Elamur Research⌕
☰
home / papers / 2026 / ACID: On the Perception of Online Classism

ACID: On the Perception of Online Classism

Summary

In NLP, hate speech research has largely ignored discrimination based on socioeconomic status (SES). Few existing hate speech definitions even mention class. To fill this gap, the authors build ACID (Automatic Classism Identification Dataset), a corpus of more than 1.15 million English YouTube comments and Twitter posts from 14 English-speaking countries. Texts were gathered through lower-SES keywords covering both poverty and working-class identity.

The analytical framework draws on the Stereotype Content Model and a grounded-theory approach. It combines a small manually annotated subset with unsupervised topic modeling and LLM-based classification. Each SES-related text is examined on three dimensions:

  • which stereotypes it invokes,
  • whether it blames individuals or structural conditions,
  • whether it is offensive.

Across countries, the stereotypes look strikingly alike. They are mostly moral judgments that cast lower-SES people as lazy, ignorant, dependent or criminal. Individual blame dominates overall, but Twitter shows more structural explanations than YouTube. Offensiveness is high, especially when it accompanies individual blame. The resource gives researchers in NLP and the social sciences a basis for detecting classism and studying it across cultures.

Key points

  • Corpus size and sources: ACID contains 505,584 YouTube and 646,957 Twitter instances. The Twitter portion reuses the DRAX dataset (tweets from August to November 2022). YouTube comments were collected in September 2025 through the YouTube Data API.
  • YouTube filtering: videos had to have titles combining an SES keyword and a country name, plus at least 100 comments. English texts were kept using fasttext-langdetect. The US is the largest country subset on both platforms.
  • Annotation consistency: a pilot annotation of 150 YouTube comments by three annotators was used to check agreement.
  • Country differences: individual blame is strongest in the UK and US, both highly individualistic societies. Structural explanations are more visible in less individualistic countries such as India.
  • Availability and caveats: data and code are released on GitHub (elisabassignana/acid-online-classism). The authors note that keyword sampling, English-only data and LLM classification errors mean the findings probably underestimate how common online classism is.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/2myisgn9aju6

@inproceedings{muti2026acid,
  title = {{ACID: On the Perception of Online Classism}},
  author = {Muti, Arianna and Bassignana, Elisa and Curry, Amanda Cercas and Durante, Federica and Hovy, Dirk and Nozza, Debora},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/2myisgn9aju6},
  url = {https://doi.org/10.63317/2myisgn9aju6}
}

References

40 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mina Arzaghi, Florian Carichon, and Golnoosh Farnadi. 2024. Understanding intrinsic socioeconomic biases in large language models. find
  2. Elisa Bassignana, Amanda Cercas Curry, and Dirk Hovy. 2025. The AI gap: How socioeconomic status affects language technology interactions. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 18647–18664, Vienna, Austria. Associati… find
  3. Paul Bose, Lorenzo Lupo, Mahyar Habibi, Dirk Hovy, and Carlo Schwarz. 2024. Beyond the stats: Realities, perception, and social media discourse on poverty. In AEA Papers and Proceedings, volume 114, pages 690–694. find
  4. American Economic Association 2014 Broadway, Suite 305, Nashville, TN 37203. find
  5. Pierre Bourdieu. 1986. The forms of capital. In John Richardson, editor, Handbook of Theory and Research for the Sociology of Education, pages 241–258. Greenwood Press, New York. find
  6. Marco Brambilla, Simona Sacchi, Patrice Rusconi, and Geoffrey P. Goodwin. 2021. The primacy of morality in impression development: Theory, research, and future directions. In Advances in Experimental Social Psychology, volume 64, pages 1–66. Elsevier. find
  7. Amanda Cercas Curry, Giuseppe Attanasio, Zeerak Talat, and Dirk Hovy. 2024a. Classist tools: Social class correlates with performance in NLP. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 12643–12655, Bangkok, Thailand. Asso… find
  8. Amanda Cercas Curry, Zeerak Talat, and Dirk Hovy. 2024b. Impoverished language technology: The lack of (social) class in NLP. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 8675–8682, Torino, Italia.… find
  9. Catherine Cozzarelli, Anna V. Wilkinson, and Michael J. Tagler. 2001. Attitudes toward the poor and attributions for poverty. Journal of Social Issues, 57(2):207–227. find
  10. Georgina Curto, Svetlana Kiritchenko, Kathleen Fraser, and Isar Nejadgholi. 2024. The crime of being poor: Associations between crime and poverty on social media in eight countries. In Proceedings of the Sixth Workshop on Natural Language Processing and Computational Social Science (NLP+CSS 2024), p… find
  11. Georgina Curto, Svetlana Kiritchenko, Muhammad Hammad Fahim Siddiqui, Isar Nejadgholi, and Kathleen C. Fraser. 2025. Tackling social bias against the poor: a dataset and a taxonomy on aporophobia. In Findings of the Association for Computational Linguistics: NAACL 2025, pages 6995–7016, Albuquerque,… find
  12. Federica Durante and Susan T Fiske. 2017a. How social-class stereotypes maintain inequality. Current opinion in psychology, 18:43–48. find
  13. Federica Durante and Susan T. Fiske. 2017b. How social-class stereotypes maintain inequality. Current Opinion in Psychology, 18:43–48. find
  14. Federica Durante, Courtney Bearns Tablante, and Susan T Fiske. 2017. Poor but warm, rich but cold (and competent): Social classes in the stereotype content model. Journal of Social Issues, 73(1):138–157. find
  15. Susan T. Fiske, Amy J. C. Cuddy, Peter Glick, and Jun Xu. 2002. A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82(6):878–902. find
  16. Maarten Grootendorst. 2022. Bertopic: Neural topic modeling with a class-based tf-idf procedure. arXiv preprint arXiv:2203.05794. find our card
  17. Geert Hofstede. 1980. Culture’s Consequences: International Differences in Work-Related Values. Sage, Beverly Hills, CA. find
  18. Geert Hofstede, Gert Jan Hofstede, and Michael Minkov. 2010. Cultures and Organizations: Software of the Mind, 3rd edition. McGraw-Hill, New York. find
  19. Shanto Iyengar. 1991. Is Anyone Responsible? How Television Frames Political Issues. University of Chicago Press, Chicago. find
  20. Sullam Jeoung, Yubin Ge, and Jana Diesner. 2023. StereoMap: Quantifying the awareness of human-like stereotypes in large language models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 12236–12256, Singapore. Association for Computational Linguistics… find
  21. Jessica A. Jordan, Joanna R. Lawler, and Jennifer K. Bosson. 2021. Ambivalent classism: The importance of assessing hostile and benevolent ideologies about poor people. Basic and Applied Social Psychology, 43(1):46–67. find
  22. John T Jost and Mahzarin R Banaji. 1994. The role of stereotyping in system-justification and the production of false consciousness. British journal of social psychology, 33(1):1–27. find
  23. Michael B Katz. 2013. The undeserving poor: America’s enduring confrontation with poverty: Fully updated and revised. Oxford University Press. find
  24. Svetlana Kiritchenko, Georgina Curto Rex, Isar Nejadgholi, and Kathleen C. Fraser. 2023. Aporophobia: An overlooked type of toxic language targeting the poor. In The 7th Workshop on Online Abuse and Harms (WOAH), pages 113–125, Toronto, Canada. Association for Computational Linguistics. find
  25. Katerina Korre, Arianna Muti, Federico Ruggeri, and Alberto Barrón-Cedeño. 2024. Untangling hate speech definitions: A semantic componential analysis across cultures and domains. find
  26. Michèle Lamont. 2000. The Dignity of Working Men: Morality and the Boundaries of Race, Class, and Immigration. Harvard University Press. find
  27. Yang Liu. 2024. Quantifying stereotypes in language. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1223–1240, St. Julian’s, Malta. Association for Computational Linguistics. find
  28. Gandalf Nicolas, Xuechunzi Bai, and Susan T. Fiske. 2021. The stereotype content dictionaries: A semi-automated tool to quantify the warmth and competence of words, groups, and texts. Behavior Research Methods, 53(6):2609–2627. find
  29. Gandalf Nicolas, Xuechunzi Bai, and Susan T. Fiske. 2022. A spontaneous stereotype content model: Taxonomy, properties, and prediction. Personality and Social Psychology Bulletin, 48(12):1678–1695. find
  30. Carla Perez Almendros, Luis Espinosa Anke, and Steven Schockaert. 2020. Don‘t patronize me! an annotated dataset with patronizing and condescending language towards vulnerable communities. In Proceedings of the 28th International Conference on Computational Linguistics, pages 5891–5902, Barcelona, S… find
  31. Jaspreet Ranjit, Brihi Joshi, Rebecca Dorn, Laura Petry, Olga Koumoundouros, Jayne Bottarini, Peichen Liu, Eric Rice, and Swabha Swayamdipta. 2024. OATH-frames: Characterizing online attitudes towards homelessness with LLM assistants. In Proceedings of the 2024 Conference on Empirical Methods in Nat… find
  32. Max Rose and Frank R Baumgartner. 2013. Framing the poor: Media coverage and us poverty policy, 1960–2008. Policy Studies Journal, 41(1):22– 53. find
  33. Giuseppe Russo, Debora Nozza, Paul Röttger, and Dirk Hovy. 2026. The pluralistic moral gap: Understanding judgment and value differences between humans and large language models. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1:… find
  34. Mario Sainz. 2024. Identifying hostile versus paternalistic classism profiles: a person-based approach to the study of ambivalent classism. Current Psychology, 43(33):27316–27326. find
  35. Smriti Singh, Shuvam Keshari, Vinija Jain, and Aman Chadha. 2024. Born with a silver spoon? investigating socioeconomic bias in large language models. find
  36. M Carmen Terol Cantero, Maite Martin-Aragón Gelabert, Carolina Vázquez Rodríguez, Ana Lledó Boyer, and Jose Enrique García Soler. 2023. Perceptions of poverty in spain: differences in the attitudinal profiles between women and men. Frontiers in Psychology, 14:1229685. find
  37. Alice Tihelková. 2015. Framing the’scroungers’: the re-emergence of the stereotype of the undeserving poor and its reflection in the british press. Brno studies in English, 41(2):121–139. find
  38. Bernard Weiner, Danny Osborne, and Udo Rudolph. 2011. An attributional analysis of reactions to poverty: The political ideology of the giver and the perceived morality of the receiver. Personality and Social Psychology Review, 15(2):199–213. find
  39. Zehui Yu, Indira Sen, Dennis Assenmacher, Mattia Samory, Leon Fröhling, Christina Dahn, Debora Nozza, and Claudia Wagner. 2025. The unseen targets of hate: A systematic review of hateful communication datasets. Social Science Computer Review, 43(5):1114–1144. © The Author(s) 2024. find
  40. Caleb Ziems, William Held, Omar Shaikh, Jiaao Chen, Zhehao Zhang, and Diyi Yang. 2024. Can large language models transform computational social science? Computational Linguistics, 50(1):237–291. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/muti-2026-acid-perception-online-classism/