This study investigates how master students majoring in Business Studies engage with the corpus toolkit AntConc in a data-driven learning (DDL) intervention targeting non-finite phraseological patterns in academic writing. Thirty-five Tunisian EFL students participated in a five-week course combining guided practice with independent concordance searching. Six annotated patterns, drawn from a business-domain corpus, were deployed for structural and functional analysis. Learner activity during DDL tasks was tracked through screen recordings for a sub-sample of 20 students and supplemented by written reflection logs. Tracking captured query variety, search refinement, time on task, use of AntConc features, pattern documentation, and discourse function recognition, adapted from established frameworks (Pérez-Paredes et al., 2011, 2012). Learning outcomes were assessed through pre-, post-, and delayed tests. Results showed significant gains in the use of target structures, though retention varied across learners. Correlations indicated that varied querying and systematic notetaking were associated with higher gains. Thematic analysis of the reflection logs corroborated these findings, revealing increased phraseological awareness but persistent challenges in discourse function identification. Thereby, the study demonstrates the value of integrating behavioural monitoring into corpus-based instruction, supporting the design of scaffolded, tool-mediated environments that foster phraseological competence and learner autonomy.
Amor et al. (Sun,) studied this question.