ScholarWorks@중앙대학교

초록

This study investigates the impact of Generative Artificial Intelligence(GenAI) voice interaction on English-speaking competence and perception in an English communication course. Thirty-five Korean university students in the experimental group engaged in GenAI conversations as end-of-class takeaways using a one-minute paper(OMP) approach, while 32 control group students completed the OMP without GenAI assistance. The study assessed speaking competence and perception of GenAI usefulness. The results showed that the experimental group made more gains in vocabulary complexity using diverse vocabulary(root type-token ratio) and rich words(lexical density) as well as syntactic complexity(the ratio of clauses per T-units) at the final assessment. Improvements in accuracy were observed, with clause-level accuracy(error-free clauses) at the midterm and T–unit-level accuracy at the final assessment(p = .048, 95% CI [-.00, .99]), with the latter showing marginal significance. In contrast, the control group outperformed only in lexical density at the midterm. In addition, the experimental group perceived that GenAI conversations helped improve speaking fluency and accuracy. Regression analysis further revealed that GenAI conversations were associated with fewer verb errors at the final assessment. These findings suggest that targeted GenAI use as end-of-class takeaways can potentially enhance productive language skills, supporting diverse, sophisticated, and meaningful language use.

키워드