Elamur Research⌕
☰
home / papers / 2013 / Exploring Morphosyntactic Annotation over a Spanish Corpus for Dependency Parsing

Exploring Morphosyntactic Annotation over a Spanish Corpus for Dependency Parsing

Summary

The authors investigate how morphosyntactic information in a treebank affects statistical dependency parsing of Spanish. They sought answers to three questions: whether all morphological features should be added to the model, which feature combinations work best, and why some features are more useful than others. The data come from AnCora-UPF, a part of the AnCora corpus with detailed language-specific annotation and 41 syntactic relation types. Using different feature sets (number, gender, person, mood, tense, finiteness, part-of-speech tags), the authors created 82 versions of the corpus. For each version, a MaltParser model was tuned automatically with MaltOptimizer and evaluated by LAS, UAS and recall for individual relation types. The best overall set turned out to be [finiteness gender number spos]. Thus, for Spanish a small number of well-chosen features is sufficient, and full annotation is not needed. The work shows that the result is influenced both by a feature's frequency and by how features interact with one another. This is worth taking into account when annotating treebanks for morphologically rich languages.

Key points

  • The training set contains 3263 sentences (93 803 tokens), and the test set 250 sentences (7089 tokens). The average sentence length is about 28 tokens.
  • The baseline model had no morphological features and used the parameters from Phase 2 of MaltOptimizer. For each of the 82 corpus versions, Phase 3, which selects features, was run separately.
  • The features that occur most frequently in the corpus almost always yield the largest gain in quality. At the same time, the contribution of each feature on its own does not make it possible to predict which combination will be best.
  • For specific constructions, the best sets differ from the best overall one. For verbal arguments and the sentence root it is [finiteness number person spos]. For the internal structure of noun phrases it is [gender mood number person spos tense].
  • When the more detailed spos tags were placed in the POSTAG column instead of the Tree Tagger tags, LAS increased by about 0.5 points. The authors attribute this primarily to the fact that spos splits the IN tag into preposition and conjunction.

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: Dependency parsing · Treebanks

Cite

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

@inproceedings{ballesteros2013exploring,
  title = {{Exploring Morphosyntactic Annotation over a Spanish Corpus for Dependency Parsing}},
  author = {Ballesteros, Miguel and Mille, Simon and Burga, Alicia},
  booktitle = {Proceedings of the Second International Conference on Dependency Linguistics (DepLing 2013)},
  year = {2013},
  url = {https://aclanthology.org/W13-3703/}
}

References

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

  1. A. Atutxa, E. Agirre, and K. Sarasola. 2012. Contribution of Complex Lexical Information to Solve Syntactic Ambiguity in Basque. In Proceedings of COLING, pages 97–114. find
  2. M. Ballesteros and J. Nivre. 2012a. MaltOptimizer: A System for MaltParser Optimization. In Proceedings of the 8th LREC, pages 2757–2763. find
  3. M. Ballesteros and J. Nivre. 2012b. MaltOptimizer: An Optimization Tool for MaltParser. In Proceedings of the System Demonstration Session of the 13th EACL, pages 58–62. find
  4. K. Bengoetxea and K. Gojenola. 2009. Application of feature propagation to dependency parsing. In Proceedings of IWPT, pages 142–145. find
  5. B. Bohnet and J. Kuhn. 2012. The Best of Both Worlds - A Graph-based Completion Model for transition-based parsers. In Proceedings of EACL, pages 77–87. find
  6. B. Bohnet and J. Nivre. 2012. A Transition-Based System for Joint Part-of-Speech Tagging and Labeled Non-Projective Dependency Parsing. In Proceedings of EMNLP-CoNLL, pages 1455–1465. find
  7. S. Buchholz and E. Marsi. 2006. CoNLL-X Shared Task on Multilingual Dependency Parsing. In Proceedings of CoNLL-06, pages 149–164. find our card
  8. A. Burga, S. Mille, and L. Wanner. 2011. Looking Behind the Scenes of Syntactic Dependency Corpus Annotation: Towards a Motivated Annotation Schema of Surface-Syntax in Spanish. In Proceedings of DepLing ’11, pages 104–114. find
  9. B. Cowan and M. Collins. 2005. Morphology and reranking for the statistical parsing of Spanish. In Proceedings of EMNLP, pages 795–802. find
  10. G. Eryigit, J. Nivre, and K. Oflazer. 2008. Dependency parsing of Turkish. Computational Linguistics, 34(3):357–389. find
  11. D. Klein and Ch. Manning. 2003. Accurate Unlexicalized Parsing. In Proceedings of ACL-03, pages 423–430. find our card
  12. F. Mambrini and M.C. Passarotti. 2012. Will a Parser Overtake Achilles? First experiments on parsing the Ancient Greek Dependency Treebank. In Proceedings of the 11th TLT, pages 133–144. find
  13. R. McDonald, F. Pereira, K. Ribarov, and J. Hajič. 2005. Non-Projective Dependency Parsing using Spanning Tree Algorithms. In HLT-EMNLP, pages 523–530. find our card
  14. I.A. Mel’čuk. 1988. Dependency Syntax: Theory and Practice. State University of New York Press, Albany. find
  15. S. Mille, L. Wanner, V. Vidal, and A. Burga. 2009. Towards a rich dependency annotation of Spanish corpora. Procesamiento del Lenguaje Natural, 1(43):325–333. find
  16. S. Mille, A. Burga, G. Ferraro, and L. Wanner. 2012. How Does the Granularity of an Annotation Scheme Influence Dependency Parsing Performance? In Proceedings of COLING 2012, pages 839–852. find
  17. S. Mille, A. Burga, and L. Wanner. 2013. Ancora-UPF: A Multi-Level Annotation of Spanish. In Proceedings of DepLing. find
  18. J. Nivre, J. Hall, S. Kübler, R. McDonald, J. Nilsson, S. Riedel, and D. Yuret. 2007a. find
  19. The CoNLL 2007 Shared Task on Dependency Parsing. In Proceedings of CoNLL-ST-07, pages 915–932. find
  20. J. Nivre, J. Hall, J. Nilsson, A. Chanev, G. Eryiǧit, S. Kübler, S. Marinov, and E. Marsi. 2007b. Maltparser: A Language-Independent System for Data-Driven Dependency Parsing. Natural Language Engineering, 13:95–135. find
  21. J. Nivre. 2003. An Efficient Algorithm for Projective Dependency Parsing. In Proceedings of IWPT-03, pages 149–160. find our card
  22. R. Schwartz, O. Abend, and A. Rappoport. 2012. Learnability-based Syntactic Annotation Design. In Proceedings of COLING 2012, pages 2405–2422. find
  23. M. Seraji, B. Megyesi, and J. Nivre. 2012. Dependency parsers for Persian. In Proceedings of 10th Workshop on Asian Language Resources, COLING 2012, pages 35–44. find
  24. M. Taulé, M. A. Martı́, and M. Recasens. 2008. Ancora: Multilevel Annotated Corpora for Catalan and Spanish. In Proceedings of the 6th LREC, pages 96– 101. find
  25. R. Tsarfaty, D. Seddah, Y. Goldberg, S. Kübler, M. Candito, J. Foster, Y. Versley, I. Rehbein, and L. Tounsi. 2010. Statistical Parsing of Morphologically Rich Languages (SPMRL): What, How and Whither. In Proceedings of the NAACL HLT 2010 First Workshop on SPMRL, pages 1–12. find
  26. R. Tsarfaty, J. Nivre, and E. Andersson. 2012a. Cross-Framework Evaluation for Statistical Parsing. In Proceedings of EACL, pages 44–54. find
  27. R. Tsarfaty, D. Seddah, S. Kübler, and J. Nivre. 2012b. Parsing Morphologically Rich Languages: Introduction to the Special Issue. Computational Linguistics, 39(1):15–22. find
  28. Y. Zhang and S. Clark. 2008. A Tale of Two Parsers: Investigating and Combining Graph-based and Transition-based Dependency Parsing. In Proceedings of EMNLP, pages 562–571. find our card
  29. Y. Zhang and J. Nivre. 2011. Transition-Based Parsing with Rich Non-Local Features. In Proceedings of ACL, pages 188–193. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/ballesteros-2013-exploring-morphosyntactic-annotation-over-spanish/