Introduction

Phonetics and phonology

Much recent work in phonetics aims to provide rules, in the framework of generative phonology, that will characterize aspects of speech previously thought to be outside the province of grammatical theory. These phonetic rules operate on a symbolic representation from the phonology to derive a physical representation which, like speech, exists in continuous time and space. The precise nature of phonological representations depends on the theory of phonology, but certain general distinctions between phonological and phonetic representations can be expected. Only in the phonology are there discrete and timeless segments characterized by static binary features, though phonological representations are not limited to such segments. Even in the phonology, segments may become less discrete as features spread from one segment to another, and less categorical if features assume non-binary values. However, only in the phonetics are temporal structure made explicit and features interpreted along physical dimensions; the relations between phonological features and physical dimensions may be somewhat complex.

Thus phonological representations involve two idealizations. They idealize in time with segmentation, by positing individual segments which have no duration or internal temporal structure. Temporal information is limited to the linear order of segments and their component features. Phonological representations idealize in space with labeling, by categorizing segments according to the physically abstract features. These idealizations are motivated by the many phonological generalizations that make no reference to quantitative properties of segments, but do make reference to categorial properties. Such generalizations are best stated on representations without the quantitative information, from which more specific and detailed representations can then be derived.