7 - The segment: primitive or derived?

Summary

Introduction

The segmental or articulated character of speech has been one of the cornerstones of phonology since its beginnings some two-and-a-half millennia ago. Even though segments were broken down into component features, the temporal coordination of these features was still regarded as a given. Other common characteristics of the segment, not always made explicit, are that they have a roughly steady-state character (or that most of them do), and that they are created out of the same relatively small set of features used in various combinations.

Autosegmental phonology deviates somewhat from this by positing an underlying representation of speech which includes autonomous features (autosegments) uncoordinated with respect to each other or to a CV core or “skeleton” which is characterized as “timing units.” These autonomous features can undergo a variety of phonological processes on their own. Ultimately, of course, the various features become associated with given Cs or Vs in the CV skeleton. These associations or linkages are supposed to be governed by general principles, e.g. left-to-right mapping (Goldsmith 1976), the obligatory contour principle (Leben 1978), the shared feature convention (Steriade 1982). These principles of association are “general” in the sense that they do not take into account the “intrinsic content” of the features (Chomsky and Halle 1968: 400ff.); the linkage would be the same whether the autosegments were [± nasal] or [± strident].