Elamur Research⌕
☰
home / papers / 2022 / Towards Learning Arabic Morphophonology

Towards Learning Arabic Morphophonology

Paper · 2022 · Salam Khalifa, Jordan Kodner and Owen Rambow

Summary

Inflection systems often fail on the sound changes that happen when morphemes combine. Arabic is a hard case: its morphology mixes templates with affixation, and most of its dialects have few resources. The authors induce explicit, human-readable morphophonological rules for Cairene Egyptian Arabic (EGY), with the longer-term aim of reusing such rules for related dialects.

No suitable dataset existed, so they built one. Surface forms come from the ECAL pronunciation lexicon. Segmented underlying forms are derived from CALIMA EGY analyses, which CamelTools selects by matching orthography, lemma, POS and features.

Rules are extracted in four steps:

  • align each (UR, SF) pair with Levenshtein edits;
  • trim the context to a narrow window around the changes;
  • replace stem consonants with a generic C;
  • at prediction time, pick the longest and most specific matching left side, using frequency to break ties.

The rules regenerate surface forms with about 82% accuracy on EVAL and roughly 69% on types never seen in training. The most frequent rules correspond to processes already described in grammars of Egyptian Arabic. This suggests an interpretable alternative to neural inflection models.

Key points

  • The dataset pairs ECAL surface forms (SF) with underlying forms (UR) built from CALIMA EGY segmentations. Segmentation comes from the Buckwalter POS tag, and boundaries are marked with - (prefix), = (suffix) and # (word). Only open-class words (verbs, nouns, adjectives) are kept.

  • Split sizes in types are TRAIN 13,170, DEV 5,180 and EVAL 6,974. Overlap-free subsets were also created: DEV-OOV has 2,189 types and EVAL-OOV has 2,271.

  • Extraction produced 4,661 rules, about 35.4% of the TRAIN size. Each rule rewrites all changes in one step, and unchanged words map to a "copy" rule.

  • Accuracy results:

    SplitAll typesOOV types
    TRAIN90.1%–
    DEV80.5%69.4%
    EVAL82.3%68.9%

  • The error analysis covered the 31% of DEV-OOV forms that were produced incorrectly:

    • 32% of these errors had a correct rule already in the inventory, so the selection heuristic was at fault;
    • 68% needed a rule that had never been seen;
    • in a sample of 100 of those unseen rules, every phenomenon was already covered by existing rules, and only the surrounding context was new.

Interesting ideas

  • I1. Create supervision for a task with no annotated corpus by automatically joining two independent resources (a pronunciation lexicon and a morphological analyzer), choosing the best-matching analysis on several fields at once.
  • I2. Derive rules from character edit alignments, then generalize them by keeping only a small window around the change and abstracting stem consonants to a placeholder, while vowels and affixes stay explicit.
  • I3. Let one rule encode every change of a word at once, in two-level style, so the system never has to discover rule ordering.
  • I4. Report accuracy on full running-text splits and also on overlap-removed OOV splits, so memorization and generalization are measured separately.
  • I5. Sort errors into "the right rule existed but selection chose wrongly" versus "the needed rule was never learned" to decide whether to fix the ranking or the rule induction.

For Mova

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

  • I1 (learn): silver morphology pairs could come from aligning treebank forms with analyzer or WordNet lemma entries and keeping only entries where all fields agree, giving cheap training data.
  • I2 (tokens): English spelling changes at morpheme joins (consonant doubling, y→i, e-deletion) could be learned as readable rewrite rules from lemma+affix → form pairs, replacing hand-written exceptions.
  • I3 (back): when generating surface words from lemma plus features, single whole-change rules keep each output traceable to exactly one rule.
  • I4 (learn): every lemmatizer or repair-rule evaluation could include an unseen-type subset as a built-in control against inflated scores.
  • I5 (learn): self-induced repair rules could be audited the same way, showing whether failures come from rule choice or from missing coverage.

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

🎓Stony Brook UniversityStony Brook · United States43 works · 2,571 citations

 Fields, topics and concepts

📚 Shelves 4The 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 works

🧭 Topics 2Research subjects and languages the work is about.Semitic languages · 2,908 worksMorphology · 405 works

Cite

Cite as Salam Khalifa, Jordan Kodner and Owen Rambow. 2022. Towards Learning Arabic Morphophonology. Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2022.wanlp-1.27

@inproceedings{khalifa2022learning,
  title = {{Towards Learning Arabic Morphophonology}},
  author = {Salam Khalifa, Jordan Kodner and Owen Rambow},
  booktitle = {Proceedings of the Seventh Arabic Natural Language Processing Workshop (WANLP)},
  year = {2022},
  doi = {10.18653/v1/2022.wanlp-1.27},
  url = {https://doi.org/10.18653/v1/2022.wanlp-1.27}
}

References

32 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Ernest T. Abdel-Massih, Zaki N. Abdel-Malek, and El-Said M. Badawi. 1979. A Reference Grammar of Egyptian Arabic. Georgetown University Press. find
  2. Bashar Alhafni, Nizar Habash, and Houda Bouamor. 2020. Gender-aware reinflection using linguistically enhanced neural models. In Proceedings of the Second Workshop on Gender Bias in Natural Language Processing, pages 139–150, Barcelona, Spain (Online). Association for Computational Linguistics. find
  3. Evan L Antworth. 1991. Introduction to two-level phonology. Notes on Linguistics, 53:4–18. find
  4. Khuyagbaatar Batsuren, Gábor Bella, Aryaman Arora, Viktor Martinovic, Kyle Gorman, Zdeněk Žabokrtský, Amarsanaa Ganbold, Šárka Dohnalová, Magda Ševčíková, Kateřina Pelegrinová, Fausto Giunchiglia, Ryan Cotterell, and Ekaterina Vylomova. 2022. find
  5. The SIGMORPHON 2022 Shared Task on Morpheme Segmentation. In Proceedings of the 19th SIGMOR-PHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 103–116, Seattle, Washington. Association for Computational Linguistics.
  6. Kenneth Beesley. 1998. Arabic morphology using only finite-state operations. In Proceedings of the Workshop on Computational Approaches to Semitic Languages (CASL), pages 50–7, Montereal.
  7. Caleb A Belth, Sarah RB Payne, Deniz Beser, Jordan Kodner, and Charles Yang. 2021. The greedy and recursive search for morphological productivity. In Proceedings of the Annual Meeting of the Cognitive Science Society, volume 43. ↗
  8. Ellen Broselow. 2017. Syllable Structure in the Dialects of Arabic. The Routledge handbook of Arabic linguistics, pages 32–47. ↗
  9. Tim Buckwalter. 2002. Buckwalter Arabic morphological analyzer version 1.0. Linguistic Data Consortium (LDC) catalog number LDC2002L49, ISBN 1-58563-257-0. ↗
  10. Tim Buckwalter. 2004. Buckwalter Arabic Morphological Analyzer Version 2.0. LDC catalog number LDC2004L02, ISBN 1-58563-324-0. ↗
  11. Noam Chomsky and Morris Halle. 1968. The Sound Pattern of English. Harper & Row New York. find
  12. Verna Dankers, Anna Langedijk, Kate McCurdy, Adina Williams, and Dieuwke Hupkes. 2021. Generalising to German plural noun classes, from the perspective of a recurrent neural network. In Proceedings of the 25th Conference on Computational Natural Language Learning, pages 94–108, Online. Association for Computational Linguistics. find
  13. Alexander Erdmann, Salam Khalifa, Mai Oudah, Nizar Habash, and Houda Bouamor. 2019. A Little Linguistics Goes a Long Way: Unsupervised Segmentation with Limited Language Specific Guidance. In Proceedings of the 16th Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 113–124, Florence, Italy. Association for Computational Linguistics. find
  14. Hassan Gadalla, Hanaa Kilany, Howaida Arram, Ashraf Yacoub, Alaa El-Habashi, Amr Shalaby, Krisjanis Karins, Everett Rowson, Robert MacIntyre, Paul Kingsbury, David Graff, and Cynthia McLemore. 1997. CALLHOME Egyptian Arabic transcripts LDC97T19. Web Download. Philadelphia: Linguistic Data Consortium. find
  15. 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
  16. Nizar Habash, Ramy Eskander, and Abdelati Hawwari. 2012. A Morphological Analyzer for Egyptian Arabic. In Proceedings of the Workshop of the Special Interest Group on Computational Morphology and Phonology (SIGMORPHON), pages 1–9, Montréal, Canada.
  17. Nizar Habash, Reham Marzouk, Christian Khairallah, and Salam Khalifa. 2022. Morphotactic modeling in an open-source multi-dialectal Arabic morphological analyzer and generator. In Proceedings of the 19th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 92–102, Seattle, Washington. Association for Computational Linguistics. ↗
  18. Nizar Habash and Owen Rambow. 2006. MAGEAD: A morphological analyzer and generator for the Arabic dialects. In Proceedings of the International Conference on Computational Linguistics and the Conference of the Association for Computational Linguistics (COLING-ACL), pages 681–688, Sydney, Australia. ↗
  19. Nawar Halabi. 2016. Modern standard Arabic phonetics for speech synthesis. Ph.D. thesis, University of Southampton. find
  20. Bruce Hayes. 2008. Introductory Phonology. Blackwell Textbooks in Linguistics. Wiley. find
  21. Go Inoue, Salam Khalifa, and Nizar Habash. 2022. Morphosyntactic Tagging with Pre-trained Language Models for Arabic and its Dialects. In Findings of the Association for Computational Linguistics: ACL 2022, pages 1708–1719, Dublin, Ireland. Association for Computational Linguistics.
  22. Salam Khalifa, Sara Hassan, and Nizar Habash. 2017. A morphological analyzer for Gulf Arabic verbs. In Proceedings of the Workshop for Arabic Natural Language Processing (WANLP), Valencia, Spain.
  23. Salam Khalifa, Nasser Zalmout, and Nizar Habash. 2020. Morphological Analysis and Disambiguation for Gulf Arabic: The Interplay between Resources and Methods. In Proceedings of the 12th Language Resources and Evaluation Conference, pages 3895– 3904, Marseille, France. European Language Resources Association.
  24. Hanaa Kilany, Hassan Gadalla, Howaida Arram, Ashraf Yacoub, Alaa El-Habashi, and Cynthia McLemore. 2002. Egyptian Colloquial Arabic Lexicon. LDC catalog number LDC99L22. ↗
  25. Christo Kirov and Ryan Cotterell. 2018. Recurrent neural networks in linguistic theory: Revisiting pinker and prince (1988) and the past tense debate. Transactions of the Association for Computational Linguistics, 6:651–665. ↗
  26. Jordan Kodner and Salam Khalifa. 2022. SIGMORPHON–UniMorph 2022 shared task 0: Modeling inflection in language acquisition. In Proceedings of the 19th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 157–175, Seattle, Washington. Association for Computational Linguistics. find
  27. Jordan Kodner, Salam Khalifa, Khuyagbaatar Batsuren, Hossep Dolatian, Ryan Cotterell, Faruk Akkus, Antonios Anastasopoulos, Taras Andrushko, Aryaman Arora, Nona Atanalov, Gábor Bella, Elena Budianskaya, Yustinus Ghanggo Ate, Omer Goldman, David Guriel, Simon Guriel, Silvia Guriel-Agiashvili, Witold Kieraś, Andrew Krizhanovsky, Natalia Krizhanovsky, Igor Marchenko, Magdalena Markowska, Polina Mashkovtseva, Maria Nepomni-ashchaya, Daria Rodionova, Karina Scheifer, Alexandra Sorova, Anastasia Yemelina, Jeremiah Young, and Ekaterina Vylomova. 2022. SIGMORPHON– UniMorph 2022 shared task 0: Generalization and typologically diverse morphological inflection. In Proceedings of the 19th SIGMORPHON Workshop on Computational Research in Phonetics, Phonology, and Morphology, pages 176–203, Seattle, Washington. Association for Computational Linguistics. find
  28. Karthik Narasimhan, Regina Barzilay, and Tommi Jaakkola. 2015. An unsupervised method for uncovering morphological chains. Transactions of the Association for Computational Linguistics, 3:157– 167. ↗
  29. Ossama Obeid, Nasser Zalmout, Salam Khalifa, Dima Taji, Mai Oudah, Bashar Alhafni, Go Inoue, Fadhl Eryani, Alexander Erdmann, and Nizar Habash. 2020. CAMeL tools: An open source python toolkit for Arabic natural language processing. In Proceedings of The 12th Language Resources and Evaluation Conference, pages 7022–7032, Marseille, France. European Language Resources Association.
  30. Rico Sennrich and Barry Haddow. 2016. Linguistic input features improve neural machine translation. In Proceedings of the First Conference on Machine Translation: Volume 1, Research Papers, volume 1, pages 83–91.
  31. Dima Taji, Jamila El Gizuli, and Nizar Habash. 2018. An Arabic dependency treebank in the travel domain. In Proceedings of the Workshop on Open-Source Arabic Corpora and Processing Tools (OS-ACT), Miyazaki, Japan. ↗
  32. Shijie Wu, Ryan Cotterell, and Mans Hulden. 2021. Applying the transformer to character-level transduction. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 1901–1907, Online. Association for Computational Linguistics. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/khalifa-2022-learning-arabic-morphophonology/