L'accent cleaning par l'intelligence artificielle : enjeux éthiques, responsabilité épistémique et reproduction des hiérarchies linguistiques dans les technologies vocales AI-Powered Speech Recognition: Ethical Issues, Epistemic Responsibility, and the Reproduction of Linguistic Hierarchies in Voice Technologies
Article Sidebar
Page Numbers:
353-368
Digital Object Identifier:
10.58578/ijemt.v4i2.11742
Viewed: 0 times
Downloaded: 0 times
LYAS Publisher cordially invites qualified professionals to serve as Editors or Reviewers. Your expertise will make an important contribution to maintaining and strengthening the academic quality of our publications. Interested applicants are kindly requested to complete the application form at the following link: Editors & Reviewers
Please feel free to contact us if you need any further information about the submission process or if you have any additional questions.
Authors:Nourreddine Bensaid1*, Malika Bahmad2
Copyright :Authors retain copyright and grant the journal right of first publication.
Main Article Content
Abstract
The widespread deployment of AI based speech technologies is profoundly reshaping contemporary language practices. These systems, primarily trained on corpora from standardized or dominant varieties, tend to normalize phonetic features that deviate from the reference norm a phenomenon known as "accent cleaning." This paper investigates this process through a dual question: to what extent do automatic speech recognition (ASR) systems contribute to the reproduction of linguistic and social hierarchies, and what responsibility do language researchers bear in the face of these biases? Our central hypothesis is that these models prioritize certain phonetic norms at the expense of diversity, producing forms of symbolic exclusion for speakers of marginalized accents. An exploratory comparative analysis is conducted on three ASR systems (Google SpeechtoText, OpenAI Whisper, Amazon Transcribe), based on a speech corpus from French speaking speakers of three contrasting varieties (Standard French, Moroccan French, Sub-Saharan French). Anticipated results suggest significant asymmetries in word error rates (WER) and active normalization mechanisms. The study aims to empirically document these biases and to examine the researcher's role within a digital ethics framework.
Citation Metrics:
Downloads
Citation Metrics & Similar Scopus Articles
-
Wang J. (2027)Analysis and Applications of Neuromorphic Memristors in Artificial Intelligence ComputingNano Micro Letters, 19(1)
-
Li X. (2027)Localization challenges and technological gaps in cleaning robots for fixed-tilt coplanar photovoltaic arrays: A review and perspectiveUnconventional Resources, 17
-
Miyazono S. (2027)Improved Efficiency and Lesion Detection in Small Bowel Capsule Endoscopy Using the Open-Source Artificial Intelligence Model SEE-AIDen Open, 7(1)
Article Details
Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
References
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? ???? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (pp. 610–623). Association for Computing Machinery. https://doi.org/10.1145/3442188.3445922
Bourdieu, P. (1982). Ce que parler veut dire: L’économie des échanges linguistiques. Fayard.
Calvet, L.-J. (1974). Linguistique et colonialisme: Petit traité de glottophagie. Payot.
Calvet, L.-J. (2024). Linguistique et colonialisme: Petit traité de glottophagie. Lambert-Lucas.
Commission nationale de l’informatique et des libertés. (2017). Comment permettre à l’Homme de garder la main? Les enjeux éthiques des algorithmes et de l’intelligence artificielle. https://www.cnil.fr/fr/comment-permettre-lhomme-de-garder-la-main-rapport-sur-les-enjeux-ethiques-des-algorithmes-et-de
Dent, R. J. (2022). Modeling regional accents of French for inclusive speech recognition [Master’s dissertation, University of Malta]. OAR@UM. https://www.um.edu.mt/library/oar/handle/123456789/137280
Graham, C., & Roll, N. (2024). Evaluating OpenAI’s Whisper ASR: Performance analysis across diverse accents and speaker traits. JASA Express Letters, 4(2), Article 025206. https://doi.org/10.1121/10.0024876
Koenecke, A., Nam, A., Lake, E., Nudell, J., Quartey, M., Mengesha, Z., Toups, C., Rickford, J. R., Jurafsky, D., & Goel, S. (2020). Racial disparities in automated speech recognition. Proceedings of the National Academy of Sciences, 117(14), 7684–7689. https://doi.org/10.1073/pnas.1915768117
Maison, L., & Estève, Y. (2023). Some voices are too common: Building fair speech recognition systems using the Common Voice dataset [Preprint]. arXiv. https://doi.org/10.48550/arXiv.2306.03773
Milroy, J. (2001). Language ideologies and the consequences of standardization. Journal of Sociolinguistics, 5(4), 530–555. https://doi.org/10.1111/1467-9481.00163
Villani, C., Schoenauer, M., Bonnet, Y., Berthet, C., Cornut, A.-C., Levin, F., & Rondepierre, B. (2018). Donner un sens à l’intelligence artificielle: Pour une stratégie nationale et européenne. https://www.vie-publique.fr/rapport/37225-donner-un-sens-lintelligence-artificielle-pour-une-strategie-nation