The association between generative artificial intelligence use and foreign language classroom anxiety

This study examines the association between Generative Artificial Intelligence use and foreign language classroom anxiety (FLCA) among undergraduates in an under-resourced private university in Western China. Data were collected from 630 first- and second-year students using the Foreign Language Classroom Anxiety Scale (FLCAS). Analyses included descriptive statistics, t-tests, analysis of variance (ANOVA), Pearson correlations, and hierarchical regression. Results showed that gender significantly predicted frequency of Artificial Intelligence (AI) use (female > male, p < .01), but not FLCA or its two factors, Classroom Performance Anxiety (CPA) and Learning Process Anxiety (LPA). English proficiency significantly influenced both AI use frequency and FLCA (p < .05): higher-proficiency students used AI more frequently but reported lower anxiety, whereas age and socioeconomic status showed no significant effects. Regression analyses indicated that, after controlling for proficiency, frequency of AI use significantly predicted overall FLCA (B = 0.253, p < .001), CPA (B = 0.288, p < .001), and LPA (B = 0.235, p < .001). Moderation analysis further showed that English proficiency attenuated the relationship between AI use frequency and anxiety, particularly for LPA. The final models explained 11.9% of the variance in overall FLCA, 10.8% in CPA, and 11.3% in LPA. These findings suggest that frequent AI use may be associated with heightened anxiety in under-resourced contexts, while higher English proficiency serves an attenuating role, underscoring the need for scaffolded AI literacy training and differentiated instructional support when integrating AI tools.