Does Generative Artificial Intelligence Enhance Language Learning? A Meta-Analysis

• In Event: Teaching and Learning with GenAI

Abstract

Generative Artificial Intelligence (GenAI) is reshaping language learning, yet its effectiveness remains debated. This meta-analysis of 23 experimental/quasi-experimental studies addresses two questions: whether GenAI impacts language learning, and if factors like subject, education level, duration, region, etc., affect effectiveness. Results show GenAI has a significant positive effect compared with traditional methods (g = 0.62, 95% CI [0.49, 0.75], Z = 9.63). Furthermore, its influence on academic outcomes (g = 0.82, 95% CI [0.58, 1.06]) exceeds its impacts on cognitive outcomes (g = 0.32, 95% CI [0.09, 0.54]) and affective outcomes (g = 0.44, 95% CI [0.10, 0.78]). Factors such as region, education level, teaching duration, and technical interventions significantly moderate the effectiveness. Findings highlight differentiated strategies and context-sensitive implementation.

Authors

• Jinwen Qu, East China Normal University

• Yinuo Li, East China Normal University

• Jun Qian, East China Normal University

• Jing Leng, East China Normal University

• Lixin Ye, East China Normal University