In defense of frequency generalizations and usage-based linguistics. An answer to Frederick Newmeyer’s “Conversational corpora : when big is beautiful”

In his paper “Conversational corpora : when big is beautiful”, Newmeyer sets himself the goal of evaluating the relationship between corpus size and conclusions drawn from corpora regarding questions of grammatical theory. He formulates a strong critique against corpus research based on too small (conversational) corpora and in doing so, explicitly rejects the usage-based approach to language in which they are embedded. He argues that, unless they are based on large (conversational) corpora, frequency analyses do not give sufficiently reliable analyses compared to introspection-based analyses. In this response, I will counter some of the critique that Newmeyer levels against usage-based (or frequency-based) models, showing that, first of all, his criticism needs to be reevaluated and secondly, frequency-based analyses (and a usage-based approach more generally) do imply a radically different view on grammar which surpasses some of the shortcomings of introspection-based models.