Variable domains and variable relevance: interpreting phonetic exponents

Spoken language is a resource which is systematically deployed in the management of social interaction— its primary site of occurrence. The patterns and structures in language are emergent properties of and shaped by the exigencies and contingencies of social interaction. However, despite significant advances in modelling speech perception and understanding, and an increasing acknowledgment of the relevance of phonetic detail, there continues to be an overemphasis on issues of lexical distinctiveness and lexical access with the consequence that all sorts of systematically controlled fine phonetic detail do not find their way into contemporary models. Moreover, despite the explanatory potential of emergent, rich, category representations in self-organising episodic/exemplar models there is a continuing willingness to accept and model the kinds of category schemata to be found in rather traditional linguistic descriptions.