Introduction

Few problems have engaged the creativity of language variationists to the extent that the collection of phonological data has. In studying phonology, researchers have to discern how phonetic variation fits together to form phonological primitives. The variation may be phonetic in nature, that is, dependent on factors such as rate of speech, degree of stress or other prosodic factors, and elasto-dynamic constraints on articulators. It may also be due to social factors, as with style-shifting and social and class variation. In addition, researchers have to consider how variation interacts with the speech production/speech perception opposition. The means of studying production generally involve impressionistic auditory transcription or acoustic analysis, while analysis of perception usually entails cognitive experiments. Different kinds of variables also require different approaches. As broad categories, consonants, vowels, prosody, and, though it has barely been studied by variationists, voice quality, all require distinct sorts of analyses, and within each category individual variables need their own kinds of analysis.

The shifting sands of theory and technology create more challenges. Theoretical stances in phonology, such as generativism, autosegmental phonology, optimality theory, and exemplar theory, have at times induced variationists to adjust aspects of how they study data. However, variationists have often been content to let phonology work out its own issues without adapting phonological theories to sociolinguistics or vice versa (see Honeybone, this volume).