The present work has partly characterized the phonetic interpretation system of language through an investigation of the surface distributional patterns of acoustic phonetic measurements, correlating those patterns with features of the phonological structure: vowel classes, stress, and following consonants. It was found at all three levels that important phonetic differences exist between dialects, in addition to and separate from the phonological differences that distinguish them. At two levels, in the patterns of distribution of the mean nuclei of each vowel class, as well as in the patterns of vowel reduction, rule-systems which generate the observed distributions of acoustical measurements were characterized. Because languages differ in these phonetic patterns, and because they can be described by precise and language-specific rule systems, I believe it is appropriate to call this level of linguistic description, namely, the system of phonetic implementation of surface phonological structure as sound, ``phonetic grammar''.
In this chapter, the existence of dialect-specific phonetic interpretation processes of coarticulation was demonstrated. Linguistic description and theory can now begin to explore the internal structure and functioning of this intricate system of interacting phonetic processes.
An important conclusion of this work is that historical phonetic rules, which restrict the range of possible chain shifts, appear to play a role in the phonetic interpretation system. Universal principles of phonetic implementation are supplemented by dialect-particular applications of these general phonetic rules, in order to describe characterize the system of average (one might say, ``target'') vowel nuclei. This general idea is quite similar to the basic theoretical approach taken in Liberman & Pierrehumbert (1984), an approach that was proposed in order to account for observed patterns in pitch contours. The idea of a phonetic grammar consisting of interacting universal and dialect-specific factors was taken to a logical extreme in the phonetic grammar of mean vowel nuclei was presented for Jamaican Creole nuclei in Chapter 6. Partial characterizations of conceptually similar phonetic grammars are also contained in the descriptions of the other dialects.
In the studies of vowel reduction presented in the discussions of the four dialects, there appeared to be a phonetic ``reduction target'', towards which vowels shift when phrasal stress is reduced. This reduction target seems to be somewhat different in different dialects. In Jamaican, the short vowels reduce in the direction of a mid-central position in vowel space. In Chicago, a high-central position is the reduction target. In Alabama (in the speaker studied), a high and relatively front position appears to represent the reduction target. The L.A. Chicano pattern seems to have a target that lies in a intermediate between the targets in Chicago White English and Alabama English: it is somewhat to the front of a central position, but not as far to the front as is the reduction target in the Alabama speech studied. It was found that the pattern of vowel reduction was fairly consistent across speakers within a dialect, so that Judy and Jim from Chicago, for example, have very similar reduction patterns, and Rita is not very different. However the differences across dialects are quite striking. Also, the application of the process seems to be conditioned by phonological vowel length in Jamaican.
We may conclude that vowel reduction as a function of phrasal stress is a widespread phenomenon in English, if not also throughout the world's languages. However, the process is not identical in different dialects. The main difference between the dialects appears to be in the location of the reduction target, which appears to range from mid-central to high-central to high-front. The reduction target is never peripheral in the front-back dimension of vowel space, though it may be quite high. The patterns observed are consistent with the hypothesis that this reduction target is identical with the phonetic realization of //, which may additionally represent the ``basis of articulation'' of Sievers (1901), which was taken up as the ``neutral position'' in SPE (Chomsky and Halle 1968). In this work, the phonological status of this position is not understood, as it was in SPE, as the basis for deriving the marked and unmarked values of the phonological features, though proposals might be made in that direction in the future. However, the location of this position is a crucial element in the explanation of the surface patterns of vowel reduction. Phonetic grammar must specify this parameter in some way.
The system by which surface phonological forms are implemented in phonetic production has been partly characterized for a number of dialects in several chapters. Both endpoints of the mapping performed by this system were explored in detail: in Chapter 2, the acoustic dimensions were related understandably to the articulatory configurations that give rise to them, while in Chapter 3, the surface phonological structure of the vowel system of a useful fictional dialect, ``Reference American'' was characterized using autosegmental structure, underspecification theory, and privative features. In the large-scale acoustical studies of surface patterns of phonetic performance conducted in each of the dialect chapters, patterns in the system of phonetic interpretation were explored and characterized, leading to considerable progress in some aspects. In addition to the characterization of abstract phonological structure and the characterization of the physical aspects of the mapping from surface phonological structure to formant patterns, various parts of the intermediate system between physical and phonological structure were explored. The existence of differences across dialects in this system was demonstrated. The examination of surface distributional patterns of acoustic phonetic measurements and the correlation of those patterns with features of the phonological structure, including the underlying vowel classes, phrasal stress, and in this chapter, certain following consonants, has led in the end to insightful, interesting, and surface-true analyses of important aspects of the system of phonetic interpretation, or phonetic grammar.