This paper outlines the theoretical background of data-driven learning (DDL) and its hands-on application in French Language Teaching, as well as the didactic potential of using text corpora functionalities to extract collocations. The second part focuses on the analysis of a project conducted among students of Romance Philology. Its objective was a corpus analysis of selected French terms and their Polish equivalents on the theme of social issues. The study reported a good understanding of the principles of corpus analysis by the students and an attempt to draw conclusions based on word frequency and extracted collocations. It was found that DDL could effectively deepen the development of lexical competence, especially as regards lexical combinations. Another conclusion points out that this approach allows the acquisition of extralinguistic competences – the student’s autonomy in learning a foreign language, critical thinking and heuristic skills.