UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A transformational study of Japanese reflexivization Matsuda, Hiroshi Jan 29, 1975

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


831-UBC_1975_A8 M38_7.pdf [ 5.6MB ]
JSON: 831-1.0093433.json
JSON-LD: 831-1.0093433-ld.json
RDF/XML (Pretty): 831-1.0093433-rdf.xml
RDF/JSON: 831-1.0093433-rdf.json
Turtle: 831-1.0093433-turtle.txt
N-Triples: 831-1.0093433-rdf-ntriples.txt
Original Record: 831-1.0093433-source.json
Full Text

Full Text

A TRANSFORMATIONAL STUDY OF JAPANESE REFLEXIVTZATION by HIROSHI MATSUDA B.A., Kwanseigakuin University, Japan, 1970 B.A., Kwanseigakuin University, Japan, 1973 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in the Department of LINGUISTICS We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April, 1975 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 i ABSTRACT Japanese reflexivization has only recently come to lin guists' attention in the framework of Chomsky's linguistic model, the transformational generative grammar. The present thesis, also being based on this newly de veloped model, is aimed at investigating the nature of the Japanese reflexive system. In effect, the investigation is to become a procedure whereby inconsistencies of the current hypothetical treatments of Japanese reflexivization can be excluded so that a revised proposal can be presented. The discussion toward a revised proposal is designed to pass through the two phases. First, in Chapter I, by examining Oyakawa's hypothesis, the discussion will focus on the structural positions of the antecedent and its re flexive, which will outline the basic property of Japanese reflexivization. It is also claimed that Langacker's two notionst 'command' and 'precede', are crucial in account ing for the question of the directionality, Forward and Backward reflexivization. The result of the examination will be presented as the revised proposal (I) to conclude 11 the chapter. Next, in Chapter II, the second phase of the revision will result from the examination of Akatuka's treatment of Japanese reflexivization. The examination will eventually suggest that the ill-formedness treated by Akatuka stems rather from the inadequacy of what has been allegedly treat ed as a Japanese reflexive pronoun than from a syntactic reason. Consequently, the alternative to replace the alleged reflexive form will be presented with syntactic evidence in which the existence of the non-human reflexive pronoun is to be pointed out as well. The conclusion to the chapter will be presented as the revised proposal (II) showing that the coreferentiality between the antecedent and the genuine re flexive is really a phenomenon observable only in a simplex sentence. In Chapter III, some problems left unsolved will be dis cussed. Having shown that the possible solutions to the prob lems are not far from being ad hoc so long as we adhere to the standard transformational approach, the alternative so lutions will be looked into in the framework of Jackendoff's Interpretive Theory in which, unlike the standard transforms-* • • 111 tional theory, the antecedent-reflexive relation is to be accounted for in the semantic component rather than the syntactic component. This chapter will be concluded by showing that the interpretive approach is equipped with more explanatory power than the standard transformational approach. All the information obtained through the revision in the preceding chapters will play a crucial role in formu lating Japanese reflexivization by the interpretive approach. In Chapter IV, the conclusions to the preceding three chapters are to be recapitulated so that a schematic represen tation of Japanese reflexivization is presented as both the conclusion to the present thesis and a tentative conclusion to the issue of Japanese reflexivization. iv TABLE OF CONTENTS Chapter Page I. OYAKAWA'S HYPOTHESIS AND THE REVISED PROPOSAL (I) 1 1.1 Oyakawa's Hypothesis 2 1.2 An Examination Of Oyakawa's Hypothesis 18 NOTES FOR CHAPTER I 49 II. AKATUKA'S LIKE-NP CONSTRAINT AND THE REVISED PROPOSAL (II) 5* 2.1 Akatuka's Like-NP constraint 55 2.2 The Genuine Reflexive And The Revised Proposal (II) 76 NOTES FOR CHAPTER II 105 III. SOME RESIDUAL PROBLEMS AND THE INTERPRETIVE THEORY 107 3.1 The Interpretive Theory By Jackendoff 108 3.2 Non-preferential Zibun 'self And Zibunzisin 'oneself* 117 NOTES FOR CHAPTER III 130 IV. CONCLUSION 132 BIBLIOGRAPHY 6 CHAPTER I OYAKAWA'S HYPOTHESIS AND THE REVISED PROPOSAL (I) 1 INTRODUCTION This chapter is mainly concerned with the coreferen-tiality of the two noun phrases, namely, the antecedent-T reflexive relation observable in such sentences as followst (1) a. KuniOjga Kenta^o zibun^jno uti de korosita. self *s house at killed "(Lit.) Kunio^ killed Kenta at self^'s house." b. Kenta-ga Kunio^ni zibun^^^no uti de koros-by self *s house at kill rare-ta. (Pass) "(Lit.) Kenta* was killed by Kunio at self.'s house." To account for the above-mentioned relation, the dis cussion is to be mainly based on Oyakawa's two elaborated conditions t the subject-antecedent condition and the high-2 est human NP condition, A close examination of the two con ditions eventually suggests that we revise Oyakawa's hypoth esis itself to avoid its inconsistency. The result of the following discussion will be presented as the revised pro posal to conclude the chapter. 1.1 OYAKAWA•S HYPOTHESIS We shall observe Oyakawa's two conditions which are the most crucial constituents of his hypothesis. In 1.1.1, the  subject-antecedent condition is to be observed and in 1.1.2, the highest human NP condition. 1.1.1 The sub.iect-antecedent condition It is claimed by Oyakawa (1973) that the antecedent of a Japanese reflexive pronoun must be the subject of the sen tence. This condition is called the sub.iect-antecedent con  dition.2 In order to clarify the nature of the above condition, let us observe the underlying structure of (1) first in terms of diagrams. (2.a) and (2,b) correspond to (l.a) and (l.b) respectively. 3 (2) a. Kenta. •Kunio* In the above, (l.b) is the passive version of (l.a). In both sentences, the reflexive zibun *self* only refers back to the subject. Hence, the reading indicated by the index j. is ungrammatical. The following (3) is a schematized re presentation of the sub.iect-antecedent condition where NPa is the subject noun phrase of Si and NPr is to be reflexiviz-ed under the referential identity with NPa. (3) Thus, the above condition plays a crucial role in deter mining the antecedent of Japanese reflexive pronoun, whereas this condition does not work in English as illustrated belowi (4) a. Tom^ showed Jim^ a picture of himself^ ... b. Iomega Jim^ ni zibun^j no syasin o miseta. to self *s picture showed "(Lit.) Tom. showed Jim self.'s picture." c. Betty^ showed Jim., a picture of himself d. Betty.ga Jim.ni zibuni#.no syasin o miseta. to self *s picture showed "(Lit.) Betty^ showed Jim self^'s picture." Although the reflexive 'himself could be coreferential with either 'Tom' or 'Jim' in (4,a)i its Japanese translation (i. e. 4.b) only has one reading as shown by the index i. Also, (4.c) indicates that English reflexive pronouns can be co-referential with the direct object of the verb rather than 5 the subject. This is not the case with its Japanese counter part (i.e. 4.d) since it is impossible for the reflexive zibun 'self* to be coreferential with 'Jim'. Hence, the asterisk for the index j.. Oyakawa also discusses the command condition which is inter-dependent with the sub.iect-antecedent condition. First, consider the following. (5) a. Kunio^wa Kunio^ga yonda hon o Kenta ni yatta. read book to gave "Kunio gave the book he read to Kenta." b. Kunio.wa zibun^ga yonda hon o Kenta ni yatta. self read book to gave "(Lit.) Kunio. gave the book self, read to Kenta." c. *Zibun^wa Kunio.ga yonda hon o Kenta ni yatta. self read book to gave "(Lit.) *Self. gave the book Kunio. read to Kenti." 1 Oyakawa attributes the ungrammaticality of (5»c) to the violation of the command condition which, as:..-#irs:to pro posed by Langacker (1969»l67);, is .defined r as follows: 6 (6) The command condition [l] Neither A nor B dominates the other £ 2 3 The S-node that most immediately dom inates A also dominates B if £ 13 and [ 2 J above meet, we say A •commands' B. Now, let us examine whether it is valid or not for the command condition to block such ungrammatical sentences as (5.c). The schematical underlying structure of (5»c) follows as (7). (7) In the above, NPa and NPr meet the command condition. In other words, NPa 'commands' NPr, but not vice versa because the S-node immediately dominating NPr (i.e. S2) does not dom-inate NPa. Since there are two coreferential noun phrases, NPa and NPr in (7), reflexivization may seem to apply either forward (i.e. 5.b) or backward (i.e. 5.c). However, reflexiv ization does not operate backward. Obviously, it is the com  mand condition that" blocks the backward application of re flexivization. From the foregoing, Oyakawa (1973 * 118) gives the def inition of the sub.iect-antecedent condition along with the  command condition in the following wayi (8) The sub.iect-antecedent condition The antecedent of the reflexive must be the subject of a sentence and commands the coreferential NP to be reflexivized. The above condition, according to Oyakawa, not only blocks the ungrammatical (5«c)i but also tells us predict ably that the following sentences are ambiguous. (9) a. Kunio.wa Kenta^ni zibun. .no kuruma o untensuru to self ' s car drive koto o tanonda. that asked "(Lit.) Kunio. asked Kenta. to drive self. .*s car."1 J 1J 8 (9) b. Kunio.wa Kenta.ga zibun..no koto o yuu no o x j 1 j self 's thing say that kiratta. hated "(Lit.) Kunio^ hated Kenta^to talk about self^." c. Kunio.wa Kenta^ni zibun^^.no syukudai o yar-by self *s homework do sase-ta. (Caus) "(Lit.) Kunio. made Kenta- do self^-'s homework." d. Kunio.wa Kenta^ni zibun. .no mono o araw-by self 's stuff wash sase-ta. (Caus) "(Lit.) Kunio. made Kenta. wash self.^*s stuff." Notice that Kenta above is underlyingly the subject of the embedded sentence, which would be shown more clearly in the following simplified underlying structure of (9.a). 9 (10) NPjEj, above can be reflexivized, being coreferential with either NPi or NP2f for NP^ and NP2 are respectively the subject of Si and S2 and command NPjj. at the same time. Therefore, nothing prevents NPij, from being ref lexivized under the referential identity with either NPX or NP2. The above (10) is, in fact, ambiguous and so are the other sentences in (9). (9.c) and (9.d) are only different from the other sen tences in that a causative sase 'make (someone) do (some thing)* is involved in them. But the account of their am-biguity is not out of line with that for (9.a). We would assume the underlying structure of (9»c), for example, in the following two ways* (11) In the above, NPzj. is coreferential with NP2 that is the sub ject of the embedded S2. Therefore, in the S2 cycle, reflex ivization applies and results in the reading indicated by J[. in (9.c), while the resultant second reading is possible when NP^ is coreferential with NP^, the subject of the matrix sen tence as shown in (11). Thus, according to Oyakawa, the subject-antecedent con  dition along with the command condition determines the ante-11 cedent of the reflexive in a crucial manner in Japanese re flexivization. 1.1.2 The Highest Human NP Condition Another of Oyakawa's significant conditions is the high  est human NP condition. First, let us consider the following which is to provide us with the preliminary access to our main concern in the sec tion. (12) a. History^ repeats itself^. b.*Rekisi.ga zibun.o kurikaesu. history self repeat "(Lit. )*Historj^ repeats selfj.." Unlike English, the Japanese reflexive pronoun has only one form^ zibun 'self regardless of gender, number and per son. Furthermore, the Japanese reflexive pronoun carries J~+ human feature with it^ implying that non-human noun phrases are unable to be the antecedent of zibun 'self. Hence, in the above example, rekisi 'history' is never eli gible for the antecedent of the reflexive in (12.b). We would neednaieondition, therefore, to block ungrammatical 12 sentences such as (12.b). The condition which follows is due to Oyakawa (1973«95). (13) The humanness condition The antecedent of the reflexive must be human. Next, observe the following of Oyakawa's sentencesi (14) a. Zibun^no gakusee no kageki na koodoo ga self 's student *s radical behaviors Yamada-sensee^o kyoosyoku kara sirizok-sase-ta. Prof. teaching from retire (Caus) "(Lit.) The radical behaviors of self's students made Prof. Yamada. retire from teaching." b. Syatyoo no kettei to zibun^no iken no kuitigai ga president's decision and self's opinion discrepancy Tanaka-butyoo^ni aimai na taido o tor-sase-ta. Director by unclear attitude maintain (Caus) "(Lit.) The discrepancy between the president's decision and self^'s opinion caused Director Tanaka. maintain an unclear attitude." 1 (15) a. Zibun^no gakusee no totta kageki na koodoo ga self 's student adopted radical behaviors Yamada-sensee o kyoosyoku kara sirizok-sase-ta. Prof. teaching from retire (Caus) 13 "(Lit.) The radical behaviors that self,*s students adopted made Prof. Yamada, retire from teaching." (15) b. Zibun.no yarakasita tumaranai hema ga self 's made silly mistake Suzuki-si o meir-sase-ta. Kir. be depressed (Caus) "(Lit.) The silly mistake that self, made depressed Mr. Suzuki." (16) a. Zibun.no gakusee ga kageki na koodoo o self ' s student radical behaviors totta koto ga Yamada-sensee^o kyoosyoku Kara adopted that Prof, teaching from sirizok-sase-ta. retire (Caus) "(Lit,)' That self.'s students adopted radical behaviors caused Prof. Yamada. retire from teaching." b. Syatyoo ga zibun.ni dake kaisya no kimitu o president self to only company's secret morasita koto ga kaikeigakari^o tomadow-sase-ta. told that accountant be puzzled (Caus) "(Lit.) That the president told the company's secret only to self, puzzled the accountant^." Notice that the above sentences manifest one peculiar syn-14 tactic feature; that is, the subject of the sentence is non-human. According to Oyakawa, the sentences in (14), (15) and (16) have a. "nominally complex NP subject", a "sen-tentially complex NP subject",andJa "sentential NP subject" respectively. Whatever the structure of the subject may be, what is note-worthy here is that all sentences are perfectly grammatical despite the absence of a human subject. In other words, the subject-antecedent condition does not hold in the above examples, which would suggest that an other condition is needed in accounting for the reflexive-antecedent relation in these sentences. In order to give a reasonable explanation to the subject matter here, Oyakawa assumes the highest human NP condition in terms of the hierarchical relation of the two ^referen tial noun phrases. Oyakawa (1973«112) defines the condition as follows: (17) The highest human NP condition "(When a sentence does not satisfy the sub  ject-antecedent condition.) only the high est human noun in the structural hierarchy is allowed to be coreferential with the re flexive." 15 To see how the condition works, consider the follow ing (18.a), (I8.b) and (I8.c) which represent the underly ing structure of (l*.a), (15.a) and (l6.a) respectively. (18) a. NP no kageki na koodoo ^ radical behavior; APr no gakusee -*a student Yamada-sensee. Prof." 4 zibun self Yamada-sensee* Prof. sirizok-sase-ta retire (Caus) kyoosyoku teaching b. V£__(Same as NP^, r no gakusee •s student Yamada-sensee 4k zibun self kageki na koodoo totta radical behaviors adopted 16 (18) c. zibun self In the above schematization, NPlt the subject of Si, is non-human regardless of its structure. Therefore, as shown in (16), the highest human NP condition takes place, allowing only the hierarchically highest human noun phrase (i.e. NPa) to be coreferential with NPr. Thus, in Oyakawa's hypothesis, the sub.iect-antecedent  condition and the highest human NP condition are applying in a mutually exclusive way. In addition, according to Oyakawa, both conditions dictate the direction in which reflexivization operates ; that is, the former condition ac counts for Forward reflexivization, whereas the latter, for 17 Backward reflexivization. Since the two conditions are mutu ally exclusive, and do not apply to the same coreferential noun phrases at the same time, the ungrammaticality of the following (19)» in which Forward reflexivization has been applied to the sentences in (14.a), (15.a) and (l6.a) re spectively, is self-explanatory. (19) a. *Yamada-sensee.no gakusee no kageki na koodoo ga Prof. *s student radical behaviors zibun^o kyoosyoku kara sirizok-sase-ta. self teaching from retire (Caus) "(Lit.) *The radical behaviors of Prof. Yajaada.'s students made self, retire from teaching." b. *Yamada-sensee.no gakusee no totta kageki na Prof, 's student adopted radical koodoo ga zibun^o kyoosyoku kara sirizok-behaviors self teaching from retire sase-ta. (Caus) "(Lit.) *The radical behaviors that Prof. Yamada^'s students adopted made self^ retire from teaching." c. *Yamada-sensee.no gakusee ga kageki na koodoo o Prof, 's student radical behaviors 18 totta koto ga zibun^o kyoosyoku kara sirizok-adopted that self teaching from retire sase-ta. (Caus) "(Lit.) *That Prof. Yamada^s students adopted the radical behaviors caused self, retire from teaching." With the information obtained through the foregoing observation, we shall examine Oyakawa's sub.iect-antecedent  condition and highest human NP condition in section-1.2 and, as a result of this examination, it will be suggested that Oyakawa's hypothesis as based on the two conditions be revised. 1.2 AN EXAMINATION OF OYAKAWA'S HYPOTHESIS The examination of the sub.iect-antecedent condition and the highest human NP condition is in effect an examination of the hypothesis. In 1.2.1, we shall present the counter argument to Oyakawa's treatment of Forward and Backward re flexivization. In 1.2.2, some counter-examples to the two conditions are to be treated, whereby the inconsistencies in Oyakawa*s hypothesis become obvious. As both a conclusion to the chapter and a result of the 19 preceding examination, a revised proposal will be presented. 1.2.1 Forward and Backward Reflexivization As to the direction in which Japanese reflexivization takes place, Oyakawa (1973«123-124) saysj (20) "....the choice between Forward and Backward Reflexivization is uniquely predetermined by the hierarchical structure of a given sentence in terms of functions, such as subject, and a relation, i.e. command, of two coreferential NPs in the sentence. In this sense, thereC"is no need to use linear order to account for Jap anese Reflexivization. That is, the 'direction ality* expressed by terms forward and backward does not count in the syntactic operation of re-flexiviza'tion, for all the neccessary informa tion is provided outside of the notion." What is highly implausible in the above quotation is that Oyakawa mentions the syntactic relation —"subject" — and "a relation" of coreferential noun phrases — "command" — as the elements to predetermine the "directionality". In fact, contrary to Oyakawa's expectation, the two elements never ^predetermine" Forward and Backward reflexivization "uniquely". In order to show that Oyakawa's treatment of the ques tion of "directionality" is implausible, we can present the fol lowing two types of sentences 20 First, Oyakawa himself cites (21) from Kuno (1972). The structure underlying (21) follows as the diagram (22), (21) Sono keiken ga Hanako^ni zibun^ga baka dearu that experience to self fool is koto o osieta. that taught (22) In this underlying structure, the subject NPi of S± cannot be the antecedent of the reflexive pronoun, for the huiaanness  condition disqualifies, the non-human subject sono keiken 'that 21 experience' as a candidate for the antecedent of the reflexive. In other words, the humanness condition predicts that Si does not meet the sub.iect-antecedent condition. Hence, as the high-* est human NP condition dictates, only the hierarchically high est human noun phrase (i.e. NP2J is eligible for b?ing the antecedent of the reflexive. Consequently, reflexivization , takes place forward between NP2 and NP;, changing the latter noun phrase into the reflexive pronoun zibun 'self'. There fore, the following account of Backward reflexivization by Oyakawa (1973*119) is implausible. (23) "One way to explain why Backward Reflexivization* insteadc' of Forward Reflexivization, takes place in them would be to say that the subject-ante cedent condition cannot be,met for: an obvious rea sons the position of subject.is occupied by some thing else which does not naturally qualify for the antecedent, of the reflexive. The fact; that the condition is not satisfied calls for Backward Reflexivization*" Example (21) aboveodoes not meet the sub.iect-antecedent con  dition, yet, rather than calling for Backward Reflexivization as Oyakawa expects, it "calls for" Forward Reflexivization. Likewise, the forward manifestation of Japanese reflex ivization observable in the causative sentences below cannot 22 be accounted for if we adhere to (23) above,. (2*0 a. Senkyo no kekka ga Kunio^ni sibun^no election result by self *s mizyukusa o sator-sase-ta. immaturity realize (Caus) "(Lit.) The result of election made Kunio* realize selfimmaturity." b. Kenta no zyogen ga Kunio*ni zibun*no •s advice by self 's ayamati o nattokus-sase-ta. mistake understand (Caus) "(Lit.) Kenta's advice made Kunio, under stand self£$s mistake." c. Issatu no han ga Kunio^ni zibun*, no taido o one book by self *s attitude ketteis-sase-ta. decide (Caus) "(Lit.) One book made Kunio. decide self.'s attitude." 1 1 Recall that causativization precedes reflexivization to make Oyakawa*s highest human NP condition valid as we have observed in (14) through (16).* In (24.a) above* for example, only the forward application of reflexivization is possible since the highest human noun phrase is the one which precedes 23 its coreferent after causativization has applied. The fol lowing schematization represents this point more clearly. (25) senkyo no kekka election Ts result W.. no mlzyukusa ""s" immaturity satorf sase-ta realize (CausT" Kunio 'J The second type of sentences which also show that Oyakawa's treatment of Backward reflexivization is implausible are the thematized sentences,-* (26) a. Kunio^wa zibun^no tomodati ga sinda. self •s friend died "(Lit.) As for Kunio^ self's friend died." b. Kunio^wa zibun^ga katta takarakuzi ga attata. self bought lot fell on "(Lit.) As for Kunio*, the lot which self. bought won tne prize." It Is clear that the sub.iect-antecedent condition is not accountable for the above sentences since in (26) the subject itself contains the reflexive pronoun and the ante cedent is not the subject but the theme. The diagrams (27.a) and (27.b) represent the schematized underlying structures of (26.a) and (26.b) respectively. (27) a. So b. So 25 Notice that NPr above, which is one constituent of a nominally complex NP subject (i.e. 26.a) or a sententially complex NP subject (i.e* 26.b), is coreferential with NPa, the theme of So. Thus, it is impossible for SA to meet the subject-ante  cedent condition since it does not have coreferential noun phrases in the SA cycle. Then, the highest human noun phrase NPa is to be identified as the antecedent of the reflexive.^ Although Oyakawa seems to fail to notice it, the highest  human NP condition plays a crucial role in determining the antecedent-reflexive relation even in the thematized sen tences. Hence, the sentences in (26) above are counter examples to Oyakawa*s hypothesist that is, the highest human  NP condition applies forward. Therefore, Oyakawa's explana tion of the "directionality" in Japanese reflexivization as quoted in (20) and (23) is highly implausible. In order to account plausibly for this subject matter, it is necessary to develop an alternative explanation. Now, recall that Oyakawa in (20) discards the linear information of two coreferential noun phrases as irrelevant in Japanese reflexivization, saying^that "there is no need to use linear order to account for Japanese reflexivization." Nevertheless, we are able to give a very plausible account for the "directionality" of reflexivization by talcing into consideration the very thing Oyakawa rejects as irrelevant, namely, the linear order of coreferential noun phrases. The "directionality" of Japanese reflexivization is simply pre determined by whether or not the noun phrase precedes the coreferential noun phrase to be reflexivized. What follows represents this relationship clearlyi (28) NPa precedes commands NPr. Forward reflexivization follows commands Backward reflexivization where: NPa and NPr are coreferential human noun phrases To sum up the foregoing, we may say that the very thing Oyakawa discards as irrelevant (i.e. linearity) plays a sig nificant role in predetermining which direction Japanese re flexivization takes, along with the information of the hier archical order of noun phrases concerned (i.e. command). In addition, we do not have to rely on the two different elements 27 to account for the "directionality" as Oyakawa does t the syntactic relation 'subject*, on the one hand, and the hier archical order 'command', on the other* Instead* the linear and hierarchical positions of coreferential noun phrases cor rectly predict which direction Japanese reflexivization op-arates in. 1.2,2 More Counter-examples To Oyakawa's Hypothesis Now, let us proceed to the further examination of Oyakawa's hypothesis, dealing with some more counter-examples. First, observe the following sentences in (29) which are cited from Oyakawa » (29) a. Zibuiij,4^no kenzyuu o, nakuslta keikan^ ga self 's service lost policeman revolver butyoo^ni nakituita. chief to implored "(Lit.) The policeman, who lost self's service revolver implored the chief. b. Hensyuutyoo*wa zibun* *no denki o kaita editor self J's biography wrote syoosetukajni mannenhitu o okutta. novelist to fountain pen presented "(Lit.) The editor* presented a fountain pen to the novelist* who wrote self.**s biography." 3 1J 28 The above (29.a) should fall under Backward reflexivization. Butt Oyakawa attributes this to an example of Forward reflex ivization, giving the following deep structure. (30) In the S2 cycle of (30), keikan 'policeman' can be reflexiv ized under the referential identity with NP2, the subject of S2* before reiativization takes place. Oyakawa, however, does not mention why butyoo 'chief must be discarded as an illegal candidate for the reflexive. In other words, he does not men tion how the application of reflexivization between NP3 and NPi+ is blocked. Comparing (30) with (31), the underlying structure of (29,b), we can see this point clearly. 29 Notice that (3D is the underlying structure of the ambiguous sentence* NP^ of this structure may be coreferential with either NP^ or NP3. Then, why is it that butyoo 'chief* in (30) has no chance Of being reflexivized under the referential identity with NPl*, the highest human noun phrase of Si? To explain this inconsistency, one might say that Si does not meet the highest human NP condition because it has the human subject zibun no kengyuu o nakusita keikan 'the policeman who 30 lost his service revolver* in the cycle. Then, consider the sentence (32) which is followed by its simplified under lying structure (33). (32) Zibun^#*no kao-zyasin ga insatus-rare-ta koohosya^no self *s face-picture printed (Pass) candidate's posutaa ga yuukensya ..ni kubar-rare-ta. poster voter to hand-to (Pass) "(Lit.) The candidate's posters in which self's portrait was printed was handed to the voters." (33) NP? no posutaa ^ *s poster NHt- no kao-zvasin /\^s portrait f koohosya. koohosya, candidate yuukensya, kubar-rare-ta voter Jhand-to (PasiT 1 :T''~i NP< v t candidate ^>—^--^^^ [•yuukensya. (^ame as NP?) insatus-rare^ca * voter print (Hss")~~" The sub.iect-antecedent condition is not met in the S2 cycle above because S2 has the non-human subject NP3, koohosya no  kaozyasin 'candidate's portrait*. Hence, the highest human 31 noun phrase NP2 koohosya •candidate' is to be identified as the antecedent of NP/+. Then, in the Sj cycle, Si does not have the human subject in the same sense as in (29*a). Therefore, the same argument that discards the unacceptable reading of (29.a) can no longer hold to block the ungrammati-cal sentence as indicated by £ in (32). This is a serious counter-example to Oyakawa's hypothesis. To avoid this inconsistency, we must revise Oyakawa's hypothesis in some way* Notice that, in all the examples above, the antecedent of the reflexive has a common property in that the antecedent is the left-most human noun phrase. More precicely, in both (29.a) and (30), NP2 is the leftTmost human noun phrase in the S2 cycle and reflexivization applies forward in the same manner as Oyakawa's. Next, in the Si cycle, NPi, which is now a constituent of Sj, is the left most human noun phrase after NP2 was deleted in virtue of reflexivization* Notice, also, that NP2 and NP^ here are identical noun phrases. Hence, a non-ambiguous sentence or the ungraramatical reading indicated by index j in (29.a). The ungrammatical reading or non#ambiguity of (32) can be 32 accounted for along exactly the same line. The left-most noun phrase in the S2 cycle and the SA cycle is identical in (32), too. The ambiguity of (29.b) is also predictable} that is, in (31) NP3 is the left-most human noun phrase in terms of S2, and the left-most position is occupied by the human noun phrase NPX in the SA cycle. NP3 and NPA are not identi cal noun phrases this time. Hence, the ambiguous sentence (29.b). Second, the following sentences are another type of counter-examples to Oyakawa's hypothesis, and support the notion 'left-most' in Japanese reflexivization as well. (34) a. Zibun^ *^ga tootoo maketakoto ga Kunio^ ni self at last defeated that by KentajO mihaos-sase-ta. think-better-of (Caus) "(Lit.) That self* was defeated at last made Kunio^ think better of Kenta." b. Zibun#jga gosinsur-rare-ta koto ga Kunio^ni self diagnosed(Pass) that by erroneously isya o koros-sase-ta. doctor kill (Caus) "(Lit.) That self, was diagnosed erroneously made KentS*^ kill the doctor." 33 These sentences have the sentential NP subject with a non-human noun koto 'that* as its head noun and, at the same time, there are two human noun phrases in the hierarchically highest position, (35) below represents the structure under-: lying (3^,a), for example. (35) at last defeated The/above Sx does not satisfy the subject-antecedent condition because of its non-human subject NPX, so that the highest hu  man NP condition applies, producing the confusing situation. In other words, there are two human noun phrases, NP3 arid NPif, in the same hierarchical level. As the index j[ in (34.a) shows, NP4 is not eligible for the antecedent of NP2. Oyakawa would account for this saying that reflexivization must precede the 3* application of causativization^ since NP3 Kunio used to be the subject of an embedded sentence, while NP24. Kenta used to be the direct object as shown in the following sehematical underlying structure (36), However, this treatment only makes the situation worse. (36) Although NP3 Kunio is the highest Human noun phrase, it does not command NP2, the coreferent noun phrase, anymore since S3 immediately dominating NP3 does not dominate S2 in (36). Hence, nothing prevents reflexivization from applying forward to produce the ungrammatical sentence. (That Si does not meet the subject-antecedent condition never Implies reflexivization takes place backward, (cf. 21, 24, and 26)) In neither case, does the highest human NP condition hold to account for the 35 grammatical sentence (3^.a). It is important to notice that once again the linearly left-most position of the human noun phrase in (35) is oc cupied by the legal candidate for the antecedent of NP2 to be reflexivized. With the crucial notion 'left-most', we can systematical ly account for what Oyakawa*s hypothesis cannot. It may be said, therefore, that Oyakawa*s hypothesis must be revised to be equipped with more descriptive adequacy with the in formation of the linear order (left-most), and the hierarchi cal order (command) as well, of the two coreferential noun phrases. 1.2.3 The Top-most Human NP Condition The next argument suggesting the need for a revision of Oyakawa*s hypothesis stems from the following Oyakawa*s sen tences. (37) Hosyuseitoo no ooboosa to zibun,no sizisitekita conservative *s unreason- and self *s supported party ableness seitoo no fuhai ga sono gakusee no otooto no party *s corrup- that student 's younger *s tion brother 36 sinyuu^no sisoo o museifu-syugi e katamuk-sase-ta. best *s thought anarchism toward lean (Caus) friend "(Lit.) The unreasonableness of the conservative : party and the corruption of the party that self* had supported made the thought of the student's younger brother's best friend^ lean towards anarchism." As seen in its English translation, (3?) has a non-human subject so that the highest human NP condition is responsible for the antecedent-reflexive relation. Now, notice that the antecedent sinyuu 'best friend' is the highest human noun phrase in the nominally complex structure. To show this clear ly, Oyakawa gives the schematised surface structure of relevant noun phrases of (37) as followst (38) thought [-human] sono gakusee otooto that student younger brother sinyuu best friend 4, zibun self 37 Oyakawa (1973i112) supports the validity of his highest  human NP condition from the preceding fact, saying thati (39) "....the complexity of a nominally-complex NP can be increased as much as you want, yet only the highest human noun in the structural hierarchy is allowed to be coreferential with the reflexive." However, we are able to show that (39) above never supports the validity of the condition in question. Rather, what Oyakawa says as (39) is the case regardless of the highest  human NP condition or the sub.iect-antecedent condition. To clarify this point, let us consider (40) below. (40) a. Zibunjno denki o dasita koto ga sono gakusee no self *s published that that student 's biography otooto no sinyuu no haha^o yuumei ni sita. younger 's best 's mother famous made brother friend "(Lit.) That self's biography was published made that student's younger brother's best friend's mother^famous." b. Sono gakusee no otooto no sinyuu no haha^ ga that student 's younger 's best 's mother brother friend zibun^no denki o dasita. self's biography published 38 '(Lit.) The students younger brother's best friend's mother, published self,'s biography." 1 (4l,a) is a schematic representation of the nominally complex noun phrase which (40) above has in common. (4l.b) and (4l.c) represent the underlying structure of (40.a) and (40.b) respec tively. (4l) a. NPl [+humsu3 mother sinyuu best friend sono gakusee otooto that student younger brother b. NP C-humarff NPr no denki dasita 's published biography haha mother koto (Same as above) yuumel ni sita that famous made 39 c. S nana mothe r The nominally complex noun phrase (41.a) is different from the one in (37) in that the noun phrase (41.a) is itself human. In (4l.b), this nominally complex structure is in the direct Object position. The highest human noun phrase haha 'mother* is the antecedent of the reflexive as (39) dictates. It is thus clear that the highest human NP condition is responsible for (40.a) because of its non-human subject. Now* we must notice that the same noun phrase haha'motti er', the highest human noun phrase in the nominally complex structure, is also the antecedent in (40.b) as shown in (4l.c). Nevertheless* the sentence at issue cannot be an example of the highest human NP condition since its subject (i.e. NPj.) . .-. ...... | is human. (Recall that the highest human NP condition and the subject-antecedent condition are mutually exclusive.) 40 Thus, the subject-antecedent condition is responsible for (4o.b), yet (39) is still true with (40.b). Por the foregoing reason, we must reject the claim that the quotation (39) supports the highest human NP condition. Furthermore, in order to validate (39) which is the case regardless of the two conditions, we need a condition allow ing no constituent but the highest human noun phrase within the nominally complex noun phrase to be eligible for the ante-cedente of the reflexive. Therefore, we are proposing that such a condition be call ed the top-most human NP condition, and that this condition be defined as followst (42) The top-most human NP condition The antecedent of the reflexive must be the top-most human noun phrase if the eligible antecedent is the nominally complex noun phrase. It must be worth recapitulating that the quotation (39) is the case with both the subject-antecedent condition and the highest human NP condition so that (39) no longer supports *1 the highest human NP condition only, and, as a result, the condition (42) above is independent of Oyakawa's two conditions. 1.2.4 The Revision Of Oyakawa's Hypothesis To conclude the examination of Oyakawa's hypothesis, we shall present a revised proposal based on the foregoing ex amination. First, observe the following. (*3) a. NPa2 NPr wherei NPai(NPa2) and NPr are coreferential human noun phrases. NPai(NPa2) is the subject of Si(S2). The schematization (43) is a simplified representation of the sentence structure which meets the subject-antecedent condition. (43.b) could be the underlying structure of either non-ambiguous or ambiguous sentences. When NPaj is not equal to NPa2, sen-42 tences are ambiguous with NPr being coreferential with either the former noun phrase or the latter (e.f. 9). When NPai equals to NPa2» non-ambiguous sentences result. It is important to notice here that the crucial notion •left-most* discussed previously is also valid to describe the above configurations. Therefore, we may say about these configurations that the antecedent of its reflexive is the left-most human noun phrase in a give S cycle. Also, notice that reflexivization applies forward in the above cases, since the antecedent (the left-most human noun phrase), precedes its coreferent^(c.f. 28). Next, let us take a look at the configurations in (44). (44) a. Si NPr wherei NPa and NPr are coreferential human noun phrases. NPs is the subject of SA. *3 (44) b. NPs NPr [-human) where: NPa and NPr are coreferential human noun phrases. NPa is the theme of Si. NPs is the subject of S2. These two configurations are intended to show the cases where S2 does not satisfy the subject-antecedent condition and NPa, the highest human noun phrase, can be the antecedent of the reflexive. As shown in 1.2.1, (44) is also peculiar in that it is not an instance of the subject-antecedent condition, yet reflexivization takes place forward.(cf. 21, 24 and 26). Nevertheless, NPa is the left-most human noun phrase and commands NPr to be reflexivized. Thus, (44) can be accounted for along the same line as (43). Notice that we are also able to pre dict the forward manifestation of reflexivization quite sys tematically 1 that is, in (44) the antecedent precedes its reflexive as does that in (43). What follows is the case where the backward manifestation of reflexivization is observable. 44 (*5) a. NPs NPa £-human) "NPr" wheret NPa and NPr are coreferential human noun phrases. NPs is the subject of SA. b. NP Inhuman] NPr The configuration (45.a) represents the sentence structure whose subject position is occupied by a nominally complex non-human noun phrase, while (45.b) corresponds to the sen tence structure that has either the sententially complex NP subject or the sentential NP subject (c.f. 14, 15 and 16). In any case, a given sentence that has one of the underlying structures of (45) comes about to manifest Backward reflex ivization. Closer examination easily makes us aware that the de-*5 scription of preceding (43) and (44) does not differentiate from that of (45) above t that is, the antecedent of the reflexive is NPa which is the left-most human noun phrase, commanding its coreferent as shown in the configurations. The backward application of reflexivization is due to the linear order of the two coreferential noun phrases t the noun phrase to be reflexivized precedes its coreferent, which is quite predictable from (28). Japanese reflexivization, therefore, may be defined in terms of the linear and hierarchical order of two coreferential noun phrases in the following wayt (46) Japanese Reflexivization In a given S that dominates two coreferential noun phrases, NPa and NPr, change NPrvinto the reflexive pronoun zibun 'self if NPa is the left most human noun phrase and also commands NPr. This definition must be accompanied by the top-most human NP  condition proposed in (42) so that no left-branching noun phrase of a nominally complex noun phrase can be the ante-46 cedent of the reflexive. From the foregoing, it may also he said that what Oyakawa really implies by his two conditions is the linear and hier archical positions of two coreferential noun phrases. The  subject-antecedent condition is one special case of Forward reflexivization where the left-most human noun phrase is accidentally the subject of a sentence. Furthermore, our definition (46) in terms of the linear and hierarchical order of the antecedent and its reflexive is able to account for 8 Japanese reflexivization in a uniform manner , whereas Oyakawa uses two different factors t the syntactic relation •subject* and the hierarchical position * command*. *7 CONCLUSION Our discussion of Japanese reflexivization has been based on Oyakawa's two elaborated conditions, the subject-antecedent  condition and the highest human NP condition* Our closer ex amination of the conditions eventually led us to the conclusion that information about the linear order of two coreferential noun phrases is crucially needed as well as the hierarchical information, although Oyakawa in his hypothesis dismisses the linear information as irrelevant. Moreover, it may be said that Japanese reflexivization is, as correctly observed by Oyakawa, "the unitary operation" resulting in either Forward or Backward reflexivization as the surface phenomenon. This surface manifestation is pre determined depending on whether or not the left-most human noun phrase precedes its coreferential noun phrase. To conclude the chapter, we present, as a result of the foregoing discussion, the revised proposal (1), which will be given greater elaboration in the next chapter. 48 (47) The Revised Proposal (I) Japanese Reflexivization (i) In a given S that dominates two coreferential noun phrases, NPa and NPr, change NPr into the re flexive pronoun zibun *self if NPa is the left-most human noun phrase and also commands NPr. (ii) When the antecedent is the nominally complex noun phrase, only the top most human noun phrase is eligible for the antecedent. (iii) When NPa precedes NPr, reflexivization applies forward, resulting in Forward reflexivization superficially. When NPa follows NPr, Backward reflexive ization results. *9 NOTES FOR CHAPTER I 1. All example sentences are to be written in Roman alphabet with the following conventions. i. The conjugation of verbs is shown only when it is relevant to the discussion, ii. The following abbreviations are adopted throughout the present thesis* (Pass) — passive marker rare (Caus) — causative sase 'make (someone) do (something)' (Lit.) — literal translation 2. Oyakawa's definition of the subject-antecedent condition is to be quoted later as (8) in this section. 3. In relation to this, observe the following. (12) b*. Rekisi.wa sorezitai.o kurikaesu. history itself repeat 'History repeats itself.' Unlike (12.b), this sentence is grammatical suggesting that there seems to be a non-human reflexive pronoun in Japanese. This point is to be discussed later in 1.2.1, Chapter II. 4. Oyakawa does not mention the precedence of causativiza-tion over reflexivization explicitly. However, we may say so for the following reason. Consider Oyakawa's examples. (i) a. Zibun*ga hutatabi erabareta koto ga Satoo-san.o self again elected that Mr. odorok-sase-ta. be surprised (Caus) "(Lit.) That self* was elected again surprised Mr. Satoo£." b. Zibun*no hon ga uredasita koto ga syoosetuka*ni self *s book start-selling that writer to syoorai e no kiboo o ataeta. future to hope gave "(Lit.) That self.'s book started selling well gave the writer* a hope for his future." 50 (i) c. Zibun.no an ga saiyoosareta koto ga sono self *s plan adopted that the kikakubuin* o utyootennisita. member of delighted the plan ning staff "(Lit.) That self.'s plan was adopted delighted the membef\jOf the planning staff." In order to account for the reflexive-antecedent relation in (i), Oyakawa gives the underlying structure of (i.a), for example, as followsi Oyakawa's explanation (1973»107) readsi (iii) "This underlying structure presents a situa tion where the antecedent Satoo-san 'Mr. Sato' for Backward Reflexivization is the embedded subject NPa so that the subject-antecedent condition might be considered as valid. How ever, there are other cases like (22d-e) which are not causative sentences, so they cannot be explained by this condition." In the quotation, (22.d) and (22.e) correspond to (i.b) and (i.c) respectively. Thus, according to Oyakawa, all examples in (i) fall under the highest human NP condition. Notice that the con dition in question and the sub.iect-antecedent condition are mutually exclusive. Therefore, in order that Si in (ii) meets the highest human NP condition. S3 cannot have the human sub-ject NPa. In other words, the highest human NP condition can be responsible for (i.a) only when NPa is a constituent of Sif a higher sentence. Hence, causativization precedes reflexivization. followsithematized sentence", we imply the sentences as 51 (i) Zoo wa nana ga nagai. elephant trunk long "(Lit.) As for the elephant, the trunk is long," (ii) UBC ni wa Nitobe-garden ga aru. in is "(Lit.) As for in UBC, the Nitobe garden is there." What we must bear in mind here as relevant to our main concern is that the thematic noun phrases Chomsky-adjoin to the left-most position, and also that the intervening particle deletion depends upon the kind of particle. In the above ex amples, for instance, the intervening particle no (genitive) must be deleted to derive (i) from the following (i*), where as ni (locative) may not be deleted in (ii) whose non-thematiz-ed structure is (ii*) below. (i*) Zoo no hana ga nagai. elephant *s trunk long "(Lit.) Elephant's trunk is long." (ii») Nitobe-garden ga UBC ni aru. in is "The Nitobe garden is in UBC." 6. Consider the sentences in (i). (i) a. Kunio.ga Kenta.o zibun.#,no uti de korosita. 1 3 self1 J,s house at killed "(Lit.) Kunio*^ killed Kenta at self's house." b. Kenta,ga Kunio,ni zibun.#.no uti de koros-1 Jby self1 J,s house at kill rare-ta. (Pass) "(Lit.) Kentaj was killed by Kunio at self's house." c. Kunio.ni wa Kenta.ga zibun*, .no uti de koros-xby 3 self 1 J,s house at kill rare-ta. (Pass) "(Lit.) As for by Kunio,, Kenta, was killed at self^'s house." J 52 In these examples, (i.c) is different from (i.b), the passive version of (i.a), in that the b-sentence has the thematic noun phrase Kunio ni wa *As for by Kunio*. It must be noticed here that Kunio. the thematic noun phrase, has no chance of being the antecedent of the reflexive but only Kenta. the subject of the sentence can be. Ggawa (1974*137) correctly observes this and gives a solu tion of the problem in the following ways (ii) 1. The subject-antecedent condition takes precedence over the theme-antecedent condition. 2, The theme-antecedent condition is sus pended, by virtue of the subject-ante cedent condition, when the subject which intervenes between a thematic NP and its identical NP is human. 3. If the intervening subject is not human, reflexivization applies. The theme-antecedent condition above is the condition that de-termines the thematic noun phrase as the antecedent of the re flexive . Notice that we may modify what Ggawa says in (ii) into such a way as (iii) below since the thematic noun phrase is also the highest human noun phrase. (iii) The highest human NP condition does not apply if the human subject intervenes between a theme and its reflexive. Noticeably, (iii) above is the paraphrase of (1?) in 1.1.2. In other words, (ii) is not necessary. Rather, the highest  human NP condition along with the subject-antecedent condition plays a significant role in determining the antecedent of the reflexive even in the thematized sentence structure without any modification. Therefore, Qgawa's solution is virtually redundant as well as such a condition as the theme-antecedent  condition. 7. See note 4 c, above. 8. What follows is Oyakawa*s (1973*124) conclusion to the question of "directionality" in Japanese reflexivization. "....what we have called Forward and Backward Reflexiv ization are the surface results of the unitary operation." This conclusion of Oyakawa*s is supported by our analysis of Japanese reflexivization in terms of the linear and hierarchi-53 cal order of two noun phrases concerned, for the definition we have proposed in (ko) clearly implies that Japanese re flexivization is able to be accounted for in a uniform way regardless of the direction in which it applies* CHAPTER II AKATUKA'S LIKE-NP CONSTRAINT AND THE REVISED PROPOSAL (II) 2 INTRODUCTION We are concerned with the ungrammaticality of such sen tences as theset (48) a. *Kunio^wa zibunjO tataita. self hit "Kunio hit himself." b. *Kunio^wa zibun^ni kuruma o katta. self to car bought "Kunio bought a car to himself." These sentences differ entirely from those discussed previously in that each simplex sentence above perfectly satisfies the reflexivization condition (4?), yet each is ungrammatical.1 Therefore, a further condition would be needed to block such sentences as (48). In regard to this, Akatuka proposes the syntactic con straint, the Like-NP Constraint, and the two relevant trans-55 formational rules which are fully examined in the course of the following discussion. The examination of them will even tually suggest that Akatuka's constraint and two rules be dis carded as inadequate and that the possibility;of an alternative solution be looked into. Consequently, we shall have to de termine whether or not the alleged Japanese reflexive pronoun zibun 'self is a genuine reflexive form. As the conclusion to the chapter, we shall present our proposal based on the ex amination. 2.1 AKATUKA'S LIKE-NP CONSTRAINT In the following three sections, we shall examine in detail Akatuka's Like-NP Constraint and two related trans formational rules — Inalienable-Possessor-Deletion and Unspecified-Body-Deletion. 2.1.1 The Like-NP Constraint Akatuka (1772t30) proposes the constraint as quoted below in (49). using the notion 'peer'; whose definition by Postal (1970tl78-179) also followsi 56 (49) The Like-NP Constraint "(The constraint) discards the sentences as ungrammatical if the reflexive and its ante cedent are in peer relation." (50) The peer relation "Two NP, NPi and NP2, neither of which dom inates the other....in a phrase marker P are peers with respect to a node Si, just in case the paths between each of these NP and Si are such that they contain no NP-nodes not sepa rated from the starting point NP, NPA or NP2» by a node SA." The underlying structure of (48.a) is given in (51) to see how the constraint is able to account for the ungrammatically. (51) S Kunio* NPo V Kunio. tataita _i hit In this underlying structure, NPi is perfectly eligible to be the antecedent of NP2 since the former is the left-most human noun phrase which also commands the latter. However, reflexiv ization between these two coreferential noun phrases must be blocked. For, according to Akatuka, NPA and NP2 are "in peer relation". Hence, (48.a) is ungrammatical. Furthermore, it 57 is important to notice that the constraint in question remains 2 in effect only between "the two maximum NP*s in the simplex sentence"."' Therefore, the following sentences given by Akatuka do not violate the Like-NP Constraint. (52) a. HirosijWa zibun^no titi o sonkei site irji. self *s father respect is "(Lit.) Hirosi*^ respects self's father." b. Hirosi^wa oyahukoomono no zibun^o hazita. unfilial son self ashamed of "(Lit.) Hirosi^ was ashamed of unfilial self^" c. Hirosi^wa kagami ni ututta zibun^o nagameta. mirror in reflected self looked at "(Lit.) Hirosi, looked at self* who was reflected in the mirror." In (52.a) above, the reflexive zibun 'self is the left-branch ing noun phrase of the direct object titi 'father*, while in (52.b) and (52.c) the reflexive occupies the direct object position as the head noun of the relative clause. Hence, the above sentences are exempt from the Like-NP Constraint. The following diagrams illustrate the preceding point more clearly. (53.a), (53.b) and (53.c) correspond to (52.a), (52.b) and (52.c) respectively. 58 (53) Hire-si^ NP? no titj >/\^s father Hirosi. zibun self sonkei site iru respect is b. oyaJftuKoomono a5a unfilial son ii ashamed of / 59 In these diagrams, NPX and NP2 are not in peer relation because they are not "the two maximum NP*s in the simplex sentence". Therefore, reflexivization may take place between the two coreferential noun phrases, resulting in the grammatical sen tences. Thus, as far as such sentences in (48) are concerned, the Like-NP Constraint is crucial. 2.1.2 Inalienable-Possessor-Deletion and Unspecified-Body- Deletion Akatuka proposes two transformational rules which are inter-related with the constraint above. Let us observe the sentences in (54) in relation to the two rules at issue. (54) a. •Kunio^wa zibun^o tataita. self hit "Kunio hit himself." b, Kunio^wa zibun.no hoc- o tataita. self 's cheek hit "(Lit.) Kunio., hit self^s cheek." c. Kunio wa Kenta o tataita. hit "Kunio hit Kenta." 60 d. Kunio wa Kenta no hoo o tataita. •s cheek hit "Kunio hit Kenta on his cheek." Akatuka (19?2»33) accounts for the ungrammatically of (5*.a) in the following wayt (55) "....if the action identified by the verb affects the subject NP, then the specific body part must be mentioned...., otherwise the sentence is ill-formed." Contrary to what is mentioned in (55)» the specific body part does not have to be mentioned if the subject noun phrase is not affected by the action identified by the verb in such a case as (5*.c) where not the subject Kunio but the direct object Kenta was hit. In order to explain why it is the case here, Akatuka first assumes that "a class of Japanese verbs of physical contact are really 3-place predicates in the deep-LL er level". By the "3-place predicates", Akatuka seems to imply that the verb in this category obligatorily requires the actor, the one who is affected by the action, and the specific body part the action affects to be in the deeper level. The two transformational rules we are primarily concerned with 61 here are based on this assumption. Let us consider the following (56) in which the under lying structure of (5*.b) and (5*.d) are given as (56.a) and (56.b) respectively. (56) a. b. Kenta In the above schematization, three places in question are occupied by NPj, NP2 and NP3. According to Akatuka, Inalienable-Possessor-Deletion is responsible for the derivation of (5*.b) and (5*Kd) from (56.a) and (56.b) respectively, deleting the direct object (i.e. NP2). For 62 the derivation of (54.a) and (5*..c), another transformational rule, Unspecified-Body-Deletion, is to he applied to the under lying structures in (56). The application of the rule yields (5*.a) and (54.c) from (56.a) and (56.b) respectively, deleting NP3 above. The Like-NP Constraint, then, discards (5*.a) as ungrammat i cal. Akatuka also accounts for the ungrammaticality of the fol lowing (58.b) along the same lines as above. The sentences in (57) and (58) are all cited from Akatuka. (57) a. Tanaka wa Satoo o nagutta. hit "Tanaka hit Satoo." b. Satoo wa Tanaka ni nagur-rare-ta. by hit (Pass) "Satoo was hit by Tanaka." (58) a. Tanaka wa Satoo no atama o nagutta. •s head hit "Tanaka hit Satoo on his head." b. *Satoo no atama wa Tanaka ni nagur-rare-ta. •s head by hit (Pass) "(Lit.) *Satoo's head was hit by Tanaka." 63 (58) c. Satoo wa Tanaka ni atama o nagur-rare-ta. by head hit (Pass) "Satoo was hit on his head by Tanaka." d. Satoo^wa Tanaka ni zibun^no atama o nagur-by self 's head hit rare-ta. (Pass) "(Lit.) Satoo. was hit on self^'s head by Tafiaka." 1 (57.b) is the passive sentence derived from its active counter part (57.a). To yield the passive sentence (57.b), the subject-object inversion took place and, then, the passive marker rare o was attached to the main verb naguru •hit'. This same deriva tional process, however, cannot result in the grammatical passive version of (58.a). Either (58.c) or (58.d) is really the correct passive sentence of (58.a). According to Akatuka, in the "deep er level" of (57.a) and (58.a), which Doth share the 3-place predicate naguru 'hit' as shown in the following schematic rep resentation, they have the underlying structure in common. 64 (59) S Satoo Inalienable-Possessor-Deletion deletes NP2 in (59) to result in (58.a), while Unspecified-Body-Deletion is responsible for the derivation of (57.a), deleting NP-j of (59). As for (58.c) and (58.d), their derivational process would be illustrated in the following way. (60) a. Satoo wa Tanaka ni Satoo no atama o nagur-rare-ta. by 's head hit (Pass) "(Lit.) Satoo was hit on Satoo's head by Tanaka." b. SatoOjWa Tanaka ni zibun.no atama o nagur-rare-ta. by self 's head hit (Pass) "(Lit.) SatoojWas hit on self's head by Tanaka." c. Satoo wa Tanaka ni 0 atama o nagur-rare-ta. by head hit (Pass) - .  . •'; •9'' • "(Lit.) Satoo was hit on 0 head by Tanaka." The application of passivization to (59) results in (60.a). 65 Reflexivization changes Satoo in Satoo no 'Satoo*s* into zibun 'self* with the subject noun phrase, Satoo. being the antecedent in (60.b). When ^-pronominalization operates in stead of reflexivization, (60.c) results. Thus, it is claim ed by Akatuka that the above-mentioned peculiar phenomenon in Japanese passivization can be accounted for only when we as sume that "a class of Japanese verbs of the physical contact are really 3-place predicates in the deeper level". Now, with the information from the preceding observation, we shall examine Akatuka's two transformational rules and constraint so that we can present some counter-examples and counter-arguments to them. 2.1.3 Examination of Akatuka*s two rules and constraint First, consider the following sentences which are the crucial counter-examples to the Like-NP Constraint. (6l) a. *Konboo o motta Kunio.wa zibun.o tataita. •L mm. stick had self hit "(Lit.) Kunio± who had a stick hit self^" 66 (61) b. *Okane o motta Kunio ^wa zibun^ni kuruma o katta. money had self to car bought "(Lit.) Kunio. who had money bought self. a car. •" Compare the above with the sentences in (48). (6l.a) and (6l.b) differ from (48.a) and (48.b) respectively only be cause the subject noun phrase is the head noun of the relative construction, whereas the subject of (48) is the maximum noun phrase. Recall that the reflexive pronoun cannot occupy the direct/indirect object position if it is in peer relation with the subject nounnphrase. Therefore, the sentences in (6l) must be blocked as unacceptable. The Like-NP Constraint, however, does not hold in (61). For, the constraint is valid only when the two coreferential noun phrases are maximum noun phrases. By showing the relevant part of (6l.a) and (6l.b) schematical ly, we can demonstrate this point more precisely. (62) a. konboo motta stick had 67 (62) b. Kunio, NP okane money motta had Notice, in the above, that neither NPg nor NP^ of NPlt the subject noun phrase of (6l), is the maximum noun phrase. The notion 'peer1 is the one defined between two maximum noun phrases as we observed in regard to the examples in (52). Hence, reflexivization may take place between NP2, the head noun phrase of NP^, and its coreferential noun phrase in the direct object position (i.e. 6l.a), or the indirect object position (i.e. 6l.b). As a result, the ungrammatical sen tences are produced. Thus;,' it is obvious that the Like-NP Constraint is not plausible enough to be responsible for the incorrect applica tion of Japanese reflexivization. Second, we observed that Akatuka's one ground for assum ing the "3-place predicate" stems from Japanese passivization. 68 In order that we can present the counter-argument to this account of the "3-place predicates", let us compare (57) and (58) in the previous section with what follows. (63) a. Kunio wa Kenta no inu o kakusita. *s dog hid "Kunio hid Kenta's dog." h. 'Kenta no inu wa Kunio ni kakus-rare-ta. 's dog by hid (Pass) "Kenta*s dog was hidden by Kunio." c. KentajWa Kunio ni zibun^no inu o kakus-rare-ta. by self 's dog hid (Pass) "(Lit.) Kenta. had self's dog hidden by Kunio." 1 d. Kenta-wa Kunio ni 0 inu o kakus-rare-ta. by dog hid (Pass) "(Lit.) Kenta had 0 dog hidden by Kunio." (64) a. Kunio wa Kenta no biiru o nonda. *s beer drank "Kunio drank Kenta*s beer." b. 'Kenta no biiru wa Kunio ni nom-rare-ta. *s beer by drink (Pass) "Kenta*s beer was drunk:: by Kunio." 69 c. Kenta.wa Kunio ni zibun.no biiru o nom-rare-ta. by self 's beer drink (Pass) "(Lit.) Kenta* had self!s beer drunk by Kunio." 1 d. Kenta wa Kunio ni 0 biiru o nom-rare-ta. by beer drink (Pass) "(Lit.) Kenta had 0 beer drunk by Kunio." (64) a/ Kunio wa Kenta no hon o utta. •s book sold "Kunio sold Kenta*s book." b. 'Kenta no hon wa Kunio ni ur-rare-ta. *s book by sell (Pass) "Kenta's book was sold by Kunio." c. Kenta^wa Kunio ni zibun.no hon o ur/-rare-ta. by self 's book sell (Pass) "(Lit.) Kenta* had self.'s book sold by Kunio." 1 d. Kenta*wa Kunio ni 0 hon o ur-rare-ta. by book sell (Pass) "(Lit.) Kenta had 0 book sold by Kunio." The above b-sentences are in marginal acceptance unless they are direct translations from a foreign language. Either the c-sentences or the d-sentences are acceptable passive sen-70 tences of the a-sentences. Notice, here, that the verbs in volved in the above examples are not the verbs of "physical contact". Nevertheless, we are able to observe the same be havior of Japanese passivization as that claimed by Akatuka in (57) and (58) previously t the ordinal subject-object inversion cannot result in grammatical passive sentences. Therefore, we must say that what Akatuka claims is not an isolated phenomenon on the verbs of "physical contact" but one observable in Japanese passivization in general t that is, under a certain condition, the person who is affected by the action identified by the verb may occupy the subject position by virtue of passivization.^ Akatuka's ground to assume the verbs of "physical contact" as the "3-place predicates" is thus highly ad hoc. Third, closer examination of the verbs of "physical eon-tact" also raises another serious counter-argument, which con sequently leads us to the conclusion that the constraint and the two rules by Akatuka must be discarded. Consider the following. 71 (66) a. *Kunio-wa zibun^o kaita. self scratched "(Lit.) Kunio.. scratched self^." b. Kunio.wa zibun.no senaka o kaita. self 's back scratched "(Lit.) Kunio^ scratched self^'s back." c. *Kunio wa Kenta o kaita. scratched "Kunio scratched Kenta." d. Kunio wa Kenta no senaka o kaita. *s back scratched "Kunio scratched Kenta's back." (67) a. *Kunio.wa zibun^o sasutta. self stroked "(Lit.) Kunio,^ stroked self^" b. Kunio^wa zibun.no ude o sasutta. self 's arm stroked "(Lit.) Kunio^ stroked self^'s arm." c. *Kunio wa Kenta o sasutta. stroked "Kunio stroked Kenta." d. Kunio wa Kenta no ude o sasutta. *s arm stroked "Kunio stroked Kenta's arm." 72 (68) a. *Kunio^wa zibun.o monda. self massaged "(Lit.) Kunio^ massaged self^." b. Kunio^wa zibun^no asi o monda. self ' s leg massaged "(Lit.) Kunio^ massaged self^'s leg." c. *Kunio wa Kenta o monda. massaged "Kunio massaged Kenta." d. Kunio wa Kenta no asi o monda. *s leg massaged "Kunio massaged Kenta's leg." What we must notice here is that all the verbs used in (66) through (68) are the verbs of "physical contact", yet they require the specific body part to be mentioned whether or not the action identified by the verb affects the subject noun phrase. In other words, all the c-sentences above are ungrammatical since the specific body part is not mentioned. Otherwise, they are grammatical as the d-sentenees show. Also, compare (54.c), which is grammatical, with the c-sen tences above, which are ungrammatical. The specific body part does not have to be mentioned in (54.c), whereas it has to be 73 mentioned in the c-sentences. Therefore, the ungrammatically of (54.a) in the previous section is not due to what (55) says, for some verbs seem to semantically and obligatorily require the specific body part to be mentioned and some do not, regard less of the syntactic function of the noun phrase to be affect ed by the action identified by the verb. In addition, Akatuka's classification of the verbs of "physical contact" itself is not clear. Observe the sentences in (69), (70) and (71). (69) a. *KuniOjWa zibun^o aratta. self washed "(Lit.) Kunio*^ washed self" b. Kuniofa zibun.no senaka 0 aratta. self *s back washed "(Lit.) Kunio^ washed self s back." c. *Kunio wa Kenta o aratta. washed "Kunio washed Kenta." d. Kunio wa Kenta no senaka 0 aratta. •s back washed "Kunio washed Kenta*s back." 7* (70) a. *Kunio.wa zibun^o tumanda. self held -(Lit.) KuniGj^ held self b. Kuniofa zibun.no hana o tumanda. self *s nose held "(Lit.) Kunio£ held self's nose." c. *Kunio wa Kenta o tumanda. held "•Kunio held Kenta." d. Kunio wa Kenta no hana o tumanda. •s nose held "Kunio held Kenta's nose." (71) a. •Kunio.wa zibun^o fusaida. self covered "(Lit.) Kunio^ covered self^." b. Kunio^wa zibun^no kuti o fusaida. • self 's mouth covered "(Lit.) Kunio^ covered self's mouth." c. •Kunio wa Kenta o fusaida. covered "•Kunio covered Kenta." d. Kunio wa Kenta no kuti o fusaida. 's mouth covered "Kunio covered Kenta's mouth." 75 We are not sure if the verbs involved in the above examples fall under "a class of the verbs of physical contact" or not. We may say, however, that in (69) through (71), the specific body part must be mentioned when the verbs in question are used in the sense of "physical contact". In other words, the verbs at issue are examples which semantically require the three arguments, regardless of the noun phrase to be affected by the action identified by the verb. Now, recall that Akatuka assumes the following to block the a-sentences in (66) through (71)• (i) A class of verbs of physical contact must have three arguments in the deeper level (ii) Unspecified-Body-Deletion transformation (iii) Inalienable-Possessor-Deletion (iv) The Like-NP Constraint If we adhere to Akatuka, in (66) through (71), rule (ii) is responsible for the blocking of the a and c-sentences, while the b and d-sentences are derived by the transformational rule (iii). Then, the constraint (iv) marks the a-sentences 76 as ungraramatical. Notice, however, that the ungrammatical c-sentences are to be left unmarked. Thus, if we were to follow the above-mentioned (i) - (iv), the ungrammatical sen tences result. To sum up the examination, we may conclude that closer examination of Akatuka's constraint and two related;trans formational rules suggests that these are inadequate for the foregoing three reasons. Therefore, it is necessary for us to discard them entirely. Our examination, however, has itself given no solution to the problem of why the reflexive pronoun cannot emerge in the direct/indirect object position. In the following sections, we shall reconsider the inadequacy of the alleged reflexive zibun 'self*, so that we are able to propose an alternative solution to the problem which remains unsolved. 2.2 THE GENUINE REFLEXIVE AND THE REVISED PROPOSAL (II) First, in 2.2.1, we shall treat the zibunzisin-form as another candidate for a Japanese reflexive. Then, in 2.2.2, to justify the zibunzisin-form as the genuine reflexive, the 77 treatment of the alleged reflexive zibun 'self* is to be discussed, which is virtually to suggest that reflexiviza tion in Japanese be a phenomenon of the simplex sentence. The result of the following sections will be presented as the revised proposal (II) and as a conclusion to the chapter. 2.2.1 Another Candidate For The Reflexive Pronoun Let U3 first consider the examples below to get the basic idea of the bound morpheme -zisin which would be crucial in order to tackle the problem left unsolved in the previous section. (72) a. Kunio ga syatyoo ni natta. president to became "Kunio became the president (of a company)." b. Kunio-zisin ga syatyoo ni natta. president to became "Kunio himself became the president (of a company)." (73) a. Kunio ga Kenta ni syukudai 0 yar-sase-ta. by homework do (Caus) "Kunio made Kenta do the homework." 78 (73) b. Kunio ga Kenta-zisin ni syukudai o yar-by homework do sase-ta. (Caus) "Kunio made Kenta himself do the homework." The b-sentences convey the contrastive meaning in comparison with the a-sentences. Thus, the bound morpheme -zisin func-tions as a sort of emphasizes, giving the word to which -zisin is attached the contrastive meaning. Notice, also, that the non-contrastive sentences (i.e. a-sentences) must be perfectly grammatical before the bound morpheme is attached. Now, bearing the above information in mind, compare the following sentences in (7*0 with those in (48) which is cited here again from the previous section. (48) a. •KuniOjWa zibun.^0 tataita. self hit "Kunio hit himself." b. *KuniOjWa zibun.ni kuruma o katta. self to car bought "Kunio bought a car to himself." 79 (74) a. Kunio.wa zibunzisin^o tataita. hit "Kunio hit himself." b. Kunio.wa zibunzisin.ni kuruma o katta. "Kunio bought a car to himself." Example (74) is only different from (48) in that the alleged reflexive zibun-form is replaced with the zibunzisin-form. What is important here is that (48) must be blocked as un grammatical, .while (74) is perfectly well-formed. Remember that Akatuka accounts for the ungrammatieality of (48) by say ing that the zibun-form in (48) is unacceptable because it is "in peer relation" with its coreferential noun phrase. Being "in peer relation" with its coreferent, however, the zibunzisin-form is perfectly acceptable in (7*). Why is it the case here? Furthermore, it is unlikely that the zibunzisin-form in (7*) can be treated along the same line as that in (72) and (73) I namely, as a compound word. For, there exists no non-contrast ive version for (7*) i possible non-contrastive sen tences are ungrammatical as shown in (48). The treatment of the zibunzisin-form. therefore, must differentiate from that 80 of the bound morpheme -zisin* The above fact would suggest that the ungrammatically in question should not be attributed to the syntactic relation, such as the peer relation of two coreferential noun phrases, and also that the adequacy of the zibun-form as the reflexive pronoun be reconsidered. Therefore, let us assume for the time being that the reflexive pronoun in Japanese is not the alleged ^zi^un-form but the zibunzisin-form. Hence, the ungrammatically of (48) is not due to the violation of such a syntactic constraint as the Like-NP Constraint, but is simply due to the inadequacy of the zibun-form as the reflexive pronoun. If the reflexive pronoun is really the zibunzisin-form. we are able to account for the following otherwise ignored phenomenon quite systematically in exactly the same manner as above. Consider the following examples, first. (75) a. Mizu^wa sorezitai^ni tetubun o fukunde iru. water itself in iron contain is "(Lit.) Water± contains iron in itself^" 81 (75) b. TaiyoojWa sorezitai.kara hikari o dasu. the sun itself from light emit "(Lit.) The sun. emits the light from itself'" The sorezitai-form in these examples is the non-human counter part of the zibunzisin-form. (Henceforth, zibunzisin and sorezitai are translated as 'oneself* and 'itself* respec tively for ease of reference.) Although it is used less fre quently than zibunzisin 'oneself, sorezitai 'itself* in Japanese behaves in exactly the same fashion as its human counter-part j that is, sorezitai * itself* in (75) is co-referential with the left-most non-human noun phrase which also commands sorezitai 'itself*. The following (76) re presents this more clearly. (76) a. S tetubun fukunde iru iron contain is mizu. water sorezitai. itself 82 (76) b. taiyoo. NP2 1 y\ NP the sun taiyoo. hikari dasu the sun light emit sorezitai. itself 1 In the diagrams, NPi is the left-most noun phrase commanding NP£» sorezitai 'itself. Hence, NPX and NP2 perfectly meet, reflexivization with NPA being the antecedent (cf. 47). (Incidentally, these sentences must be blocked as ungrammati cal if we adhere to Akatuka*s Like-NP Constraint, since the two noun phrases are in peer relation.) Again, it must be noticed that the claim that sorezitai 'itself is a compound word, consisting of the demonstrative sore 'it' and the bound morpheme -zitai does not hold 1 that is, the bound morpheme can be attached only to the grammatical sentences in order that the sentences to which -zitai is attach ed convey the contrastive meaning. The following (77) and (78) show an example. 83 (77) a. Mizu ni eiyoo ga aru. water in nutrition is "(Lit.) Nutrition is in water." b. Mizu-zitai ni eiyoo ga aru. water in nutrition is "(Lit.) Nutrition is in water itself." (78) a. Kunio no zizyoden ga rippana syoosetu da. •s autobiography excellent novel is "Kunio*s autobiography is an excellent novel." b. Kunio no zizyoden-zitai ga rippana syoosetu da. *s autobiography excellent novel is "Kunio*s autobiography itself is an excellent novel." Thus, sorezitai 'itself* behaves just like its human counter part, zibunzisin 'oneself*. So long as we treat the zibun-form as the reflexive, we are bound to fail to realize this relation between zibunzisin * oneself* and sorezitai 'itself, which share the above-mentioned common characteristics. In other words, unless we reject the notion to treat the zibun-form as the reflexive, we cannot cope with the non-human re flexive pronoun. 84 To sum up, as an alternative proposal to Akatuka's Like- NP Constraint, we have so far claimed that the genuine re flexive pronoun in Japanese is not the alleged zibun-form but the zibunzisin-form. Furthermore, should it be so, the otherwise ignored sorezitai 'itself can be treated in exact ly the same manner as zibunzisin 'oneself. Then, what is the zibun-form? This question will be the main topic in the following section, in order that.we can further elaborate our proposal. 2.2.2 The Treatment Of Zibun 'self If the reflexive pronouns are really zibunzisin 'one self and sorezitai 'itself in Japanese, the reason why we have the zibun-form such as in (80) is our main concern in this section. (80) Kunio^wa Kenta ni zibun^o waraw-rare-ta. by self laugh at (Pass) "(Lit.) Kunio.^ had Kenta laugh at self First, observe the sentences in (81) through (84). Watasi.wa Kenta ni I to watasi^ ga iku to itta. I zibun. self 2 go that said "I said to Kenta that I would go." Anata.wa Kenta ni you to anata. ga iku to itta. you zibun. self 1 go that said •You said to Kenta that you would go." Kunio.wa Kenta ni fkare^ ga iku to itta. to \ he go that said zibun. self 1 "Kunio said to Kenta that he would go." Watasi^wa watasi, no hon o yonda. j 1 •s book read zibun. self 1 "I read my book." Anata.wa you anata£ no hon o yonda, you zibun. self 2 's book read •You read your book." Kunio ^wa kare. no hon o yonda. he zibun. self 1 "Kunio read his book." 86 (83) a. Watasi^wa Kenta ni by watasi.o syookai-sase-ta. introduce (Caus) zibun. self "I made Kenta introduce me." b. Anata^wa Kenta ni you by J anata^ o syookai-sase-ta. you introduce (Caus) zibuni Self-1 "You made Kenta introduce you." Kunio.wa Kenta ni by kare. o syookai-sase-ta, i l he zibun. self 3 introduce (Caus) "Kunio^made Kenta introduce him^." (84) a. Watasi.wa Kenta ni I by I " watasi. ni kuruma o kau I to car buy zibun. self 1 koto o nattoku-sase-ta. that agree (Caus) "I made Kenta agree to buy me a car. / anaxa. to car b. Anata-wa Kenta ni you by « anata^ ni kuruma o kau you zibun. self 2 buy koto o nattoku-sase-ta. that agree (Caus) "You made Kenta agree to buy you a car. 87 (84) c. Kunio.wa Kenta ni by < kare. ni kuruma o kau he -to car buy zibun. self 1 koto o nattoku-sase-ta. that agree (Caus) "Kunio*, made Kenta agree "to buy him^ a car." As shown above, the zibun-form is fully interchangeable^ with the personal pronouns' without changing the meaning of the sentences. The zibun-form. however, differentiates from others in the following two aspects. (i) Zibun 'self* refers to the first, the second and the third person as shown in (81) through (84). (ii) Also, zibun 'self* is entirely free from a major feature of the Japanese language, the speech level. • o (It is not the case with the other pronouns.) In relation to what the above (ii) implies, it would be easier to observe the examples below. (85) a. (Watasi^wa) ima now watasi. no uti de gozaimasu. I •s house at be(formal) zibun* self 1 *bre^ I (informal) "I am at my home now." 88 (85) a*. (Ore-wa) ima I now ore. I 1 zibun. self 1 •watasi. no uti da. *s house be(informal) I (formal) "I am at my home now." b. (Anata-wa) ima now anata. no uti ni irassyaimasu ka? y . *s house at be(formal) zibun. self x •omae. you (informal) "Are you at your home now?" b'. (Omae*wa) ima you now omae- no uti ni iru ka? vou 's house at be(informal) zibun. self 1 •anata*(formal) I you 'Are you at your home, now?" c. (Ano kata.wa) ima that person now ano kata- no uti ni that per son , ,_ ' s house at zibun* self;1 •aitu*^ .that person(informal) irassyaimasu. be(formal) "That person is at his home now." 89 (85) c». (AitUjWa) ima that person now -aitu. no uti ni iru. that person . 's house at be (informal) *ang kata l that person(formal) That person is at his home now. ti In Japanese the verb of a sentence generally controls the the verbs in formal form, whereas the a*, bf, and c'-sentences have the informal form. The zibun-form can be co-occurrent o with both forms, but the others cannot.' In addition to the above observation, we oare able to present four more reasons to postulate the zibun-form as a personal pronoun, rather than a reflexive pronoun. First, consider the following sentences by Hirakouji (1973*17-18). All sentences are cited here with minor changes. (86) a. Zibun wa sanzyuusan-sai de arimasu. speech level. In the above, the a, b, and c-sentences have self thirty-three-years copula 11 I am thirty-three years old. tt a'. Zibun-zisin wa sanzyuusan-sai de arimasu. self thirty-three-years copula "(I don't know about others, but) I am thirty-three years old." 90 (86) b. Zibun wa siai-no-tame kesseki simasita. self game-because of absent did "I was absent (from the class) because of the game." b*. Zibun-zisin wa siai-no-tame kesseki simasita. self game-because of absent did (I don't know about others, but) I was absent (from the class) because of the game." Hirakouji treats zibun 'self* in (86.a) and (86.b) as jargon, differentiating them from those in (86.a') and (86. b*). Hirakouji*s explanation (1973»18) readst (87) "They(86.a-b) used to be uttered by people who belonged to the army. Some students who commit themselves to a university sports club which places them under a strict discipline often use sentence like (2)....The zibun*s in (3) and (4) are not necessarily such jargon. They are used in a ordinary conversation." In the quotation, (2), (3) and (4) corespond to (86.b), (86.a*) and (86.b*) respectively. Notice that we can plausibly account for the zibun-form above, postulating zibun 'self* as a personal pronoun ; that is, although it is jargon, zibun *self* in (86.a) and (86.b) is a personal pronoun which is free from the speech level. (In fact, zibun 91 •self* to refer to the first person was employed in the army in order to maintain the simplest way of communication with out the formality. Also, it is used in a university sports club or the like where the army-like strict order is kept.) As to (86.a*) and (86.V), they are the contrastive versions of (86.a) and (86.b) (cf. ?2 and 73). Second, consider the following examples. (88) a. Zibun wa doo desu ka? self how is "How about you?" b. Zibun ga kinasai t self come "(You) cornel" In these examples, the zibun-form refers to the second person. Unless zibun 'self is treated as a personal pronoun, this occurrence of zibun 'self must be differentiated from the other occurrences. However, as shown in (i) previously, that zibun •self refers to the second person as a personal pro noun is quite predictable. Thus, with zibun 'self as a personal pronoun, we can give a systematic account for both 92 (86) and (88) in exactly the same manner. Third, observe the following causative sentences. (89.a) is Akatuka's. (89) a. Taroo.wa Ziroo.ni zibun.».o nagur-sase-ta. l 0 1 3 •••• \ by self hit (Caus) "(Lit.) Taroo.made Ziroo hit self.." b. Kunio.wa Kenta^ni zibun^..o kak-sase-ta. by self paint (Caus) "(Lit.) Kunio^ made Kenta paint self^" c. Kunio4wa Kenta.ni zibun.o syookai-sase-ta. i 3 1*3 # . by self introduce (Caus) "(Lit.) KuniOj^ made Kenta introduce self.^" In all the above causative sentences, the zibun-form only refers to the subject of the matrix sentence. Akatuka attribute the ungrammatical reading (i.e. the one indicated by the index ji.) to the violation of the Like-NP Constraint. for NP2 and NP3 in the following schematization of (89.a) are in peer relation. 93 (90) Taroo. nagur •Ziroo1 hit However, we have observed in the previous section that zibunzisin •oneself can be in peer relation with its ante cedent. Hence, (91) is ambiguous where the zj,bun.-form in (89) has been replaced by zibunzisin •oneself. (91) a. Taroo^wa Ziroo ..ni zibunzisin-jO nagur-sase-ta. by oneself hit (Caus) "(Lit.) Taroo*^ made Ziroo j hit self.^." b. KuniOjjWa Kenta^ni zibunzisin^o kak-sase-ta. by oneself paint (Caus) "(Lit.) Kunio. made Kenta. paint self..." c. Kunio^wa Kenta ..ni zibunzisin^ ..o syookai-sase-ta. by oneself introduce (Caus) "(Lit.) Kunio. made Kenta. introduce self..." 94 In the above, unless the zibun-form is treated as a personal pronoun on the one hand and the zibunzisin-form. on the other hand as a reflexive, the ambiguity of (91) . cannot ibe accounted for systematically. Notice that we could account for the ambiguity in question quite systematical ly in the following way, using the underlying structure of (91.b) in terms of schematization (i.e. 92). (92) The derivation of the sentence indicated by the index 2 in (91.b) would bet (i) Reflexivization — NP3 gets reflexivized under the coreferential identity with NPg. being changed into zibunzisin 'oneself in the S2 cycle. 95 (ii) Causativization — kak 'paint' is combined with the causative sase, forming kak-sase. Also, be cause of this, the S2 node is to be deleted, (iii) As a result of (ii), the sentence with the index X is derived having NP2 and NP-j as its constituents, When NPi and NP2 are coreferential, the sentence indi cated by i results in the following way» (i) Nothing happens in the S2 cycle. (ii) Causativization — Same as above. 11 (iii) Pronominalization — NP-j, now a constituent of the matrix sentence Si, gets pronominalized. As a result, (89.b) is derived, (iv) -zisin-attachment — If the bound morpheme -zisin is attached to the result of (iii), the sentence with the index i in (91.h) results. Notice that the ambiguity of (91.h) is just a fortuitous matter, for the bound morpheme -zisin could be attached to NPi or NP2, resulting in the following grammatical (93.a) or (93.b) respectively. 96 (93) a. Kunio^-zisin wa Kenta ni zibun^o kak-sase-ta. by self paint (Caus) "(Lit.) Kunio^ himself made Kenta paint self^." b. Kuniojwa Kenta-zisin ni zibun-o kak-sase-ta. by self paint (Caus) "(Lit.) Kunio^ made Kenta himself paint self^." Therefore, it is of importance to notice that the sen tence with the reflexive pronoun is only the one indicated by the index j. in (91), while the reading with i is accept able only when it conveys the contrastive meaning. In other words, reflexivization in Japanese is a phenomenon observable only in the simplex sentence. Moreover, this would also ex plain why the reading with the index i, in (91) is the dominant one. What we have claimed above is not only peculiar to the causative sentences. Consider the following examplesi (94) a. Kunio.wa zibun.<4o tataita Kenta.o niranda. l i 3 j self hit stared at "(Lit.) Kunio^ stared at Kenta who hit self^." a. Kunio^wa zibunzisin^© tataita Kenta^o niranda. oneself hit stared at "(Lit.) Kunio*^ stared at Kenta^ who hit self.j." 97 (9*) b. Kunio.wa zibun^#io kaita Kenta^0 tataeta. self painted admired "(Lit.) Kunio.- admired Kenta who painted self" b'« Kunio^wa zibunzisin^o kaita Kenta o tataeta, oneself painted admired "(Lit.) Kunio. admired Kenta. who paint ed self." J c. KuniojWa zibun.*.o syookaisita Kenta ni self introduced to tikazuita. approached "(Lit.) Kunio. approached Kenta who introduced self-V' c'. Kunio-wa zibunzisin^^o syookaisita Kenta ni oneself introduced to tikazuita. approached "(Lit.) Kunio. approached Kenta* who introduced self." The diagram (95) below represents the underlying structure that the sentences in (9*) have in common. 98 (95) Kenta. NP< Kunio. Kenta1 j In the above diagram, NP2 is the relative clause with NP3 being its head noun. When NP5 is coreferential with NP^, reflexivization takes place in the S2 cycle, yielding the a', b* and c'-sentences with the index j[, which is the dominant reading of them. Otherwise, with NP^ being its coreferent, NP5 gets pronominalized, resulting in the a, b .and c-sentences. The -zis in-attachment is responsible for the a1, b* and c^-sentences with the index i, which are one case of the contrastive versions of the a, b and c-sentences. Hence, the ambiguity is once again coincident. The fourth reason to justity our treatment of the zibun« form as a personal pronoun comes from the following passive sentences* 99 Compare (96) with (1). (1) a. Kunio.ga Kenta^o zibun^^no uti de korosita. self *s house at killed "(Lit.) Kunio-Jeilled Kenta at selfs house." b. Kenta^wa Kunio^ni zibun .^no uti de by self 's house at koros-rare-ta. kill (Pass) "(Lit.) Kentai was killed by Kunio at self., *s house. (96) a. Kunio.wa Kenta.ni zibun.*.o syookais-by self introduce rare-ta. (Pass) "(Lit.) Kunio*^ had Kenta introduce self" b, Kunio.wa Kenta^ni zibunzisin..0 syookais-by oneself introduce rare-ta. (Pass) "(Lit.) Kunio^ had Kenta,. introduce selfy." Recall that (l.b) was crucial to propose the sub.iect- antecedent condition in the previous chapter (cf. 1.1.1) , since in (l.a) and its passive (l.b) zibun •self* only refers 100 to the subject noun phrase. Hence, the application of passiv ization must precede reflexivization so long as we treat the zibun-form as the reflexive pronoun. However, it does not seem to be the case with (96.b), since zibunzisin •oneself* refers back to both Kunio and Kenta. To account for this, we assume the following deriva tional processes, using the underlying structure of (96). (97) Si The sentence with the index j. in (96,b) would be derived in the following way: (i) Reflexivization — NPjj, is to be changed into zibunzisin 'oneself*, being coreferential with NP2 prior to passivization. (ii) Passivization — NP3 is to be deleted by Equi-NP  deletion* Also, the verb syookais 'introduce' is combined with the passive marker rare. Now, NP2 and NPjj, are the constituents of the passive sen tence Slt the S2 node having been eliminated, (iii) The result of (ii) is the sentence with the index 1 in (96.b). The sentence with the index i in (96.b) would be derived in this way1 (i) Passivization — Same as above (ii). (ii) Pronominalization — NP^ being changed into the zibun-form under the referential identity with NPA, (96.a) results, (iii) -zisin-attachment — -zisin being attached to NPij., the sentence with index i in (96.b) results. Again, the ambiguity of (96.b) is just fortuitous, because the bound morpheme -zisin could be attached to NPA or NP2 rather than NP4, yielding the grammatical sentences, (98.a) 102 or (98.b) respectively. (98) a. KuniOj-zisin wa Kenta ni zibun.0 syookais-by self introduce rare-ta. (Pass) "(Lit.) Kunio. himself had Kenta introduce self" b. Kunio.wa Kenta-zisin ni zibun^o syookais-by self introduce rare-ta. (Pass) "(Lit.) Kunio, had Kenta himself introduce self" Thus, postulating zibun •self* as a personal pronoun on the one hand, and zibunzisin •oneself* as the reflexive on the other, we are able to account for the ambiguity very plausibly. Therefore, as a conclusion to this section we may say that what has been treated as Japanese reflexivization (by Akatuka, Oyakawa etc.) through the zibun-form really consists of two different syntactic phenomena » reflexivization in the simplex sentence and pronominalization in the complex sentence. 103 CONCLUSION Through the examination of Akatuka'S Like-NP Constraint and related two transformational rules, we have encountered the question of the adequate form of the reflexive pronoun in Japanese. In order to give a plausible account of the problem, we have postulated the zibun-form as a personal pronoun on the one hand, and the zibunzisin-form as the genuine re flexive on the other, with the following resultst (i) sorezitai 'itself, the non-human counter-part of zibunzisin 'oneself, can be treated as the reflexive pronoun, (ii) What has been classified as reflexivization in Japanese so far really consists of two different syntactic processes i reflexivization and pro nominalization. As a conclusion to this chapter, we may present what the foregoing discussion has resulted in as our revised proposal (II) in the following: 104 (99) The Revised Proposal (II) Japanese Reflexivization (i) In a given simplex sentence that dominates the two coreferential noun phrases, NPa and NPr, change the latter into zibunzisin 'one self (sorezitai 'itself) if NPaUs the left most human (non-human) noun phrase which also commands NPr. (ii) When NPa is in a higher sentence, change NPa into zibun 'self. (Pronominalization12) 105 NOTES FOR CHAPTER II 1. The deep structure of these sentences is also unaccept able as shown belowi i. *Kunio wa Kunio o tataita. hit -Kunio hit Kunio." ii. *Kunio wa Kunio ni kuruma o katta. to car bought "Kunio bought Kunio a car." In Chapter III, this point is to be discussed in detail. 2. By the notion "maximum NP", Akatuka seems to imply that an NP which is not dominated by another NP node is a "maximum NP". Therefore, the head noun of the relative construction is not a "maximum NP", and neither is the left-branching NP. 3. Noriko A. McCawley, "A Study of Japanese Reflexivization," (Unpublished doctoral dissertation, Illinois University 1972) P.30. 4. Ibid., p.34. 5. It is well-known that there are two types of passive sen tences in Japanese t the plain passive-derived by the subject-object inversion plus passive marker rare attachment and the so-called affective passive. What we are concerned with here is the characteristics of the latter type of passive in general, not peculiar to the verbs of "physical contact." 6. At present, the condition under which the personal pronouns are to be interchanged with zibun •self is not clear. 7. In Japanese, personal pronouns which refer to the third person are missing. Kare 'he' in (81) through (84) and kanozyo 'she* are sometimes classified as such personal pronouns. How ever, this classification is highly questionable since the gender is quite foreign in Japanese. 8. As for the discussion in detail about the speech level, see. Akiko Shinoda (1973)» for example. 9. In these examples, ano hito 'that person' and aitu 'that person* are not personal pronouns. Nevertheless, our main claim would remain in effect. 10. This, usage of zibun 'self* is more frequently observable in the Kansai dialect which is spoken in Osaka area. 106 11. The term 'pronominalization' in the present thesis only referstothe derivationof zibun 'self* in order to show that zibun 'self', and zibunzisin 'oneself are derived in a different manner. As for the pronominalization in Japanese, we leave its general formulation for future study. 12. Notice that this pronominalization is in the above-mentioned sense (cf. note 10). Also, (ii) in the revised proposal (II) is not the case with sore 'it/that' because it is not a personal pronoun. See the examples belowt i. Rekisi^wa sorezitai^o kurikaesu. history itself repeat "History repeats itself." ii. SensoOjWa rekisi.ga sorezitai***© kurikaesu war history itself repeat koto o simesu. that show "(Lit.) The war shows that history, repeats itself±." 1 iii. *Sensoo.wa rekisi ga sore.,0 kurikaesu koto o war history it repeat that simesu. show "(Lit.) The war*, shows that history repeats Although iii above might be acceptable as a direct translation from a foreign language, it is otherwise unacceptable. Hence, sore 'it/that' must be treated differently from zibun 'self. CHAPTER III SOME RESIDUAL PROBLEMS AND THE INTERPRETIVE THEORY 3 INTRODUCTION In the previous chapter, we claimed that the genuine re flexive in Japanese is not the zibun-form but the zibunzisin-forra, and that zibun 'self* and zibunzisin •oneself are the respective results from pronominalization and reflexivization. We must admit, however, that with this approach there still is a difficulty in determining the adequate deep structure to which the transformational rules apply. Consider the following, for example. (100) a. Kunio^wa zibunzisin-o tataita. oneself hit "Kunio hit himself." a*. *Kunio wa Kunio o tataita. hit "•Kunio hit Kunio." 108 (100) b. Kunio-wa Kenta ni zibun-^o tatak-sase-ta. by self hit (Caus) -(Lit.) Kunio*^ made Kenta hit self b*. *Kunio wa Kenta ni Kunio o tatak-sase-ta. by hit (Caus) "•Kunio made Kenta hit Kunio," In the present chapter, we shall look into a possible account for this in Jackendoff s proposal known as the In terpretive Theory. As a result, it will be shown that the approach in the framework of Jackendoff's proposal can shed light on the following two problems as well, which the stan dard transformational approach seems unable to account for. (i) The occurrence of zibun * self'with the particle de •by/with* (ii) Non-coreferential zibun 'self and zibunzisin 'oneself 3»1 The Interpretive Theory by Jackendoff In order that the coreferentiality can be interpreted in the semantic component, Jackendoff (1972ill2) formulates English reflexivization in the following wayi 109 (101) Reflexivization in English NPA (X'coref ^^^ef J *n the environment.... OBLIGATORY According to Jackendoff, this rule says thatt (102) "....in the proper contexts for reflexivization NP2 is coreferential with NPA if and only if it is reflexive." It is important to notice here (.that, the revised proposal (II) in the previous chapter being preserved,1 the rule (101) can remain in effect even in Japanese reflexivization s for example, the reflexive zibunzisin •oneself1 in (100.a) is to be interpreted as coreferential with Kunio. the left-most human noun phrase which commands the reflexive, by the rule (101). As the reflexive zibunzisin •oneself* is presented in the deep structure in the Interpretive Theory, such a ungrammatical deep structure as (100.a*) can be disregarded. Likewise, zibun *self' in (100.b) can be easily in terpreted as coreferential with its eligible antecedent Kunio by the rule something like follows* 110 (103) Pronominalization NPi o(coref £ of2pro| in the environmen,fc* • • OBLIGATORY In (lOO.b) both.. Kunio and Kenta. the left-most human noun phrase in the matrix sentence and in the embedded sentence respectively, will qualify as NP^ and zibun 'self* as NP2. As indicated in (lOO.b) the zibun-form can be only coreferen tial with Kunio. the matrix subject. Therefore, we must maintain all the conditions mentioned previously in the re vised proposal (II) as "the environment" in order for the rule to result in the correct interpretation. Using the revised proposal (II) as "the environment", reflexivization and pronominalization in Japanese may be formulated in the following way in terms of the Interpretive Theory t (104) a. Reflexivization TNP2 NPi (/coref ^£2, ] if /(i) NPi is the left-most noun phrase • commanding NP2 and (ii) both NPi and NP2 are in a given simplex sentence Ill (104) b. Pronominalization fNP2 — x —• NP, </coref ] if (i) same as above and (ii) NPA is in a higher sentence This formulation not only avoids such ungrammatical deep structures as in (100) but also accounts for the coreferen-tiality observable in the examples belowi (105) a. Kunio^wa zibun^de syukudai o sita. self by homework did "Kunio did the homework by himself." b. Zibun^de syukudai o sita koto ga Kunio-ni self by homework did that to zisin o ataeta. confidence gave "(Lit.) That (he) did the homework "by himself gave Kunio confidence." The zibun-form in (105) is a perfect instance of our anal ysis in the preceding chapters since the pronoun zibun •self* is coreferential with the left-most human noun phrase which commands zibun 'self as the index shows in (105). However, 112 the standard transformational rule by which one of the two coreferential noun phrases is to be changed into the zibun-form cannot be accountable for the occurrence of zibun 'self* in the above examples because the possible deep structures for the sentences in question are entirely unacceptable as shown below t (106) a. *Kunio wa Kunio de syukudai o sita. by homework did "•Kunio did the homework ^by Kunio." b. •Kunio de syukudai o sita koto ga Kunio ni by homework did that to "•(Lit.) That (he) did the homework :by. Kunio gave Kunio \\J.I confidence." It seems to be because of this difficulty that the pronoun zibun •self' with the particle de_ 'by/with* has been rarely treated in the current linguistic works in terms of the stan dard transformational approach.-' Contrary to this, the interpretive approach does not have to assume the ungrammatical deep structure (106). Ac cording to the rule (104), Kunio in (105) is to be qualified as NPi and zibun •self1 as NP2 in (104). 113 Next, we shall treat mere complicated examples. (107) a. Kunio .wa zibun*, de zibun.no syukudai o sita. self by self 's homework did "(Lit.) Kunio. did self**s homework by self*^" 1 b. Zibun^de zibun^no syukudai o sita koto ga self by self * s homework did that Kunio*.ni zisin o ataeta. to confidence gave "(Lit.) That (he) did aelf^'s homework by self* gave Kunio. r - confidence." 1 l (108.a) and (108.b) below are the schematical representation of (107.a) and (107.b) respectively. (108) a. S 114 (108) b. (Same as VP of 108.a) In these schematizations, pronominalization between NPX and NP3 is accountable by the standard transformational approach, in which NP3 is to be changed into zibun 'self*. But, pro nominalization between NPX and NP2 is to remain unaccounted for by the same approach because of the ungrammatical deep structures. The interpretive approach, once again, shows no difficul ty in explaining the antecedent-reflexive relation in (107). In both (108.a) and (108.b), the rule (104.b) gives the two interpretations as follows* (109) a. NPX(Kunio) +coref NP? (zibun'self) b. NPX(Kunio) +coref NP3(zibun'self) The interpretive approach thus accounts for such complicated instances as (107). whereas the standard transformational 115 approach seems unable to account for them. Jackendoff (19?2tll2) also imposes the following con dition on English reflexivization. (110) The Consistency Condition If the table of coreference marks two NPs coreferential, these NPs must in fact be able to describe the same in dividual. Jackendoff uses (111) below to show how the condition works. (111) a. *The boy^shot herself^. b. *Finkelstein^ shot yourself^. The reflexive rule marks the two noun phrases in (111) co-referential as indicated by the index i,. This interpreta tion, however, is to be rejected because of the Consistency  Condition whereby the male individual in (111.a) cannot de scribe the same individual as the female 'herself. The third person 'Pinkelstein* and the second person individual •yourself cannot be the same in (lll.b). In other words, the condition rejects (111.a) because of its gender dis agreement, whereas (lll.b) because of its person disagree-116 merit. Notice that this condition must be adopted into Japanese reflexivization in order to block such ungrammatical sentences as followsi (112) a. •Rekisifa zibunzisin^o kurikaesu. history oneself repeat "•History repeats oneself." b. •Kuniojwa sorezitai^o tataita. itself hit "•Kunio hit itself." It is obvious that the reflexive rule (104.a) cannot bloek the ungrammaticality of (112). The reflexive rule gives the following interpretations. (113) a. (Rekisi'history*) +coref (zibunzisin*oneself*) b. (Kunio) +coref (sorezitai*itself*) Then, it is the Consistency Condition that discards (113) above * that is, the non-human noun phrase rekisi 'history* cannot describe the same individual as the human noun phrase zibunzisin 'oneself in (113.a). Likewise. Kunio and sorezitai 'itself in (113.b) cannot be the same because of their human-117 ness disagreement. Thus, Japanese reflexivization too requires the Consist  ency Condition. To sum up, we may say that the Interpretive Theory is equipped with more descriptive adequacy than the standard transformational approach on the basis of the following two points* (i) Without assuming ungrammatical deep structures, the interpretive approach can account for such an instance as (100) plausibly, (ii) Zibun 'self* with the particle de 'by/with1 can be treated plausibly only when the zibun -form is assumed to be present in the deep structure. 3.2 Non-coreferential zibun*self and zibunzisin*oneself' First, let us observe (114) below. (114) a. Nihon-zin^wa zibunzisin#io kantan ni koros-eru. Japanese oneself easily kill can "Japanese can easily kill one's own self." 118 (114) b. Ano nihon-zin^wa zibunzisin.© kantan ni that Japanese oneself easily koros-eru. kill can "(Lit.) That Japanese can easily kill self^." Although it is hard to see from its English translation, (ll4.a) is acceptable only when zibunzisin •oneself is not corefer ential with nihon-zin •Japanese1 i namely, (114.a) does not mean that •Japanese can easily kill Japanese* but something to the effect that 'Japanese can easily commit suicide'. More examples of this type followi (115) a. Ningen^wa zibunzisin.© aisu. men oneself love "Men love one's own self." b. Ano ningen^wa zibunzisin^ 0 aisu. that man oneself love "(Lit.) That man (person) gloves self* (116) a. Kodomo^wa zibunzisin*^© kontorooru deki-nai. children oneself control can not "Children cannot control one's own self." b. Ano kodomo^wa zibunzisinjo kontorooru deki-nai. that child oneself control can not "(Lit.) That child*^ cannot control self" 119 In accounting for the zibunzisin-form in the a-sentences by the standard transformational approach, we shall encounter the two problems. For one thing, the possible underlying structures are ungrammatical as discussed in section 3 of this chapter and, for another, even if we admit the ungrammati cal underlying structures, zibunzisin 'oneself in the a-sentences above cannot be derived through the transformational reflexive rule for there are not two coreferential noun phrases for reflexivization to apply to. (Recall that (114.a),for example, does not mean 'Japanese can easily kill Japanese.') Now, compare the b-sentences with the a-sentences in (114) through (116). The coreferentiality observable in the b-sentences is acceptable as shown by the index. Why is it that zibunzisin 'oneself behaves differently in each example? In order to explain this phenomenon, we are proposing that a feature QfspecificTJ is needed in Japanese reflexiv ization. In (114) through (116), what differentiates the a-sentences from the b-sentences is the demonstrative ano •that* attached to the left-most human noun phrases of the former sentences. For example, the left-most human noun 120 phrase is nihon-zin 'Japanese' in general in (114.a), while the left-most human noun phrase in (114.0) is specific ano  nihon-zin 'that Japanese' and, as a result, only the latter sentence shows the acceptable coreferentiality. Notice that the parallel phenomenon can be observed in Japanese pronominalization t the zibun-form. (117) a. Nihon-zin-wa Japanese 4 rzibun,,.. no koto o hanas-nai •s affair talk not piihon-zin I Japanese "(Lit.) Japanese, do not talk one's.own affair."1 1 b. Sono nihon-zin^wa the Japanese (zibun. self 1 sono nihon-zin ^ the Japanese no koto o •s affair hanas-nai. talk not •'(Lit.) The Japanese, does not talk about self's affair." (118) a. Ningen^wa men 'azibu^.ga kiraina koto o su-nai. self hate thing do not •ningen men "(Lit.) Men do not do what one hates." 121 (118) b. Sono ningen.wa the man zibun. self 1 sono ningen the man ga kiraina hate koto o su-nai. thing do not "(Lit.) The man(person). does not do what self-^ hates." (119) Kodomo^wa children l zibun*. no kanzyoo ga kontorooru self , . ^ , •s temper control rkodomo children deki-nai. can not "(Lit.) Children cannot control one*! own temper." b. Sono kodomoiwa ( zibun. the child { I sono kodomo [ the child kontorooru deki-nai. control can not no kanzyoo ga 's temper "(Lit.) The child, cannot control self's temper." 1 In the above examples, if the eligible antecedent is a spe cific noun phrase, the zibun-form is coreferential (i.e. b-sentences), but otherwise zibun 'self* is non-coreferential 122 (i.e. a-sentences). Therefore, we propose such a condition as followsi (120) The Specific Antecedent Requirement For the zibun(zisin)-form to be coreferential, there must exist a noun phrase which describes a specific individual in a given sentence. Consider the interpretation procedure of (114) and (11?), for example, to see how the condition (120) validates itself. i) By the rules (104), the interpretations below are given to (114) and (117). <»*•••>• (SgiJUi2) » <S£tB8gK£>+coref 'liB1"' <w;.)i(«fs +ooref * ii) The condition (120) rules out (ll4.a) and (117.a). <U*..M<§igg£^) +coref 123 iii) As a result, (114.b) and (11?.b) are to be interpreted as coreferential. This procedure thus tells us correctly that the coreferen-tiality of the a-sentences in (114) through (119) is un acceptable. But, the acceptable reading of the sentences (i.e. non-coreferential reading) is still left unaccounted for. More precisely, (114.a) and (11?.a) can be interpreted only on the basis of -coref and zibun(zisin) *(one)self* must be interpreted as a noun, not a pronoun of any kind. In short, the following should holdt (121) - (u*.,),(^to, _eoref bV (117.a),(%^n) -coref Therefore, we must incorporate the condition (120) into Japanese reflexivization/pronominalization (104) as a third condition so that such an interpretation as (121) can be possible* 124 (122) a. Reflexivization NPi tfcoref [^|ef J it (i) Same as (104.a) and (ii) Same as (104.a) and (iii) NPi must be a noun phrase which describes a specific individual, b. Pronominalization NPi +coref [^rG] if (i) Same as (104.b) and ^(ii) Same as (104.b) and (iii) Same as above It is because of the violation of the third condition above that the interpretation (121) results from the rules (122). The following schematization represents the foregoing. (123) Reflexivization Pronominalization i NPi -coref NP2 NPj +coref NP2 i The Consistency Condition I e.g. (Il4.a),(115.a) (116.a),(117.a) (118.a),(119.a) etc. e.g. (100),(105),(107).1(111) *(112),*(ll4.a) etc. 125 Next, let us observe the following examples which would further support the incorporation of (120) into (104) as a third condition. (124) a. Gakusee. Wa zibunzisin*.o kadaihyooka suru. student x . oneself over-estimate do a Qakusee-tati4 * student(PITT "(Lit.) Students over-estimate one's own self." b. Gakusee. student b! Gakusee-tati . student (PI.)1 wa zibun#iga benkyoo siteir-self study not-doing nai koto ni kizuk-nai. that realize not "(Lit.) Students do not realize that one is not studying." (124.a) and (124.b) are different from (124.3*) and (124.b») in that the latter have the plural marker -tati attached to the subject noun phrase. As shown by the index i, zibunzisin •oneself* and zibun 'self* in (124) are not coreferential with the left-most noun phrase gakusee(-tati) * student(s)*, yet both examples in (124) are grammatical. It might be claimed that the Consistency Condition is responsible for the non-coreferentiality in (124.a*) and 126 (124,b') because gakuseetati 'students1 with the plural marker cannot describe the same individual as the zibun(aisin)-form. However, we must recall that the Consistency Condition is able to disqualify gakusee-tat i 'students' as NPA in (104), but is unable to give the non-coreferential interpretation for (124,a') and (124,b'). In other words, the combination of (104) and the Consistency Condition cannot derive the in terpretation of (125) below, <125) >• <'2^,)'(iS^» (^i^eTf-"' <^.»'>'<«!§K£2Bt*> -«-«' (Sir1) Contrary to this, the rules (122) with (120) being their third condition can correctly account for the non-coreferen-tiality of (124), According to (122), gakusee-tatft 'students' does ndt qualify itself for NPA because it is not a "specific individual", so that (122) results in the interpretation of (125.a) and (125.b) for (124.a') and (124,b') respectively. (Notice that (124.a) and (124,b) are also given the non-co-referential interpretation for the same reason,) The Interpretive Theory thus plausibly accounts for the 127 non-coreferentiality of the zibun and zibunzisin-form. Al though they seem to be equipped with more explanatory adequacy than the standard transformational approach, the interpretive rules (122) need further elaboration since there are such sen tences as followst (126) a. Kenta^dake wa zibunzisin^jjni toohyoo sita. only oneself vote did "(Lit.) Only Kenta.^ voted for self^." b. Kenta^hitori ga zibun^no kuruma de kita. alone self 1s car by came "(Lit.) Kenta^ alone came by self^x's car." These sentences are ambiguous j that is, both the coreferential and the non-coreferential readings are possible. For example, (126.a) could mean either that 'only Kenta voted for Kenta,* or that * although others voted for someone else rather than themselves, only Kenta voted for one's own self. In order to account for the ambiguity, the reflexive rule (122.a) must interpret(126.a) as both coreferential and non-coreferential, which is impossible in terms of the rule. Since we are not able to solve this problem for the present, we are simply pointing out its existence and leave it for future study. 128 CONCLUSION The intent of this chapter has been to treat some re sidual problems, a treatment which has been, in effect, a comparison of the standard transformational approach with the Interpretive Theory by Jackendoff. This comparison eventually suggests the precedence of Jackendoffs proposal over the standard transformational approach on the basis of the following points i i) Due to the difficulty in determining the adequate deep structures, the standard transformational approach cannot derive such a sentence as (100) plausibly, whereas the interpretive approach can systematically give the correct interpretation for the antecedent-reflexive relation, ii) For the same reason as above, the standard trans formational approach fails to account for the zibun-form plus the particle de 'by/with', whereas the interpretive approach accounts for it in the same manner as for the other zibun occurrences. 129 iii) The interpretive rules (122) can be responsible for non-coreferential zibun 'self* and zibunzisin •oneself* as observable in (114) through (119). 130 NOTES FOR CHAPTER III 1. With the information in the revised proposal (II), we shall propose reflexivization and pronominalization rules in terms of the Interpretive Theory()later as (104) and (122). 2. See note 11 in the previous chapter. 3. One might say that a plausible explanation is possible in the framework of the standard transformational approach, assum ing the underlying structure of (105) as followsi i. Kunio wa Kunio no tikara de syukudai o sita. *s ability by homework did "(Lit.) Kunio did the homework with Kunio*s ability." ii. Kunio no tikara de syukudai o sita koto ga *s ability by homework did that Kunio ni zisin o ataeta. to confidence gave "(Lit.) That (he) did the homework with Kunio's ability gave Kunio the confidence." The application of pronominalization results ini iii. a. Kunio.wa zibun.no tikara de syukudai o sita. self 's ability by homework did "(Lit.) Kunio. did the homework with self.'s ability." x b, Zibun.no tikara de syukudai o sita koto ga self 's ability by homework did that Kunio ni zisin o ataeta. to confidence gave "(Lit.) That (he) did the homework with self.'s ability gave Kunio^the confidence." Every native speaker of Japanese may accept iii as a paraphrase of (105). Therefore, what is needed to yield (105) from iii seems to bean optional transformational rule something like tikara-deletion which deletes tikara 'ability' in iii. However, we must say that this treatment is very ad hoc for the two reasons belowt First, the rule deletes the^lexical item tikara 'ability* rather arbitrarily. Second, and more important, iii above is 131 not the only paraphrase of (105). The sentences in iv, for example, could be also the paraphrase of (105.a). iv. Kunio.wa zibun.no selfx,s kangae de syukudai o sita, idea by homework did nooryoku rye faculty etc. "(Lit.) Kunio^ did the homework with self^/s idea/ faculty etc. " Thus, in order to yield (105.a) from both iii and iv, we must increase such a rule as tikara-deletion up to as many as three (or presumably more). Hence, fthis approach cannot be far from being ad hoc. 4. The rule (104,b) actually gives a third interpretations NP2(zibuja'self •) -coref NP3(ii1bjin.,self1) Notice that in (108.a) NP2 is not the left-most noun phrase and alsothat in (108.b) NP2 and NP3 are in the same simplex sentence.' CHAPTER IV CONCLUSION In Chapter I and Chapter II, we have examined Oyakawa's and Akatuka's hypothetical treatments of Japanese reflexiv ization, resulting in a revised proposal. The revised proposal virtually suggests thati 1) Japanese reflexivization is a phenomenon in a simplex sentence with the zibunzisin-form being the genuine reflexive, otherwise— a. the violation of Akatuka's Like-NP Constraint (cf. 48 and 7*) cannot be accounted for plau sibly b. the ambiguity of certain type of sentences (cf. 96) is unaccountable c. the same behavior of the non-human reflexive sorezitai 'itself* as its human counter-part zibunzisin 'oneself cannot be coped with (the humanness condition ho longer holds in Japanese reflexivization), 2) as a result of the foregoing, the treatment of the 133 zibun-form must be different from that of the genuine reflexive pronouns. 3) the antecedent is the left-most noun phrase which commands zibun 'self or zibunzisin(sorezitai) 'one self (itself )' , and, as a result— a. the subject-antecedent condition is one special instance of the above-mentioned condition b. the highest human NP condition is also a specific case where the antecedent is human and does not meet the subject-antecedent condition c. there are some sentences in which neither the high  est human NP condition nor the subject-antecedent  condition can give a plausible explanation (cf. 32 and 34), whereas the revised proposal can. 4) the linear order (as well as the hierarchical order) of the two coreferential noun phrases is crucial in predetermining Forward and Backward operation of the rule, whereas Oyakawa discards the linearlity as ir relevant in Japanese. 13* In addition, comparing the standard transformational ap proach with Jackendoff*s interpretive approach in the last chapter, we could clarify thati 5) in the Interpretive Theory, the ungrammatical deep structure can be avoided so that— a. such sentences as (7*0 and (75) are accountable with no difficulty b. the zibun-form with the particle de 'by/with* can be accounted for as an instance of pro nominalization (cf. 106, 107 and 108) 6) there exists the non-coreferential zibun(zlsin)-form which requires the J~+ specific] feature to be attached to the antecedent (cf. 114 through 119) 7) there also exists the ambiguity for which a rule must give both the coreferential and the non-coreferen tial interpretations at the same time (cf. 126). (At present, we are not able to account for this ambiguity, therefore, the existence of the problem is simply point ed out as warranting further investigation.) 8) as a result of the comparison, the interpretive ap-135 proach is more advantageous as far as the foregoing points are concerned. In order to conclude the present thesis, the revised proposal is recapitulated in terms of the diagram belowt y commands^ NPa<^ and/or /NPr precedes NPa C+specif icj is coreferential with NPr NPa and NPr are in the same simplex sentence NPr zibunzisin oneself B-humanJ NPr* -k sorezitai nrr -» itself (rhuman] (Reflexivization) NPa G-specif icj is not coreferential with NPr NPa is in a higher sentence NPr zibun self (Pronominalizat ion) 136 BIBLIOGRAPHY Akamajian, Adrian, and Chisako Kitagawa. "Pronominalization, Relativization, and Thematization i Inter related System of Ooreference in Japanese and English," University of Massachusetts, (Reproduced by the Indiana University Linguistic Club), 197*. Akatuka, Noriko. "A Study of Japanese Reflexivization," Unpublished Ph. D. dissertation, Illinois University, 1972. Hasegawa, Kinsuke. "Transformations and Semantic Interpre tation," Linguistic Inquiry. Vol. 3, No. 2 (1972), pp. 141-159. Hirakouji, Kenji. "Jibun Forms in Japanese," Papers in Japanese Linguistics. Vol. 2. No. 2 (1973) PP. 17-43. Jackendoff, Ray S. "An Interpretive Theory of Pronouns and Reflexives," Massachusetts Institute of Technology, (Reproduced by the Indiana University Linguistic Club), 1968. Semantic Interpretation in Generative Grammar. Cambridge, Massachussets 1 The MIT Press, 1972. Kuno, Susumu. Notes on Japanese Grammar. Mathematical Linguistics and Automatic Translation, Report No. NSF-27. Cambridge, Massachussets 1 Harvard University, 197©. "Pronominalization, Reflexivization, and Direct Discourse," Linguistic Inquiry. Vol. 3, No. 3, (1972), pp. 269-320. The Structure of the Japanese Language. Cambridge, Massachussets t The MIT Press, 1973. Kuroda, S.-Y. "On Kuno's Direct Discourse Analysis of the Japanese Reflexive zibun," Papers in Japanese  Linguistics. Vol. 2, No. 2 (1972) pp. 136-147. Langacker, Ronald W. "On Pronominalization and the Chain of Command," in David A. Reibel and Sanford A. Schane, eds., Modern Studies in  English. Prentice-Hall, Englewood Cliffs, New Jersey, 1969. 137 Muraki, Masatake. "Presupposition, Pseudo-clefting and Themat-ization," Unpublished Ph, D. dissertation. The University of Texas at Austin, 1971. Ogawa A. Toshimitsu. "A Study of Japanese Relativization," Unpublished Master's Thesis. University of British Columbia, 1974. Oyakawa, Takatsugu. "Japanese Reflexivization..I,." Papers in Japanese Linguistics. Vol. 2, No. 1 (1972), pp. 9^-135. Postal, Paul M. Cross-over Phenomena. New York t Holt, Rinehart and Winston, Inc., 1971. Shinoda, Aiko. "Classification of Japanese Speech Levels and styles," Papers in Japanese Linguistics. Vol. 2, No. 1 (1972), pp.~o*"6-81. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items