---
title: "Some Dialects Are More Equal Than Others — Eida 2026"
description: "NLP work on Egyptian Arabic has mostly meant the prestigious Cairene variety (CEA)."
authors: ["Mai Mohamed Eida", "Ryan Dolan", "Paul de Nijs", "Jonathan Dunn"]
year: 2026
venue: "arXiv:2609.23955"
source: "https://arxiv.org/abs/2609.23955"
doi: "10.48550/arxiv.2609.23955"
arxiv: "2609.23955"
licence: "CC BY 4.0 (arXiv)"
licence_url: "http://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/eida-2026-some-dialects-more-equal-than/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/eida-2026-some-dialects-more-equal-than/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/9/) / Some Dialects Are More Equal Than Others: Non-Prestigious Arabic Dialectal Bias in LLMs

# Some Dialects Are More Equal Than Others: Non-Prestigious Arabic Dialectal Bias in LLMs

**Authors:** Mai Mohamed Eida, Ryan Dolan, Paul de Nijs, [Jonathan Dunn](https://md.research.elamur.ai/free/en/authors/jonathan-dunn/) · **Year:** 2026 · **Venue:** arXiv:2609.23955 · **Licence:** [CC BY 4.0 (arXiv)](http://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2609.23955) · [PDF](https://arxiv.org/pdf/2609.23955) · [DOI 10.48550/arxiv.2609.23955](https://doi.org/10.48550/arxiv.2609.23955) · [arXiv 2609.23955](https://arxiv.org/abs/2609.23955) · [OpenAlex](https://openalex.org/W7214044436)

**Citations:** 0 citations (openalex-snapshot · 2026-10-14)

## Summary
NLP work on Egyptian Arabic has mostly meant the prestigious Cairene variety (CEA). The authors ask whether this leaves LLMs biased against Sa'idi Egyptian Arabic (SEA), the country's second most spoken dialect. They test two levels.

- **Model preferences:** they build minimal pairs from a spoken SEA corpus by swapping SEA-specific phonetic, morphological, orthographic and lexical features for their CEA equivalents. They add an ungrammatical baseline made of nonce forms. Five open models then score each version with length-normalized log probabilities, using Targeted Syntactic Evaluation (TSE).
- **Task accuracy:** they inject the same features into three Arabic MMLU benchmarks and compare accuracy across dialect versions with paired significance tests.

Nearly all models rank CEA or MSA above SEA. However, they still rank SEA above the nonce baseline, so they treat SEA as legitimate but less likely language. The accuracy effect is weaker and depends on the model, yet it still leans toward CEA overall. The authors conclude that broad pretraining alone does not make models fair to speakers of stigmatized sub-dialects.

## Key points
- **Models tested:** Jasmine-350M (monolingual), Jais-2-8B and Jais-1-13B (bilingual), Llama 3.1-8B (multilingual) and NileChat-4B (continuously pretrained on Egyptian Arabic).
- **TSE set:** 134 minimal pairs. Most change a single feature, and a "mixed" condition changes several at once. Scores are summed token log probabilities divided by sequence length, following Leong and Linzen (2023).
- **MADAR comparison:** MSA is the top choice for every model except NileChat-4B. Among the dialects, the order is Aswan, then Alexandria, then Cairo.
- **Downstream benchmarks:** ArabicMMLU (MSA), AraDiCE-ArabicMMLU-EGY and DialectalArabicMMLU.
  - Jais-2-8B shows the most consistent significant favoring of CEA.
  - Llama-3.1-8B shows the least, with significance in only a couple of conditions.
- **Data construction:** SEA features came from 1980s dialect surveys and were checked against a recent spoken Sa'idi corpus. Variants were generated with regular expressions and then validated by hand. Data, code and prompts are released on GitHub.

## Interesting ideas
- A nonce-form baseline matched in length and script tells you whether a model sees a variant as dispreferred or as outright broken, and it controls for "typo" effects. — For Mova (learn): Add length-matched nonce variants as a negative control when scoring the tagger or parser on dialectal English, so that "lower but valid" can be separated from "treated as noise".
- Change one dialect feature at a time, then several at once, and track how performance drops as the number of features grows. This turns a vague "dialect gap" into a dose-response curve. — For Mova (tree): Run UD dialect conversion on minimal pairs that differ by one feature, and plot LAS of the perceptron parser and gparse against the number of injected features.
- Look at the model's second choice: when the prestige form wins, check whether the dialect form still beats the ungrammatical form. This reveals whether the model has some knowledge of the variety at all. — For Mova (answer): When a reader prefers a standard paraphrase, check that it still ranks the dialectal version above a scrambled one, which shows whether the dialectal input was understood at all.
- Inject dialect features into existing benchmark items while keeping their content fixed, then run paired significance tests. This measures robustness without writing new questions. — For Mova (answer): Rewrite school-course tests and fable QA items with regional English features, and test paired accuracy to confirm that frames and answers do not change.
- Feature lists taken from old descriptive surveys can be checked against a recent corpus before use, then applied by rule and verified by hand. — For Mova (tokens): Build English dialect morphology and lexical rules from the literature, filter them by frequency in a modern corpus, and apply them with reviewed regex rewrites.

*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](https://md.research.elamur.ai/free/en/shelves/linguistics/)

## Cite

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@misc{eida2026some,
  title = {{Some Dialects Are More Equal Than Others: Non-Prestigious Arabic Dialectal Bias in LLMs}},
  author = {Eida, Mai Mohamed and Dolan, Ryan and Nijs, Paul de and Dunn, Jonathan},
  year = {2026},
  doi = {10.48550/arxiv.2609.23955},
  eprint = {2609.23955},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2609.23955}
}
```

- [← previous Some Comments on a New Manual of the Semitic Languages](https://md.research.elamur.ai/free/en/papers/corriente-2026-some-comments-new-manual-semitic/)
- [next → Some Special Functions of the Adnominal Pronouns in Neo-Semitic](https://md.research.elamur.ai/free/en/papers/kapeliuk-2026-some-special-functions-adnominal-pronouns/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/9/)

## References

35 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Zaki N Abdel-Malek. 1972. The closed-list classes of colloquial Egyptian Arabic. Mouton & Co. N.V., Publishers, The Hauge. [find](https://scholar.google.com/scholar?q=Zaki+N+Abdel%2DMalek%2E+1972%2E+The+closed%2Dlist+classes+of+colloquial+Egyptian+Arabic%2E+Mouton+%26+Co%2E+N%2EV%2E%2C+Publishers%2C+The+Hauge%2E)
2. AI@Meta. 2024. Llama 3 model card. [find](https://scholar.google.com/scholar?q=AI%40Meta%2E+2024%2E+Llama+3+model+card%2E)
3. Osamh M Aldosaree. 2016. Language attitudes toward Saudi dialects. California State University, Long Beach. [find](https://scholar.google.com/scholar?q=Osamh+M+Aldosaree%2E+2016%2E+Language+attitudes+toward+Saudi+dialects%2E+California+State+University%2C+Long+Beach%2E)
4. Malik H. Altakrori, Nizar Habash, Teresa Lynn, Younes Samih, Abed Alhakim Freihat, Kirill Chirkunov, Muhammed AbuOdeh, Radu Florian, Preslav Nakov, and Alham Fikri Aji. 2026. Dialectalarabicmmlu: Benchmarking dialectal capabilities in arabic and multilingual language models. In Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026), pages 3199–3219, Palma, Mallorca, Spain. European Language Resources Association (ELRA). [find](https://scholar.google.com/scholar?q=Malik+H%2E+Altakrori%2C+Nizar+Habash%2C+Teresa+Lynn%2C+Younes+Samih%2C+Abed+Alhakim+Freihat%2C+Kirill+Chirkunov%2C+Muhammed+AbuOdeh%2C+Radu+Florian%2C+Preslav+Nakov%2C+and+Alham+Fikri+Aji%2E+2026%2E+Dialectalarabicmmlu%3A+Benc) [our card](https://md.research.elamur.ai/free/en/papers/altakrori-2026-dialectalarabicmmlu-benchmarking-dialectal-capabilities-arabic/)
5. Mohamed Anwar, Abdelhakim Freihat, George Ibrahim, Mostafa Awad, Abdelrahman Atef Mohamed Ali Sadallah, Gurpreet Gosal, Gokul Ramakrishnan, Sarath Chandran, Biswajit Mishra, Rituraj Joshi, Ahmed Frikha, Etienne Goffinet, Abhishek Maiti, Ali El Filali, Sarah Al Barri, Samujjwal Ghosh, Rahul Pal, Parvez Mullah, Awantika Shukla, and 41 others. 2025. Jais 2: A family of Arabic-centric open large language models. Technical report, IFM. [find](https://scholar.google.com/scholar?q=Mohamed+Anwar%2C+Abdelhakim+Freihat%2C+George+Ibrahim%2C+Mostafa+Awad%2C+Abdelrahman+Atef+Mohamed+Ali+Sadallah%2C+Gurpreet+Gosal%2C+Gokul+Ramakrishnan%2C+Sarath+Chandran%2C+Biswajit+Mishra%2C+Rituraj+Joshi%2C+Ahmed+Frikh) [our card](https://md.research.elamur.ai/free/en/papers/anwar-2026-jais-2-family-arabic-centric/)
6. El-Said Badawi. 1973. Mustawayat al-Arabiyah al-muasirah fı̄ Misr : bahth fi alaqat al-lughah bi-al-hadarah. Dār al-Mārif, Cairo. [find](https://scholar.google.com/scholar?q=El%2DSaid+Badawi%2E+1973%2E+Mustawayat+al%2DArabiyah+al%2Dmuasirah+f%C4%B1%CC%84+Misr+%3A+bahth+fi+alaqat+al%2Dlughah+bi%2Dal%2Dhadarah%2E+Da%CC%84r+al%2DMa%CC%84rif%2C+Cairo%2E)
7. Reem Bassiouney. 2017. Identity and dialect performance: A study of communities and dialects. Routledge. [find](https://scholar.google.com/scholar?q=Reem+Bassiouney%2E+2017%2E+Identity+and+dialect+performance%3A+A+study+of+communities+and+dialects%2E+Routledge%2E)
8. Reem Bassiouney. 2018. Constructing the stereotype: Indexes and performance of a stigmatised local dialect in Egypt. Multilingua, 37(3):225–253. [find](https://scholar.google.com/scholar?q=Reem+Bassiouney%2E+2018%2E+Constructing+the+stereotype%3A+Indexes+and+performance+of+a+stigmatised+local+dialect+in+Egypt%2E+Multilingua%2C+37%283%29%3A225%E2%80%93253%2E)
9. Peter Behnstedt and Manfred Woidich. 1985. Die ägyptisch-Arabischen dialekte. Tübinger Atlas des Vorderen Orients/Beihefte/B, 50. [find](https://scholar.google.com/scholar?q=Peter+Behnstedt+and+Manfred+Woidich%2E+1985%2E+Die+%C3%A4gyptisch%2DArabischen+dialekte%2E+T%C3%BCbinger+Atlas+des+Vorderen+Orients%2FBeihefte%2FB%2C+50%2E)
10. Peter Behnstedt and Manfred Woidich. 2013. Arabic Dialectology. In The Oxford Handbook of Arabic Linguistics. Oxford University Press. [find](https://scholar.google.com/scholar?q=Peter+Behnstedt+and+Manfred+Woidich%2E+2013%2E+Arabic+Dialectology%2E+In+The+Oxford+Handbook+of+Arabic+Linguistics%2E+Oxford+University+Press%2E)
11. El Moatez Billah Nagoudi, Muhammad Abdul-Mageed, AbdelRahim Elmadany, Alcides Inciarte, and Md Tawkat Islam Khondaker. 2023. JASMINE: Arabic GPT models for few-shot learning. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 16721–16744, Singapore. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=El+Moatez+Billah+Nagoudi%2C+Muhammad+Abdul%2DMageed%2C+AbdelRahim+Elmadany%2C+Alcides+Inciarte%2C+and+Md+Tawkat+Islam+Khondaker%2E+2023%2E+JASMINE%3A+Arabic+GPT+models+for+few%2Dshot+learning%2E+In+Proceedings+of+the+202)
12. Houda Bouamor, Nizar Habash, Mohammad Salameh, Wajdi Zaghouani, Owen Rambow, Dana Abdulrahim, Ossama Obeid, Salam Khalifa, Fadhl Eryani, Alexander Erdmann, and 1 others. 2018. The MADAR Arabic dialect corpus and lexicon. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018). [find](https://scholar.google.com/scholar?q=Houda+Bouamor%2C+Nizar+Habash%2C+Mohammad+Salameh%2C+Wajdi+Zaghouani%2C+Owen+Rambow%2C+Dana+Abdulrahim%2C+Ossama+Obeid%2C+Salam+Khalifa%2C+Fadhl+Eryani%2C+Alexander+Erdmann%2C+and+1+others%2E+2018%2E+The+MADAR+Arabic+dialect)
13. Minh Duc Bui, Carolin Holtermann, Valentin Hofmann, Anne Lauscher, and Katharina von der Wense. 2025. Large language models discriminate against speakers of German dialects. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 8223–8251, Suzhou, China. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Minh+Duc+Bui%2C+Carolin+Holtermann%2C+Valentin+Hofmann%2C+Anne+Lauscher%2C+and+Katharina+von+der+Wense%2E+2025%2E+Large+language+models+discriminate+against+speakers+of+German+dialects%2E+In+Proceedings+of+the+2025)
14. Ibrahim Eltouhamy. 2015. Language attitudes towards dialects of arabic in egypt. Master’s thesis, American Univeristy in Cairo. [find](https://scholar.google.com/scholar?q=Ibrahim+Eltouhamy%2E+2015%2E+Language+attitudes+towards+dialects+of+arabic+in+egypt%2E+Master%E2%80%99s+thesis%2C+American+Univeristy+in+Cairo%2E)
15. Eve Fleisig, Genevieve Smith, Madeline Bossi, Ishita Rustagi, Xavier Yin, and Dan Klein. 2024. Linguistic bias in ChatGPT: Language models reinforce dialect discrimination. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 13541–13564, Miami, Florida, USA. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Eve+Fleisig%2C+Genevieve+Smith%2C+Madeline+Bossi%2C+Ishita+Rustagi%2C+Xavier+Yin%2C+and+Dan+Klein%2E+2024%2E+Linguistic+bias+in+ChatGPT%3A+Language+models+reinforce+dialect+discrimination%2E+In+Proceedings+of+the+2024+)
16. Howard Giles. 1970. Evaluative reactions to accents. Educational review, 22(3):211–227. [find](https://scholar.google.com/scholar?q=Howard+Giles%2E+1970%2E+Evaluative+reactions+to+accents%2E+Educational+review%2C+22%283%29%3A211%E2%80%93227%2E)
17. Atiqa Hachimi. 2015. “good arabic, bad arabic” mapping language ideologies in the arabic-speaking world. Zeitschrift für arabische Linguistik, (61):35– 70. [find](https://scholar.google.com/scholar?q=Atiqa+Hachimi%2E+2015%2E+%E2%80%9Cgood+arabic%2C+bad+arabic%E2%80%9D+mapping+language+ideologies+in+the+arabic%2Dspeaking+world%2E+Zeitschrift+f%C3%BCr+arabische+Linguistik%2C+%2861%29%3A35%E2%80%93+70%2E)
18. Niloofar Haeri. 1997. The Sociolinguistic Market of Cairo: Gender, Class, and Education, 1st edition. Routledge, London. [find](https://scholar.google.com/scholar?q=Niloofar+Haeri%2E+1997%2E+The+Sociolinguistic+Market+of+Cairo%3A+Gender%2C+Class%2C+and+Education%2C+1st+edition%2E+Routledge%2C+London%2E)
19. Muhammad Furquan Hassan, Faiza Khan Khattak, and Laleh Seyyed-Kalantari. 2025. Dialectic preference bias in large language models. Proceedings of the AAAI Symposium Series, 5(1):365–369. [find](https://scholar.google.com/scholar?q=Muhammad+Furquan+Hassan%2C+Faiza+Khan+Khattak%2C+and+Laleh+Seyyed%2DKalantari%2E+2025%2E+Dialectic+preference+bias+in+large+language+models%2E+Proceedings+of+the+AAAI+Symposium+Series%2C+5%281%29%3A365%E2%80%93369%2E)
20. Abdelghany A Khalafallah. 1969. A descriptive grammar of Saidi Egyptian colloquial Arabic, volume 32. Walter de Gruyter GmbH & Co KG. [find](https://scholar.google.com/scholar?q=Abdelghany+A+Khalafallah%2E+1969%2E+A+descriptive+grammar+of+Saidi+Egyptian+colloquial+Arabic%2C+volume+32%2E+Walter+de+Gruyter+GmbH+%26+Co+KG%2E)
21. Fajri Koto, Haonan Li, Sara Shatnawi, Jad Doughman, Abdelrahman Sadallah, Aisha Alraeesi, Khalid Almubarak, Zaid Alyafeai, Neha Sengupta, Shady Shehata, Nizar Habash, Preslav Nakov, and Timothy Baldwin. 2024. ArabicMMLU: Assessing massive multitask language understanding in Arabic. In Findings of the Association for Computational Linguistics: ACL 2024, pages 5622–5640, Bangkok, Thailand. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Fajri+Koto%2C+Haonan+Li%2C+Sara+Shatnawi%2C+Jad+Doughman%2C+Abdelrahman+Sadallah%2C+Aisha+Alraeesi%2C+Khalid+Almubarak%2C+Zaid+Alyafeai%2C+Neha+Sengupta%2C+Shady+Shehata%2C+Nizar+Habash%2C+Preslav+Nakov%2C+and+Timothy+Baldwi) [our card](https://md.research.elamur.ai/free/en/papers/koto-2024-arabicmmlu-assessing-massive-multitask-language/)
22. William Labov. 2006. The social stratification of English in New York city. Cambridge University Press. [find](https://scholar.google.com/scholar?q=William+Labov%2E+2006%2E+The+social+stratification+of+English+in+New+York+city%2E+Cambridge+University+Press%2E)
23. Wallace E Lambert, Richard C Hodgson, Robert C Gardner, and Samuel Fillenbaum. 1960. Evaluational reactions to spoken languages. The journal of abnormal and social psychology, 60(1):44. [find](https://scholar.google.com/scholar?q=Wallace+E+Lambert%2C+Richard+C+Hodgson%2C+Robert+C+Gardner%2C+and+Samuel+Fillenbaum%2E+1960%2E+Evaluational+reactions+to+spoken+languages%2E+The+journal+of+abnormal+and+social+psychology%2C+60%281%29%3A44%2E)
24. Cara Su-Yi Leong and Tal Linzen. 2023. Language models can learn exceptions to syntactic rules. Proceedings of SCiL. [find](https://scholar.google.com/scholar?q=Cara+Su%2DYi+Leong+and+Tal+Linzen%2E+2023%2E+Language+models+can+learn+exceptions+to+syntactic+rules%2E+Proceedings+of+SCiL%2E)
25. Fangru Lin, Shaoguang Mao, Emanuele La Malfa, Valentin Hofmann, Adrian de Wynter, Xun Wang, Si-Qing Chen, Michael J. Wooldridge, Janet B. Pierrehumbert, and Furu Wei. 2025. Assessing dialect fairness and robustness of large language models in reasoning tasks. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 6317–6342, Vienna, Austria. Association for Computational Linguistics. [↗](https://doi.org/10.18653/v1/2025.acl-long.317)
26. Tal Linzen, Emmanuel Dupoux, and Yoav Goldberg. 2016. Assessing the ability of LSTMs to learn syntax-sensitive dependencies. Transactions of the Association for Computational Linguistics, 4:521–535. [find](https://scholar.google.com/scholar?q=Tal+Linzen%2C+Emmanuel+Dupoux%2C+and+Yoav+Goldberg%2E+2016%2E+Assessing+the+ability+of+LSTMs+to+learn+syntax%2Dsensitive+dependencies%2E+Transactions+of+the+Association+for+Computational+Linguistics%2C+4%3A521%E2%80%93535%2E) [our card](https://md.research.elamur.ai/free/en/papers/linzen-2016-assessing-ability-lstms-learn-syntax/)
27. Mai Mohamed Eida and Nizar Habash. 2025. Beyond Cairo: Sa’idi Egyptian Arabic literary corpus construction and analysis. In Proceedings of the 5th International Conference on Natural Language Processing for Digital Humanities, pages 292–304, Albuquerque, USA. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Mai+Mohamed+Eida+and+Nizar+Habash%2E+2025%2E+Beyond+Cairo%3A+Sa%E2%80%99idi+Egyptian+Arabic+literary+corpus+construction+and+analysis%2E+In+Proceedings+of+the+5th+International+Conference+on+Natural+Language+Processi)
28. Mai Mohamed Eida, Mayar Nassar, and Jonathan Dunn. 2024. How well do tweets represent sub-dialects of Egyptian Arabic? In Proceedings of the Eleventh Workshop on NLP for Similar Languages, Varieties, and Dialects (VarDial 2024), pages 41–55, Mexico City, Mexico. Association for Computational Linguistics. Basel Mousi, Nadir Durrani, Fatema Ahmad, Md. [find](https://scholar.google.com/scholar?q=Mai+Mohamed+Eida%2C+Mayar+Nassar%2C+and+Jonathan+Dunn%2E+2024%2E+How+well+do+tweets+represent+sub%2Ddialects+of+Egyptian+Arabic%3F+In+Proceedings+of+the+Eleventh+Workshop+on+NLP+for+Similar+Languages%2C+Varieties%2C+)
29. Arid Hasan, Maram Hasanain, Tameem Kabbani, Fahim Dalvi, Shammur Absar Chowdhury, and Firoj Alam. 2025. AraDiCE: Benchmarks for dialectal and cultural capabilities in LLMs. In Proceedings of the 31st International Conference on Computational Linguistics, pages 4186–4218, Abu Dhabi, UAE. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Arid+Hasan%2C+Maram+Hasanain%2C+Tameem+Kabbani%2C+Fahim+Dalvi%2C+Shammur+Absar+Chowdhury%2C+and+Firoj+Alam%2E+2025%2E+AraDiCE%3A+Benchmarks+for+dialectal+and+cultural+capabilities+in+LLMs%2E+In+Proceedings+of+the+31st+)
30. Eileen Pan, Anna Seo Gyeong Choi, Maartje Ter Hoeve, Skyler Seto, and Allison Koenecke. 2025. Analyzing dialectical biases in LLMs for knowledge and reasoning benchmarks. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 20882–20893, Suzhou, China. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Eileen+Pan%2C+Anna+Seo+Gyeong+Choi%2C+Maartje+Ter+Hoeve%2C+Skyler+Seto%2C+and+Allison+Koenecke%2E+2025%2E+Analyzing+dialectical+biases+in+LLMs+for+knowledge+and+reasoning+benchmarks%2E+In+Findings+of+the+Associatio)
31. Dennis R Preston. 1993. Folk dialectology. American dialect research, pages 333–378. [find](https://scholar.google.com/scholar?q=Dennis+R+Preston%2E+1993%2E+Folk+dialectology%2E+American+dialect+research%2C+pages+333%E2%80%93378%2E)
32. Jayanta Sadhu, Maneesha Rani Saha, and Rifat Shahriyar. 2025. Social bias in large language models for Bangla: An empirical study on gender and religious bias. In Proceedings of the First Workshop on Language Models for Low-Resource Languages, pages 204–218, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Jayanta+Sadhu%2C+Maneesha+Rani+Saha%2C+and+Rifat+Shahriyar%2E+2025%2E+Social+bias+in+large+language+models+for+Bangla%3A+An+empirical+study+on+gender+and+religious+bias%2E+In+Proceedings+of+the+First+Workshop+on+)
33. Neha Sengupta, Sunil Kumar Sahu, Bokang Jia, Satheesh Katipomu, Haonan Li, Fajri Koto, Osama Mohammed Afzal, Samta Kamboj, Onkar Pandit, Rahul Pal, Lalit Pradhan, Zain Muhammad Mujahid, Massa Baali, Alham Fikri Aji, Zhengzhong Liu, Andy Hock, Andrew Feldman, Jonathan Lee, Andrew Jackson, and 3 others. 2023. Jais and jais-chat: Arabic-centric foundation and instruction-tuned open generative large language models. Preprint, arXiv:2308.16149. [↗](https://arxiv.org/abs/2308.16149) [our card](https://md.research.elamur.ai/free/en/papers/sengupta-2023-jais-jais-chat-arabic-centric/)
34. Guokan Shang, Hadi Abdine, Ahmad Chamma, Amr Mohamed, Mohamed Anwar, Abdelaziz Bounhar, Omar El Herraoui, Preslav Nakov, Michalis Vazirgiannis, and Eric P. Xing. 2025. Nile-chat: Egyptian language models for Arabic and Latin scripts. In Proceedings of The Third Arabic Natural Language Processing Conference, pages 306–322, Suzhou, China. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Guokan+Shang%2C+Hadi+Abdine%2C+Ahmad+Chamma%2C+Amr+Mohamed%2C+Mohamed+Anwar%2C+Abdelaziz+Bounhar%2C+Omar+El+Herraoui%2C+Preslav+Nakov%2C+Michalis+Vazirgiannis%2C+and+Eric+P%2E+Xing%2E+2025%2E+Nile%2Dchat%3A+Egyptian+language+mod)
35. Kees Versteegh, Mushira Eid, Alaa Elgibali, Manfred Woidich, and Andrzej Zaborski, editors. 2006–2009. Encyclopedia of Arabic Language and Linguistics. Brill, Leiden. [find](https://scholar.google.com/scholar?q=Kees+Versteegh%2C+Mushira+Eid%2C+Alaa+Elgibali%2C+Manfred+Woidich%2C+and+Andrzej+Zaborski%2C+editors%2E+2006%E2%80%932009%2E+Encyclopedia+of+Arabic+Language+and+Linguistics%2E+Brill%2C+Leiden%2E)