"Prosody and Syntax"

[1] Introduction: We would like to argue in this paper, first, that in order to make syntactic tests for scope interpretation of wh-questions in Japanese repeatable, we must take prosody into consideration. Second, we propose that focused wh-phrases come to be associated with some formal feature showing up in a higher functional projection at LF, which is distinct from any version of wh-movement proposed in the literature. The two claims are related but independent. First, we assume that the observations reported by Selkirk-Tateishi (1988) concerning the downstep of lexical accents in Japanese are basically correct even with the addition of an accented predicate as in (1). The pitch of each lexical accent therefore gradually declines toward the end of the utterance with the pitch level reset at the onset of each 'major phrase.' (Lexical accents are indicated by capitalization, and higher pitch levels by larger fonts.) (1) [ JOhn-wa [ MAry-ga oKAsi-o nuSUnda-to ] oMOtteiru ] -TOP -NOM sweets-ACC stole-that think 'John thinks that Mary stole the sweets.' In addition, we assume the existence of Emphatic Prosody (henceforth EPD) associated with each focused item in Japanese, which consists of: (i) a sharply raised pitch of the first high tone of a focused item, plus (ii) an abrupt drop to low pitch, which is prolonged until the next EPD or the end of the scopal domain of the focused item is encountered. cf. Bolinger (1965), Jackendoff (1972) and Ishihara (2000) Compare (1) above and (2) below. In (2), EPD is indicated by a bold-face large-font capital followed by underlined items enclosed by downward arrows '↓__↓'. (In this and other examples that follow, we will indicate the lexical accents of only relevant items, disregarding those that do not concern us.) (2) [ John-wa [ MA↓ry-ga okasi-o nusunda-to ] omotteiru↓ ] 'It is Mary that John thinks stole the sweets.' [2] Prosody and Wh-scope: Observe now the contrast between (3a) and (3b). They involve an identical Wh-question, but only (3a) is accompanied by EPD and interpretable, while (3b) sounds quite odd and prohibits us from obtaining any plausible interpretation. (On the examples, the rising question intonation is indicated by nO↑ at the end of an utterance. The readers are reminded here that elimination of the lexical accents of the two predicates nusunda and omotteiru in (3a) and their retention in (3b) are crucial to detect the contrast. It should be also kept in mind that no emphatic stress is intended in any of the lexical accents in (3b).)