Investigating EFL students’ emotional responses in AI-assisted language learning: a Broaden-and-Build Theory approach

This study investigates the emotional responses of EFL students to AI-assisted language learning using ChatGPT and examines how these emotions influence engagement. Employing a mixed-methods approach, 215 Vietnamese university students used ChatGPT for language learning over a 12-week period. Data was collected through emotion-focused questionnaires and semi-structured interviews. Results reveal a dynamic emotional landscape, with positive emotions (interest, enjoyment, pride) increasing over time, while anxiety decreased. Positive emotions were associated with broader engagement in learning activities, supporting the Broaden-and-Build Theory. However, moderate levels of frustration also positively impacted engagement, challenging simplistic views of negative emotions in learning. Students’ perceptions of AI intelligence influenced their emotional responses, sometimes fostering confidence but occasionally leading to feelings of inadequacy. The study highlights the complex interplay between emotions and engagement in AI-assisted language learning, suggesting that imperfect AI systems can create valuable learning opportunities. These findings have significant implications for the design and implementation of AI-assisted language learning tools, emphasizing the need for emotionally adaptive systems that balance challenge with support and foster a sense of partnership between learners and AI.