Derivation of Japanese clefts Masanori Nakamura / Senshu University 1. Outline There have been controversies over how exactly the cleft construction in Japanese, exemplified in (1b), is derived. Superficially, (1b) is similar to (1a), the relevant difference being that in (1b), the entire clause in (1a) has been nominalized by no and turned into the topic, with the focused material sandwiched between the wa-marked clausal topic and the copula da. This paper shows that previous analyses fail to provide a unified account of peculiar properties of Japanese clefts and puts forth an alternative proposal. 2. Previous Analyses One notable property of the Japanese cleft is that it exhibits island effects, as shown in (2). The question is: what kind of movement is involved? Hoji (1987) argues that it is null operator movement within the presuppositional clause, positing the derivation of (1b) in (3) where the null operator identifies the focused phrase as its antecedent. On the other hand, Hiraiwa and Ishihara (2002) claim that it is focus movement, maintaining that (1b) is derived as in (4): the focused material in the underlying no da construction undergoes syntactic movement, as in (4b), followed by the remnant topicalization, as in (4c). 3. Data and Proposal There is evidence that the cleft construction involves BOTH null operator movement AND focus movement. It has been pointed out that DP null operators can neither launch from subject positions ((5b, c)) nor move across finite T ((5d)) (Stowell 1986). If the Japanese cleft involves null operator movement, it is predicted that the counterparts of (1b) where either nominative subject or accusative object is clefted should be ill-formed. This prediction is borne out, as shown in (6a,b). Notice that the focus movement analysis fails to rule out (6a,b) in a principled way. One can show, however, that the postulation of null operator movement is not enough. Hoji and Ueyama (2003) note that Japanese clefts tolerate resumption, giving examples like (7). In (7) the resumptive pronoun so-re ‘it’ appears in the object position of the presuppositional clause, which indicates that no null operator movement has taken place in the clause. This is why (7) is legitimate in sharp contrast to (6b). Now, it is well known that island effects are nullified by the use of resumption, as shown in (8) (Erteschik-Shir 1992). The null operator analysis predicts then that the use of a resumptive pronoun in the presupposition should rescue otherwise ill-formed island-violating cleft structure. This prediction is false, as in (9), confirming the existence of another movement. Specifically, it must be the accusative object tanken-o ‘a dagger’ that has been illegally extracted out of the island. This prompts us to suggest that (1b) has the derivation in (10), where the presuppositional clause involves null operator movement and the no da clause involves focus movement of the dative PP, followed by clausal ellipsis (indicated by strikethrough). This hybrid analysis will be shown to retain all the virtues of the two previous analyses and, at the same time, account for new data that they cannot explain. 4. Further Evidence and Implications The proposed analysis differs crucially from the previous ones in that it posits ellipsis. We present novel evidence for ellipsis in Japanese clefts. When it comes to extraction of DP out of an island in Japanese, ellipsis of the island does not repair the violation (Takahashi 1994). This is why the ellipsis in (9) does not repair the island violation caused by the focus movement of the accusative DP. However, we show that argument PP is eligible for island repair under Japanese sluicing/stripping. Given this, our analysis predicts that the same contrast between DP and PP should extend to clefts, a prediction borne out by (9) and (11). In well-formed (11), what has been focus-moved is the argument PP Bill-no bokuzyoo-kara ‘from Bill’s ranch.’ Our analysis is superior not only empirically but also theoretically. It reduces all the relevant characteristics of Japanese clefts to the independently motivated features of ellipsis: the two instances of movement in (10) are driven by the need to meet the Parallelism requirement (Fox and Lasnik 2003) and the intricate interactions among null operator movement, resumption, and island (non-)repair yield the observed patterns of grammaticality.

Examples (1) a. Ken-ga Mari-ni atta. b. [Ken-ga e atta no]-wa Mari-ni da. Ken-NOM Mari-DAT met Ken-NOM met C-TOP Mari-DAT COP ‘Ken met Mari.’ ‘It was Mari that Ken met.’ (2) *[Ken-ga [[ e atta] hito]-o sagasiteiru no]-wa Mari-ni da. Ken-NOM met person-ACC looking.for C-TOP Mari-DAT COP Lit. ‘It was Mari that Ken is looking for the person who met.’ (3) [CP OPi [TP Ken-ga ei atta] no]-wa [VP Marii-ni da] (null operator mov’t + in-situ focus) (4) a. [TopP [FocP [CP [TP Ken-ga Mari-ni atta] no] da]] (underlying no da construction) b. [TopP [FocP Marii-ni [CP [TP Ken-ga ei atta] no] da]] (focus mov’t) c. [TopP [CP [TP Ken-ga ei atta] no]j [FocP Marii-ni ej da]] (remnant topicalization) (5) a. John was easy [OP to please e]. b. *John is easy [OP to believe (that) e knows the news]. c. *John is easy [OP to believe e to know the news]. d. *John is easy [OP to demonstrate [that Bill killed e]]. (6) a. *[e Kookyuusya-o katta no]-wa syachoo-ga da. (nominative clefting) luxury.car-ACC bought C-TOP president-NOM COP ‘It was the president that bought a luxury car.’ b. *[Syachoo-ga e katta no]-wa kookyuusya-o da. (accusative clefting) president-NOM bought C-TOP luxery.car-ACC COP ‘It was a luxury car that the president bought.’ (7) [Kokuren-ga kibisiku so-re-o hihansita no]-wa amerika-no seisaku-o da. the.U.N.-NOM harshly that-thing-ACC criticized C-TOP USA-GEN policy-ACC COP Lit. ‘It was the USA’s policy that the United Nations harshly criticized it.’ (8) a. *I’d like to meet the linguist that Peter knows a psychologist that works for e. b. I’d like to meet the linguist that Peter knows a psychologist that works for her. (9) *[Keizi-ga [[a-no ban so-re-o katta] otoko]-ni aitagatteiru no]-wa detective-NOM that-GEN night that-thing-ACC bought man-DAT want.to.meet C-TOP tanken-o da. dagger-ACC COP Lit. ‘It is a dagger that the detective wants to meet the person who bought it that night.’ (10) [CP OPi [TP Ken-ga ei atta]] no]-wa [FocP Marij-ni [CP [TP [vP Ken-ga ej atta]] no] da]. (11) [Ken-ga [[a-no ban so-ko-kara deta] uma]-o sagasiteiru no]-wa Ken-NOM that-GEN night that-place-from got.out horse-ACC looking.for C-TOP Bill-no bokuzyoo-kara da. Bill-GEN ranch-from COP Lit. ‘It is from Bill’s ranch that Ken is looking for the horse which escaped from it that night.’ References Erteschik-Shir, N. (1992) Resumptive pronouns in islands. In Island constraints: Theory, acquisition and processing, 89–109. Dordrecht: Kluwer. Fox, D. and H. Lasnik. (2003) Successive-cyclic movement and island repair: The difference between Sluicing and VP Ellipsis. Linguistic Inquiry 34:143-154. Hoji, H. (1987) Japanese clefts and chain binding/reconstruction effects. WCCFL VI. Hoji, H. and A. Ueyama. (2003) Resumption in Japanese. WECOL 2003. Hiraiwa, K. and S. Ishihara. (2002) Missing links: cleft, sluicing, and “no da” construction in Japanese. MIT Working Papers in Linguistics 43: 35-54. Stowell, T. (1986) Null antecedents and proper government. NELS 16: 476-493. Takahashi, D. (1994) Sluicing in Japanese. Journal of East Asian Linguistics 3: 265-300.