mlsanityData to 14 Oct 2026

Data Caricatures: On the Representation of African American Language in Pretraining Corpora

arXiv 2503.10789 ↗First posted 13 Mar 2025cs.CLcited by 3

Not on the map. Why?

Abstract (arXiv)

With a combination of quantitative experiments, human judgments, and qualitative analyses, we evaluate the quantity and quality of African American Language (AAL) representation in 12 predominantly English, open-source pretraining corpora. We specifically focus on the sources, variation, and naturalness of included AAL texts representing the AAL-speaking community. We find that AAL is underrepresented in all evaluated pretraining corpora compared to US demographics, constituting as few as 0.007% and at most 0.18% of documents. We also find that more than 25% of AAL texts in C4 may be perceived as inappropriate for LLMs to generate and to reinforce harmful stereotypes. Finally, we find that most automated filters are more likely to conserve White Mainstream English (WME) texts over AAL in pretraining corpora.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 32 of its 86 references that are in the map. First posted on arXiv 13 Mar 2025.

Builds on

32 of 86 references.

References · 32

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  3. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  4. 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,326
  5. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,100
  6. 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,411
  7. 2024The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scalecited by 1,160
  8. 2020Language (Technology) is Power: A Critical Survey of "Bias" in NLPcited by 956
  9. 2016Bag of Tricks for Efficient Text Classificationcited by 876
  10. 2021Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpuscited by 492
  11. 2024Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Researchcited by 483
  12. 2024DataComp-LM: In search of the next generation of training sets for language modelscited by 398
20 more
  1. 2018SWAG: A Large-Scale Adversarial Dataset for Grounded Commonsense Inferencecited by 333
  2. 2021Annotators with Attitudes: How Annotator Beliefs And Identities Bias Toxic Language Detectioncited by 264
  3. 2023From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Modelscited by 260
  4. 2019Racial Bias in Hate Speech and Abusive Language Detection Datasetscited by 209
  5. 2016Demographic Dialectal Variation in Social Media: A Case Study of African-American Englishcited by 196
  6. 2022What Does it Mean for a Language Model to Preserve Privacy?cited by 195
  7. 2024Towards Measuring and Modeling "Culture" in LLMs: A Surveycited by 167
  8. 2024QuRating: Selecting High-Quality Data for Training Language Modelscited by 157
  9. 2024Infini-gram: Scaling Unbounded n-gram Language Models to a Trillion Tokenscited by 153
  10. 2024How to Train Data-Efficient LLMscited by 93
  11. 2017Racial Disparity in Natural Language Processing: A Case Study of Social Media African-American Englishcited by 71
  12. 2020Demoting Racial Bias in Hate Speech Detectioncited by 62
  13. 2024Linguistic Bias in ChatGPT: Language Models Reinforce Dialect Discriminationcited by 56
  14. 2020Investigating African-American Vernacular English in Transformer-Based Text Generationcited by 53
  15. 2023Evaluation of African American Language Bias in Natural Language Generationcited by 36
  16. 2024Participation in the age of foundation modelscited by 32
  17. 2025Rejected Dialects: Biases Against African American Language in Reward Modelscited by 20
  18. 2021Mitigating Racial Biases in Toxic Language Detection with an Equity-Based Ensemble Frameworkcited by 19
  19. 2020Dialect Diversity in Text Summarization on Twittercited by 17
  20. 2022Corpus-Guided Contrast Sets for Morphosyntactic Feature Detection in Low-Resource English Varietiescited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Dialect vs Demographics: Quantifying LLM Bias from Implicit Linguistic Signals vs. Explicit User Profilescited by 4
  2. 2026DialectLLM: A Dialect-Aware Dialog[ue] Generation Framework Beyond Standard American Englishcited by 0
  3. 2026Whose Story Gets Told? Positionality and Bias in LLM Summaries of Life Narrativescited by 0