Control and Value Appraisals in AI-Mediated Language Learning: A Multi-Dimensional Analysis of EFL Learners' Emotions, Engagement and Affective Judgements

Abstract

The rapid expansion of AI-supported instructional tools has reshaped how emotions, motivation and engagement operate in language learning environments. Guided by Control–Value Theory (CVT), this study examines how learners' emotional appraisals, technology-related affective attitudes and engagement behaviours collectively shape their judgements about the balance between AI-driven support and human instruction. A sample of 738 Chinese English as a Foreign Language (EFL) learners from eight universities completed validated measures of Foreign Language Enjoyment (FLE), Foreign Language Classroom Anxiety (FLCA), academic engagement/disengagement, technology-related affective attitudes and perceptions of AI–human interaction balance. Using Structural Equation Modelling (SEM) (AMOS 24) and complementary analyses in SPSS 27, the results showed that enjoyment and positive affect toward technology significantly promoted engagement, whereas anxiety exerted a negative influence. Engagement, in turn, strongly predicted learners' evaluations of AI–human instructional balance and functioned as a mediating mechanism linking emotional and attitudinal appraisals to AI-related perceptions. Significant direct effects from FLE, FLCA and technology attitudes to AI–human balance also emerged, indicating a partially mediated structural pattern. These findings highlight the centrality of control and value appraisals in shaping both engagement and affective judgements within AI-mediated learning settings. The study extends CVT to technology-enhanced language learning and provides practical guidance for designing emotionally supportive and pedagogically balanced AI-integrated EFL environments. Implications for teaching, instructional design and policy, along with recommendations for future research, are presented.