This paper is a response to Lavandera's question regarding the limits of the study of language variation. Sociolinguistics is characterized by its desire to limit representational meaning much more narrowly than formal linguistics. In addition while formal linguistics views language as species-specific and designed to accomodate logical representations, sociolinguistics views language in the context of common biological inheritance. The proper goal of sociolinguistic theory might be stated as the apportionment of the variance on any sub-section of a linguistic system to the functions of representation, identification, and accomodation, and to predict for any new language the probable distribution of information. Thus, variation studies go beyond grammatical description to explanations of variable constraints which will lead to conclusions about the form of grammar. Although in its early days sociolinguistic analysis developed to study sociolinguistic stratification, the full value of variation analysis has only gradually become clear. In addition to its usefulness in describing phonological variation, variation analysis can be a tool in the description of syntactic and semantic analysis, and variation theory as a whole can be a heuristic device for determining the shape of linguistic theory. (AM)