Psychological mechanisms of AI integration in ESL teaching: teacher self-efficacy and classroom practice in Ghanaian senior high schools.

The integration of artificial intelligence (AI) into language education has gained increasing attention, yet research has predominantly focused on student outcomes, with limited attention to teachers' roles in mediating these technologies. This study investigates the psychological mechanisms underlying ESL teachers' integration of AI tools in reading and writing instruction in Ghanaian senior high schools. Grounded in the Technology Acceptance Model and teacher self-efficacy theory, the study adopts a qualitative research design drawing on semi-structured interviews, classroom observations, and stimulated recall interviews. The findings indicate that while teachers generally perceive AI tools as useful and moderately easy to use, these perceptions do not directly translate into pedagogical practice. Instead, teacher self-efficacy emerges as a central psychological mechanism shaping the extent and depth of AI integration in classroom contexts. Teachers with higher confidence demonstrate more structured and pedagogically meaningful use, whereas others rely on limited, surface-level applications. The findings also reveal a persistent gap between positive attitudes and enacted practice, shaped by contextual constraints such as limited infrastructure and time, as well as teachers' capacity to adapt AI tools within instructional processes. The study contributes to AI in education research by emphasizing the role of teacher psychology as a mediating layer between technological potential and instructional adaptation, offering a more nuanced understanding of how AI-supported practices are enacted in resource-constrained educational contexts.