Cognitive and emotional dynamics of artificial intelligence-assisted English learning: A mixed-methods study in a dual instructional context

Artificial intelligence (AI) tools have become increasingly integrated into university-level English learning, yet their pedagogical role within input-based instruction and English-medium education remains insufficiently understood. This mixed-methods study examines how first-year students in a Thai higher education institution use AI applications to support English learning, how they emotionally experience AI-assisted learning, and how the frequency of AI use relates to these emotional dynamics. The study was situated within a year-long storytelling-based English curriculum grounded in input-based pedagogy informed by Krashen's Input Hypothesis, alongside broader English-medium instruction across non-language courses. Survey data from 82 students, using validated subscales on AI use and emotional responses, were analyzed alongside reflective essays and interviews with 11 students. Quantitative findings indicate that students engage with AI moderately to frequently, primarily for grammar checking, writing assistance, and idea generation. Students reported increased confidence and excitement but also experienced frustration and concerns about dependency. Correlation analyses revealed strong positive associations between AI use and confidence and excitement, with moderate relationships with frustration and dependency. Qualitative findings portray AI as a supportive “learning partner” that enhances clarity and self-directed revision, while simultaneously encouraging reliance on automated guidance. The study contributes to current debates on AI in language education by demonstrating how, within input-based and English-medium learning environments, AI can both facilitate learner autonomy and risk over-dependence, accentuating the need for pedagogical strategies that cultivate reflective, critical, and gradually more independent language learning.