---
title: "Tibyan Corpus — Alrehili 2024"
description: "Arabic grammatical error correction (GEC) has very little training data."
authors: ["Ahlam Alrehili", "Areej Alhothali"]
year: 2024
venue: "arXiv:2411.04588"
source: "https://arxiv.org/abs/2411.04588"
doi: "10.48550/arxiv.2411.04588"
arxiv: "2411.04588"
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/alrehili-2024-tibyan-corpus-balanced-comprehensive-error/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/alrehili-2024-tibyan-corpus-balanced-comprehensive-error/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2024](https://md.research.elamur.ai/free/en/papers/2024/page/5/) / Tibyan Corpus: Balanced and Comprehensive Error Coverage Corpus Using ChatGPT for Arabic Grammatical Error Correction

# Tibyan Corpus: Balanced and Comprehensive Error Coverage Corpus Using ChatGPT for Arabic Grammatical Error Correction

**Authors:** [Ahlam Alrehili](https://md.research.elamur.ai/free/en/authors/ahlam-alrehili/), [Areej Alhothali](https://md.research.elamur.ai/free/en/authors/areej-alhothali/) · **Year:** 2024 · **Venue:** arXiv:2411.04588 · **Licence:** [CC BY 4.0 (arXiv)](http://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://arxiv.org/abs/2411.04588) · [PDF](https://arxiv.org/pdf/2411.04588) · [DOI 10.48550/arxiv.2411.04588](https://doi.org/10.48550/arxiv.2411.04588) · [arXiv 2411.04588](https://arxiv.org/abs/2411.04588) · [OpenAlex](https://openalex.org/W4409032850)

**Citations:** 6 citations (openalex-snapshot · 2026-10-14) · [3](#cited-by) in our library, of the 140884 papers whose bibliographies we read

## Summary
Arabic grammatical error correction (GEC) has very little training data. Most work depends on QALB-14 and QALB-15, which together hold roughly 20.5K parallel examples. Those corpora also cover error types unevenly and correct punctuation inconsistently. The authors build Tibyan, a new parallel GEC corpus, by using ChatGPT as a data augmenter.

They begin with short "guide" pairs of one to seven words, each an erroneous phrase plus its correction. These come from three Arabic books on common linguistic mistakes and from the A7'ta corpus. ChatGPT expands each guide pair into a full sentence pair, one version with the error and one without. Linguistic experts then review the output and refine it over several rounds of feedback.

Error types are profiled with the ARETA annotation tool. Cases ARETA could not label were sorted by hand. The final resource has about 600K tokens, and roughly 49% of them carry errors across seven broad categories. The authors argue that Tibyan widens error-type coverage and balances error types better than existing Arabic data, which makes it a basis for future seq2seq GEC models.

## Key points
- Guide pairs come from three Arabic books on common errors plus A7'ta (466 short pairs from the book Linguistic Error Detector, Saudi Press).
- The seven error categories span punctuation errors, orthographic, morphological, syntactic and semantic errors, and words that are wrongly split or merged. ARETA breaks these down into 26 fine-grained tags (OA, OH, OT, PC, MG, SP, etc.).
- Tibyan is about 600K tokens, and around 49% of tokens are erroneous.
- In the combined Tibyan counts, the most frequent tags are:
  - Ha/Ta/Ta-Marbuta confusion (OT), about 12%
  - error combinations, about 11%
  - Hamza errors (OH), about 9%
  
  Tags such as OG, OS and ON remain rare.
- ARETA mislabels tokens that have several errors, split words (it reads them as XT + OM) and multi-error merges. Its X and UNK outputs were therefore reclassified manually.

## Interesting ideas
- Short, attested error→correction snippets serve as seeds that an LLM expands into full natural sentences, so each synthetic example stays tied to a real, documented phenomenon rather than invented noise. — For Mova (learn): LLM teacher tiers could expand a list of known hard constructions (PP attachment, coordination scope, garden paths) into full sentences, which witness parsers then annotate as targeted silver data.
- Each generated item is a pair: one faulty and one correct version differing in a controlled way, which yields built-in minimal pairs. — For Mova (learn): minimal pairs make clean negative controls, so an evaluation could check that parse or frame outputs change exactly where the controlled difference lies and nowhere else.
- An automatic typer (ARETA) labels the bulk of the data, and humans touch only the residue it marks as X/UNK, which keeps expert effort small and focused. — For Mova (learn): route only the cases that the repair rules or witness parsers cannot categorize to the higher LLM teacher tier or to manual review.
- The paper finds that the error classifier breaks when token boundaries differ between the faulty and corrected text (split and merge errors). Boundary alignment must therefore be solved before any error typing. — For Mova (tokens): the tokenizer should explicitly align merged and split tokens against a normalized form, so tagging and later stages do not misread one boundary slip as two separate errors.
- The authors separate errors typical of native writers (spelling conventions, often typing slips) from those typical of learners (vowel length, gender) and suggest that tools should target each group differently. — For Mova (tokens): typo-like character errors could be handled by a normalization layer in the en crate, while agreement errors are left for the tagger and parser to flag.

*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/)

## Cited by

Cited by

3 papers of our library cite it (of the 140884 whose bibliographies we read); each has a page here.

- [ArbESC+: Arabic Enhanced Edit Selection System Combination for Grammatical Error Correction Resolving conflict and improving system combination in Arabic GEC](https://md.research.elamur.ai/free/en/papers/alrehili-2025-arbesc-arabic-enhanced-edit-selection/)
- [Nahw: A Comprehensive Benchmark of Arabic Grammar Understanding, Error Detection, Correction, and Explanation](https://md.research.elamur.ai/free/en/papers/eacl-2026-mubarak-nahw-comprehensive-benchmark-arabic-grammar/)
- [Advancing Complex Medical Communication in Arabic with Sporo AraSum: Surpassing Existing Large Language Models](https://md.research.elamur.ai/free/en/papers/lee-2024-advancing-complex-medical-communication-arabic/)

## Cite

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

```bibtex
@misc{alrehili2024tibyan,
  title = {{Tibyan Corpus: Balanced and Comprehensive Error Coverage Corpus Using ChatGPT for Arabic Grammatical Error Correction}},
  author = {Alrehili, Ahlam and Alhothali, Areej},
  year = {2024},
  doi = {10.48550/arxiv.2411.04588},
  eprint = {2411.04588},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2411.04588}
}
```

- [← previous The Translation of Circumlocution in Arabic Short Stories into English](https://md.research.elamur.ai/free/en/papers/shehab-2024-translation-circumlocution-arabic-short-stories/)
- [next → TinyLlama: An Open-Source Small Language Model](https://md.research.elamur.ai/free/en/papers/zhang-2024-tinyllama-open-source-small-language/)
- [↑ back to papers of 2024](https://md.research.elamur.ai/free/en/papers/2024/page/5/)

## References

29 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Muhammad Hasan Bakalla. Arabic culture: through its language and literature. Taylor & Francis, 2023. [find](https://scholar.google.com/scholar?q=Muhammad+Hasan+Bakalla%2E+Arabic+culture%3A+through+its+language+and+literature%2E+Taylor+%26+Francis%2C+2023%2E)
2. Anwar G Chejne. The Arabic language: Its role in history. U of Minnesota Press, 1968. [find](https://scholar.google.com/scholar?q=Anwar+G+Chejne%2E+The+Arabic+language%3A+Its+role+in+history%2E+U+of+Minnesota+Press%2C+1968%2E)
3. Charles A Ferguson. Diglossia. word, 15(2):325–340, 1959. [find](https://scholar.google.com/scholar?q=Charles+A+Ferguson%2E+Diglossia%2E+word%2C+15%282%29%3A325%E2%80%93340%2C+1959%2E)
4. Clive Holes. Modern Arabic: Structures, functions, and varieties. Georgetown University Press, 2004. [find](https://scholar.google.com/scholar?q=Clive+Holes%2E+Modern+Arabic%3A+Structures%2C+functions%2C+and+varieties%2E+Georgetown+University+Press%2C+2004%2E)
5. Christopher Bryant, Zheng Yuan, Muhammad Reza Qorib, Hannan Cao, Hwee Tou Ng, and Ted Briscoe. Grammatical error correction: A survey of the state of the art. Computational Linguistics, 49(3):643–701, 2023. [find](https://scholar.google.com/scholar?q=Christopher+Bryant%2C+Zheng+Yuan%2C+Muhammad+Reza+Qorib%2C+Hannan+Cao%2C+Hwee+Tou+Ng%2C+and+Ted+Briscoe%2E+Grammatical+error+correction%3A+A+survey+of+the+state+of+the+art%2E+Computational+Linguistics%2C+49%283%29%3A643%E2%80%93701%2C)
6. Behrang Mohit, Alla Rozovskaya, Nizar Habash, Wajdi Zaghouani, and Ossama Obeid. The first qalb shared task on automatic text correction for arabic. In Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP), pages 39–47, 2014. [find](https://scholar.google.com/scholar?q=Behrang+Mohit%2C+Alla+Rozovskaya%2C+Nizar+Habash%2C+Wajdi+Zaghouani%2C+and+Ossama+Obeid%2E+The+first+qalb+shared+task+on+automatic+text+correction+for+arabic%2E+In+Proceedings+of+the+EMNLP+2014+Workshop+on+Arabic)
7. Alla Rozovskaya, Houda Bouamor, Nizar Habash, Wajdi Zaghouani, Ossama Obeid, and Behrang Mohit. The second qalb shared task on automatic text correction for arabic. In Proceedings of the Second workshop on Arabic natural language processing, pages 26–35, 2015. [find](https://scholar.google.com/scholar?q=Alla+Rozovskaya%2C+Houda+Bouamor%2C+Nizar+Habash%2C+Wajdi+Zaghouani%2C+Ossama+Obeid%2C+and+Behrang+Mohit%2E+The+second+qalb+shared+task+on+automatic+text+correction+for+arabic%2E+In+Proceedings+of+the+Second+worksh)
8. Serena Jeblee, Houda Bouamor, Wajdi Zaghouani, and Kemal Oflazer. Cmuq@ qalb-2014: An smt-based system for automatic arabic error correction. In Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP), pages 137–142, 2014. [find](https://scholar.google.com/scholar?q=Serena+Jeblee%2C+Houda+Bouamor%2C+Wajdi+Zaghouani%2C+and+Kemal+Oflazer%2E+Cmuq%40+qalb%2D2014%3A+An+smt%2Dbased+system+for+automatic+arabic+error+correction%2E+In+Proceedings+of+the+EMNLP+2014+Workshop+on+Arabic+Natura)
9. Ossama Obeid, Wajdi Zaghouani, Behrang Mohit, Nizar Habash, Kemal Oflazer, and Nadi Tomeh. A web-based annotation framework for large-scale text correction. In The Companion Volume of the Proceedings of IJCNLP 2013: System Demonstrations, pages 1–4, 2013. [find](https://scholar.google.com/scholar?q=Ossama+Obeid%2C+Wajdi+Zaghouani%2C+Behrang+Mohit%2C+Nizar+Habash%2C+Kemal+Oflazer%2C+and+Nadi+Tomeh%2E+A+web%2Dbased+annotation+framework+for+large%2Dscale+text+correction%2E+In+The+Companion+Volume+of+the+Proceedings+)
10. Nora Madi and Hend S Al-Khalifa. A7’ ta: Data on a monolingual arabic parallel corpus for grammar checking. Data in brief, 22:237, 2019. [find](https://scholar.google.com/scholar?q=Nora+Madi+and+Hend+S+Al%2DKhalifa%2E+A7%E2%80%99+ta%3A+Data+on+a+monolingual+arabic+parallel+corpus+for+grammar+checking%2E+Data+in+brief%2C+22%3A237%2C+2019%2E)
11. Riadh Belkebir and Nizar Habash. Automatic error type annotation for arabic. arXiv preprint arXiv:2109.08068, 2021. [↗](https://arxiv.org/abs/2109.08068) [our card](https://md.research.elamur.ai/free/en/papers/belkebir-2021-automatic-error-type-annotation-arabic/)
12. Nizar Habash and David Palfreyman. Zaebuc: An annotated arabic-english bilingual writer corpus. In Proceedings of the Thirteenth Language Resources and Evaluation Conference, pages 79–88, 2022. [find](https://scholar.google.com/scholar?q=Nizar+Habash+and+David+Palfreyman%2E+Zaebuc%3A+An+annotated+arabic%2Denglish+bilingual+writer+corpus%2E+In+Proceedings+of+the+Thirteenth+Language+Resources+and+Evaluation+Conference%2C+pages+79%E2%80%9388%2C+2022%2E)
13. Tomoya Mizumoto, Mamoru Komachi, Masaaki Nagata, and Yuji Matsumoto. Mining revision log of language learning sns for automated japanese error correction of second language learners. In Proceedings of 5th International Joint Conference on Natural Language Processing, pages 147–155, 2011. [find](https://scholar.google.com/scholar?q=Tomoya+Mizumoto%2C+Mamoru+Komachi%2C+Masaaki+Nagata%2C+and+Yuji+Matsumoto%2E+Mining+revision+log+of+language+learning+sns+for+automated+japanese+error+correction+of+second+language+learners%2E+In+Proceedings+of)
14. Abdullah Alfaifi and Eric Atwell. Arabic learner corpora (alc): a taxonomy of coding errors. In The 8th International Computing Conference in Arabic, 2012. [find](https://scholar.google.com/scholar?q=Abdullah+Alfaifi+and+Eric+Atwell%2E+Arabic+learner+corpora+%28alc%29%3A+a+taxonomy+of+coding+errors%2E+In+The+8th+International+Computing+Conference+in+Arabic%2C+2012%2E)
15. S Farwaneh and M Tamimi. Arabic learners written corpus: A resource for research and learning. retrieved 2 september, 2012, from the the university of arizona. the Center for Educational Resources in Culture, Language and Literacy web site: [http://l2arabiccorpus.cercll](http://l2arabiccorpus.cercll). arizona. edu, 2012. [↗](http://l2arabiccorpus.cercll)
16. Nizar Habash and Owen Rambow. Arabic tokenization, part-of-speech tagging and morphological disambiguation in one fell swoop. In Proceedings of the 43rd annual meeting of the association for computational linguistics (ACL’05), pages 573–580, 2005. [find](https://scholar.google.com/scholar?q=Nizar+Habash+and+Owen+Rambow%2E+Arabic+tokenization%2C+part%2Dof%2Dspeech+tagging+and+morphological+disambiguation+in+one+fell+swoop%2E+In+Proceedings+of+the+43rd+annual+meeting+of+the+association+for+computati) [our card](https://md.research.elamur.ai/free/en/papers/aclbulk-2005-habash-arabic-tokenization-part-speech-tagging/)
17. Nizar Habash, Owen Rambow, and Ryan Roth. Mada+ tokan: A toolkit for arabic tokenization, diacritization, morphological disambiguation, pos tagging, stemming and lemmatization. In Proceedings of the 2nd international conference on Arabic language resources and tools (MEDAR), Cairo, Egypt, volume 41, page 62, 2009. [find](https://scholar.google.com/scholar?q=Nizar+Habash%2C+Owen+Rambow%2C+and+Ryan+Roth%2E+Mada%2B+tokan%3A+A+toolkit+for+arabic+tokenization%2C+diacritization%2C+morphological+disambiguation%2C+pos+tagging%2C+stemming+and+lemmatization%2E+In+Proceedings+of+the+2)
18. Haoran Wu, Wenxuan Wang, Yuxuan Wan, Wenxiang Jiao, and Michael Lyu. Chatgpt or grammarly? evaluating chatgpt on grammatical error correction benchmark. arXiv preprint arXiv:2303.13648, 2023. [↗](https://arxiv.org/abs/2303.13648)
19. Tao Fang, Shu Yang, Kaixin Lan, Derek F Wong, Jinpeng Hu, Lidia S Chao, and Yue Zhang. Is chatgpt a highly fluent grammatical error correction system. A comprehensive evaluation. CoRR, abs/2304.01746, 2023. [↗](https://arxiv.org/abs/2304.01746)
20. Mengsay Loem, Masahiro Kaneko, Sho Takase, and Naoaki Okazaki. Exploring effectiveness of gpt-3 in grammatical error correction: A study on performance and controllability in prompt-based methods. arXiv preprint arXiv:2305.18156, 2023. [↗](https://arxiv.org/abs/2305.18156)
21. Yue Zhang, Leyang Cui, Enbo Zhao, Wei Bi, and Shuming Shi. Robustgec: Robust grammatical error correction against subtle context perturbation. arXiv preprint arXiv:2310.07299, 2023. [↗](https://arxiv.org/abs/2310.07299)
22. Yaxin Fan, Feng Jiang, Peifeng Li, and Haizhou Li. Grammargpt: Exploring open-source llms for native chinese grammatical error correction with supervised fine-tuning. In CCF International Conference on Natural Language Processing and Chinese Computing, pages 69–80. Springer, 2023. [find](https://scholar.google.com/scholar?q=Yaxin+Fan%2C+Feng+Jiang%2C+Peifeng+Li%2C+and+Haizhou+Li%2E+Grammargpt%3A+Exploring+open%2Dsource+llms+for+native+chinese+grammatical+error+correction+with+supervised+fine%2Dtuning%2E+In+CCF+International+Conference+o)
23. Masahiro Kaneko and Naoaki Okazaki. Controlled generation with prompt insertion for natural language explanations in grammatical error correction. arXiv preprint arXiv:2309.11439, 2023. [↗](https://arxiv.org/abs/2309.11439)
24. Sang Yun Kwon, Gagan Bhatia, El Moatez Billah Nagoud, and Muhammad Abdul-Mageed. Chatgpt for arabic grammatical error correction. arXiv preprint arXiv:2308.04492, 2023. [↗](https://arxiv.org/abs/2308.04492)
25. Ilya Sutskever, Oriol Vinyals, and Quoc V Le. Sequence to sequence learning with neural networks. Advances in neural information processing systems, 27, 2014. [find](https://scholar.google.com/scholar?q=Ilya+Sutskever%2C+Oriol+Vinyals%2C+and+Quoc+V+Le%2E+Sequence+to+sequence+learning+with+neural+networks%2E+Advances+in+neural+information+processing+systems%2C+27%2C+2014%2E)
26. Felix Stahlberg and Shankar Kumar. Seq2edits: Sequence transduction using span-level edit operations. arXiv preprint arXiv:2009.11136, 2020. [↗](https://arxiv.org/abs/2009.11136)
27. Wajdi Zaghouani, Behrang Mohit, Nizar Habash, Ossama Obeid, Nadi Tomeh, Alla Rozovskaya, Noura Farra, Sarah Alkuhlani, and Kemal Oflazer. Large scale arabic error annotation: Guidelines and framework. 2014. [find](https://scholar.google.com/scholar?q=Wajdi+Zaghouani%2C+Behrang+Mohit%2C+Nizar+Habash%2C+Ossama+Obeid%2C+Nadi+Tomeh%2C+Alla+Rozovskaya%2C+Noura+Farra%2C+Sarah+Alkuhlani%2C+and+Kemal+Oflazer%2E+Large+scale+arabic+error+annotation%3A+Guidelines+and+framework%2E) [our card](https://md.research.elamur.ai/free/en/papers/zaghouani-2014-large-scale-arabic-error-annotation/)
28. Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q Weinberger, and Yoav Artzi. Bertscore: Evaluating text generation with bert. arXiv preprint arXiv:1904.09675, 2019. [↗](https://arxiv.org/abs/1904.09675)
29. Abdullah Alfaifi and ES Atwell. An evaluation of the arabic error tagset v2. In Proceedings of the AACL 2014-The American Association for Corpus Linguistics conference. The American Association for Corpus Linguistics, 2014. [find](https://scholar.google.com/scholar?q=Abdullah+Alfaifi+and+ES+Atwell%2E+An+evaluation+of+the+arabic+error+tagset+v2%2E+In+Proceedings+of+the+AACL+2014%2DThe+American+Association+for+Corpus+Linguistics+conference%2E+The+American+Association+for+Co)