메타분석에 의한 AI 챗봇 기반 영어교육 프로그램의 교육효과 비교

Abstract

This study endeavors to assess the impact of AI chatbot-based English education programs on the effectiveness of language education for EFL learners. Additionally, it aims to examine moderator variables that influence the efficacy of these programs. A total of seventy-nine dissertations and academic journals, spanning from 2002 to October 2023, were chosen as the subjects for analysis. The effect size was computed utilizing a random effects model. A meta-regression analysis was conducted to validate moderation effects, and Egger's regression test was employed to assess publication bias. The impact of AI education programs was statistically significant, demonstrating a moderate effect size (0.44; 95% Confidence intervals: 0.34-0.54). The results of comparing educational effects in cognitive and non-cognitive domains revealed a higher effect size in the cognitive domain. The effect sizes based on educational outcomes ranked highest to lowest as follows: achievement, motivation, interest, confidence, attitude, and immersion. This study provides valuable insights into areas that need improvement to enhance the educational effects of utilizing AI chatbots. In future designs of English education programs utilizing AI chatbots, enhancing the pronunciation recognition component to ameliorate negative attitudes towards AI and diversifying conversation patterns to elevate learner immersion are anticipated to improve overall educational effectiveness.