A Corpus-Based Assessment of French CEFR Lexical Content

The French CEFR vocabulary profiles as presented in the Referentiels - while a valuable resource for a wide range of applications focused on lexical content - result from introspection, intuition-based judgements, and unquanliftable experience. The result is a specification of vocabulary that has been largely untested from a corpus-based, empirical perspective. In contrast, the English profiles were corpus-derived using current computational practices by English Profile, an ongoing project. In this article we summarize our evaluation of the French profiles by comparing their content with two sizable contemporary corpora and derived artifacts. We quantify the vocabulary overlap and uniqueness across all three of these resources. This leads us to make recommendations about possible ways of refining French CEFR vocabulary content by retaining, adding, and removing vocabulary items.