Elamur Research⌕
☰
home / papers / 2019 / Universal dependencies for Scottish Gaelic: syntax

Universal dependencies for Scottish Gaelic: syntax

Summary

Scottish Gaelic is a low-resource language and until now has lacked a reliable syntactic parser. The author creates the first moderately sized treebank for it within UD. The corpus contains 1021 sentences from the ARCOSG corpus: fiction, news, narratives, and formal and popular texts. The existing manual morphological annotation of ARCOSG was automatically converted to the CoNLL-U format. During conversion, the tags were mapped to UPOS, and lemmas and grammatical features were added. The dependency trees were built iteratively. First, a small portion was annotated manually. Then MaltParser parsed the rest of the text, and the annotator checked the output and added the corrected trees to the training data. The annotation scheme was brought as close as possible to the Irish treebank. Contentious decisions concerning tokenization, copular constructions, the verbal noun, and cosubordination are discussed separately, with comparisons to the Irish and Breton corpora. In ten-fold cross-validation, the udpipe parser achieved a LAS of about 0.79. This resource provides a foundation for NLP tools for an endangered language. It also opens up the possibility of joint experiments with Irish.

Key points

  • Treebank size: 1021 sentences, approximately 20 thousand tokens. The subcorpora of conversations, interviews, and sports commentary have not yet been included. More than 750 sentences are still awaiting addition, and 30 sentences with cosubordination have been set aside.
  • ARCOSG consists of 76 texts and has more than 200 morphological tags. The author converts it from the Brown Corpus format to CoNLL-U and maps the tags to UD v2.
  • Tokenization differs from ARCOSG. Multiword units such as ann an are split into separate tokens, and the prefixes h-, n-, t- are attached to the following word. Inflected prepositions (orm) remain a single word, as in the Irish corpus, rather than being split into two, as in the Breton one.
  • Cosubordination is considered in two variants. The first relies on the acl relation and yields non-projective trees. The second treats the construction as ellipsis and uses orphan. The author prefers the first, as it reflects the language more accurately.
  • The LAS results for the udpipe (parsito) parser are as follows: 0.796 for the projective transition system, 0.789 for swap, and 0.792 for link2. The author notes that the average sentence length (19.6 words) is relatively short, so these scores may be somewhat inflated.

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

Cited by

7 papers of our library cite it (of the 140673 whose bibliographies we read); 1 of them have a page here, 6 not yet.

Cite

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

@inproceedings{batchelor2019universal,
  title = {{Universal dependencies for Scottish Gaelic: syntax}},
  author = {Batchelor, Colin},
  booktitle = {Proceedings of the Celtic Language Technology Workshop},
  year = {2019},
  url = {https://aclanthology.org/W19-6902/}
}

References

26 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Batchelor, Colin. 2014. gdbank: The beginnings of a corpus of dependency structures and type-logical grammar in Scottish Gaelic. In Proceedings of the First Celtic Language Technology Workshop, pages 60–65, Dublin, Ireland, August. Association for Computational Linguistics and Dublin City University… find
  2. Batchelor, Colin. 2016. Automatic derivation of categorial grammar from a part-of-speech-tagged corpus in Scottish Gaelic. In Actes de la conférence conjointe JEP-TALN-RECITAL 2016, volume 6 : CLTW, Paris, France, July. find
  3. Cox, Richard A. V. 2017. Geàrr-Ghràmar na Gàidhlig. Clann Tuirc, Tigh a’ Mhaide, Ceann Drochaid, Alba. find
  4. Lamb, William and Samuel Danso. 2014. Developing an Automatic Part-of-Speech Tagger for Scottish Gaelic. In Proceedings of the First Celtic Language Technology Workshop, pages 1–5, Dublin, Ireland, August. Association for Computational Linguistics and Dublin City University. find
  5. Lamb, William, Sharon Arbuthnot, Susanna Naismith, and Samuel Danso. 2016. Annotated Reference Corpus of Scottish Gaelic (ARCOSG), 1997–2016 [dataset]. Technical report, University of Edinburgh; School of Literatures, Languages and Cultures; Celtic and Scottish Studies. http://dx. doi.org/10.7488/ds… — [doi:10.7488/ds/1411](https://doi.org/10.7488/ds/1411) find
  6. Lamb, William. 1999. A diachronic account of Gaelic news-speak: The development and expansion of a register. Scottish Gaelic Studies, XIX:141–171. find
  7. Lamb, William. 2003. Scottish Gaelic, 2nd edn. Lincom Europa, Munich, Germany. find
  8. Lynn, Teresa and Jennifer Foster. 2016. Universal Dependencies for Irish. In Actes de la conférence conjointe JEP-TALN-RECITAL 2016, volume 6 : CLTW, Paris, France, July. find
  9. Lynn, Teresa, Jennifer Foster, and Mark Dras. 2013. Working with a small dataset - semi-supervised dependency parsing for Irish. In Proceedings of the Fourth Workshop on Statistical Parsing of Morphologically-Rich Languages, pages 1–11, Seattle, Washington, USA, October. Association for Computationa… find
  10. Lynn, Teresa, Jennifer Foster, and Mark Dras. 2017. Morphological Features of the Irish Universal Dependency Treebank. In TLT 2017: Proceedings of the 15th International Workshop on Treebanks and Linguistic Theories, Aachen, Germany. find
  11. Lynn, Teresa. 2016. Irish Dependency Treebanking and Parsing. Ph.D. thesis, Dublin City University and Macquarie University. find
  12. Moseley, Christopher, editor. 2010. Atlas of the World’s Languages in Danger. UNESCO Publishing, Paris, 3rd edition. find
  13. Naismith, Susanna and William Lamb. 2014. Scottish Gaelic Part-of-Speech Annotation Guidelines. Celtic and Scottish Studies, University of Edinburgh. find
  14. Nilsson, Jens and Joakim Nivre. 2008. MaltEval: an evaluation and visualization tool for dependency parsing. In LREC 2008. find
  15. Nivre, Joakim and Chiao-Ting Fang. 2017. Universal dependency evaluation. In Proceedings of the NoDaLiDa 2017 Workshop on Universal Dependencies (UDW 2017), pages 86–95, Gothenburg, Sweden, May. Association for Computational Linguistics. find
  16. Nivre, Joakim, Johan Hall, and Jens Nilsson. 2006. MaltParser: A data-driven parser-generator for dependency parsing. In Proceedings of the Fifth International Conference on Language Resources and Evaluation (LREC’06), Genoa, Italy, May. European Language Resources Association (ELRA). find our card
  17. Ross, Susan, Mark McConville, Wilson McLeod, and Richard Cox. 2019. Stiùireadh Gràmair. https://dasg.ac.uk/grammar/grammar.pdf, accessed 23 May 2019. ↗
  18. (SQA), Scottish Qualifications Authority. 2009. Gaelic Orthographic Conventions. Scottish Qualifications Authority, Dalkeith. find
  19. Stifter, David. 2006. Sengoidelc: Old Irish for beginners. Syracuse University Press, Syracuse, New York. find
  20. Straka, Milan and Jana Straková. 2017. Tokenizing, pos tagging, lemmatizing and parsing ud 2.0 with udpipe. In Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 88–99, Vancouver, Canada, August. Association for Computational Linguistics. find our card
  21. Tyers, Francis M. and Vinit Ravishankar. 2018. A prototype dependency treebank for Breton. In Actes de la conférence Traitement Automatique de la Langue-Naturelle, TALN 2018 : Volume 1 : Articles longs, articles courts de TALN, pages 197–204, Rennes, France, May. find
  22. Uı́ Dhonnchadha, Elaine. 2009. Part-of-speech tagging and partial parsing for Irish using finite-state transducers and constraint grammar. Ph.D. thesis, Dublin City University. find
  23. Universal Dependencies contributors. 2016. UD Guidelines, https://universaldependencies.org/ guidelines.html. Accessed 9 July 2019. ↗
  24. University of Glasgow. 2019. Corpas na Gàidhlig, Digital Archive of Scottish Gaelic (DASG), https://dasg.ac.uk/corpus/. Accessed 9 July 2019. ↗
  25. Zeman, Daniel, Jan Hajič, Martin Popel, Martin Potthast, Milan Straka, Filip Ginter, Joakim Nivre, and Slav Petrov. 2018. find
  26. CoNLL 2018 shared task: Multilingual parsing from raw text to universal dependencies. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 1–21, Brussels, Belgium, October. Association for Computational Linguistics. find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/batchelor-2019-universal-dependencies-scottish-gaelic-syntax/