Dynamic Accent Accommodation: Training Learners for World Englishes using Generative Speech

Abstract

Traditional English Language Teaching (ELT) and Computer-Assisted Language Learning (CALL) have largely privileged native-speaker pronunciation models, despite the fact that much international communication occurs among speakers from Outer and Expanding Circle English contexts. This paper proposes Dynamic Accent Accommodation (DAA), a conceptual framework that integrates generative speech synthesis with World Englishes pedagogy to develop learners’ receptive flexibility across diverse English accents. Drawing on World Englishes theory, Communication Accommodation Theory, and the Lingua Franca Core, the framework introduces a three-phase Graduated Exposure Model supported by an adaptive human-in-the-loop speech generation architecture. The paper explains the technical logic of accent manipulation, pedagogical task sequencing, and evaluation metrics for measuring auditory processing efficiency, decoding accuracy, and communicative confidence. It also addresses ethical issues related to algorithmic bias, accent authenticity, and linguistic equity. The principal contribution of the study is the formulation of a theoretically grounded and pedagogically operational framework that can guide future empirical research on AI-supported listening instruction and Global Englishes education. Rather than promoting accent imitation, DAA emphasizes intelligibility, perceptual adaptation, and inclusive multilingual communication.