Four challenges for usage-based linguistics.

Dirk Geeraerts has played a key role in launching Cognitive Linguistics as a full-fledged theory of linguistics and in expanding its sphere of influence in Western Europe. Dirk is furthermore one of the first and strongest advocates for the incorporation of empirical methods - and quantitative, corpus- based methods in particular - into cognitive linguistic research. The “Quantitative Turn” (Janda 2013) is in large part due to his relentless insistence on methodological rigour. In this chapter, I want to take a closer look at what is currently methodological “good practice” in the field and draw attention to some of the assumptions that underlie our methodology and thereby shape our findings yet have gone unquestioned. Four challenges are highlighted - data annotation, statistical analysis, model validation and experimental design - and their theoretical foundations and implications discussed.