The study discusses the application of a data-driven learning (DDL) method to teaching Polish as a foreign language. The presented study aims to describe DDL in teaching Polish as a foreign language by taking as an example alternations of phonemes ó:o in nouns (cf. samochód 'car':[bez] samochodu '[without] car').
The author analyses the data from a manually annotated one-million-word subcorpus of the National Corpus of Polish. The research builds on the procedure of DDL which consists of three steps: identification, classification, and generalisation.
The present approach has a potential to enrich classroom activities and has a favourable impact on the development of language teaching methods.