LinguaPilot: A GAI-based framework for self-regulated EFL reading

Abstract

Generative artificial intelligence (AI) shows great promise for computer-assisted language learning (CALL), but its native application may lead to limited benefits. This study introduces LinguaPilot, an innovative framework and online platform that integrates generative AI technologies to enhance self-regulated learning (SRL) among university-level English as a Foreign Language (EFL) learners. Following Zimmerman's Cyclical Phases Model, LinguaPilot incorporates AI features into pre-reading, during-reading, and post-reading stages, corresponding to forethought, performance, and self-reflection phases, respectively. Through extensive experiments in text complexity classification and question generation, this study has uncovered the optimal results among various optimization techniques, demonstrating effective methods for harnessing AI's potential. Accessible via an intuitive online interface, LinguaPilot makes a tentative fulfillment of AI's transformative role in enhancing self-regulated language learning and pioneers the application of generative AI in EFL reading.