THE IMPACT OF AI-ENHANCED TASK-BASED LANGUAGE TEACHNG (TBLT) ONORAL FLUENCY IN HIGHER EDUCATION: A MIXED-METHODS STUDY

This study investigates the efficacy of integrating Generative AI (GenAI) scaffolding within a Task-Based Language Teaching (TBLT) framework to improve English oral fluency among university students. While TBLT has long been established as a premier methodology for communicative competence, limited practice opportunities and delayed feedback often hinder progress. In this 16-week study involving 60 EFL (English as a Foreign Language) students, an "Intelligent-TBLT" model was employed, utilizing AI tools like ChatGPT-4 and ELSA Speak to provide immediate feedback and digital scaffolding. Results from a mixed-methods analysis indicate significant improvements in speech rate, reduced hesitation, and increased learner confidence. The findings suggest that while AI cannot replace human social nuances, it serves as a critical "digital collaborator" that reduces cognitive load and fosters learner autonomy.