Shana Poplack

Variationist tools for the study of language change

Abstract Despite important and continuing advances in historical linguistics, key questions about language change remain, including the graduality or abruptness of change, the nature of the transition period, universal vs. language-specific processes and objects of change, the possibility and extent of contact-induced change, and the role of social structure. Three major methodological, analytical, and theoretical issues may contribute to this state of affairs.

The first relates to what Labov has called the bad data of language history, many of which have survived by chance. The evidence they provide is necessarily limited compared with what can be obtained from synchronic materials: they may be rife with scribal error, unrepresentative of spoken language, or simply silent on the questions of interest. A second issue involves the analytical consequences of such data. Although isolated facts can readily be adduced to support a number of change scenarios, we cannot decide between conflicting analyses unless we situate them within the larger linguistic system. Third, most change originates in spontaneous speech, whose inherent variability poses a serious challenge for the uniformity assumption embodied in traditional methods of linguistic reconstruction.

In this paper I describe ongoing efforts to address each of these issues, making use of novel data sources and the variationist framework for linguistic analysis. We circumvent the bad data problem by exploiting a variety of synchronic speech sources (e.g. linguistic enclaves, bilingual communities in intense contact, minority language situations) as well as diachronic speech surrogates (semiliterate correspondence, popular theatre), which can be inferred to instantiate different stages in the evolution of change. We approach the variability problem from a variationist perspective, which rests on the observation that, in discourse, speakers select among discrete alternatives carrying the same referential value or grammatical function (broadly construed). These choices are not random, but constrained (usually quantitatively rather than qualitatively) by specific aspects of the linguistic (and/or extralinguistic) context. The key theoretical construct is the linguistic variable, made up of the class of variants among which speakers alternate in the expression of a given meaning or function. Each variable has an internal structure of its own, as emerges from the hierarchy of factors constraining the choices speakers make among competing forms; analysis of this structure offers a way to detect the mechanisms and direction of change. Finally, explicit comparisons with relevant benchmarks (representing e.g., earlier and/or pre-contact stages) enable us to identify stable variation and distinguish contact-induced from internally-motivated change.

Our studies of morphosyntactic variability tracked over long periods of time converge in showing that change tends to be surprisingly slow, infrequent, highly circumscribed and independent of language contact. I discuss the implications of these findings for widespread assumptions about the rate and extent of linguistic change, and the role of contact in triggering or accelerating it.