Phonological Primes: Gestures or Features?

In their studies of the articulatory patterns that underlie speech, Browman and Goldstein (see e.g. Browman and Goldstein 1989, 1990, this issue) have brought to light some of the important respects in which the physical activity of s:peaking contributes to the way phonological systems are structured. From this point of view, their work has several interesting implications.

One is that much of what has usually been considered as lying in the domain of discrete phonological rules may be better understood from the point of view of nondiscrete or gradient properties of articulatory organization. Thus, in various papers Browman and Goldstein have produced evidence suggesting that many types of prosodically- conditioned reduction, contextual allophony and casual speech variation reflect dynamic properties of the activity of speaking and are better modeled at the level of physical speech production than at that of more abstract categorical representation.

Somewhat more ambitiously (and controversially), they propose that the dynamic properties of speech production, often viewed by phonologists as having little interest for the study of grammatical organization, playa large or even predominant role in shaping the structure of what is interpreted: the phonological system of rules and representations itself.

In this issue of Phonetica, Browman and Goldstein propose that a gesture-based model of phonology and phonetics can provide an alternative to models taking segments and features as their basic units. Their explicit intention is to show that "gestures are basic units of contrast among lexical items as well as units of articulatory action" and to "help clarify the differences among gestures, features, and segments."

This commentary will address both of these goals. It will examine several areas in which articulatory phonology as presently conceived by Browman and Goldstein appears insufficient to account for some of the generalizations. that are usually thought to lie in the domain of phonological (and phonetic) theory, and will suggest ways in which a theory of this son might be extended in order to accommodate these generalizations.

It will finally consider the status of articulation-based models of phonetic interpretation in phonological theory as a whole.