Formats and implementations of exercises for collocation learning: Learning outcomes and students’ beliefs

Abstract Collocation‐focused exercises in language courses often require students to choose the right words from a given set of candidates to reassemble broken‐up collocations. Other exercises invite the students themselves to supply the missing words to complete collocations. One would expect exercises to serve the purpose of retrieval practice after students are first exposed to the collocations, but textbook analyses have revealed that they are also used in a spirit of trial and error. Collocation exercises thus show variation in format and in implementation, the combinations of which may yield different learning outcomes. The present study applied a 2 × 2 counterbalanced within‐participant design, where 56 ESL learners (international students enrolled in a TESOL program) tackled multiple‐choice or gap‐fill exercises on 32 verb–noun collocations (e.g., “catch fire”), half of which they had first been given a chance to study. This initial exercise was followed in the same session by a practice test to remind the participants of the correct collocations. One week later, they sat a delayed posttest. The gap‐fill format implemented as retrieval practice produced the best learning outcomes whereas the multiple‐choice format implemented in a trial‐and‐error fashion produced the poorest, despite feedback on both the initial exercise and the practice test. Despite its poorer effectiveness, one third of the participants expressed a preference for the multiple‐choice format in retrospective interviews.