The renewed interest in qualitative methods has led to a re-examination of assumptions about human nature and the distinction between the natural and human sciences. In this article we take a different approach and suggest that we focus on the role of language in our investigations. We note that a late development in the natural sciences has been a greater attention to the use of language in empirical inquiry. Language may be considered as a form of action by which phenomena are distinguished, grouped, and ordered in such a way as to see relationships between qualitatively heterogeneous events. The evaluation of language accounts—broadly conceived as theories, models, case studies, concepts, narrative descriptions, and so forth—should therefore be according to instrumental rather than metaphysical criteria. The consequences of this approach to language are demonstrated in a model investigative situation of social interaction analysis. Implications are drawn about descriptive activity, the status of our descriptive accounts, subject matter, and appropriate tasks for an empirical psychology.