Elamur Research⌕
☰
home / papers / 2021 / Overcoming the challenges in morphological annotation of Turkish in universal dependencies framework

Overcoming the challenges in morphological annotation of Turkish in universal dependencies framework

Summary

Turkish is an agglutinative language with highly productive word formation. The authors show that the UD guidelines are poorly suited to such languages and identify three problems. First, UD does not record derivation, so the lemma becomes the already derived form, and the morphemes preceding the last derivational affix are lost. Second, the framework has no means of marking null morphemes, in particular the present-tense copula, which in Turkish is always null. Third, the polysemous verb ol- ("to be", "to become") was annotated inconsistently in the BOUN and IMST-UD treebanks. A team of three linguists and four computer scientists proposes solutions that stay within the existing UD toolkit. Information about the root and the hidden copula is recorded in the MISC column, and six uses of ol- are distinguished, each with its own UPOS, XPOS and dependency type. The authors discussed these issues with the community on the official UD portal. Under the new scheme, 2 000 sentences of the BOUN Treebank have already been re-annotated, yielding thousands of corrections. The work can serve as a model for other agglutinative languages and help train more accurate lemmatizers and parsers.

Key points

  • Re-annotation of 2 000 sentences of the BOUN Treebank: 5 418 POS tags were changed, 12 654 morphological features were added and 9 513 were removed. In addition, 701 words were split into several syntactic words, 5 456 dependency relations and 7 333 heads were changed, and 2 895 MISC fields were filled in.
  • The BOUN Treebank contains 9 761 manually annotated sentences. It is a sample from the Turkish National Corpus (over 50 million words) and one of the largest Turkish UD collections. At the time of writing, UD v2.8 comprised 202 treebanks for 114 languages.
  • Derivation: a single Turkish word can contain up to six derivational suffixes. The authors considered the approaches of Quechua (entries such as Root_VDeriv) and Finnish TDT (the Derivation feature), but chose a marker in MISC, e.g. df=iş for the lemma işçi.
  • The null present-tense copula is marked in MISC (nullcop=3p), e.g. in the sentence İşçiler grevde.
  • Scheme for ol-: in the sense "to become" it is a VERB with an xcomp complement; as a copula in embedded clauses it is tagged AUX with the cop dependency; as the embedded form of var/yok it receives the new XPOS Exist; as an auxiliary verb it is AUX with aux; in light-verb constructions it is compound:lvc.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Universal Dependencies · Treebanks

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2021.law-1.12

@inproceedings{bedir2021overcoming,
  title = {{Overcoming the challenges in morphological annotation of Turkish in universal dependencies framework}},
  author = {Bedir, Talha and Şahin, Karahan and Gungor, Onur and Uskudarli, Suzan and Özgür, Arzucan and Güngör, Tunga and Basaran, Balkiz Ozturk},
  booktitle = {Proceedings of the Joint 15th Linguistic Annotation Workshop (LAW) and 3rd Designing Meaning Representations (DMR) Workshop},
  year = {2021},
  doi = {10.18653/v1/2021.law-1.12},
  url = {https://doi.org/10.18653/v1/2021.law-1.12}
}

References

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

  1. Yesim Aksan, Mustafa Aksan, Ahmet Koltuksuz, Taner Sezer, Ümit Mersinli, Umut Ufuk Demirhan, Hakan Yilmazer, Gülsüm Atasoy, Seda Öz, Ipek Yildiz, et al. 2012. Construction of the Turkish national corpus (tnc). In LREC, pages 3223–3227. find
  2. Nart B Atalay, Kemal Oflazer, and Bilge Say. 2003. The annotation process in the Turkish treebank. In Proceedings of 4th International Workshop on Linguistically Interpreted Corpora (LINC-03) at EACL 2003. find our card
  3. Çağrı Çöltekin. 2010. A freely available morphological analyzer for Turkish. In Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC’10), Valletta, Malta. European Language Resources Association (ELRA). find
  4. Aysel Kapan. 2019. Derivational networks of nouns and adjectives in Turkish. Master’s thesis, Boğaziçi University, İstanbul, Turkey. find
  5. Meltem Kelepir. 2001. Topics in Turkish syntax: Clausal structure and scope. Ph.D. thesis, Massachusetts Institute of Technology. find
  6. Jaklin Kornfilt. 1996. On copular clitic forms in Turkish. ZAS Working Papers. find
  7. Geoffrey Leech and Roger Garside. 1991. Running a grammar factory: The production of syntactically analysed corpora or treebanks. English Computer Corpora: Selected Papers and Research Guide, pages 15–32. find
  8. Mitchell Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. 1993. Building a large annotated corpus of english: The penn treebank. find our card
  9. Kemal Oflazer, Bilge Say, Dilek Zeynep Hakkani-Tür, and Gökhan Tür. 2003. Building a Turkish treebank. In Treebanks, pages 261–277. Springer. find
  10. Çağrı Çöltekin. 2015. A grammar-book treebank of Turkish. In Proceedings of the 14 th Workshop on Treebanks and Linguistic Theories (TLT 14), pages 35–49. find
  11. Marie-Catherine De Marneffe, Timothy Dozat, Natalia Silveira, Katri Haverinen, Filip Ginter, Joakim Nivre, and Christopher D Manning. 2014. Universal Stanford dependencies: A cross-linguistic typology. In LREC, volume 14, pages 4585–4592. find our card
  12. Marie-Catherine De Marneffe, Bill MacCartney, Christopher D Manning, et al. 2006. Generating typed dependency parses from phrase structure parses. In Lrec, volume 6, pages 449–454. find our card
  13. Marie-Catherine De Marneffe and Christopher D Manning. 2008. The Stanford typed dependencies representation. In Coling 2008: proceedings of the workshop on cross-framework and cross-domain parser evaluation, pages 1–8. find our card
  14. Aslı Göksel. 2001. The auxiliary verb ol at the morphology-syntax interface. The Verb in Turkish. Amsterdam: John Benjamins, pages 151–181. find
  15. Aslı Göksel and Celia Kerslake. 2005. Turkish: A Comprehensive Grammar. Comprehensive grammars. Routledge. find
  16. Balkız Öztürk and Eser Erguvanlı Taylan. 2018. Türkçede iyelik yapıları ve geçişli adlar. In Dilbil-imde Güncel Tartışmalar. Ankara: Dilbilim Dernegi Yayınları. find
  17. Slav Petrov, Dipanjan Das, and Ryan McDonald. 2011. A universal part-of-speech tagset. arXiv preprint arXiv:1104.2086. find
  18. Annette Rios, Anne Göhring, and Martin Volk. 2012. Parallel treebanking spanish-quechua: how and how well do they align? Linguistic Issues in Language Technology, (7/13):online. find
  19. Umut Sulubacak, Gülşen Eryiğit, Tuğba Pamay, et al. 2016a. Imst: A revisited Turkish dependency treebank. In Proceedings of TurCLing 2016, the 1st International Conference on Turkic Computational Linguistics. Ege University Press. find
  20. Umut Sulubacak, Memduh Gökırmak, Francis Tyers, Çağrı Çöltekin, Joakim Nivre, and Gülşen Eryiğit. 2016b. Universal dependencies for Turkish. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 3444–3454. find
  21. Jorge Hankamer. 2005. Why there are two –ki’s in Turkish. Ms., USC. find
  22. Utku Türk, Furkan Atmaca, Şaziye Betül Özateş, Gözde Berk, Seyyit Talha Bedir, Abdullatif Köksal, Balkız Öztürk, Tunga Güngör, and Arzucan Özgür. 2020. Resources for Turkish dependency parsing: Introducing the boun treebank and the boat annotation tool. arXiv preprint arXiv:2002.10416. — [arXiv:2002.10416](https://arxiv.org/abs/2002.10416) find
  23. Jenna Kanerva, Filip Ginter, Niko Miekka, Akseli Leino, and Tapio Salakoski. 2018. Turku neural parser pipeline: An end-to-end system for the CoNLL 2018 shared task. In Proceedings of the CoNLL 2018 Shared Task: Multilingual parsing from raw text to universal dependencies, pages 133– 142. find
  24. Utku Türk, Furkan Atmaca, Şaziye Betül Özateş, Abdullatif Köksal, Balkiz Öztürk, Tunga Güngör, and Arzucan Özgür. 2019a. Turkish treebanking: Unifying and constructing efforts. In Proceedings of the 13 th Linguistic Annotation Workshop, pages 166– 177, Florence, Italy. Association for Computationa… find
  25. Jorge Hankamer. 2004. An ad-phrasal affix in Turkish. MIT Working Papers in Linguistics, 46:289–299. find
  26. Utku Türk, Furkan Atmaca, Şaziye Betül Özateş, Balkız Öztürk, Tunga Güngör, and Arzucan Özgür. 2019b. Improving the annotations in the Turkish universal dependency treebank. In Proceedings of the Third Workshop on Universal Dependencies (UDW, SyntaxFest 2019), pages 108–115. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/bedir-2021-overcoming-challenges-morphological-annotation-turkish/