Elamur Research⌕
☰
home / papers / 2018 / Fine-Grained Arabic Dialect Identification

Fine-Grained Arabic Dialect Identification

Summary

Most earlier work on Arabic Dialect Identification (DID) sorted text into at most five broad regional dialects plus Modern Standard Arabic (MSA). This paper moves to a much harder setting: recognizing the specific city dialect, covering 25 cities from Rabat to Muscat plus MSA, including several cities within the same country. The authors organize the dialects in a three-level hierarchy (5 regions, 10 sub-regions, 25 cities) and train classifiers over a broad range of features. They use two parallel corpora of commissioned travel-domain translations: a large CORPUS-6 (five regional representatives plus MSA) and a smaller CORPUS-26 (all 25 cities plus MSA). The richer CORPUS-6 resource is used to support the 26-way task. On sentences of roughly seven words on average, the best system reaches 67.9% accuracy at identifying the city, cutting the error of the Zaidan and Callison-Burch (2014) approach by 9% in relative terms. Accuracy rises above 90% when 16 words of input are available. The work matters because it shows city-level DID is feasible and that commissioned data can transfer to real social media text.

Key points

  • Task scale: 26 classes (25 city dialects + MSA), grouped as 5 regions → 10 sub-regions → 25 cities. CORPUS-26 is the parallel corpus of Bouamor et al. (2018).
  • Results: 67.9% city-level accuracy on short sentences (about 7 words). On fixed-length inputs, accuracy is 69.4% at 6 words and over 90% at 16 words, and it peaks at roughly 51–52 words. A writer's dialect needs about 1.52 sentences on average to be identified correctly.
  • Region-level F1 from the confusion matrix of the 26-way model ranges from 91 (Maghreb, Tunisia) down to 78 (Gulf) and 75 (MSA). Libya (80) and the Gulf are the most confusable regions.
  • Twitter test: CORPUS-6 MODEL was run on one million tweets collected using dialectal function words as seeds. For each city, annotators checked the 500 most confident predictions (probability > 0.9). Accuracy was 99.6% for Cairo, 91.4% for Doha, 87.0% for Beirut, 82.6% for Rabat and 20.2% for Tunis.
  • In an inspected sample of 100 tweets labeled "Tunis", about 70% were actually Libyan. This reflects shared vocabulary in the seed list and dialect continuity near the Tunisia–Libya border.

Interesting ideas

  • I1. Hierarchical label design (region → sub-region → city): Evaluating at several granularities shows whether errors are near-misses or cross-region mistakes. The same scheme could suit Ukrainian regional varieties or any geographically graded classification task.
  • I2. Measuring how much input is enough: The authors build test items of controlled length by concatenating random same-dialect sentences, then plot accuracy against word count. This is a cheap way to set minimum-input requirements for any text classifier, such as deciding how many posts per user are needed for profiling.
  • I3. Leveraging a rich subset to help a sparse fine-grained task: A larger corpus for a few representative classes is used to support the many-class problem with little data. This pattern carries over to other low-resource label spaces.
  • I4. High-confidence harvesting with human spot checks: Only top-probability predictions on unlabeled social media are kept and verified by dialect-familiar annotators. This both evaluates domain transfer and offers a route to growing a corpus, though the Tunis/Libya case shows that seed words and neighbouring varieties need careful auditing.
  • I5. Confusion analysis as linguistic insight: The confusion matrix is read as evidence of dialect similarity, effectively redrawing the dialect map. This turns model errors into a research result rather than just a score.

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: Natural language processing

Cited by

85 papers of our library cite it (of the 140884 whose bibliographies we read); 15 of them have a page here, 70 not yet.

Cite

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

@inproceedings{salameh2018fine,
  title = {{Fine-Grained Arabic Dialect Identification}},
  author = {Salameh, Mohammad and Bouamor, Houda and Habash, Nizar},
  booktitle = {Proceedings of the 27th International Conference on Computational Linguistics},
  year = {2018},
  url = {https://aclanthology.org/C18-1113/}
}

References

42 references, 10 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Muhammad Abdul-Mageed, Hassan Alhuzali, and Mohamed Elaraby. 2018. You Tweet What You Speak: A City-Level Dataset of Arabic Dialects. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation, LREC 2018, Miyazaki, Japan. find
  2. Abdel-Rahman Abu-Melhim. 1991. Code-switching and Linguistic Accommodation in Arabic. In Perspectives on Arabic Linguistics III: Papers from the Third Annual Symposium on Arabic Linguistics, volume 80, pages 231–250. John Benjamins Publishing. find
  3. Mahmoud Al-Ayyoub, Aya Nuseir, Kholoud Alsmearat, Yaser Jararweh, and Brij Gupta. 2017. Deep Learning for Arabic NLP: A Survey. Journal of Computational Science. find
  4. Mohamed Al-Badrashiny and Mona Diab. 2016. LILI: A Simple Language Independent Approach for Language Identification. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers. find
  5. Nora Al-Twairesh, Hend Al-Khalifa, and Abdulmalik AlSalman. 2016. AraSenTi: Large-Scale Twitter-Specific Arabic Sentiment Lexicons. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, Berlin, Germany. find
  6. Reem Bassiouney. 2009. Arabic Sociolinguistics. Edinburgh University Press. find
  7. Yonatan Belinkov and James Glass. 2016. A Character-level Convolutional Neural Network for Distinguishing Similar Languages and Dialects. In Proceedings of the Third Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial3), Osaka, Japan. find
  8. Elabbas Benmamoun. 2012. Agreement and Cliticization in Arabic Varieties from Diachronic and Synchronic Perspectives. In Reem Bassiouney, editor, Al’Arabiyya: Journal of American Association of Teachers of Arabic, volume 44-45, pages 137–150. Georgetown University Press. find
  9. Fadi Biadsy and Julia Hirschberg. 2009. Using Prosody and Phonotactics in Arabic Dialect Identification. In Proceedings of Interspeech, Brighton, UK. find
  10. Ann Bies, Zhiyi Song, Mohamed Maamouri, Stephen Grimes, Haejoong Lee, Jonathan Wright, Stephanie Strassel, Nizar Habash, Ramy Eskander, and Owen Rambow. 2014. Transliteration of Arabizi into Arabic Orthography: Developing a Parallel Annotated Arabizi-Arabic Script SMS/Chat Corpus. In Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP), Doha, Qatar. find
  11. Houda Bouamor, Nizar Habash, Mohammad Salameh, Wajdi Zaghouani, Owen Rambow, Dana Abdulrahim, Ossama Obeid, Salam Khalifa, Fadhl Eryani, Alexander Erdmann, and Kemal Oflazer. 2018. The MADAR Arabic Dialect Corpus and Lexicon. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. find
  12. Soumia Bougrine, Hadda Cherroun, and Djelloul Ziadi. 2017. Hierarchical Classification for Spoken Arabic Dialect Identification using Prosody: Case of Algerian Dialects. CoRR, abs/1703.10065. ↗
  13. Andrei Butnaru and Radu Tudor Ionescu. 2018. UnibucKernel: A Kernel-based Learning Method for Complex Word Identification. In Proceedings of the Thirteenth Workshop on Innovative Use of NLP for Building Educational Applications, New Orleans, Louisiana. find
  14. Slavomír Čéplö, Ján Bátora, Adam Benkato, Jiří Milička, Christophe Pereira, and Petr Zemánek. 2016. Mutual Intelligibility of Spoken Maltese, Libyan Arabic, and Tunisian Arabic Functionally Tested: A Pilot Study. Folia Linguistica, 50(2):583–628. find
  15. Kareem Darwish, Hassan Sajjad, and Hamdy Mubarak. 2014. Verifiably Effective Arabic Dialect Identification. In Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), Doha, Qatar. find
  16. Mona Diab and Heba Elfardy. 2012. Simplified Guidelines for the Creation of Large Scale Dialectal Arabic Annotations. In Proceedings of The eighth international conference on Language Resources and Evaluation (LREC 2012), Istanbul, Turkey. find
  17. Mahmoud El-Haj, Paul Rayson, and Mariam Aboelezz. 2018. Arabic Dialect Identification in the Context of Bivalency and Code-Switching. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. find
  18. Heba Elfardy and Mona Diab. 2012. Token Level Identification of Linguistic Code Switching. In Proceedings of the 24th International Conference on Computational Linguistics (COLING), IIT Mumbai, India. find
  19. Heba Elfardy and Mona Diab. 2013. Sentence Level Dialect Identification in Arabic. In Proceedings of the Association for Computational Linguistics, Sofia, Bulgaria. find
  20. Asma Etman and Louis Beex. 2015. Language and Dialect Identification: A Survey. In Proceedings of The 2015 SAI Intelligent Systems Conference (IntelliSys), London, UK. find
  21. Nizar Habash and Owen Rambow. 2006. MAGEAD: A Morphological Analyzer and Generator for the Arabic Dialects. In Proceedings of COLING/ACL, Sydney, Australia. find our card
  22. Nizar Habash, Abdelhadi Soudi, and Timothy Buckwalter. 2007. On Arabic Transliteration. In Arabic Computational Morphology, volume 38 of Text, Speech and Language Technology, chapter 2, pages 15–22. Springer. find
  23. Nizar Habash, Mona Diab, and Owen Rambow. 2012a. Conventional Orthography for Dialectal Arabic. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC-2012), pages 711–718, Istanbul, Turkey. find our card
  24. Nizar Habash, Ramy Eskander, and Abdelati Hawwari. 2012b. A Morphological Analyzer for Egyptian Arabic. In NAACL-HLT 2012 Workshop on Computational Morphology and Phonology (SIGMORPHON2012), Montréal, Canada. find our card
  25. Nizar Habash, Fadhl Eryani, Salam Khalifa, Owen Rambow, Dana Abdulrahim, Alexander Erdmann, Reem Faraj, Wajdi Zaghouani, Houda Bouamor, Nasser Zalmout, Sara Hassan, Faisal Al shargi, Sakhar Alkhereyf, Basma Abdulkareem, Ramy Eskander, Mohammad Salameh, and Hind Saddiki. 2018. Unified guidelines and resources for arabic dialect orthography. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018), Miyazaki, Japan. find our card
  26. Nizar Y Habash. 2010. Introduction to Arabic natural language processing. Morgan & Claypool Publishers. find
  27. Niloofar Haeri. 1991. Sociolinguistic Variation in Cairene Arabic: Palatalization and the qaf in the Speech of Men and Women. find
  28. Kenneth Heafield. 2011. KenLM: Faster and Smaller Language Model Queries. In Proceedings of the Sixth Workshop on Statistical Machine Translation. find our card
  29. Salam Khalifa, Sara Hassan, and Nizar Habash. 2017. A Morphological Analyzer for Gulf Arabic Verbs. find
  30. The Third Arabic Natural Language Processing Workshop WANLP 2017. find
  31. Shervin Malmasi, Marcos Zampieri, Nikola Ljubešić, Preslav Nakov, Ahmed Ali, and Jörg Tiedemann. 2016. Discriminating between Similar Languages and Arabic Dialect Identification: A Report on the Third DSL Shared Task. In Proceedings of the Third Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial3), Osaka, Japan. find our card
  32. Christopher D. Manning, Prabhakar Raghavan, and Hinrich Schütze. 2008. Introduction to Information Retrieval. Cambridge University Press, New York, NY, USA. find
  33. Arfath Pasha, Mohamed Al-Badrashiny, Mona Diab, Ahmed El Kholy, Ramy Eskander, Nizar Habash, Manoj Pooleery, Owen Rambow, and Ryan M Roth. 2014. MADAMIRA: A Fast, Comprehensive Tool for Morphological Analysis and Disambiguation of Arabic. In Proceedings of the Language Resources and Evaluation Conference (LREC), Reykjavik, Iceland. find our card
  34. Fatiha Sadat, Farzindar Kazemi, and Atefeh Farzindar. 2014. Automatic Identification of Arabic Language Varieties and Dialects in Social Media. In Proceedings of the Second Workshop on Natural Language Processing for Social Media (SocialNLP), Dublin, Ireland. find
  35. Wael Salloum, Heba Elfardy, Linda Alamir-Salloum, Nizar Habash, and Mona Diab. 2014. Sentence Level Dialect Identification for Machine Translation System Selection. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics, Baltimore, Maryland, USA. find
  36. Abdulhadi Shoufan and Sumaya Alameri. 2015. Natural Language Processing for Dialectical Arabic: A Survey. In Proceedings of the Second Workshop on Arabic Natural Language Processing, Beijing, China. find
  37. Toshiyuki Takezawa, Genichiro Kikui, Masahide Mizushima, and Eiichiro Sumita. 2007. Multilingual Spoken Language Corpus Development for Communication Research. Computational Linguistics and Chinese Language Processing, 12(3):303–324. find
  38. Christoph Tillmann, Saab Mansour, and Yaser Al-Onaizan. 2014. Improved Sentence-Level Arabic Dialect Classification. In Proceedings of the First Workshop on Applying NLP Tools to Similar Languages, Varieties and Dialects, Dublin, Ireland. find
  39. Janet Watson. 2007. The Phonology and Morphology of Arabic. Oxford University Press. find
  40. Omar F. Zaidan and Chris Callison-Burch. 2011. Crowdsourcing Translation: Professional Quality from Non-Professionals. In Proceedings of ACL, Portland, Oregon, USA. find our card
  41. Omar F. Zaidan and Chris Callison-Burch. 2014. Arabic Dialect Identification. Computational Linguistics, 40(1):171–202. find our card
  42. Marcos Zampieri, Shervin Malmasi, Nikola Ljubešić, Preslav Nakov, Ahmed Ali, Jörg Tiedemann, Yves Scherrer, and Noëmi Aepli. 2017. Findings of the VarDial Evaluation Campaign 2017. In Proceedings of the Fourth Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial), Valencia, Spain. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/salameh-2018-fine-grained-arabic-dialect-identification/