---
title: "QAMAR — paper by Faqihi et al. (2026) — summary and ideas"
description: "The authors build QAMAR, a manually verified resource that gives every Quranic word five layers: its Modern Standard Arabic… 27 references."
authors: ["Sara Faqihi", "Karim Bouzoubaa", "Rachida Tajmout", "Driss Namly"]
year: 2026
venue: "Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script"
source: "https://aclanthology.org/2026.abjadnlp-1.38/"
doi: "10.18653/v1/2026.abjadnlp-1.38"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/faqihi-2026-qamar-new-fully-verified-accurate/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/faqihi-2026-qamar-new-fully-verified-accurate/)

[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/40/) / QAMAR: A New Fully Verified and Accurate Quranic Arabic Morphological Analysis Resource.

# QAMAR: A New Fully Verified and Accurate Quranic Arabic Morphological Analysis Resource.

**Authors:** Sara Faqihi, [Karim Bouzoubaa](https://md.research.elamur.ai/free/en/authors/karim-bouzoubaa/), [Rachida Tajmout](https://md.research.elamur.ai/free/en/authors/rachida-tajmout/) and [Driss Namly](https://md.research.elamur.ai/free/en/authors/driss-namly/) · **Year:** 2026 · **Venue:** [Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script](https://md.research.elamur.ai/free/en/venues/abjadnlp/) ([Workshop on NLP for Languages Using Arabic Script 2026](https://md.research.elamur.ai/free/en/venues/abjadnlp/2026/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.abjadnlp-1.38/) · [PDF](https://aclanthology.org/2026.abjadnlp-1.38.pdf) · [DOI 10.18653/v1/2026.abjadnlp-1.38](https://doi.org/10.18653/v1/2026.abjadnlp-1.38) · [OpenAlex](https://openalex.org/W7140141707)

**Citations:** 0 citations (OpenAlex, 2026-10-14)

## Summary
Quranic text is written in Classical Arabic using the Uthmani script. Existing morphological corpora of the Qur'an share several recurring flaws: words with more than one conflicting analysis, little or no human checking, stems and lemmas that do not match each other, script-specific marks leaking into the annotations, and missing diacritics. The authors build QAMAR, a manually verified resource that gives every Quranic word five layers: its Modern Standard Arabic (MSA) equivalent, stem, lemma, root, and POS. Stem, lemma and root follow traditional Arabic grammar rather than conventions borrowed from other languages. The resource can serve as gold data for normalization, lemmatization, root extraction and POS tagging. It can also be used to benchmark morphological analyzers. The authors measure QAC, NQC, output of the SAFAR toolkit (SGC) and LLaMA-4-generated analyses (LGC) against QAMAR under strict and diacritic-insensitive matching. No compared resource matches QAMAR well across all layers, and the LLM output is the weakest. The authors present QAMAR as a reference for Arabic NLP tools and for future LLM tuning.

## Key points
- Each word in QAMAR carries an MSA form, stem, lemma, root and POS. The definitions are explicit:
  - stem: the word minus its clitics;
  - lemma: the dictionary form, with affixes also removed;
  - root: usually three consonants.
- Problems found in earlier resources:
  - QAC treats as "stem" what QAMAR calls the lemma. About 10.97% of QAC stems keep the Uthmani elongation mark, and some roots are wrong.
  - In NQC, 99.95% of stems lack diacritics.
  - QMAC gives multiple lemmas to about 13.28% of words.
- To pick an LLM, the authors started from 34 candidate models and probed them on ten Quranic words. Survivors were tested on a representative sample, and the winner annotated the whole Qur'an.
- Full-corpus results:
  - Under strict matching, SGC has the best mean agreement (77.81%), followed by NQC (65.78%), QAC (56.15%) and LGC (55.44%).
  - Under broad matching, NQC leads with a 90.60% mean.
  - NQC is best on roots (93.81%) and QAC is best on POS (94.64%).
- On a restricted sample where the Uthmani and MSA forms coincide, SGC is best under strict matching (81.28%). NQC is best under broad matching (90.49%).

## Interesting ideas
- Scoring each layer twice, once with full diacritics and once on base letters only, tells apart "wrong word" errors from "wrong fine marking" errors. A single accuracy figure would hide this difference.
- A restricted evaluation subset, limited to items where original and normalized forms are identical, isolates the analyzer's quality from noise introduced by normalization.
- Before comparing resources, the authors spell out how each one defines "stem" and "lemma," because apparent errors are often just mismatched definitions.
- Model selection is a staged funnel: many candidates, a tiny probe set, a representative sample, and only then the full run. This saves compute and drops models that cannot follow the format at all.
- Treating "exactly one analysis per token in context" as a hard quality criterion exposes resources that hedge with multi-valued labels.

## For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

- **I1** (learn): Evaluations of lemmas and features could report both an exact-match score and a relaxed score (e.g., ignoring case or minor feature differences), showing whether errors are lexical or merely in detail.
- **I2** (tokens): Tagger and lemmatizer scores could be computed separately on tokens that need no spelling normalization, so tokenizer or normalizer faults are not blamed on morphology.
- **I3** (tree): When converting between UD dialects or comparing treebanks, Mova could first build an explicit map between differing label conventions, so that convention gaps are not counted as parser errors.
- **I4** (learn): Candidate LLM teachers could pass a cheap ten-item probe before they are allowed to join the silver-data or teacher-tier pipeline.
- **I5** (learn): When merging silver data from several witness parsers, Mova could flag or reject tokens that still carry multiple conflicting analyses, instead of letting them silently pass as gold.

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

## Institutions

- [Mohammed V University](https://md.research.elamur.ai/free/en/universities/mohammed-v-university/) — University, Rabat, Morocco · 8 works · 103 citations
- [Ecole Mohammadia d'Ingénieurs](https://md.research.elamur.ai/free/en/universities/ecole-mohammadia-d-ingenieurs/) — University, Rabat, Morocco · 4 works · 32 citations

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

🧭 Topics (2): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works · [🧩 Morphology](https://md.research.elamur.ai/free/en/topics/morphology/) · 405 works

## Cite

Cite as: Sara Faqihi, Karim Bouzoubaa, Rachida Tajmout and Driss Namly. 2026. QAMAR: A New Fully Verified and Accurate Quranic Arabic Morphological Analysis Resource. Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script.

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

```bibtex
@inproceedings{faqihi2026qamar,
  title = {{QAMAR: A New Fully Verified and Accurate Quranic Arabic Morphological Analysis Resource.}},
  author = {Sara Faqihi, Karim Bouzoubaa, Rachida Tajmout and Driss Namly},
  booktitle = {Proceedings of the 2nd Workshop on NLP for Languages Using Arabic Script},
  year = {2026},
  doi = {10.18653/v1/2026.abjadnlp-1.38},
  url = {https://doi.org/10.18653/v1/2026.abjadnlp-1.38}
}
```

- [← previous Q2EI: Query-to-Entity Inference for Semantic Condensation in Domain-Specific Retrieval](https://md.research.elamur.ai/free/en/papers/sun-2026-q2ei-query-entity-inference-semantic/)
- [next → Qayyem: A Real-time Platform for Scoring Proficiency of Arabic Essays](https://md.research.elamur.ai/free/en/papers/elbahnasawi-2026-qayyem-real-time-platform-scoring/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/40/)

## References

27 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. F. T. Abbas, M. S. Ahmad, F. Alam, E. Altinisik, E. Asgari, Y. Boshmaf, S. Boughorbel, S. Chawla, S. A. Chowdhury, F. Dalvi, K. Darwish, N. Durrani, M. G. Elfeky, A. K. Elmagarmid, M. Y. Eltabakh, M. Fatehkia, A. Fragkopoulos, M. Hasanain, M. Hawasly, and 22 others. 2025. [Fanar: An arabic-centric multimodal generative AI platform](https://md.research.elamur.ai/free/en/papers/abbas-2025-fanar-arabic-centric-multimodal-generative/). arXiv. ArXiv preprint arXiv:2501.13944. [↗](https://arxiv.org/abs/2501.13944)
2. D. Akra, T. Hammouda, and M. Jarrar. 2025. Quranmorph: Morphologically annotated quranic corpus. arXiv. ArXiv preprint arXiv:2506.18148. [↗](https://arxiv.org/abs/2506.18148)
3. R. Al-Rasheed, A. Al Muaddi, H. Aljasim, R. Al-Matham, M. Alhoshan, A. Al Wazrah, and A. AlOsaimy. 2025. [Evaluating RAG pipelines for arabic lexical information retrieval: A comparative study of embedding and generation models](https://md.research.elamur.ai/free/en/papers/rasheed-2025-evaluating-rag-pipelines-arabic-lexical/). In Proceedings of the 1st Workshop on NLP for Languages Using Arabic Script, pages 155–164.
4. I. A. Al-Sughaiyer and I. A. Al-Kharashi. 2004. Arabic morphological analysis techniques: A comprehensive survey. Journal of the American Society for Information Science and Technology, 55(3):189–213. [↗](https://doi.org/10.1002/asi.10368)
5. Z. Alyafeai, M. Pieler, H. B. Teufel, J. Tow, M. Bellagente, D. Phung, N. Pinnaparaju, R. Adithyan, P. Rocha, M. Zhuravinskyi, and C. Riquelme. 2024. [Arabic stable LM: Adapting stable LM 2 1.6b to arabic](https://md.research.elamur.ai/free/en/papers/alyafeai-2024-arabic-stable-lm-adapting-stable/). arXiv. ArXiv preprint arXiv:2412.04277. 2010. Understanding the Quran: a new grand challenge for computer science and artificial intelligence. In Proceedings of the GCCR’2010 Grand Challenges in Computing Research. UKCRC. [↗](https://arxiv.org/abs/2412.04277)
6. M. H. Bashir, A. M. Azmi, H. Nawaz, W. Zaghouani, M. Diab, A. Al-Fuqaha, and J. Qadir. 2023. Arabic natural language processing for Qur’anic research: a systematic review. Artificial Intelligence Review, 56(7):6801–6854. [↗](https://doi.org/10.1007/s10462-022-10313-2)
7. M. Boudchiche, A. Mazroui, M. O. A. O. Bebah, A. Lakhouaja, and A. Boudlal. 2017. Alkhalil morpho sys 2: A robust arabic morpho-syntactic analyzer. Journal of King Saud University-Computer and Information Sciences, 29(2):141–146. [↗](https://doi.org/10.1016/j.jksuci.2016.05.002)
8. K. Bouzoubaa, Y. Jaafar, D. Namly, R. Tachicart, R. Tajmout, H. Khamar, H. Jaafar, S. L. Aouragh, and A. Yousfi. 2021. [A description and demonstration of SAFAR framework](https://md.research.elamur.ai/free/en/papers/bouzoubaa-2021-description-demonstration-safar-framework/). In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, pages 127–134.
9. C. Brierley, M. Sawalha, and E. Atwell. 2012. [Open-source boundary-annotated corpus for arabic speech and language processing](https://md.research.elamur.ai/free/en/papers/brierley-2012-open-source-boundary-annotated-corpus/). In Proceedings of the International Conference on Language Resources and Evaluation (LREC), pages 1011–1016.
10. C. Brierley, M. Sawalha, and H. El-Farahaty. 2019. Translating sacred sounds: Encoding tajwīd rules in automatically generated IPA transcriptions of Quranic arabic. In The Routledge Handbook of Arabic Translation, pages 46–64. Routledge. [find](https://scholar.google.com/scholar?q=C%2E+Brierley%2C+M%2E+Sawalha%2C+and+H%2E+El%2DFarahaty%2E+2019%2E+Translating+sacred+sounds%3A+Encoding+tajw%C4%ABd+rules+in+automatically+generated+IPA+transcriptions+of+Quranic+arabic%2E+In+The+Routledge+Handbook+of+Arabic)
11. J. Dror, D. Shaharabani, R. Talmon, and S. Wintner. 2004. Morphological analysis of the Qur’an. Literary and Linguistic Computing, 19(4):431–452. [↗](https://doi.org/10.1093/llc/19.4.431)
12. K. Dukes and N. Habash. 2010. [Morphological annotation of Quranic arabic](https://md.research.elamur.ai/free/en/papers/dukes-2010-morphological-annotation-quranic-arabic/). In Proceedings of the International Conference on Language Resources and Evaluation (LREC), E. Atwell, N. Habash, B. Louw, B. A. Shawar, pages 2530–2536. T. McEnery, W. Zaghouani, and M. El-Haj. 311
13. E. Filipovska, A. Mladenovska, M. Bajrami, J. Dobreva, V. Hillman, P. Lameski, and E. Zdravevski. 2024. Benchmarking OpenAI’s apis and other large language models for repeatable and efficient question answering across multiple documents. In 2024 19th Conference on Computer Science and Intelligence Systems (FedCSIS), pages 107–117. IEEE. [find](https://scholar.google.com/scholar?q=E%2E+Filipovska%2C+A%2E+Mladenovska%2C+M%2E+Bajrami%2C+J%2E+Dobreva%2C+V%2E+Hillman%2C+P%2E+Lameski%2C+and+E%2E+Zdravevski%2E+2024%2E+Benchmarking+OpenAI%E2%80%99s+apis+and+other+large+language+models+for+repeatable+and+efficient+question)
14. T. Huber and C. Niklaus. 2025. LLMs meet Bloom’s taxonomy: A cognitive view on large language model evaluations. In Proceedings of the 31st International Conference on Computational Linguistics, pages 5211–5246. [find](https://scholar.google.com/scholar?q=T%2E+Huber+and+C%2E+Niklaus%2E+2025%2E+LLMs+meet+Bloom%E2%80%99s+taxonomy%3A+A+cognitive+view+on+large+language+model+evaluations%2E+In+Proceedings+of+the+31st+International+Conference+on+Computational+Linguistics%2C+pages)
15. A. Hurst, A. Lerer, A. P. Goucher, A. Perelman, A. Ramesh, A. Clark, A. J. Ostrow, A. Welihinda, A. Hayes, A. Radford, and 1 others. 2024. [Gpt-4o system card](https://md.research.elamur.ai/free/en/papers/openai-2024-gpt-4o-system-card/). arXiv. ArXiv preprint arXiv:2410.21276. [↗](https://arxiv.org/abs/2410.21276)
16. Y. Jaafar and K. Bouzoubaa. 2015. Arabic natural language processing from software engineering to complex pipeline. In 2015 First International Conference on Arabic Computational Linguistics (ACLing), pages 29–36. IEEE. [find](https://scholar.google.com/scholar?q=Y%2E+Jaafar+and+K%2E+Bouzoubaa%2E+2015%2E+Arabic+natural+language+processing+from+software+engineering+to+complex+pipeline%2E+In+2015+First+International+Conference+on+Arabic+Computational+Linguistics+%28ACLing%29%2C)
17. A. Kong, S. Zhao, H. Chen, Q. Li, Y. Qin, R. Sun, X. Zhou, E. Wang, and X. Dong. 2024. Better zero-shot reasoning with role-play prompting. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 4099–4113, Mexico City, Mexico. Association for Computational Linguistics. [↗](https://doi.org/10.48550/arxiv.2308.07702)
18. D. Namly and K. Bouzoubaa. 2025. An innovative arabic light stemmer developed using a hybrid approach. International Journal of Electrical and Computer Engineering (IJECE), 15(2):2356–2363. [↗](https://doi.org/10.11591/ijece.v15i2.pp2356-2363)
19. D. Namly, K. Bouzoubaa, A. El Jihad, and S. L. Aouragh. 2019. Improving arabic lemmatization through a lemmas database and a machine-learning technique. In Recent Advances in NLP: The Case of Arabic Language, pages 81–100. Springer. [↗](https://doi.org/10.1007/978-3-030-34614-0_5)
20. M. Sawalha, C. Brierley, and E. Atwell. 2014. Automatically generated, phonemic arabic-IPA pronunciation tiers for the boundary annotated Qur’an dataset for machine learning (version 2.0). In Proceedings of LRE-Rel 2: 2nd Workshop on Language Resource and Evaluation for Religious Texts, LREC 2014 post-conference workshop, pages 42–47, Reykjavik, Iceland. The University of Leeds. [find](https://scholar.google.com/scholar?q=M%2E+Sawalha%2C+C%2E+Brierley%2C+and+E%2E+Atwell%2E+2014%2E+Automatically+generated%2C+phonemic+arabic%2DIPA+pronunciation+tiers+for+the+boundary+annotated+Qur%E2%80%99an+dataset+for+machine+learning+%28version+2%2E0%29%2E+In+Proceedi)
21. N. Sengupta, S. K. Sahu, B. Jia, S. Katipomu, H. Li, F. Koto, O. M. Afzal, S. Kamboj, O. A. Pandit, R. Pal, L. Pradhan, Z. M. Mujahid, M. Baali, X. Han, A. Aji, Z. Liu, A. Hock, A. Feldman, J. Lee, and 4 others. 2023. [Jais and jais-chat: Arabic-centric foundation and instruction-tuned open generative large language models](https://md.research.elamur.ai/free/en/papers/sengupta-2023-jais-jais-chat-arabic-centric/). arXiv. ArXiv preprint arXiv:2308.16149. [↗](https://arxiv.org/abs/2308.16149)
22. K. Shaalan, A. Allam, and A. Gomah. 2003. Towards automatic spell checking for arabic. In Proceedings of the 4th Conference on Language Engineering, Egyptian Society of Language Engineering (ELSE), pages 21– 22, Cairo, Egypt. [find](https://scholar.google.com/scholar?q=K%2E+Shaalan%2C+A%2E+Allam%2C+and+A%2E+Gomah%2E+2003%2E+Towards+automatic+spell+checking+for+arabic%2E+In+Proceedings+of+the+4th+Conference+on+Language+Engineering%2C+Egyptian+Society+of+Language+Engineering+%28ELSE%29%2C+pa)
23. Sibawayh. 1898. Kitab Sibawayh, volume 1. Al-Matba‘ah al-Kubra al-Amiriyyah. Arabic edition. [find](https://scholar.google.com/scholar?q=Sibawayh%2E+1898%2E+Kitab+Sibawayh%2C+volume+1%2E+Al%2DMatba%E2%80%98ah+al%2DKubra+al%2DAmiriyyah%2E+Arabic+edition%2E)
24. Qwen Team and 1 others. 2024. Qwen2 technical report. arXiv. ArXiv preprint arXiv:2407.10671. [↗](https://arxiv.org/abs/2407.10671)
25. K. Tnaji, K. Bouzoubaa, and S. L. Aouragh. 2021. A light arabic POS tagger using a hybrid approach. In International Conference on Digital Technologies and Applications, pages 199–208, Cham. Springer International Publishing. [find](https://scholar.google.com/scholar?q=K%2E+Tnaji%2C+K%2E+Bouzoubaa%2C+and+S%2E+L%2E+Aouragh%2E+2021%2E+A+light+arabic+POS+tagger+using+a+hybrid+approach%2E+In+International+Conference+on+Digital+Technologies+and+Applications%2C+pages+199%E2%80%93208%2C+Cham%2E+Springer+)
26. Su-Youn Yoon. 2023. Short answer grading using one-shot prompting and text similarity scoring model. arXiv preprint arXiv:2305.18638. [↗](https://arxiv.org/abs/2305.18638)
27. I. Zeroual and A. Lakhouaja. 2016. A new quranic corpus rich in morphosyntactical information. International Journal of Speech Technology, 19(2):339–346. [↗](https://doi.org/10.1007/s10772-016-9335-7)

*Updated 1 November 2026*