Recurrent strings in corpus-based pedagogical research: A reappraisal of the field

Formulaic competence is a hotly debated issue in teaching circles, not only because of its role in L2 communication but also due to the inherent complexity of the identification criteria for formulaic strings. While the mixed approach, combining meaning-based and corpus- based identification measures, remains a natural solution, the subjective character of the criteria, together with the required involvement of native experts, diminishes its attractiveness for every-day pedagogical purposes. We would like to explore the potential of “corpus-only” identification tools. Specifically, our objective is to show that meaningless n-grams (of the, in a, etc.) generated by frequency searches contain useful pedagogical data, and that, coupled with MI scores frequency-based measures accurately characterize learners’ formulaic competence. Because of the relative simplicity of the identification procedure and free availability of corpus tools, frequency-based and distribution-based measures may become an important new pedagogical tool at the disposal of language teachers