A custom GPT for English learning and well-being: A case study of Japanese EFL students

Abstract

This small-scale qualitative study examined how second language (L2) students' interactions with a researcher-developed custom GPT reflected Seligman's (2011) PERMA framework and how students perceived its use for English learning. Participants were six Japanese university students enrolled in a positive psychology-based English-language seminar. Over five weeks, students completed weekly PERMA-based written conversations with five domain-specific GPT versions. Thematic analysis of 30 conversation transcripts showed that student-GPT interactions reflected all five PERMA domains, with follow-up prompts often helping students move from brief responses toward concrete examples from their own lives. Post-survey responses indicated that students viewed the GPT as a psychologically safe and supportive space and valued its English feedback, particularly for noticing mistakes, learning alternative vocabulary, and practicing written interaction. However, some students found more personal or complex prompts difficult and expressed a need for more tailored support. Taken together, the findings suggest that PERMA-based GPT use may offer a low-stakes way to support well-being reflection while providing meaningful opportunities for English use and feedback in L2 contexts.