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
/
On the so-called Japanese reflexives zibun and zibunzisin
(USC Thesis Other)
On the so-called Japanese reflexives zibun and zibunzisin
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. UMI films the text directly 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 afreet reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also 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 UMI directly to order. UMI A Bell & Howell Information Company 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ON THE SO-CALLED JAPANESE REFLEXIVES ZIBUN AND ZIBUNZISIN Copyright 1998 by Akemi Kagawa A Thesis Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree MASTER OF ARTS (East Asian Languages and Cultures) August 1998 Akemi Kagawa Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 1393172 UMI Microform 1393172 Copyright 1999, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. U NIV ERSITY O F S O U T H E R N C A L IFO R N IA TH E GRA DUATE SCH O O L U N IV ER SITY RARK LOS A N G ELES. C A LIFO R N IA 80007 This thesis, •written by JU&EMI.. KAGAWA____________________________ under the direction of kj&r. Thesis Committee, and approved by all its members, has been pre sented to and accepted by the Dean of The Graduate School, in partial fulfillment of the requirements for the degree of MASTER OF ARTS____ THESIS COMMITTEE / Reproduced with permission of the copyright owner. Further reproduction prohibited without permission ACKNOWLEDGEMENTS ii I would like to express my gratitude to my friends and colleagues at University of Southern California, who provided invaluable comments during various stages of writing this thesis. I am grateful to Ayumi Ueyama for her thoughtful suggestions and comments. I am also much indebted to Keiko Miyagawa, who has been more than generous with her time in answering my countless questions about grammatical judgments and analyses and in engaging in long discussions. Much appreciation is owed to my committee chair, Professor Hajime Hoji for his unending guidance and his painstaking reviews o f the drafts. Special thanks are due also to Professor Yen-hui Audrey Li for her encouragement and to Professor Nam-kil Kim for valuable advice and comments. Writing this thesis would have been mush more painful without my family. I thank Takao, Hanako-Gabrielle, and Shiori-Michaelle for their encouragement, patience, and support throughout the long months of this past year. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. iii CONTENTS Acknowledgments ii Abstract v 1. Introduction 1 1.1 The theoretical background 1 1.2 Anaphors and pronouns 3 1.3 A problem for the Binding Theory 4 2. Summary o f the previous literature 9 2.1 Zibun as a pronoun 9 2.1.1 Fukui (1984): Zibun bound by the closest A’-binder 9 2.1.2 Problems with Fukui’s proposal: Split antecedence 12 2.1.3 Ueda (1986): Zibun as a bound pronominal 17 2.1.4 A problem with Ueda’s claim I: Soo-su vs. do so 25 2.1.5 A problem with Ueda’s claim II: Disjoint reference 27 2.2 Zibun as an anaphor 29 2.2.1 Aikawa (1993): Zibun as non-reflexivizer 29 2.2.2 Problems with Aikawa’s proposal 35 2.2.3 Katada (1991): Zibun as an operator anaphor 37 2.2.4 Problems with Katada’s analysis 41 2.3 Hoji (1997): Otagai 44 3. Properties of zibun and zibunzisin 49 3.1 Zibun and zibunzisin are not anaphors 49 3.1.1 The absence of locality 49 3.1.2 The c-command restriction 50 3.2 Other properties of zibun and zibunzisin 52 3.2.1 Weak Crossover Effects 52 3.2.2 Split Antecedent 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4. Reanalysis o f zibun and zibunzisin 54 4.1 [pro[z/6wn]] and [^xo[zibunzisiri\\ 54 4.2 The advantages of postulating the structure [pro [...]] 57 4.3 The absence of local anaphors in Japanese 60 4.4 Remaining Issues 61 5. Conclusion 66 Appendix 67 Reference 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT This thesis critically examines the empirical evidence put forth in the literature for the view that zibun and zibunzisin are (local) anaphors, and concludes that they are not (local) anaphors. It further argues that the relevant anaphoric relation is not between zibun!zibunzisin and their antecedents, but it is between pro in [m>pro [s zibun/zibunzisin]] and the antecedent of pro. The internal structures for zibun and zibunzisin thus accounts for the absence of Condition B effects with zibun and zibunzisin. The possibility of the absence of local anaphors altogether in Japanese which is suggested by the present thesis, together with Hoji’s (1997) analysis of the so-called Japanese reciprocal otagai, will be discussed in terms of the absence of formal agreement features in Japanese. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1. Introduction 1.1. Theoretical Background This thesis is an investigation o f the so-called Japanese reflexives zibun and zibunzisin in relation to their characterization in generative grammar, in particular, their binding theoretic status. Binding Theory is concerned with the distribution of the three types of NP: anaphors (i.e., reflexives (e.g., himself, herself, etc.) and reciprocals (e.g., each other)), pronouns, and R-expressions. (1) Binding Theory (Chomsky, 1995, p. 96) Condition A: An anaphor must be bound in a local domain. Condition B: A pronominal must be free in a local domain. Condition C: An r-expression must be free. (2) Definitions (a) Bound: a is bound by 3 if and only if a and P are coindexed, 3 c- commands a, and 3 is in an A-position. (b) C-command, a c-commands 3 if and only if a does not dominate 3, 3 does not dominate a , and the first branching node dominating 3 also dominates a. Chomsky (1995) states that “the local domain, or governing category as it is frequently called, involves reference to government” (p. 101), and gives the following definitions. (c) The governing category for a is the minimal complete functional complex (CFC) that contains a and a governor of a and in which a ‘s binding condition could, in principle, be satisfied. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 (d) Complete functional complex (CFC) is a projection containing all grammatical functions compatible with its head. The sentences in (3) exemplify how the principles regulate the anaphoric relationship(s) between NPs. (3) a. Johnj has hurt himself|/*himi. b. Johni has hurt his* friend. c. Johnj believes that Mary hurt himj/*himselfi. d. Johni has hurt * Johni. e. Johni has hurt *John^s friend. (3a) is grammatical with himself, but ungrammatical with him. The occurrence of him self satisfies Condition A since it is bound by the antecedent John in its governing category. However, the occurrence o f him violates Condition B because it should not be bound by John in its governing category. (3b) is grammatical since the governing category of his is the NP that includes it, and the antecedent John occurs outside the governing category. (3c) with him self is ungrammatical since it is not bound in its governing category, i.e., the embedded clause. It is a violation of Condition A. In contrast, (3 c) with him is grammatical because the sentence satisfies Condition B: John occurs outside the governing category of him. Condition C rules out both (3d) and (3e): the R-expression John must be free. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 1.2. Anaphors and pronouns As illustrated in section 1.1, the binding theory contains three conditions, each regulating the distribution and interpretation of one specific type of NP. In this section, I would like to briefly summarize how anaphors behave differently from pronouns when binding is concerned. The reason for this is that one of the purposes of this thesis is to argue against the widely-held view that zibun and zibunzisin in Japanese are reflexive anaphors corresponding to English reflexives such as myself. Condition A and Condition B, as stated in (1), suggest that there is complementarity in the distribution of bound anaphors and bound pronouns. That is, whereas anaphors must be bound in a local domain, pronouns must be free. Further, with respect to an antecedent, anaphors must have a c-commanding antecedent, whereas pronouns do not need an antecedent. The paradigm is summarized in (4). (4) a. John; invited himself. b. Johnj should buy himself a car. b’. * Johni should buy yourself a car. c. * Himself invited Johni. d. *Johni hurt him;. e. John* hurt [hisi mother], f. Hisi mother hurt John. (4a) and (4b) are grammatical because him self has its antecedent, John, in its local domain. In contrast, (4b’) is ungrammatical although in a conversation between Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 two people, m yself refers to the speaker, and yourself refers to the addressee. Therefore, it seems that the ungrammaticality of (4b’) is not due to the interpretability of the reflexive yourself Rather, the fact that yourself does not satisfy the requirement makes the sentence ungrammatical. (4d) is ungrammatical since him must be free, and (4f) is grammatical because the pronoun his does not have its antecedent in its local domain. (4c) is ungrammatical since the antecedent John does not c-command the anaphor him self In contrast, (4f) is grammatical because pronoun binding does not require a c-commanding antecedent. In addition, pronoun binding allows a split antecedent (i.e., a plural pronoun can have more than one antecedent in a sentence), whereas anaphor binding does not. A case of split antecedents is shown in (5), where a plural pronoun they can be interpreted as having both John and Bill as its antecedents. (5) Johnj told Billj that theyt would win. (where k can be / + j) That anaphor binding does not permit split antecedents is illustrated in (6). (6) a. *John; told Billj about themselvest. (where k= i + j) b. Johnj told Billj about himselfj. c. Johnj told Billj about himselfj. 1.3. A Problem for the Binding Theory In Japanese, zibun is regarded as the most representative reflexive. However, it has been observed that the so-called Japanese reflexive zibun demonstrates a different range of characteristics from that o f English reflexives. A very well- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. known characteristic is long-distance binding of zibun. In English, reflexives and their antecedents must be in the same clause (i.e., local binding), whereas, in Japanese, the antecedent of zibun can be outside its governing category (i.e., long distance binding) (Aikawa (1993); Katada (1991); Ueda (1986), just to mention a few). (7) a. Johnj thinks that Billj hates himselfj/*;. b. Johnj-ga Billj-ga zibun^-o kiratteiru to omotteiru John NOM Bill NOM self ACC hate COMP think ‘John; thinks that Billj hates selfj’ In the English example of (7a), only the complement subject B ill, but not the matrix subject John, can be the antecedent of himself. In contrast, in the Japanese example of (7b), either the matrix subject John or the complement subject Bill can be the antecedent of zibun.1 It must be added here that in the previous literature, it is endorsed that the antecedent of zibun must be a subject, even in the case of long-distance reflexivization, as in (8). In English, such a subject orientation is not observed. Tukui (1984), in feet, assumes that in a sentence like (7b), the reading "zibun = the complement subject, B iir is very weak, while the reading "zibun = the matrix subject, John” is dominant. Fukui’s account for the near impossibility of "zibun = the complement subject” reading is that zbun is subject to Condition B. I will discuss Fukui's argument (1984) in detail in section 2.2.4 of chapter 2. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (8) a. (Fukui’s (6), 1984) Johnj-ga Billj-ni zibun^j-no koto-o hanasita. NOM DAT self GEN things ACC talked ‘Johni talked to Billj about self/*j’ b. (Aikawa's (32), 1993) [Johnj-ga Billj-ni [[Maryt-ga zibun^j-o hihansita] NOM DAT NOM self ACC criticized to] itta. COMP said ‘Johnj said to Billj that Maryk criticized se lffj’ In addition to the most explored “reflexive” zibun, Japanese is said to have another “reflexive,” zibunzisin. Zibunzisin has recently drawn more attention because the behavior of it is said to be different from that of zibun. For example, it is reported that while zibun need not be bound locally, zibunzisin must be (Aikawa (1993); Katada (1991)). (9) (Katada’s (3), 1991) Johnj-ga [Billj-ga Mike^-m zibunzisin?.i/.kj-no koto-o hanasita] NOM NOM DAT self GEN thing ACC told to itta. COMP said ‘Johnj said that Billj told Miket about self?*^’ Katada (1991) claims that in (9) zibunzisin takes only the embedded subject as its antecedent, concluding that zibunzisin must be bound locally. Note also that in (9) the subject orientation of zibunzisin is exemplified: the embedded indirect object Mike cannot be the antecedent of zibunzisin. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 The so-called Japanese refelexives behave differently from English reflexives as discussed above. Various attempts have been made in capturing the properties inherent to zibun and zibunzisin. Oshima (1979, 1983), for example, argues that grammar does not capture the interpretation of zibun since it is essentially a discourse anaphor. Others try to account for the properties of zibun or zibunzisin within the rules of grammar. As will be illustrated in chapter 2, Fukui (1984) and Ueda (1986) argue that zibun is a pronoun, while Aikawa (1993) and Katada (1991) consider it to be an anaphor. My position on the issue of the Japanese zibun and zibunzisin is as follows. As to pragmatic constraint, I cannot totally deny that pragmatic factors affect how their antecedents are determined. Such factors as ‘empathy’ (Kuno, 1987) might be involved in their interpretation. However, in this thesis, I would like to concentrate on grammatical aspects. As Fukui (1984) writes, there seems to be some grammatical constraint that regulates the distribution of zibun and zibunzisin. Regarding the nature of zibun and zibunzisin, I believe that they cannot be captured as anaphors. A detailed discussion will be provided in chapters 3 and 4. The organization of this thesis is as follows. Chapter 2 briefly summarizes the previous literature on zibun and zibunzisin. Section 2.1 introduces the studies in which zibun is claimed as a pronoun. Section 2.2 will review the studies which take the position that zibun is an anaphor. I will argue that all the views introduced in the subsection are unsound. In section 2.3, Hoji’s (1997a) analysis of otagai, ‘each other,’ the so-called Japanese reciprocal, will be introduced since this thesis is Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 inspired by his analysis o f otagai and I believe that zibun and zibunzisin should be analyzed along the lines of Hoji’s analysis of otagai. In chapter 3 ,1 will present the empirical materials that support my proposal that zibun and zibunzisin are not anaphors: section 3.1.1 will present the evidence that zibun and zibunzisin need not have their antecedent in their local domain. In section 3.1.2,1 will show that zibun and zibunzisin need not be c-commanded by their antecedent when coreference is at stake. That zibun and zibunzisin show typical weak crossover (WCO) effects is presented in section 3.2.1. Section 3,2.2.deals with split antecedents. In chapter 4, I will discuss the internal structure of zibun and zibunzisin and relevant discussions. I will argue that, based on the absence of Condition B effects in bound variable construal, the internal structure of zibun and zibunzisin must be understood as [pro[zibun\\ and [pro[zibunzisin]] respectively.2 Accordingly, the binding relationship between zibun!zibunzisin and their antecedents should be considered as that between the pro and the antecedents of zibun!zibunzisin. My claim in this thesis, together with Hoji’s (1997a) analysis of otagai, ‘each other’ the so-called Japanese reciprocal, cast a doubt on the existence of the local anaphor in Japanese. The possibility that Japanese lacks local anaphors altogether will be discussed in relation to the absence of formal agreement features in Japanese. Chapter 5 concludes this thesis. 2 I assume, following Hoji (1997a), that pro is a phonetically empty argument in this thesis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 2. Summary of the previous literature 2.1. Zibun as a pronoun In the following, I will review the arguments that zibun is a pronoun. The claims are based on the assumption that zibun has the following properties: (i) split antecedence (as in Fukui, 1984), and (ii) strict identity reading under VP deletion (as in Ueda, 1986). 2.1.1. Fukui (1984): Zibun bound by the closest A’-binder Fukui (1984) presumes that the subject position in Japanese is an A’-position (he calls his proposal the ‘Adjunct Subject Hypothesis (ASH)’). Fukui assumes that, under the ASH, a seemingly complicated phenomena concerning the binding of zibun is given an unified account. Take the subject orientation of zibun for instance. Fukui endorses Kuroda’s claim that zibun shows subject-object asymmetry in adjunct clauses (1965, p. 139 & 144 -145). That is, if zibun occurs in the subject position in an adjunct clause, it can refer to the matrix subject (cf. (10a)), whereas if it appears in the object position in an adjunct clause, it cannot refer to the matrix subject (cf. (IOb)). (10) Adjunct clauses a. (Kuroda’s (29), 1965, chap. 5; Fukui’s (27a)) Georgej-ga [s zibunj-ga e ,- sitai] toki-ni sigotoj-o suru (koto) want time at work do-Pres ‘Georgei does (his) work when self feels like doing it.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 b ((Kuroda’s (48), 1965, chap. 5; Fukui’s (28a)) ?*Johni-wa [s Bill-ga zibun;-o mihatte iru] aida-wa sigoto-o suru be watching while work do ‘John; does (his) work while Bill watches self.’ In contrast, zibun does not show such an asymmetry in complement clauses. In (lib), for example, zibun that occurs in the object position can refer to the matrix subject John as well as to the complement subject Bill, although, according to Fukui, the reading “ zibun = BilF is very weak3 (11) Complement clauses a. (Fukui’s (29a)) Johnj-ga [szibiuii-ga tensai -da]-to omotte -iru (koto) NOM genius Cop COMP think Pres ‘John; thinks that self is a genius.’ b (Kuroda’s (58), 1965, chap. 5) Johni-wa [Billj-ga zibuni^r?j -o miru-no]-o tome-ta ‘John; prohibited Billj from seeing selfi/j.’ Fukui’s proposals can be summarized as in (12)4 and (13). 3In section 2 .2 .4 ,1 will argue against Kuroda and Fukui's factual claim concerning the subject-object asymmetry. 4I understand Fukui's proposal should be read as (12'). (12’) a. The subject position o f a sentence a is an A-position if a is properly governed. b. The subject position o f a sentence a is an A’-position if a is not properly governed. According to Fukui, a sentence S’ is properly governed by the matrix verb if both the S’ and the matrix verb are immediately dominated by a maximal projection VP (Fukui 1984, p. 27). For the illustration of the relevant notion, see (14) below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 (12) (Fukui’s (32a, b)) a. The subject position o f a S which is properly governed is an A- position. b. The subject position of a S which is not properly governed is an A’- (13) (Fukui’s (37)) Zibun must be bound by the closest A’-binder. Based on the above proposals, Fukui assumes the difference between the structures of complement clauses and that of adjunct clauses, as in (14), is the key to explain the subject-object asymmetries. (14) a. Complement clauses (Fukui’s (36a)) S ’ / \ position S COMP / \ VP, / \ NP V / \ COMP s / \ NP VP2 / \ NP V Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. Adjunct clauses (Fukui’s (36b)) S ’ / \ S COMP / L \ NP ( s ) VP, / \ \ S COMP V / \ NP VP2 / \ NP V Fukui assumes that in (14a), since a maximal projection VPi dominates both the encircled S’ and the matrix verb, the encircled S’ is (properly) governed by the matrix verb. Therefore, the subject position of the complement clauses is an A- position. In contrast, in (14b), because a maximal projection VPi does not dominate the adjunct clause (i.e. the encircled S’), the adjunct clause is not (properly) governed by the matrix verb. Thus, the subject position of the adjunct clauses is an A’-position. Fukui endorses that, in (10a) and (1 la, b), zibun can be bound by the matrix subject since it is the closest A’-binder, whereas in (10b), zibun cannot be bound by the matrix subject because the adjunct subject is intervening and hence the matrix subject is not the “closest” A’-binder. 2.1.2. Problems with Fukui's proposal: Split antecedence In the previous section, we saw that it is almost impossible to read “ zibun = the complement subject” when zibun is in the complement object position (see (lib)). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 Fukui assumes that this is because zibun is subject to the Disjoint Reference Rule. He believes that zibun has a ‘pronoun’-like property. His claim is based on an observation in which zibun-tati, ‘the plural form o f zibun,' seems to take the split antecedents, as in (15). (15) (Fukui’s (33a)) Johni-ga Maryj-ni [s Bill-ga zibun-tatii,j-o waruku itteiru] to NOM to NOM ACC badly saying COMP siraseta (koto) reported ‘Johni reported to Maryj that Bill was speaking ill of selfi-j.’ In (15), Fukui states, ‘the plural form of zibun, that is zibun-tati, has John and Mary as its antecedents. It has been said that pronouns have split antecedents, whereas anaphors do not (cf. Freidin, 1992) as exemplified in (16). (16c) shows that the pronoun their, the plural form of the pronoun him in (16a), can have split antecedents. As in (I6d), the anaphor themselves, the plural form of the anaphor himself in (16b), cannot have split antecedents. (16) a. Johni told Bill2 about himi/2 ./hisi/2 schedule. b. Johni told Bill2 about himselfj. c. Johni told Bill2 about theiri*2 schedule. d. *Johni told Bill2 about themselvesi.2. By showing that zibun-tati can have split antecedents, Fukui concludes that zibun is a pronoun, not an anaphor. If (15) and (16c, d) were compared, it seems that Fukui assumes zibun corresponds to the “ / mot” part of him self and zibun-tati to the “them” part of themselves. ■ ’. I ■ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 Cho (1990), however, has taken a position against Fukui and argues that the split antecedents of zibun cannot be attributed to pronominal properties of zibun, but to the properties of the plural morphine tati. Cho points out that the functions of tati involve (i) an interpretation o f‘x and other(s)’ and (ii) an interpretation of general plurality. The functions are illustrated in (17). (17) a. (an example of ‘x and others)’ case) Asita Johm-tati-ga L.A-ni kuru. Tomorrow John-PL-NOM L.A.-to come ‘Tomorrow John;-k will come to L.A.’ b. (an example of general plurality) Asita gakusei-tati-ga S.F.-ni iku. Tomorrow student-PL-NOM S.F.-to go ‘Tomorrow the students will go to S.F.’ In (17a), ‘ John-tatV refers to "John and somebody,’ for example ‘ John and BiW or ‘ John and Mike and Fred, whereas tati in (17b) refers to general plurality. Gakusei-tati’ can mean ‘student 1, student 2, and so on.’ Now, let us look at the interpretation o f ‘x and other(s)’of tati in relation to zibun. In Japanese, the number agreement between zibun and its antecedent is not obligatory. As shown in (18b), although the antecedent is a singular John, zibun-tati can refer to ‘ John and other(s).’ (18) (Cho’s (8a) and (8b), 1990) a. John-ga zibun-ga katta-to itta. John-NOM self-NOM won-COMP said ‘Johni said that he; w on.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 b. John-ga zibun-tati-ga katta-to itta. John-NOM self-PL-NOM won-COMP said ‘Johnj said that theyj^ w on.’ Thus, it is possible to assume that in (18b) zibun corresponds to John and tati to ‘other(s).’ The property o f tati, ‘x and other(s),’ is found in bound variable construal, too, as illustrated in (19). (19) a. Daremo-ga zibun-ga katta-to itta. everyone-NOM self-NOM won-COMP said ‘Everyone said that self w on.’ b. Daremo-ga zibun-tst\-gz. katta-to itta everyone-NOM self-PL-NOM won-COMP said ‘Everyone said that selves won.’ The intended bound reading for (19a) and (19b) can be paraphrased as in (20a) and (20b) respectively. (20) a. For every person x, x said x won. b. For every person x, x said x and other(s) won.5 In addition to the argument based on the properties of tati, we can argue against the claim that zibun-tati shows split antecedents in the following two ways. First, as the example in (21) illustrates, zibun is not plural-denoting. 5H. Hoji (personal communication, November, 1997) suggested to me that, in (19b), daremo can correspond to “all the people” and the sentence can be considered as analogous to “all the people said that they won.” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 (21) *[Johnj to Billj]-ga [cp zibun^j no seikaku ga kiraida]-to itta. and NOM self GEN personality hate COMP said ‘Johnj and Billj said that they do not like their^j personality.’ The sentence in (21), zibun cannot be interpreted as having both John and Bill as its antecedents. The antecedent o f zibun is either John or Bill. This ‘bound variable’ like reading is illustrated in (22). (22) x-ga [x-no seikaku ga kiraida]-to itta. ‘x said that x does not like x ’s personality.’ We thus take the unavailability o f plural denotation in zibun as evidence that the split antecedence reading is impossible for zibun. Second, the sentence like (23a) can be raised against Fukui’s claim that zibun- tati shows split antecedents. (23) a. *John-ga Maryj-ni [s Bill-ga zibunr o waruku itteiru] to NOM to NOM ACC badly saying COMP siraseta (koto) reported ‘John reported to Maryj that Bill was speaking ill of selfj.’ b. (Fukui’s (33a)) Johnj-ga Maryj-ni [s Bill-ga zibun-tatij^-o waruku itteiru] NOM to NOM ACC badly saying to siraseta (koto) COMP reported ‘Johnj reported to Maryj that Bill was speaking ill of self,,j.’ Under Fukui’s assumption, tati in zibun-tati is merely a plural morpheme. Therefore, in (23b), zibun-tati should merely connote selves. This and the fact that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 M ary cannot be the antecedent of zibun as illustrated in (23 a) suggest that Mary cannot be one of the antecedents of zibun-tati in (23b). Fukui’s claim that the antecedents of zibun-tati include M ary in (23b) is thus open to objection. In this section, we have seen that (i) zibun is singular-denoting; and (ii) x-tati has the interpretation of ‘x and other(s). ’ Based on these, it can be concluded that the claim of Fukui (1984) that zibun-tati shows split antecedents cannot be accepted . The alleged property of the split antecedents with zibun-tati might come from the plural morpheme tati, but not from the pronominal nature of zibun, contrary to Fukui’s claim. 2.1.3. (Jeda (1986): Zibun as a bound pronominal As discussed in section 1.1, Condition A deals with the distribution of anaphors (reciprocals and reflexives). In Japanese, it has been assumed that otagai is a reciprocal corresponding to English each other,6 and zibun is regarded as the most representative reflexive. Ueda (1986) takes notice of the behavioral differences of otagai and zibun and concludes that the difference can be accounted for if it is assumed that otagai is an anaphor, while zibun is not an anaphor, but a pronominal. (24a), (24b), and (24c) are the properties Ueda assumes zibun has, but otagai does not. 'Tfoji (1997a) argues against this widely-held view. His arguments will be introduced in section 2.3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 (24) a. The antecedent must be a subject. b. Zibun can be bound by a topic NP. c. Zibun can be discourse-bound.7 I doubt that zibun and otagai have such behavioral differences as in (24). Ueda, for example, states that a topic NP can be an antecedent of zibun when zibun is contained in the subject NP of the associated clause. In contrast, otagai cannot be bound by a topic NP, even if it is contained in the subject NP. According to Ueda, (25a), (25b), and (25c) illustrate his points. (25) a. (Ueda’s (17b)) Kaorui-wa [s[np[s zibun;-ga sukidatta] gakusei]-ga Tokyo-e kaette TOP self NOM liked student NOM to return simatta. PER ‘As for Kaoru, the student she liked had gone to back to Tokyo. b. (Ueda’s (18a)) *Daisukej-wa [s[Hiraoka-ga TOP NOM hukoo-ni sita] unhappy made ‘As for Daisuke, Hiraoka made c. (Ueda’s (19a)) *Karera,-wa [sotagaij -no musume -ga sin-de simatta] they TOP each other GEN daughter NOM die PER ‘As for them, each other’s daughters have died.’ 7 Oshima (1979) also discusses this phenomenon. [ n p zibunj-no mukasi-no koibito]-o self GEN previous lover ACC his ex-lover unhappy.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 Ueda claims that (25a) is grammatical since zibun is in the subject NP, whereas (25b) is ungrammatical since zibun is contained in the object NP. In contrast to Ueda’s judgment, I believe that (25b) is grammatical and some of my native informants agree with me. Those who judged (25b) to be marginal agree that the sentence in (25b’) is acceptable. (25b ’) Daisukej-wa [s[ yoku syatyoo -ga [N -p zibun;-no tuma]-ni TOP often the president NOM self GEN wife to denwa-o katete kuru koto]-ga ki ni natteiru. call fact NOM worry about ‘As for Daisuke, (he worries about the fact that) the president often calls his wife.’ Regarding (25c), Ueda assumes that it is ungrammatical even if otagai is contained in the subject NP since it cannot be bound by a topic NP. I do not believe that (25c) is terrible. My native informants agree with my judgment on (25c) and the sentence like (25c’). (2 5 c’) [John to Mary];-wa [s otagaij-no haiguusya-ga sin-de simatta] and TOP each other’s spouse NOM die PER koto o ato de sitta.). he fact ACC later found ‘As for John and Mary, (they later found out that) each other’s spouse have died.’ In addition, I do not agree with Ueda’s claim about discourse-binding (see 24c). That is, although both zibun and otagai can be bound across discourse, discourse- bound zibun is found only in the matrix subject position, whereas discourse-bound Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 otagai is not limited to the matrix subject position. (26a), (26b), and (26c) are Ueda’s examples. (26) a. (Ueda’s (20)) Speaker A: SpeakerB: Johni-ga dareka-o okutta nodesu ka? NOM someone sent Dec-Int ‘Did John send someone?’ lie, zibunj-ga kitan-desu. no, self NOM come-Dec-For ‘No, (lit.) self came.’ b. (Ueda’s (21) from Koster (1982, p. 5)) Speaker A: Dare-ga Johnj-o okutta nodesu ka? Who NOM ACC sent Dec-Int ‘Who sent John.?’ Speaker B: *Bill-ga zibuni-o okutta ndesu. NOM self ACC sent Dec ‘Bill sent (lit.) self.’ c. (Ueda’s (23)) [m > John to Mary]i-wa sengetsu wakareta. Sikasi, Bill-wa TOP last month part but TOP [s sore-ga [np otagaii -no gooi-no ue- de-no koto] datta] it NOM each other GEN agreement GEN based-on thing was koto-o sitte iru. COMP ACC know Prog Pres ‘John and Mary parted last month. But Bill knows that it was done by mutual agreement.’ I do not agree with Ueda’s judgment on (26b). In fact, I do not believe that discourse-bound zibun is found only in the matrix subject position. For example, in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 (26b’), although discourse-bound zibun is in the object position, the sentence is grammatical. (26b’) Speaker A: Johni-ga sugoku okotte iru. NOM terribly is angry with Doo sita no daroo. what happened wonder ‘John is in furious. I wonder what happened.’ Speaker B: Sensei-ga zibunj-o hihansitanda yo. teacher NOM self ACC criticized you know. ‘His teacher criticized him, you know .’ Further, in (26c’), even though zibun is not in the matrix subject position, the sentence is grammatical. (26c’) Johnj-wa sengetsu gakkoo-o yameta. . Sikasi, Bill-wa TOP last month school ACC quit but TOP [s sore-ga [zibun; -no ketui] datta] koto-o sitte iru. it NOM self GEN decision was COMP ACC know ‘John dropped school last month. But Bill knows that it was self s decision.’ Based on the examples in (25) and (26), Ueda believes that the Japanese otagai lacks the properties in (24) and behaves like the English anaphors, except for the differences that are due to the lack of AGR in Japanese. Thus, otagai can be postulated as an anaphor. As to zibun, its ‘peculiar’ characteristics can be accounted for if zibun is assumed to be a pronominal. Ueda states that his presumption is supported by the observation that zibun allows both the strict and the sloppy reading under the VP ellipsis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 As Hoji (1991, chap. 5, p. 3-4) reviews, in the standard analysis of the VP deletion, the sloppy and the strict reading are accounted for as illustrated in (27) through (34). The sentence like (27) allows the three interpretations as shown in (27’). (27)(Hoji’s (4)) Felix kissed his dog after Max did. (27’)a. Felix kissed Felix’s dog after Max kissed Max’s dog. b. Felix kissed Felix’s dog after Max kissed Felix's dog. c. Felix kissed John’s dog after Max kissed John’s dog. In (27’c), his is used pragmatically (i.e. referentially). That is, (27’c) is the interpretation in which his refers to some individual salient in the context of discourse. The interpretation in (27’a) and that in (76’b) are called the sloppy reading and the strict reading respectively. (28) illustrates the ‘derived VP rule (DVPR)’ proposed in Partee (1973). (28) (Hoji’s (5)) Felix (Xx (x kissed his dog)) after Max did. The differences in interpretation of the first conjunct of (27) (i.e.. (27’a), (27’b), and (27’c)) can be achieved by representing it as in (291), (2911), or (29 III). (2911) illustrates the rule of anaphora in Chomsky (1976, p. 202). (29) (Hoji’s (6)) Pronoun interpretation: I Xx (x kissed his=Fei« dog) II Xx (x kissed x’s dog) III Xx (x kissed his=z dog) (z = someone in the context of discourse) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 The operations in (30) illustrate that the interpreted X predicate is copied into the empty VP slot (i.e., Williams’s (1977) LF rule). (30) Copy of predicate: I after Max (did) Xx (x kissed his=FC i« dog) II after Max (did) Xx (x kissed x’s dog) III after Max (did) Xx (x kissed his=z dog) Accordingly, pronouns allow both the strict reading (c.f., (30) I) and the sloppy reading (c.f., (30) II). In contrast, anaphors are presumed to undergo obligatorily the rule II of (29). Thus, the lack of the strict reading in the sentences with anaphor is predicted as shown in (31) through (34). (31) John kicked himself after Bill did. (32) DVPR John (Xx (x kicked himself)) after Bill did. (33) Pronoun interpretation: II Xx (x kicked x) (34) Copy o f predicate: II after Bill (did) Xx (x kicked x) Now lets us look at Ueda’s example. (35) (Ueda’s (37b), 1986) John-ga [szibun-ga kat-te iru] inu-o nagu-ru to Bill-mo John-NOM self-NOM keep-Prog-Pres dog-ACC hit-Pres when Bill also soo sita so do-Past ‘When John hit the dog self kept, Bill did so, too.’ Ueda presumes that in (35) the second conjunct allows the interpretation that Bill hit Bill’s dog as well as the interpretation that Bill hit John’s dog. Accordingly, Ueda Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 assumes that zibun is a pronominal. Then, what kind o f pronominal is zibun? Ueda focuses on the two possible interpretations of a pronominal, i.e., bound variable interpretation and coreferential interpretation and argues that the difference in interpretation (in other words, [+/- bound] features) should be added to the classification of the categories of nominal expression based on the two features with binary values, i.e., [+/- anaphor] and [+/- pronominal], as illustrated in (36). Accordingly, the category of pronominal can be expressed as [-anaphor, +pronominal, +bound] or [-anaphora, +pronominal, -bound] (cfi, (37)). (36) Chomsky’s classification (1982) Anaphors: [+anaphor, -pronominal] Pronominals: [-anaphor, +pronominal] R-expressions: [-anaphor, -pronominal] (37) Ueda’s classification of pronominals a: [-anaphor, +pronominal, +bound] b: [-anaphor, +pronominal, -bound] In the literature, it is widely accepted that the Japanese overt pronouns, kare, kcmozyo, karera cannot be construed as a bound variable (Hoji, 1991, 1995; Kitagawa, 1981; Nakai, 1976, 1977; Nakayama, 1982; Saito, 1981; Saito and Hoji, 1983, among others). Hence, kare is the pronominal with the features in (37b), according to Ueda, Zibun, on the other hand, can be bound by a quantificational NP. Thus, Ueda classifies zibun as (37a). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 2.1.4. A problem with Ueda’s claim I: Soo-su vs. do so As illustrated in (35), repeated here as (38), Ueda (1986, p. 97) uses the soo su construction in testing the sloppy reading in Japanese assuming that the structure is the Japanese counterpart English do so. (38) (Ueda’s (37b)) John-ga [s zibun-ga kat-te iru] inu-o nagu-ru to Bill-mo John-NOM self-NOM keep-Prog-Pres dog-ACC hit-Pres when Bill also soo sita so do-Past ‘When John hit the dog self kept, Bill did so, too.’ Cho (1990)8 and Hoji (1991), however, doubt that the Japanese soo-su is really a counterpart of the English do so construction Let us consider the sentences in (39) for example. (39) a. (Hoji’s (14), 1991, chap. 5, p. 5) Johnrga kare;-no ronbun-o Ll-ni okuru-to NOM he GEN article ACC LI to sent-when Bill-mo suguni soo sita. also immediately so did ‘When Johnisent hisi article to LI, Bill immediately “did so” too.’ b. (Hoji’s (16), 1991, ch. 5, p. 6) Johnrga aituj -no ronbun-o Ll-ni okuru-to NOM that guy GEN article ACC LI to sent-when Bill-mo suguni soo sita. also immediately so did ‘When John; sent that guy’s; article to LI, Bill immediately “did so” too.’ ®Cho notes that his claim is taken from H. Hoji through personal communication Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 26 In Japanese kare and the a series of the Japanese demonstratives are considered to resist bound variable construal (Hoji, 1991).9 Thus, (39a, b) should not allow sloppy reading, given the assumption that sloppy reading is based on bound variable anaphora as Sag (1976), Williams (1977) and others suggest. However, sloppy reading appears to be available in (39a, b). This might imply that (i) si-ta is not the same auxiliary verb as English did’ and (ii) soo si-ta does not correspond to English did so. The observation thus casts doubt over the claim that the Japanese soo su is analogous to the English do so. In fact, Hoji (1991) argues that the meaning of soo su should be translated as ‘do that way’ but not ‘do so.’ Recall as noted in footnote 10, repeated here as (40), soo means ‘in that way.’ (40) koo ‘in this way’ soo ‘in that way’ aa ‘in that way’ doo ‘in what way’ The translation o f (39a, b) should be like (41). 9 In literature, it is customary to refer kare as ‘pronoun’ and translate ‘he’ although its status as 'pronoun' is not clear yet. Japanese employs very productive deictic system called the ko.so,a,do paradigm (Sakuma, 1936). Nominals: koxe ‘this (thing)’ (close to the speaker); sore ‘that (thing) (away from the speaker and close to the listener); ore ‘that (thing) over there’ (away from both the speaker and the listenner); dore ‘which (thing)’ (among 3 or more). Prenominals: kono ‘this’; sono ‘that’; ono ‘that over there'; dono ‘which’. Place: koko ‘here’; soko ‘there’; a(so)ko ‘that place over there’; doko ‘where’. Manner koo ‘in this wav’; soo ‘in that way’; «a ‘in that way’; doo ‘in what way’. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 (41) (Hoji’s (26), 1991, chap. 5, p. 8) When John; sent hisj (that guy’S i) paper to LI, Bill immediately did the same. Accordingly, Hoji argues, what appears to be the sloppy reading for (39a, b) might not involve bound variable construal at all. Hoji, in fact, suggests that soo su can be considered as a case of “deep anaphora,” in the sense of Hankamar and Sag (1976), rather than a case of “surface anaphora.” In this section, Hoji’s (1991) argument against Ueda’s claim that zibun allowing both strict and sloppy reading in VP deletion is introduced. Namely, Ueda’s claim may be due to incorrect selection o f counterpart form of English do so, not to a pronominal property of zibun. 2.1.5. A problem with Ueda’s claim II: Disjoint reference As Ueda himself admits, an additional problem for Ueda’s claim is that zibun sometimes allows local-binding, as illustrated in (42), a violation of Condition B, if zibun is a pronominal. (42) (Ueda’s (59a), (60a), and (61a, b)) a. Johni-wa zibunj-o aisi-te iru. TOP ACC love-Prog-Pre ‘John; loves self.’ b. Johni-wa zibun;-o seme-ta. TOP ACC blame-Past ‘John blamed self.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 c. *Hirosii-wa zibunj-o korosi-ta TOP ACC kill-Past ‘Hirosi; killed self).’ d. *Johrii-wa zibuni-o nagut-ta TOP ACC hit-Past ‘John; hit self;. ’ Judging the sentences in (42) to be marginal or marked, Ueda states that “ zibun is exempt from the Binding Theory (B) when the verb of the clause in which zibun occurs represents abstract activity” (p. 107). Therefore, zibun occurring with such verbs as “aisite iru,” “nikunde iru,” like (42a) and (42b), disobey Condition B while zibun occurring with such verbs as “korosita,” “ketta” is subject to Condition B, as in (42c) and (42d).1 0 Ueda supports this argument by pointing out that “if these sentences (such as (42a) and (42b)) are embedded in another sentence, the matrix subject is always a preferred antecedent of zibun” (p. 106), as shown in (43). Ueda concludes that zibun favors LD-binding over local-binding. (43) (Ueda’s (62)) a. Billi-wa [Johnj-ga zibun ^ -o nikun-de i-ru] to omot-ta TOP NOM ACC hate-Prog-Pres COMP think-Past ‘Bill; thought that Johnj hated hinvj.’ b. Billi-wa [Johnj-ga zibun ^j-o bengosu-ru daroo]-to omot-ta TOP NOM ACC defend-Pres will COMP think-Past ‘Billi thought that Johnj would defend him ^.’ 1 0 Ueda notes that (42c) is grammatical if it is interpreted as representing the abstract activity of killing one's own will. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 I would like to argue against Ueda’s claim that, in (43a) and (43b), zibun prefers the long-distance antecedent Bill to the local antecedent John. Consider the sentences in (44). (44) a. Mary,-wa [Johnj-ga zibunj/j-o nikun-de i-ru] -to omotte TOP NOM ACC hate-Prog-Pres COMP think naita cry-Past ‘Mary; cried thinking that Johnj hated selfLj.’ b. Maryi-wa [Johnj-ga zibunj/j-o seme-te-iru] -to TOP NOM self ACC blame-Prog-Pres COMP sinpaisi-ta worry-Prog ‘Maryj worried that Johnj blames selfy.’ I find the sentences in (44) ambiguous, and all of my six native informants share the opinion: in (44a, b), the sentences with abstract verbs are embedded, but zibun still can refer to the embedded subject John as easily as the matrix Mary. Thus, both Mary and John can be the antecedent of zibun in these sentences. 2.2. Zibun as anaphor 2.2.1. Aikawa (1993): Zibun as non-reflexivizer Aikawa (1993), focusing on the difference in local binding between zibun and zibunzisin, presumes that zibun cannot be, but zibunzisin can be locally bound when Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 its antecedent is a QP or a WH-word.1 1 She believes that this suggests that zibun is not a true anaphor but zibunzisin is because “when an anaphor/pronoun occurs with a QP/WH antecedent, the only way for the anaphor/pronoun to be associated with the QP/WH antecedent is by binding” (Aikawa, 1993, p. 42). (45) ( Aikawa’s (le) & (2e), chap. 2) a. John-ga zibun-o tunetta. NOM ACC pinch-Past ‘Johni pinched zibun.’ b. John-ga zibunzisin-o tunetta. ‘John pinched zibunzisin.’ (46) (Aikawa’s (3a) & (3b), chap. 2) a. ?* Dare/Dareka-ga zibun-o tunetta (no). Who/Someone NOM ACC pinch-ed ‘Who pinched zibun?/Someone pinched zibun.’ b. Dare/Dareka-ga zibunzisin-o tunetta (no). ‘Who pinched zibunzisin?/Someone pinched zibunzisin.’ Aikawa believes that, in (45b) and (46b), the type of antecedent (NP or QP) does not affect the interpretation of zibunzisin. Therefore, it is construed as a bound variable. Zibun, on the other hand, is associated with an NP through coreference in (45a). A coreference reading for zibun is not available in (46a). Hence, it is ungrammatical. UI do not agree with Aikawa’s judgment on (46). I will argue against Aikawa’s claim in the following section. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 Aikawa attempts to explain the alleged behavioral difference between zibun and zibunzisin, using the notion of reflexivity proposed by Reinhart and Reuland (1993): zibunzisin is a reflexivizer, whereas zibun is not.1 2 Aikawa argues that the semantic function of a reflexivizer is that “two arguments of a predicate must be identical variables” (p. 61). In other words, “one of the co-arguments of a predicate (an anaphor) must be construed as a bound variable of the other (the antecedent)” (p. 61). Hence, when it occurs in the domain o f reflexivityfor a predicate, a reflexivizer must be construed as a bound variable, whereas a non-reflexivizer cannot be.1 3 (47) summarizes the above discussion. (47) (Aikawa’s (40), chap. 2) a. Zibunzisin is a reflexivizer. Therefore, when it occurs in the domain of reflexivity for a predicate P, it must be construed as a bound variable. b. Zibun is not a reflexivizer. Therefore, when it occurs in the domain of reflexivity for a predicate P, it cannot be a bound variable. (When zibun occurs outside the domain of reflexivity of a P, however, it can be construed as a bound variable.) 1 2 Aikawa defines reflexivizer as an "indicator” that evinces that two arguments o f a predicate to be identical variables. That is, when a predicate has an anaphor (a reflexivizer) as one of its co-arguments, the co-arguments should be interpreted as identical. This, in turn, means, according to Aikawa, “one o f the co-arguments o f a predicate (an anaphor) must be construed as a bound variable o f the other (the antecedent)” (p. 61). 1 3 Aikawa notes that domain o f reflexivity for a reflexive predicate is co-aiguments o f a reflexive predicate. Reinhart & Reuland (1993) defines reflexive predicate as follows; “a predicate is reflexive iff two o f its arguments are coindexed.” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 Zibun, which Aikawa assumes to be a non-reflexivizer, has two properties: subject orientation and long-distance binding. Aikawa explains the first by proposing that zibun must receive phi-features and referential index from a subject. Since zibun lacks phi-features, in order to be interpreted, it has to be bound to the first accessible X°-eIement that can provide phi-features. That is, zibun is bound to AGR through coindexation at LF (see (48a) below). Through spec-head agreement between the subject and its associated AGR the phi-features of the subject are assigned to AGR When the subject undergoes quantifier raising, the index of the subject can be assigned to A.GR/zibun (see (48b)).1 4 (48) a. (Aikawa’s (23), chap. 3) AgrP / \ / \ VP AGR / \ ... zibuni... “ Aikawa claims that the subject needs to undergo quantifier raising so that AGR can share the same referential index with the subject. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 b. (Aikawa (24), chap. 3)1 5 AgrP / \ NP; AgrP / \ t; AGRi [+ a , +P, +x~features] [+ c l, +P, +x~features] VP / \ ..zibunt... Aikawa believes that what makes the LD binding of zibun possible is the “fact” that the nature of AGRs in Japanese are anaphoric.1 6 It makes the transfer of an index o f one AGR to another become possible. Therefore, the index of zibun1 s remote binder can be transferred from the AGR coindexed with the binder to the AGR closest to zibun. Further, Aikawa assumes, since AGRs in Japanese are blind to phi-features, there is nothing that blocks the LD-binding of zibun. (49) illustrates Aikawa’s LF-analysis of the LD-binding of zibun. According to Aikawa, it shows that if there exist more than one zibun in the deepest clause, all o f them share the same antecedent. 1 5 '[+ a, +p, -^-features]' means phi-features: [+person, +number, +gender] 1 6 Aikawa endorses the proposal by Borer ( 1989) and Progovac (1991): (i) there are two types of AGR: anaphoric AGR and referential AGR; (ii) language without overt agreement have anaphoric AGR; and (iii) anaphoric AGR can be bound to a higher AGR. She notes “Japanese lacks overt agreement, hence, AGRs in Japanese are 'anaphoric’” (p. 162). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 (49) (Aikawa’s (13), chap. 4) AgrP / \ NPi " AgrP / \ ti AGRjn / \ AGRj2 / \ AGRil / \ ...zibuni... (Where / is a bound variable index.) Zibunzisin, a reflexivizer, on the other hand, must be bound locally. Aikawa believes that this property can be explained if it is assumed that the zisin-part raises to the head of a predicate (V°) at LF so that it can function as a reflexivizer, while the zibun-part is bound to the first accessible AGR by coindexation as bare zibun does. (50a) illustrates the LF-movement in zibunzisin: the zibun part of zibunzisin moves to V° so that it can reflexive mark the predicate. The zibun part associates with AGR through coindexation. (50b) shows the LF of zibunzisin binding. Analogues to zibun binding explained in the earlier part of this section, the zibun part o f zibunzisin requires the quantifier raising of the subject so that it can be bound to the first accessible AGR and it can function as a reflexivizer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 (50) a. (Aikawa’s (39), chap. 3) AgrP / \ AGRj VP / \ / \ ziburii-tj V°-zisirij b. (Aikawa’s (40), chap. 3) AgrP / \ NPi AgrP / \ ti AGRi VP / \ / \ zibun-lj W°-zisin} 2.2.2. Problems with Aikawa’s proposal I do not agree with Aikawa’s proposal that zibun must be distinguished from zibunzisin. Showing the behavioral differences of zibun and zibunzisin in the sentences such as (46), repeated here as (51), she claims that zibun cannot be locally Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 bound, but zibunzisin can be. This, in turn, indicates, according to Aikawa, that zibun is not a true anaphor, whereas zibunzisin is. (51) (Aikawa’s (3a) & (3b), chap. 2) a. ?* Dare/Dareka-ga zibun-o tunetta (no), who/someone NOM ACC pinch-ed ‘Who pinched zibun?/Someone pinched zibun.’ b. Dare/Dareka-ga zibunzisin-o tunetta(no). ‘Who pinched zibunzisin?/Someone pinched zibunzisin.’ I, as well as my native informants, doubt that (51a) is terrible. Further, the sentence such as (52), for instance, is also grammatical, and my native-speaker informants share the opinion. This indicates that zibun can be locally bound by the Q-NP. (52) Dare/Dareka-ga zibun-ni iyake-ga sasiteiru (no). w ho/som eone NOM ACC be disgust-ed ‘Who is disgusted with zibun?/Someone is disgusted with zibun.’ Aikawa’s analysis of subject-orientation of zibun also faces a problem. As introduced in section 2.2.1, Aikawa proposes that zibun, lacking phi-features, must receive them from a subject so that it can be interpreted. Aikawa’s claim is that through spec-head agreement between the subject and its associated AGR, the phi- features of the subject are assigned to AGR. Zibun receives the phi-features by binding to AGR through coindexation at LF. However, it has been proposed that Japanese does not have any formal agreement features (Fukui (1986); it is implied also in Kuroda (1988), as pointed out by Hoji (1996c)). Aikawa’s proposal becomes pointless if, in fact, Japanese does not have any formal agreement features. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 2.2.3. Katada (1991): Zibun as an operator anaphor Katada (1991) assumes that zibun is a long-distance anaphor with subject- orientation, whereas zibunzisin is a local anaphor that is subject-oriented.1 7 These properties are, according to her, illustrated in (53). (53) (Katada’s (3)) a. Johni-ga [Billj-ga Miket-ni zibun -no koto-o NOM NOM DAT self GEN things ACC hanasita to] itta. told that said b. Johnj-ga [Billj-ga Miket-ni zibunzisin r^-k -no koto-o NOM NOM DAT self GEN things ACC hanasita to] itta. told that said According to Katada, this difference of the binding behavior of zibun and zibunzisin as well as their property of subject-orientation can be accounted for in terms of LF- raising via VP-adjunction and the Empty Category Principle (ECP) (Chomsky, 1981). The core of Katada’s proposal is that zibun is an operator anaphor.1 8 The empirical justification of her proposal is the fact that zibun is unmarked for phi- features. Katada writes: 1 7 I do not agree that zibunzisin is a local anaphor. In (53b), for example, I as well as my native informants believe that the antecedent of zibunzisin can be the matrix subject John. I will argue against Katada’s claim in section 2.2.4. I8 I will argue against this claim in section 2.2.4. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 The(se) lexically unmarked agreement features of zibun imply that it possesses a “semantic range.”1 9 It is this property that distinguishes operator anaphors from non-operator anaphors. (p. 291 -292) Based on the above assumption and treatment o f operators in general, Katada induces that zibun, an operator anaphor, may undergo LF-raising to an Apposition. The internal structures of zibun and zibunzisin that Katada assumes are as in (54a) and (54b) respectively. (54) a. NP b. NP;. 1 / \ N’ 1 Spec i N’ 1 1 N i 1 n p 2 1 1 N i 1 zibun 1 zibun 1 zisin (55a) and (55b) show the movement of zibun and zibunzisin respectively. (55) a. (=Katada’s (17)) b. ( =Katada’s (20)) NP[+op] NPl[-op] 1 / \ N’ Spec N’ zibunj 1 1 1 [+op] N zibun; NP2 N t 1 1 [+op] ! 1 1 t; T t; zisin long distance local 1 9 However, what “semantic range” exactly means is not clearly stated. Katada assumes that other operators such as quantifiers and null operators also possess "semantic range.” Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 (56) shows their LF-raising resulted in its requisite LF representation respectively. (56) a. (Katada’s (18)) .. NPi-ga [vpzibun[vp...NP2 -ni...[cp*...ti-Case...]]] I l+lex-gvnd]2 0 (where CP* stands for zero or more occurrences of clauses) b. (Katada’s (21) .. NP-ga[vp2 [cpi[vpiNP-ga [vpi zibun;[vpi...[t;-zisin[...]]]]] I _______I [+ant-gvnd]2 1 According to Katada, (56a) illustrates that the domain of VP adjunction for zibun can be interpreted in multiple VP-adjunction sites since the domain of VP adjunction for zibun is unlimited. Now let us look at (57). (57) a. Johni-ga [vp 2 zibun; [vp[[Billj-ga[vpi Miket-ni tj-Case...]]]]] b. Johnrga [vre [cpO Bill;-ga [vpi zibun; [vpi Mikek-ni t;-Case...]]]]] Katada states that (57) indicates the LF movement of zibun in (53a) (repeated here). Namely, (57a) illustrates the LF of (53a) with the coindexation of John and zibun, (57b) with the coindexation of B ill and zibun. 20[+lex-gvnd] reads "the trace is lexically governed.” 21[+ant-gvnd] reads 'the trace is antecedent-governed by zibun Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 (53) (Katada’s (3)) a. Johnj-ga [Billj-ga Miket-ni zibun -no koto-o NOM NOM DAT se lf GEN things ACC hanasita to] itta. told that said Katada assumes that the properties of zibun, the LD-binding and the subject- orientation, can be accounted for as follows. First, as to the LD-binding of zibun, in (57a), it is possible since there is no ECP violation. This is because she presumes that in Japanese case markers are lexical governors (following Saito (1985)), and that Case marker stranding is allowed at LF. In other words, the trace of zibun in (57a) can be lexically-governed by the nominative Case marker gar, satisfying the Empty Category Principle. These conditions, together with the assumption in (56a), make the LD-binding of zibun possible. Second, the subject-orientation of zibun can be accounted for in terms of c-command relationship: the subject NPs John and B ill c-command zibun in its VP-adjunction sites in (57b), whereas the non subject Mike does not. Lets us look at (58) now. (58) a. * Johnj-ga [vp2 zibun,[\T>[[Billj-ga[vpi Miket-ni tj-zisin...]]]]] I __________________ t * antecedent governed b. Johni-ga [vp2 [cp[ip Billj-ga [vpizibuni[vpt Miket-ni tj-zisin...]]]]] I ____________T O K antecedent governed Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 (58) indicates the LF movement of zibun in (53b) (repeated here). Namely, (58a) illustrates the LF with the coindexation of John and zibun, (58b) the coindexation of B ill and zibun. (53) (Katada’s (3)) b. Johni-ga [Billj-ga Miket-ni zibunzisin?*i^k -no koto-c NOM NOM DAT self GEN things ACC hanasita to] itta. told that said Katada assumes that only zibun part of zibunzisin raises at LF. The trace of zibun must be antecedent-governed, since it cannot be lexically governed (because there is no Case marker). Hence, the ungovemed zibun cannot move father than antecedent government allows. (56) is not possible since zibun adjoins to a higher VP, crossing the first VP. In contrast, in (56b), zibun adjoins to the first VP. Thus, the trace can be antecedent-governed. 2.2.4. Problems with Katada’s analysis As introduced in section 2.2.3, Katada, paying attention to locality and subject orientation, claims that zibun is a long-distance reflexive with subject orientation, whereas zibunzisin is a local reflexive with subject orientation.2 2 I doubt the — According to Katada (1991), Japanese has three distinct forms of reflexives: zibun, zibunzisin, and karezisin ‘ he-self (p. 288). As for locality and subject orientation, karezisin is a local reflexive with no particular orientation (p. 290). The fuller study of karezisin lies outside the scope of this thesis. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 necessity of this classification. First, let us look at the sentences in (59) and (60) which show the possibility of non-subject binding. (59) Johni-ga Maryj-ni [zibun\ir no ten]-o osienakatta (koto) NOM DAT self GEN score ACC not told PAST (fact) ‘(the fact that) John did not tell Mary selfi/j’s score.’ (60) John-ga Ann-ni [Billi-ga Maryj-ni [zibunj/j/zibunzisini/j-no tameni jikan-o GEN for time-ACC tukau-bekida]-to itta]-to tsugeta. use should that said that reported ‘John reported Ann that Bill; told Maryj that (Mary) should use time for selfij.’ In these examples, zibun and zibunzisin can refer to the non-subject Mary. Katada’s analysis of zibun as operator anaphor completely excludes the possibility of non-subject orientation of zibun!zibunzisin. Second, according to Katada, zibunzisin is a local-anaphor. However, in sentences like (61), zibunzisin, like zibun, can be bound by the matrix subject John. (61) Johnj-ga [Billj-ga Maryk -ni [[zibunzisin^-t]-no koto]-o hanasita]-to things-ACC told that omotte okotte iru. think angry ‘John; is angry thinking that Billj told Mary about selfy.’ The central to Katada’s proposal is that zibun is an operator anaphor that undergoes LF-raising to A’-position. I doubt the dichotomy, i.e. operator anaphor (e.g., zibun) versus nonoperator anaphor (e.g., karezisin), because the distinction is on the basis of their antecedent’s distinctive agreement features. That is, although Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 the antecedent of zibun can be any person, number, and gender, that of karezisin should be 3rd person, singular, and masculine. Therefore, she concludes that “ karezisin is inherently (lexically) marked for these agreement features, whereas zibun is not” (Katada, 1991, p. 291). According to Katada, zibun, unmarked for agreement features, possesses a ‘semantic range’ like other well-known operators, and hence it is an operator.2 3 I would like to point out that it has been proposed that Japanese does not have any formal agreement features (Fukui (1986); it is also implied in Kuroda (1988), as pointed out by Hoji (1996c)). If that is indeed the case, I wonder how Katada supports her claim that zibun is an operator. Hoji (1990) and Aikawa (1993) point out another problem with Katada’s analysis, which is repeated here. First, let us look at the sentence in (62). (62) (Katada’s (44)) *John;-ga [zibuni-ga kare;-no hahaoya-o semeta to] itta. NOM NOM he GEN mother-ACC blamed that said ‘John; said that selfi blamed hisi mother.’ Katada states that (62) is ungrammatical because zibun, as an operator, is in an A’- position, and, therefore, kare is A’-bound by zibun}* However, Katada’s analysis does not explain the sentence like (63b). 23As stated in footnote 19, what she means by a ‘semantic range’ is not clear. Aikawa (1993) also points out this problem. 24It has been assumed that the so-called Japanese pronoun kare cannot be construed as a bound variable anaphora (Hoji (1990, 1997b); Kitagawa (1981); Nakai (1976), among others). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 (63) a. (Hoji’s (140a), 1990, chap. 3, p. 47) Daremo karemoi-ga soitu;-no ronbun-o mottekita. everyone-NOM the guy-GEN paper-ACC brought 'Everyone; brought the guy;’s paper.’ b. (Aikawa’s (9b), originally from Abe’s (9), 1991) ?*Daremo karemo;-ga zibun;-ga soitu;-no hahaoya-o semeta] to] everyone-NOM self-NOM the guy-GEN mother-ACC blamed that itta. said ‘Everyone; said that self blamed the guy;’s mother.’ As illustrated in (63 a), the Japanese demonstrative soitsu is able to be construed as a bound variable. Let us now look at (63b). It should be grammatical because its structure is parallel to that o f (62), i.e. zibun is in an A’-position, and as shown above soitsu can be A’-bound. However, it is not the case. 2.3. Otagai: Hoji (1997a) In section 2.1 and 2 .2, we saw that none of the previous proposals can explain the true nature of zibun and zibunzisin. If we turn our eyes to the other “so-called” local anaphor in Japanese, otagai (‘each other’), we realize that it has been recently thrown new light on by Hoji (1997a). Since I believe that zibun and zibunzisin should be analyzed along the lines of Hoji’s (1997a) analysis of otagai, in this sub section I would like to introduce his proposals and findings. Otagai has been treated as a reciprocal anaphor, and therefore, a local anaphor in Japanese. Hoji, however, argues that otagai is not a (local) anaphor. Hoji Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 provides empirical evidence that supports his claim that otagai is not a (local) anaphor, as summarized in (64). The sentences in (65), (66), (67), and (68) are the examples of (64a), (64b), (64c), and (64d) respectively. (64) a. The antecedent of otagai need not be in local domain of it. (cf. (65)) b. Otagai need not be bound by its antecedent if coreference is at stake, (cf. (66)“ ) c. Otagai allows split antecedent.(cf. (67)) d. When the relevant referential association is that of bound variable anaphora, Weak Crossover (WCO) effects are observed, (cf. (68)) (65) (Hoji’s (7a); exemplifies that the antecedent o f otagai need not be in local domain of it)2 6 [John to Bill]i-wa [cp Mary-ga [prox otagai]-ni horeteiru to] [John and Bill]-TOP [c p Mary-NOM [otagai-DAT is-in-love that] omoikonde-i-ta. believed ‘[each of John and Bill] believed that Mary was in love with the other.’ ‘[each of John and Bill]t believed that Mary was in love with himi.’ (66) (Hoji’s (8a); examples that otagai need not be bound by its antecedent if coreference is at stake) [\prox otagai]-no koibito]-ga [John to Bill] i-o yuuwaku-sita otagai-GEN lover-NOM [John and Bill]-ACC seduced (to yuu uwasa-ga matizyuu-no wadai-ni natte ita) ‘(The rumor that) each otheri’s lover seduced [John and Bill]i (had become a hot topic of the town.)’ 2 5 Kuno and Kim (1994) also points this out 2 6 The structure [np \pro ...] will be discussed later in this section. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 (67) (exemplifies that otagai allows split antecedent) a. (Hoji’s (9a)) Ieyasui-wa Nobunaga2-ni [Singen-ga sin-e-ba \prox ,2 otagai]-no Ieyasu-TOP Nobunaga-DAT [Singen-NOM d ie-if otagai-GEN ryoodo-ga sibaraku-wa antai-da to] tuge-ta territory-NOM for-a-while safe-be that] told ‘Ieyasui told Nobunaga2 that, if Singen dies, theiri-2 territories will be safe for a while.’ b. (Hoji’s (10a)) [subete-no Kyuusyuu-no daim yoojrga [Sikoku-no dokoka- all-GEN Kyuusyuu-GEN war-lord-NOM Sikoku-GEN someplace- no daimyoo]2-ni [Singen-ga sin-e-ba [pro\-2 otagai]-no GEN war-lord-DAT [Singen-NOM die-if otagai-GEN ryoodo-ga sibaraku w a antai-da to] tuge-ta (koto) territory-NOM for-a-while safe-be that told (fact) ‘[every feudal king in Kyuusyuu]i told [a feudal king of some place in Sikoku]2 that, if Singen dies, theiri-2 (respective) territories will be safe for a while.’ (68) (exemplifies that when the relevant referential association is that o f bound variable anaphora, Weak Crossover (WCO) effects are observed) a. (Hoji’s (11)) (Watasi-wa) [kanari-no kazu-no nihonzin huuhuji-ga [prox otagai]- (I-TOP) a good number-GEN Japanese couple-NOM otagai- no (katsute no) onsi-o batoo suru (no-o mita) GEN(former) teacher-ACC harshly criticize (COMP-ACC saw) ‘(I saw) [a good number of Japanese couple] j harshly criticize theiri (former) teachers].’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. b. (Hoji’s (13)) *(W atasi-wa) [\pro\ otagai]-no (katute no) onsi]-ga [kanari-no (I-TOP) -GEN (former) teacher-NOM a good kazu-no nihonzin huuhu]i-o batoo suru (no-o mita) number-GEN Japanese couple-ACC harshly criticize (COMP-ACC saw) ‘(I saw) theiri (former) teachers harshly criticize [a good number o f Japanese couples] i.’ Hoji proposes that the internal structure o f otagai is in fact [ n p pro [N otagai]]. He directs his attention to the absence of Condition B effects in sentences with otagai, such as (69), when bound variable anaphora is concerned.2 7 2 7 In Hoji (1995), he demonstrates that Principle B effects are not observed in Japanese when coreference is at stake (as illustrated in (i)), whereas, when bound variable anaphora is at stake, Principle B effects are observed (as illustrated in (ii». (i) (Hoji's(21), 1995) (Tovota-ga) [Arizona koozyoo], -ga soko,-o suisensita. (Toyota-NOM) Arizona factory-NOM it-ACC recommended. '(It was Toyota that) [(its) Arizona factory], recommended it,. (ii) (Hoji’s (20a, b), 1995) a. *(Toyota-ga) [Arizona koozyoo-sae], -ga soko,-o suisensita. (Toyota-NOM) [Arizona factory-even]-NOM it-ACC recommended. '(It was Toyota that) [even (its) Arizona factory], recommended it, b. (Toyota-ga) [Arizona koozyoo-sae], -ga [sokoi-no sitauke]-o (Toyota-NOM) [Arizona factory-even]-NOM [it-GEN subsidiary]-ACC suisensita. recommended '(It was Toyota that) [even (its) Arizona factory], recommended it,s subsidiary.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 (69) (Hoji’s (28b)) [kanari-no kazu-no huuhu]i-ga (hissi-ni-natte) [otagai] i-o [a good number-GEN couple]-NOM (very hard) -ACC urikonde ita (no w a sensyuu-no kaigi-de da) was promoting (COMP TOP last week-GEN meeting-at be) ‘(It was at the meeting last week that) [each of [a good number of couples]]i was promoting otagaii with utmost enthusiasm.’ (70) (Hoji’s (30a, b, c)) For a good number o f couples, it is true o f each o f those couples that a. the husband x and the wife y were promoting x and y. b. the husband x was promoting the w ife y , and y was promoting x. c. the husband x was promoting x, and the w ife^ was promoting y. In (69), the distributive readings (i.e., (70)) are possible even though the relevant binding is local. This contradicts with the previous findings that Japanese observes Condition B effects when bound variable anaphora is under consideration as indicated in (71) and (72). (71) (Hoji’s (26a)) •[Toyota to Nissan]rga (hissi-ni-natte) sokoi-o urikondeita [Toyota and Nissan]-NOM (very hard) it-ACC was promoting (no wa sensyuu-no kaigi-deda) (COMP TOP last week-GEN meeting-at be) ‘(It was at the meeting last week that) [each of Toyota and Nissan] i was promoting iti with utmost enthusiasm.’ (72) (Hoji’s (27a)) [Toyota to Nissan]r ga (hissi-ni-natte) sokoi-no kogaisya-o [Toyota and Nissan]-NOM (very hard) it-GEN subsidiary-ACC urikondeita (no wa sensyuu-no kaigi-de da) was promoting (COMP TOP last week-GEN meeting-at be) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 ‘(It was at the meeting last week that) [each of Toyota and Nissan]i was promoting iti’s subsidiary with utmost enthusiasm.’ Hoji argues that it is due to the structure ([nppro [...]]), the distributive readings (i.e., (70a, b, c)) are available in (69). In other words, what is bound by the Q-NP is not otagai, but pro in [nvpro [N otagai]]. Therefore, Condition B is not violated in (69) because the binding is not local. 3. Properties of zibun and zibunzisin In the following sections, I will present the empirical materials that support my claim that zibun and zibunzisin are not anaphors. In section 3.1,1 will show that zibun and zibunzisin do not have the properties of anaphors, i.e. anti-locality will be illustrated in section 3.1.1 and c-commanding restriction 3.1.2. Section 3.2 concerns additional properties of zibun and zibunzisin: 3.2.1 is concerned with weak crossover effects in bound variable construal, and section 3.2.2 illustrates that zibun and zibunzisin do not allow split antecedents. 3.1. Zibun and zibunzisin are not anaphors 3.1.1. The absence of locality The antecedent of zibun and zibunzisin need not be in their local domain, as illustrated in (73) through (75). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (73) Johni-ga [Mary2 -ga zibuni/ 2/zibunzisin 1/ 2 - 0 kiraidato] Johni NOM Mary2 NOM selfi/2 ACC hate COMP omotte naha (koto ) thought cried (fact) ‘(the fact that) Johni thought that Mary2 hates selfi,2 and cried’ (74) (from Freidin (46), 1992, chap. 7, modified) Johnrga [Mary2 -ga zibuni/2 / zibunzisini;2 ni tuite kaita kiji]- Johni NOM [Mary2 NOM selfi/2 about wrote article]- ga matigaidatta to hapyoosita. NOM false COMP announced. ‘Johni announced that the article Mary2 wrote about selfia was false.’ (75) Johni-sae [naze Bill2 -ga zibunt.2/zibunzisini(2 -o suisensita no ka] Johni even why Bill2 NOM selft,2 ACC recommended Q wakaranai (koto) not know (fact) ‘(The fact that) even Johni does not know why Biil2 recommended self,2.’ 3.1.2. The c-command restriction That zibun and zibunzisin need not be c-commanded by their antecedent when coreference is at stake is illustrated in (76) though (77).2 8 2 8 The examples in (76) is based on McCawley (1976). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 (76) [[ziburti/zibunzisini-no kobito]-ga Mike to sitasikusite iru koto]- selfi GEN lover NOM with be friends fact ga Johni-o otikomaseteiru. NOM Johni ACC distress ‘The fact that selfi’s lover is friends with Mike distresses Johni.’ (77) [Sikatanasini Johni-ga sono sigoto-o hikiuketa koto]-ga (ato de) unwillingly Johni NOM that work-ACC took on fact NOM (later) zibun\/zibunzisin t - no tatiba-ga warukunaru genin ni natta. selfi GEN position-NOM worsen cause became ‘The fact that Johni took on the work unwillingly put self! in an awkward position.’ In such sentences like (78a) and (78b), there is no overt antecedent that c- commands zibun or zibunzisin (78) a. [zibun ga zibun rasiku ikirareru koto]- ga taisetuda. self NOM self like Iive-can COMP NOM important ‘It is important that self can become self.’ b. [zibunzisin ga jinsei ni manzoku dekiru koto]-ga jyuuyooda. self NOM life satisfy can COMP important ‘It is important that self can be satisfied with life.’ ^The relevance of sentences like (78a) to the present discussion has been brought to my attention by A. Ueyama (personal communication, April, 1997), who attributed it to H. Hoji. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.2. Other properties of zibun and zibunzisin 52 3.2.1. Weak Crossover Effects When the bound variable construal is considered, typical weak crossover effects are observed in pronoun binding in English. (79) a. Hisi mother likes Johni. b. *Hisi mother likes everyonei. That zibun and zibunzisin also show this weak crossover effects is illustrated in (80) and (81). (80) Gakutyoo-ga [subete-no kyoju]t-ni zibunt/zibunzisini-o Chancellor NOM [every professor] i-DAT selfi-ACC suisensaseta. recommended ‘The chancellor made [every professor]i recommend selfi.’ (81) a *[Zibuni/zibunzisini-no oya]-ga [subete-no gakusei]i-ni selfi GEN parents NOM [every student] i DAT gakkoo-o yamesaseta. school ACC quit ‘Selfi’s parents made [every student]t quit school.’ b. [[Zibuni/zibunzisinrno oya]- ga Johni-ni gakkoo-o selfi GEN parents NOM DAT school ACC yamesaseta] koto ga karej-no kokoro-o kurakusita. quit fact NOM his feeling ACC disturbed ‘The fact that selfi’s parents made Johni quit school made him; upset.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 Zibunlzibunzisin has the Q-NP [subete-no kyooju] as a c-commanding antecedent in (80). In (81a), on the other hand, the Q-NP [subete-no oya] does not c-command zibunl zibunzisin. (81b) illustrates that the c-command is not necessary in the case of coreference . The sentences in (82) also show that the c-command is necessary in the case of bound variable anaphora (cf (82a)) whereas it is not in the case of coreference (cf. (82b)). (82) a. *[Kangaenasini [subete no kyoju]i-ga natuyasumi ni thoughtlessly every professor-NOM summer vacation during ryoko sita] koto]- ga (ato de) zibuni/zibunzisiiii no tatiba-o traveled fact NOM (later) selfi’s position-ACC warukusuru genin ni natta. worsen cause became ‘The fact that every professori thoughtlessly traveled during the summer put selfi in an awkward position.’ b. [Tamatama Johnr ga sono tegami-o yonda] koto]- ga by chance NOM that letter-ACC read fact NOM (ato de) zibuni/zibunzisini no tatiba-o warukusuru genin ni. (later) self’s position-ACC w orsen cause natta became ‘The fact that John! happened to read the letter put selfi in an awkward position.’ 3.2.2. Split Antecedent As discussed in section 1.2, in English, pronoun binding permits split antecedents whereas anaphor binding does not. In section 2 .1 .2 ,1 introduced Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 Fukui’s (1984) claim that zibun-tati allows split antecedents and, therefore, zibun has a pronoun-like characteristics I argued against Fukui’s claim by pointing out (i) that zibun is singular-denoting, and (ii) that what appears to be split antecedence cases with zibun-tati is due to the plural morpheme tati. Zibun and zibunzisin do not take split antecedents as illustrated in (15), repeated here as (83). (83) (=(15), Fukui’s (33a), 1984) Johnj-ga Maryj-ni [s Bill-ga zibun-tati;-j-o waruku itteiru] NOM to NOM ACC badly saying to siraseta (koto) COMP reported ‘John; reported to Maryj that Bill was speaking ill of selfl-j.’ Although Fukui claims that ‘the plural form of zibun’ (i.e., zibun-tati) has John and M ary as its antecedents, his argument is unsound due to the characteristics of zibun described above. Zibun!zibunzisin do not permit split antecedents. The fact that they are singular terms prevents them from allowing split antecedents. 4. Reanalysis of zibun and zibunzisin 4.1. [pro [zibun]] and \pro [zibunzisin]] In chapter 3, it is shown that zibun and zibunzisin have the following properties: (i) they can be bound by non-local antecedents; (ii) if coreference is at stake, they need not be c-commanded by their antecedents; (iii) they show weak Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 crossover effects when bound variable anaphora is at stake; and (iv) being singular- denoting, they do not allow split antecedence. These data indicate that zibun and zibunzisin are not anaphors. Then, what are zibun and zibunzisin? An alternative analysis is to claim that they are simply pronominals. One possible objection to this analysis is to indicate that a sentence like (84) would be ruled out by Condition B, if zibun and zibunzisin were pronominals. (84) Johni-ga zib unjzibunzisin ,-o suisensita. NOM self ACC recommended. However, if we recall what Hoji (1995) points out about Condition B effects in Japanese, it becomes clear that such an argument is unsound. As noted in Footnote 27, Hoji (1995) demonstrates that Condition B is not observed in Japanese when coreference is at stake. For example, (85) shows that soko ‘it,’ can be coreferential with its antecedent in its local domain. (85) (Hoji’s (21)) (Toyota-ga) [Arizona koozyoo]i-ga sokoi-o suisensita (n desu) (Toyota-NOM) Arizona factory-NOM it-ACC recommended ‘(It was Toyota that) [(its) Arizona factory]; recommended it;.’ Recall now, as introduced in chapter 2, section 2.3, Hoji (1995) points out, when bound variable anaphora is at stake, Condition B effects are observed. Now see the sentences in (86). (86) (Hoji’s (20a, b)) a. *(Toyota-ga) [Arizona koozyoo-sae]; -ga soko;-o suisensita (n desu) (Toyota-NOM) [Arizona factory-even]-NOM it-ACC recommended ‘(It was Toyota that) [even (its) Arizona factory]; recommended it;.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 b. (Toyota-ga) [Arizona koozyoo-sae]i-ga [sokoi-no sitauke]-o (Toyota-NOM) [Arizona factory-even]-NOM [it-GEN subsidiary]-ACC suisensita (n desu) recommended ‘(It was Toyota that) [even (its) Arizona factory]i recommended it;’s subsidiary.’ (86a) shows that the anaphoric relation between soko and the Q-NP Arizona koozoo sae is unavailable. (86b) indicates that when the relevant binding is not local, the anaphoric relation is available. In section 2.3 o f chapter 2 ,1 stated that Hoji (1997a) notices the absence of Condition B effects in sentences with otagai when bound variable construal is concerned. That is, it is because the internal structure of otagai is [\?pro [s otagai]], the binding of otagai by its antecedent is not local in the sentences like (69) (repeated here). Therefore, in (69) for example, Condition B is not violated. I add (69’), a version of (69), in which otagai is represented as {pro [otagai]]. (69) (Hoji’s (28b), 1997a) [kanari-no kazu-no huuhu]i-ga (hissi-ni-natte) [otagai] i-o [a good number-GEN coupIe]-NOM (very hard) -ACC urikonde ita (no wa sensyuu-no kaigi-de da) was promoting (COMP TOP last week-GEN meeting-at be) ‘(It was at the meeting last week that) [each of [a good number of couples]]i was promoting otagaii with utmost enthusiasm.’ (69’) [kanari-no kazu-no huuhu]i-ga (hissi-ni-natte) \prox otagai]-o [a good number-GEN couple]-NOM (very hard) -ACC urikonde ita (no wa sensyuu-no kaigi-de da) was promoting (COMP TOP last week-GEN meeting-at be) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 If we consider zibun and zibunzisin, we realize that they can be analyzed along the lines of Hoji’s (1997a) analysis of otagai. In the example in (87), bound variable readings are possible. This indicates that in the sentence with zibunl zibunzisin, Condition B effect is not observed. (87) Chomskyi-sae-ga ziibuni/zibunzisini-o suisensita. even-NOM self! recommended. ‘Even Chomsky; recommended self;.’ I believe that it is due to the structure \pro[. .]\ that the bound variable readings are available in (87), as suggested in Footnote 14 in Hoji (1997a). I.e., what is bound by the Q-NP is not zibun or zibunzisin, but pro in |pro[ziburi\\ and [pro[zibunzisin]]. Therefore, Condition B is not violated in (87) since the binding is not local. 4.2. The advantages of postulating the structure [[pro[...]] The analysis of zibun and zibunzisin having the internal structure [pro[zibun]] and [pro[zibunzisin]] respectively gives an answer to the following two problems. First, if zibun and zibunzisin are not anaphors, the sentence like (88a) may be problematic since they seem to allow a local antecedent. (88) a. John; sae zibun/zibunzisini-o suisensita. even self-ACC recommended. ‘Even John; recommended self.’ b. John; sae [pro,[zibun]y\pro\[zibunzisin]]-o suisensita. I argue that the sentence (88a) has the internal structure as in (88b). That is, the anaphoric relation between John and zibun or zibunzisin is, in fact, that of John and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 pro in [pro [zibun]] or [pro [zibunzisin]]. Therefore, John is not a local antecedent of zibun or zibunzisin. The second problem concerns the contrast between English and Japanese sentences, as illustrated in (89) and (90). (89) a. * Johni voted for him;. b. *Everyonej voted for hinii. (90) a. John,-ga zibun; /zibunzisini-ni toohyoosita. -NOM selfi-ACC voted for ‘Johnj voted for self.’ b. Daremoi-ga zibun/zibunzisini-ni toohyoosita. everyonei-NOM selfl-ACC voted for ‘Everyonei voted for self.’ This problem too can be accounted for by assuming (i) the internal structure of zibun and zibunzisin to be [pro [zibun]] and [pro [zibunzisin]] respectively, and (ii) the formal relation called Formal Dependency (Hoji 1997b). I assume that (90b) is represented as in (91). (91) DaremOj-ga [pro,[zibun]]/[pro;[zibunzisin]]-ni toohyoosita. everyonei-NOM selfi-ACC voted for ‘Everyonei voted for self;.’ Hoji (1997b) proposes that the establishment at LF of the Formal Dependency, a formal relation, is a necessary condition for bound variable construal. According to Hoji, there are three necessary conditions for the establishment of Formal Dependency between A and B as shown in (92). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 (92) If FD (A, B),3 0 then a. B has a property [+P]3 1 *3 2 b. A c-commands B, and c. A is not in the local domain of B. Only when the conditions in (92) are satisfied, FD (A, B) is established, and FD (A, B) is argued to be interpreted as follows. (93) FD(A, B) has the interpretive consequence such that the value of B is to be the same as that of A. Given (92) and (93), the grammaticality of (91) can be accounted for as follows. (91) has the LF representation as illustrated in (94a), and the Formal Dependency is established between t and pro as shown in (94b). In (94a), all the conditions in (92) are satisfied. I.e., (i) pro is [+p]; (ii) / c-commands pro; and (iii) / is not in the local domain of pro. : 1 0 We are concerned only with FDs in which A and B of FD (A,B) are in argument positions. 3 1 Hoji states that the concept [+P] is inspired by Fiengo & May's (1994) indexical type p and the two concepts are closely related with each other, noting that the difference(s) between them will be discussed in his forthcoming work. 3 2 In Hoji (1997b), [+/-p[ is meant to account for the contrasts between the so series and a series of Japanese Demonstrative Paradigms (cf. (53) in Hoji (1997b) as well as Footnote 9 in this thesis). It has been observed that a member of the so series can yield bound variable anaphora interpretation, while a member of the a series cannot (Hoji 1991, 1995, 1997b among others). Hoji assumes that the contrast is due to a general difference between so and a, and he calls the relevant property [+p]. Therefore, so can be [+P], whereas a cannot. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 (92) a. Daremoj-ga [ti [^ro,[zibun]]/[proi[zibunzisin]]]-ni toohyoosita. everyonei-NOM selfj-ACC voted for ‘Everyonei voted for selfi.’ b. FD (t, pro) 4.3. The absence of local anaphors in Japanese I have argued that the so-called Japanese ‘reflexive’ zibun and zibunzisin are, in fact, not anaphors and the internal structure of them is [ n p pro[$ zibun]] and [n -p pro[^zibunzisin]] respectively. My claim, together with Hoji’s (1997a) analysis of the so-called Japanese reciprocal otagai ‘each other’ cast a doubt on the existence of anaphors in Japanese. As Hoji (1997a) points out in footnote 14, if we follow the lines of reasoning of Reinhart and Reuland (1991, 1993), it is possible to build up a hypothesis, i.e. Japanese lacks anaphors altogether Reinhart and Reuland, along the lines of Chomsky (1986), suggest that the defining properties of a local anaphor is formal agreement feature(s). According to them, anaphors are referentially defective NPs (i.e., not having the property R or [-R]j3), whereas pronouns are referentially independent (i.e. having the property R.or [+R]). Reinhart and Reuland consider that R should be a purely syntactic property. The difference of referentiality between anaphors and pronouns comes from the fact that anaphors lack “ According to Reinhart and Reuland (1993), they follow Chomsky ( 1981) and use the term “the property R (referential independence)” as in the GB framework. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 a full specification of <j>-features. Because of this, Reinhart and Reuland states, “they do not project an argument that can be interpreted independently. We take the lack of <j>-features to be the property responsible for their anaphoric nature” (Reinhart & Reuland, 1993, p. 658). They also notice that anaphors do not have a full specification for structural Case. Assuming that the full specification for structural Case is among the <J>-features, Reinhart and Reuland gives (95). (95) (Reinhart & Reuland’s (81), (1993)) An NP is +R iff it carries a full specification for <J)-features and structural Case. Now let us consider Japanese case. It has been proposed that Japanese does not have any formal agreement features (Fukui (1986); it is implied also in Kuroda (1988), as pointed out by Hoji (1996c)). If their proposal is correct and if we follow Reinhart and Reuland’s assumption, then it would not be surprising if we conclude that Japanese does not have local anaphors. 4.4. Remaining Issues My analysis of zibun and zibunzisin raises a number of questions. The following is no more than an indication of some of the remaining questions. First, I have proposed above that the internal structure of zibun and zibunzisin is [np pro[s zibun]] and [nppro[^ zibunzisin]] respectively. But it is not clear what exactly pro Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 is. As illustrated in (96) and (97), for example, its behavior is not easily accounted for. (96) a. [Dono gakusei]i-ga \pro\ [hahaoya]]-ni tegami-o kaita no. which student NOM mother to letter ACC wrote Q ‘[Which student]i wrote a letter to that gu yi’s/selfi’s mother?’ b. [Dono gakuseijrga [soitur no [hahaoya]]-ni tegami-o which student NOM that guy GEN mother to letter ACC kaita no wrote Q ‘[[Which student]! wrote a letter to that guyi’s mother?’ c. [Dono gakusei]i-ga [zibuni-no [hahaoya]]-ni tegami-o kaita no which student NOM self GEN mother to letter ACC wrote Q ‘[[Which student]i wrote a letter to selfi’s mother?’ As indicated in (96), pro in [pro [hahaoya]] (cf. (96a)) can be interpreted as either soitu ‘that guy’ (cf. (96b)), or zibun (cf. (96c)). This seems to suggest that pro can appear instead of soitu or zibun. However, the examples in (97) show that is not necessarily the case. In (97), read ec as pro. As noted in footnote 2 in chapter 1 , pro in this thesis means a phonetically empty argument. (97) [Dono gakusei]r mo [[zibuni/???soitui/ *eci]-ga tensai-da]-to each student self/that guy NOM genius COMP omotta thought ‘[[Each one of the student] i thought that selfi/???that guyi/*ect is a genius.’ Clearly, in (97), ec (i.e., pro) cannot appear instead of zibun. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 Secondly, the exact semantico-fimctional properties of zibun and zibunzisin are not clear. Let us look at (98) for instance. (98) Johni-ga \prox [zibun]]/[proi[zibunzisin]]-ga kiraida. NOM self NOM hate ‘John! hates self.’ In (98), pro, being co-indexed with John. Now, what is the value o f [pro[zibun]] or [pro[zibunzisin]]? One possibility is “John.” Then, what does zibun or zibunzisin in [pro[zibun]] or [pro[zibunzisin]] indicate? I believe it is closely related to what many scholars have tried to capture in terms ofpoint-of-view. The fuller discussion of how point-of-view affects the use of zibun and zibunzisin as a whole lies outside the scope of this thesis. I will, therefore, just make the following two points. First, native informants suggest me that, in English, although the sentence like ‘John hits his self.’ is unacceptable, ‘John despises his self.’ is acceptable. Zibun or zibunzisin in [pro[zibun]] or [pro[zibunzisin]] might be closely related to se lf in ‘John despises his self.’" ’ 4 The second point relates to how pro, zibunl zibunzisin, and pro[zibun]]/[[pro[zibunzisin]] functions. The following is what I suspect. First, pro in [pro[zibun]] or [pro[zibunzisin]] points to the antecedent. Second, zibuni or zibunzisin indicates the referent of the point-of-view (i.e., whose (e.g., the speaker, the subject of the sentence, etc.) point-of-viewT). Thirdly, the function of ^This point is brought to my attention first by H. Hoji (personal communication, 1997). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 [pro[zibun]] or |/vo[zibunzisin]] is to match the antecedent and the referent of the point-of-view. The third of the remaining question is that the acceptability of the sentences with zibun and zibunzisin varies in accordance with the kind of predicate used, as pointed out in Ueda (1986). (99) a. Johnj-ga zibunj/zibunzisini-o suisensita. Johni-NOM selfi-ACC recommended. ‘Johni recommended selfi.’ b. Johni-ga zibuni/zibunzisini-o nagusameta. Johni-NOM selfi-ACC consoled. ‘Johni consoled selfi.’ c. ???/*Johnrga zibuni/zibunzisini-o aratta. Johni-NOM selfi-ACC washed. ‘Johni washed selfi.’ d. *Johni-ga zibuni/zibunzisini-o korosita. Johni-NOM selfi-ACC killed. ‘Johni killed selfi.’ Notice even though [/?ro[zibun]] or |/>ro[zibunzisin]] and their antecedent are in exactly the same structural relations in the sentences in (99), the acceptability differs. What is of great interest is the fact that (99d) becomes perfectly acceptable if korosita is understood as restraining one’s feelings, as also noted in Ueda (1986). The examples in (99) suggest that the relevant referential interpretation between pro in [pro[zibun]] or [/vo[zibunzisin]] on the one hand and its antecendnt on the other is closely connected with lexico-semantic and pragmatic factors. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 That the distribution of zibun zibvnzisin is more restricted than otagai should also be a puzzle.3 5 In section 2.2.4 in chapter 2 , 1 have shown the examples that indicate the antecedents of zibun and zibunzisin are not necessarily in subject position (and, therefore, the restriction should not be simply described by ‘subject- orientation’). However, there seems to be a clear contrast between the position of the antecedent of zibun!zibunzisin and that of otagai. (100) a. ?*Johnrga Mary2-ni [pro2 zibun]/[/>ro2 zibunzisin]-no hon-o NOM DAT self GEN book ACC kaesita. returned ‘Johni returned Mary2 self2’s book.’ b. ?[John to Bill]-ga [Mary to Suzan]i-ni \prox otagai]-no ronbun-o. NOM DAT GEN paper ACC watasita passed ‘John and Bill passed [Mary and Suzanji each othen’s paper.’ The examples in (100) illustrate that the antecedent of otagai can be in non-subject position, whereas the antecedent of zibun!zibunzisin should be in subject position in some cases. 3 5 This was brought to my attention by K. Miyagawa (personal communication, March, 1998). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 5. Conclusion In this thesis, I have argued that the internal structure of zibun and zibunzisin is [nppro[s zibun]] and [nppro[s zibunzisin]] respectively and that the anaphoric relation between zibun/zibunzisin and their antecedents are in fact that between pro in [npp/-o[n zibun]] or [[np /vo[n zibunzisin]] and the antecedent of pro. The empirical materials presented in chapter 3 have supported the claim that zibun and zibunzisin are not (local) anaphors. The observation of the absence of Condition B effects when bound variable construal is considered, as discussed in chapter 4, has led to the postulation of the internal structure of zibun and zibunzisin being [np/vo[n -zibun]] and [np/vo[n zibunzisin]]. Although the syntactic analysis of zibun and zibunzisin such as above has its advantages, as illustrated in chapter 4, there remain questions not answered. In fact, the syntactic and/or semantico-fimctional properties ofpro, zibun, zibunzisin, [np pro[s zibun]], or [np/vo[n zibunzisin]] are not entirely clear yet. In order to account for the exact nature of zibun and zibunzisin, it seems unavoidable to clarify non- syntactic constraints. I believe that only when the interactions of diverse types of linguistic constraints are identified, we could fully capture the behavior o f zibun and zibunzisin. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Appendix: ‘Pronominal* like use of reflexive36: The Korean caki and zibun, zibunzisin I have argued that zibun and zibunzisin, which have been assumed as anaphors, are not anaphors. In this section, I would like to mention the cases in which zibun, zibunzisin, as well as the Korean caki are used like a pronoun. I believe this ‘pronominal’ like use o f ‘reflexives’ calls for further consideration. Nam-Kil Kim (personal communication, March, 1998) pointed out that the Korean reflexive caki is sometimes used like personal pronouns. According to him, it is very unnatural to use the Korean pronominal ku (‘he’) in speech, and instead of using the ku, the Korean speakers use zero-forms (deletion), epithets, definite descriptions, or reflexives/7 Nam-Kil Kim also drew my attention to the sentences like (101a), (102a), and (103), given in Sohn (1994). M In Footnote 2 ,1 note that pro means a phonetically empty argument in this thesis. I am not claiming that [/>ro[zibun]] and [/>ro[zibunzisin|] are pronouns, having the binding- theoretic [+p] feature. 3 7 Kang (1988) notes: The use of Korean pronominal ku, whether bound variable or referential, some marginaritv in colloquial speech, and is avoided in general. This is presumably because the independent use of Korean pronominal ku is a fairly recent development in Korean language, dating back to 1920s, even though the ku as a specifier (as in ku + N) has a long history in Korean. Due to this historical fact, Korean speakers are in general less accustomed to the usage of the pronoun in colloquial speech, so that the language learners in their early stage are not frequently exposed to the usage of the pronoun ku. (p. 196) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 68 (101)a. (Sohn’s (335c)) Minca-nun caki-ka ceyil silheha-n-ta Minca-TC self-NM most dislike-IN-DC ‘As for Minca, she dislikes herself most.’ b. *?Minca-nun kunyo-ka ceyil silheha-n-ta. Minca-TC she-NM most dislike-IN-DC ‘As for Minca, she dislikes herself m ost.’ (102)a. (Sohn’s (336)) Minca-nun tongsayng-i caki-pota te ttokttokha-ta Minca-TC sibling-NM self-than more smart-DC ‘As for Mina, her younger sibling is smarter than her.’ b. ?Minca-nun tongsayng-i kunyo-pota te ttokttokha-ta Minca-TC sibiing-NM she-than more smart-DC ‘As for Mina, her younger sibling is smarter than her.’ (103)a. (Sohn’s (337b)) Minca-nun cip-ey caki-ppwun i-ess-ta. TC home-at self-only be-PST-DC ‘As for Minca, she is the only person at home.’ b. * Minca-nun cip-ey kunyo-ppwun i-ess-ta. TC home-at she-only be-PST-DC ‘As for Minca, she is the only person at home.’ In these sentences, the reflexive caki is used instead of the so-called third person pronouns, ku ‘he’ or kunyo ‘she.’- ’8 In contrast, in (101b), (102b), and (103b), the “ Sohn (1994) does not explicitly state that (101a), (102a), and (103a) illustrate the cases where the reflexive is used as pronouns. His intention is to illustrate the cases where non subject topics serve as antecedents o f the reflexives. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 ‘personal pronouns’ are used. My Korean informants reported that (i) the ‘a’ sentences are acceptable; and (ii) the ‘b’ sentences are somehow ‘degraded,’ ranging from ‘less acceptable’ to ‘unacceptable.’ In fact, these speakers find the ‘b’ sentences to be very unnatural and avoid using ku and kunyo in everyday conversation. Now let us consider the so-called Japanese personal pronouns (cf. Footnote 9). It has been pointed out that there are no third person pronouns in Japanese (Fiengo and Haruna 1987; Hoji 1990; Martin 1975/1988; Sakuma 1951/1983, among others). Kuno (1973) states: Japanese lacks authentic third person pronouns. In colloquial speech, in which there are many levels of first person and second person pronouns used, no third person pronouns are used. Where English would use he, she, it, they, Japanese would either (i) have nor overt forms, (ii) have attribute nouns such as titles, or (iii) have full-fledged noun phrases.3 9 (p. 17) The word kare ‘he’ and kanozyo ‘she’ do appear in Japanese.4 0 However, according to Sakuma (1951/1983), these words are used due to translation necessities, and, in fact, Fiengo and Haruna (1987) notes that “somehow karelkanozyo give sentences the flavor of having been translated from an Indo- European language” (p. 116). 39Kuno (1973) also mentions the cases in which zibun is used in place o f kare. 4 0 The past works, such as Fiengo and Haruna (1987), Hoji (1990a, 1990b, 1991), Kitagawa (1981), and Kuroda (1965) maintain that kare is not a (personal) pronoun. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 Now let us look at the examples in (104), (105), and (106). They are the Japanese counterparts of (101), (102), and (103), respectively. My Japanese informants reported that the sentences with zibun or zibunzisin (i.e., the ‘a’ sentences) sound more natural than the sentences with kare or kanozyo (i.e., the ‘b’ sentences) for the relevant meanings and that they use the ‘a’ sentences in general. (104)a. Maryj-wa zibuni/zibunzisini-ga itiban kirai-da Mary-TOP self-NOM most dislike ‘As for Mary;, shej dislikes herself; most.’ b. ?Mary;-wa kanozyo;-ga itiban kirai-da Mary-TOP she-NOM most dislike ‘As for Mary;, she; dislikes herself; most.’ (105)a. Mary;-wa imooto-ga zibun/zibunzisini-yori atama-ga-ii Mary-TOP younger sister- NOM self-than smart ‘As for Mary;, (her;) younger sister is smarter than her;.’ b. Mary.-wa imooto-ga kanozyo;-yori atama-ga-ii Mary-TOP younger sister-NOM she-than smart ‘As for Mary;, (hen) younger sister is smarter than her;.’ (106)a. Mary;-wa ie-ni zibun/zibunzisinj?? dake ita. Mary-TOP home-at self; only was ‘As for Mary;, she; was the only person at home.’ b. ??*Mary;-wa ie-ni kazyo dake ita. Mary-TOP home-at she; only was ‘As for Mary;, she; was the only person at home.’ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 Nam-Kil Kim further pointed out that the Korean reflexive caki replaces the second person pronouns, as indicated in (107). (107) Swukcey-num caki-ka hay-la!4 1 homework-TOP self-NOM do-lM ‘A s for homework, you do!’ In Japanese too, the first or second person use of zibun is observed, as noted in Iida (1996), Kuroda (1965), and Martin (1976), among others. For example, Kuroda (1965) states that “ zibun was used as the first person noun in the Japanese armed forces” (p. 163). Iida (1996) points out that in the sentences like (108), zibun refers to the speaker. (108) (Iida’s (1) & (2), 1996, chap. 5) a. Tarooi- wa zibun;,< ^ - 0 hihansita. Taroo-TOP self-ACC criticized ‘Tarooi criticized him selfi/'m e.’ b. Taroo, -w a [Hanakoj-ga zibunj/j/^k-o hihansita] koto-o sitteiru Taroo-TOP Hanako-NOM self-ACC criticized COMP ACC know ‘Tarooi knows that Hanakoj criticized himj/herselfj/^me.’ Iida (1996) uses ’% ’ to show that the acceptability of the relevant interpretation varies among her informants, and for some of them, it is not possible at all. Iida calls the use of zibun as in (108) is a first person pronoun use o f zibun, and argues that a separate lexical entry o f zibun should be proposed for this type o f zibun. For me as well as the majority o f my Japanese informants, the reading ‘zibun = the speaker’ is 4’This sentence is suggested to me by N.-k. Kim (personal communication, March, 1998) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 possible. That is, we can very easily imagine that those who belong to a right-wing organization, the Self-Defense Force, or a college athletic club use zibun as ‘I.’ As to the second person pronoun use of zibun, my informants told me that in the Kansai area (the western part of Japan) dialects, zibun is taken as one form of the second person pronoun. (109) Zibun, ashita nani suruno. You tomorrow what do-Q ‘What are you going to do tomorrow?’ This usage of zibun is not limited to the Kansai dialect, as indicated in (110).4 2 (110)a. A: Kinoo, shukudai-o wasurete, sensei-ni okorareta. yesterday homework ACC forgot teacher-by scolded Watashi-no sensei-wa hontoo-ni kibisiinda. I GEN teacher TOP really strict ‘Yesterday, I was scolded by (my) teacher for not doing the homework. My teacher is very strict.’ B: zibun/zibunzisin ga waruinda-yo. you NOM fault ASSERT ‘(It is) your fault, you know.’ b. soozi-wa okaasan-ga suru keredo, shukudai-wa cleaning TOP mother NOM do but homework TOP zibun/zibunzisin-ga si-nakerebaikemasen-yo you NOM do-must ASSERT ‘As for cleaning, I will do, but as for the homework, you must do, o.k.?’ 4 2 I am grateful to K. Miyagawa for bringing example such as (10 la) to my attention. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In this appendix, I have discussed the ‘pronominal’ like use o f ‘reflexives’ in Japanese and Korean. I believe that in order to fully capture the behavior of reflexives, this usage, too, needs to be fully investigated in future research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 References Aikawa, T. (1993). Reflexivitv in Japanese and LF analysis of zibun binding. Doctoral dissertation, Ohio State University, distributed as MIT Occasional Paper in Linguistics, No. 4, Cambridge, MA. Cho, D.-I. (1990). Caki and zibun. In E.-J. Baek, (Ed.), Papers from the 7th international conference in Korean linguistics (ICKL 7, p. 38-48). Osaka: International Circle of Korean Linguistics & Osaka University of Economics and Law. Chomsky, N. (1995). The minimalist program. Cambridge, MA: The MIT Press. Fiengo, R. & Haruna (1987). Parameters in binding theory: Some suggestions based on an analysis of Japanese. In T. Imai & M. Saito (Eds.), Issues in Japanese linguistics. Dordrecht: Foris Publication. Fiengo, R. & R. May (1994). Indices and identity. Cambridge, MA: The MIT Press. Freidin, R. (1992). Foundations of generative syntax. Cambridge, MA: The MIT Press. Fukui, N. (1984). Studies on Japanese anaphora I: The adjunct subject hypothesis and ^zibun”. Manuscript, MIT. Hankamer, F. & I. Sag (1976). Deep and surface anaphora. Linguistic Inquiry. 7. 391-426. Hoji, H. (1990a). On the so-called overt pronouns in Japanese and Korean. In Eung-In Baek (Ed.), Papers from the Seventh International Conference on Korean Linguistics (p. 61- 8). Osaka: International Circle of Korean Linguistics & Osaka University of Economics and Law. Hoji, H. (1990b). Theories of anaphora and aspects of Japanese syntax. Manuscript, University of Southern California. Hoji, H. (1991). Kare. In R. Ishihara & C. Georgopolous (Eds.), Interdisciplinary approaches to language: Essavs in honor of S.-Y. Kuroda. Dordrecht: Reidel. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 Hoji, H. (1995). Demonstrative binding and principle B. NFJ.S 25 255- 271. GLSA, University of Massachusetts, Amherst. Hoji, H. (1996a). Sloppy identity and principle B. Manuscript, University of Southern California. To appear in H. Bennis, P. Pica, & J. Rooryck (Eds.), Atomism and Binding. Dordrecht: Foris. Hoji, H. (1996b). Null object and sloppy identity in Japanese. Manuscript, University of Southern California. Hoji, H. (1997a). Otaeai. Manuscript, University of Southern California. Paper presented at WCCFL 16, University of Massachusetts, Amherst. Hoji, H. (1997b). Formal dependency, organization o f grammar, and Japanese demonstratives. Manuscript, University o f Southern California. To appear in Akatsuka, H. Hoji & S. Iwasaki (Eds.), Japanese/Korean Linguistics. 7. Stanford, CA: CSLI. Iida, M. (1996). Context and binding in Japanese. Stanford, CA: CSLI Kang, M. Y. (1988). Topics in Korean syntax: Phrase structure, variable binding and movement. Doctoral dissertation, MIT, Cambridge, Mass. Katada, F. (1991). The LF representation of anaphora. Linguistic Inquiry. 22* 287-313. Kitagawa, C. (1981). Anaphora in Japanese: Kare and zlbitn. Covote Paper. 'L University of Arizona, Tucson. Koster, J. (1982). Counter-opacitv in Korean and Japanese. Manuscript, Tilburg University. Kuno, S. (1973). The structure of the Japanese language. Cambridge, MA: The MIT Press. Kuno, S. (1987). Functional syntax: Anaphora, discourse and empathy. Chicago: University of Chicago Press. Kuroda, S.-Y. (1965). Generative grammatical studies in the Japanese language. Doctoral dissertation, MIT. Martin, S. (1975/1988). A reference grammar of Japanese. New Haven: Yale University Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 McCawley, N. A. (1976). Reflexivization: A transformational approach. In M. Shibatani (Ed.), Syntax and Semantics 5: Japanese Generative Grammar (p. 51-115). New York: Academic Press. Nakai, S. (1976). A study o f anaphoric relations in Japanese. Manuscript, University of Massachusetts. Nakai, S. (1977). Kare and kanozyo. Eigo Eibungaku Kenkvuu. 16. Nakayama, S. (1982). On English and Japanese pronouns. Master’s thesis, University of Tokyo, Tokyo. Oshima, S. (1979). Conditions on rules: Anaphora in Japanese. In G. Bedell, E. Kobayashi, & M. Muraki (Eds.), Explorations in linguistics: Papers in honor of Kazuko Inoue. Tokyo: Kenkyusha. Reinhart, R. & E. Reuland (1993). Reflexivity. Linguistic Inquiry. 24: 657- 720. Sag, I. (1976). Deletion and logical form. Doctoral dissertation, MIT, Cambridge, Mass. Sakuma, K. (1951/1983). Gendai nihongo no hvoogen to eohoo. Tokyo. Kuroshio Shuppan. (The references in the text are to the 1983 version.) Saito, M. (1981). Notes on Anaphora in Japanese (I). Manuscript., MIT, Cambridge, Mass. Saito, M. & H. Hoji (1983). Weak crossover and move a in Japanese. Natural Language & Linguistic Theory. 1.2. Sonh, Ho-min (1994). Korean. London: Routledge. Ueda, M. (1986). Notes on a Japanese (reflexive) pronoun zibtm. Manuscript, University of Massachusetts. Williams, E.S. (1977). Discourse and logical form. Linguistic Inquiry. 8. 101-139. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. w IMAGE EVALUATION TEST TARGET (Q A -3 ) / - ✓ :< 9 150mm IIVWGE, Inc 1653 East Main Street Rochester. NY 14609 USA Phone: 716/482-0300 Fax: 716/288-5989 0 1993. Applied Image, Inc.. A ll Flights Resented Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
In search of the sun: Yukio Mishima's body aesthetic
PDF
Long distance binding, logophoricity and Chinese reflexives
PDF
Religion and literary practice in the essays and fiction of Endo Shusaku
PDF
Negotiations of the transPacific, United States-Japan divide in the writings of Lafcadio Hearn and Yone Noguchi
PDF
A commercial and optimistic worldview of the afterlife of the Song people: Based on stories from the "Yijian Zhi"
PDF
Second language acquisition of the ba-construction in contemporary Mandarin Chinese
PDF
Korean-Japanese writer, Yu Miri overcoming Korea and Japan
PDF
Intersubjectivity and the mother-daughter dyad in Korean American women literature
PDF
Representations of the mother as origin and force of life in Japanese literature and history
PDF
Discovering Japan: Anime and learning Japanese culture
PDF
The evolving vocabulary of otherness in pre-imperial China: From 'belligerent others' to 'cultural others'
PDF
Tanizaki and film: An introduction with three early writings in translation
PDF
Ito Noe: Living in freedom. A critique of personal growth in Japanese society
PDF
A picture of human composition: An interpretation of Mencius' theory about human nature
PDF
Yamamoto Sumika: A new kind of living goddess in Japan
PDF
"The Infant in the Void": A spiritual journey
PDF
The feminized city: Reading Wang Anyi's "Ballad of Eternal Sorrow"
PDF
The death penalty in Japan and China: A comparative study
PDF
Confrontation and compromise: The worlds of the supernatural and humans in Pu Song-Ling's "Liaozhai Zhiyi"
PDF
Maternal devotion: the symbiotic relationship between mothers and sons in Yi Jian Zhi
Asset Metadata
Creator
Kagawa, Akemi
(author)
Core Title
On the so-called Japanese reflexives zibun and zibunzisin
Degree
Master of Arts
Degree Program
East Asian Languages and Cultures
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
language, linguistics,OAI-PMH Harvest
Language
English
Contributor
Digitized by ProQuest
(provenance)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c16-24913
Unique identifier
UC11336906
Identifier
1393172.pdf (filename),usctheses-c16-24913 (legacy record id)
Legacy Identifier
1393172.pdf
Dmrecord
24913
Document Type
Thesis
Rights
Kagawa, Akemi
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