Title
Ni Passive and Ni yotte Passive in Japanese
Author(s)
Nishi, Izumi
Citation
沖縄大学紀要 = OKINAWA DAIGAKU KIYO(10): 17-46
Issue Date
1993-03-15
URL
http://hdl.handle.net/20.500.12001/5773
Ni Passive and Ni yotte Passive in Japanese* Izumi Nishi
Research on the Japanese passive has a long history, including that done by Japanese philologists (Kokugogakusha). Yet some problems concerning the Japanese passive remain to be solved. One of those problems is concerned with a dichotomy between the ni passive and the ni yotte passive. The purpose of this paper is to reexamine the syntactic properties of these two types of Japanese passives and reconfirm that the aforementioned dichotomy is syntactically a real one.
The nature of the passive has become clearer through the perspective of Universal Grammar (UG), conceived within the 'principles-and-parameters1 approach, especially since the work by Jeaggli (1986). I will first focus on the problems to be treated in this paper and then review the analysis of passive construction within the framework of the 'principles-and-parameters' approach, mainly following Jeaggli (1986). By doing so, I will present a universal and cross-linguistic characterization of the passive. In order to apply Jeaggli's (1986) theory to the Japanese passive construction, I will refer to the work by Washio (1989-1990), in which the Japanese passive construction is analyzed in terms of principles and parameters and is related to the English get passive. Then, I will argue for Kuroda (1979), in which he first proposes that a major dichotomy in the Japanese passive exists between ni passive and ni yotte passive. Along the same line, Hoshi's (1991) work will be introduced to reconfirm that the dichotomy in question plays a relevant role in the syntax as well as the semantics of Japanese. An interesting
* I would like to express my deep gratitude to Roslyn Williams and Timothy Ahern for improving this paper and correcting stylistic errors. Needless to say, the remaining errors and inadequacies are strictly my own.
phenomena of the interaction between the two types of the Japanese passive and an aspect marker te iru will be observed, following Kuroda (1979), and its consequences will be considered. Finally I will present a rough scheme of mechanisms to derive each construction of the ni and the ni yotte passive and suggest that predication is cross-linguistically involved in expressing the adversity.
1. The Problems
The logical subjects in the passive constructions of Japanese are, if necessary, expressed by two markers:1 one with ni and the other with ni yotte.1 Following Kuroda (1979), I call the former type of passive construction the ni passive and the latter the ni yotte passive. A passive sentence as in (1) may take either ni or ni yotte to express its logical subject, of which the thematic role is agent in this particular example.
(1) John-ga Mary-ni/-ni yotte ker -are -ta
-Nom -BY kick-Pass-Pst
'John was kicked by Mary'
The logical meanings of the sentences in (1), whether with -ni or -ni yotte, do not differ. An example as in (2) seems to be more acceptable with ni than with ni yotte.
!As for the terminology of 'logical subject1,1 follow its use in Marantz (1981) among others. 2The grammatical status of ni is case particle, and a noun with ni attached to it is thus
considered to be Noun Phrase. That of ni yotte is rather unclear; however, as obviously seen from its morpheme boundary between ni and yotte, this consists of the two morphemes. The former is the same case particle as the above ni alone, and the latter is derived from the verb yor-u. The phrase of a noun with ni yotte attached to it will be regarded as Adjunct below.
(2) John-ga Mary-ni/?ni yotte mitume-rare-ta
-Norn -BY gaze at -Pass-Pst 'John was gazed at by Mary'
The passive sentences in which their passive subjects have a strong sense of being affected seem never to be allowed to take ni yotte.3
(3)
a. John-ga
ame -ni/ *-ni yotte hur-are -ta
-Norn rain -BY fall-Pass-Pst 'John got rained on'4
b. Mary-ga John -ni/ *-ni yotte sakini sono-uta -o
-Nom -BY before that-song-Acc
utaw-are -ta sing -Pass-Pst
'John sang the song before she did, which affected Mary1
Passives as in (3) are sometimes called 'indirect passive' (see among others
Howard and Howard (1976)), because passive subjects usually do not directly involve themselves in the events expressed there. Some further peculiar syntactic points in the indirect passive are that the active counterpart of (3a) 'It rained (on John)1 does not have any object to promote to the subject unless
3'Being affected' in such examples as (3) usually implies 'being adversely affected'; however, there exists a set of exceptions to this. One example of these exceptions is as follows:
(i) John-ga sensei-ni/*-ni yotte ronbun-o home-rare-ta
-Nom teacher -BY paper -Ace praise-Pass-Pst "John's paper was praised by his teacher'
4Some native speakers of English, especially American speakers, prefer (i) for the translation of the sentence (3a)
(i) John had it rain (on him)
Others never admit this as grammatical. The grammatical judgments on (i) differ to a large extent among native speakers of English. I refer the reader to Oehrle and Nishio (1981), Haegeman (1985), and Washio (1993) on this point and the references cited there.
'John' is treated as the object of "rain on1 in (3a) and that there remains an object marked by -0 in (3b).5
It is in general believed that the passive sentences objectively describing a fact prefer taking ni yotte to ni. Inoue (1976) furthermore claims that when the surface subject is [-animate] and indicating some collective activity, ni yotte is preferred over ni in expressing its logical subject, as in (4c) below.6
(4) a. Genzi-monogatari-ga Murasakisikibu *-ni/-ni yotte Genzi-tale -Nom Murasakisikibu -BY kak -are -ta
write-Pass-Pst
The Tale of Genzi was written by Murasakisikibu b. kono-hasi -ga watasi-no -yuuzin *-ni/-ni yotte
tuku-this-bridge-Nom I -Gen-friend -BY make-rare-ta (koto)
Pass-Pst (fact)
'This bridge was built by a friend of mine'
c. yuusi *-ni/-ni yotte kurusii
sokoku-volunteers-BY with-difficulty one's own country-hukki-undoo -ga tuzuke -rare-ta
return-movement-Nomcontinue-Pass-Pst
5Some native speakers of English, however, consider the sentence (ib) as marginally good,
not totally out. They seem to treat John as the object of rained on1 and promote it to [NP,S] position. On this issue, the reader is referred to Nishi (in preparation),
(i) a. It rained on John b. John was rained on
6Koto 'the fact that' is added to some of the Japanese examples so as to avoid the
unnaturalness caused by the lack of topic phrase in a matrix sentence. That is to say, in (4b) for instance, if a topic marker wa is attached to kono-hasi instead of the nominative marker ga, the sentence will become perfectly acceptable without koto.
'The movement for the reversion (of Okinawa to Japanese control) was carried on with much difficulty by volunteers1 (Inoue 1976: 83)
(5a) objectively describes a fact but shows the opposite pattern of grammaticality. Kuroda (1979) observes that ni and ni yotte are both equally accepted in (5b).
(5) a. tukue-ga kagu -ni/*-ni yotte hukum -are -ru (koto) desk-Nom furniture-BY contain-Pass-Prs (fact) ' 'Desk' is contained in the category of 'furniture'' b. ano -mati-ga nippon -gun -ni/-niyotte
that-town-Nom Japanese-army -BY hakai -s-are -ta (koto) destruction-do-Pass-Pst (fact)
'That town was destroyed by the Japanese army.1
((5a) is adapted from Kinsui (1991:11), and (5b) is adapted from Kuroda (1979: 327))
The patterns of appearance of ni and ni yotte thus remain mysterious. In order to clear up the situation, it is essential to see if there is any syntactic difference between the ni passive and ni yotte passive. And it is also important to view the Japanese passive from the perspective of Universal Grammar (UG).
2. Universal Characterization of the Passive
In this section, I will cross-linguistically characterize the passive within the 'principles-and-parameters' framework, following Chomsky (1981) and
-21-Jeaggli (1986). The passive constructions analyzed here are mainly from English, but some from other European languages.
Chomsky (1981: 124) mentions that passive in English such as (6) is decomposed into the two crucial properties stated in (7).
(6) a. Mary kicked John b. John was kicked by Mary (7) a. [NP,S] does not receive a 6-role.
b. [NP,VP] does not receive Case within VP.
The theory that can explain the properties of the passive in (7) will lead to a universal characterization of the passive. For the time being, I characterize the property (7a) as 6-role absorption and the property (7b) as Case absorption, following Jeaggli (1986).
2.1 6-role absorption
Jeaggli (1986) considers the passive morpheme en to absorb a 0-role of external argument. That is to say, the 0-role assigned by the predicate 'kicked John' in (6a) is absorbed in (6b).7 Some of evidence for the absorption of 0-role is shown in (8).
(8) It was estimated that they spent almost 6,000 million yen on the tree-planting ceremony in Okinawa.
(8) is an example of the English passive, and the [NP,S] position here is not assigned any 0-role; as a result, only the expletive 'it' is allowed to occupy that position.8 Remember that expletive elements are not referential and no
7The 0—role in question is 'agent'.
®There exist two members in the set of expletive elements in English. One is it, and the other
there.
9-roles need to be assigned to license them. That is why I conclude that the [NP,S] position in (8) has somehow become non-8-position, and that
according to Jaeggli (1986), its 6-role (i.e., that of external argument) is absorbed by the passive morpheme en. The term 'G-role absorption' is, however, misleading in that the passive morpheme en does not absorb the 0-role but the morpheme itself is in effect assigned the 0-0-role due to a subpart of 0-Criterion, a principle of Universal Grammar. Chomsky (1981: 36) characterizes the 9-Criterion as follows.
(9) Each argument bears one and only one 0-role, and each 0-role is
assigned to one and only one argument.
The latter part of (9), put differently, states that all 0-roles must be uniquely assigned. In (6), the 0-role of external argument of the predicate 'kick John' cannot simply be absorbed by a passive morpheme only to go up in smoke, but the principle of 0-Criterion forces the 0-role to be actually assigned to an
argument, i.e., a passive morpheme.9/10 The external 0-role in question in
effect will further be assigned to NP in fey-phrase, if any. For more details on
by-phrases, I refer the reader to Jeaggli (1986: 599-605) and Baker (1988: 334-339).
2.2 Case Absorption
In the English passive, Case absorption is obligatory for NP at [NP,VP]. In the D-structure representation (10), underlying the sentence John was kicked, the subject position is assigned no 0-role, as mentioned above, and the object
is assigned no Case.
9As for the status of passive morpheme, especially the point as to how we can consider it to be "an argument", see Jeaggli (1986: 590-591).
10The 0-role in question in effect will be further assigned to an NP in by -phrases in English. For the details concerning the mechanisms of this process, I refer the reader to Jeaggli (1986: 599-655) and Baker (1988: 334-339).
(10) [e] was kicked John
If John is not assigned any Case within the VP, the only way for it to receive Case is to move to the subject position and receive nominative Case there. Otherwise, (10) will be ruled out by Case filter. The movement of John to the [NP,S] position does not violate the 9-Criterion because the [NP,S] position here is assigned no 0-role. The NP-preposing is thus guaranteed and even forced by the subsystems of UG: 0-theory and Case theory. If Case is not needed for an element in the post-verbal position, movement is not forced. This prediction is borne out, as in (8) (repeated here as (11)), assuming that clausal complements do not require Case in English Qeaggli, 1986 :595).
(11) It was estimated that they spent almost 6,000 million yen on the tree-planting ceremony in Okinawa
2.3 Impersonal Passive
Case absorption can be interpreted as obligatory without any conditions in English. Intransitive verbs thus fail to be passivized because the passive morpheme en cannot find any NP from which Case is absorbed.
(12) *It was danced by many students
In some languages such as German and Dutch, it is possible to passivize certain intransitive verbs. The constructions are called impersonal passive; (13a) is an example of this type of construction in German and (13b) in Dutch.
(13) a. Es wurd von vielen Studenten getanzt it was by many students danced
'There was dancing by many students'
b. Er wordt hier door de jonge lui veel gedanst there is here by the young people a lot dance-Pass There was a lot of dancing here by the young people'
The accusative Case absorption by the passive morpheme is obligatory in German and Dutch whenever possible. Even if impossible, as happens, for example, with intransitive verbs, the passive morphemes can be attached to the verb and passivize the sentence without absorbing any Case. That results in impersonal passive.11 The syntactic difference between the languages not having impersonal passive (i.e., English type) and those having impersonal passive (i.e., German-Dutch type) is attributed to a parametric difference as follows.
(14) In languages such as English, passive morpheme must always absorb (accusative) Case; on the other hand, in languages such as German and Dutch, passive morpheme must absorb (accusative) Case whenever possible.
It follows that in English type languages, on the one hand, impersonal passive is impossible because the condition of passivization expressed in the first half part of (14) is not met; in German-Dutch type languages, on the other hand, impersonal passive is possible since the condition expressed in the second half of (14) is weak enough to accommodate the case of passivizing intransitive verbs.
nAlternative theory is proposed in Jeaggli (1986: 596-598). The above analysis in question, however, seems to me to be more fitting to the concepts of 'economy' in Chomsky (1991).
3. Two Dichotomies in The Japanese Passive 3.1. Direct Passive and Indirect Passive
The basic properties of the passive are characterized within the 'principles-and-parameters1 framework in the previous section. Each of Japanese sentences in (15) corresponds to the English ones in (6). And these cases such as (15) are analyzed by the same way as in English cases in (6); that is, the properties of the passive stated in (7) above, repeated here as (16), are applied to (15a) with other principles in UG, deriving (15b) as a result. Therefore, the analysis of the passive within the framework of the 'principles-and-parameters' presents no difficulties in analyzing the Japanese direct passive, which corresponds to the English passive construction.
(15) a. Mary-ga John-o ket -ta (koto)12 -Nom -Acckick-Pst (fact) 'Mary kicked John'
b. John-ga Mary-ni ker-are -ta (koto) -BY -Pass-Pst 'John was kicked by Mary1 (16) a. [NP,S] does not receive a G-role.
b. [NP,VP] does not receive Case within VP.
As a matter of fact, Japanese has another set of passives, called indirect passives, exemplified in (3), repeated here as (17) with some necessary adaptation.
(17) a. John-ga ame -ni hur-are -ta -Nom rain -BY fall-Pass-Pst 'John got rained on1
12The verb root ker is changed into ket, followed by the past suffix ta.
b. Mary-ga John -ni sakini sono-uta -o -Nom -BY before that-song-Acc utaw-are -ta
sing -Pass-Pst
'John sang the song before she did, which affected Mary1
The problems here, with respect to the analysis stated above within the 'principles-and-parameters' system, are as follows: in (17a) the verb hur (rain)
is an intransitive verb and has no [NP/VP] whose Case should be absorbed by (16b); in (17b) the object sono-uta (the song) still remains unchanged, which implies that the passive property (16b) fails to apply there. We therefore
conclude that the indirect passive construction lacks the property (16b) in Japanese; that is, the passive morpheme rare does not have to absorb Case in indirect passives. This conclusion is not strange at all, however, if one remembers the case of impersonal passives in German and Dutch, repeated here as (18).
(18) a. Es wurd von vielen Studenten getanzt it was by many students danced There was dancing by many students'
b. Er wordt hier door de jonge lui veel gedanst there is here by the young people a lot dance-Pass "There was a lot of dancing here by the young people1
Thus the problem does not lie in the fact that the passive morpheme does not absorb Case in the case of the indirect passive. The question to probe into is
whether or not the Japanese indirect passive holds the property (16a); in other
words, we are not sure yet whether the Japanese passive morpheme rare
absorbs a 0-role assigned to the [NP,S] position. In order to answer the
-27-question, we must see the characteristics of the other dichotomy in the Japanese passive.
3.2. Ni Passive and Ni yotte Passive
Kuroda (1979) and Hoshi (1991) among others contend that the major dichotomy in the Japanese passive exists between ni passives and ni yotte passives, not between direct passives (DP) and indirect passives (IDP). Kuroda (1979), for instance, points out some differences in the nature of the subject position in each type of passive construction. One difference is, according to him, that ni passives impose a selectional restriction on the subject NP but ni yotte passives do not. Compare the example of a ni passive sentence with that of a ni yotte passive one in (19), cited in Kuroda (1979: 330-331).
(19) a. *Fermat-no teiri -ga John-ni syoomeis-are -ta -Gen theorem-Nom -By prove -Pass-Pst Termat'si theorem was affected by Johns proving iti' b. Fermat-no teiri-ga John-ni yotte syoomeis-are-ta
Termat's theorem was proven by John'
Kuroda (1979) claims that the ni passive induces a changed state of its subject, that is, 'Fermat's theorem1 in (19a). But an abstract entity like Termat's theorem' is not affected at all by John's proving it. That is why (19a) is starred. On the other hand, in (19b) the ni yotte passive does not induce a change of this sort in its subject, and the sentence is thus grammatical. Hoshi (1991) attributes the property of this selectional restriction to the property of ni itself that assigns an affectee 0-role to the subject of its construction.
3.2.1. Hoshi's (1991) argument for Kuroda's dichotomy
Hoshi (1991: 65) restates the above-mentioned analysis by Kuroda (1979) in modern terms as follows:
...the subject position of ni passives is uniformly a 8-position while that of ni yotte passives is a non-0-position.
He also posits the Generalized Projection Principle (GPP), which purports to
express both the argument 0-relation and the adjunct G-relation.13 This
principle is needed on independent grounds, as shown in the footnote 13 below.
(20) Generalized Projection Principle (GPP) (=(15) in Hoshi 1991) All the 0-relations are represented at all levels in syntax.
Given the principle (20), in order to argue for Kuroda's dichotomy, it now suffices to show that the subject oriented adverbs can modify the subject of the ni passive but not the subject of the ni yotte passive. Following Hoshi (1991), compare the set of ni passives in (21) with that of ni yotte passives in (22).
13Lasnik & Fiengo (1974) suggests that some adjuncts like 'intentionally1 must have a 0-relation
with a subject as in (i).
(i) John is intentionally easy to please
That this 0-relation between the adjunct and the subject must hold at D-structure as well is one of his arguments for the rule of Complement Object Deletion for (i). Compare (i) with those non-sentences in (ii) (Lasnik p. 546).
(ii) a. *It is intentionally easy to please John b. To please John is intentionally easy
If 'John' does lie at its surface subject position even at D-structure, the GPP (20) can nicely explain the grammaticality in (i) and (ii) above.
(21) a. daitooryoo-ga orokanimo CIA-ni koros-are -ta (DP) president -Nom stupidly -BY kill -Pass-Pst 'Stupidly, the president! was affected by the CIA's killing
himi'
b. daitooryoo-ga orokanimo CIA-ni hisho-o koros-are-ta (IDP) secretary-Ace
'Stupidly, the president! was affected by the CIA's killing hist secretary1
c. daitooryoo-ga orokanimo hisyo-ni sin-are -ta (IDP) die-Pass-Pst 'Stupidly, the president! was affected by hisi secretary's
dying'
(22) a. ??daitooryoo-ga orokanimo CIA-ni yotte koros-are-ta (DP)
-BY
'Stupidly, the president was killed by the CIA'
b. ??daitooryoo-ga orokanimo CIA-ni yotte hisyo-o koros-are-ta (IDP)
'(lit.) Stupidly, the president was killed secretary by the CIA'
( = Stupidly, the president's secretary was killed by the CIA) c. *daitooryoo-ga orokanimo hisyo-ni yotte sinareta
'(lit.) The president was secretary died' (the intended meaning is as that of (21c))
The above sets of examples clearly show that a major dichotomy exists
between ni passives and ni yotte passives because the subject oriented adverb
orokanimo does not fail to be related to each subject in (21) when the agents of 'killing' and 'dying' are marked by ni, whether its construction is a direct
passive sentence as in (21a) or an indirect passive one as in (21b, c). In the — 30 —
case that those agents are followed by ni yotte whether it is a direct passive or an indirect passive, the sentences in (22) are marginal at best. The reason that the sentences in (22) are much worse than those in (21) is as follows. In the case of the ni yotte passive, as Kuroda (1979) assumes, the subject position is non-9-position. That amounts to saying that daitooryoo has moved there to obtain Case though its 0-role is already given at D-structure. The subject oriented adverb orokanimo must satisfy its 6-requirement with its subject. This cannot be done only by checking with its surface subject because of the Generalized Projection Principle (20), which requires that the 8-relation between the subject and orokanimo hold not only at S-structure but at D-structure. Those sentences in (22) are therefore ruled out by (20) (Hoshi, 1991: 67).
The contrast is also captured by a selectional restriction induced by secondary predicate. 'Nude' (-hadaka-de), for instance, always creates a relation of the secondary predication with a subject, and it observes the Generalized Projection Principle. That is why the sentence of a raising predicate (23) is not allowed to be with 'nude1.14
(23) *Nude, John is certain to have won
This explanation is also applicable to the paradigm of Japanese passives, which are adapted from Hoshi (1991: 68).
(24) a. John-ga hadaka-de keisatu-ni taihos-are -ta (DP) -Nom nudity -in police -BY arrest -Pass-Pst 'Johni, who was naked, was arrested by the police, which
affected himi1
14The sentence (23) is due to Lasnik (1974), cited in Hoshi (1991). — 31 —
b. John-ga hadaka-de keisatukan-ni asi-o ker-are-ta (IDP) policeman leg-Ace
'John, who was naked, hisi leg was kicked by a policeman, which affected himi'
c. John-ga hadaka-de ame-ni hur-are-ta15 (IDP) rain-BY fall
'John was rained on naked'
(25) a. ??John-ga hadakade keisatu-ni yotte taihos-are-ta -BY
John was arrested by the police naked'
b. ??John-ga hadakade keisatu-ni yotte asi-o ker-are-ta
'(lit.) John was kicked the leg by the police nude' ( = John, nude, was kicked on the leg by the police(man))
Assuming that the secondary predication between 'nude' and the subject is an
example that the Generalized Projection Principle applies to, the paradigm in
(24) and (25) is explained in the same way as the above case of
subject-oriented adverbs. That is, only ni passives do not involve the movement;
thus, the secondary predication is represented at all levels in syntax, thereby
satisfying the GPP. Ni yotte passives involve the movement to fill the surface
subject position, which is a non-6-position. The predicational relation between 'nude' and the subject is only represented at S-structure, failing to
15It is interesting to notice that the secondary predication holds between 'secretary' and
'naked', not between 'John' and 'naked' in (i) (i) John-ga hadaka-de hisho -ni sin-are-ta
secretary-BY die-Pass-Pst 'John'sj secretary died naked, which affected himj'
Hoshi's (1991) Generalized Projection Principle apparently fails to provide the reason why 'John's secretary' is naked, not 'John1 himself. It however does not contradict the interpretation of (i) only if hisho maintains some subject status in (i); what we really have to account for is why 'hadaka-de' only holds a predicational relation with 'John's secretary", never with 'John'.
-32-observe the GPP. The passive sentences in (24) are therefore labeled marginal at best.
We have seen in this section how Kuroda's (1979) dichotomy is substantiated by Hoshi's (1991) argument which introduces and is based on the Generalized Projection Principle (20) and that the dichotomy in question has a certain advantage over the more traditional dichotomy that is believed to exist between the direct passive and the indirect one.
4. Derivation of the Ni Passive
It is rather simple and straightforward to derive ni yotte passives because they are derivable in the same manner as English passives. That is to say, the universal characterization based on the 'principle-and-parameters' approach is applicable to the ni yotte passive construction, as seen in 2 and 3.1. In the case of ni passives, however, it is not that simple to derive its construction because we conclude in 3 above that the surface subject must be a 0-position in all levels in syntax; otherwise, it will violate the Generalized Projection Principle (20). This amounts to saying that to derive the passive construction in question we do not rely on (7a), one of the essential properties of passives.
(26) [NP,S] does not receive a 6-role. (=7a)
As elaborated in 2.1, (26) is equivalent to saying that the 0-role to the subject position is "absorbed". That is why the object can move to the passive subject position in English passives and Japanese ni yotte passives.
The question posed in 3.1 has so far remained unanswered and untouched; the question is whether the Japanese passive morpheme rare absorbs a 0-role assigned to the [NP,S] position in the case of indirect passives. Following the conclusion in the previous section, I have to change
-33-the question as follows: whether the Japanese passive morpheme rare absorbs a 8-role assigned to the [NP,S] position in the case of ni passives. The answer is positive, since the absorption of the 9-role is part of the universal properties of the passive. The 6-role to the [NP,S] position is indeed disposed of but it gains another 9-role, which we simply call an affectee role.
4.1. Semantic Property of the Ni Passive
We have examined the syntactic differences between the ni and the ni yotte passive in the previous section, but not yet the semantic differences between them.
(1), repeated here as (27a, b), takes both ni and ni yotte; however, there is a certain difference in meaning.
(27) a. John-ga Mary-ni yotte ker -are -ta -Nom -BY . kick-Pass-Pst 'John was kicked by Mary'
b. John-ga Mary-ni ker -are -ta -BY
'Johni was kicked by Mary, which affected
It is known that the passive with ni yotte states a fact objectively and that the passive with ni, rather than merely stating a fact, expresses that the passive subject is adversely affected. The two passives in (27) clearly show this semantic difference, indicated by the two different English translations. Reviewing all the Japanese examples in this paper, in effect, one can see this difference in meaning between the ni and the ni yotte passive. To sum up, if a
-34-passive only takes ni with it, it carries an connotation that the ni -34-passive puts some empathy on the passive subject. If a passive allows itself to take either ni or ni yotte with it as in (27), the reading of each case differs as just stated above. If a passive only takes ni yotte, never ni with it, the sentence tends to be an objective description.
There is a certain set of ni passives that seems to resist the above semantic generalization. That is, the set of those whose passive subjects are inanimate ones. The examples in (5) can take ni and their passive subjects are inanimate. Another example of this sort is added in (29), from Inoue (1976).
(28) a. tukue-ga kagu -ni/*ni yotte hukum -are -ru (koto) (=5a) desk -Nom furniture-BY contain-Pass-Prs (fact) 1 'Desk1 is contained in the category of 'furniture'' b. ano -mati-ga nippon -gun -ni/-ni yotte
that-town-Nom Japanese-army -BY hakai -s-are -ta (koto) (=5b) destruction-do-Pass-Pst (fact)
That town was destroyed by the Japanese army' (29) kono-ie -ga itabei -ni/-ni yotte kakom
-are-this -house -Nom wooden-fence -BY surround-Pass teiru (koto)
-Prs (fact)
'This house is surrounded by wooden fence'
Kuroda (1979), through closely examining the aspectual properties in such passive forms, maintains that the appearance of ni in (28b) and (29), in spite of its inanimate subjects, can be treated as part of the ni passive, the subject of
mo^ (199350
which is affected.16 The distribution of ni and ni yotte in (28a), however, still remains unaccounted for.17
4.2.2. Interpretation of Aspects in Passives
Kuroda (1979) substitutes an aspect marker ta in (5b) for te iru and discovers an interesting phenomenon of the interaction of the aspect marker te iru and the passive construction. The resulting sentences are in (31).
(31) a. ano-mati -wa nippon -gun -ni hakais -are-te iru that-town-Top Japanese-army-BY destroy-Pass
b. ano-mati -wa nippon -gun -ni yotte hakais -are-te iru that-town-Top Japanese-army -BY destroy-Pass
Te iru expresses either the progressive aspect or the perfective aspect, depending on the semantic properties of the verb it is attached to. If the aspect of the sentence is progressive, the English translation of (31) must be as in (32).
(32) That town is being destroyed by the Japanese Army
As Kuroda (1979) observes, however, one can find it easy to obtain the progressive reading only from (31b), not from (31a). For the perfective reading, one can obtain it from (31a) as well as from (31b).
Kuroda (1779: 328) generalizes the phenomena from the examples in (31) as follows.
16More details of Kuroda's analysis will be dealt with in 4.1.1 and 4.1.2 below, where the
property reading ni passive is treated.
17Kinsui (personal communication) suggests that Kageyama's (1991) format might help to
account for the suppression of the appearance of ni yotte in (28a). According to Kageyama, the verb fukum-u 'contain' does not imply the changed state of the logical object (passive subject); hence, ni yotte is not allowed to use in its passive construction.
(33) The ni passive may be used only if the aspect of the sentence is perfect.
(33) does not say anything about the case of the ni yotte passive; as a matter of fact, (33) logically implies that if the sentence is not perfect, the ni passive cannot be used. This prediction is borne out, as stated above; (31b) is interpreted as perfect and as progressive.
Though the judgment is delicate, the following pair of the passive construction from Kuroda (1976: 329) seems to support the generalization; that is, the aspect of (34b) is either perfect or progressive, but that of (34a) is only perfect.
(34) a. Mary-no -koto -ga FBI-ni tyoosas -are-teiru. -Gen-thing(s)-Nom -BY investigate-Pass "Things about Mary have once been investigated by FBI1 b. Mary-no-koto-ga FBI-ni yotte tyoosas-are-te iru
'Things about Mary are being investigated by FBI1 or 'Same as the above'
The generalization, Kuroda (1979) assumes, is also applicable to another aspect marker ta, which expresses either simple past or past perfective. In the case of the ni passive, the aspect of a sentence with ta is preferably interpreted as past perfective, but in the case of the ni yotte passive, it is interpreted as either simple past or past perfective.
Although Kuroda's (1979) generalization deals with the interesting phenomena of the interaction between Japanese passive sentences and aspect markers, it is not without problems. A simple ni passive sentence as in (35), whrere te iru is used, seems to naturally express the aspect of progressive.
(35) Mary-ga John-ni ais -are-te iru -Nom -BY love-Pass 'Mary is loved by John'
If the verb is involved in action, rather than state, a perfective reading is easier to obtain.
(36) Mary-ga John-ni ker -are-te iru -Nom -BY kick-Pass
'Mary has (once) been kicked by John1
Therefore, the result of the interaction of the passive and aspect markers is not straightforward, and the semantic features of root verbs must involve itself in expressing the aspect of the whole sentence. In order to deal with all these problems, to express my honest feeling, another full-fledged paper is needed.
As seen in 4.1.3. below, however, Kuroda's generalization can explain the difference in aspectual interpretation between the ni and the ni yotte passive within the indirect passive. I assume therefore that it is essentially a right generalization.
4.12. Property Reading Passive
There is another set of ni passives whose subjects are also inanimate. Those are called the 'property reading' passives in Kinsui (1991) since the predicates of these passives express a property of passive subjects. Kinsui (1993) considers this type of the passive to have the same construction as the ni passive and develop itself into another semantic class, though he has not given any arguments for it.
mo-5f (199330 (37) a. neko-wa ookuno-hito -ni kaw-are ais -are -te iru
cat -Top many -people-BY keep-Pass love-Pass 'Cats are kept and loved by many people1 b. kono-hon -wa ookuno-hito -ni yom-are -ta
this-book-Top many -people -BY read-Pass-Pst This book used to be read by many people1
The aspect of (37b) is perfect, not simple past; thus its translation into English is as This book used to be read by many1. This is what Kuroda's generalization predicts if it is the case that (35b) is part of the ni passive. (37a) however seems to me not to have only a perfective reading; it does not necessarily imply that cats have been kept and loved by many up to this point but also that they are still kept and loved by many and may be so in the future.
Here again further research is needed to carefully examine the nature of the 'property reading' passive and the validity of Kuroda's generalization on the interaction between passive constructions and aspects.
4.2.3. Indirect Passive
Before summing up this section, a case of the indirect passive is briefly referred to with regard to the dichotomy between the ni and ni yotte passive. It has long been believed that the indirect passive always has the adversity connotation. But this is not the case. Consider and compare each example in (38).
(38) a. Jane-wa kamera-o keisatu-ni bossyuus -are-ta -Top camera-Ace police -BY confiscate-Pass-Pst 'Jane's camera was confiscated by the police, which affected
Jane1
b. Jane-wa kamera-o keisatu-ni yotte bossyuus-are-ta 'Jane's camera was confiscated by the police'
The indirect passive usually takes ni; that is why it is considered to be always adversity passive, where its passive subject is adversely affected. But when the context allows it to be a ni yotte passive as in (38), its semantic contrast exactly corresponds to the cases of direct passives examined above. The degree of adversity on the passive subject in (38a) is definitely felt stronger than that in (38b). This is another piece of evidence showing that the dichotomy of the ni and the ni yotte passive is overriding that of the direct and indirect passive in the grammar of Japanese. Things are, however, not that simple; since if it is the fact that the appearance of ni yotte in the indirect passive is rare, I have to provide some account for it. I will return to the problem in 4.2.1 below.
It is also possible here to check Kuroda's generalization (33) by changing the aspect marker ta in (38) to te iru.
(39) a. Jane-wa kamera-o keisatu-ni bossyuus-are-te iru b. Jane-wa kamera-o keisatu-ni yotte bossyuus-are-te iru
There are two English translations possible for (39): A. Jane has an experience that her camera was confiscated by the police (perfective reading); B. As for Jane, her camera is being confiscated by the police (progressive reading). It seems to me that the aspect of (39a) tends to be interpreted as perfect; thus, the English translation A is appropriate. The aspect of (39b) is somewhat ambiguous, that is, either perfective or progressive. This is exactly what Kuroda's generalization implies, and it has thus provided a piece of evidence for the ni and ni yotte dichotomy in the Japanese passive.
4.2.4. Summary
Examining a semantic difference between the ni and ni yotte passive in this section, I have provided semantic grounds for an affectee 9-role to be assigned to the passive subject.
4.2. The Mechanism 4.2.1. The Ni yotte Passive
As to how to derive the ni yotte passive, as mentioned above, things are fairly straightforward; it is the result of the universal characterization of the passive. The passive morpheme rare absorbs the Case of the object, and the object thus must move to the subject position whose 9-role is already absorbed by rare in order to satisfy Case Filter.
If ni yotte passives are admitted within the indirect passive, as shown in the previous section, it is not yet clear how to derive them. Since such examples as in (38) are marginal, the construction of those sentences might be introduced and supplied by analogy.
4.2.2. The Ni Passive
The problems concerning the derivation of the ni passive is not that simple and is controversial among linguists. We have seen so far that in the ni passive, the passive subject is always affected. This affectee role is provided by the passive morpheme rare and licensed by the passive predicate, or the ni phrase itself.18'19 The passive morpheme rare, in the case of the ni passive, does two tasks: the absorption of the subject 0-role and the creation of a new affectee G-role. This apparently contradictory function of rare comes to look less strange if some example of the get-passive in English is considered. (40) is in fact the English translation of (38a), repeated here as (41).
18For more technical details on the derivation, the reader should refer to Washio (1989-1990). 19The possibility that the ni phrase itself is involved in the predication is suggested by Yoshida (personal communication). — 41
(40) Jane got [her camera confiscated by the police]
(The example was adapted from Washio (1989-1990: 257)) (41) Jane-wa kamera-o keisatu-ni bossyuus -are-ta
-Top camera-Ace police -BY confiscate -Pass-Pst
Washio (1989-1990) considers the bracketed part in (40) to be a passive construction in a small clause, wherein the subject 0-role is absorbed. The affectee role is created and assigned to 'Jane1 by 'get'. In English the two mechanisms of absorbing and creating a G-role are realized by the passive morpheme and 'get', respectively. In Japanese, both are realized by rare alone.
I should elaborate a little bit more the idea of licensing the affectee role by predication. It is my idea that one of the syntactic operations to express adversity in a sentence is predication. Some Japanese adversity passives are best translated into English by using the ethical dative 'on'. (42), a typical example of the Japanese ni passive, for instance, is translated into (43), where there is no passive construction involved.
(42) watashi-wa hisyo -ni yame-rare-ta I -Top secretary-BY quit -Pass-Pst (43) My secretary quit on me
Menn (1972) proposes such a structure as (44) for (43) and claims that certain selectional restrictions are observed between S" and Eth Dat. To put this differently, a predicational relation exists between them. I will not give further arguments for this analysis here but will deal with it in Nishi (in preparation).
(44)
My secretary quit on me
The last remaining problem in this section is concerned with the question whether the passive rare induces Case-absorption or not. Considering the case of indirect passives, as exemplified in (17) above, it
seems not obligatory or not necessary at all in the ni passive to absorb the
Case of the object in that those examples show that either the passive morpheme passivized the intransitive verb in (17a) or the object remains as an
object (17b). This effect in effect follows from Burzio's generalization.20
(45) a. John-ga ame -ni hur-are -ta (=17) -Norn rain -BY fall-Pass-Pst
'John got rained on1
b. Mary-ga John -ni sakini sono-uta -o -Nom -BY before that-song-Acc utaw-are -ta
sing -Pass-Pst
'John sang the song before she did, which affected Mary1
4.2.3 Summary
20Burzio's Generalization: If some NP governed by V is assigned no case, then the VP of which V is the head assigns no 6-role. .«
^lO-5f (199350
Detailed and technical arguments aside, I have presented each mechanism of deriving the ni passive and the ni yotte passive. Moreover, that some universal characterization on adversity is induced by predication in syntax is suggested.
5. Conclusion
In this paper, the relevance of the dichotomy of the ni and the ni yotte passive has been reconfirmed on syntactic and semantic grounds. Special attention has been paid to Kuroda's generalization on the interaction between the aspect markers and the passive constructions.
The passive subject in the ni passive is affected and thus given an affectee 0-role, and the passive subject position in the ni yotte passive is a non-0-position. The affectee 0-role is licensed by predication, some detailed characteristics and properties of which will be discussed in Nishi (in preparation).
References
Baker, M. C. (198S)Incorporation: A Theory of Grammatical Function Changing, University of Chicago Press, Chicago, Illinois.
Bedell, G., E. Kobayashi and M. Muraki (1979)Explorations in Linguistics: Papers in Honor ofKazuko Inoue, Kenkyusha, Tokyo.
Chomsky, N. (19Sl)Lectures on Government and Binding, Foris, Dordrecht. Chomsky, N. (1986a)Barriers, MIT Press, Cambridge.
Chomsky, N. (1986b)Knowledge of Language: Its Nature, Origins and Use, Preager, New York.
Chomsky, N. (1991) "Some Notes on Economy of Derivation and Representation", in R. Freidin ed., Principles and Parameters in Comparative Grammar, MIT Press, Cambridge, Massachussetts. Fukui, N. (1986) A Theory of Category Projection and Its Application, Doctoral
Dissertation, MIT.
Haegeman, L. (1985) "The Get Passive and Burzio's Generalization", 66, 53-77. Hoshi, H. (1991) "The Generalized Projection Principle and Its Implications
for Passive Constructions", Journal of Japanese Linguistics, 13,53-116. Howard, I. and A. M. Niyekawa-Howard (1976) "Passivization", in M.
Shibatani ed., Japanese Generative Grammar: Syntax and Semantics, 5, Academic Press, New York, 201-237.
Inoue, K. (1976)Henkeibunpoo to Nihongo, Taisyuukan, Tokyo. Inoue, K. (1989) Nihonbunpoo Shoojiten, Taishuukan, Tokyo, Jeaggli, O. A. (1986) "Passive", Linguistic Inquiry, 17,587-622.
Kageyama, T. (1991) "On the 'Passive Sense"', Gengorironto Nihongokyooikuno Soogokasseika, Tsuda Nihongo Kyooiku Center.
Kinsui, S. (1991) "Judoobun no Rekishi ni Tsuite no Ichikoosatsu", Kokugogaku, 164,1-14.
Kinsui, S. (1993) "Judoobun no Koyuu Hikoyuu-sei ni Tsuite", Kindaigo
Kenkyuu, 9,474-508.
Kuno, S. (1973)The Structure of the Japanese Language, MIT Press
Kuroda, S.-Y. (1979) "On Japanese Passives", in G. Beddel, E. Kobayashi and
M. Muraki eds., Exploration in Linguistics: Papers in Honor ofKazuko Inoue, Kenkyusha, Tokyo, 305-347.
Lasnik, H. and R. Fiengo (1974) "Complement Object Deletion", Linguistic Inquiry 5,535-572.
Marantz, A. (1981) "Grammatical Relations, Lexical Rules, and Japanese Syntax", in Coyote Paters: Working Papers in Linguisticsfrom A->Z, University of Arizona.
Menn, L. (1972) "On Me", Linguistic Inquiry, 3, 228-232.
Miyagawa, S. (1989)Structure and Case Marking in Japanese, Academic Press, New York.
Nishi, I. (in preparation) "The 'adversity' passive in English and Japanese", MS.
Oehrle, R. T. and H. Nishio (1981) "Adversity", in A. K. Farmer and C. Kitagawa eds., Coyote Papers: Working Papers in Linguisticsfrom A ->Z,
2, University of Arizona, Tucson, Arizona, 163-185.
Saito, M. (1982). "Case-marking in Japanese: A Preliminary Study", MS. MIT.
Teramura, H. (19S2)Nihongo no Syntax to Imi I, Kuroshio Shuppan, Tokyo. Washio, R. (1989-1990) "The Japanese Passive", The Linguistic Review, 6,
227-263.
Washio, R. (1993) "When Causative Means Passive: A Cross-Linguistic
Perspective", Journal of East Asian Linguistics, 2,45-90.
