Exploring the Role of Grammar and Word Choice in Bias Toward African American English (AAE) in Hate Speech Classification
Open accessThe DOI matches its Crossref record; 10 of the 10 references it cites were found in a registry or an index.
Abstract
Language usage on social media varies widely even within the context of American English. Despite this, the majority of natural language processing systems are trained only on “Standard American English,” or SAE, the construction of English most prominent among white Americans. For hate speech classification, prior work has shown that African American English (AAE) is more likely to be misclassified as hate speech. This has harmful implications for Black social media users as it reinforces and exacerbates existing notions of anti-Black racism. While past work has highlighted the relationship between AAE and hate speech classification, no work has explored the linguistic characteristics of AAE that lead to misclassification. Our work uses Twitter datasets for AAE dialect and hate speech classifiers to explore the fine-grained relationship between specific characteristics of AAE such as word choice and grammatical features and hate speech predictions. We further investigate these biases by removing profanity and examining the influence of four aspects of AAE grammar that are distinct from SAE. Results show that removing profanity accounts for a roughly 20 to 30% reduction in the percentage of samples classified as ’hate’ ’abusive’ or ’offensive,’ and that similar classification patterns are observed regardless of grammar categories.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
Automated Hate Speech Detection and the Problem of Offensive Language
Davidson, T., Warmsley, D., Macy, M., & Weber, I. (2017) · Proceedings of the International AAAI Conference on Web and Social Media · doi:10.1609/icwsm.v11i1.14955 · Cited by 2,610
The Risk of Racial Bias in Hate Speech Detection
Sap, M., Card, D., Gabriel, S., Choi, Y., & Smith, N. A. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1163 · Cited by 817 · cited 6×
“also propose the approach of racial priming, or educating data annotators on racial and dialectical differences before annotating training data, and show that this approach reduces bias [35].”
BackgroundResultInfluentialRacial Bias in Hate Speech and Abusive Language Detection Datasets
Davidson, T., Bhattacharya, D., & Weber, I. (2019) · Proceedings of the Third Workshop on Abusive Language Online · doi:10.18653/v1/w19-3504 · Cited by 414 · cited 4×
“In particular, prior works provide evidence that hate speech classification systems and sentiment analysis tend to classify AAE text as more likely to be hate speech or another form of toxic speech [8, 17, 22, 35].”
BackgroundMethodResultInfluentialExamining Gender and Race Bias in Two Hundred Sentiment Analysis Systems
Kiritchenko, S., & Mohammad, S. (2018) · Proceedings of the Seventh Joint Conference on Lexical and Computational Semantics · doi:10.18653/v1/s18-2005 · Cited by 384
Online Harassment and Content Moderation: The Case of Blocklists
Jhaver, S., Ghoshal, S., Bruckman, A., & Gilbert, E. (2018) · ACM Transactions on Computer-Human Interaction · doi:10.1145/3185593 · Cited by 270
Self-citationBlodgett, S. L., & O'Connor, B. (2017) · arXiv · doi:10.48550/arxiv.1707.00061 · Cited by 153
Toward a Description of African American Vernacular English Dialect Regions Using “Black Twitter”
Jones, T. (2015) · American Speech · doi:10.1215/00031283-3442117 · Cited by 144 · cited 1×
“Furthermore, AAE differs drastically by region and generation [20, 21].”
BackgroundThe N Word: Its History and Use in the African American Community
Rahman, J. (2012) · Journal of English Linguistics · doi:10.1177/0075424211414807 · Cited by 142
Designing Toxic Content Classification for a Diversity of Perspectives
Kumar, D., Kelley, P. G., Consolvo, S., Mason, J., Bursztein, E., Durumeric, Z., Thomas, K., & Bailey, M. (2021) · arXiv · doi:10.48550/arxiv.2106.04511 · Cited by 28
Cited by
Using large language models in psychology
Demszky, D., Yang, D., Yeager, D. S., Bryan, C. J., Clapper, M., Chandhok, S., Eichstaedt, J. C., Hecht, C., Jamieson, J., Johnson, M., et al. (2023) · Nature Reviews Psychology · doi:10.1038/s44159-023-00241-5 · Cited by 369
Self-citationAI generates covertly racist decisions about people based on their dialect
Hofmann, V., Kalluri, P. R., Jurafsky, D., & King, S. (2024) · Nature · doi:10.1038/s41586-024-07856-5 · Cited by 233
Feng, S., Park, C. Y., Liu, Y., & Tsvetkov, Y. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.656 · Cited by 149
Repairing the Cracked Foundation: A Survey of Obstacles in Evaluation Practices for Generated Text
Gehrmann, S., Clark, E., & Sellam, T. (2023) · Journal of Artificial Intelligence Research · doi:10.1613/jair.1.13715 · Cited by 123
Understanding Gender Bias in AI-Generated Product Descriptions
Kelly, M., Tahaei, M., Smyth, P., & Wilcox, L. (2025) · Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3715275.3732169 · Cited by 4
Metadata
- Authors
- Harris, C., Halevy, M., Howard, A., Bruckman, A., & Yang, D.
- Venue
- 2022 ACM Conference on Fairness Accountability and Transparency
- Publisher
- ACM
- Published
- 2022-06-20
- Pages
- 789-798
- OpenAlex
- W4283165820
- Cited by
- 35
- References
- 10
- FWCI
- 4.49
- Type
- conference paper
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.