The Language Ideologies of GenAI: An Appraisal Analysis of Chatbots’ Sociolinguistic Judgments and Vocal Stereotyping
Journal of Education, Language, and Ideology · 29.10.2026 · 4 (2) · 330-356
DOI: 10.67372/kdd5da47 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Education, Language, and Ideology, online erschienen am 29.10.2026. Verfasst von Zeynep Arslan und 1 weiteren Autor:innen.
Abstract (Originalsprache)
The study examined the language ideologies of common Generative AI (GenAI) bots. Four chatbots were given a sociolinguistic evaluation task, ChatGPT 3.5, Gemini, Copilot, and Perplexity. Based on speech samples of three distinct varieties of English (African American Vernacular English, British English, and Singaporean English) the chatbots were prompted to describe the person who produced each sample. The researchers posited that chatbots’ outputs could reveal what types of language ideologies each bot has, likely based on the LLM and human feedback used to train the bot. The chatbots descriptions were analyzed using Appraisal Theory, and characteristics of each bot’s language ideologies were mapped out. The findings reveal that all the chatbots were capable of indexing linguistic markers to social identities. However, surprisingly the bots vary significantly in terms of the language ideologies they subscribe to. While some bots actively cautioned against making judgments about a speaker’s race, educational background or socioeconomic class based on a speech sample, other bots assigned judgments that reflected a greater degree of vocal stereotyping. Furthermore, the same chatbots were given the same prompt 14 months later to determine whether the bots’ language ideologies had shifted over time as they receive more human input and training. The implications of the study point to how researchers can elucidate the language ideologies, including implicit biases, of chatbots.
Zitieren
APA
Arslan, Z., & Sayer, P. (2026). The Language Ideologies of GenAI: An Appraisal Analysis of Chatbots’ Sociolinguistic Judgments and Vocal Stereotyping. Journal of Education, Language, and Ideology, 4(2), 330-356. https://doi.org/10.67372/kdd5da47
Vancouver
Arslan Z, Sayer P. The Language Ideologies of GenAI: An Appraisal Analysis of Chatbots’ Sociolinguistic Judgments and Vocal Stereotyping. Journal of Education, Language, and Ideology. 2026;4(2):330-356. doi:10.67372/kdd5da47
BibTeX
@article{Arslan2026,
author = {Arslan, Zeynep and Sayer, Peter},
title = {The Language Ideologies of GenAI: An Appraisal Analysis of Chatbots’ Sociolinguistic Judgments and Vocal Stereotyping},
journal = {Journal of Education, Language, and Ideology},
year = {2026},
volume = {4},
number = {2},
pages = {330--356},
doi = {10.67372/kdd5da47}
}
EUVIMED