There is general agreement in present-day linguistics that the subject is at its best when it is empirical. However, there are a number of apparently incompatible views on what makes language study truly empirical, and even what counts as the right sort of data for the linguist to study. Siobhan Chapman offers a fresh approach to this debate by comparing it to some remarkably similar disagreements about data, methodology and the nature of empiricism in mid-twentieth century philosophy, disagreements that were largely provoked by reactions to the ideas of the Vienna Circle. Her main focus is a comparison of the work of J. L. Austin and the less well know work of Arne Naess. Despite significant differences, both said things about language that have striking resonance with much more recent claims in linguistics, particularly in fields such as corpus linguistics that deal with ‘real life’ examples of language use.