My work is informed by theories and practices in the field of linguistics that view language structure as rooted in language use. These usage based approaches to language study share the assumption that grammar is shaped by social and cognitive activities of speakers. Scholars working in these areas of linguistic inquiry base their analyses on naturally occurring language, emphasizing the roles of frequency, function, and cognition to explain the emergence and conventionalization of linguistic form and meaning. This view of language structure as dynamic--in both its development and change--contrasts foundationally and methodologically with formal linguistic theories that often treat grammar as abstract, separate from its uses by speakers in communities. From a functionalist perspective "grammars code best what speakers do most" (Du Bois 1985: 363). In this view, then, the study of grammatical structure becomes the study of discourse, social interaction, and expression.
The majority of my research projects have been analyses of distributions of lexical and grammatical elements in American English conversations. Identification of associations between frequently occurring linguistic elements and their functions provides a view of the expressive and interactional underpinnings of grammatical categories and constructions as they appear in conversational contexts. For example, analyses of conversational data suggest that speakers' attitudes, evaluations, and metalinguistic commentaries (subjectifying and metapragmatic elements) have a robust influence on the distribution of utterances found in English interactive discourse, countering the traditional view that discourse is primarily referential (2002, 2009). I also found evidence of formal differences between inclusive we (referring to the speaker and the addressee) and exclusive we (referring to the speaker and others, but not the addressee) that suggest a communicative basis for this well documented cross linguistic feature (inclusive and exclusive pronouns) (2004, 2014). Another project highlighted the stance-related uses of a particular type of generalizing expression in English conversations--utterances containing subject noun phrases that index or evoke classes (2007).
Because traditional linguistic categories have been derived without discourse data, the fitness--or sometimes lack of fitness--between these categories and actual conversational tokens provides a potential test of how well these grammatical and lexical classes hold up when applied to interactive discourse. Evaluating (dis)continuities between analytical categories and conversational data has been a unifying aspect of both my research and teaching that very much engages me.