Effects of Multimodal AI on Emotional Engagement and Cognitive Strategies in Speech Training

Abstract

This study, framed within sociocultural theory, investigates the impact of multimodal AI-assisted speech training on foreign language learners' emotional engagement and cognitive strategies. This research employed a mixed-methods approach to analyze the emotional experiences and cognitive regulation mechanisms for Chinese university students in an AI-assisted speech training environment. Results reveal that learners exhibit moderate-to-high enjoyment and moderate-to-low anxiety in this environment. Foreign language enjoyment and foreign language anxiety are significantly negatively correlated. No significant linear relationship was found between the frequency of AI use and emotional experience. Qualitative analysis indicates that learners actively and critically integrate AI feedback, demonstrating considerable cognitive strategizing and agency. The study identifies limitations of AI in providing emotional support and simulating authentic pressure, especially its difficulty in replacing the emotional interaction and contextual creation functions of human teachers. The results unveil the double-edged sword effect of AI in speech training, underscore the necessity of constructing a human-machine collaborative teaching ecosystem, and provide empirical evidence for the design and integration of intelligent pedagogical tools.