How Does GenAI Agency Drive Academic Engagement Among Second Language Learners? An Empirical Study Based on Self-Determination Theory

Generative artificial intelligence (GenAI) holds significant potential for enhancing academic engagement among English learners, yet its effectiveness largely depends on students' agency in its use. Grounded in Self-Determination Theory (SDT), this study examines the mediating role of students' AI usage agency in influencing academic engagement. Data were collected via an online questionnaire comprising five validated scales from 644 undergraduate English students across multiple Chinese universities. Structural equation modeling (SEM) was conducted using AMOS 24, yielding several key findings. GenAI support and students’ agency beliefs both significantly and positively predicted the fulfillment of their basic psychological needs (autonomy, competence, and relatedness). Furthermore, the fulfillment of these three basic psychological needs exerted significant direct positive effects on students' learning engagement, with competence demonstrating the strongest impact. Moreover, GenAI support and student agency exerted significant indirect effects on learning engagement through the mediating role of basic psychological needs. Findings indicate that in GenAI-enhanced learning environments, external technological support and students’ intrinsic agency jointly drive deep academic engagement by satisfying their basic psychological needs. The study concludes by discussing educational implications and future research directions.