---
title: "Text and Speech-based Tunisian Arabic Sub-Dialects Identification — Abdallah 2020"
description: "The paper asks whether standard dialect identification (DID) can still separate Arabic varieties when they come… 16 references, cited by 2 papers here."
authors: ["Najla Ben Abdallah", "Saméh Kchaou", "Fethi Bougares"]
year: 2020
venue: "Proceedings of the Twelfth Language Resources and Evaluation Conference"
source: "https://aclanthology.org/2020.lrec-1.787/"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/abdallah-2020-text-speech-based-tunisian-arabic/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/abdallah-2020-text-speech-based-tunisian-arabic/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2020](https://md.research.elamur.ai/free/en/papers/2020/page/6/) / Text and Speech-based Tunisian Arabic Sub-Dialects Identification

# Text and Speech-based Tunisian Arabic Sub-Dialects Identification

**Authors:** Najla Ben Abdallah, Saméh Kchaou and [Fethi Bougares](https://md.research.elamur.ai/free/en/authors/fethi-bougares/) · **Year:** 2020 · **Venue:** [Proceedings of the Twelfth Language Resources and Evaluation Conference](https://md.research.elamur.ai/free/en/venues/lrec/) · **Publisher:** [European Language Resources Association (ELRA)](https://md.research.elamur.ai/free/en/publishers/european-language-resources-association-elra/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2020.lrec-1.787/) · [PDF](https://aclanthology.org/2020.lrec-1.787.pdf) · [OpenAlex](https://openalex.org/W3030503100)

**Citations:** 5 citations (OpenAlex, 2026-10-14) · [2](https://md.research.elamur.ai/free/en/papers/abdallah-2020-text-speech-based-tunisian-arabic/cited-by/) in our library, of the 140,884 papers whose bibliographies we read

## Summary
The paper asks whether standard dialect identification (DID) can still separate Arabic varieties when they come from cities in the same country. Such varieties share much of their vocabulary and sound. The authors built a new spoken corpus of four Tunisian sub-dialects: Tunis, Sfax, Sousse and Tataouine. They took recordings from online sources such as interviews and social media, cut them into utterances, and transcribed them by hand. The corpus is freely distributed. They also give a short phonetic sketch of each variety at the vowel, consonant and syllable levels.

Using the audio and the matching transcripts, they trained several speech-based and text-based classifiers and scored them on one shared test set. Speech was far stronger: the best audio system reached an F-1 of 93.75%, while text stopped at 54.16%. Confusion-matrix analysis shows that the text errors come from heavy lexical overlap between neighbouring varieties, and the audio system fixes most of them. The study supports native speakers' belief that you can hear regional origin, and it is a first step toward a dialect map of Tunisia.

## Key points
- The corpus has 1673 utterances and about 45.95 minutes of transcribed speech, all segmented and transcribed by hand.
  - Sfax: 298 utterances, 8.37 min.
  - Tunis: 411 utterances, 12.94 min.
  - Sousse: 626 utterances, 13 min.
  - Tataouine: 338 utterances, 11.64 min.
- Speakers are men and women aged 18–50. Long WAV recordings were split and annotated with the Transcriber tool.
- Data was hardest to find for Tataouine, because little of it is online, and for Sousse, because the Tunisian coast forms a dialect continuum. Tunis was easiest, since it is the variety used in broadcast media.
- On the same test set, the best speech DID system scored F-1 93.75%, against 54.16% for the text DID system.
- The paper adds a small phonetic description of the four sub-dialects and plans more cities, more data, and rule-based systems using phonotactics and prosody.

## Interesting ideas
- The authors compare audio and its own transcript on an identical test set, so the only difference between the two systems is which signal the model sees. This shows cleanly where the useful information lives.
- They read the confusion matrix to see which classes get mixed up and why: neighbouring varieties with shared vocabulary account for most text errors.
- When the lexicon overlaps heavily, word-level features alone cannot separate classes, and the deciding cues sit at another level (here, sound).
- The authors turn a folk claim (natives can tell where a speaker is from) into a testable experiment and report whether the data supports it.
- They pair the statistical classifier with a human-readable linguistic description and plan rule-based systems built on it. This keeps the classifier's decisions open to explanation.

## For Mova

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

- **I1** (learn): when comparing witness parsers or teacher tiers, score them on exactly the same items so that gaps reflect what each model uses, not differences in the data.
- **I2** (answer): break down reader failures by which question types get confused with each other, instead of reporting one accuracy number, to find where the system lacks the right cue.
- **I3** (tokens): register or dialect tagging of English text should combine spelling and morphology cues with words, because shared vocabulary alone may be too weak a signal.
- **I4** (learn): treat assumed linguistic "facts" behind repair rules as hypotheses and check them on held-out data with negative controls before relying on them.
- **I5** (tree): when converting between UD dialects or treebank conventions, write down the distinguishing patterns as explicit rules next to any learned model, so every conversion can be explained.

*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

Also on the byline: [University of Maine](https://ror.org/01adr0w49).

📚 Shelves (1): [🗣️ Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/) · 4,876 works · 20,420.2

🧭 Topics (1): [🐪 Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 1,334 works · 2,090.4

## Cite

Cite as: Najla Ben Abdallah, Saméh Kchaou and Fethi Bougares. 2020. Text and Speech-based Tunisian Arabic Sub-Dialects Identification. Proceedings of the Twelfth Language Resources and Evaluation Conference.

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

```bibtex
@inproceedings{abdallah2020text,
  title = {{Text and Speech-based Tunisian Arabic Sub-Dialects Identification}},
  author = {Najla Ben Abdallah, Saméh Kchaou and Fethi Bougares},
  booktitle = {Proceedings of the Twelfth Language Resources and Evaluation Conference},
  year = {2020},
  url = {https://aclanthology.org/2020.lrec-1.787/}
}
```

- [← previous Linear Neural Parsing and Hybrid Enhancement for Enhanced Universal Dependencies](https://md.research.elamur.ai/free/en/papers/attardi-2020-linear-neural-parsing-hybrid-enhancement/)
- [next → The Parallel Meaning Bank: A Framework for Semantically Annotating Multiple Languages](https://md.research.elamur.ai/free/en/papers/abzianidze-2020-parallel-meaning-bank-framework-semantically/)
- [↑ back to papers of 2020](https://md.research.elamur.ai/free/en/papers/2020/page/6/)

## References

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

1. Abdul-Mageed, M., Alhuzali, H., and Elaraby, M. (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, May. European Language Resources Association (ELRA). [find](https://scholar.google.com/scholar?q=Abdul%2DMageed%2C+M%2E%2C+Alhuzali%2C+H%2E%2C+and+Elaraby%2C+M%2E+%282018%29%2E+You+tweet+what+you+speak%3A+A+city%2Dlevel+dataset+of+Arabic+dialects%2E+In+Proceedings+of+the+Eleventh+International+Conference+on+Language+Resources)
2. Barkat, M. (1999). Identification of arabic dialects and experimental determination of distinctive cues. [find](https://scholar.google.com/scholar?q=Barkat%2C+M%2E+%281999%29%2E+Identification+of+arabic+dialects+and+experimental+determination+of+distinctive+cues%2E)
3. Biadsy, F., Hirschberg, J., and Habash, N. (2009). Spoken Arabic dialect identification using phonotactic modeling. In Proceedings of the EACL 2009 Workshop on Computational Approaches to Semitic Languages, pages 53–61, Athens, Greece. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Biadsy%2C+F%2E%2C+Hirschberg%2C+J%2E%2C+and+Habash%2C+N%2E+%282009%29%2E+Spoken+Arabic+dialect+identification+using+phonotactic+modeling%2E+In+Proceedings+of+the+EACL+2009+Workshop+on+Computational+Approaches+to+Semitic+Lang)
4. Bougrinea, S., Cherrouna, H., and Ziadib, D. (2017). Hierarchical classification for spoken arabic dialect identification using prosody: Case of algerian dialects. In arXiv:1703.10065v1 [cs.CL], March. [↗](https://arxiv.org/abs/1703.10065)
5. Djellab, M., Amrouche, A., Bouridane, A., and Mehallegue, N. (2016). Algerian Modern Colloquial Arabic Speech Corpus (AMCASC): regional accents recognition within complex socio-linguistic environments. In Language Resources and Evaluation 1â29. [find](https://scholar.google.com/scholar?q=Djellab%2C+M%2E%2C+Amrouche%2C+A%2E%2C+Bouridane%2C+A%2E%2C+and+Mehallegue%2C+N%2E+%282016%29%2E+Algerian+Modern+Colloquial+Arabic+Speech+Corpus+%28AMCASC%29%3A+regional+accents+recognition+within+complex+socio%2Dlinguistic+environments)
6. Matthew S. Dryer et al., editors. (2013). WALS Online. Max Planck Institute for Evolutionary Anthropology, Leipzig. [find](https://scholar.google.com/scholar?q=Matthew+S%2E+Dryer+et+al%2E%2C+editors%2E+%282013%29%2E+WALS+Online%2E+Max+Planck+Institute+for+Evolutionary+Anthropology%2C+Leipzig%2E)
7. El-Haj, M., Rayson, P., and Aboelezz, M. (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, May. European Language Resources Association (ELRA). [find](https://scholar.google.com/scholar?q=El%2DHaj%2C+M%2E%2C+Rayson%2C+P%2E%2C+and+Aboelezz%2C+M%2E+%282018%29%2E+Arabic+dialect+identification+in+the+context+of+bivalency+and+code%2Dswitching%2E+In+Proceedings+of+the+Eleventh+International+Conference+on+Language+Resou)
8. Kchaou, S., Bougares, F., and Hadrich-Belguith, L. (2019). LIUM-MIRACL participation in the MADAR Arabic dialect identification shared task. In Proceedings of the Fourth Arabic Natural Language Processing Workshop, pages 219–223, Florence, Italy, August. Association for Computational Linguistics. [find](https://scholar.google.com/scholar?q=Kchaou%2C+S%2E%2C+Bougares%2C+F%2E%2C+and+Hadrich%2DBelguith%2C+L%2E+%282019%29%2E+LIUM%2DMIRACL+participation+in+the+MADAR+Arabic+dialect+identification+shared+task%2E+In+Proceedings+of+the+Fourth+Arabic+Natural+Language+Proces)
9. Maddieson, I. (2013). Consonant inventories. In Matthew S. Dryer et al., editors, The World Atlas of Language Structures Online. Max Planck Institute for Evolutionary Anthropology, Leipzig. [find](https://scholar.google.com/scholar?q=Maddieson%2C+I%2E+%282013%29%2E+Consonant+inventories%2E+In+Matthew+S%2E+Dryer+et+al%2E%2C+editors%2C+The+World+Atlas+of+Language+Structures+Online%2E+Max+Planck+Institute+for+Evolutionary+Anthropology%2C+Leipzig%2E)
10. Malmasi, S. and Zampieri, M. (2016). Arabic dialect identification in speech transcripts. In Proceedings of the Third Workshop on NLP for Similar Languages, Varieties and Dialects, page 106â113, Osaka, Japan, December. [find](https://scholar.google.com/scholar?q=Malmasi%2C+S%2E+and+Zampieri%2C+M%2E+%282016%29%2E+Arabic+dialect+identification+in+speech+transcripts%2E+In+Proceedings+of+the+Third+Workshop+on+NLP+for+Similar+Languages%2C+Varieties+and+Dialects%2C+page+106a%CC%82113%2C+Osak)
11. Mubarak, O., rajah, E. A., and Epps, J. (2006). Novel features for effective speech and music discrimination. In IEEE Engineering on Intelligent Systems, pages 342– 346. [find](https://scholar.google.com/scholar?q=Mubarak%2C+O%2E%2C+rajah%2C+E%2E+A%2E%2C+and+Epps%2C+J%2E+%282006%29%2E+Novel+features+for+effective+speech+and+music+discrimination%2E+In+IEEE+Engineering+on+Intelligent+Systems%2C+pages+342%E2%80%93+346%2E)
12. Sadat, F., Kazemi, F., and Farzindar, A. (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), pages 22–27, Dublin, Ireland, August. Association for Computational Linguistics and Dublin City University. [find](https://scholar.google.com/scholar?q=Sadat%2C+F%2E%2C+Kazemi%2C+F%2E%2C+and+Farzindar%2C+A%2E+%282014%29%2E+Automatic+identification+of+Arabic+language+varieties+and+dialects+in+social+media%2E+In+Proceedings+of+the+Second+Workshop+on+Natural+Language+Processin)
13. Salameh, M., Bouamor, H., and Habash, N. (2018). Fine-grained Arabic dialect identification. In Proceedings of the 27th International Conference on Computational Linguistics, pages 1332–1344, Santa Fe, New Mexico, USA, August. Association for Computational Linguistics. [our card](https://md.research.elamur.ai/free/en/papers/salameh-2018-fine-grained-arabic-dialect-identification/)
14. Shon, S., Ali, A., and Glass, J. (2017). MIT-QCRI Arabic dialect identification system for the 2017 multi-genre broadcast challenge . In Automatic Speech Recognition and Understanding Workshop (ASRU), 2017. [find](https://scholar.google.com/scholar?q=Shon%2C+S%2E%2C+Ali%2C+A%2E%2C+and+Glass%2C+J%2E+%282017%29%2E+MIT%2DQCRI+Arabic+dialect+identification+system+for+the+2017+multi%2Dgenre+broadcast+challenge+%2E+In+Automatic+Speech+Recognition+and+Understanding+Workshop+%28ASRU%29%2C)
15. Zaidan, O. F. and Callison-Burch, C. (2014). Arabic dialect identification. Comput. Linguist., 40(1), March. [our card](https://md.research.elamur.ai/free/en/papers/zaidan-2014-arabic-dialect-identification/)
16. Zampieri, M., Malmasi, S., LjubeÅ¡iÄ, N., Nakov, P., Ali, A., Tiedemann, J., Scherrer, Y., and Aepli, N. (2017). Findings of the vardial evaluation campaign 2017. In In Proceedings of the Fourth Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial), Valencia, Spain. [our card](https://md.research.elamur.ai/free/en/papers/zampieri-2017-findings-vardial-evaluation-campaign/)