Description
Abstract
This chapter explores the effectiveness and perceptions of ChatGPT as an Automated Writing Evaluation (AWE) tool in supporting second language (L2) English writing development among international students. Previous research has highlighted the benefits of AWE systems for providing immediate feedback; however, these tools primarily focus on sentence-level errors. With the advent of large language models (LLMs) such as ChatGPT, there is a need to evaluate their potential as AWE tools, particularly for addressing global aspects of writing to complement existing systems. This study applied Chapelle’s (2001) CALL framework to investigate ChatGPT’s language learning potential through qualitative and quantitative analysis of interactions with eight international students enrolled in ESL courses at a U.S. university. Data collected from student-ChatGPT conversation logs, screen recordings, surveys, semi-structured interviews, and stimulated recalls were analyzed to identify learners’ interaction patterns, their perceived effectiveness of ChatGPT’s feedback, and how these changed across three sessions. Findings indicate that ChatGPT is valued for addressing global aspects of writing, such as content matter, organization, and flow, but is less effective in correcting sentence-level issues like grammar. Participants reported mixed feelings about the feedback, often finding it overly general and positive. Despite these limitations, learners acknowledged ChatGPT’s usefulness for idea generation and process-oriented writing. Implications suggest that ChatGPT can serve as a supplementary tool to complement teacher feedback, enhancing writing development in language learning contexts. .