The tireless study of human conduct by social scientists is predicated on the well-founded assumption that regularities in human behavior can be detected, described, and explained. Sociolinguistic studies of speech in its social context demonstrate this well, with a great variety of research showing how variations in speech can be systematically related to speaker characteristics and to facets of the situation (e.g., Ervin-Tripp, 1969; Hymes, 1972; Labov, 1972; Trudgill, 1978). We have argued elsewhere (Smith, Giles, & Hewstone, 1980), however, that preoccupation with the linguistic side of the sociolinguistic endeavor has fostered a casual and sometimes rather naive use of social variables in explanations for linguistic variation. In particular, Giles and Hewstone (in press) expressed serious misgivings about the way that the concept of situation has been employed in sociolinguistics. They drew attention to what they called the “taxonomic” approach to the definition of social situations, an approach that emphasizes the static, objective features of situations and neglects the dynamic nature of social interaction and its consequences for speech. Proponents of this kind of approach strive to create lists and taxonomies of influential situational variables on the basis of commonsense assumptions about how to operationalize situations in meaningful ways, usually without checking these assumptions empirically, or checking them only by referring to the very data that their taxonomies are meant to explain. Moreover, taxonomists provide few suggestions as to how elements of situations combine to influence behavior and tend to ignore the ongoing or unfolding nature of social interaction.