10/10/05
Yasuhiko Miura
Passives in Japanese: How are they to be classified?
1. Main claims
(1)Kuroda’s (1979) analysis of Japanese passives classifying them into ni and niyotte passives should be retained in terms of nipassives (contra Hoshi 1994), while it needs some articulation in terms of ni yottepassives (in line with Hoji to appear).
Specifically,
(2)nipassives, regardless of direct or indirect passives, have a structure in which the passive morpheme rare takes a subject NP and a complement clause as in (3).
(3)NP3-ga [VP NP1-ni (NP2-o) V]-rare
(4)ni yottepassives involve the argument-taking rare as in (3) as well as the argument-reducing rare as in (5).[1] The latter is further divided into the Surface OS type as in (5) and the Deep OS type as in (6) in terms of Ueyama (1998).
(5)a. NP1 NP2 V(D-S)
b.NP2 i-ga NP1-niyotte ti V-rare(S-S/PF)
(6)NP2i-ga [ proi [(NP1-niyotte) ti (NP1-niyotte) V-rare]]
base-generatedLF movement
2. Arguments for the claims
Argument for (2):
(7)ni direct passive sentencescan have resumption at the base-position of the object of the verb as in (8).
(8)[ NP2i- nom … NP2-ni … RESUMPi-acc V-(r)are]-ru/ta
Argument for (4):
(9)ni yotte passive sentencescan have resumption at the base-position of the object of the verb as in (10).
(10)[ NP2i- nom … NP2-niyotte … RESUMPi-acc V-(r)are]-ru/ta
(11)Reconstruction effects are observed in the ni yotte passive constructions. With resumption, however, the effects cannot be obtained. (Hoji to appear: 6-7)
3. Classification of Japanese Passives
3.1 Direct passives vs. Indirect passives[2]
(12)Direct passives:
Taroo-ga sensei-ni/niyotte sika-rare-ta
-NOM -by scold-PASS-PAST
‘Taro was scolded by the teacher.’
cf.Sensei-ga Taroo-o sikat-ta
-NOM -ACC scold-PAST
‘The teacher scolded Taro.’
(13)Indirect passives:
a.Taroo-ga sensei-ni Hanako-o sika-rare-ta
-NOM -by -ACC scold-PASS-PAST
‘Taro had Hanako scolded by the teacher.’
b. Taroo-wa ame-ni hu-rare-ta
-TOP rain-by fall-PASS-PAST
‘Taro was rained on.’
(14) English passives:
a. John broke the vase.
b. The vase was broken (by John).
(15) Characteristics of -en in the generative tradition (Hoji to appear:(3))
a. The V of V-en loses its external argument.
b. The internal argument of the V of V-en can no longer be ‘licensed’ as an object; i.e., the Case cannot be assigned/checked in the object position.
3.2 -ni vs. -niyotte in the direct passives
(i) If the passive subject is an abstract noun, ni yotte is obligatory.
(16) (Kuroda 1979’s (17) and (18))
a.Kaikai-ga gityoo-niyotte sengens-are-ta
opening-NOM chairman-by announce-PASS-PAST
‘The opening of the meeting was announced by the chairman.’
b. *Kaikai-ga gityoo-ni sengens-are-ta
(ii) If the passive subject is concrete but inanimate, ni yotte is obligatory.
(17)(Kuroda 1979’s (19) and (20))
a.Siroi booru-ga Oo-niyotte takadakato utiage-rare-ta
white ball-NOM -by high hit:up-PASS-PAST
‘A white ball was hit high in the air by Oo.’
b. *Siroi booru-ga Oo-ni takadakato utiage-rare-ta
(iii) Under the interpretation in which the adverb such as orokanimo ‘stupidly’ is associated with the surface passive subject, the sentence is unacceptable with ni yotte.
(18)(Kuroda 1979’s (91) and (93))[3],[4]
a.Daitooryoo-ga orokanimo CIA-ni koros-are-te simat-ta
president-NOM stupidly CIA-by kill-PASS should:not:have-PAST
‘The president stupidly let the CIA kill him, which he should not have let happen.’
b. ??Daitooryoo-ga orokanimo CIA-ni yotte koros-are-te simat-ta
3.3 Three types of passives in the literature:[5]
(19)a.NP2-ga NP1-ni V-(r)are-Tns ni direct passives
b.NP2-ga NP1-niyotte V-(r)are-Tns ni yotte passives
c.NP2-ga NP1-ni NP3-o V-(r)are-Tns ni indirect passives
4. Analyses of Japanese Passives
4.1 Controversy b/w non-uniform and uniform theories of direct/indirect passives
Non-uniform theory: N. McCawley 1972, Kuno 1973, Inoue 1976
It assigns direct and indirect passives different structures.
(20)a.Direct passives:
[S NPi-ga [VP (NP-ni/niyotte) ti V-rare-ta]]
b.Indirect Passives:
[S NP1-ga [VP NP2-ni NP3-o V]-rare-ta]
Uniform theory: Hasegawa 1964, Kuroda 1965, Makino 1972,1973, Howard and N-Howard 1976
It assigns direct and indirect passivesthe same structure.
(21)a.Direct passives:
[S NP1-ga [VP NP2-ni NP1-o V]-rare-ta]
b.Indirect Passives:
[S NP1-ga [VP NP2-ni NP3-o V]-rare-ta]
4.2 Kuroda 1979
- Kuroda (1979) provides arguments against the non-uniform theory. [6]
- A distinction within the direct passive between ni and ni yotte passives is introduced.
- Two-way distinction between ni and ni yotte passives
A. ni direct passives/ ni indirect passives (= argument-taking rare)
(22) (based on Kuroda’s 1979 (134))
a.NP3 [ NP1 NP2 V]-rare(D-S)
b.NP3-ga [ NP1-ni NP2-o t ] V-rare(S-S)
The surface difference between the ni direct passives and the ni indirect passivesis ascribed to the application of a process that deletes the object of the embedded sentence of the ni direct passives under identity with the matrix subject coreferential with it. (Kuroda 1979:183-4)
B. niyotte passives (= argument-reducing rare)
(23) (based on Kuroda’s 1979 (135))
a. NP1 NP2 V(D-S)
b.NP2 i-ga NP1-niyotte ti V-rare(S-S)
4.3 Two recent proposals
4.3.1 Hoshi 1994[7]
- Distinction between ni direct passives and indirect passives revival of the non-uniform theory
-NP-movement is involved in the ni direct passives.
- Three-way distinction among ni-direct, indirect and ni yotte passives.
A. ni direct passives (= argument-takingrare)
(24) (based on Hoshi’s 1994 (20))
a.[VP1 NP2i [V1' [VP2 e [V2' (NP1) [V2' PROi V-rareta ]]] e(V1)]](D-S)
adjunct
θ-role assignment
b.[VP1 NP2i [V1' [VP2 PRO [V2' (NP1) [V2' ti V-tj ]]] raretaj]](LF)
NP movement excorporation
c.[IP NP2i-ga [VP1 ti[V1' [VP2 PROi [V2' (NP1-ni) [V2' ti V-tj ]]] tj]] raretaj]
Case tense feature checking
B. ni yotte passives (= argument reducing rare)
(25) (based on Hoshi’s 1994 (23))
a.[IP e [VP e [V' (NP1-ni yotte) [V' NP2 V-rareta ]]]](D-S)
adjunct
b.[IP NP2i-ga [VP (ti) [V' (NP1-ni yotte) [V' ti V-tj]]] aretaj](LF)
( )
Case tense feature checking
C. ni indirect passives (= argument taking rare)
(26) (based on Hoshi’s 1994 (25))
a.[VP1 NP3 [V1' [VP2 NP1 (NP2) V-rareta ] e ]](D-S)
base-generated
θ-role assignment
b.[VP1 NP3-ga [V1' [VP2 NP1-ni (NP2-o) V-tj ] raretaj]]
excorporation
c.[IPNP3-ga [VP1 ti [V1' [VP2 NP1-ni (NP2-o) V-tj ] tj ]] raretaj](LF)
Case tense feature checking
4.3.2 Hoji to appear
- Two-waydistinction between ni and niyotte passives as in Kuroda (1979), but the structures of niyotte passives need to get articulated.
- Surface/Deep OS types distinction in terms of Ueyama (1998) is introduced.
A. ni direct passives/ ni indirect passives (= argument-taking rare)
(27) θ-role assignment
NP3-ga [VP NP1-ni (NP2-o) V]-rare
B. ni yotte passives
(28) (=argument-taking rare)
θ-role assignment
NP3-ga [VP NP4-niyotte (NP1-ni) (NP2-o) V]-rare
adjunct
(29) (=argument-reducing rare): Surface OS type
a.NP1-niyotte NP2-ga V-rare(unmarked order)
adjunct
b.NP2i-ga [NP1-niyotte ti V-rare] (marked order)
PF movement
(30) (= argument-reducing rare): Deep OS type[8],[9]
NP2i-ga [ proi [(NP1-niyotte) ti (NP1-niyotte) V-rare]]
base-generatedLF movement
4.4. Summary
NP-ni / NP-niyottedirect / Indirect
Kuroda (1979) / arg.-takingrare / arg.-reducing rare
Hoshi (1994) / arg.-reducing/-takingrare / arg.-takingrare / arg.-reducing rare
Hoji (to appear) / arg.t-takingrare / arg.-reducingrare arg.-takingrare
5. Assessment of Hoshi 1994
5.1 Hoshi’s(1994)/Saito’s(1982) argument for the non-uniform treatment of ni direct/indirect passives
Hoshi’s (1994) argument for the NP-movement analysis of the ni direct passives is based crucially on the acceptability of the example in (31) presented by Saito (1982) and his account. (31) would have the structure as in (32) under Kuroda’s (1979) analysis of the passives.
(31)Mary-ga John-o Tom-ni e sikar-are-sase-ta
-NOM -ACC -by scold-PASS-CAUSE-PAST
'Mary made/let John be scolded by Tom.' (Hoshi 1994: (12) cited from Saito 1982)
(32)[Mary-ga John-o [John-ga [Tom-ni John-o sika]-rare]-sase-ta]
Saito’s (1982) assumptions:[10]
(33)The complement NP of the embedded V in the ni-direct passives is marked with 'o'.
(34)The morphological form "o" signals the objective Case.
Saito’s (1985) hypothesis in regard to the abstract Case assignment: [11]
(35)A verb can assign objective Case to at most one NP in Japanese
Given the structure (32)of the example (31)and the hypothesis in (35), it is predicted under Kuroda’s non-NP movement analysis of the ni direct passives that the sentence (31) is unacceptable since it has two NPs which are assigned objective Case.
However, the fact is that(31) is acceptable. Saito (1982) thus concludes that Kuroda’s non-NP movement analysis is wrong given the validity of the hypothesis (35).
Given (35), the complement position of the V in (32)cannot be an objective Case position. Saito (1982) thus concludes that this position is an NP-trace position.
5.2 Falsification of Saito 1982
Summary of Saito’s (1982) proposals:
(36)(i)Complement position of the verbin the nidirect passives is a non-Case-marked position.
(ii)Thus, the surface subject NP, which is base-generated in this position, cannot be Case-marked at this position.
(iii)Therefore, the surface subject NP is moved from that position leaving NP-trace in order to get Case-marked.
Assumptions:
(37)The complement position of (transitive) verbs is either a Case-marked position (to be occupied by an overt NP) or a non-Case-marked position (to be occupied by an NP-trace).
(38)Resumption cannot surface in an NP-trace position, i.e., a non-Case-marked position.
Given Saito’s (1982) proposal in (36-(i)) and (38), the following negative prediction can be extracted with regard to the example (31).
(39)Under the configuration where a ni direct passive sentence with the structure as in (40) is embedded in a causative sentence as in (41), the sentence is unacceptable if the sentencecontains resumption at the base-position of the object of the verb.
(40)[NPi-nom… NP-ni … ti V-(r)are]
(41) *[ NP-nom … NPi-acc[NPi-nom… [NP-ni … RESUMPi-acc V]-(r)are]-sase]-ru/ta
If sentences with the structure of (41) turn out to be acceptable for any Japanese speakers, either of the following is falsified, given the validity of (38).
(42)(i) Saito’s (1982) proposal that the complement position of verbs in the nidirect passive is a non-Case-marked position. (cf. (36-(i))).
(ii) Saito’s (1985) hypothesis that a verb can assign objective Case to at most one NP in Japanese, on which (36-(i)) is based. (cf. (35))
5.3 Falsification of Hoshi (1994)
Given the structure of the ni direct passives proposed by Hoshi 1994 (see (24)), he also makes the following hypothesis.
(43)Complement position of the verb in the nidirect passive is non-Case-marked position.
(= (36-(i)))
Given (43)and (38), the negative prediction as in (44) can be extracted.
(44)If a ni direct passive sentence contains resumption at the base-position of the object of the verb as in
(45)), the sentence is unacceptable.
(45)*[ NPi- nom … NP-ni … RESUMPi-acc V-(r)are]-ru/ta
If sentences with the structure of (45)turn out to be acceptable for any Japanese speakers, Hoshi’s (1994) hypothesis in (43)is falsified, given the validity of (38).
5.4 CFJ
The paradigm to be tested:
(46)a.[ NP2- nom … NP1-ni … V-(r)are]-ru/ta
b.(=
(45)))[ NP2i- nom … NP1-ni … RESUMPi-acc V-(r)are]-ru/ta
c.[ NP3-nom … [ NP2-acc … NP1-ni … V-(r)are]-sase]-ru/ta
d.(=(41)) [ NP3-nom … [ NP2 i-acc … NP1-ni … RESUMPi-acc V-(r)are]-sase]-ru/ta
The result of the CFJ:12 informants (one is excluded due to the unacceptability of resumption.[12],[13]
(47)
Crucially, the examples which are predicted to be unacceptable by Saito’s (1982) and Hoshi’s (1994) analyses, i.e. (d)-examples and (b)-examples in (47), respectively, are not unacceptable.
(b)-examples are all in the plus range and (d)-examples are close to the plus range. Thus, both of the negative predictions in (39) and (44) are disconfirmed.[14]
6.Assessment of Hoji to appear (to be added)
6.1 non-uniform treatment of ni yotte passives
6.2 Ni yotte phrases in the indirect passives
6.3 CFJ
7. Remaining issues
(to be added)
Referrences
Hasegawa, Kinsuke, 1964. "Nihongo Bunpoo Siron [An Essay on Japanese Grammar]," Gengobunka 1, 3-46
Hoji, Hajime., to appear. “Reconstruction Effects in Passive and Scrambling in Japanese,” JK 13.
Hoji, Hajime and Ayumi Ueyama., 2003. “Resumption in Japanese: A preliminary Study,” WECOL 2003, University of Arizona.
Hoshi, Hiroto., 1994. “Theta-role Assignment, Passivization, and Excorporation,” Journal of East Asian Linguistics 3, pp.147-178.
Hoshi, Hiroto., 1999. “Passives,” in Natsuko Tsujimura, ed., The Handbook of Japanese Linguistics, 191-235,Blackwell Publishers.
Howard, Irwin and Agnes M. Niyekawa-Howard., 1976 “Passivization,” in M. Shibatani, ed., Japanese Generative Grammar: Syntax and Semantics 5, 201-237, Academic Press.
Inoue, Kazuko, 1976. Henkeibunpoo to nihongo: Joo, toogoron o tyuusin-ni (Transformational Grammar and Japanese: (I) On Syntactic Structures), Taishuukan, Tokyo.
Kuno, Susumu, 1973. The Structure of the Japanese Language, MIT Press, Cambridge.
Kuroda, S.-Y., 1965. Generative Grammatical Studies in the Japanese Language, Doctoral dissertation, Massachusetts Institute of Technology, Cambridge.
Kuroda, S.-Y., 1978. “Case Marking, Canonical Sentence Patterns, and Counter Equi in Japanese,” in J. Hinds and I. Howards (Eds.), Problems in Japanese Syntax and Semantics, Kaitakusha, Tokyo.
Kuroda, S.-Y., 1979. “On Japanese Passives,” in Bedell et al., eds., Explorations in Linguistics: Paper in Honor of Kazuko Inoue, 305-347, Kenkyuusya, Tokyo
Makino, Seiichi., 1972. "Adverbial Scope and the Passive Construction in Japanese," Papers in Linguistics 5, 73-98.
Makino, Seiichi., 1973. "The Passive Construction in Japanese," in B. Kachru et al., eds., Issues in Linguistics: Papers in Honor of Henry and Renee Kahane, 588-605, University of Illinois Press.
McCawley, Noriko Akatsuka, 1972 "On the Treatment of Japanese Passives," Papers from the Eighth Regional meeting of the Chicago Linguistics Society 8, 256-270.
Saito, Mamoru., 1982. “Case Marking in Japanese: A Preliminary Study,” ms., MIT
Saito, Mamoru, 1985. Some Asymmetries in Japanese and their Theoretical Implications. Doctoral dissertation. MIT.
Ueyama, Ayumi, 1998. Two Types of Dependency. Doctoral dissertation. Los Angeles: University of Southern California.
1
[1] The terms ‘argument-taking/argument-reducing -rare’ are from Hoji to appear. With regard to the properties of these two types of -rare, Hoji to appear:8 mentions that “(the argument-taking -rare takes an NP (as its external argument) and a VP as its complement, assigning the NP the experiencer theta-role. NP-ni within the VP that express the agent remains to be an argument of the V. The argument-reducing -rare eliminates an argument of the V to which it is attached.”
[2] Kuroda 1979 refers to Howard and Niyekawa-Howard 1976 on the terminology of ‘direct/indirect passive’.
Howard and Niyekawa-Howard (1976:202) mention that “(t)he most striking property of the Japanese passive is the fact that it often contains one more noun phrase than appears in its closest active counterpart. This noun phrase represents the individual indirectly affected by the action or event expressed in the remainder of the sentence,” and refer such passive sentences as ‘indirect’ passives for the purposes of exposition. They thus point out that “(t)his terminological distinction is based on a single syntactic criterion and is not intended to make any claim about other syntactic or semantic properties of these constructions.” (Howard and Niyekawa-Howard 1976:203)
[3] The gloss and the English translation are from Hoshi 1994:151,(8).
[4] The acceptability ?? on (18b) is by Hoshi (1994). Kuroda (1979:202) mentions that “[(18b)] is unacceptable as a surface form, without a permissible reading.”
[5] I employ the terminology in Hoshi 1994 simply for the purpose of exposition.
[6] Kuroda (1979:184) mentions that "Howard and Niyekawa-Howard [(1976)] convincingly refute those arguments [for the non-uniform theory]. Furthermore, they give an account in the uniform theory of the ambiguity phenomenon, the crucial fact for the promotion of the nonuniform theory…"
[7] Hoshi (1994) assumes the following lexical specification for rare.
(i)(Hoshi’s 1994 (21))
1)+/ Experiencer
2)+/ Passivisation: External θ-role suppression & Case absorption
Given (i), there are four types of rare with the following features in Japanese.
(ii)1)[+Experiencer/ +Passivisation]: ni direct passives
2)[Experiencer/ +Passivisation]: ni yotte passives
3)[+Experiencer/ Passivisation]: ni indirect passives
4)[Experiencer/ Passivisation]
Regarding [Experiencer/ Passivisation], Hoshi (1994:161) suggests that “this types of passive verb exists but cannot surface because of the Principle of Morphological Nonredundancy (iii), which prohibits a passive verb with no positive feature from surfacing.
(iii)Principle of Morphological Nonredundancy
Attachment of redundant morphology is prohibited. (Zubizarreta 1985)
[8]pro acts as an empty operator, which is adjoined to the IP at LF and forms the predication relation between the base-generated subject (NP2 in (30)) and the IP (or the lambda predicate that the IP gets mapped to). (Hoji to appear: 4-5)
[9] Resumption forces a given OS to be an instance of Deep OS.
[10] These are not clearly expressed in Saito 1982.
[11] As mentions by Hoshi (1994:152), Saito’s (1982) original argument for the NP-movement analysis of the ni direct passives relies on Harada’s (1973) double-o constraint.
[12] For the full result of the survey, see Appendix A.
[13] The examples in (A5) are not included in this chart. Hoji (2003: fn.61) points out that "… so-itu 'that guy' … has a clearly derogatory connotation. The fact that many speakers find the pure BVA use of so -itu less acceptable (and often highly marginal) than that of so-ko 'that place', as pointed out by Hoji (1995: footnote 21), is therefore not unexpected either…" Thus, it might be possible that some irrelevant things to the current discussion affect the judgment of the sentences.
[14]Recall that the negative prediction in (39) has been extracted from the following two hypotheses made by Saito (1982).
(36)(i)Complement position of the verbin the nidirect passives is a non-Case-marked position.
(38)Resumption cannot surface in an NP-trace position, i.e., a non-Case-marked position.
The hypothesis (36)-(i) is based crucially on Saito’s (1985) hypothesis in (35).
(35)A verb can assign objective Case to at most one NP in Japanese.
Thus, it would be possible that the hypothesis (35) is wrong, managing to save (36)-(i). Recall, however, that the negative prediction in (44)has also been extracted from (36)-(i) (=(43)) and (38). Here (36)-(i) is the direct consequence of Hoshi’s (1994) proposed structure of the ni direct passives. Therefore, the falsification of (36)-(i) directly falsifies Hoshi’s (1994) NP-movement analysis of the nidirect passives.