Foreign Language Speaking Anxiety and Enjoyment on ASR-Based and Human-Rated Read-Speech Evaluation

Foreign language speaking anxiety (FLSA) is a critical factor in students’ oral proficiency, while positive states like foreign language speaking enjoyment (FLSE) are also believed to be influential. However, the competing predictive power of these two opposing emotions is not well understood. This mixed-methods study investigates the unique predictive power of FLSA and FLSE on oral assessment scores, using a combination of human-rated and automatic speech recognition (ASR)-based evaluations. Participants were 159 first-year students from two non-English departments at a state university in Indonesia. They completed oral assessments (a read-aloud task and an impromptu speech task) rated by their teacher and analyzed using Speechace, an ASR-based language learning platform. Results showed a significant negative correlation between FLSA and FLSE. Crucially, multiple regression analyses revealed that FLSA was a strong, unique, and dominant negative predictor across almost all measures, including human-rated scores, ASR-based IELTS scores, and most notably, impromptu speech scores. In stark contrast, the predictive power of FLSE became non-significant for these core measures when FLSA was included in the model. This suggests that the influence of enjoyment on performance is overshadowed by the more dominant, competing effect of anxiety. The findings imply that FLSA is the primary affective barrier to speaking performance. For pedagogy, this highlights that managing and reducing anxiety is a more critical and direct prerequisite for improving performance than solely focusing on boosting enjoyment.