Effectiveness of second language collocation instruction: a meta-analysis

Abstract This study reports a meta-analysis of studies that investigated the effectiveness of instruction on second language collocation learning. A total of 64 research projects in 17 primary studies (the number of participants = 3,859) were included in the meta-analysis. Aggregated results confirmed the effectiveness of collocation instruction for facilitating second language collocation learning (d = 1.415, 95% CI: 1.189, 1.641). The results of moderator analysis indicated that the intervention method was a significant predictor of intervention effectiveness. To be specific, 1) explicit interventions produced larger effect sizes than implicit interventions; 2) interventions with low session frequency did not produce larger effect sizes than those with high session frequency; 3) long interventions did not show larger effect sizes than short interventions; and 4) constrained-constructed responses, selected responses, and multiple outcome measures did not produce larger effect sizes than free-constructed responses. The study also offered explanations for the results, as well as implications for the teaching and research of collocations.