DR. SHIRISHA DESHPANDE , DR.A. VIJAYALAKSHMI , DR. B. CHANDANA , DR.M. SAMADHANAM EMIMAL , DR. GITASRI MUKHERJEE , DR MEENA GULLUR
DOI: https://doi.org/Communication competence in English is easily impaired in those non-native speakers due to the error in English pronunciation, which may adversely affect the intelligibility, confidence and general proficiency in the language. This paper is an investigation into a hybrid psychological-computational approach to the detection and correction of pronunciation errors in non-native English speech. Based on the theory of the second language acquisition and psycholinguistics, the study examines the cognitive and psychological issues of learners in generating speech with phonetic correctness. A contrastive learning-based model is constructed on the computational front to learn the subtle differences between the native and non-native pronunciation patterns. The system uses acoustic characteristics, phoneme embeddings and contextual inputs to detect deviations and produce an error feedback mechanism. Policymaking in an experimental environment proves the effectiveness of the model in increasing the accuracy of error detection, and providing adaptive correction strategies depending on the profile of the learner. The combination of psychological understanding makes the system not only a solution of phonetic mistakes but also a decrease in anxiety and a perpetual improvement in pronunciation among the learners. The interdisciplinary approach will lead to the expanding research into the Intelligent Language Learning System and introduce the opportunities of more individualized, empathetic, and efficient speech training.
