AI-driven sentiment analysis for mitigating foreign language anxiety (FLA) in EAP: A proof-of-concept study

Abstract

Foreign Language Anxiety (FLA) is a persistent affective barrier in English for Academic Purposes (EAP), yet little research has examined how artificial intelligence (AI) might support learners emotionally. This article presents a proof-of-concept study investigating the feasibility of using AI-driven sentiment analysis to identify and mitigate FLA. A small Padlet-based corpus of student reflections (n = 41) was analysed using GPT-4-based sentiment classification, followed by an AI-mediated reframing activity completed by a volunteer sub-sample of learners (n = 14). Rather than making generalisable empirical claims, the study explores the potential and limitations of large language models as tools for affective scaffolding. Results indicate that students frequently express mixed emotions, combining anxiety with hope and motivation, and that AI-supported reframing may promote short-term reassurance and increased confidence. The paper discusses methodological and ethical considerations and outlines how affect-aware AI tools could be meaningfully integrated into EAP pedagogy.