Word-specific phonetics

A long-standing forte of the Laboratory Phonology series has been work on phonetic implementation of phonological representations. Numerous studies in this series have elucidated the patterns of variation in the realization of phonological categories in different segmental and prosodic contexts, and such studies now provide one of the main lines of evidence about the cognitive representation of sound structure. In a consensus view of phonetic implementation, lexemes (the phonological representations of words) are abstract structures made up of categorical, contrastive elements. The phonetic implementation system relates this abstract, long-term, categorical knowledge to the time course of phonetic parameters in particular acts of speech. In fluent mature speakers, the phonetic implementation system is a modular, feed-forward system, reflecting its nature as an extremely practiced and automatic behavior. Lexemes are retrieved from the lexicon, and assembled in a phonological buffer in which phrasal prosody and intonation are also assigned. The fully formed hierarchical structures thus assembled provide the input to the phonetic implementation rules, which compute the degree and timing of articulatory gestures. The model is feedforward because no arrows go backwards, from articulatory plans to phonological encoding, or from the phonological encoding to the lexical level (apart from some post-hoc monitoring which permits people to notice and correct speech errors). It is modular because no lexeme information can influence the phonetic implementation directly, bypassing the level of phonological buffering. Though highly successful in explaining a wide range of data, such models are now challenged by a number of studies demonstrating the existence of word-specific phonetic detail. In modular feed-forward models, the (categorical) form of the lexeme wholly determines the phonetic outcome. If two words differ at all in their phonetics, then they differ categorically, and accordingly one job of the phonology is to identify a category set which captures all systematic differences amongst words. Another feature of these models is they do not take on the job of describing systematic phonetic variation related to sociostylistic register. Though the authors of such models would no doubt acknowledge the existence of such variation, they have not undertaken to provide a formal treatment of the cognitive capabilities which permit it. These limitations in formal models of