Towards a Unified Analysis of Japanese Passives: Dealing with Individual Variability∗ T OMOKO ISHIZUKA UCLA
[email protected] July 7, 2008
1
Goal of this talk
(i) To provide syntactic analyses for systematically different judgment patterns regarding Japanese passives. (ii) To illustrate that there is individual variability with respect to certain types of passives: (1)
±óLʪßkp_
Ken-nom Naomi-dat run-rare-past ‘Ken had Naomi run on him.’ (Shibatani 1985)
• A group of speakers consistently rejects this type of passive (Grammar Q). • For some speakers, this type of passive is reported to be well-formed (Grammar L). (iii) To provide a preliminary account for how such variability arises in Japanese passives. ∗ Thanks
to Hilda Koopman, Anoop Mahajan, Dominique Sportiche, to Megha Sundara, Kuniko Nilsen, Reiko Okabe and to the audiences at the Annual meeting of Linguistic Society of America (LSA) and the 4th Formal Approaches to Japanese Linguistics (FAJL) conference, and at the Syntax-Semantics Seminar at UCLA. This work was supported in part by Lenart Travel Fellowship and UCLA, department of Linguistics.
1
T. I SHIZUKA
2
Japanese passives and Individual variability
July 7th
Background
2.1
Why we care about individual variability
Linguistics is a branch of cognitive science • The goal of ‘cognitive science’ is to describe and explain a human being’s ability to think. • The goal of ’linguistics’ is to figure out the mental state that an individual possesses when he or she speaks his or her native language(s). Syntax contributes to this goal by 1. describing the (subconscious) knowledge an individual has of his or her native language(s). • by identifying the syntactic atoms • by uncovering the laws (i.e., rules) that combine them 2. modeling this knowledge explicitly. We call this model Universal Grammar: (Chomsky 1986) Linguistic object to study is I-languages Person-specific I-languages vs. Socially-constituted E-languages, Chomsky 1981, 1986 ‘Japanese’, ‘English’ or ‘Tokyo dailect’ is an idealization. • I-language: Tomoko speaks Japanese • E-language: People in Japan speak Japanese • Presumably, no two individuals’ I-languages are exactly the same, although those of people living in the same community significantly overlap. Only the mental system of an individual’s I-languages can be studied.
2.2
In the past 50 years, syntactic theory has reached . . .
A basic theoretical understanding of the atoms and laws that give rise to language knowledge, based on introspective data. Through comparisons of typologically different languages, we now have some ideas about the options of UG, such as: • what the shared properties across languages are. • how languages can maximally or minimally differ. 2
T. I SHIZUKA
2.3
Japanese passives and Individual variability
July 7th
Japanese syntax: in the past 50 years
Japanese syntactitians have made huge progress in terms of • understanding basic and fine-grained properties of Japanese • identifying a number of central topics in Japanese e.g., scrambling, wh-questions, relative clauses, causatives, passives, multiple nominative constructions and nominative/ genitive conversions (Fukui 1986, Hasegawa 1993, Hoji 1985, Hoshi, 1994, Inoue 1976, Kageyama 1982, Kishimoto 1996, Kuno 1973, Kuroda 1992, Miyagawa 1989, Murasugi 1991, Saito 1985, Sakai 1994, Shibatani 1990, Watanabe 1996, among many many others)
2.4
For further theoretical progress: through I-languages
Syntactic theory is now • At the point where new types of data, such as the comparison of I-languages (i.e., individual variability) not only can, but in fact, must be brought into the field. • Individual grammars among the speakers of a particular language are optimal to study since they are naturally controlled in many respects (ex., lexical items, prosody patterns). ⇒ Help us understand finer-grained properties of UG. • Today’s talk addresses one such case: namely, an individual variability regarding ‘Japanese passives’.
3 3.1
Grammar Q: Grammar of the surveyed speakers Two types of passive and the standard analyses
a. Direct Passive (2)
±óLʪßkt_
Ken-nom Naomi-dat kick-pass-past ‘Ken was kicked by Naomi.’
b. Indirect (Gapless) Passive (3)
±óLèkM_
Ken-nom rain-dat descend-pass-past ‘Ken had rain descended upon him.’ 3
T. I SHIZUKA
Japanese passives and Individual variability
July 7th
Shared properties 1. Both passives contain -rare-. 2. The highest DP of the (closest) active is realized as -ni phrase.
Indirect (Gapless) passives are unique. . . 1. They appear to lack an active source (where does Ken come from?) cf.
èLMc_ ’It rained.’
2. Many of them carry the connotation that the subject is ’adversely affected’ by the event. • Unified movement analysis is not possible since indirect passives lack an active source. Standard analyses of Indirect passives • The indirect passive predicate -rare- selects for an affectee subject and a clausal complement (Kuroda 1976, Kitagawa and Kuroda 1992, Kubo 1992, Pylkk¨anen 2002 among others) vP v’
DP Ken
VP
v
rain descend
rare
[Ken-ga [rain-ni descend]-rare-past] Two types of analyses for the Direct Passive ‘Movement’ (e.g., Kuno 1973, Kubo 1992) Derived from their active counterparts by movement. vP DP Ken
v’ vP
v
Naomi Ken kick
rare
4
T. I SHIZUKA
July 7th
Japanese passives and Individual variability
‘Control (i.e., Affectee)’ (e.g., Kitagawa & Kuroda 1992) Same as indirect passives: PRO (i.e. silent pronoun) is contained in a complement clause, which is controlled by the affectee subject. vP
DP
v’
Keni
vP
v
Naomi PROi kick
rare
Pervious Analyses: how are the subjects licensed? The subject of the ‘gapless passives’ is always an affectee argument of rare.
Indirect Direct
Gapless
Affectee √ √
Movement √
⇒ Predictions made by the standard analysis: I. If the passive is well-formed, the subject of the indirect passive is affected by the event. II. As long as the subject is affected by the event, the passives should be wellformed.
3.2
Puzzles for the standard analysis
Predictions: I. If the passive is well-formed, the subject is affected. Puzzles: Indirect passives are well-formed even if the subject is not (adversely) affected. (4)
±óLʪßkJ}U_
Ken-nom Naomi-dat love-acc confession-do-pass-past. Lit. ‘Ken was confessed the love to by Naomi.
5
T. I SHIZUKA
Japanese passives and Individual variability
July 7th
±óoón3L^SHjD]åZʪßL |nó3kJ}W_
⇒ Suppose a situation:
Is the above sentence still felicitous to you?
⇒ In this case, Ken is not affected at all. Predictions: II. If the subject is affected, the sentence should be well-formed. Puzzles: Some sentences are unacceptable even though they are contextualized (introspective data). (5)
±óLʪßkp_
* Ken-nom Naomi-dat run-rare-past. Lit. ‘Ken had Naomi run on him.’ • However, this is given as well-formed in the literature (Shibatani 1985). ⇒ A sign of variability!?
Questions arise: • Do I have a different grammar from Shibatani? • Or do I just not come up with the ‘right affectee context’? Some passives require an adversely affected context An example of affectee context • Is the following well-formed? Kuroda 1979 notes that Hanako has to be closely related to Taro, (a daughter or girlfriend). (6)
*ÎLH k±Pñ_
Taro-nom teacher-dat Hanako-acc scold-pass-past. ‘Taro had a teacher scold Hanako on him.’
• But this produces the ’affected situation’ through establishing possessor relationship. • Adding a possessor relation forces a syntactic representation to change: namely, Taro is now licensed by the Possessed DP (i.e., [Taro-no Hanako]) • We can no longer tell whether -rare- licenses Taro or not ⇒ The two variables, the affectee and possessor contexts, need to be teased apart.
6
T. I SHIZUKA
3.3
Japanese passives and Individual variability
July 7th
Introducing variability and the proposed analysis
Proposal • The predictions made by the standard analysis seem not to hold for me. ⇒ How can we account for this? • My grammar seems to be different from the standard analysis in the following ways: Proposal for my grammar: I. -rare- is not a predicate that means ’be-affected’. II. -rare- never selects an argument, an affectee. (i.e. the subject of the indirect passive). This raises two questions: I. What licenses the subject of the indirect passive? II. How does the affectee connotation arise? Question 1: What licenses the subject of the passive? Proposed answers: I. either the main verb II. or the the possessed NP
I. The subject is an argument of the ‘main verb’ (i.e., Theme, Goal, Source). √ (7) a. Ken-nom Naomi-dat love-acc confession-do-pass-past. ‘Ken [Goal] √ was confessed the love to by Naomi.’ b. Naomi-nom Ken-dat love-acc confession-do-past. ‘Naomi confessed her love to Ken.’
±óLʪßkJ}U_
ʪßL±ókJ}W_
(8)
a.
b.
√
±óLʪßkR_
Ken-nom Naomi-dat escape-pass-past. ‘Ken √ was escaped from by Naomi.’
ʪßL±óKR_
Naomi-nom Ken-from escape-past. ‘Naomi escaped from Ken.’
7
[Source]
T. I SHIZUKA
(9)
a.
b.
√
Japanese passives and Individual variability
July 7th
q¬L'èkM_
Tokyo-nom heavy_rain-dat descend-pass-past. ‘Tokyo was descended upon by the rain.’ [Goal] √
'èLq¬kMc_
heavy_rain-nom Tokyo-dat descend-past. ‘Heavy rain descended upon Tokyo.’ II. The possessed NP licenses the subject (i.e., possessor). √ (10) John-nom son-dat die-pass-past. ‘John had his son die on him.’ √ (11) John-gen son-nom die-past. ‘John’s son died.’
¸çóLoPk{j_ ¸çónoPL{`
Note: Possessor here means ‘syntactic possession’, which is realized as a possessor small clause [DP-no NP]. (e.g. )
¸çónoP'qnØøH nf
⇒ All of the passives that I accept have active counterparts. Without an active source, the passive is ill-formed to me.
±óLʪßkp_
(12)
* Ken-nom Naomi-dat run-pass-past. ‘Ken had Naomi run on him.’
(13)
* { / / } Naomi-nom {Ken-acc/Ken-from/Ken-dat} run-past. Naomi ran {Ken/from Ken/to Ken }.
ʪßL ±ó ±óK ±ók pc_
Question 2: How does the affectee connotation arise? Proposed answers: I. variable due to the nature of the verb II. and the syntactic relation with the possessed NP I. Variable due to the nature of the main verb. (14)
±óLʪßkjJ}U_
Ken-nom Naomi-dat sin-acc confession-do-pass-past. Lit. ‘Ken was confessed her sin to by Naomi.’ (15)
±óLʪßkKºêbU_
Ken-nom Naomi-dat lover-acc bragging-do-pass-past. Lit. ‘Ken was bragged her lover to by Naomi.’ ⇒ Same structure but stronger affectee connotation for ‘to brag’. 8
T. I SHIZUKA
Japanese passives and Individual variability
July 7th
II. Due to the relation with the possessed NP. Ambiguity: Two readings with different syntactic structures (16)
¸çóLá¢êükjJ}U_
John-nom Mary-dat sin-acc confession-do-pass-past. (i) ‘John was confessed her sin to by Mary’ (ii) ‘John was confessed his sin to (somebody) by Mary.’
á¢êüL¸çókjJ}W_
(i) [Goal] John is a listener goal Active: Mary-nom John-dat sin-acc confession-do-past. ‘Mary confessed her sin to John
¸çóL á¢êük¸çó k jJ}U _
( ) ] Derivation: [ John-nom Mary-dat John-(dat) sin-acc confession-do-pass-past. John was confessed her sin to by Mary. (ii) [Possessor] Imagine a situation where Mary confessed John’s sin to someone else. Active: ( ) Mary-nom (somebody-dat) John-gen sin-acc confession-do-past. ‘Mary confessed John’s sin (to someone else).’
á¢êüL °Kk ¸çónjJ}W_
¸çóL á¢êük °Kk ¸çó n j J}U _
( ) ] ] Derivation: [ ( )[ John-nom Mary-dat (pro) John-gen sin-acc confession-do-pass-past. ‘John was confessed his sin to (somebody) by Mary.’ ⇒ Strong affectee connotation arises in (2) but not in (1).
3.4
Evaluating the arguments in favor of the standard analysis
Many arguments have been presented to distinguish between direct and indirect passives • Anaphor ‘zibun’-binding (Kuno 1973) • Numeral quantifier floating (Miyagawa 1989) • Ability to host numeral quantifier (Shibatani 1977, Kubo 1992) • Optionality of dative phrase (Shibatani 1985) and so on We will focus on the arguments in favor of rare being an affectee predicate
9
T. I SHIZUKA
Japanese passives and Individual variability
July 7th
Against the arguments for rare being an affectee predicate I. Compatibility with subject-oriented adverbials (Kuroda 1979) II. Modifying -rare- with an adverb (this is my argument)
I. Compatibility with subject-oriented adverbs Assumption: Adverbs, such as ’reluctantly’ or ’intentionally’, are assumed to require an agent or experiencer subject. (17)
åÒLVhfßkdK~HH_
thief-nom deliberately policeman-dat catch-pass-past. ‘The thief was caught deliberately by the policeman (the thief’s intention)’ However, it is also compatible with an unaccusative verb. (18)
åÒLVhfßkdK~c_
thief-nom deliberately policeman-dat be_caught-past. ‘The thief was_caught deliberately by the policeman.‘ (the thief’s intention)’ ⇒ The test is not sufficient to show that -rare- assigns an experiencerlike affectee theta-role.
II. -rare- cannot be modified by an adverb • -rare- is not a predicate meaning ‘be-affected’ • If -rare- is a predicate that is equivalent in meaning to ’be-affected’, it should be modifiable by a temporal adverb. • This is the case with causative -sase- (Shibatani 1990):
±óL ¦ÊªßkßßyU[_
(19)
[Causative] Ken-nom three_times Naomi-dat meal-acc eat-caus-past. ‘Ken made Naomi eat her meal three times’ (Ken did it three times, or Naomi ate three times)
(20)
[Passive] Ken-nom three_times Naomi-dat meal-acc eat-pass-past. ‘Ken had Naomi eat his meal on him three times’ *√Ken was affected three times because Naomi ate his meal. Ken was affected because Naomi ate his meal three times.
±óL ¦Êªßkßßy_
10
T. I SHIZUKA
3.5
July 7th
Japanese passives and Individual variability
Interim summary
At least two grammars exist for Japanese indirect passives: • Grammar Q: the grammar based on my judgment patterns • Grammar L: the standard analysis capturing the judgment patterns in the literature ⇒ They differ in how the nominative subject is licensed. How shall we proceed from here? Question: Am I alone? (i.e., Do other speakers have Grammar Q?)
3.6
Investigating the variability: Questionnaire survey
The purposes 1. To investigate the issue of individual variability. 2. To test the predictions. Note: Depending on the grammar each speaker has, the prediction will differ. Predictions made by standard (Grammar-L) vs. proposed (Grammar-Q) analyses (1) subject is always an affectee (2) affectee context improves (3) possessor context improves (4) well-formed only with active source
GMR-L √ √ √
GMR-Q √
-
√
Methodology • Participants: 74 native speakers of Japanese, residing in Tokyo at the time of the study (56: university students) • Tasks: participants were asked to judge the acceptability of Japanese stimuli on a 1 (impossible) to 5 (completely natural) scale. (multiple choice task regarding referential dependency on an anaphor zibun ‘self’ or silent possessors) • Stimuli: 120 sentences, (12 were presented w/context) • Conditions: 1. availability of the active source: argument of the main verb 2. availability of the active source: possessor argument 3. sensitivity to the syntactic structure of the active source 11
T. I SHIZUKA
Japanese passives and Individual variability
July 7th
Condition (1): availability of the active: argument of the verb
±óhʪßLlg
Context:
Ken-top Naomi-dat run-pass-past. ‘Ken had Naomi run on him.’ √ (mean w/o context:1.09) (mean w/context:2.01) (Literature: )
(22)
±óoʪßkR_
Ken-top Naomi-dat run-pass-past. √ ‘Ken had Naomi escaped from him.‘ (cf.mean:4.82) (Literature: ) • 73 out of 74 people gave ‘1 or 2’ to (1) without context. ⇒ 73% of participants had the same judgment pattern as Grammar-Q. ⇒ The speakers who gave ‘5’ or ‘4’ with the context did not behave consistently with respect to other items without an active source.
Condition (2): availability of the active: possessor argument
ʪßhêµoînëüàá¤ÈgBBåêµnͪL¡ Ojêµo i-ãDfD_]n_ʪßoÌå'jf`c _kKKZ agMZkfkn]Shkjc_
Context:
(23)
ʪßoêµnͪk{j_
Naomi-top Lisa-gen mother-dat die-pass-past. √ ‘Naomi had Lisa’s mother die on her.’ (mean:1.69) (mean:1.66) ( )
(24)
ʪßoͪk{j_
Naomi-top mother-dat die-pass-past. √ ‘Naomi had her mother die on her.’ (mean:4.68) ( ) ⇒ If the active possessor source is available, no need for contextualization at all. √ * vs.
ʪßnêµnJÍU
ʪßnJÍU
Condition (3): sensitivity to the syntactic position in the active Context:
*În n¬L{k L^8k²W*Îo ºkgMZ} WkWfD_ÎnfBMVjKc_ 2. * Î o n ¬ p å K F k < ~ f D _ L ¬ L K c f D k{gW~c_ 1.
(25)
*ÎL n¬k{j_
Taro-nom brother-gen dog-dat die-pass-past. Int. Taro had (his) brother’s dog die on him.
12
T. I SHIZUKA
July 7th
Japanese passives and Individual variability
(w/o context mean 1.66), Context 1:mean 2.73, Context 2: mean 3.11 Paired-T-test: t(73)=2.36, p=0.01: better in Context 2 ⇒ Question: What is the difference between the two contexts? Structural Ambiguity: [
*În n¬]
‘Context 1: No possession’ • No direct relation between Taro and the dog.
DP DP DP
NP NP
dog
Taro brother • Taro is embedded. ‘Context 2: Possession’ • Taro was a temporary owner of his brother’s dog.
DP DP
DP DP
Taro
NP
brother dog • Taro is the highest possessor. Syntactic theory predicts them to behave differently with respect to some syntactic operations, such as extractability or locality (Ross 1967). Summarizing the Data Pattern Sensitivity to the active (1) Active source:the verb licenses the subject (2) Active source: possession licenses the subject (3) structural position of the subject
Grammar-L -
Questionnaire √ √ √
• The results of the questionnaire show a quite uniform cluster of judgments: • The data pattern of 74 speakers was consistent with the predications made by the proposed analysis based on my judgments (Grammar-Q). • This is good: I am not alone . . . . but where do we go from here? 13
Grammar-Q √ √ √
T. I SHIZUKA
4
Japanese passives and Individual variability
July 7th
Grammar L: Grammar of the judgments in the literature
4.1
What could give rise to individual variabilities?
Bridging the gap between the two grammars How can we relate the two grammars and their distribution? • Regional differences? (Dutch case:SAND project) • Generational difference? (Icelandic Impersonal Passives: Maling and Sigurjonsdottir 2002) • Repeated exposure to a certain linguistic construction? (e.g., violation of CNPC) (i.e., syntactic satiation effects, Snyder 1994) ⇒ Certainly a difficult issue: but we encounter grammatical variability all the time, and it actually is an optimal object to study. ⇒ I speculate that the variability observed in Japanese passives is the result of syntactic satiation
4.2
Properties and the proposed analysis
Properties of Grammar L Question: What are the differences between the two grammars? ⇒ Grammar-L seems to generate all the cases that Grammar-Q does. • Direct passives (underlying objects of transitive and typical ditransitives) • Indirect passives (the ones which have an active source) This includes Inanimate subjects √ (26) That-island-top sea-dat surround-pass-asp-nonpast. ‘That island is surrounded by the sea’ (Kuroda 1979)
]nöowkò~fD
⇒ This is a statement expressing a state of being. ‘Island’ is not affected. This example is a puzzle if -rare- really means ‘be-affected’.
14
T. I SHIZUKA
Japanese passives and Individual variability
July 7th
Grammar-Q is a subset of Grammar-L (1) No overlap
They are independent. The overlapping judgments are accidents.
(2) Q is a subset of L
They share the same grammar: but GMR-L contains an extra mechanism to license those lacking an active source. ⇒ This is what I am pursuing . . .
4.3
Case study: a speaker of Grammar-L
Indirect passive without an active source The hashi-rare ‘run-passive’ example is acceptable with the following context:
± ó L 1 k O g D f k Ê ª ß L O g D D d Ê ª ß n ³óo^8kFUD]n[Dg±óoÎå jDÊå . . . (27)
√
±óLʪßkp_
Ken-nom Naomi-dat run-pass-past. ‘Ken had Naomi run on him.’
Background: • A Japanese linguist who majors in L1-acquisition • Very familiar with the literature of Japanese syntax, the standard analyses of Japanese passives. • Was exposed to the literature starting as an undergraduate.
15
T. I SHIZUKA
July 7th
Japanese passives and Individual variability
The availability of an affectee argument is very limited • She accepts an ‘unaffected subject’:
(28)
ʪßL±ókJ}W_L±óo]^DfDjKc _ √ ±óLʪßkJ}U_ Ken-nom Naomi-dat love-acc confession-do-pass-past. ‘Ken was confessed the love to by Naomi.
• ‘rain’ example is acceptable only if the subject is physically ‘wet’ due to the rain. • It is not felicitous under the following context:
±óo'Mj¤ÙóÈn¬ûgB ¤ÙóÈSåèLM¤ÙóÈL-bkjc_
(29)
±óoèkM_
Ken-top rain-dat descend-pass-past. *Intended: Ken was affected by the fact that it rained.
4.4
Analyzing the judgment patterns in the literature
I. Proposal for Grammar-L • Grammar-L seems to contain the core grammar of the surveyed speakers. • It can insert a null applicative head that introduces an affectee argument as a last resort. • The speakers uniformly claim that the subject has to be ’animate’: i.e., arguments that can be affected. II. Proposed structure of Grammar-L
ApplP
DP
Appl
Ken
vP
Appl
v’
16
vP
v
Naomi run
rare
T. I SHIZUKA
4.5
Japanese passives and Individual variability
July 7th
Discussion
Speculation: Syntactic Satiation • If this analysis is on the right track, this phenomenon might be a case of syntactic satiation . . . • Repeated exposure to a certain linguistic construction seems to serve as linguistic input, and the speaker develops a parsing strategy, which is then internalized, and becomes part of their syntax. • What is interesting is that this seems to be the grammar of last resort:: the optional applicative head is available only if a given passive cannot be derived by Grammar-Q. What does this tell us? • It seems not to be the case that there are two totally different grammars for passives. • There are languages that have a predicate ’be-affected’ (e.g., Chinese, Huang 1999) ⇒ This seems to be an option of UG. • There are variabilities: some speakers, like me, do not acquire the applicative head, but just reject the sentence as ill-formed. • Plasticity might have something to do with the age when the exposure takes place: still an acquisition process?
5
Conclusion
5.1
Grammar-Q: the surveyed speakers
• The well-formedness of indirect passives depends on the availability of an active source. • Not only accusative DPs, but also the dative goal and source arguments should be taken into account as the underlying subject of the passive. • Affectee connotation arises from the nature of the main verb and the subject being an external possessor.
17
T. I SHIZUKA
5.2
Japanese passives and Individual variability
July 7th
Grammar-L: Revising the standard analysis
• They have the same grammar as the surveyed speakers • In addition, they can introduce an affectee argument by having an applicative head as a last resort. • The affectee argument must be animate, and has to be directly and rather severely (physically or financially) affected by the event. Further investigation is necessary to identify the exact conditions and distribution.
6
Future direction • Experiments: complete a fuller version of the pilot questionnaire • Investigating Grammar-L: Is there an individual variability among the speakers of Grammar-L? – What are the properties of the silent applicative head? – Are there individual differences in terms of the availability of the applicative head? – What kind of context is required to have an applicative head? – Acquisition process? (how is it acquired? and when and how much exposure is necessary?)
Selected References Chomsky, N. (1981). Lectures on government and binding, Foris, Dordrecht. Chomsky, N. (1986). Knowledge of language: its nature, origin and use. Praeger, New York. Ishizuka, T. (2007). Towards a unified analysis of the passives in Japanese., Ms., University of California, Los Angeles. Available at: http://www.linguistics. ucla.edu/people/grads/ishizuka/index.htm Kuroda, S-Y. (1979). On Japanese passives, in Bedell, G., E. Kobayashi, and M. Muraki (eds.), Explorations in Linguistics: Papers in honor of Kazuko Inoue, 305-347, Kenkyusha. Kitagawa, Y. and Kuroda, S-Y. (1992). Passive in Japanese, Ms., Indiana Univ. and University of California, San Diego. Kubo, M. (1992). Japanese passives., Working papers of the department of language and cultures., Vol. 23. University of Hokkaido. Kuno (1973). Structure of the Japanese language. Cambridge, Mass.: MIT Press. Maling, J. and Sigurjonsdottir, S. (2002), The ‘new impersonal’ construction in Icelandic., Journal of comparative Germanic linguistics 5: 97-142. 18
T. I SHIZUKA
Japanese passives and Individual variability
July 7th
Miyagawa, S. (1989). Syntax and Semantics 22: Structure and Case Marking in Japanese, Academic press, New York. Pylkk¨anen, L. (2002). Introducing arguments, Doctoral dissertation, MIT, Cambridge, Mass. SAND (Syntactic Atlas of Dutch dialects): http://www.meertens.knaw.nl/projecten/sand/sandeng.html Shibatani, M. (1990). The languages of Japan. Cambridge: Cambridge University Press. Shibatani, M. (1985). Passives and related constructions: A prototype analysis, Language 61, 4. 821-848.
19
