A Deep Learning-Based Blended Model for Enhancing English Proficiency in Sports Students through Athlete Training Modules

Abstract

Sports students often face challenges in language proficiency, which can impede their overall academic and professional development. Traditional language learning methods may not fully engage them, particularly in environments primarily focused on physical training. Combining athletic training with language acquisition offers the potential to enhance both cognitive and linguistic abilities. This study aims to design and evaluate a deep learning-based blended model to improve English proficiency among sports students through specialised athlete training modules. A deep learning model, combining Binary Chimp-Assisted Residual Network (BC-RN), was developed to track and analyse student progress in both athletic training and language acquisition. The model was implemented via interactive training modules that integrated language-focused exercises alongside sports-specific drills. Data were collected from a cohort of 100 sports students, with pre- and post-assessments measuring improvements in English language proficiency (vocabulary, listening, and speaking) and physical performance (strength, agility), and pre-processed using Z-score normalisation. The findings revealed that BC-RN outperformed other existing models, achieving an accuracy of 85.61%. Furthermore, the AUC increased from 0.92 to 0.95, while the RMSE decreased from 0.25 to 0.15. These results demonstrate that BC-RN is the most effective model for predicting blended teaching performance, offering more accurate and reliable predictions. Student engagement and feedback were positive, indicating that the dual approach enhanced learning in both domains. Future research should explore the scalability of the model to other disciplines and incorporate real-time adaptive learning features to further personalise the learning experience.