---
title: "Automatic Gender Identification and Reinflection in Arabic (2019)"
description: "English often leaves the speaker's gender unexpressed in first-person sentences. 35 references, cited by 42 papers here."
authors: ["Nizar Habash", "Houda Bouamor", "Christine Chung"]
year: 2019
venue: "Proceedings of the First Workshop on Gender Bias in Natural Language Processing"
source: "https://aclanthology.org/W19-3822/"
doi: "10.18653/v1/w19-3822"
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/habash-2019-automatic-gender-identification-reinflection-arabic/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/habash-2019-automatic-gender-identification-reinflection-arabic/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2019](https://md.research.elamur.ai/free/en/papers/2019/page/5/) / Automatic Gender Identification and Reinflection in Arabic

# Automatic Gender Identification and Reinflection in Arabic

**Authors:** [Nizar Habash](https://md.research.elamur.ai/free/en/authors/nizar-habash/), [Houda Bouamor](https://md.research.elamur.ai/free/en/authors/houda-bouamor/) and Christine Chung · **Year:** 2019 · **Venue:** [Proceedings of the First Workshop on Gender Bias in Natural Language Processing](https://md.research.elamur.ai/free/en/venues/gebnlp/) ([Workshop on Gender Bias in Natural Language Processing 2019](https://md.research.elamur.ai/free/en/venues/gebnlp/2019/)) · **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/W19-3822/) · [PDF](https://aclanthology.org/W19-3822.pdf) · [DOI 10.18653/v1/w19-3822](https://doi.org/10.18653/v1/w19-3822) · [OpenAlex](https://openalex.org/W2972654913)

**Citations:** 33 citations (OpenAlex, 2026-10-14) · [42](https://md.research.elamur.ai/free/en/papers/habash-2019-automatic-gender-identification-reinflection-arabic/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
English often leaves the speaker's gender unexpressed in first-person sentences. Arabic usually has to mark it, so a single-output translator ends up guessing, typically along stereotyped lines. The authors do not try to debias the translation model itself. Instead they add a monolingual post-processing layer that can wrap any system. This layer decides whether a first-person-singular Arabic sentence is masculine, feminine, or ambiguous, and then produces the alternative gendered version so a user or downstream component can choose.

To train and test this layer, they built the first parallel gender corpus for Arabic from OpenSubtitles sentences, with manually reinflected counterparts. They compare rule-based and learned gender identifiers. Reinflection uses character-level NMT, either as one joint step or as identification followed by rewriting.

On clean Arabic input, the two-step pipeline works best. On Google Translate output, the joint model gives about 8% relative BLEU gain for feminine targets and 5.3% for masculine ones. The work shows that gender awareness can be added without retraining the underlying system.

## Key points
- The original annotated corpus has 10,242 gender-ambiguous (B) sentences, 362 feminine and 636 masculine. Adding reinflected copies of the F and M sentences produces a balanced corpus of 12,238 sentences.
- The reinflection models were trained on 8,566 TRAIN pairs plus 226,175 synthetic pairs. DEV has 1,224 sentences and TEST has 2,448.
- **On balanced Arabic input**, the two-step system reached about 98.2–98.3 BLEU on TEST, against 97.05 for a pass-through baseline. Full-sentence accuracy on DEV rose from 91.75% to 95.42% (M) and 94.68% (F), roughly a 40% error reduction. The single joint model scored below the baseline here.
- **On Google Translate output**, the single model performed best. The two-step system lagged, probably because its identifier never saw MT-style noise.
- Scoring was done on AYT-normalized text (Alif, Ya and Ta-Marbuta variants unified) to absorb the frequent spelling noise in subtitles.
- Running the gender identifier on Google Translate's Arabic output found masculine forms about 10 times as often as feminine ones. The source corpus had a ratio of roughly 2:1.

## Interesting ideas
- When the input lacks the information needed for a choice, the system returns every valid variant, labelled, instead of silently picking one; this keeps the guess visible and leaves the decision to whoever has more context.
- A three-way label (masculine / feminine / ambiguous) treats ambiguity as a real value, whether it comes from the grammar or from missing diacritics, rather than as an error.
- For rewriting tasks, a "do-nothing" baseline that copies the input reveals when a model does more harm than good; exact sentence accuracy alongside BLEU makes small corruptions visible.
- The pipeline failed on real MT output mainly because its first component was trained only on clean text; training each stage on the actual noisy output of the stage before it is the suggested fix.
- Running a classifier over a system's outputs and comparing the label distribution with the training data turns a vague suspicion of bias into a measurable ratio.

## For Mova

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

- **I1** (back): Generation from the meaning graph could output labelled alternatives whenever a required feature (gender, number, formality) is absent from the world model, rather than defaulting.
- **I2** (tokens): Morphological features in crate en could carry an explicit "underspecified" value, so later stages see that the evidence is genuinely absent instead of inheriting a forced guess.
- **I3** (learn): Any repair rule or rewriting model should be scored against an identity baseline as a negative control, so harmful "fixes" are rejected automatically.
- **I4** (tree): The parsers could be trained partly on Mova's own tokenizer and tagger output instead of only gold tokens, so parsing errors caused by upstream noise are learned rather than amplified.
- **I5** (learn): Mova's outputs (for example, default genders assigned to entities in tales) could be audited by distribution, flagging skews that the training data does not justify.

*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

- [Carnegie Mellon University](https://md.research.elamur.ai/free/en/universities/carnegie-mellon-university/) — University, Pittsburgh, United States · 539 works · 179,533 citations
- [New York University Abu Dhabi](https://md.research.elamur.ai/free/en/universities/new-york-university-abu-dhabi/) — University, Abu Dhabi, United Arab Emirates · 122 works · 4,602 citations
- [Carnegie Mellon University Qatar](https://md.research.elamur.ai/free/en/universities/carnegie-mellon-university-qatar/) — University, Doha, Qatar · 47 works · 2,773 citations

📚 Shelves (6): [🖥️ 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 · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,772 works · [🤝 Gender Studies](https://md.research.elamur.ai/free/en/shelves/social-sciences/gender-studies/) · 52 works

🧭 Topics (1): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works

💡 Concepts (1): [💡 Word form](https://md.research.elamur.ai/free/en/concepts/word-form/) · 16 works

## Cite

Cite as: Nizar Habash, Houda Bouamor and Christine Chung. 2019. Automatic Gender Identification and Reinflection in Arabic. Proceedings of the First Workshop on Gender Bias in Natural Language Processing.

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

```bibtex
@inproceedings{habash2019automatic,
  title = {{Automatic Gender Identification and Reinflection in Arabic}},
  author = {Nizar Habash, Houda Bouamor and Christine Chung},
  booktitle = {Proceedings of the First Workshop on Gender Bias in Natural Language Processing},
  year = {2019},
  doi = {10.18653/v1/w19-3822},
  url = {https://doi.org/10.18653/v1/w19-3822}
}
```

- [← previous Pragmatically Informative Text Generation](https://md.research.elamur.ai/free/en/papers/shen-2019-pragmatically-informative-text-generation/)
- [next → JuICe: A Large Scale Distantly Supervised Dataset for Open Domain Context-based Code Generation](https://md.research.elamur.ai/free/en/papers/agashe-2019-juice-large-scale-distantly-supervised/)
- [↑ back to papers of 2019](https://md.research.elamur.ai/free/en/papers/2019/page/5/)

## References

35 references, 18 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Sarah Alkuhlani and Nizar Habash. 2011. [A corpus for modeling morpho-syntactic agreement in Arabic: Gender, number and rationality](https://md.research.elamur.ai/free/en/papers/alkuhlani-2011-corpus-modeling-morpho-syntactic-agreement/). In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics (ACL’11), Portland, Oregon, USA.
2. Christine Basta, Marta Ruiz Costa-jussà, and Noe Casas. 2019. [Evaluating the underlying gender bias in contextualized word embeddings](https://md.research.elamur.ai/free/en/papers/basta-2019-evaluating-underlying-gender-bias-contextualized/). CoRR, abs/1904.08783. [↗](https://arxiv.org/abs/1904.08783)
3. Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. 2016. [Man is to computer programmer as woman is to homemaker? debiasing word embeddings](https://md.research.elamur.ai/free/en/papers/bolukbasi-2016-man-computer-programmer-woman-homemaker/). In Advances in Neural Information Processing Systems, pages 4349–4357. 100 years of gender and ethnic stereotypes. Proceedings of the National Academy of Sciences, 115(16):E3635–E3644. [↗](https://doi.org/10.48550/arxiv.1607.06520)
4. Annette Rios Gonzales and Don Tuggener. 2017. Coreference resolution of elided subjects and possessive pronouns in Spanish-English statistical machine translation. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers, volume 2, pages 657–662. [find](https://scholar.google.com/scholar?q=Annette+Rios+Gonzales+and+Don+Tuggener%2E+2017%2E+Coreference+resolution+of+elided+subjects+and+possessive+pronouns+in+Spanish%2DEnglish+statistical+machine+translation%2E+In+Proceedings+of+the+15th+Conferenc)
5. David Graff, Mohamed Maamouri, Basma Bouziri, Sondos Krouna, Seth Kulick, and Tim Buckwalter. 2009. Standard Arabic Morphological Analyzer (SAMA) Version 3.1. Linguistic Data Consortium LDC2009E73. [find](https://scholar.google.com/scholar?q=David+Graff%2C+Mohamed+Maamouri%2C+Basma+Bouziri%2C+Sondos+Krouna%2C+Seth+Kulick%2C+and+Tim+Buckwalter%2E+2009%2E+Standard+Arabic+Morphological+Analyzer+%28SAMA%29+Version+3%2E1%2E+Linguistic+Data+Consortium+LDC2009E73%2E)
6. Nizar Habash, Abdelhadi Soudi, and Tim Buckwalter. 2007. On Arabic Transliteration. In A. van den Bosch and A. Soudi, editors, Arabic Computational Morphology: Knowledge-based and Empirical Methods, pages 15–22. Springer, Netherlands. [↗](https://doi.org/10.1007/978-1-4020-6046-5_2)
7. Nizar Y Habash. 2010. Introduction to Arabic natural language processing, volume 3. Morgan & Claypool Publishers. [find](https://scholar.google.com/scholar?q=Nizar+Y+Habash%2E+2010%2E+Introduction+to+Arabic+natural+language+processing%2C+volume+3%2E+Morgan+%26+Claypool+Publishers%2E)
8. Nawar Halabi. 2016. Modern standard Arabic phonetics for speech synthesis. Ph.D. thesis, University of Southampton. [find](https://scholar.google.com/scholar?q=Nawar+Halabi%2E+2016%2E+Modern+standard+Arabic+phonetics+for+speech+synthesis%2E+Ph%2ED%2E+thesis%2C+University+of+Southampton%2E)
9. Google Translate Help. 2019. Get gender specific translations. [https://support.google.com/translate/answer/9179237](https://support.google.com/translate/answer/9179237). [↗](https://support.google.com/translate/answer/9179237)
10. Aylin Caliskan, Joanna J. Bryson, and Arvind Narayanan. 2017. Semantics derived automatically from language corpora contain human-like biases. Science, 356.6334. [↗](https://doi.org/10.1126/science.aal4230)
11. Yoon Kim, Yacine Jernite, David Sontag, and Alexander M Rush. 2016. [Character-aware neural language models](https://md.research.elamur.ai/free/en/papers/kim-2015-character-aware-neural-language-models/). In Thirtieth AAAI Conference on Artificial Intelligence.
12. Ryan Cotterell, Christo Kirov, John Sylak-Glassman, Géraldine Walther, Ekaterina Vylomova, Patrick Xia, Manaal Faruqui, Sandra Kübler, David Yarowsky, Jason Eisner, and Mans Hulden. 2017. [CoNLL-SIGMORPHON 2017 shared task: Universal morphological reinflection in 52 languages](https://md.research.elamur.ai/free/en/papers/cotterell-2017-sigmorphon-reinflection-52/). CoRR, abs/1706.09031. [↗](https://arxiv.org/abs/1706.09031)
13. Guillaume Klein, Yoon Kim, Yuntian Deng, Jean Senellart, and Alexander M Rush. 2017. [Opennmt: Open-source toolkit for neural machine translation](https://md.research.elamur.ai/free/en/papers/klein-2017-opennmt-open-source-toolkit-neural/). arXiv preprint arXiv:1701.02810. [↗](https://arxiv.org/abs/1701.02810)
14. Ahmed El Kholy and Nizar Habash. 2010. Orthographic and morphological processing for English-Arabic statistical machine translation. In Proceedings of Traitement Automatique des Langues Naturelles (TALN), Montréal, Canada. Montréal, Canada. [↗](https://doi.org/10.1007/s10590-011-9110-0)
15. Mostafa Elaraby, Ahmed Y Tawfik, Mahmoud Khaled, Hany Hassan, and Aly Osama. 2018. Gender aware spoken language translation applied to English-Arabic. In Natural Language and Speech Processing (ICNLSP), 2018 2nd International Conference on, pages 1–6. IEEE. [find](https://scholar.google.com/scholar?q=Mostafa+Elaraby%2C+Ahmed+Y+Tawfik%2C+Mahmoud+Khaled%2C+Hany+Hassan%2C+and+Aly+Osama%2E+2018%2E+Gender+aware+spoken+language+translation+applied+to+English%2DArabic%2E+In+Natural+Language+and+Speech+Processing+%28ICNLSP)
16. Joel Escudé Font and Marta R Costa-jussà. 2019. Equalizing gender biases in neural machine translation with word embeddings techniques. arXiv preprint arXiv:1901.03116. [↗](https://arxiv.org/abs/1901.03116)
17. Nikhil Garg, Londa Schiebinger, Dan Jurafsky, and James Zou. 2018. Word embeddings quantify P. Koehn. 2005. Europarl: A parallel corpus for statistical machine translation. In MT summit, volume 5. Citeseer. [find](https://scholar.google.com/scholar?q=Nikhil+Garg%2C+Londa+Schiebinger%2C+Dan+Jurafsky%2C+and+James+Zou%2E+2018%2E+Word+embeddings+quantify+P%2E+Koehn%2E+2005%2E+Europarl%3A+A+parallel+corpus+for+statistical+machine+translation%2E+In+MT+summit%2C+volume+5%2E+Cit)
18. Wang Ling, Isabel Trancoso, Chris Dyer, and Alan W. Black. 2015. [Character-based neural machine translation](https://md.research.elamur.ai/free/en/papers/jussa-2016-character-based-neural-machine-translation/). CoRR, abs/1511.04586. [↗](https://arxiv.org/abs/1511.04586)
19. Pierre Lison and Jörg Tiedemann. 2016. [OpenSubtitles2016: Extracting Large Parallel Corpora from Movie and TV Subtitles](https://md.research.elamur.ai/free/en/papers/lison-2016-opensubtitles2016-extracting-large-parallel-corpora/). In Proceedings of the Language Resources and Evaluation Conference (LREC), Portorož, Slovenia.
20. Kaiji Lu, Piotr Mardziel, Fangjing Wu, Preetam Amancharla, and Anupam Datta. 2018. [Gender bias in neural natural language processing](https://md.research.elamur.ai/free/en/papers/lu-2018-gender-bias-neural-natural-language/). CoRR, abs/1807.11714. [↗](https://arxiv.org/abs/1807.11714)
21. Minh-Thang Luong and Christopher D Manning. 2015. Stanford neural machine translation systems for spoken language domains. In Proceedings of the International Workshop on Spoken Language Translation. [find](https://scholar.google.com/scholar?q=Minh%2DThang+Luong+and+Christopher+D+Manning%2E+2015%2E+Stanford+neural+machine+translation+systems+for+spoken+language+domains%2E+In+Proceedings+of+the+International+Workshop+on+Spoken+Language+Translation%2E)
22. Ngoc Quang Luong and Andrei Popescu-Belis. 2016. Improving pronoun translation by modeling coreference uncertainty. In Proceedings of the First Conference on Machine Translation: Volume 1, Research Papers, volume 1, pages 12–20. [find](https://scholar.google.com/scholar?q=Ngoc+Quang+Luong+and+Andrei+Popescu%2DBelis%2E+2016%2E+Improving+pronoun+translation+by+modeling+coreference+uncertainty%2E+In+Proceedings+of+the+First+Conference+on+Machine+Translation%3A+Volume+1%2C+Research+Pa)
23. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. [Bleu: a method for automatic evaluation of machine translation](https://md.research.elamur.ai/free/en/papers/papineni-2002-bleu-method-automatic-evaluation-machine/). In Proceedings of ACL, Philadelphia, USA.
24. Arfath Pasha, Mohamed Al-Badrashiny, Mona Diab, Ahmed El Kholy, Ramy Eskander, Nizar Habash, Manoj Pooleery, Owen Rambow, and Ryan Roth. 2014. [Madamira: A fast, comprehensive tool for morphological analysis and disambiguation of Arabic](https://md.research.elamur.ai/free/en/papers/pasha-2014-madamira-fast-comprehensive-tool-morphological/). In Proceedings of the Language Resources and Evaluation Conference (LREC), pages 1094–1101, Reykjavik, Iceland.
25. F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay. 2011. Scikit-learn: Machine Learning in Python . Journal of Machine Learning Research, 12:2825–2830. [↗](https://doi.org/10.48550/arxiv.1201.0490)
26. Marcelo O. R. Prates, Pedro H. C. Avelar, and Luís C. Lamb. 2018. Assessing gender bias in machine translation - A case study with google translate. CoRR, abs/1809.02208. [↗](https://arxiv.org/abs/1809.02208)
27. Ella Rabinovich, Shachar Mirkin, Raj Nath Patel, Lucia Specia, and Shuly Wintner. 2016. [Personalized machine translation: Preserving original author traits](https://md.research.elamur.ai/free/en/papers/rabinovich-2017-personalized-machine-translation-preserving-original/). arXiv preprint arXiv:1610.05461. [↗](https://arxiv.org/abs/1610.05461)
28. Adam Sutton, Thomas Lansdall-Welfare, and Nello Cristianini. 2018. Biased embeddings from wild data: Measuring, understanding and removing. CoRR, abs/1806.06301. [↗](https://arxiv.org/abs/1806.06301)
29. Dima Taji, Jamila El Gizuli, and Nizar Habash. 2018. [An Arabic dependency treebank in the travel domain](https://md.research.elamur.ai/free/en/papers/taji-2019-arabic-dependency-treebank-travel-domain/). In Proceedings of the Workshop on Open-Source Arabic Corpora and Processing Tools (OS-ACT), Miyazaki, Japan. [↗](https://doi.org/10.48550/arxiv.1901.10188)
30. Eva Vanmassenhove, Christian Hardmeier, and Andy Way. 2018. [Getting gender right in neural machine translation](https://md.research.elamur.ai/free/en/papers/vanmassenhove-2018-getting-gender-right-neural-machine/). In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 3003–3008. Association for Computational Linguistics.
31. William Wright. 1955. A grammar of the Arabic language, volume I. Cambridge University Press. [find](https://scholar.google.com/scholar?q=William+Wright%2E+1955%2E+A+grammar+of+the+Arabic+language%2C+volume+I%2E+Cambridge+University+Press%2E)
32. Wajdi Zaghouani, Behrang Mohit, Nizar Habash, Ossama Obeid, Nadi Tomeh, Alla Rozovskaya, Noura Farra, Sarah Alkuhlani, and Kemal Oflazer. 2014. [Large Scale Arabic Error Annotation: Guidelines and Framework](https://md.research.elamur.ai/free/en/papers/zaghouani-2014-large-scale-arabic-error-annotation/). In Proceedings of the Language Resources and Evaluation Conference (LREC), Reykjavik, Iceland. [↗](https://doi.org/10.13140/rg.2.1.3273.6169)
33. Jieyu Zhao, Tianlu Wang, Mark Yatskar, Ryan Cotterell, Vicente Ordonez, and Kai-Wei Chang. 2019. [Gender bias in contextualized word embeddings](https://md.research.elamur.ai/free/en/papers/zhao-2019-gender-bias-contextualized-word-embeddings/). CoRR, abs/1904.03310. [↗](https://arxiv.org/abs/1904.03310)
34. Jieyu Zhao, Yichao Zhou, Zeyu Li, Wei Wang, and Kai-Wei Chang. 2018. [Learning gender-neutral word embeddings](https://md.research.elamur.ai/free/en/papers/zhao-2018-learning-gender-neutral-word-embeddings/). arXiv preprint arXiv:1809.01496. [↗](https://arxiv.org/abs/1809.01496)
35. Ran Zmigrod, Sebastian J. Mielke, Hanna Wallach, and Ryan Cotterell. 2019. [Counterfactual data augmentation for mitigating gender stereotypes in languages with rich morphology](https://md.research.elamur.ai/free/en/papers/zmigrod-2019-counterfactual-data-augmentation-mitigating-gender/). In Proceedings of the Annual Meeting of the Association for Computational Linguistics, Florence, Italy.

*Updated 1 November 2026*