Diachronic Explanations of Sound Patterns

Phonological systems show clear signs of being shaped by phonetics. Sound patterns are overwhelmingly phonetically 'natural', in that they reflect the influence of physical constraints on speech production and perception, and categorical phonological processes often mirror low-level gradient phonetic effects. The question of how best to explain and model the influence of phonetics on phonology has been approached in different ways, one of which situates the locus of explanation in the diachronic domain of language change, in particular sound change. On this view, recurrent sound patterns merely reflect recurrent sound changes with phonetic origins, typically in speech perception. Explicit models of sound change are reviewed and illustrated, in particular Ohala's listener-based model and Blevins' Evolutionary Phonology framework, and the relevance of exemplar-based models of speech production and perception is also noted. Current issues of controversy regarding the adequacy of diachronic vs. synchronic explanations for the typology of sound patterns are surveyed. 1. Introduction: the phonetic bases of phonology As even a cursory acquaintance with phonology will reveal, the vast majority of sound patterns and phonological processes in the world's languages show clear indications of being shaped by the physical constraints of speech production and perception. Much of phonology is thus arguably 'natural' from a phonetic perspective. Distributional patterns and asymmetries in the ability of different positions to support phonemic contrasts (syllable onset vs. coda, stressed vs. unstressed syllable) are strongly correlated with the relative salience of acoustic- perceptual cues and/or the relative magnitude of articulatory gestures in the positions in question. Common phonological processes usually have clear parallels in low-level patterns of phonetic variation, both in the articulatory domain (coarticulation, gestural undershoot, inter-gestural timing and overlap, boundary strengthening) and in the acoustic-perceptual domain (confusion, misperception; aerodynamic effects on the acoustic signal). Such obvious and pervasive correlations raise a number of questions which cut to the conceptual core of phonology as a discipline. What is the nature of the connection between phonetic 'substance' and the higher-order patterning that phonology deals with? How close is this connection, and how—if at all—should we allow it to inform our theoretical models of the implicit knowledge speakers have of the sound patterns of their native language? To what extent, and in what manner, should phonetics be called upon to explain empirical generalizations about the typology of sound patterns and phonological systems? What bearing does all of this have on the question of