyeko' As we have seen in the previous subsection, the semantics of the verb cannot be the sole determining factor of 'subject control' or 'object control' in certain cases. In these cases, the selection of the complementizer is important to decide 228 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. whether it will be 'subject control' or 'object control'. In this subsection, we will claim that verbs like 'seltukhata (persuade)', 'myenglyenghata (order)', 'kangyohata (force)' do not always require the covert embedded subject to be corefe rential with the object. We claim that it depends on the intention of the causer of the event described by the embedded clause. (130) Apeci-ka atul^-ul [0i ku cip-ul sa]-tolok father—NOM son—ACC the house-ACC do—COMP sultukhay-ess-ta. force-PST-IND 'Father persuaded (his) son to buy the house.' One interpretation is that the event described by the embedded clause is the content of the persuasion made by the father. Say, the father verbally persuaded the son by saying the content of the embedded clause. The second interpretation is that the father did the action of persuading his son to bring about the event of (his son's) buying the house. Due to the semantics of the verb 'seltuk-hata (persuade)', these two interpretations may be salient. If the causer, however, has a goal of his doing something rather than letting the object do something, the situation may change. (131) Apeci.j-ka [0i ku cip-ul sa]-lyeko atul-ul father-NOM house-to buy-COMP son-ACC seltukhay-ess-ta. persuade-PST-IND 'The father persuaded the son to buy the house.' The same main verb 'sultukhata (persuade)' is used, but a different complementizer 'iyeko', which is predominantly 229 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'subject control'-oriented. The action of the father's persuading the son occurred to bring about the event of 'his buying the house.' The only difference between (130) and (131) is that the complementizer used in these two examples is different and this choice of the complementizer made the dis tinction that the agent of the event described by the embedded clause will most likely be the main clause object in the example (130), and the main clause subject in the example (131) . When the example (130) is given the first interpreta tion which is that the embedded clause is the content of per suasion, it may not be compared to the interpretation of (131) . Only when the interpretation that the persuasion occurred to bring about the event of the embedded clause, can the comparison be possible because the interpretation of (131) is that the persuasion occurred to bring about the event of the embedded clause. This means that the intention of the causer is also a crucial factor in deciding whether it is going to be 'subject control5 or 'object control5. This point can be reiterated with the examples in the following (133)- (134) which has the verb 'wihyephata (threaten)'. Comrie (1984) claims that "The felicity conditions for threats specify that the content of the threat must be a situation in which high agentivity is exercised by the giver of the threat, thus predicting subject control". Due to the fact that English does not allow an infinitival clause to come 2 3 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with this verb and an NP object, the test for this prediction in English is not possible. German, however, allows an infinitival clause to come with this verb. (132) (Comrie(1984)'s (8)) Ich drohe dir, dich alleine zu lassen. 'I threaten you that I will leave you alone.' (lit.) 'I threaten you to leave you alone.' Comrie's prediction is borne out in German, thus in (132) the subject controls the covert infinitival subject position. In Korean, this pragmatic felicity condition for threats may be blocked by the fact that the preference for either subject control or object control may be shown by the choice of the complementizer. For instance, 'lyeko' is predominantly 'subject-control'-oriented, and 'tolok' is predominantly ' ob j ect-control' —oriented15. (133) Ku salam.j-i [0i ku mwulken-ul phal]-lyeko Mary-lul that person-NOM that thing-ACC sel1-COMP -ACC wihyephay-ess-ta. threaten-PST-IND 'That person threatened Mary in order to sell the thing =' (134) ku salam-i [0i ku mwulken-ul phal]-tolok Mary^lul that person-NOM that thing—ACC sell-COMP -ACC wxhyephay—ess—ta. threaten-PST-IND 'That person threatened Mary to sell the thing.' The only distinction between these two sentences is the choice of the complementizer, 'lyeko' in (133), and 'tolok' in (134). The choice of the complementizer gives the difference between these two sentences, (133) being 'subject—oriented control', and (134) being 'object-oriented control'. The event of 231 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. threatening occurs to bring about another event which is the one described in the embedded clause, and the event of the embedded clause is meant to be done by the main clause subject, then by using the complementizer 'lyeko' it becomes a subject control structure. If the event of the embedded clause is meant to be carried out by the main clause object, then the complementizer 'tolok' may be used. Thus in Korean, we should say that the intention of the causer may be one of the crucial factors to decide whether the covert embedded subject position should be controlled by the subject or by the object. In the case of the German data in (133) with 'drohe (threaten)' as the main verb, the interpretation given is 'subject control' due to the 'felicity conditions for threats'. But in Korean 'subject control' and 'object control' are both possible depending on who the causer wants to be the agent of the desired event, which is described by the embedded clause. Thus we may say that in Korean the choice of comple mentizer plays an important role axuuy wilh tne p r a g m a t i c s and the semantics of the verb. (D) Change of the Controller (D.l.) The English facts In the case of object control in English, passivization of the main clause does not change who controls the understood subject of the infinitival clause. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (135) (Comrie(1986) (36)) Zoe persuaded Albert to leave. (136) (Comrie(1986) (37)) Albert was persuaded by Zoe to leave. (135) shows that the controller of the understood infinitival subject is the main clause object, and the surface subject in (136), which is the underlying object. Even though the seman tic role of the entity that controls the covert infinitival subject does not change when the main clause is passivized, the fact of the matter is that at least at the surface level, the subject in (136) controls the covert infinitival subject position. The common factor is that in both of (135) and (136), the controller is the main theme. This is a semantic factor, rather than a syntactic factor. But when the subject control verb 'promise7 is passivized, a restriction is added. (137) (Comrie(1986) (27)) Nora was promised to be allowed to leave. (138) (Comrie(1986) (28)) ♦Oswald was promised to leave. when the main clause is passivized, the infinitival clause should also be passivized, thus while (137) is grammatical, (138) is ungrammatical. The distinction between the passivization of the 7persuade'-type as in (136) and the passivization of the 'promise'-type as in (137) is that in the former case, the controller is semantically constant in both the active and passive sentences while in the latter case, the 233 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. patient/object in the active version can be the controller in the passive version. But they share the syntactic feature which is that the controller was an object in the active ver sion, while it is the subject in the passive version. Then why does the passivization of 'promise'-type require the passivization of the infinitival clause? Is it mandatory? Can pragmatics override the syntactic restraint? The answer may be partially positive. ?(139) John was promised by the president to get a raise within two months. This sentence may be marginal, but not totally unacceptable according to the English native speakers we asked. While the main clause is passivized, the infinitival clause is not. This example is interesting not only because this confirms that pragmatics helps ameliorate a syntactic constraint, but also there does not seem to be an active counterpart of the example (139) because when the main clause is active, the sentence is not grammatical with the intended interpretation. *(140) The president promised John to get a raise within two months. This sentence is not grammatical when the intended meaning is that "John will get a raise within two months". This sentence is possible only when the intended interpretation is that the president himself will get a raise. The next sentence also 234 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. shows that the embedded clause does not have to be passivized when the main clause is passivized. ?(141) The architect was promised by the construction company president to (be allowed to) design the building. It is the architect who will design the building. Thus the main clause subject controls the covert embedded subject position. The next set of examples also proves the same point. ?(142) The singer was promised by the show producer to sing in the show. The subject of the infinite in (142) is the main clause subject 'the singer', while the object 'the singer' in the active counterpart (143) cannot be the subject of the infinitive. *(143) The show producer promised the singer to sing in the show. This sentence is grammatical if the show producer himself will sing in the show. One way to salvage the claim that there is an active counterpart for (142) is to assume 'to be allowed to' is deleted. With 'to be allowed to' added, we get the following sentence. (144) The singer was promised by the show producer to be allowed to sing in the show. The same problem which an example like (137) faces, however, may arise because this means that we have to passivize the 235 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. infinitival clause in order to passivize the main clause. The only possible active counterpart for (144) is also the one whose main clause as well as infinitival clause are both active. If the infinitival clause remains passive, while the main clause is active, the sentence is marginal16. (145) (Comrie(1986) (29)) Penelope promised her son to be allowed to leave. According to Comrie's (1986) judgment, the salient interpre tation for this sentence is that 'Penelope' will be allowed to leave. The interpretation that 'her son' will be allowed to leave is, according to him, highly marginal. When the infini tive clause in (145) is active, then the sentence is readily acceptable. (146) Penelope promised her son to allow him to leave. One reason why 'her son' in (145) may not be the controller of the subject of the infinitive is that 'promise' in English is a 'subject' control verb, not an 'object' control verb. Since the controller of the first infinitive in (146) is the subject 'Penelope', the sentence is grammatical without any problem. The next passive example poses the same problem of either passivizing both of the main and infinitive clauses, or activizing both of them. o n r Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (147) The sick student was promised by the teacher to be taken to the hospital. If only the main clause is activized, it becomes ungrammatical with the intended meaning. *(148) The teacher promised the sick student to be taken to the hospital. (149) The teacher promised the sick student to take him to the hospital. It may be possible to have the interpretation that the teacher will be taken to the hospital, but this is rather odd due to the fact that in the given context, it is supposed to be 'the sick student' who is to be taken to the hospital, not 'the teacher'. We saw that in the case of 'object' control, passivizing the main clause results in 'subject' control, although semantically the controller remains the 'theme' argument. In the case of the 'promise'-type 'subject' control, when the main clause is passivized, the infinitival clause should also be passivized as argued by Comrie (1986). If the infinitival clause remains active, the intended meaning, which is the active version of the passive sentence, is not possible. When a pragmatically felicitous condition is given, however, it may be possible to have an active infinitival clause while the main clause being passive as the examples (139) and (142) show. One difference between the 'object' control and 237 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'promise'-type subject control in passivization is that in the former case, the controller remains constant semantically while it is no longer the object control, but subject control. In the case of the 'promise'—type subject control, the subject is the controller in both active sentence and passive sentence nhils semantically in the active version, the controller is the agent, and in the passive version, the controller is a non-agent. In the following subsection, we will look at how the Korean counterpart reverses the choice of the controller. (D.2.) The Korean facts In (D.l.) we looked at one possibility of changing the controller choice. When the causer himself intends to control the embedded subject position, with certain verbs the comple mentizer 'lyeko' is used, and when the causer intends the object to control the embedded subject position, 'tolok' is used. (150) John-i emma-lul [0 cenyek-ul ilccik meku]-lyeko -NOM mother-ACC dinner-ACC early eat-COMP seltukhay-ess-ta. persuade—PST—IND 'John persuaded mother so that (he) can eat dinner early.' (151) Sensayngnim-i ku haksayng-ul [0 ku chayk-ul ilk]-tolok teacher-NOM the student—ACC the book-ACC read-COMP seltukhay-ess-ta. persuade-PST-IND 'The teacher persuaded the student to read the book.' Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The reason why John persuaded mother in (150) is for him to eat dinner early, and in (151) the reason why the teacher persuaded the student is to make him read the book. Thus the complementizer choice, depending on the intention of the causer, is one way to change the controller. Another way to change the controller is to passivize the main clause as in English. There are three verbs 'tanghata (undergo (an adverse experience))', 'toyta (become)', and 'patta (receive)' which are attached to the so-called 'verbal noun' part to passivize the verbs which include 'hata (do)'. The 'subject' control verbs and the 'object' control verbs which we have dealt with so far all have the 'hata' ending. (152) John-i Mary-eykey [0 ku il-ul ha]—ki-lo -NOM -DAT the work-ACC do-COMP-LOC yaksokhay—ess-ta. promise-PST-IND 'John promised Mary to do the work.' (153) John-i Mary-eykey [0 ku il-ul ha]-tolok seltukhay- -NOM -DAT the work-ACC do-COMP persuade- ess—ta. PST—IHD 'John persuaded Mary to do the work.' Everything pragmatically being neutral, the most salient interpretation, for (152) is that the main clause subject 'John' will do the work, and the most salient interpretation for (153) is that the main clause object 'Mary' will do the work as we saw in the above sections. When the 'hata' ending in 'yaksokhata (promise)' in (152) is replaced by 'patta', however, the controller changes. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (154) John-i Mary-eykey [0i ku il-ul ha]—ki-lo -NOM -DAT the work-ACC do-CGMP-LOC vaksokpat-ass-ta. got promise-PST-IND 'John received a promise from Mary to do the work.' The person that made the promise to do the work in (154) is 'Mary', not the subject 'John', thus now it is strongly 'object' control—oriented. In the case of the verb 'sultukhata (persuade)', the 'hata' part can be replaced by 'patta (receive)', 'tanghata (undergo)', 'tovta (become)'. (155) John.j-i Mary-eykey [0i ku il-ul ha]-tolok seltukpat- -NOM -DAT the work-ACC do-COMP got persuade- ass-ta. PST—IND 'John got persuaded by Mary to do the work.' (156) John-i Mary-eykey [0i ku il-ul ha]-tolok seltuktoy- -NOM -DAT the work-ACC do-COMP be persuaded- ess-ta. PST—IND 'John was persuaded by Mary to do the work.' (157) John^—i Mary-eykey [0- ku il-ul ha]-toiok -NOM -DAT the work-ACC do-COMP seltuktanghay-ess-ta. got persuaded—PST—IND 'John got persuaded by Mary to do the work.' The main verb in (155) has 'patta'-ending, and the one in (156) the 'toyta'-ending, and the one in (157) the 'tanghata'- ending. These endings make the controller the subject 'John' in all these examples17. In (C.I.), we argued that the complementizer '-tolok' is strongly 'object'-control oriented. Even though the complementizer used in (155)-(157) is 'tolok', the empty embedded subject is not controlled by the object 240 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'Mary' any longer, but controlled by the subject. The combination of the complementizers 'ki ' and the locative marker 'lo' and 'tolok' closely interact with the semantic roles of the controllers. Even though the surface grammatical relations change in examples (154)-(157), the semantic roles do not change. Since the controllers in these examples remain constant, so do the complementizers. In the case of English, when the 'promise'-type subject control structure is passivized, the controller remains the subject, but semantically it is a non-agent, while in an active sentence, the controller is the agent as well as the subject. When the object control structure is passivized, the semantic role of the controller remains the same, being a non agent, although it becomes 'subject' control, rather than 'object' control. In Korean, it is rather consistent. When a subject control verb 'yaksokhata (promise)' is passivized, it is no longer a subject control, but becomes an object control, and it stays agent control= When an object control verb 'seltukhata (persuade)' is passivized, it becomes subject control, but semantically remains non-agent control. Thus the only difference between English and Korean is that in the case of the passivization of the 'promise'-type subject control structure, the English case remains 'subject' control whereas the Korean one changes to object control; while the semantic role of the controller changes to the non-agent in English, in +-he scent remams the controller. Thxs shows that m nuicau uuc 2 4 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Korean, in both subject and object control, for both active and passive sentences, it is not the surface grammatical roles, but the semantic roles which are more important in choosing the controller. In English, when it is subject control, the surface grammatical roles play a grater role than the semantic roles in choosing the controller, while in the case of object control, it is the semantic role which plays a grater role in determining the controller. (E) Comparison of the syntactic causative and object control In chapter 1, we defined 'causation' as the relationship of 'cause' and 'effect'. The 'cause' is the 'causing event', which is initiated by the causer and is described in the main clause. The 'effect' is the 'desired event', which is initiat ed by a causee different from the causer and is described in the embedded clause. In chapter 2, we argued that the syntac tic causative in Korean is biclausal at all levels of repre sentation. According to the definition of 'causation' we gave, the syntactic causative sentence involves 'causation' because the 'causing event', however opaque it is semantically, is described in the main clause, and the 'desired event' is the one depicted in the embedded clause. The causer and the causee are two different entities. 242 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The object control sentence we discussed in section (B) of this chapter fits this definition of 'causation'", too. The 'causing' event is described in the main clause subject and initiated by the causer, the main clause subject to bring about the 'desired' event described in the embedded subject which may be achieved by a causee different from the causer. There are, as a matter of fact, several similarities between these two phenomena. The first similarity between these two phenomena is that there is an optionally empty embedded subject position which may be filled by various overt elements. The potential candidates for this are strikingly similar. First, an NP or a pronoun which is coreferential with the main clause object may occupy this position in both phenomena. (158) John-i Mary-eykey [Mary--ka il-e na]-tolok hay-ess-ta. -NOM -DAT -NOM get up-COMP do-PST-IND kunye.j-ka she—NOM 'John made Mary get up.' (159) John-i Mary-eykey [Mary^-ka il-e na]-tolok -NOM -DAT -NOM get up-COMP kunye^ka she-NOM m n A M M 1 « • » W »»*•— a <■« <■« m uij w u ^ u a j r c o o • order-PST-IND 'John ordered Mary to get up.' In both of the sentences— the first one is a syntactic causa tive one, and the second an object control one— the empty em bedded position is occupied by an NP or a pronoun which is co- ref erential with the main clause object. It was mentioned in 243 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. section (B) that in Korean the empty embedded subject position can be filled with an overt element, unlike in languages with the constraint of having the position empty due to structural reasons. Although redundant, filling the embedded subject slot makes it clear that the agent of the desired event is equiva lent with the main clause object because there are other possibilities for the realization of the embedded subject. Another possibility for both phenomena is to have an embedded subject which includes the main clause object. (160) John-i Mary-eykey [Mary-wa Bill-~i hakkyo-ey ka]-tolok -NOM -DAT -and -NOM school-to go-COMP hay-ess-ta. do-PST-IND 'John brought it about to Mary that Mary and Bill go to school.' (161) John-i Mary-eykey [Mary-wa Bill--i hakkyo-ey ka]-tolok -NOM -DAT -and -NOM school-to go-COMP myenglyenghay-ess-ta. order-PST-IND 'John ordered Mary that Mary and Bill go to school.' V J J V% C* ^ ^ l | ^ ^ M ^ 1 A JU Vk A a « m 1* a ^ J A J a ■ . A —■ A 1- * A m w w m ui. u . ^ ^ c f t i i u u a uxuaj. wucu uuc cniucuucu dUUJCUU J .b not equivalent to the main clause object, but actually the main clause object is a part of the embedded subject. Thus the main clause object 'Mary' does not control the overt embedded subject in its true sense although there is a part—whole relationship between the main clause object and the embedded subject. This does not violate the definition of 'causation' mentioned earlier since the causer and the causee are not 244 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. same. The part-whole relationship between the main clause object and the embedded subject can be reversed, the former being the whole, and the latter the part. (162) John-i Mary-wa Bill^eykey [Mary-man^-i hakkyo-ey ka]- —NOM —and —DAT —only—NOM school—to go— tolok hay-ess-ta. COMP do-PST-IND 'John brought it about to Mary and Bill that only Mary go to school.' (163) John-i Mary-wa Bill^-eykey [Mary-man.-i hakkyo-ey ka] — -NOM -and -DAT —only-NOM school—to go tolok mvenglyenghay—ess-ta. COMP * order-PST-IND 'John ordered Mary and Bill that only Mary go to school.' In these examples the embedded subject 'Mary' is a part of the main clause object 'Mary-wa Bill (Mary and Bill)'. In the case of the examples (160)-(163), we can assume that the main clause object has a way to have the embedded subject carry out the event in the embedded clause. This semantic and pragmatic link between these allow the main clause object and the embedded subject to be different. When the embedded subject slot is vacant, and the pragmatic factor is not specially considered, it is saliently equivalent with the main clause object due to the semantics of the main clause verb in the case of 'myenglyenghata (order)' in (159), (161), (163) having 'tolok' as the complementizer in both cases. Some complemen tizers in Korean are used for some specific group of verbs. See J.D. Kim (1988) for this matter. There are only two possible candidates in the main clause to be the agent of the 245 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. embedded event. According to our definition of 'causation', the main clause subject is not a viable option to control the empty embedded subject position. Thus, everything being neutral, the main clause object is the only available candidate to be the controller. When there is some kind of semantic and pragmatic link between the mam clause object and the embedded subject, they may not even hold a part-whole relationship. (164) Sacang-i pwucang^eykey [ku ttokttokhan sawen^-i president-NOM manager-DAT the smart employee-NQM elyewun il-ul ha]-tolok hay-ess-ta. hard work-ACC do-COMP do-PST-IND 'The president brought it about to the manager that the smart employee does the difficult work.' (165) Sacang-i pwucang^eykey [ku ttokttokhan sawen^-i president-NOM manager-DAT the smart employee-NOM elyewun il-ul ha]-tolok myenglyenghay-ess-ta. hard work-ACC do-COMP order-PST-IND 'The president ordered the manager that the smart employee does the difficult work.' The main clause object and the embedded subject are not equivalent, and do not have a part—whole relationship. The main clause object, however, has authority (/ability of persuasion) over the embedded subject, and can make the embedded subject perform the event in the embedded clause. When the president does something to the manager in (164)— for instance, order, ask, tell, or instruct— to make the event in the embedded clause happen, the manager dees something to the actual agent of the event, the embedded subject to make him (/her) perform the event. When there does not exist this 246 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. kind of link, it is rather hard for some people to accept the sentence as natural. Suppose the main clause object in (164)= (165) is 'Mary', and the embedded subject 'Tom'. Unless some kind of context given in which '"Mary' is in a position of being able to manipulate 'Tom' to carry out the desired event, some may say it is not natural^. As shown in section (B), when there is the right context, even an empty embedded subject does not have to be coreferential with the main clause object. (166) Coach-ka cwucang^eykey [Oj kong-ul han saram-ey han Coach-NOM team captain-DAT ball-ACC one person-per one key-sshik kat]-tolok hay-ess-ta. CL-each have-COMP do-PST-IND 'The coach brought it about to the team captain that (the players) have one ball per person.' (167) Coach-ka cwucangi=eykey [Oj kong-ul han saram-ey han coach-NOM team captain-DAT ball-ACC one person-per one key-sshik kat]—tolok myenglyenghay—ess—ta. CL-each have-COMP order—PST—IND 'The coach ordered the team captain that (the players) have one ball per person.' The main clause object is singular whereas the empty embedded sub]6ct must bs piursl ggcgusg th? phrsss ' nun ssr2si“sy iisn key-sshik (one ball per person)' implies that there is more than one person getting the bails. 'Senswutul (players)' can be optionally placed in the embedded subject position to make it clear who gets the balls. The main clause object 'cwucang (team captain)' is in a position to tell the players what to do. As long as this kind of context can be kept in mind, even having a name like 'Mary' in the main clause object position in (164)—(165) will be acceptable. 'Mary' in this case has 247 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. some close link with the embedded subject to influence the embedded subject to perform the embedded event. The covert or overt embedded subject, however, cannot be coreferential with the main clause subject. As a part of the definition of 'causation' defined for this thesis says, the causer and the causee are different. The examples in (164)-(167) show that the causee which is the main clause object is completely different from the actual agent of the desired event. This means that the main clause subject is the causer and the main clause object is the causee of this causer. But this causee is the causer of the causee which is the agent of the desired event because it is not the first causee who performs the desired event, but the embedded subject, whether overt or covert, is the one who actually does that. The relationship among the main clause subject and the main clause object and the embedded subject is the following: (168) Main clause subject Main clause Object Embedded Subject 1st causer 1st causee 2nd causer 2nd causee The main clause subject, therefore, is not the immediate causer of the 'desired event' in the embedded clause although it is the causer that initiates the event which ultimately triggers the 'desired event'. As we have seen in the above, the candidates to fill the embedded subject position in both the syntactic causative and object control are practically the same, showing there are syntactic and semantic similarities here. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Another thing these two structures share is the choice of complementizer. In Korean, there are two generic complemen tizers 'ko' and 'ket' which are rather commonly used. The second one is sometimes considered to be a nominalizer. The usage of other complementizers is strongly connected with different types of verbs. The complementizer 'tolok' which is used for the syntactic causative and object control is mainly used to mark a complement of manipulative verbs. Another function of the complementizer ’tclck7 is to show that the embedded clause led by it is the 'effect' or the 'result' caused by the event described in the main clause. The comple mentizer 'tolok' is mostly used when there are distinct causer and causee. When the causer and causee are equivalent, meaning the agent of the 'causing' event and the 'desired event' are same, the complementizer 'lyeko' is used19. Another similarity between these two phenomena of the syntactic causative and object control is that it may be the ______ — J-l J_ J_ 1 -----1 3 J 3 - 1 M k A J . U A ta a c tu a u u iic ciuucuucu i* a . an sc; xu uu w j. u u c u i majr uwu */ « • u complement of the main verb. In the case of the syntactic causative, the direct object of the verb 'hata' is not overtly present, thus it is not clear what it may be. The following sentence has 'muetinka (something)' as the direct object of the main verb 'hata (do)' (169) John-i Mary-eykey [ ku il-ul ha]-tolok [mwuetinka]-lul -NOM -DAT the work-ACC do-COMP something-ACC hay-ess—ta. do-PST-IND 'John did something to Mary so that (she) will do the work.' 249 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. If the embedded clause in this sentence were an argument of the main verb, then the only possibility is that it is the direct object since the indirect object position is occupied by dative 'Mary'. An NP 'mwuetinka (something)', however, occupies the object position slot preventing the embedded clause from taking the position. The only possibility is that the embedded clause is an adverbial clause. In the case of the object control sentence, the embedded clause can be marked by 'tolok' or 'kes'. (170) Sensayngnim-i haksayng-eykey [0 ku il-ul ha]-tolok teacher-NOM student-DAT the work-ACC do-COMP cishihay-ess-ta. instruct-PST-IND 'The teacher instructed the student to do the work.' (171) Sensayngnim-i haksayng-eykey [0 ku il-ul ha—l]-kes-ul teacher-NOM student-DAT the work-ACC do-FUT-COMP-ACC cishihay-ess-ta. instruct-PST-IND 'The teacher instructed the student (her) doing the work.' In the case of (171), it is obvious that the embedded clause is the content of 'the teacher's order' while it is not so clear in case of (170) . As we have been claiming, the embedded clause in (170) is the desired event or the effect which may be realized as the result of the event in the main clause. Thus it is possible that the content of 'the teacher's order' may not be equivalent with the content of the embedded clause. This possibility may be shown by the next example. OKA Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (172) Sensayngnim-i haksayng-eykey [0 ku il-ul ha]—tolok teacher-NOM student-DAT the work-ACC do-COMP kulen cishi-lui nay—ess—ta. such instruction—ACC do-PST—IND 'The teacher gave such an instruction to the student so that (he) will do the work.' In this example,, the direct object of the verb 'hata(do) ' is 'kulen cisbi (such an instruction)', not the embedded clause, and the content of the instruction is not the same as the content of the embedded clause. The reason why the teacher gave 'kulen cishi (such an instruction)' to the student is to make him (/her) carry out the desired event, which is 'doing the work'. If the embedded clause in (172) is marked by 'kes', the sentence becomes ungrammatical. (173) *Sensayngnim-i haksayng-eykey [0 ku il-ul ha-l]-kes-ul —COMP—ACC kulen cishi-lul hay-ess-ta. The ungrammaticality of this sentence arises because when the C l S U S S ■ » c f C b j S C t t. tH i rj’ h i c f h o Q ^ the instruction, there is another candidate for the direct object, 'kulen cishi (such an instruction)'. If the embedded clause led by 'tolok' in (172) were to be the direct object of the verb, the sentence (172) should be as ungrammatical as (173). The fact that the sentence (172) is grammatical convinces us to think that the embedded clause is an adverbial clause, not an argument of the verb. We claim, thus, in this study that the 'tolok'—led embedded clause in the syntactic 251 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. causative and object control clause are adverbial clauses, rather than arguments ox the main verb. As we have seen, the syntactic causative and object control share equivalent syntactic structures while semantically both containing 'causation', the main clause representing the 'causing' event and the embedded clause the 'desired' event, and the causer and the causee being different. We thus con clude that these phenomena share the same semantic properties involving 'causation' as well as the same syntactic structure, proving that for them there is a match between syntax and semantics in the kind of 'causation' defined for this study. In the next chapter, we will briefly deal with the issue of match/mismatch of syntax and semantics of 'causation' in English, Japanese, Korean, and Thai. 252 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 5. Syntax and semantics of 'cause' and 'effect' In attempting to define the causative construction, Shibatani (1976) characterizes the causative situation as the following: (1) a. The relation between the two events is such that the speaker believes that the occurrence of one event, the "caused event," has been realized at t2, which is after tl, the time of the "causing event." b. The relation between the causing and the caused event is such that the speaker believes that the occurrence of the caused event is wholly dependent on the occurrence of the causing event; the depen dency of the two events here must be to the extent that it allows the speaker to entertain a counter- factual inference that the caused event would not have taken place at that particular time if the causing event had not taken place, provided that all else had remained the same. With this characterization, sentences like 'I told John to go', and ' 1 know that John went', acuuiuiny uu &moatanx, cannot be counted as causative sentences. The first sentence does not entail that 'John really went', and In the second sentence, the event of 'John's going' may have happened, but it did not depend on 'my knowing it'. According to the above characterization, Shibatani says that sentences like 'I caused John to go' and 'I made John go', as well as 'I opened the door' and 'I sent John to the drugstore1' are all causative sentences. The reasons that 253 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. these sentences are counted to be causative ones are that the speaker can be certain that the caused event has happened, and s/he can believe that the caused event would not have happened, if s/he did not do something to cause the caused event. Shibatani claims that if the speaker knew that the caused event might have happened regardless of whether s/he did something to cause it, then these expressions could not be used by the speaker. In reality, the above characterizations of the causative situation will exclude the Korean syntactic causative construction we dealt with in Chapter 2. (2) John-i Mary-eykey [0 ku chayk-ul ilk]-key2 hay-ess-ta. -NOM -DAT the book-ACC read-COMP do-PST-IND 'John made Mary read the book.' What this sentence tells us is that 'John' did something to Mary so that she would read the book. This sentence does not say, though, that 'Mary actually read the book'. It does not tell us that 'Mary's reading the book' could only happen because of 'John'. It simply tells us that in the described situation, the causing event created by John is closely related to the desired event. It is likely that 'Mary read the book'. We cannot exclude, however, the possibility that 'Mary did not read the book'. Thus either of the following sentences could follow the sentence (2). (3) a. Kulayse Mary-ka ku chayk-ul ilk-ess-ta. so -NOM the book-ACC read-PST-IND 'So Mary read the book.' 2 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. Kulena Mary-nun ku chayk-ul ilk-ci ani-ha-ko, but -TOP the book-ACC read-CI NEG-do-CQN talun chayk-ul ilk-ess-ta. another book-ACC read-PST-IND 'But Mary did not read the book, and read another book.' If we take Shibatani's characterization of the causative situation, the sentence (2) cannot be considered a causative one because of the possibility that the desired event is not realized. However, sentence (2) has been treated as a causa tive one in the literature, and even Shibatani himself uses the Korean examples with 'hata (do)' as the causative verb as causative sentences. His examples (28b), (33), (43), (62), and (66) are just like the sentence (2) with 'hata' as the causa tive verb. An alternative way which allows us to include sentences like (2) as causative sentences is to modify Shibatani's characterization of the causative situation. (4) The causing event happened in order to bring about the desired event. With this modification, we can include the sentence (2) as a causative sentence. Whether the desired event of 'Mary's reading the book' happened or not, the causing event occurred in order to achieve this goal. We only deal with 'causation' where the causer and the causee are different. We change the definition (4) as following: (4') A initiated A' event (=causing event) to make B carry out B' event (=desired event). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This modification of the characterization of the causative situation expands the scope of the notion of causativity. For instance, now the sentence 'John told Mary to go home' has to be also considered as a causative sentence because the event 'John's telling Mary' occurred in order to bring about the event of 'Mary's going home'. In this chapter, we want to look at the "cause and effect" relation between the causing event and the desired event regardless of whether the latter event occurred or not as long as the causing event was intended to bring about the desired event. In the typical sense of "cause and effect" relation, Shibatani's characterization of the causative situation intended to capture the cases where 'effect' is actually realized, and our modified version is to capture the general, and broader "cause and effect" relation, where the 'effect' does not need to be realized. We will show that the causing event could be opaque as in the case of the verb 'hata (do)', 'kangyohata (force)', 'pwuthakhata (request)', or transparent in the sense that the causing event is clearly expressed in a concrete manner. An example of the latter case will be 'John bought the book so that Mary could read it'. In order to bring about the desired event, 'Mary's reading the book', the causing event, 'John's buying the book' occurred. In this case, the content of the causing event is transparent, or clear because the event is well expressed by the main clause. Whether the causing event 256 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is transparent or not, the shared characteristic which is common to all of these structures is that there is a desired result or effect to be accomplished for which the causing event occurs. We will argue, following the modified definition of causative as in (4), that so-called object control is a causative structure, and the 'so that' clause structure is also a causative structure. We will argue that in some languages, for instance Korean, the syntax of the causative structure is consistent, using either 'key' or 'tolok' as the complementizer, whether the causing event is transparent or not, while in some other languages like English, there is a syntax and semantics mismatch. In a language like Japanese, the syntactic causative utilizes the 'affixed' causative morpheme 'sase', but the object control construction and the Japanese version of 'so that' clause construction share the same syntactic structure which uses the complementizer 'yoonif (A) Semantics of cause and effect In Chapter 3, we dealt with the syntactic causative con struction in Korean, and argued that no matter which (case) marker the causee is in, the syntactic causative construction is biciausai. When we talk about causation, we have to mention 'cause' and 'effect'. The 'effect' which is meant to be brought about by the 'cause' is the event described by the 257 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. embedded clause. The 'effect' or the desired event is usually transparent, thus it is rather easy to understand what the event is. Take the following sentence, for instance. (5) John-i Mary-eykey [0 ku chayk-ul ilk]-tolok hay-ess-ta. -NOM -DAT the book-ACC read-COMP do-PST-IND 'John made Mary read the book.' The desired event or the 'effect' aimed to be triggered is 'Mary's reading the book'. John did something verbally, physically, in action, or in writing in order to bring about this desired event. As we mentioned in the above, it is not guaranteed that the desired event is actually realized. Whereas the desired event, that is the 'effect', is trans parent, the causing event is not. It is certain that John did something, but the main verb 'hata (do)' which encodes the causing event is the most general action verb, and does not give any clear idea of what John did. This works for the English causative verbs like 'make', 'have', 'get', and 'cause'. These verbs clearly show that there is causation, but do not reveal any clue about what the causer did to bring about the desired event. In the case of the sentence (5), John did something to Mary so that Mary could read the book. If the causee is nominative- marked as in (5'), then the interpretation is that John did something which was intended for Mary to read the book. (5') John-i [Mary-ka ku chayk-ul ilk]-tolok hay-ess-ta. -NOM -NOM the book-ACC read-COMP do-PST-IND 'John did something so that Mary read the book.' 258 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This does not say anything about whether John did something directly to Mary, unlike the sentence (5) which suggests that John did something directly to Mary to bring about the desired event. When the causee is nominative-marked, the closest English counterpart would be the following: (6) John did (something) so that Mary could read the book. This sentence shows that whatever John did is the direct cause of the desired event of 'Mary's reading the book'. Thus the causative verbs like 'hata (do)' do not clearly show what the causer did to bring about the desired event. The verbs which are commonly called object control verbs also contain causing events in them in the sense that the event described by a so-called object control verb or manipulative verb occurs to bring about the desired event described in the embedded clause. Take the verb 'order', for instance. / n \ m u i< a a m U m w *** 3 J t f . i. ^ .£ * ^ * _ I _ _ - _ » l- - - - - - - - - - - \ • / v ^ u c i c u n a i jr l u x J L U x a i i ( . l i e p i U J C C t J j y Friday. The event of 'the teacher's giving an order to Mary' occurred to bring about the event of 'Mary's finishing the project by Friday'. Actually it may not be the case that in this English sentence the embedded clause may be the content of the order given by the teacher3. In the following Korean equivalent with 'tolok' as the complementizer, there is no ambiguity of that sort. ICO _ / _ S Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (8) Sensayngnim-i Mary-eykey [0 ku il-ul kumyoil-kkaci teacher-NOM -DAT the work-ACC Friday-until kkutnay]-tolok myenglyenghay—ess-ta. finish-COMP order-PST-IND 'The teacher ordered Mary to finish the work by Friday. With using the complementizer 'tolok', the intended meaning is that the teacher gave an order to Mary to bring about the event of Mary's finishing the work by Friday'. If one wants to make sure that the embedded clause is the content of the order, the complementizer ' kes' may be used in place of 'tolok'. (9) Sensayngnim-i Mary-eykey [0 ku il-ul hal kes]-ul teacher-NOM -DAT the work-ACC do COMP-ACC myenglyenghay-ess-ta. order-PST-IND 'The teacher ordered Mary to do the work.' When an embedded clause is marked by 'kes', it becomes an argument of the verb. The sentence (9) does not mean that the teacher's giving an order to Mary was made to bring about the desired event, 'Mary's doing the work', instead the embedded clause is the content of the teacher's order. A literal F.ngiish translation for (9), which is not grammatical in English, thus, would be the following: (10) The teacher ordered to Mary her doing the work. As in the case of the syntactic causative verb 'hata (do)', the causee in (8) does not need to be in dative, and it may be marked nominative. n n r \ W Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (11) Sensayngnim-i [Mary-ka ku il-ul kumyoil-kkaci teacher-NOM —NKOM the work-ACC Friday-until kkutnay]—tolok myengiyengnay-ess-ta. finish—COMP order-PST~IND 'The teacher ordered that Mary finish the work by Friday.' The teacher gave an order (to someone), and this order was intended to bring about the desired event of 'Mary's finishing the work by Friday'. There is no guarantee that Mary finished the work by Friday. Thus the event described in the embedded clause could have been achieved or not achieved. The important point is that the order was made to bring about the desired event. Manipulative verbs such as 'myenglyenghata (order)' behave like the syntactic causative verb 'hata (do)' in the sense that in both cases the realization of the desired event is not guaranteed. The causing event for (11) is the teacher's giving an order. It is not clear, however, whether the order is made verbally or in writing or through a third party. This opaque ness of the causing event can be seen more clearly with a verb like 'kangyohata (force)'. (12) John-i [Mary-ka ku il-ul ha]-tolok kangyohay-ess-ta. -NGM -MOM the work-ACC do-COMP force-PST-IND 'John forced Mary to do the work.' John did something forceful in order to bring about Mary's eating the food. There are many ways to force someone to do something. One can do it by withholding salary, by imprison- 261 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ing the person, or by phoning many times. It is not clear what the exact causing event may be, which is expressed by the verb 'kangyohata (force)'. Manipulative verbs like 'order', and 'force' do not clearly show what the exact causing event may be, except that they show there is a manipulative situation in which the causer did something to bring about the desired event. The causing event does not necessarily need to be manipula tive. As long as the causing event occurred, manipulative or not, for the purpose of bringing about the desired event, it still semantically includes the relation of 'cause' and 'effect'. Take 'ttaylita (beat)' for example. (13) Mary-ka Bill-ul [0 ku il-ul ha]-tolok ttayli-ess-ta. -NOM -ACC the work-ACC do-COMP beat-PST-IND 'Mary beat Bill so that he do the work.' The reason why Mary beat Bill is to make him do the work. Bill may or may not do the work as the result of Mary's beating him, but Mary's intention of beating Bill was to make him do the work. 'Ttaylita (beat)' is not traditionally called a manipulative verb in the literature, but the situation in this case is that the beating event occurred to bring about the desired event of 'Bill's doing the work'. In this sense, from the extended point of view on manipulative verbs, 'ttaylita' is semantically a manipulative verb. The causing event is expressed transparently by the main verb, thus it is easy to see what the causing event is. When the main verb is the 262 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. syntactic causative verb 'hata (do)', the causing event is least clear. The causer did something (to the causee) to bring about the desired situation. The hearer, in this case, cannot imagine what the causing event may have been because the verb 'hata' does not give any clue about the event. It could be anything. In the case of the so-called manipulative verbs, the causing event may not be clearly expressed, either, but there is some kind of 'speech act' denoted by the main verb. In the case of 'ttaylita (beat)', there is no vagueness of what the causing event is since the event described by the main verb is a clear event of 'one person beating another'. This clarity may be seen in the next example, too. (14) Mary-ka [Bill-i ttena-ci ani ha]—tolok wul-ess-ta. -NOM -NOM leave-CI NEG do-COMP cry-PST-IND 'Mary cried so that Bill will not leave (her).' The causing event is 'Mary's crying', and this event occurred to bring about the desired event of 'Bill's not leaving'. The causing event is obvious, thus there is no possibility of ambiguity or misunderstanding. It is not guaranteed, though, that the desxred event actually occurreu. xne resuxt couxd be that Bill might or might not leave as the result of Mary's crying. Although 'wulta (cry)' is not qualified to be a 'manipulative verb' in the traditional sense, the event of 'Mary's crying' certainly contains a manipulative situation in the semantic sense. 'Mary's crying' occurred to manipulate an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. event, 'Bill's not leaving', to happen. The next example illustrates a lesser degree of manipulation in the causing event than 'crying'. (15) Mary-ka [Bill-i umak-ul tut]-tolok pang-eyse -NOM -NOM music-ACC listen-COMP room-from nawa-ass-ta. C ui i i c uui."JTOX— iW Lf 'Mary came out of the room so that Bill can listen to music.' The causing event in this case is 'Mary's coming out of the room'. This event itself is non-manipulative. It occurred, however, in order to bring about the desired event of 'Bill's listening to the music'. Mary's staying in the room may have been some kind of hindrance to Bill's listening to the music. Thus Mary left the room for the purpose of facilitating Bill's listening to the music. The next example shows again that the causing event does not have much of manipulation, if not none, in itself. (16) Mary-ka [Bill-i chayk-ul ilk]-tolok ca-ass-ta. -NOM -NOM book-ACC read-COMP sleep-PST-IND 'Mary slept so that Bill (could) read the book.' 'Cata (sleep)' does not contain any sense of manipulation as a speech act. But still it provides the cause of 'Mary's sleeping' which occurred in order to bring about the effect of 'Bill's reading the book'. Semantically speaking, the above data show that the causing event does not necessarily need to be a manipulative one like 264 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 'order' or 'force'. As long as the causing event occurred to bring about the desired event, that satisfies the relation of 'cause' and 'effect'. It may be said, thus, that in Korean, not only the syntactic causative construction and the so- called object control construction with a manipulative verb as the main verb involve semantic causation, but so also does any construction in which a causing event described by the main clause occurs to bring about the desired event described by the embedded clause. (B) Syntax of 'cause' and 'effect' In the above, we modified Shibatani's characteristics of the causative situation in order to make them workable for the Korean causative construction. This led us to include the so- called object control construction as well as the Korean version of the English 'so that' clause construction as causative construction in the semantic sense that in all of these constructions there is the 'cause', that is the causing event which is manifested in the main clause, and there is the 'effect', that is the desired event which is manifested in the embedded clause. In order to investigate how the syntax of 'causation' defined for this study is manifested in some languages, we choose four languages; Korean, English, Japanese, and Thai. Since we broadened the semantic base of causation in the 265 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. previous subsection, we will deal with the structures which semantically qualify as causative structures. (B.l) Korean In chapter 3, we argued that the syntactic causative in Korean is biclausal no matter which marker is used with the causee. In chapter 4, we dealt with object control as well as subject control, and we identified the object control structure as biclausal, too. We argued in subsection (4.1) that we should broaden the semantic base of causation. This allowed us to include so-called object control, and the Korean equivalent of the 'so that' English structure in the scope of semantic causation. The one thing all these constructions share is that they all utilize the complementizer 'key' or 'tolok'. The basic function of these complementizers is to denote that the embedded clause represents the event which may be brought about by the causing event expressed by the main clause. First let us look at a syntactic causative sentence. (17) John-i Mary-eykey [0 ku nolay-lul pwulu]-tolok -NOM -BAT the song-ACC sing-COMP hay-ess-ta. do-PST-IND 'John made Mary sing the song.' The complementizer 'tolok' shows that the embedded clause is the desired event brought upon by the causing event which is 266 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. not clearly shown in the main event. Thus the causing event is 'John's doing something' which is done in order to bring about the desired event depicted in the embedded clause which is 'Mary's singing the song'. All the Korean examples we dealt with in section (A) have 'tolok' as the complementizer. This shows that in Korean, the semantics of causation is reflected in the syntactic structure. Thus when a sentence contains semantic causation, regardless of whether the causing event is clearly encoded by the main clause (mainly by the event described by the main verb) or not, the desired event, that is the 'effect', is encoded in the embedded clause led by the complementizer 'tolok' or 'key'. We can conclude that in Korean when a sentence is semantically causative, then its syntactic structure directly reflects that by the choice of the complementizer, which is used to show that the embedded clause is the effect accom— plishsd csusi«5 svsnt dsscribsd the n i u c l u u s e Xi* Korean, thus, we can say that in causation there is a match between semantics and syntax. (B.2) English The syntactic causative sentence with 'have' or 'make' as the main verb allows a bare-infinitive clause as the embedded clause, whereas a causative sentence with 'cause' as the main verb allows a to-infinitival clause. It is not our concern 267 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. here to find out why the bare-infinitive clause is used for one and the to—infinitival clause for the other. Our interest is to find out if there is a syntax and semantics match or mismatch in the causative construction. All of the verbs which are dubbed order/permit type (object control) in Sag and Pollard (1991) permit no bare-infinitive clause, but only to-infinitival clauses4. The to-infinitival clause is not, however, the only option for the embedded clause. Many of the verbs listed in this category can have an embedded clause which is led by ' so that'. The clause led by 'so that' is the 'effect' which may occur due to the 'cause' which is described in the main clause. Thus most of the verbs listed as 'order/permit type' in Sag and Pollard (1S91) have two options for the embedded clause: to-infinitival clause and clause led by 'so that'. Thus there are three types of clause, which may be used depending on which verb is used as the main verb. In section (A), we expanded the notion of 'causation', and the sentences with 'order/permit type' verbs may be called causative ones since there is a 'cause' and 'effect' relation ship between the event described in the main clause, and the event described in the embedded clause. In the case of English, thus, three types of sentence structure are used for the semantic causatives: to-infinitive, bare-infinitive, and 'so that' as the embedded clause. The last structure is used in an example like the following one: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (18) John left the room in order that Mary might sleep. 'Leave' is not a manipulative verb, or 'order/permit type'. But the event of 'John's leaving the room' specifically occurred in order to make the event of 'Mary's sleeping' happen. Thus the former event is the 'cause', and the latter, the 'effect'. Thus this sentence is semantically a causative sentence. The fact that this sentence involves 'causation' is not due to the choice of the main verb, but to the semantic relationship between the causing event and the desired event. This relationship was clearly shown by the usage of 'so that' which is used to show the 'cause' and 'effect' relation between the event described by the main clause, and the event described by the embedded clause. This shows that it is not always the case that the main verb alone is the sole factor in making a sentence a causative one, but it rather depends on whether there is a 'cause' and 'effect' relation between the two events, regardless of whether the causing event is clearly expressed or not. In this sentence, the to—infinitival or bare-infinitive clause is not an option to be the embedded clause. In conclusion, we argue that in English three different structures are used to express 'causation': bare-infinitive, to—infinitive, and 'in order that' structure. Sentences with verbs like 'compel', 'encourage', and 'urge' may choose an embedded clause led by either 'to-infinitive' or 'in order that'. Some sentences, with verbs like 'have' or 'make', only 269 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. allow bare-infinitive, and some only 'in order that' struc ture. We claim, thus, that in English there is no one-to-one match between syntax and semantics in the case of 'causation'. (B.3) Thai According to Vichit—Vadakan (1976), there are three types of periphrastic causatives in Thai. (19) (his (la-c)) a. Saakhaa tham kracok taak. Saka cause mirror break 'Saka caused the mirror to break.' b. Saakhaa hay dek wing. Saka have child run 'Saka had the child run.' c. Saakhaa tham hay kaw?ii lom. Saka cause ? chair fall 'Saka caused the chair to fall.' He claims that these three periphrastic causatives are diffsrsnt with zTSSpsct to ths rsl^tionship between the Causing situation and the caused situation. He says that (19a) expresses inadvertent cause, (19b) expresses intentional cause, and (19c) is neutral as far as 'intent' is concerned. He does not identify what 'hay' in (19c) is, and leaves it undefined. 'Hay' literally means 'give', thus can be used as a main verb as in (20) as well as dative and benefactive marker as in (21). 270 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (20) (his (3a)) Khaw hay khoongkhwan nisaa. he give present Nisa 'He gave Nisa a present.' (21) (his (3b)) Khaw sang cotmaay hay chan, he send letter give I 'He sent a/the letter to me.' or 'He sent a/the letter for me.' It is not clear whether 'hay' in (21) stopped functioning as a verb or not. 'Hay' also functions as a complementizer and as a connective, according to him, linking a purpose clause to a sentence. (22) (his 3c) Saakhaa sang hay nisaa khian cotmaay. Saka order have Nisa write letter 'Saka ordered Nisa to write a letter.' (23) (his 3d) Saakhaa khian cotmaay (phia) hay nisaa toop. Saka write letter so that COMP Nisa answer 'Saka wrote a/the letter so that Nisa would answer.' 'Hay' in (22) is used as a complementizer when the verb is a manipulative verb like 'sang (order)'. In his footnote (p461), Vichit—Vadakan says that 'hay' in (22) has a similar usage with 'hay' in one of the periphrastic causatives as in (19c). He says that 'hay' in (23) is used as a connective, which he terms a complementizer in (23). The next sentence also shows the same function of 'hay'. (24) John ook pay cak hong hay Mary aan nangsii. exit go from room COMP read book 'John left the room so that Mary can read the book.' 271 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. According to our definition, sentences like (22)-(24) are semantically all causatives because the event described by the main clause is the causing event, and the event described by the embedded clause is the desired event. Thus it is not odd to see that the complementizer 'hay' is used in the periphra stic causative sentence (19c), in the sentence (22) with a manipulative verb as the main verb, and in the equivalent of the English 'so that' clause structure. In all of these structures, we can safely say that the desired event is marked by the complementizer 'hay'. In a periphrastic causative sentence (19c), Vichit—Vadakan did not assign any term for 'hay', and left a question mark, not knowing what to call it. In our analysis, it has the same function which 'nay' in (22)- (24) has: marking the desired event. We may say, thus, that in Thai as in Korean, the same complementizer marks the desired event regardless of whether it is periphrastic causative, a structure with a manipulative verb, or a structure equivalent to cne auvjixsn sentence using 'in order that'. The only difference between Korean and Thai is that while in Korean there is no periphrastic causative which does not contain the complementizer either 'key' or 'tolok', in Thai there is a periphrastic causative as in (19a) which does not contain the complementizer 'hay' marking the desired event5. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (B.4) Japanese In Japanese, a syntactic causative sentence usually uses a morphological process. The causative morpheme 'sase' is added to the stem of the verb to create a causative sentence. /*)C\ /ntiKi'ncl/rr / * 1 O O HY /£1U \ \ ^ j y y * / ^ \ / f y VJLt> / f Taroo ga Jiroo ni Hanako ni eigo o oshiesaseta. NOM DAT DAT English ACC teach. CAU. PRF 'Taro made/ let Jiroo teach English to Hanako. The fact that this sentence is a causative one shows that there is the relationship of 'cause' and 'effect' between whatever Taroo did and whatever Jiroo did (or may do) . What 'Taroo' did is not clear in (25), but what Jiroo did (or may do) is clear. Thus the desired event is 'Jiroo's teaching Hanako English'. Although what 'Taroo' did is not very clear, it is certain that what Taroo did was intended to bring about the desired event. When the main verb is a manipulative verb, the complementxzer 'yooni' is used to denote the desired event. (26) Sense! ga Jiroo ni [sono hon o yomu) yooni meireishita. teacher NOM DAT that book ACC read COMP ordered The teacher's giving an order to Jiroo is the causing event which occurred to make the desired event 'Jiroo's reading the book' happen. Is 'yooni' used to mark the desired event ( = effect) in a structure of 'cause' and 'effect'? The answer 273 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. may be affirmative. The main verb 'sikaru (scold)' in the next sentence is not a manipulative verb in the traditional sense. (27) Sensei wa [sitsumon o kotaeru] yooni Taroo o teacher TOP question ACC answer COMP ACC sikarimashita. scolded 'The teacher scolded Taroo so that (he) will answer the question.' The event of 'the teacher's scolding Taroo' occurred to make the event of 'Taroo's answering the question'. Thus the former event is the causing event (cause), and the latter, the de sired event (effect). Semantically speaking, this sentence is also a causative one, and the desired event is marked by the complementizer 'yooni' as in case of (26). In Japanese, the syntactic causative uses a morphological process of forming a complex verb, and no complementizer is used to mark the desired event. But sentences with a mani pulative verb and the Japanese equivalent of the English 'so that' construction both use the complementizer 'yooni' to mark the desired event. We may say, therefore, that in Japanese, there is match between syntax and semantics when the causation is expressed in sentences with a manipulative verb and the equivalent of the 'so that' structure, but the syntactic causative sentence does not resort to this kind of matching, but to a morphological process of merging the embedded verb and the causative verb 'sase'. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (C) Scale of match between syntax and semantics of cause and effect We modified the definition of the causative situation suggested by Shibatani, and our modified version includes the manipulative verb construction (/object control) and the equivalent of the English 'so that' clause construction. In Korean, the syntactic structure is consistent in all the constructions we looked at. This is shown by the fact that the same complementizer 'tolok' or 'key' is used throughout those constructions. The desired event described in the embedded clause is marked by the same complementizer regardless of what the causing event may be and how clearly the causing event is semantically expressed. In English, however, three syntactic constructions are used for the constructions of semantic 'cause' and 'effect' relation. The bare-infinitive clause is used in the syntactic causative with 'have', 'make' as the main verb and the permissive construction with 'let' as the main verb. The to- inf initival clause is used for the object control structure and an embedded clause led by 'so that' is used for many of the sentences with one of the manipulative verbs and the structures which do not allow a bare-infinitive or to- inf initive clause for subcategorization and some other grammatical reasons. In the sense that many of the manipu lative verbs allow both the 'to-infinitive' clause and a clause led by 'so that', these verbs share the same A t O Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. structure with an embedded clause led by 'so that' as the sentences which do not allow 'to-infinitive', but only allow an embedded clause with 'so that'. In Thai, one of the three periphrastic (=syntactic) causative structures has the same syntactic structure as the object control structure and the Thai equivalent structure of the English 'so that' structure. All these structures utilize 'hay' as the complementizer. But one of the syntactic causatives has an embedded clause without this complementizer, and another one uses as the main verb this complementizer which is originally a verb meaning 'give'. Although there is a match between syntax and semantics for causatives, there are some exceptions among syntactic causatives. In Japanese, the syntactic causative uses a morphological process of compounding the embedded verb and the main verb 'sase', not using any complementizer to mark the boundary between main clause and embedded clause. In the object control structure (— the manipulative verb construction) and the Japanese equivalent of the English 'so that' structure, the same complementizer 'yooni' is used to mark the desired event described in the embedded clause. Among the three structures used in English, two use the same structure in Japanese, except for the syntactic causative. We can develop a scale of match-mismatch between syntax and semantics in the case of causation for the four languages we looked at. 276 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (28) Scale of match-mismatch MISMATCH MATCH English ==> Japanese ==> Thai ==> Korean Korean shows a one-to-one match between syntax and semantics of causation, Thai is close to that, except that one syntactic causative is exceptional. Japanese has two of the three semantic causatives sharing the same syntactic structure, and two of the English semantic structures (manipulative verb structure/object control and 'so that' structure) share the same 'so that' syntactic structure although many of the manipulative verbs allow both 'to-infinitival' clause and 'so that' clause. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 6. Conclusion The main goal of this study is to argue that in Korean the syntactic causative, the object control structure, and the 'so that' clause with two different agents, one for the event of the main clause, the other for the event of the embedded clause, all involve 'causation' in the semantic sense with the definition of 'causation' as ! , A initiated A' event to make B carry out B' event". In all three of these phenomena, the main clause contains the 'causing event' while the embedded clause contains the 'desired event' regardless of whether the 'desired event' is actually realized or not. This semantic commonality in these linguistic phenomena is also matched by the selection of the complementizer 'tolok' which is mostly used to encode that the embedded clause marked by it is the 'effect' which occurred as the result of the event encoded in the main clause. We believe that we have argued successfully that as these phenomena share the semantic causation meaning, 4“ V» OTr a l c n c h s iro f c t m +• i n e +■ 11 it c * .tv / n U / n n W M V j h/Jf W w ^ W U d 4 . U l j W W JkW A Cl O l> U C complementizer. In chapter 3, we claimed that the syntactic causative in Korean is biclausal at all levels, unlike some linguists have claimed before. This was quite crucial because unless the syntactic causative is biclausal, it cannot be syntactically compared to the object control structure and the 'so that' 278 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. clause structure. In chapter 2 we ran some subjecthood tests to prove that the nominative causee is indeed the embedded subject. By confirming the subjecthood of the nominative causee we could argue that there is an embedded clause marked by the complementizer 'tolok' when the causee is nominative- marked. We also argued that even when the causee is marked accusative or dative, the syntactic causative is biclausal. We showed that in this case there is an empty subject slot in the embedded clause which can be optionally filled. Various elements can fill this position as we showed in chapter 3 including NPs, pronouns, anaphors, numeral classifiers which are either equivalent to the causee in reference or include the causee as a part of it or being a part of the causee. We also showed, however, that it is not necessary that the embedded subject in the syntactic causative should be equivalent even when it is not overt as long as there is a pragmatic and semantic connection between the causee and the embedded subject. In chapter 4, this same idea was pursued with the topic of subject control and object control. Just like the syntactic causative we looked at in chapter 3, whether it is subject oriented or object oriented, the embedded subject does not have to be equivalent to the main clause subject in the case of sentences with 'nolyekhata (try)', 'yaksokhata (promise)', and 'wenhata(want)'-type verbs and with the object in the case of sentences with 'seltukhata (persuade)'-type verbs. In the 279 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. latter case especially, the possible candidates with which to fill the overt or covert embedded subject position, are equiva lent to the ones for the syntactic causative. We also argued that in Korean it is not always the semantics of the verb which exclusively determines subject-oriented or object- oriented control, but at times the choice of the complementi zer is crucial. The sentences with verbs like 'malhata (tell)', and 'wihyephata (threaten)' may be treated as subject-oriented or object-oriented depending on whether 'lyeko' is used or 'tolok' is used as the complementizer. We claimed that the syntactic causative and so-called object control in Korean share not only the fact that the elements, covert or overt, which can be the embedded subject are basically the same, but also the fact that the same complemen tizer 'tolok' is used matching the semantics and syntax of 'causation' which these two phenomena share, We also tried to show that the 'tolok'-led embedded clause in the syntactic causative and the traditional object control structure may be an adverbial clause rather than an argument of the main verb. In chapter 5, we claimed that the 'so that' clause structure in Korean with two different agents also contains semantic 'causation' having the causing event expressed in the main clause and the desired event in the embedded clause. So this phenomenon shares several properties with the syntactic causative and the so-called object control structure in that the same complementizer 'tolok' is used and the causing event 280 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and the desired event are encoded in two different clauses, and the causer (the agent of the causing event) and the causee (the agent of the desired event) are different, and the embedded clause is an adverbial clause. We tried to briefly show that the syntax and semantics match in the phenomenon of •'causation'' with the definition we set in this study may not totally be language-specific only for Korean. Thai also shows a rather tight match between the syntax and semantics of the three linguistic phenomena we looked at while English does not have this kind of match, and Japanese has a partial match. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FOOTNOTES Chapter 2: 1. 'Subject honorification', '-kkeyse' marking, and '- myense' are also used as 'subjecthood' tests in K.A. Kim (1991). If the head noun is something like 'kay (dog)', which is not an inalienable part or the person's skill, it does not trigger the subject honorification. *Sensayngnim-uy kay-ka wucheypwu-lul mu-shi-ess-ta. teacher-Gen dog-Nom mailman-Acc bit—HON—PST—IND 'The teacher's dog bit the mailman.' 3. Control phenomena are well-documented in the literature (cf; Foster1984j Larson, 1991‘ Manzini, 1983; Radford, 1988; Riemsdijk and Williams, 1986, etc), nohanan (1983) defines control as "the relation between a lexically unrealized grammatical function and another grammatical function in the matrix clause." The unrealized gramma tical function is called the 'controllee', and its antecedent the 'controller'. Farkas (1988) has a similar definition for control as "a relation of referential dependence between the 'missing subject' and some argument of the matrix clause." In this paper, we adopt these definitions as the working definition of control. 282 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4. The incorporation also occurs in verbalizing a native Korean noun. 5. There is no empty subject position when the causee is in the nominative because in this case the causee is assumed to be the embedded subject. 6. It may be somewhat difficult to understand how a light verb can have its own object. 7. In his analysis, the light 'hata' or 'suru' does not have any capacity to assign case or theta-role to the sentential arguments. 8. This is consistent with Grimshaw and Hester's (1988) 7icw 0 X 1 the verbal uCuu • 9. Exceptionally 'man (only) + i (nominative case)' or 'man (only) + ul (accusative case)' can occur, although the reversed order 'i + man' or 'ul + man' is not allowed. Chapter 3: 1. The lexical causative morpheme changes its form accord ing to the phonological environment. 283 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2. '-Shi—' is sometimes written as '-si-'. This is phonemic. 3. He does not deal with the syntactic causative with dative causee. 4. Youn (1990) acknowledges that "subject honorification", and "reflexivization" tests are used to test the underlying subject in Snibatani (1973). He seems to use, however, these tests for testing 'subject' in general. Therefore, it is not clear whether the final subject is his sense is the subject in the final level or not. 5. Gerdts calls 'key' a complementizer. It is not clear in what sense she is using the term. Ovbiously in her mono- rl aiigaI analvcic a rnmnl omonf * ? tov cknu 1 r»of Ko o n element leading an independent clause. 6. For the sake of consistency, we use '—HON- ' instead of '—SH— ' used by Gerdts in order to indicate the subject honorific marker '—shi-'. 7. 0'Grady (1991) argues that the PRO in this sentence involves 'obligatory control' (cf. Bouchard, 1984). The PRO in an 'obligatory control' structure is assumed to be subject to the locality condition. He claims that the PRO in (43) is locality condition. He claims that the 284 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PRO in (43) is controlled by a noun phrase 'John' that is two clauses higher. He assumes that '[PRO ttena]-key' is a clause, (with which we agree), thus violating the locality condition. He claims that the fact that a sentence like (43) is not acceptable can be an evidence for biclausaiity of the syntactic causative whose causee is in nominative, rather than for monoclausality of it. 8. There is another type of reflexive pronoun, which is 'caki'. Refer to D.W. Yang (1988), and S.S. Hong (1985) for the distribution of this reflexive pronoun. 9. When the NP refers to honorable people, then it could be marked by the honorific nominative marker 'kkeyse', while the numeral quantifier is marked by the plain n o m i n a t i v e marker 'ka'. The NP can a l s o marked by the topic marker 'nun', while the numeral quantifier is marked by the nominative 'ka' or the accusative case 'lul' because the topic marker is not considered one of the regular case markers in Korean. 10. We claim that in the example (6) there is no zero anaphcr positron in the embedded clause, which is coreferential with the dative (/accusative)-marked causee in the main clause. 285 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11. The parenthesized elements in the English translation are the ones which are not overtly expressed in the Korean data. Chapter 4: 1. Overt nominals can occur in the embedded subject position in some other languages, too. See Kuno (1976), IIda (1S31), and Sakaguchi (1330) for Japanese and Zee (1987) for Serbo-Croatian. 2. As far as we know, this is the only [+AGR] feature found in a verb. 3. The reason why the subject honorific marker can be affixed to the embedded verb is because the main clause subject can be the controller of the embedded subject. 4. This is due to the fact that the main clause subject is the one most readily available to be the controller. But this preference may change as a context given clearly shows otherwise. 5. This phenomena is not well—studied yet. We will pursue this matter in further research. 286 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6. The main clause subject and the overt embedded subject are not corei'erentiai with each other in the example (29). But there is ovbiously a clear semantic link between them. 7* This is not the only available interpretation. The embedded subject could be the combination of the main clause subject and the oblique NP, the combination of the mam clause subject and a third party, the combi™ nation of the main clause subject, the oblique and a third party. It also could be a third party only, which is not present in the sentence. 8. Using the accusative marker 'lul' instead of the dative marker is much more easily acceptable to us. 9. But when the overt embedded subject is a pronoun, it is caSily COuSlucxcu COiTcfcircntial with the j T i c2 J .i l C’ldUSe object. Many people find it hard to accept that the pronoun is a third party. This applies to a name in the embedded object position. In both cases, we do not find it acceptable that the pronoun or the name can be a third party. Thus we do not agree with H.S. Lee in this although we agree and show later that an NP which is not coreferential with the main object can occur in the embedded subject position. 287 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10. As we mentioned in footnote (9), we believe that even in this example the pronoun 'ku (he)' cannot be coreferential with a third party. 11. This shows that even in English pragmatics relaxes the syntactic constrain a little. This is also true in the case of object control with a main verb like 'persuade' as we show in a later section. 12. Williams does not clearly say whether his properties of of obligatory control apply only to English or are meant to be universally workable. This subsection is based on the assumption that they are universal or at least universal tendancies. 13. Adjuncts may also be deleted. 14. This fact was once mentioned in the section on the verb 'hata (do)' in Chapter 2. We repeat this here for the sake of convenience. 15. We use the word 'predominantly' here because as we have seen there are other possibilities. - S n n ^OO Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16. It is arguable to say that the embedded clause is passivized by inserting "be allowed to" between the main clause and the embedded clause because the actual embedded clauses in the examples (144)-(145) are still active. 17. This is again a strong tendency because there are other possibilities. 18. There must be a semantic connection between the main clause object and the embedded subject where it is clear that the main clause object can somehow manipulate the embedded subject to carry out the event. 19. Again this coreferentiality between the main clause subject and the covert embedded subject is strongly preferred, but not the only possibility as we have seen. _i_ i __________t -i_“ “ 1 • u / i. _ _ w i i . * • 1 § f i _ » (9U U IC U tU C l UUlllpXCUICUUl^CSX. XXAC ' AUt'd' f ' AJ. WXUdy ‘ f ’ 1C ' may also be used in the place of 'lyeko'. Chapter 5: 1. It is arguable that the caused event is not absolutely realized in this case because a sentence like 'But John did not go to the drugstore, but went to his friend's house.' 289 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2. For the purpose of this study, 'key' and 'tolok' may be used interchangeably. Thus wherever 'key' is used in this study, it may be replaced by 'tolok', and vice versa. 3. We mean by the content of the order the actually given order. If the teacher gave the order of reading the book this week to the student, then the content of his (/her) order is 'reading the book'. We contrast this with the effect which may occur as the result of the order. 4. 'Let' is not included in their order/permit type verbs, but it is indeed one of them. This verb, however, allows a bare—infinitive clause. The complementizers 'key' and 'tolok' in Korean are affixes, and they are affixed to the embedded clause. Without them the embedded verb form must be changed. Thus their presence is obligatory. 2 9 0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References Ahn, Hee-Don. 1990: "On Light Verb Construction in Korean and Japanese," Japanese and Korean Linguistics. ed. by Hajime Hoji, 221-238. Stanford: The Center for the Study of Language and Information. Aissen, Judith L., and David M. Perlmutter. 1983: "Clause Reduction in Spanish." Studies in Relational Grammar 1. ed. by Perlmutter. Chicago: University of Chicago Press. Bach, Emmon. 1979: "Control in Montague Grammar." Linguistic Inquiry 10.515-31. Bach, Emmon. 1980: "In Defense of Passive." Linguistics and Philosophy 3.297-342. Borer, Hagit. 1989: "Anaphoric AGR." The Mull Subject Parameter ed. by O. Jaeggli and K. Safir, 69-109. Kluwer Academic Publishers. Bouchard, Denis. 1984: On the Content of Empty Categories. Dordrecht: Foris. Burzio, Luizi. 1981: Intransitive Verbs and Italian Auxiliaries. Doctoral dissertation. MIT, Cambridge, Massachusetts. Burzio, L. 1986: Italian Syntax: a Government—Binding Approach. Dordrecht: D. Reidel. Chierchia, Gennaro. 1983: "Outline of a Semantic Theory of (Obligatory) Control." West Coast Conference on Formal Linguistics 2.19-31. Stanford, CA: Stanford Linguistics Association. Chierchia, Gennaro. 1984: Topics in the Syntax and Semantics of Infinitives and gerunds. Amherst: University of Massachusetts dissertation. Chierchia, Gennaro. 1989: "Structured Meanings, thematic roles and Control." In Chierchia et al., 131-66. Chierchia, Gennaro; Barbara Partee? and Raymond Turner(eds). 1989: Properties, types and meaning II. Dordrecht: Kluwer. Cho, Eui-Yon. 1987: "Korean Causatives in Relational Grammar: Argument against the clause union analysis." Harvard Studies in Korean Linguistics 2 ed. by S. Kuno et al., 3— 26. Seoul: Hanshin. 291 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cho, Eui-Yon. 1988: Some Interaction of Grammar and Pragmatics in Korean. Ph.D. dissertation. University of Illinois. Choe, Hyun-sook. 1988: Restructuring Parameters and Complex Predicates. Ph.D. dissertation. MIT. Choi, Young-Seok. 1988: A Study of Ascension Constructions in Korean. Ph.D. dissertation. University of Hawaii. Chomsky, Noam. 1980: "On Binding." Linguistic Inquiry 11.1-46. Chomsky, Noam. 1981: Lectures on Government and Binding. Dordrecht: Foris. Comrie, Bernard. 1976. "The syntax of causative constructions: cross-language similarities and divergences." The Grammar of Causative Constructions Syntax and Semantics 6 ed. by M. Shibatani. New York: Academic Press. Comrie, Bernard. 1984: "Subject and Object Control: Syntax, Semantics, and Pragmatics." Berkeley Linguistics Society 10. 450-64. Berkeley, CA: Berkeley Linguistics Society. Comrie, Bernard. 1986: "Reflections on Subject and Object Control." Journal of Semantics 4.47-65. Dell, Francois. 1981. "On Certain Sentential Complements in Tagalog." Philippine Journal of Linguistics 12.11—28 Dowty, David. 1989: "On the Semantic Content of the Notion 'Thematic Role'." In Chierchia et al.. S9~13C. Dowty, David and Pauline Jacobson. 1989: "Agreement as a Semantic Phenomenon." Proceedings of the Fifth Eastern states Conference on LinguisticsfESCOL '38t. ed. by Joyce Powers and K. deJong, 95-108. Columbus, OH: The Ohio State University. Farkas, D. 1988: "On obligatory control." Linguistics and Philosophy 11. 27-58. Foley, William, and Robert Van Valin. 1984: Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. Gerdts, Donna. 1990: "Revaluation and inheritance in Korean Causative Union" Studies in Relational Grammar 3 ed. Postal, Paul and Brian Joseph. Chicago and London: The university of Chicago Press. 292 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gibson, Jeanne & Eduardo Raposo. 1986: "Clause Union, the Stratal Uniqueness Law, and the Chomeur Relation." Natural Language and Linguistic Theory 4f3>. 295-332. Gritnshaw, Jane and Annin Mester. 1988: "Light Verbs and Theta- Marking, "Linguistic Inquiry 19:205-232. Han, Hak-Sung. 1987a: "On Case Conversion in Korean.", Harvard Studies in Korean Linguistics 2 ed. S. Kuno et al., 67-79. Seoul: Hanshin. Han, Hak-Sung. 1987b: The Configurational Structure of the Korean Language. Ph.D. dissertation. University of Texas at Austin. Higgins, F. Roger. 1973: The Pseudo-cleft Construction in English Cambridge, MA: MIT dissertation. Hong, Sungshim. 1985: A and A' Binding in Korean and English: Government—Binding Parameters. Ph.D. dissertation. University of Connecticut. Huang, C.T.James. 1989: "Pro-drop in Chinese: a Generalized Control Theory." The Null Subject Parameter ed. by O.Jaeggli and K.Safir 185-214. Dordrecht: Kluwer. lida, Masayo. 1991: Context and Binding in Japanese. Stanford, CA: Stanford University dissertation. Jackendoff, Ray S. 1972: Semantic Interpretation in Generative Grammars Cambridge, MA: MIT Press. Jackendoff, Ray S. 1974: "A Deep Structure Projection Rule." Linguistic Inouirv 5.481—506. Jo, Mi—Jeung. 1986: Fixed Word Order and the Theory of the preverbal focus position in Korean. Ph.D. dissertation. University of Washington. Kayne, Richard. 1975: French Syntax. MIT Press, Cambridge, Massachusetts. Kang, Myung-Yoon. 1988: Topics in Korean Syntax: Phrase Structure, Variable Binding and Movement. Ph.D. dissertation. MIT. Kang, Young-Se. 1986: Korean Syntax and Universal Grammar. Ph.D. dissertation. Harvard University. Kim, Alan Hyun-Oak. 1990: "Numeral Quantifier Constructions and Definiteness Restrictions" ICKL7 Papers from the Seventh International Conference on Korean Linguistics ed. Eung-Jin Baek., 121-132. Seoul: Hanshin. 293 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kim, Kyungan. 1991: 5 1 Tough Movement and Case Marking in Korean" ms. University of Southern California. Kim, Kyungan. 1992a: "Causatives and Word Order in Korean" ms. University of Southern California. Kim, Kyungan. 1992b: "Lexical Borrowing in Korean," ms. USC. Kim, Kyungan. 1993: "Syntactic Causatives in Korean," ms. USC. Kim, Nam-Kil. 1378: "xoiok Sentential Complements in Korean." Papers in Korean Linguistics ed. Chin-w. Kim, South Carolina: Hornbeam Press. Kim, Nam-Kil. 1986: "Predication in Korean." Korean Linguistics. ed. Joe J. Ree, 4.47-60. Seoul: the International Circle of Korean Linguistics. Kim, Nam-Kil. 1984: The Grammar of Korean Complementation Hawaii: Center for Korean Studies, University of Hawaii. Kim, Sun-Hee. 1994: "On the Distribution and Interpretation of pro.1 1 Harvard Studies in Korean Linguistics V ed. S. Kuno et at., 509-522, Seoul: Hanshin. Kim, Yong-Bum. 1988: A fragment of Korean Phrase Structure Grammar. Ph.D. dissertation. University of Victoria. Koster, J. 1984: "On Binding and Control," Linguistic Inquiry 15, 417-459. Kuno, Susumu. 1976: "Subject Raising." Syntax and Semantics, vol 5: Japanese Generative Grammar, ed. Masayoshi Shibatani, 17-49. New York: Academic Press. Lakoff, George. 1965: On the Nature of Syntactic Irregularity. Published as Irregularity in Syntaxfl970). New York: Holt, Rinehart and Winston. Larson, Richard. 1991: "Promise and the Theory of Control" Linguistic Inquiry 22. 103-139. Lasnik, Howard. 1986: "On the Necessity of Binding Condi tions," Essays on Anaphora, Dordrecht: Xiuwer Academic Publishers: Dordrecht. Lee, Chung-Min. 1983: "(In)Definites, Case Markers, Classifiers and Quantifiers in Korean," Harvard Studies in Korean Linguistics III ed. S. Kuno, et al., 469-488. Seoul: Hanshin Lee, H.B. 1970b: "On Negation in Korean." Language Research 6.2. 294 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lee, H.B. 1972: "Problems in the Description of Korean Negation," Language Research 8.2. Lee, H.B. 1987: "On Empty Categories in Korean," Ene 12- 1.261-294. Lee, H.S. (1986): "Pronominal Anaphors and Binding Domain: with Reference to a Quasi-reflexive caki in Korean," ms. UCLA. McCawiey, J.D. 1968: "Lexical Insertion in a Transformational Grammar without Deep Structure," Papers from the Fourth Regional Meeting, ed. Darden, et al. Chicago: Chicago Linguistic Society. Manzini, M. R. 1983: "On Control and Control Theory," Linguistic Inquiry 14. 421-446. Mohanan, K. P. 1983: "Functional and Anaphoric Control," Linguistic Inquiry 14. 641-674. O'Grady, William. 1991: Categories and Case. Amsterdam/ Philadelphia: John Benjamins Publishing Co. Oshima, S. 1979: "Conditions on Rules: Anaphora in Japanese," Exploration in Linguistics, ed. G. Bedell, et al. Tokyo: Kenkyusha. Park, Kabyong. 1992: Light Verb Constructions in Korean and Japanese. Ph.D. dissertation. University of North Carolina, Chapel Hill. Perlmutter, David M., and Paul M. Postal. 1974: Lectures in Relational Grammar. LSA Summer Institute, University of Massachusetts, Amherst. Radford, Andrew. 1988: Transformational Grammar. Cambridge: Cambridge University Press. Riemsdijk, Riemsdijk. V., and Edwin Williams. 1986: Introduction to the Theory of Grammar Cambridge, London: The MIT Press. Rosenbaum, Peter S. 1967: The Grammar of English Predicate Complement Constructions. Cambridge, MA: MIT Press. Sag, Ivan A. and Carl Pollard. 1991: "An Integrated Theory of Complement Control." Language 67. 63—113. Sakaguchi, Mari. 1990: "Control Structures in Japanese" Japanese and Korean Linguistics, ed. Haiime Hoji, 303-318. Stanford: The Center for the Study of Language and Infor mation. TOC Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Schacter, Paul. 1976: "The Subject in Philippine Languages: Topic Actor, Actor-topic, or none of the above." Subject and Topic ed. Charles N. Li, 491-518. New York: Academic Press. Schacter, Paul. 1977: "Reference-related and Role-related properties of Subjects." Syntax and Semantics, vol 8: Grammatical Relations, ed. Peter Cole and Jerrold Sadock. 279-306. New York: Academic Press. Searle, John R. 1963: Speech Acts: An Essay in the Philosophy of Language. Cambridge University Press. Cambridge. Searle, John R. 1975: "A Taxonomy of Illocutionary Acts." Language. Mind, and Knowledge. 344-69. Minneapolis: University of Minnesota Press. Shibatani, Masayoshi. 1973: "Lexical versus Periphrastic Causatives in Korean," Journal of Linguistics 9. 281-297. Shibatani, Masayoshi. 1990: The languages of Japan Cambridge: Cambridge University Press. Song, Seok—Choong. 1971: "A Note on Negation in Korean," Linguistics 76. 59-76. Song, Seok-Choong. 1973: "Some Negative Remarks on Negation in Korean," Language Research 9.2. Song, Seok—Choong. 1988: Explorations in Korean Syntax and Semantics/ Institute of East Asian Studies. University of California, Berkeley. Suh, Jinhee. 1990: Scope Phenomena and Aspects of Korean Syntax. Ph.D. dissertation. Uoc. Williams, Edwin S. 1980: "Predication." Linguistic Inquiry 11. 203-38. Yang, Dong—Whee. 1952: "Control and Binding in Korean." Ene 7-2. 257-283. Yang, Dong—Whee. 1984: "Hwaktay Thongcey Iron (Extended Control Theory)" Ehakyenkwu 20-1. 19-29. Yang, Dong—Whee. 1S85: "On the Integrity of Control Theory." NELS 15: 389-408. Yang, Dong—Whee. 1988: Hankwue-uy tavvonghwa /Anaphora in Korean. Seoul: Hankuk Yenkuwon. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Yang, In-Seok. 1972; Korean syntax; Case Markers. Delimiters. Complementation and Relativization. Seoul: Paek Hap Sa. Youn, Cheong. 1990; A Relational Analysis of Korean Multiple Nominative Constructions, Ph.D. dissertation. State University of New York at Buffalo. Zubizarreta, M. L. 1985: "The Relation Between Morphophonology and Morpho-syntax: The Case of romance Causatives,1 1 Linguistic Inquiry lo, 247-89. Zee, Draga. 1987; "On Obligatory Control in Clausal Complements.1 1 Working Papers in Grammatical Theory and Discourse Structure; Interactions of Morphology. Syntax and Discourse, ed. Masayo Ida, Stephen Wechsler, and Draga Zee, 139-68. Stanford, CA; Center for the Study of Language and Information and Chicago: University of Chicago Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Correspondence and faithfulness constraints in optimality theory: A study of Korean consonantal phonology
PDF
Borrowing and borrowability
PDF
Conditions on null objects in Basque Spanish and their relation to leismo and clitic doubling
PDF
A case grammar of the parker manuscript of the "Anglo-Saxon chronicle" from 734 to 891
PDF
Temporal Structure Of Spoken Korean: An Acoustic Phonetic Study
PDF
Aspects of questions in Japanese and their theoretical implications.
PDF
Beyond clowns and kings: Aesthetic and ideological subversion in baroque tragicomedy
PDF
Communicative rights and responsibilities in an East Los Angeles barrio: An analysis of epistemic modal use
PDF
Antigone: A study in critical method
PDF
Controversies over grammar: Contexts and purposes in the seventeenth and eighteenth centuries
PDF
A critical edition of the Escorial manuscript of 'historia de los indios de la neuva espana' of fray toribio de benavente (motolinia) (spanish text)
PDF
A comparative study of focus constructions
PDF
"A broken bundle of mirrors": Identity in the work of John Barth
PDF
A descriptive grammar of San Bartolome Zoogocho Zapotec
PDF
"Emilia Galotti" and its aesthetic response: A case study for the reader-response criticism
PDF
Charlotte Smith: Life of a novelist, novels of a life
PDF
A study on the content and determinants of close exchange relationships in channels of distribution
PDF
Ellipsis constructions in Chinese
PDF
The Grammar Of English Causative-Transitivity
PDF
'Beowulf' and 'the hobbit': elegy into fantasy in j. R. R. Tolkien's creative technique
Asset Metadata
Creator
Kim, Kyungan
(author)
Core Title
Causativity in Korean: Syntactic causative, control, and purpose
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)
Advisor
Comrie, Bernard (
committee chair
), Hawkins, Jack (
committee member
), Woodard, Roger (
committee member
)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c17-493010
Unique identifier
UC11352599
Identifier
9625252.pdf (filename),usctheses-c17-493010 (legacy record id)
Legacy Identifier
9625252-0.pdf
Dmrecord
493010
Document Type
Dissertation
Rights
Kim, Kyungan
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
Tags
language, linguistics