Emotional Experiences and Evaluative Judgement in AI-Supported EFL Reading Circles

Generative AI can provide immediate and interactive support in EFL learning, yet learners must judge whether its outputs are trustworthy and educationally useful. Guided by control-value theory and evaluative judgement, this qualitative embedded single-case study examined reported feelings, comparisons across successive accounts, and evaluative reasoning in two AI-supported reading-circle cycles. Reflective coursework from 29 Chinese university EFL students and interviews with four focal students provided the primary data; field notes and classroom observation from ten presentation lessons contextualised implementation. The coding record contained 136 mentions of emotion-related expressions, with satisfaction most frequently recorded. Paired comparisons connected feelings with particular functions and consequences of AI use, including comprehension support, unsuitable analysis, correction efforts, and reliance concerns. Focal accounts connected standards concerning accuracy, relevance, effort, and ownership with quality verdicts and reported or contemplated use decisions. Interpretation distinguished reported attempts to manage interaction, judgements about particular outputs, and claimed responsibility for final use. Appreciation of efficient assistance could coexist with reservations about accuracy, task suitability, and authorship. The findings inform pedagogical propositions concerning explicit criteria, source checking, and responsibility for AI-supported work.