CULTURAL IDENTITY AND TECHNOLOGICAL REPRESENTATION IN NIGERIAN NEWS CASTING: EVALUATING AI'S LANGUAGE SPECIFICITY
Ezikov Svyat (Orbis Linguarum) · 30.10.2026 · 24 (3) · 201-213
DOI: 10.37708/ezs.swu.bg.v24i3.13 ↗Kurz gefasst
Fachartikel in der Zeitschrift Ezikov Svyat (Orbis Linguarum), online erschienen am 26.10.2026. Verfasst von Samuel Oyeyemi Agbeleoba.
Abstract (Originalsprache)
This study empirically investigates the capacity of GPT-4, a leading commercial Large Language Model (LLM), to interpret, represent, and evaluate the linguistic nuances of Nigerian English (NE) within Nigerian news broadcasting. Employing a mixed-methods approach, combining critical discourse analysis of AI-generated text with an empirical evaluation, the study used a multi-sample "matched-guise probing" technique. This involved testing GPT-4's performance across journalistic and analytical tasks—including content rewriting, professionalism evaluation (on a 1–10 scale), sentiment analysis, and named-entity recognition—using four distinct text samples, each with a Standard English (SE) and a Nigerian English (NE) variant. The primary contribution is the empirical finding of a consistent bias against Nigerian English: GPT-4 systematically revised NE features towards non-Nigerian linguistic norms, a process termed "algorithmic standardization." It assigned significantly lower professionalism scores to NE texts, citing local linguistic features as "informalities" or "errors." This bias extended to core NLP tasks, with lower confidence in sentiment analysis for NE texts and failure to recognize culturally embedded named entities. These results demonstrate a form of algorithmic gatekeeping that devalues a legitimate World English, highlighting the risks of perpetuating linguistic prejudice. The study calls for more inclusive, culturally aware AI systems and strategic AI implementation in Nigerian media, contributing empirical data to AI ethics and postcolonial digital studies.
Zitieren
APA
Agbeleoba, S. O. (2026). CULTURAL IDENTITY AND TECHNOLOGICAL REPRESENTATION IN NIGERIAN NEWS CASTING: EVALUATING AI'S LANGUAGE SPECIFICITY. Ezikov Svyat (Orbis Linguarum), 24(3), 201-213. https://doi.org/10.37708/ezs.swu.bg.v24i3.13
Vancouver
Agbeleoba SO. CULTURAL IDENTITY AND TECHNOLOGICAL REPRESENTATION IN NIGERIAN NEWS CASTING: EVALUATING AI'S LANGUAGE SPECIFICITY. Ezikov Svyat (Orbis Linguarum). 2026;24(3):201-213. doi:10.37708/ezs.swu.bg.v24i3.13
BibTeX
@article{Agbeleoba2026,
author = {Agbeleoba, Samuel Oyeyemi},
title = {CULTURAL IDENTITY AND TECHNOLOGICAL REPRESENTATION IN NIGERIAN NEWS CASTING: EVALUATING AI'S LANGUAGE SPECIFICITY},
journal = {Ezikov Svyat (Orbis Linguarum)},
year = {2026},
volume = {24},
number = {3},
pages = {201--213},
doi = {10.37708/ezs.swu.bg.v24i3.13}
}
EUVIMED