RUNNING HEAD: Socially Meaningful Variation Producing Socially Meaningful Linguistic Variation

It is axiomatic that spoken language is highly variable, and that variability can be observed at nearly every level of linguistic structure, from the acoustic instantiation of speech sounds to the the information structure of long stretches of spoken discourse. Many of the chapters in this volume discuss empirical and theoretical studies of the production of linguistic structures. A challenge to these has been to model the generation of variable linguistic forms. This chapter focuses on one type of variation, that which is related to social categories and social functions. We intentionally define 'social' broadly in this chapter. Consequently, this chapter reviews numerous bodies of literature, including studies of linguistic differences between groups that differ in macrosociological categories like gender, age, social class, and regional origin, linguistic differences that reflect stances, attitudes, and ideologies, and linguistic differences that are elicited in experimental tasks that manipulate different social variables. The purpose of this chapter is to consider empirical and theoretical studies of language variation, and to suggest how they might be linked to the psycholinguistic models of production that are the focus of this volume. The specific focus of this chapter is on variability in speech sounds. The reason for this focus is partly practical and partly theoretical. Practically, this is because of the expertise of the authors: both of us are laboratory phonologists who study social variation in speech-sound production, processing, representation, and acquisition. Another practical reason relates to the content of the existing literature on variation: there is simply a larger body of research on speech-sound variation than on syntactic variation. The other motivation is theoretical, and relates to the imbalance in existing literature. Early variationist studies of language were based on linguistic models that posited a strong disconnect between 'competence' and 'performance'. Sociolinguistic variation was seen as variation in performance in a group of speakers whose competence was Socially Meaningful Variation 3 identical. In this view, phonological variation was seen as cases of variation in the performance of phonological categories whose abstract competence-level representations were equivalent across dialects. In contrast, while syntactic variation was less amenable to this kind of modeling, as syntactic structures were seen as the product of a finite-state grammar that operated at the will end this chapter with a brief speculation on how our conclusions relate to the production of other types of linguistic variation, particularly syntactic variation. By way of introduction, imagine the …