Elamur Research⌕
☰
home / papers / 2016 / SAMER: A Semi-Automatically Created Lexical Resource for Arabic Verbal Multiword Expressions Tokens Paradigm and their Morphosyntactic Features

SAMER: A Semi-Automatically Created Lexical Resource for Arabic Verbal Multiword Expressions Tokens Paradigm and their Morphosyntactic Features

Paper · 2016 · Mohamed Al-Badrashiny, Abdelati Hawwari, Mahmoud Ghoneim et al.

Summary

The paper builds SAMER, a lexical resource for Arabic verbal multiword expressions (AVMWE) in Modern Standard Arabic: verbal idioms, light-verb constructions and verb–particle constructions. Listing every inflected and syntactic variant of such expressions by hand is slow, subjective and incomplete, and Arabic's rich morphology makes the problem worse. The authors start from a manually built seed list of about 4,000 AVMWE. Each word is lemmatized and POS-tagged. A paradigm detector (MPD) then searches the Arabic Gigaword and the ATB treebanks for flexible matches. These matches may contain inflected forms, verb derivatives, inserted gaps and changed word order.

For every expression type, the resource records:

  • a frequency-ranked list of variants,
  • an abstract syntactic template,
  • the most frequent surface form,
  • a nine-feature morphosyntactic vector for each word.

Checked against a gold annotated list, the detector reaches a weighted F-score of about 80%. The resource is useful for lexicography and for NLP tasks such as parsing and machine translation that must recognize MWEs despite their surface variation.

Key points

  • The detector found 732,335 matches covering 1,884 of the 4,000 seed AVMWE. These collapse into 38,408 grouped token forms, about 20 distinct forms per type on average.
  • The resource is delivered as five tables: all matches (with corpus pointers), flexibility, MWE-lexeme, sorted grouped tokens, and MWE types.
  • The weighted F-score is 80.61% on the dev set and 80.04% on unseen test data.
  • Gapping: 15.5% of types allow no gaps, 52.1% allow gaps between all words, and 32.4% allow gaps only in some positions.
  • Fixedness:
    • 15.7% of types never vary, and 4.7% also match derivatives of their verb.
    • Non-verbal parts stay fixed far more often than verbs (51.7% vs 17.7%).
    • The text calls mood the most rigid feature and gender the most flexible, but the Table 5/6 labels appear to state the opposite.

Interesting ideas

  • I1. Harvest a lexicon's real-world variants by turning each seed entry into lemmas and matching it against a large corpus, so the paradigm is observed rather than hand-written.
  • I2. Describe each MWE slot's flexibility as the union of feature values seen across matches; a set such as {sg, pl} means number may vary there, while a singleton means it is frozen.
  • I3. Let the matcher also accept deverbal derivatives of the verb, because nominalizations keep the verb's argument structure.
  • I4. Classify expressions by gap behaviour (none, everywhere, only in some positions). The error analysis also shows that gap size and reordering permissions must be set per expression, not globally.
  • I5. Inspect false matches where a variant fills a different syntactic function, as when a derived adverb is matched where an object noun was expected.

For Mova

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

  • I1 (tokens): Build English phrasal-verb and light-verb entries ("take into account", "give up on") by lemma-matching a seed list over corpora, storing the attested inflections per slot.
  • I2 (world): Compile per-slot allowed-feature sets into the verb-class knowledge so frozen slots ("kick the bucket" singular, definite) block literal readings.
  • I3 (frames): Map nominal forms like "the decision to…" or "his refusal of…" onto the same frame and roles as the verbal MWE.
  • I4 (tree): Store per-MWE gap limits and reordering flags so the parser can attach a split phrasal verb ("turn the lights off") correctly without overmatching elsewhere.
  • I5 (learn): Add a check that a matched variant keeps the seed's dependency role, and use such mismatches as negative controls when evaluating silver MWE annotations.

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

🎓George Washington UniversityWashington · United States51 works · 8,148 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 worksIdioms & multiword expressions · 217 works

Cite

Cite as Mohamed Al-Badrashiny, Abdelati Hawwari, Mahmoud Ghoneim and Mona Diab. 2016. SAMER: A Semi-Automatically Created Lexical Resource for Arabic Verbal Multiword Expressions Tokens Paradigm and their Morphosyntactic Features. Proceedings of the 12th Workshop on Asian Language Resources (ALR12).

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

@inproceedings{albadrashiny2016samer,
  title = {{SAMER: A Semi-Automatically Created Lexical Resource for Arabic Verbal Multiword Expressions Tokens Paradigm and their Morphosyntactic Features}},
  author = {Mohamed Al-Badrashiny, Abdelati Hawwari, Mahmoud Ghoneim and Mona Diab},
  booktitle = {Proceedings of the 12th Workshop on Asian Language Resources (ALR12)},
  year = {2016},
  url = {https://aclanthology.org/W16-5414/}
}

References

15 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Hassan Al-Haj, Alon Itai, and Shuly Wintner. 2013. Lexical representation of multiword expressions in morphologically-complex languages. International Journal of Lexicography, page ect036. ↗
  2. Rania Al-Sabbagh, Jana Diesner, and Roxana Girju. 2013. Using the semantic-syntactic interface for reliable arabic modality annotation. In Proceedings of the Sixth International Joint Conference on Natural Language Processing, pages 410–418, Nagoya, Japan, October. Asian Federation of Natural Language Processing.
  3. Timothy Baldwin, Colin Bannard, Takaaki Tanaka, and Dominic Widdows. 2003. An empirical model of multiword expression decomposability. In Proceedings of the ACL 2003 Workshop on Multiword Expressions: Analysis, Acquisition and Treatment - Volume 18, MWE ’03, pages 89–96, Stroudsburg, PA, USA. Association for Computational Linguistics.
  4. Nicoletta Calzolari, Charles J Fillmore, Ralph Grishman, Nancy Ide, Alessandro Lenci, Catherine MacLeod, and Antonio Zampolli. 2002. Towards best practice for multiword expressions in computational lexicons. In LREC. find
  5. Marine Carpuat and Mona Diab. 2010. Task-based evaluation of multiword expressions: A pilot study in statistical machine translation. In Human Language Technologies: The 2010 Annual Conference of the North American Chapter of the Association for Computational Linguistics, HLT ’10, pages 242–245, Stroudsburg, PA, USA. Association for Computational Linguistics. find
  6. Mona T. Diab and Pravin Bhutada. 2009. Verb noun construction mwe token supervised classification. In Proceedings of the Workshop on Multiword Expressions: Identification, Interpretation, Disambiguation and Applications, MWE ’09, pages 17–22, Stroudsburg, PA, USA. Association for Computational Linguistics. find
  7. Mahmoud Ghoneim and Mona T Diab. 2013. Multiword expressions in the context of statistical machine translation. In IJCNLP, pages 1181–1187. ↗
  8. Antton Gurrutxaga and Iaki Alegria. 2012. Measuring the compositionality of nv expressions in basque by means of distributional similarity techniques. In Nicoletta Calzolari (Conference Chair), Khalid Choukri, Thierry Declerck, Mehmet U?ur Do?an, Bente Maegaard, Joseph Mariani, Asuncion Moreno, Jan Odijk, and Stelios Piperidis, editors, Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC’12), Istanbul, Turkey, may. European Language Resources Association (ELRA). find
  9. Abdelati Hawwari, Kfir Bar, and Mona Diab. 2012. Building an arabic multiword expressions repository. Proc. of the 50th ACL, pages 24–29.
  10. Abdelati Hawwari, Wajdi Zaghouani, Tim O’Gorman, Ahmed Badran, and Mona Diab. 2013. Building a lexical semantic resource for arabic morphological patterns. In Communications, Signal Processing, and their Applications (ICCSPA), 2013 1st International Conference on, pages 1–6, Feb. find
  11. Abdelati Hawwari, Mohammed Attia, and Mona Diab. 2014. A framework for the classification and annotation of multiword expressions in dialectal arabic. ANLP 2014, page 48.
  12. Malvina Nissim and Andrea Zaninello. 2013. Modeling the internal variability of multiword expressions through a pattern-based method. ACM Transactions on Speech and Language Processing (TSLP), 10(2):7. ↗
  13. 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 LREC, Reykjavik, Iceland.
  14. Agata Savary. 2008. Computational Inflection of Multi-Word Units, a contrastive study of lexical approaches. Linguistic Issues in Language Technology, 1(2):1–53. find
  15. Andrea Zaninello and Malvina Nissim. 2010. Creation of lexical resources for a characterisation of multiword expressions in italian. In Nicoletta Calzolari (Conference Chair), Khalid Choukri, Bente Maegaard, Joseph Mariani, Jan Odijk, Stelios Piperidis, Mike Rosner, and Daniel Tapias, editors, Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC’10), Valletta, Malta, may. European Language Resources Association (ELRA). find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/badrashiny-2016-samer-semi-automatically-created-lexical/