Elamur Research⌕
☰
home / papers / 2014 / Transliteration of Arabizi into Arabic Orthography: Developing a Parallel Annotated Arabizi-Arabic Script SMS/Chat Corpus

Transliteration of Arabizi into Arabic Orthography: Developing a Parallel Annotated Arabizi-Arabic Script SMS/Chat Corpus

Paper · 2014 · Ann Bies, Zhiyi Song, Mohamed Maamouri et al.

Summary

Arabic social media text is often typed in Arabizi, an informal Latin-letter romanization that uses digits for sounds Latin lacks (e.g., 3 for ع). It is usually dialectal and mixed with English or French. Most Arabic dialect tools expect Arabic script, so converting Arabizi is a prerequisite, but no naturally parallel Arabizi–Arabic script data existed. The LDC team, with NYU Abu Dhabi and Columbia collaborators, built such a corpus from Egyptian SMS and chat collected for the DARPA BOLT program. Each Arabizi token was first flagged as foreign, name, sound or punctuation. The text was then automatically transliterated with the 3ARRIB system and manually corrected to the CODA conventional orthography for dialectal Arabic. The result covers more than 300K words in six aligned layers. Flagging was highly consistent between annotators, and transliteration agreement was lower but solid. The authors analysed every disagreement to find where guidelines and tooling should improve. The corpus is the first resource of its kind for training and evaluating Arabizi-to-script transliteration.

Key points

  • Of 1270 Egyptian SMS/chat conversations reviewed, over 42.6K sentence units containing Arabizi (more than 300K words) were annotated. Sentence units already written fully in Arabic script were left untouched.
  • Each XML file has six layers:
    • source;
    • token flags;
    • 3ARRIB automatic transliteration;
    • manual correction;
    • re-tokenized corrected transliteration;
    • human translation.
  • About 23K words were flagged as foreign and 6610 tokens as Sounds. Emoticons and emoji were replaced by # in the transliteration, and English words were transliterated as-is without being corrected.
  • Inter-annotator agreement was measured on 26 dually annotated conversations (about 3400 words). Agreement exceeded 98% for all flags and reached 86.4% for transliteration.
  • Disagreements broke down as follows:
    • 60% were CODA spelling issues, mostly Alif/Hamza;
    • 23% were task-guideline lapses;
    • 12% reflected genuine ambiguity;
    • under 5% were typos.

Interesting ideas

  • I1. The pipeline runs an automatic transliterator first and has humans correct its output, keeping both versions as separate layers; this yields a parallel resource plus a built-in error record for the tool.
  • I2. Rather than leaving many spellings, output is normalized to one conventional orthography (CODA), so variable dialect spellings collapse to a single target form.
  • I3. The team deliberately avoids separating borrowings from code switching because the boundary is unreliable, and uses a simple "foreign" flag instead.
  • I4. Annotator disagreements are sorted into guideline errors, tooling-catchable errors, genuine ambiguity and typos, so that only the unavoidable ambiguity counts as a real ceiling.
  • I5. The authors propose putting a morphological analyzer inside the annotation interface to catch rule-violating spellings as they are entered.

For Mova

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

  • I1 (learn): Storing silver parses next to their human or teacher corrections would give repair-rule induction an explicit list of the parser's systematic errors.
  • I2 (tokens): Messy English input (texting spellings, repeated letters for emphasis) could be mapped to one canonical form before lemmatization, while the original surface string is kept for explanation.
  • I3 (tokens): Foreign or uncertain-origin tokens could get a coarse flag rather than a forced fine-grained decision, which avoids brittle errors further down the pipeline.
  • I4 (learn): Disagreements among witness parsers could be triaged the same way, separating true ambiguity from fixable errors before silver data is accepted.
  • I5 (answer): Graders for school-course test answers could validate morphology on the fly and flag malformed word forms before scoring the answer.

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

🎓University of PennsylvaniaPhiladelphia · United States221 works · 63,234 citations🎓Columbia UniversityNew York · United States241 works · 54,053 citations🎓New York University Abu DhabiAbu Dhabi · United Arab Emirates122 works · 4,602 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksHuman-Computer Interaction · 141 works

🧭 Topics 1Research subjects and languages the work is about.Semitic languages · 2,908 works

Cite

Cite as 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. Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/v1/w14-3612

@inproceedings{bies2014transliteration,
  title = {{Transliteration of Arabizi into Arabic Orthography: Developing a Parallel Annotated Arabizi-Arabic Script SMS/Chat Corpus}},
  author = {Ann Bies, Zhiyi Song, Mohamed Maamouri, Stephen Grimes, Haejoong Lee, Jonathan Wright, Stephanie Strassel, Nizar Habash, Ramy Eskander and Owen Rambow},
  booktitle = {Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP)},
  year = {2014},
  doi = {10.3115/v1/w14-3612},
  url = {https://doi.org/10.3115/v1/w14-3612}
}

References

27 references, 17 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mohamed Al-Badrashiny, Ramy Eskander, Nizar Habash, and Owen Rambow. 2014. Automatic Transliteration of Romanized Dialectal Arabic. In Proceedings of the Conference on Computational Natural Language Learning (CONLL), Baltimore, Maryland, 2014.
  2. Tim Buckwalter. 2004. Buckwalter Arabic Morphological Analyzer Version 2.0. LDC catalog number LDC2004L02, ISBN 1-58563-324-0. ↗
  3. Achraf Chalabi and Hany Gerges. 2012. Romanized Arabic Transliteration. In Proceedings of the Second Workshop on Advances in Text Input Methods (WTIM 2012).
  4. Eleanor Clark and Kenji Araki. 2011. Text normalization in social media: Progress, problems and applications for a pre-processing system of casual English. Procedia - Social and Behavioral Sciences, 27(0):2 – 11. find
  5. Kareem Darwish, Walid Magdy, and Ahmed Mourad. 2012. Language processing for arabic microblog re- trieval. In Proceedings of the 21st ACM International Conference on Information and Knowledge Management, CIKM ’12, pages 2427– 2430, New York, NY, USA. ACM. find
  6. Kareem Darwish. 2013. Arabizi Detection and Conversion to Arabic. CoRR, arXiv:1306.6755 [cs.CL]. ↗
  7. Leon Derczynski, Alan Ritter, Sam Clark, and Kalina Bontcheva. 2013. Twitter part-of-speech tagging for all: Overcoming sparse and noisy data. In Proceedings of the International Conference Recent Advances in Natural Language Processing RANLP 2013, pages 198–206, Hissar, Bulgaria, September. INCOMA Ltd. Shoumen, Bulgaria.
  8. Heba Elfardy, Mohamed Al-Badrashiny, and Mona Diab. 2013. Code Switch Point Detection in Arabic. In Proceedings of the 18th International Conference on Application of Natural Language to Information Systems (NLDB2013), MediaCity, UK, June. find
  9. Ramy Eskander, Mohamed Al-Badrashiny, Nizar Habash and Owen Rambow. 2014. Foreign Words and the Automatic Processing of Arabic Social Media Text Written in Roman Script. In Arabic Natural Language Processing Workshop, EMNLP, Doha, Qatar.
  10. Ramy Eskander, Nizar Habash, Owen Rambow, and Nadi Tomeh. 2013. Processing Spontaneous Orthography. In Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT), Atlanta, GA. find
  11. Andrew T. Freeman, Sherri L. Condon and Christopher M. Ackerman. 2006. Cross Linguistic Name Matching in English and Arabic: A “One to Many Mapping” Extension of the Levenshtein Edit Distance Algorithm. In Proceedings of HLT-NAACL, New York, NY.
  12. Kevin Gimpel, Nathan Schneider, Brendan O’Connor, Dipanjan Das, Daniel Mills, Jacob Eisenstein, Michael Heilman, Dani Yogatama, Jeffrey Flanigan, and Noah A. Smith. 2011. Part-of-speech tagging for twitter: Annotation, features, and experiments. In Proceedings of ACL-HLT ’11. ↗
  13. Stephan Gouws, Donald Metzler, Congxing Cai, and Eduard Hovy. 2011. Contextual bearing on linguistic variation in social media. In Proceedings of the Workshop on Languages in Social Media, LSM ’11, pages 20–29, Stroudsburg, PA, USA. Association for Computational Linguistics. find
  14. Nizar Habash, Mona Diab, and Owen Rambow (2012a).Conventional Orthography for Dialectal Arabic: Principles and Guidelines – Egyptian Arabic. Technical Report CCLS-12-02, Columbia University Center for Computational Learning Systems.
  15. Nizar Habash, Mona Diab, and Owen Rabmow. 2012b. Conventional Orthography for Dialectal Arabic. In Proceedings of the Language Resources and Evaluation Conference (LREC), Istanbul.
  16. Nizar Habash, Ramy Eskander, and Abdelati Hawwari. 2012c. A Morphological Analyzer for Egyptian Arabic. In Proceedings of the Twelfth Meeting of the Special Interest Group on Computational Morphology and Phonology, pages 1–9, Montréal, Canada.
  17. Kevin Knight and Jonathan Graehl. 1997. Machine Transliteration. In Proceedings of the Conference of the Association for Computational Linguistics (ACL).
  18. Linguistic Data Consortium. 2014. BOLT Program: Romanized Arabic (Arabizi) to Arabic Transliteration and Normalization Guidelines, Version 3.1. Linguistic Data Consortium, April 21, 2014. find
  19. Marco Lui, Jey Han Lau, and Timothy Baldwin. 2014. Automatic detection and language identification of multilingual documents. In Proceedings of the Language Resources and Evaluation Conference (LREC), Reykjavik, Iceland. ↗
  20. Mohamed Maamouri, Ann Bies, Seth Kulick, Michael Ciul, Nizar Habash and Ramy Eskander. 2014. Developing a dialectal Egyptian Arabic Treebank: Impact of Morphology and Syntax on Annotation and Tool Development. In Proceedings of the Language Resources and Evaluation Conference (LREC), Reykjavik, Iceland. find
  21. Yaser Al-Onaizan and Kevin Knight. 2002. Machine Transliteration of Names in Arabic Text. In Proceedings of ACL Workshop on Computational Approaches to Semitic Languages. find
  22. 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.
  23. Alan Ritter, Sam Clark, Mausam, and Oren Etzioni. 2011. Named entity recognition in tweets: An experimental study. In Proceedings of the Conference on Empirical Methods in Natural Language Processing, EMNLP ’11.
  24. Wael Salloum and Nizar Habash. 2011. Dialectal to Standard Arabic Paraphrasing to Improve Arabic-English Statistical Machine Translation. In Proceedings of the First Workshop on Algorithms and Resources for Modelling of Dialects and Language Varieties, pages 10–21, Edinburgh, Scotland. ↗
  25. Zhiyi Song, Stephanie Strassel, Haejoong Lee, Kevin Walker, Jonathan Wright, Jennifer Garland, Dana Fore, Brian Gainor, Preston Cabe, Thomas Thomas, Brendan Callahan, Ann Sawyer. Collecting Natural SMS and Chat Conversations in Multiple Languages: The BOLT Phase 2 Corpus. In Proceedings of the Language Resources and Evaluation Conference (LREC) 2014, Reykjavik, Iceland. find
  26. Clare Voss, Stephen Tratz, Jamal Laoudi, and Douglas Briesch. 2014. Finding romanized Arabic dialect in code-mixed tweets. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC’14), Reykjavik, Iceland.
  27. Omar F Zaidan and Chris Callison-Burch. 2011. The arabic online commentary dataset: an annotated dataset of informal arabic with high dialectal content. In Proceedings of ACL, pages 37–41.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/bies-2014-transliteration-arabizi-into-arabic-orthography/