The Nexus between Engagement and Pronunciation Gains in AI-Powered Tools: A Correlational Inquiry among Malian Tertiary Students

Abstract

This study investigates the relationship between student engagement and pronunciation gains when using AI powered tools among Malian tertiary level English language learners. Conducted as a quantitative correlational inquiry with 150 undergraduate students, the research employed a pretest/posttest design to measure pronunciation improvement after a 12week intervention using the AI driven application ELSA Speak. Engagement was operationalized through behavioral metrics such as time investment, session frequency, and task completion, aggregated into a composite Immersion Index. Results revealed a statistically significant improvement in pronunciation scores from pretest to posttest, with a large effect size (Cohen’s d = 1.37). Strong positive correlations were found between all engagement indices and pronunciation gains, with the Immersion Index showing the strongest association (r = .72). Regression analysis confirmed engagement as a robust predictor of learning outcomes, explaining 52% of the variance in gains. Mediation analysis indicated that intrinsic motivation partially mediated the engagement gain relationship, accounting for 20% of the effect. Subgroup analysis showed a stronger correlation for female participants. The findings underscore the critical role of sustained and motivated engagement in maximizing the effectiveness of AI assisted pronunciation training, particularly in resource constrained contexts such as Mali. The study contributes to the growing literature on AI mediated language learning in African higher education and offers practical implications for curriculum design, teacher training, and educational policy aimed at leveraging AI tools to enhance English phonetic proficiency.