Introduction

Japanese is the classic example of what is often called a ‘pitch-accent language’, and regularly figures in discussions of the typology of tone and accent. Like Northern Bizkaian Basque, Tokyo Japanese has accented and unaccented words, and has an a that can contain at most one accent. Such as may well contain more than one accented morpheme, but only one of these survives on the surface. In terms of the discussion in chapter 3, Northern Bizkaian Basque and Tokyo Japanese are tone languages, with lexically distinctive accent.

The discussion of Japanese prosodic structure in this chapter follows a path from small to large. We begin in section 10.2 at the level of the word and move on to the a in section 10.3. Section 10.4 deals with the tonal structure of utterances with one α, and section 10.5 with the phonetic implementation of the tones, some of which may fall victim to truncation. The tonal structures in the one-α utterance are summarized in section 10.6 in an OT analysis of tonal associations in short words, when often not all tones can associate. Then, as we move on to utterances with two αs in section 10.7, I present the two classic arguments that Pierrehumbert and Beckman (1988) gave for the phonetic underspecification of tone: the argument in favour of interpolation over spreading and that for the moraic association of boundary tones, respectively. The Intermediate Phrase (ip) is discussed in section 10.8, where Japanese downstep is described and the data are presented that were used by Pierrehumbert and Beckman to show that Japanese downstep is cumulative.