Exploring the Dynamics of Learners’ Engagement With Generative Artificial Intelligence in Second Language Argumentative Writing

ABSTRACT Drawing on a three‐dimensional engagement framework and self‐determination theory's concept of psychological needs, this study adopted a mixed‐method approach to investigate the dynamics of 12 Chinese English‐as‐a‐second‐language (ESL) learners’ engagement with a generative artificial intelligence (GAI) tool, DeepSeek, during a 9‐week argumentative writing project. Data were collected through reflective journals, DeepSeek chat histories, and semi‐structured interviews and analyzed using a thematic analysis approach. The results revealed participants’ complex and dynamic engagement with DeepSeek across behavioral, cognitive, and emotional dimensions. In terms of behavioral and emotional dimensions, they tended to engage with DeepSeek in the initial and final phases of the writing project but disengage from it in the intermediate phase. In the cognitive dimension, their engagement with the tool emerged in the intermediate and final phases but not in the initial phase. Furthermore, this study demonstrated that participants’ behavioral, cognitive, and emotional (dis)engagement with DeepSeek in different project phases was associated with the satisfaction and frustration of their psychological needs for autonomy, competence, and relatedness. These findings will be informative for language teachers and educational institutions seeking to enhance effective integration of GAI tools into L2 writing education.