The Island Effect in the Postverbal Construction in Japanese

The Island Effect in Postverbal Constructions in Japanese Kohji Kamada Chiba University 1-33, Yayoicho, Inage Ward, Chiba-shi, Chiba, 263-8522 Japan k-kamada@L.chiba-u.ac.jp Abstract It has been generally assumed that a violation of island constraints indicates that the relevant syntactic phenomena involves movement. That is, if what look like displacements violate island constraints but remain acceptable, this means that they should not be derived by movement. A careful examination of postverbal constructions in Japanese reveals that no movement is involved in the derivation of the construction despite the fact that in some cases island effects are observed. The effects, which have up to now been dealt with purely in syntax, can receive a better account in terms of language processing. This suggests that the human parser should undertake explanations of part of the output of the competence system. 1 Introduction Japanese is descriptively a verb-final language. In some cases, however, non-verbal elements come at the end of sentences, as shown in (1).1, 2 (1) a. Taro-ga ano mise de tabe-ta yo, Taro-NOM that shop at eat-PAST FP susi-o. sushi-ACC ‘Taro ate sushi at that shop.’ b. Taro-ga susi-o tabe-ta yo, Taro-NOM sushi-ACC eat-PAST FP, ano mise de. that shop at In (1a), the object susi-o ‘sushi-ACC’ appears in postverbal position, and in (1b), the adverbial phrase ano mise de ‘at that shop’ does so. I refer to these phenomena as the postverbal construction in Japanese (JPVC), and refer to 1 The relevant elements are in boldface. 2 The abbreviations used in glossing the data are as follows: ACC = accusative, DAT =dative, FP = sentence-final particle, NEG = negative, NOM = nominative, TOP = topic. elements in sentence-final position as postverbal elements (PVE).3 Some researchers (e.g., Endo, 1989; Kaiser, 1999; Whitman, 2000; Tanaka, 2001; and Abe, 2004) claim that the PVE is derived by movement because of the obedience of the PVE to island constraints such as the so-called Complex NP Constraint (CNPC), as shown in (2). In (2), e is used to mark the position associated with the moved element, namely the PVE, and the identical subscript indicates that the PVE corresponds to e. (2) *?[NP [CP [ei Sonkeisiteiru] sensei]-ga respect teacher-NOM fueteimasu yo, gakuseitati-gai. increase FP students-NOM ‘The number of the teachers who theyi respect is increasing, studentsi.’ In (2), the PVE is extracted out of the NP that contains the relative clause, thereby violating the CNPC. The example in (3), however, is acceptable although it violates the CNPC. (3) [NP [CP ei Sonkeisiteiru] gakuseitati]-ga respect students -NOM fueteimasu yo, Tanaka sensei-oi. increase FP Tanaka teacher-ACC ‘The number of the students who respect himi is increasing, Mr. Tanakai.’ It has been generally assumed that a violation of island constraints indicates that the relevant syntactic phenomena involves movement. That is, if what look like displacements violate island constraints but are still acceptable, this means that they should not be derived by movement. 3 I do not deal with the case in which clauses appear in postverbal position, as shown below. (i) …

Content truncated to 500 words