Raising to Object in Japanese: An HPSG Analysis

31st Pacific Asia Conference on Language, Information and Computation (PACLIC 31), pages 72–80 Cebu City, Philippines, November 16-18, 2017 Copyright c⃝2017 Akira Ohtani Raising to Object in Japanese: An HPSG Analysis Akira Ohtani Faculty of Informatics, Osaka Gakuin University 2-36-1 Kishibe-minami, Suita, Osaka 564-8511, Japan ohtani@ogu.ac.jp Abstract This paper discusses the so-called raising to object (RTO), which provides interesting problems with respect to the syntactic/seman- tic status of an accusative-marked NP. We ar- gue that two types of matrix verb, control and raising, must be recognized in the construc- tion. The linearization approach can capture the possibility of word order variation, espe- cially, the distribution of accusative-marked NP in the construction. Moreover, we suggest that RTO involves a non-thematic NP related to the embedded predicate via predication. 1 Introduction In some languages, an argument that belongs seman- tically to an embedded clause is realized syntacti- cally as an object of a matrix clause, this “raising to object” (RTO) is schematized as follows: (1) [matrix subject . .. objecti [embedded ∆i .. .] .. .] � The term “raising” has its origin in the transforma- tional analysis of such constructions in which the subject of the lower clause is “raised” to become the object of the matrix verb (Postal, 1974; Lasnik and Saito, 1991; among others). In Japanese, it has been noted in the literature on transformational syntax that examples such as (2) share syntactic properties with English counterparts: (2) a. Yamada-wa Yamada-TOP Tanaka-oi Tanaka-ACC [ ti baka fool da ] COP to COMP omotta. thought ‘Yamada thought Tanaka [to be a fool.]’ b. Yamada-wa Yamada-TOP [Tanaka-ga Tanaka-NOM baka fool da ] COP to COMP omotta. thought ‘Yamada thought [that Tanaka was a fool.]’ (Kuno (1976): pp. 23-24, Slightly altered.) As those glosses indicate, (2a) and (2b) show the same case alternation patterns that English exhibits. There are a number of conditions which must be satisfied in order to form a grammatical RTO con- struction, but in this paper, we focus on the predi- cational relation between the accusative-marked NP and the complement predicate. More specifically, we argue that RTO involves a non-thematic NP re- lated to the embedded predicate via predication. 2 Word Order and Embedded Predicate While there can be no doubt that Kuno’s (1976) RTO phenomenon exists in Japanese (Tanaka, 2002), there are at least two questions that cannot be ac- counted for by his analysis. One of the problems is concerned with the word order of an accusative-marked NP, which can be generally scrambled. Consider (3): (3) a. Yamada-wa Yamada-TOP Tanaka(-no Tanaka-GEN koto)-o matter-ACC baka fool isda to that omotta. thought ‘Yamada thought Tanaka to be a fool.’ (Kuno 1976: 24) b. Yamada-wa baka da to Tanaka*(-no koto)-o omotta. (Kuno 1976: 35) 72 Kuno’s observation indicates that Tanaka-o ‘Tanaka -ACC’ can not be located to the right of the comple- ment clause, while Tanaka-no koto-o ‘Tanaka-GEN matter-ACC’ can. The question arising from this contrast is: How can we derive the difference be- tween Tanaka-o and Tanaka-no koto-o …

Content truncated to 500 words