Beyond Native Norms: Decolonizing AI Speech Recognition for Accent Equity in Nigerian ESL Classrooms
Center of Artificial Intelligence · 07.08.2026 · 1 (2) · 111-121
DOI: 10.65591/cai-206-2026 ↗Kurz gefasst
Fachartikel in der Zeitschrift Center of Artificial Intelligence, online erschienen am 07.08.2026. Verfasst von Mustapha Bayaro und 3 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref). Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The increasing adoption of Artificial Intelligence (AI)-powered Automatic Speech Recognition (ASR) has transformed pronunciation instruction by providing learners with immediate, individualized feedback. However, contemporary speech recognition models exhibit substantial Word Error Rate (WER) disparities ranging from 20% to over 40% higher error rates when processing non-native, regional, or Outer Circle Englishes compared to Inner Circle benchmarks. This paper examines the relationship between native-centric AI acoustic models and pronunciation development in Nigerian ESL classrooms. Informed by Flege's Speech Learning Model–Revised, Best's Perceptual Assimilation Model, Schmidt's Noticing Hypothesis, Swain's Output Hypothesis, Vygotsky's Sociocultural Theory, Krashen's Affective Filter Hypothesis, and the Technology Acceptance Model, the paper synthesizes current scholarship on L2 speech learning, World Englishes, and AI-assisted instruction. It argues that while ASR technologies enhance phonological awareness and autonomous practice, they frequently misclassify sociolinguistically legitimate features of educated Nigerian English as pronunciation errors due to algorithmic bias. To address this feedback paradox, the paper proposes the Eco-Contextual Automatic Speech Recognition (EC-ASR) Pedagogical Model, which integrates AI feedback with teacher mediation, sociolinguistic validation, and communicative practice. The paper concludes that pronunciation instruction must prioritize international intelligibility over native-speaker imitation and recommends developing open-access African-accented speech corpora.
Zitieren
APA
Bayaro, M., Olorunfemi, P. A., Michael, C. I., & Adeleke, H. A. (2026). Beyond Native Norms: Decolonizing AI Speech Recognition for Accent Equity in Nigerian ESL Classrooms. Center of Artificial Intelligence, 1(2), 111-121. https://doi.org/10.65591/cai-206-2026
Vancouver
Bayaro M, Olorunfemi PA, Michael CI, Adeleke HA. Beyond Native Norms: Decolonizing AI Speech Recognition for Accent Equity in Nigerian ESL Classrooms. Center of Artificial Intelligence. 2026;1(2):111-121. doi:10.65591/cai-206-2026
BibTeX
@article{Bayaro2026,
author = {Bayaro, Mustapha and Olorunfemi, Philip Abayomi and Michael, Chinyere Immaculate and Adeleke, Helen Adeola},
title = {Beyond Native Norms: Decolonizing AI Speech Recognition for Accent Equity in Nigerian ESL Classrooms},
journal = {Center of Artificial Intelligence},
year = {2026},
volume = {1},
number = {2},
pages = {111--121},
doi = {10.65591/cai-206-2026}
}
EUVIMED