--- title: "A Corpus and Phonetic Dictionary for Tunisian Arabic Speech" description: "Tunisian Arabic has almost no resources for automatic speech recognition. 14 references, cited by 12 papers here." authors: ["Abir Masmoudi", "Mariem Ellouze Khmekhem", "Yannick Estève", "Lamia Hadrich Belguith", "Nizar Habash"] year: 2014 venue: "Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)" source: "https://aclanthology.org/L14-1385/" 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/masmoudi-2014-corpus-phonetic-dictionary-tunisian-arabic/" --- [Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/masmoudi-2014-corpus-phonetic-dictionary-tunisian-arabic/) [home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2014](https://md.research.elamur.ai/free/en/papers/2014/page/3/) / A Corpus and Phonetic Dictionary for Tunisian Arabic Speech Recognition # A Corpus and Phonetic Dictionary for Tunisian Arabic Speech Recognition **Authors:** [Abir Masmoudi](https://md.research.elamur.ai/free/en/authors/abir-masmoudi/), Mariem Ellouze Khmekhem, [Yannick Estève](https://md.research.elamur.ai/free/en/authors/yannick-esteve/), [Lamia Hadrich Belguith](https://md.research.elamur.ai/free/en/authors/lamia-hadrich-belguith/) and [Nizar Habash](https://md.research.elamur.ai/free/en/authors/nizar-habash/) · **Year:** 2014 · **Venue:** [Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)](https://md.research.elamur.ai/free/en/venues/lrec/) ([LREC 2014](https://md.research.elamur.ai/free/en/venues/lrec/2014/)) · **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/L14-1385/) · [PDF](https://aclanthology.org/L14-1385.pdf) · [OpenAlex](https://openalex.org/W2250960632) **Citations:** 30 citations (OpenAlex, 2026-10-14) · [12](https://md.research.elamur.ai/free/en/papers/masmoudi-2014-corpus-phonetic-dictionary-tunisian-arabic/cited-by/) in our library, of the 149,439 papers whose bibliographies we read ## Summary Tunisian Arabic has almost no resources for automatic speech recognition. It also lacks a standard spelling. The authors address both gaps. First, they built TARIC (Tunisian Arabic Railway Interaction Corpus) from real conversations between clerks and customers at ticket offices of the Tunis railway station. The audio was transcribed by hand with full diacritics and normalized to a CODA-style conventional orthography. They then analyzed the corpus for dialogue acts, borrowed vocabulary, lexical variation and disfluencies. They also catalogued the main phonological variations of the dialect, such as the competing realizations of several consonants and the dropping of the glottal stop depending on position. From this study they built a rule-based grapheme-to-phoneme converter that produces a pronunciation dictionary, the component linking acoustic and language models in ASR. Native-speaker experts judged its output on two test sets. Word-level error was about 9%, and most errors came from rule ordering and conflicting rules. The work gives a dialect with very little data a usable starting point for ASR. ## Key points - TARIC contains 20 hours of audio, 4,662 dialogues, 18,657 statements and 71,684 words. - Recording used two laptops running Audacity, with separate microphones for the clerk and the client. - Sessions were spread across holidays, weekends, festival days and ordinary weekdays. - About 20% of TARIC tokens are foreign borrowings, mostly French. Borrowings brought three extra phonemes into the dialect (/V/, /G/, /P/). - The G2P system has two parts: - roughly 80 context-sensitive rules of the form Replacement \<= {Left-Cond}+{Graph}+{Right-Cond}; - a lexicon of 30+ exceptions that is checked before any rule fires. - Evaluation covered about 3K unique words from TARIC and about 3K from a Tunisian blog corpus. The blog text was first converted to the Tunisian CODA and diacritized by hand. Word error rate was 8% on TARIC and 10% on the web set. - Planned future work includes handling undiacritized input and trying data-driven G2P. ## Interesting ideas - Look up a small hand-checked exception list before applying general rules, so irregular items never reach the rules and the rules stay simple. - Write each rewrite rule as a target symbol plus explicit left and right context conditions, and treat rule ordering as a known error source (for example, long-vowel rules must precede short-vowel rules). - Test on both an in-domain set and an out-of-domain set brought to the same orthographic standard, so the gap between them reflects domain shift rather than spelling noise. - Treat disfluencies (repetitions, self-corrections such as "Tunis no Sousse", hesitations, cut-off words) as distinct lexical classes instead of noise. - Segment service dialogues into request, answer and confirmation acts, so each turn has an explicit function. ## For Mova How these ideas could serve Mova, our small open language model — its files, stages and tests: - **I1** (tokens): Irregular lemmas and morphological forms could sit in an exception table consulted ahead of the rule-based lemmatizer, keeping each rule auditable. - **I2** (learn): Self-induced repair rules could carry explicit context slots, and the system could flag pairs of rules whose conditions overlap but whose outputs disagree. - **I3** (learn): Parser and reader evaluations could pair in-domain and out-of-domain test sets after a shared normalization step, making transfer loss measurable on its own. - **I4** (frames): A self-correction pattern could mark the first value as retracted, so the frame keeps only the corrected filler (Sousse as destination) and records why. - **I5** (frames): Speech-act labels could link a question frame to the frame that answers it and to a later confirmation, helping track which booking the world model should commit to. *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 - [Columbia University](https://md.research.elamur.ai/free/en/universities/columbia-university/) — University, New York, United States · 241 works · 54,053 citations - [Laboratoire d’Informatique de l’Université du Mans](https://md.research.elamur.ai/free/en/labs/laboratoire-d-informatique-de-l-universite-du-mans/) — Research institute, Le Mans, France · 12 works · 2,995 citations - [University of Sfax](https://md.research.elamur.ai/free/en/universities/university-of-sfax/) — University, Sfax, Tunisia · 14 works · 231 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 · [🎙️ Speech & spoken language](https://md.research.elamur.ai/free/en/topics/speech/) · 480 works ## Cite Cite as: Abir Masmoudi, Mariem Ellouze Khmekhem, Yannick Estève, Lamia Hadrich Belguith and Nizar Habash. 2014. A Corpus and Phonetic Dictionary for Tunisian Arabic Speech Recognition. Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14). The paper itself, not this page. BibTeX from the library's record of the work. ```bibtex @inproceedings{masmoudi2014corpus, title = {{A Corpus and Phonetic Dictionary for Tunisian Arabic Speech Recognition}}, author = {Abir Masmoudi, Mariem Ellouze Khmekhem, Yannick Estève, Lamia Hadrich Belguith and Nizar Habash}, booktitle = {Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)}, year = {2014}, doi = {10.63317/2x7os8jjo9dx}, url = {https://doi.org/10.63317/2x7os8jjo9dx} } ``` - [← previous A Constituent-Based Approach to Argument Labeling with Joint Inference in Discourse Parsing](https://md.research.elamur.ai/free/en/papers/kong-2014-constituent-based-approach-argument-labeling/) - [next → A Model for Processing Illocutionary Structures and Argumentation in Debates](https://md.research.elamur.ai/free/en/papers/budzynska-2014-model-processing-illocutionary-structures-argumentation/) - [↑ back to papers of 2014](https://md.research.elamur.ai/free/en/papers/2014/page/3/) ## References 14 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet. 1. Algamdi, M. (2003). KACST Arabic Phonetics Database.Fifteenth International Congress of Phonetics Science, Barcelona. pages. 3109-3112. [find](https://scholar.google.com/scholar?q=Algamdi%2C+M%2E+%282003%29%2E+KACST+Arabic+Phonetics+Database%2EFifteenth+International+Congress+of+Phonetics+Science%2C+Barcelona%2E+pages%2E+3109%2D3112%2E) 2. Algamdi, M., Elshafei, M., & Almuhtasib, H. (2002).,Speech Units for Arabic Text-to-speech. Fourth Workshop on Computer and Inforamtion Sciences. pages. 199-212. [find](https://scholar.google.com/scholar?q=Algamdi%2C+M%2E%2C+Elshafei%2C+M%2E%2C+%26+Almuhtasib%2C+H%2E+%282002%29%2E%2CSpeech+Units+for+Arabic+Text%2Dto%2Dspeech%2E+Fourth+Workshop+on+Computer+and+Inforamtion+Sciences%2E+pages%2E+199%2D212%2E) 3. Biadsy, F., Habash, N., and Hirschberg, J., (2009), [Improving the Arabic Pronunciation Dictionary for Phone and Word Recognition with Linguistically-Based Pronunciation Rules](https://md.research.elamur.ai/free/en/papers/biadsy-2009-improving-arabic-pronunciation-dictionary-phone/), The 2009 Annual Conference of the North American Chapter of the ACL, pages 397–405, Boulder, Colorado. 4. Bisani, M., Ney, H., (2008).Joint-sequence models for grapheme-to-phoneme conversion. Speech Communication 50, 434–451. [find](https://scholar.google.com/scholar?q=Bisani%2C+M%2E%2C+Ney%2C+H%2E%2C+%282008%29%2EJoint%2Dsequence+models+for+grapheme%2Dto%2Dphoneme+conversion%2E+Speech+Communication+50%2C+434%E2%80%93451%2E) 5. Diehl, F., Gales, M. J. F., Tomalin, M., & Woodland, P. C. (2008).Phonetic pronunciations for Arabic speech-to-text systems. IEEE International Conference on Acoustics, Speech and Signal Processing .pages. 1573-1576. [find](https://scholar.google.com/scholar?q=Diehl%2C+F%2E%2C+Gales%2C+M%2E+J%2E+F%2E%2C+Tomalin%2C+M%2E%2C+%26+Woodland%2C+P%2E+C%2E+%282008%29%2EPhonetic+pronunciations+for+Arabic+speech%2Dto%2Dtext+systems%2E+IEEE+International+Conference+on+Acoustics%2C+Speech+and+Signal+Processing+%2Ep) 6. El-Imam. Y., (2004). Phonetization of Arabic: rules and algorithms. In Computer Speech and Language 18, pages 339–373. [find](https://scholar.google.com/scholar?q=El%2DImam%2E+Y%2E%2C+%282004%29%2E+Phonetization+of+Arabic%3A+rules+and+algorithms%2E+In+Computer+Speech+and+Language+18%2C+pages+339%E2%80%93373%2E) 7. Gales, M. J. F., Diehl, F., Raut, C. K., Tomalin, M., Woodland, P. C., & Yu, K. (2007).Development of a phonetic system for large vocabulary Arabic speech recognition. IEEE Workshop on Automatic Speech Recognition & Understanding. pages. 24-29. [find](https://scholar.google.com/scholar?q=Gales%2C+M%2E+J%2E+F%2E%2C+Diehl%2C+F%2E%2C+Raut%2C+C%2E+K%2E%2C+Tomalin%2C+M%2E%2C+Woodland%2C+P%2E+C%2E%2C+%26+Yu%2C+K%2E+%282007%29%2EDevelopment+of+a+phonetic+system+for+large+vocabulary+Arabic+speech+recognition%2E+IEEE+Workshop+on+Automatic+Speec) 8. Habash, Nizar. (2010) Introduction to Arabic Natural Language Processing, Synthesis Lectures on Human Language Technologies, Graeme Hirst, editor. Morgan & Claypool Publishers. [find](https://scholar.google.com/scholar?q=Habash%2C+Nizar%2E+%282010%29+Introduction+to+Arabic+Natural+Language+Processing%2C+Synthesis+Lectures+on+Human+Language+Technologies%2C+Graeme+Hirst%2C+editor%2E+Morgan+%26+Claypool+Publishers%2E) 9. Habash, N., Soudi, A., and Buckwalter T. (2007). On Arabic Transliteration. Book Chapter. In Arabic Computational Morphology: Knowledge-based and Empirical Methods. Editors Antal van den Bosch and Abdelhadi Soudi. [↗](https://doi.org/10.1007/978-1-4020-6046-5_2) 10. Habash, N., Diab, M., Rambow, O. (2012).[Conventional Orthography for Dialectal Arabic](https://md.research.elamur.ai/free/en/papers/habash-2012-conventional-orthography-dialectal-arabic/). In: Proceedings of the Language Resources and Evaluation Conference (LREC), Istanbul. 11. Hiyassat, H. A. R. (2007). Automatic Pronunciation Dictionary Toolkit for Arabic Speech Recognition Using SPHINX Engine. Ph.D. thesis, Arab Academy for Banking and Financial Sciences, Amman, Jordan. [find](https://scholar.google.com/scholar?q=Hiyassat%2C+H%2E+A%2E+R%2E+%282007%29%2E+Automatic+Pronunciation+Dictionary+Toolkit+for+Arabic+Speech+Recognition+Using+SPHINX+Engine%2E+Ph%2ED%2E+thesis%2C+Arab+Academy+for+Banking+and+Financial+Sciences%2C+Amman%2C+Jordan%2E) 12. Maamouri, M., Buckwalter, T., Cieri, C. (2004). Dialectal Arabic Telephone Speech Corpus: Principles, Tool Design, and Transcription Conventions. In: NEMLAR International Conference on Arabic Language Resources and Tools, Cairo, September, pages. 22-23. Paris-sud, Centre d'Orsay. [find](https://scholar.google.com/scholar?q=Maamouri%2C+M%2E%2C+Buckwalter%2C+T%2E%2C+Cieri%2C+C%2E+%282004%29%2E+Dialectal+Arabic+Telephone+Speech+Corpus%3A+Principles%2C+Tool+Design%2C+and+Transcription+Conventions%2E+In%3A+NEMLAR+International+Conference+on+Arabic+Language) 13. Masmoudi, A., Estève, Y., Ellouze Khmekhem, M., Hadrich Belguith, L., (2014), Phonetic tools for the Tunisian Dialect, The 4 th International Workshop on spoken Language Technologies for Under-resourced Languages, Russia. [find](https://scholar.google.com/scholar?q=Masmoudi%2C+A%2E%2C+Est%C3%A8ve%2C+Y%2E%2C+Ellouze+Khmekhem%2C+M%2E%2C+Hadrich+Belguith%2C+L%2E%2C+%282014%29%2C+Phonetic+tools+for+the+Tunisian+Dialect%2C+The+4+th+International+Workshop+on+spoken+Language+Technologies+for+Under%2Dresourc) 14. Zribi, I., Boujelban, R,. Masmoudi, A., Ellouze Khmekhem, M., Hadrich Belguith, L., and Habash, N., (2014), [A Conventional Orthography for Tunisian Arabic](https://md.research.elamur.ai/free/en/papers/zribi-2014-conventional-orthography-tunisian-arabic/) , In 19th edition of the Language Resources and Evaluation Conference , Reykjavik, Iceland. *Updated 1 November 2026*