Comparative Effects of AI and Peer Collaboration on L2 Writing Proficiency: An Empirical Study
4
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This study examines the effectiveness of ChatGPT as a pedagogical tool in L2-English writing instruction over a semester. With L1-Korean college students as participants, the study compares the outcomes of a peer-interacted group (n = 34) with those of an AI-interacted group (n = 42). Both groups wrote argumentative essays responding to the same prompts and completed identical pre- and post-writing activities, including brainstorming and debating given essay topics. The only difference lay in the learners’ interaction partners during the activities: peers or ChatGPT. The results of a pretest and four posttests (three immediate and one delayed), scored on content, language use, and organization, show that the AI-interaction group consistently outperformed the peer-interaction group, specifically in content on the first posttest and language use on subsequent posttests. Furthermore, the peer-interaction group’s scores sharply declined on the delayed posttest, while the AI-interaction group maintained consistent performance, suggesting lasting benefits of interacting with ChatGPT. Follow-up interviews, however, revealed the learners’ ambivalence toward using ChatGPT in their writing process. The students appreciated the AI’s prompt and effective editing; however, they raised concerns about its lack of creativity and originality as well as the ethics of incorporating its content suggestions into their essays. Pedagogical implications and potential avenues for future research are discussed.
Abstract
This study examines the effectiveness of ChatGPT as a pedagogical tool in L2-English writing instruction over a semester. With L1-Korean college students as participants, the study compares the outcomes of a peer-interacted group (n = 34) with those of an AI-interacted group (n = 42). Both groups wrote argumentative essays responding to the same prompts and completed identical pre- and post-writing activities, including brainstorming and debating given essay topics. The only difference lay in the learners’ interaction partners during the activities: peers or ChatGPT. The results of a pretest and four posttests (three immediate and one delayed), scored on content, language use, and organization, show that the AI-interaction group consistently outperformed the peer-interaction group, specifically in content on the first posttest and language use on subsequent posttests. Furthermore, the peer-interaction group’s scores sharply declined on the delayed posttest, while the AI-interaction group maintained consistent performance, suggesting lasting benefits of interacting with ChatGPT. Follow-up interviews, however, revealed the learners’ ambivalence toward using ChatGPT in their writing process. The students appreciated the AI’s prompt and effective editing; however, they raised concerns about its lack of creativity and originality as well as the ethics of incorporating its content suggestions into their essays. Pedagogical implications and potential avenues for future research are discussed.