Exploring University Students’ Acceptance of AI Chatbots in Language Learning: An Extended TAM Model Based on the Interaction Hypothesis

With the growing use of generative AI in language education, AI chatbots have become key tools in university English learning. Drawing on the Technology Acceptance Model(TAM) and the Interaction Hypothesis in second language acquisition to propose an integrated framework. The framework examines how comprehensible input, negotiation of meaning, pushed output, and feedback influence learners’ perceived ease of use, perceived usefulness, and behavioral intention to use AI chatbots. A total of 527 valid responses were collected through a questionnaire survey. Structural equation modeling was employed for data analysis. The results show that all core paths of TAM are supported. Perceived ease of use significantly enhances both perceived usefulness and behavioral intention. Comprehensible input, pushed output, and feedback significantly affect perceived usefulness. Feedback also significantly increases perceived ease of use. Although negotiation of meaning shows no significant direct effect, it demonstrates a meaningful indirect effect through perceived usefulness. Mediation analyses further reveal that all four interaction-related factors influence behavioral intention, with feedback exhibiting a notable chain mediation effect. This study provides theoretical insights into the acceptance mechanisms of AI-based language learning tools. It also offers practical implications for improving learner experience through interface design.