Evaluating the Efficacy of Explainable Artificial Intelligence Frameworks in Supporting ESL Learners within Tertiary Education

Abstract

This research assesses how XAI supports ESL learners at the tertiary education level. With the growing use of AI in language learning, many systems are “black boxes” and provide little transparency in the feedback generated. This opacity limits learners’ understanding, trust, and effective use of feedback, especially in academic situations for which ESL learners need more support. This research explores XAI’s role in improving comprehension, engagement, and autonomy in AI-enhanced environments. A mixed-methods approach was taken with ESL students at a higher education institution. Data was collected from language performance tasks, structured questionnaires and semi-structured interviews to capture as much of the quantitative and qualitative data as possible. The XAI framework in the research provided interpretable feedback which provided learners with a clearer understanding of the rationale of the corrections and suggestions. Learners using XAI systems achieved better language performance, feedback interpretation and engagement than learners using conventional AI systems. Increased trust and confidence in the learning process were reported by participants. This emphasizes, in educational technology, the value of Explainable AI. This research illustrates the importance of incorporating explainability into AI language learning systems, and adds to the emerging research in the AI and ESL education space. It advocates for the use of explainable AI in higher education as a means to promote more effective learner-centric approaches.