Modeling Phonology in Time

Abstract

UC Berkeley Phonology Lab Annual Report (2011) Modeling phonology in time Keith Johnson (University of California, Berkeley) keithjohnson@berkeley.edu 1. Introduction The aim of theoretical linguistics is to construct models that help us understand language in some meaningful sense. In this regard, linguistic theory is much like the other branches of social science in seeking to achieve explicit, falsifiable, predictive, and complete theories by implementing those theories in mathematical models of the systems we seek to understand. Examples of linguistic models include the formal descriptions of generative linguistics (e.g. Chomsky, 1965), psycholinguistic simulations of language processing (e.g. Frazier & Fodor, 1978), and sociolinguistic models of factors relating to linguistic variation (e.g. Labov, 1972). The key decisions in linguistic modeling (and models in the social sciences more generally) concern what to model and how to model it. Where generative linguistics focussed on modeling linguistic structure, psycholinguistics seeks to model the aspects of human cognition that are involved in using language, and sociolinguistic models seek to account for choices among linguistic variables in communities of speakers. One key point here is that different choices of what to model are interrelated. The speaker's selection of one linguistic variable versus another is influenced by cognitive processing, and the set of all sociolinguistic choices ultimately determines linguistic structure. The question of how to model language is also answered in different ways by different theoreticians. Where generative linguistics uses methods from formal logic and finite mathematics, psycholinguists model behavior with simulations of cognitive processes, and sociolinguists use regression models to identify causally-linked factors in language change. 2. Sound change and explanatory phonology This paper seeks to make a contribution toward an explanatory theory of phonological structure by modeling sound change. My approach assumes that phonological structure is shaped by the phonetic, cognitive, and social forces that impinge upon human speech communication. An explanatory theory of phonological structure is thus dependent upon a workable model of sound change (see Hume & Johnson, 2001). In passing, I would note that the structure of optimality theory (McCarthy & Prince, 1993) lends itself to an interpretation as a theory of sound change; with GEN as a source of phonetic variation and EVAL as a set of constraints on the inclusion of phonetic variants in a subsequent stage of the language. The key to success in modeling sound change, is to incorporate as many explanatory factors as needed. From before there was a difference between phoneticians and phonologists, scholars have espoused the idea that phonetic factors influence phonology (see e.g. Baudouin de Courtenay,