Mobile-Assisted Listening Comprehension Assessment: Enhancing English Listening Skills Through Interactive Mobile Platforms

The ability to comprehend spoken English is a basic skill requirement among the students but most encounter difficulties due to variations in accents, speech rate and the absence of personalized feedback in the conventional educational approaches. Mobile-assisted learning platforms offer potential solutions, but effective integration of advanced speech recognition technology has not been extensively explored. This research targets the development and testing of a mobile-supported learning platform based on the use of the Wav2Vec 2.0 model as a self-supervised learning method to enhance learner listening. The contribution to the research is the use of Wav2Vec 2.0 with a certain mobile adaptation to learners. The model is also able to learn through untagged audio and can be refined, allowing it to isolate different accents and fast speech where traditional language learning fails. L2-AR corpus was the training data and consisted of several listening comprehension exercises and the actual speech by English in the real world. It was notable that an impressive improvement was found in the experimental comparison of listening comprehension between the experimental group when applying the Wav2Vec 2.0-based platform with a mean gain of$25.2+-6.3$points, compared to the control group that only had a slight improvement. Embedded Wav2Vec 2.0 into a mobile-assisted learning system could greatly enhance the listening-comprehension skills of learners depending on the presentation of real-time customized feedback and adaptive learning. Indeed, the piece demonstrates the positive value of using advanced models of speech recognition in language acquisition and facilitating scalable and interactive education.