AI-Mediated Feedback in Blended EFL Instruction: A Conceptual Framework for Enhancing Speaking Fluency in the Saudi University Context

AI-mediated feedback has become an increasingly prominent component of blended EFL instruction in Saudi higher education. Despite ongoing national reforms under Saudi Vision 2030 that emphasize digital transformation, many undergraduate learners continue to experience persistent difficulties in oral fluency, often due to limited opportunities for sustained speaking practice and heightened evaluation anxiety within crowded instructional settings. Moreover, existing research frequently focuses on isolated AI tools, such as Automatic Speech Recognition (ASR) or chatbots, without articulating a coherent, pedagogically grounded developmental sequence that connects these technologies. Addressing this gap, the present study adopts an integrative theoretical review to propose a three-stage conceptual framework for AI-mediated feedback in blended EFL instruction. The framework comprises private ASR-based rehearsal to support phonological accuracy, including segmental and suprasegmental features, while reducing speaking anxiety; structured chatbot-mediated interaction to foster pragmatic competence and interactional fluency through simulated communicative tasks; and generative-AI-supported reflection to enhance feedback literacy, metacognitive awareness, learner autonomy, and self-regulated performance refinement. The model is deliberately aligned with the cultural and affective needs of Saudi learners, particularly female undergraduates, by emphasizing privacy, guided practice, and non-judgmental learning spaces. The study concludes by outlining pedagogical, assessment, and policy implications to support the ethical and sustainable integration of AI tools in Saudi universities and proposes a future research agenda to empirically validate the framework’s impact on oral fluency development.