Answers are grounded in this paper and cite their sources.
Using process-tracing technologies such as webcam-based eye-tracking, screen recording, and stimulated recall, this study explored how South Korean STEM graduate students employed ChatGPT in their self-initiated and naturalistic English writing settings. The analysis of three 30-minute writing sessions and stimulated recalls, as well as semi-structured interviews from four participants, revealed the complex and dynamic patterns of tool use and behaviors in ChatGPT-mediated writing. At the tool level, participants employed a range of digital tools alongside ChatGPT, guided by personal preferences and writing community norms. At the behavioral level, they demonstrated that using ChatGPT in L2 writing involved not only diverse gaze and non-gaze behaviors but also individualized sequences of these behaviors to meet specific writing goals. The closer analysis of the two participants’ cases showed idiosyncratic patterns in tool usage and behavior. One participant used ChatGPT in a brief and targeted way, focusing only on specific vocabulary and expression searches. The other adapted ChatGPT-related behaviors dynamically based on the purpose and stage of writing. By unobtrusively tracing L2 writers’ real-time evolutions during the writing process, this study suggests that ChatGPT has urged us to rethink conventional writing stages and behaviors and to develop new research methods that can capture such dynamic processes of AI-mediated L2 writing.
No takes yet. Share an insight, caveat, or question.
Sung et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: