Introduction

The primary goal of Browman and Goldstein's study appears to be that of modeling speech at a level of detail far greater than that to which a phonologist ordinarily aspires. For this reason, phonologists might be tempted to consider their gestural framework not as an alternative to the standard autosegmental model (the model presented in Goldsmith 1976; Clements 1985; and others) but rather as a way of beefing up the autosegmental representations so that they can begin to deal with certain subphonemic aspects of articulatory timing.

In these comments, however, I will assume, that Browman and Goldstein are presenting a distinct theoretical alternative and that they advocate the adoption of gestural representations to the exclusion of autosegmental representations, not just as an interpretive appendix to them. I will begin by outlining the formal differences between gestures and autosegments. I will explain then why the distinct properties of the theory of autosegmental timing remain more useful in phonological analyses. In the last section of my comment, I will try to show that the points on which Browman and Goldstein's representations diverge from standard autosegmental theory have direct application to the description of certain phonological phenomena which have so far remained unexplained.

Gestures and autosegments

To understand what differentiates the gestural and autosegmental models we must look for answers to two questions: what types of units are the phonological representations made of and what types of relations obtain between these units?