AI 챗봇을 활용한 영어교육 프로그램이 언어교육 효과에 미치는 영향에 대한 메타분

Meta-analysis on the effect of English education programs using AI Chatbots on language education effectiveness

This study aimed to analyze the effect size of English education programs using AI chatbots on EFL learners’ language education effectiveness as well as moderator variables affecting program effectiveness. Forty-four dissertations and academic journals published from 2002 to March 2023 were selected as the analysis subjects, and the effect size was calculated using a random effect model due to high heterogeneity among studies. A meta-regression analysis was performed to confirm moderation effects, and publication bias was verified using Egger’s regression test. The effect size of AI education programs was significant, with a moderate effect size (0.428; 95% CI: 0.310-0.546), and major variables affecting language education effectiveness included school level by age and length for a single class. In terms of school level, it showed a higher effect size in the order of college students, high school students, and elementary school students. The most effective length of a single class was found to be 51-60 minutes. The time range of 51-60 minutes is mainly associated with university student learning. From both the school level and the length of a single class perspectives, the effect size related to university students is the largest. This can be attributed to the higher likelihood of university students engaging in self-directed learning compared to younger learners. It is believed that self-directed learning time has contributed significantly to improving learning effectiveness. This study provides useful information for developing and application of AI education programs for EFL learners.