What drives college students to use AI for L2 learning? Modeling the roles of self-efficacy, anxiety, and attitude based on an extended technology acceptance model

Abstract

Prior research highlights the critical role of AI in enhancing second language (L2) learning. However, the factors that practically affect L2 learners to engage with AI resources are still underexplored. Given the widespread availability of digital devices among college students, they are particularly poised to benefit from AI-assisted L2 learning. As such, this study, grounded in an extended Technology Acceptance Model (TAM), investigates the predictors of college L2 learners' actual use of AI tools, focusing on AI self-efficacy, AI-related anxiety, and their overall attitude toward AI. Data was gathered from 429 L2 learners at Chinese universities via an online questionnaire, utilizing four established scales. Through structural equation modeling (SEM) via AMOS 24, the results indicate that AI self-efficacy could negatively affect AI anxiety, and positively influence both learners' attitude toward AI and their actual use of AI tools. Besides, AI anxiety negatively predicted the actual use of AI. Moreover, AI self-efficacy was a positive predictor of AI use through reducing AI anxiety, enhancing attitude toward AI, or a combination of both. This study also discusses the theoretical and pedagogical implications and suggests directions for future research.