Hybrid Machine Learning Models for Improving EFL Reading Fluency via Real-Time Feedback in Adaptive Systems

Comprehensive research on first language reading demonstrates that reading fluency is essential for proficient reading. Nonetheless, schemata, cognitive and metacognitive strategies, and comprehension contribute to effective reading. Therefore, it is imperative to develop effective strategies for improving the fluency of EFL readers. Auditory reading models for repeated reading and comprehensive reading programs for English as a foreign language learners have shown promising outcomes. This study examines how EFL learners can enhance their fluency by employing an auditory reading model-assisted repeated reading technique. The proposed system follows a technique of preprocessing, feature selection, and model training. Feature selection is conducted on preprocessed data, followed by the implementation of data normalisation. An AE is employed to derive low-dimensional characteristics from experimental reading data as the initial phase of the AERF methodology. The findings indicate that AERF demonstrates resilience, maintaining effective performance even when utilised on skewed datasets. The proposed method significantly enhances performance indicators, achieving an accuracy of 96.87%, an R of 98%. This study underscores the importance of integrating adaptive technologies that provide real-time feedback into EFL learning environments, illustrating their effectiveness in improving EFL reading fluency.