Similarity in phonological theory

forthcoming

Although the sound patterns across the languages of the world vary widely, commonalities seem to recur in how human beings categorise sounds: we seem consistently more likely to ‘bundle’ /p t k/ than /v s g/, /f s h/ than /q x r/. A large fraction of ‘phonological theory’ as it now exists has resulted from interest in the properties such classes of sounds might share; in the distributional typology of sound classes, and the extent to which linguists' intuitions about the rarity or ubiquity of such classes match reality; and in the mechanism by which such classes arise over time. Cognitive representations of the SIMILARITY between the sounds involved in these patterns, although essential to the complete picture, have remained poorly-understood. This project seeks to remedy this gap in our understanding via the creation of a large cross-linguistic database of potentially similarity-driven phonological phenomena (assimilations and dissimilations). This will offer a good-quality empirical foundation on which to base theories of category formation and models of the similarity between individual segments. Analysis of this data, taking into account the interdependent nature of spatial variation and temporal variation, will link approaches to phonological theory and historical phonology with current state-of-the-art work in distributional typology; this connects sub-fields of linguistics whose intersection remains under-explored and overdue for further development.