# Linguistic Bias in ChatGPT: Language Models Reinforce Dialect Discrimination

arXiv: https://arxiv.org/abs/2406.08818 · HTML: https://mlsanity.com/paper/2406.08818/

## Key facts

- Cited by 56 papers in the map (data through 14 Oct 2026).
- Builds on 6 of its 53 references that are in the map.
- First posted on arXiv 13 Jun 2024.

## Builds on (most cited first, 6 of 6)

- [The State and Fate of Linguistic Diversity and Inclusion in the NLP World](https://mlsanity.com/paper/2004.09095/) (2020)
- [Unintended Impacts of LLM Alignment on Global Representation](https://mlsanity.com/paper/2402.15018/) (2024)
- [Multi-VALUE: A Framework for Cross-Dialectal English NLP](https://mlsanity.com/paper/2212.08011/) (2022)
- [Prompting Multilingual Large Language Models to Generate Code-Mixed Texts: The Case of South East Asian Languages](https://mlsanity.com/paper/2303.13592/) (2023)
- [Dialect prejudice predicts AI decisions about people's character, employability, and criminality](https://mlsanity.com/paper/2403.00742/) (2024)
- [Evaluation of African American Language Bias in Natural Language Generation](https://mlsanity.com/paper/2305.14291/) (2023)

## Built on by (most cited first, 30 of 56)

- [Assessing Dialect Fairness and Robustness of Large Language Models in Reasoning Tasks](https://mlsanity.com/paper/2410.11005/) (2024)
- [Rejected Dialects: Biases Against African American Language in Reward Models](https://mlsanity.com/paper/2502.12858/) (2025)
- [What's Not Said Still Hurts: A Description-Based Evaluation Framework for Measuring Social Bias in LLMs](https://mlsanity.com/paper/2502.19749/) (2025)
- [Ads that Talk Back: Implications and Perceptions of Injecting Personalized Advertising into LLM Chatbots](https://mlsanity.com/paper/2409.15436/) (2024)
- [Unsupervised Concept Vector Extraction for Bias Control in LLMs](https://mlsanity.com/paper/2502.19721/) (2025)
- [Invisible Filters: Cultural Bias in Hiring Evaluations Using Large Language Models](https://mlsanity.com/paper/2508.16673/) (2025)
- [EnDive: A Cross-Dialect Benchmark for Fairness and Performance in Large Language Models](https://mlsanity.com/paper/2504.07100/) (2025)
- [Fragile Preferences: A Deep Dive Into Order Effects in Large Language Models](https://mlsanity.com/paper/2506.14092/) (2025)
- [Trans-EnV: A Framework for Evaluating the Linguistic Robustness of LLMs Against English Varieties](https://mlsanity.com/paper/2505.20875/) (2025)
- [Different Demographic Cues Yield Inconsistent Conclusions About LLM Personalization and Bias](https://mlsanity.com/paper/2601.18486/) (2026)
- [Performance in a dialectal profiling task of LLMs for varieties of Brazilian Portuguese](https://mlsanity.com/paper/2410.10991/) (2024)
- [DiffSampling: Enhancing Diversity and Accuracy in Neural Text Generation](https://mlsanity.com/paper/2502.14037/) (2025)
- [Data Caricatures: On the Representation of African American Language in Pretraining Corpora](https://mlsanity.com/paper/2503.10789/) (2025)
- [MEDEQUALQA: Evaluating Biases in LLMs with Counterfactual Reasoning](https://mlsanity.com/paper/2510.12818/) (2025)
- [How Does "English (US)" Become the Default? Triangulating Structural Bias Towards American English Across the LLM Pipeline](https://mlsanity.com/paper/2604.04204/) (2026)
- [Data-Centric AI Governance: Addressing the Limitations of Model-Focused Policies](https://mlsanity.com/paper/2409.17216/) (2024)
- [Artificial Impressions: Evaluating Large Language Model Behavior Through the Lens of Trait Impressions](https://mlsanity.com/paper/2510.08915/) (2025)
- [Steering LLMs toward Korean Local Speech: Iterative Refinement Framework for Faithful Dialect Translation](https://mlsanity.com/paper/2511.06680/) (2025)
- [IDIOLEX: Unified and Continuous Representations for Idiolectal and Stylistic Variation](https://mlsanity.com/paper/2604.04704/) (2026)
- [Topics as Proxies for Sociodemographics: How Conversational Context Affects LLM Answers](https://mlsanity.com/paper/2606.02776/) (2026)
- [Finding A Voice: Exploring the Potential of African American Dialect and Voice Generation for Chatbots](https://mlsanity.com/paper/2501.03441/) (2025)
- [Harnessing Test-time Adaptation for NLU tasks Involving Dialects of English](https://mlsanity.com/paper/2503.12858/) (2025)
- [Ethics and Persuasion in Reinforcement Learning from Human Feedback: A Procedural Rhetorical Approach](https://mlsanity.com/paper/2505.09576/) (2025)
- [I Think, Therefore I Am Under-Qualified? A Benchmark for Evaluating Linguistic Shibboleth Detection in LLM Hiring Evaluations](https://mlsanity.com/paper/2508.04939/) (2025)
- [DiaLLM: An Investigation into the Robustness-Generation Gap in English Dialect Adaptation](https://mlsanity.com/paper/2607.07669/) (2026)
- [The Language of Weather: Social Media Reactions to Weather Accounting for Climatic and Linguistic Baselines](https://mlsanity.com/paper/2407.07683/) (2024)
- [Diversidade lingu'istica e inclus~ao digital: desafios para uma ia brasileira](https://mlsanity.com/paper/2411.01259/) (2024)
- [Whispers of Many Shores: Cultural Alignment through Collaborative Cultural Expertise](https://mlsanity.com/paper/2506.00242/) (2025)
- [Comparative Analysis of Large Language Models for the Machine-Assisted Resolution of User Intentions](https://mlsanity.com/paper/2510.08576/) (2025)
- [Automated Coding of Communication Data Using ChatGPT: Consistency Across Subgroups](https://mlsanity.com/paper/2510.20584/) (2025)

## Abstract (arXiv)

We present a large-scale study of linguistic bias exhibited by ChatGPT covering ten dialects of English (Standard American English, Standard British English, and eight widely spoken non-"standard" varieties from around the world). We prompted GPT-3.5 Turbo and GPT-4 with text by native speakers of each variety and analyzed the responses via detailed linguistic feature annotation and native speaker evaluation. We find that the models default to "standard" varieties of English; based on evaluation by native speakers, we also find that model responses to non-"standard" varieties consistently exhibit a range of issues: stereotyping (19% worse than for "standard" varieties), demeaning content (25% worse), lack of comprehension (9% worse), and condescending responses (15% worse). We also find that if these models are asked to imitate the writing style of prompts in non-"standard" varieties, they produce text that exhibits lower comprehension of the input and is especially prone to stereotyping. GPT-4 improves on GPT-3.5 in terms of comprehension, warmth, and friendliness, but also exhibits a marked increase in stereotyping (+18%). The results indicate that GPT-3.5 Turbo and GPT-4 can perpetuate linguistic discrimination toward speakers of non-"standard" varieties.