Introduction

Recent studies have raised a number of questions about the nature and role of phonological representations in the analysis of intonation. One area of uncertainty involves the division of labor between the phonological component and phonetic implementation. Hausa shows that there is a phonology of intonation separate from the phonetics; attempts to frame the phonological generalizations in purely phonetic terms lead to loss of explanatory force. Expressing such intonational features as downdrift and ‘key raising’ (Newman and Newman 1981) as phonological not only explains their distribution more effectively but also reduces the size and perhaps the complexity of the phonetic implementation component. A second issue involves the nature of prosodic constituents, i.e. phonological domains, within which phonological rules apply. Hausa offers evidence of intonational phrases, prosodic constituents which, although they are constrained by syntactic constituent structure, do not mirror it exactly.

Third, our work addresses the representation of tone features. This is still a subject of controversy in the description of lexical tone. And when we look at the representation of intonational melodies which are superimposed on lexical tone, the problem is compounded even further. In Hausa all syllables are specified for either a High or a Low tone by the time rules of intonation apply. What sorts of features do phonological rules of intonation insert, and where in the representation are they inserted?

Prosodic constituency

We will address the question of constituency first. We have argued in the past (Leben et al. forthcoming; Inkelas et al. 1987) for the intonational phrase in Hausa on the basis of a number of phonological and phonetic effects that are bounded by phrases.