AI in Academic English: Predicting EAP Learners’ Speaking Improvements Through TAM Constructs

This study examines the acceptance and effectiveness of EAP Talk, an AI-powered tool designed to improve English for Academic Purposes (EAP) learners' speaking skills. Utilizing the Technology Acceptance Model (TAM), the research explores how perceived usefulness (PU), perceived ease of use (PEOU), and attitudes (AT) affect behavioral intention (BI) to use the tool and how these constructs predict actual improvement (AI) in speaking performance. A sequential explanatory mixed-methods approach was employed, with quantitative data from 94 participants complemented by qualitative insights from semi- structured interviews. The results revealed significant relationships between TAM constructs and actual improvement in speaking skills, with behavioral intention having the strongest predictive effect. Qualitative findings supported the quantitative data, highlighting the tool's effectiveness and areas for potential enhancement, particularly in feedback mechanisms. This study bridges the gap in existing literature by linking TAM constructs to measurable performance improvements in EAP learners' speaking skills.