Elamur Research⌕
☰
home / papers / 2025 / Leveraging AI to Bridge Classical Arabic and Modern Standard Arabic for Text Simplification

Leveraging AI to Bridge Classical Arabic and Modern Standard Arabic for Text Simplification

Summary

Classical Arabic (CA), the language of the Quran and Hadith, uses archaic vocabulary and dense syntax. Many of its words and constructions have no counterpart in Modern Standard Arabic (MSA), and LLMs trained on modern text handle it poorly. Arabic simplification corpora also exist only for contemporary MSA material. The paper introduces the Hadith Simplification Dataset, which pairs 250 CA Hadith with MSA rewrites. The texts come from the LK Hadith Corpus and were limited to Sahih Al-Bukhari and Sahih Muslim, sampled at random, capped at 100 words and deduplicated. Only the Matn was kept, without the narrator chain, and diacritics were stripped. GPT-4 produced the rewrites from a prompt with task instructions and domain guidance. An Islamic Studies graduate then checked each pair for theological fidelity. Classic readability formulas gave misleading results, because explaining old terms makes texts longer. The lexical SAMER scale instead showed a clear shift toward Level 1 words and fewer Level 4–5 words. The resource and its pipeline aim to make religious texts more accessible without distorting them, and the author suggests they could transfer to poetry, legal and historical CA.

Key points

  • The Hadith Simplification Dataset holds 250 CA→MSA pairs and is released at github.com/ShathaTm/CA_MSA.

  • The source is the LK Hadith Corpus (Altammami et al., 2020), which already separates Matn from Isnad, so only Matn was used.

  • Selection kept only Bukhari and Muslim (for authenticity), sampled randomly, kept items of ≤100 words and removed duplicates; diacritics were deleted before prompting.

  • GPT-4 did the simplification, and its prompt asked it to:

    • skip narrator chains,
    • swap archaic words for common MSA ones,
    • gloss religious terms in parentheses.

    An Islamic Studies graduate verified every output.

  • Surface readability metrics failed for this task. SAMER's five-level word scale showed more Level 1 words and fewer Level 4–5 words after simplification.

Interesting ideas

  • I1. Simplification here often adds words, because archaic terms get a short gloss in parentheses, so length-based readability formulas wrongly report the output as harder; word-level difficulty bands capture the real change. — For Mova (answer): Graded readers and school tests could rate passage difficulty by the frequency bands of the lemmas they contain rather than by sentence or word length.
  • I2. The prompt tells the model to explain specialized terms inline rather than only replace them, treating elaboration as a legitimate form of simplification. — For Mova (back): When generating text from the meaning graph, rare concepts could receive a brief parenthetical gloss drawn from WordNet, with each gloss traceable to the node that triggered it.
  • I3. An LLM produces the drafts and a domain expert checks every one, yielding a small but trustworthy parallel set where fidelity matters more than scale. — For Mova (learn): A strong LLM teacher tier could draft a compact gold set of hard cases, with a human signing off on each item before it is used for evaluation.
  • I4. Using the corpus's existing split, the authors drop the chain of transmitters and keep only the content being transmitted. — For Mova (frames): Reported-speech chains ("X said that Y said…") could be handled as a separate attribution layer, so that frames for the core proposition are built cleanly while the source is still recorded.
  • I5. Data selection combines a source-reliability filter, a length cap and deduplication to keep the sample manageable and free of redundancy. — For Mova (learn): Silver-data pools could apply the same three filters (trusted sources, bounded sentence length, near-duplicate removal) before agreement among witness parsers is computed.

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: Machine translation · Natural language processing

Topics: Semitic languages

Cite

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

@inproceedings{altammami2025leveraging,
  title = {{Leveraging AI to Bridge Classical Arabic and Modern Standard Arabic for Text Simplification}},
  author = {Altammami, Shatha},
  booktitle = {Proceedings of the New Horizons in Computational Linguistics for Religious Texts},
  year = {2025},
  url = {https://aclanthology.org/2025.clrel-1.8/}
}

References

32 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. William Coster and David Kauchak. 2011. Simple english wikipedia: a new text simplification task. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 665–669. find our card
  2. M Al-Dawsari. 2004. The assessment of readability books content (boys-girls) of the first grade of intermediate school according to readability standards. Sultan Qaboos University, Muscat. find
  3. Sumit Kumar Dam, Choong Seon Hong, Yu Qiao, and Chaoning Zhang. 2024. A complete survey on llm-based ai chatbots. arXiv preprint arXiv:2406.16937. ↗
  4. Noora Al-Shameri and Hend Al-Khalifa. 2024. Arabic paraphrased parallel synthetic dataset. Data in Brief, page 111004. find
  5. Abdel Karim Al Tamimi, Manar Jaradat, Nuha Al-Jarrah, and Sahar Ghanem. 2014. Aari: automatic arabic readability index. Int. Arab J. Inf. Technol., 11(4):370–378. find
  6. Mahmoud El-Haj and Paul Rayson. 2016. Osman—a novel arabic readability metric. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC’16), pages 250–255. find
  7. Suha S Al-Thanyyan and Aqil M Azmi. 2023. Simplification of arabic text: A hybrid approach integrating machine translation and transformer-based lexical model. Journal of King Saud University-Computer and Information Sciences, 35(8):101662. find
  8. Ashraf Hatim Elneima, AhmedElmogtaba Abdelmoniem Ali Abdelaziz, and Kareem Darwish. 2024. Osact6 dialect to msa translation shared task overview. In Proceedings of the 6th Workshop on Open-Source Arabic Corpora and Processing Tools (OSACT) with Shared Tasks on Arabic LLMs Hallucination and Dialect to MSA Machine Translation@ LREC-COLING 2024, pages 93–97. find
  9. Ahmet Yusuf Alan, Enis Karaarslan, and Ömer Aydin. 2024. A rag-based question answering system proposal for understanding islam: Mufassirqas llm. arXiv preprint arXiv:2401.15378. ↗
  10. Núria Gala, Anaïs Tack, Ludivine Javourey-Drevet, Thomas François, and Johannes C Ziegler. 2020. Alector: A parallel corpus of simplified french texts with alignments of misreadings by poor and dyslexic readers. In Proceedings of the Twelfth Language Resources and Evaluation Conference, pages 1353– 1361. find
  11. Bashar Alhafni, Reem Hazim, Juan Piñeros Liberato, Muhamed Al Khalil, and Nizar Habash. 2024. The samer arabic text simplification corpus. arXiv preprint arXiv:2404.18615. ↗ our card
  12. Shatha Altammami, Eric Atwell, and Ammar Alsalka. 2020. The arabic-english parallel corpus of authentic hadith. International Journal on Islamic Applications in Computer Science And Technology, 8(2). find
  13. Yuan Gao, Ruili Wang, and Feng Hou. 2023. How to design translation prompts for chatgpt: An empirical study. arXiv preprint arXiv:2304.02182. ↗
  14. Yejin Bang, Samuel Cahyawijaya, Nayeon Lee, Wenliang Dai, Dan Su, Bryan Wilie, Holy Lovenia, Ziwei Ji, Tiezheng Yu, Willy Chung, et al. 2023. A multi-task, multilingual, multimodal evaluation of chatgpt on reasoning, hallucination, and interactivity. arxiv. arXiv preprint arXiv:2302.04023. ↗ our card
  15. Nizar Y Habash. 2010. Introduction to Arabic natural language processing. Morgan & Claypool Publishers. find
  16. Karima Kadaoui, Samar M Magdy, Abdul Waheed, Md Tawkat Islam Khondaker, Ahmed Oumar El-Shangiti, El Moatez Billah Nagoudi, and Muhammad Abdul-Mageed. 2023. Tarjamat: Evaluation of bard and chatgpt on machine translation of ten arabic varieties. arXiv preprint arXiv:2308.03051. ↗ our card
  17. Noam Benkler, Drisana Mosaphir, Scott Friedman, Andrew Smart, and Sonja Schmer-Galunder. 2023. Assessing llms for moral value pluralism. arXiv preprint arXiv:2312.10075. ↗
  18. Tomoyuki Kajiwara and Mamoru Komachi. 2016. Building a monolingual parallel corpus for text simplification using sentence similarity based on alignment between word embeddings. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 1147–1158. find
  19. Jonathan AC Brown. 2017. Hadith: Muhammad’s legacy in the medieval and modern world. Simon and Schuster. find
  20. Dominique Brunato, Andrea Cimino, Felice Dell’Orletta, and Giulia Venturi. 2016. Paccss-it: A parallel corpus of complex-simple sentences for automatic text simplification. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 351–361. find
  21. Nouran Khallaf and Serge Sharoff. 2022. Towards arabic sentence simplification via classification and generative approaches. arXiv preprint arXiv:2204.09292. ↗ our card
  22. Samuel Cahyawijaya, Holy Lovenia, and Pascale Fung. 2024. Llms are few-shot in-context low-resource language learners. arXiv preprint arXiv:2403.16512. ↗
  23. Md Tawkat Islam Khondaker, Abdul Waheed, El Moatez Billah Nagoudi, and Muhammad Abdul-Mageed. 2023. Gptaraeval: A comprehensive evaluation of chatgpt on arabic nlp. arXiv preprint arXiv:2305.14976. ↗
  24. Canyu Chen and Kai Shu. 2023. Can llm-generated misinformation be detected? arXiv preprint arXiv:2309.13788. ↗
  25. Qingyu Lu, Baopu Qiu, Liang Ding, Liping Xie, and Dacheng Tao. 2023. Error analysis prompting enables human-like translation evaluation in large language models: A case study on chatgpt. find our card
  26. Naoual Nassiri, Safae Berrichi, Abdelhak Lakhouaja, Violetta Cavalli-Sforza, and Azzeddine Mazroui. 2022. Lexical simplification of arabic educational texts through a classification approach. In The International Conference on Artificial Intelligence and Smart Environment, pages 581–587. Springer. find
  27. OpenAI. 2023. Gpt-4 technical report. openai.com/research/gpt-4. https:// ↗ our card
  28. Keqin Peng, Liang Ding, Qihuang Zhong, Li Shen, Xuebo Liu, Min Zhang, Yuanxin Ouyang, and Dacheng Tao. 2023. Towards making the most of chatgpt for machine translation. arXiv preprint arXiv:2303.13780. ↗ our card
  29. Horacio Saggion and Graeme Hirst. 2017. Automatic text simplification, volume 32. Springer. find
  30. Horacio Saggion, Sanja Štajner, Stefan Bott, Simon Mille, Luz Rello, and Biljana Drndarevic. 2015. Making it simplext: Implementation and evaluation of a text simplification system for spanish. ACM Transactions on Accessible Computing (TACCESS), 6(4):1–36. find
  31. Advaith Siddharthan. 2014. A survey of research on text simplification. ITL-International Journal of Applied Linguistics, 165(2):259–298. ↗
  32. Zhemin Zhu, Delphine Bernhard, and Iryna Gurevych. 2010. A monolingual tree-based translation model for sentence simplification. In Proceedings of the 23rd International Conference on Computational Linguistics (Coling 2010), pages 1353–1361. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/altammami-2025-leveraging-ai-bridge-classical-arabic/