Sociophonetic Analysis of ASR Performance: Investigating Socially Patterned Bias in Newcastle English

Summary

Automatic Speech Recognition (ASR) systems are widely deployed in everyday communication, education, healthcare, and industry, yet their performance remains uneven across speakers, particularly when dialectal variation diverges from the mainstream accents represented in training data. This study, authored by Kevin Tang and posted to arXiv (2603.24549), investigates ASR bias through a sociolinguistic analysis of Newcastle English. The work applies a sociophonetic lens to examine how socially patterned accent variation affects recognition accuracy, providing a case study of how speakers of non-mainstream British dialects may be systematically disadvantaged by current ASR technology. The paper is relevant to researchers in NLP, speech technology, and computational sociolinguistics, as well as practitioners concerned with fairness and inclusivity in voice-based systems. The full abstract and preprint are available at https://arxiv.org/abs/2603.24549.