Using Modern Languages to Parse Ancient Ones: a Test on Old English

What it is about

Old English has treebanks only in the constituency style of YCOE. The authors convert 292 sentences of two West Saxon prose texts to Universal Dependencies by hand, starting from the YCOE morphology, and split them into 149 training, 73 development and 70 test sentences. The parser is UUParser, trained once per combination of Old English and three modern Germanic treebanks (Swedish, Icelandic, German), each cut to 60,000 tokens. Old English alone gives a LAS of 47.23. Adding Icelandic gives 58.70, and adding Icelandic and German gives a slightly lower LAS but the best label accuracy. The paper then studies the errors on advmod, obl and acl:relcl (for example, the particle "ne" and the postposition "ongean"). A rule table that fixes labels by word form, UPOS and the old tag raises LAS for the best model to 59.30 (the label accuracy rises by one to two points).

Ideas

• I1. Keep a conversion table from the source tags to UD labels, as the paper does (its Table 4): "ne" as PART becomes advmod:neg, as CCONJ becomes cc, a source adverb tagged for location becomes advmod:lmod, and modals become aux. Each rule is a row with a count of firings, so the effect of each is visible.
• I2. Transfer from a related variety: when Mova's English parser is trained with a small Early Modern treebank from the languages/en-x-emode texts, report the LAS on held-out Early Modern sentences with and without it. The paper's result is that closely related languages help most.
• I3. Relative clauses without a marker are the paper's main error source. Count attachment errors in the Early Modern sample by whether a relative marker is present, and report the two rates.

For Mova

Chapters: 4. Verb endings: -est, -st, -eth, -th and 06-strong-verbs.md (the verb forms that the paper's errors come from), 8. Word order and phrase structure, and 25. Relative clauses for relative clauses. Code: en/src/parse.rs (the perceptron parser), en parse-eval for LAS, en/src/emode.rs for the closed forms and the post-parse fix. Measures: LAS and label accuracy before and after each conversion rule; the relative-clause error rate by marker presence; run twice to check determinism.

Topics: Treebanks

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.18653/v1/2023.sigtyp-1.4

@inproceedings{villa2023using,
  title = {{Using Modern Languages to Parse Ancient Ones: a Test on Old English}},
  author = {Villa, Luca Brigada and Giarda, Martina},
  booktitle = {Proceedings of the 5th Workshop on Research in Computational Linguistic Typology and Multilingual NLP},
  year = {2023},
  doi = {10.18653/v1/2023.sigtyp-1.4},
  url = {https://doi.org/10.18653/v1/2023.sigtyp-1.4}
}

References

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

2. Ika Alfina, Daniel Zeman, Arawinda Dinakaramani, Indra Budi, and Heru Suhartanto. 2020. Selecting the ud v2 morphological features for indonesian dependency treebank. In 2020 International Conference on Asian Language Processing (IALP), pages 104–109. find
3. Ailsa Meechan-Maddon and Joakim Nivre. 2019. How to parse low-resource languages: Cross-lingual parsing, target language annotation, or both? In Proceedings of the Fifth International Conference on Dependency Linguistics (Depling, SyntaxFest 2019), pages 112–120, Paris, France. Association for Computational Linguistics. find
4. Oscar Bandle, Kurt Braunmüller, Ernst Hakon Jahr, Allan Karker, Hans-Peter Naumann, Ulf Tele-mann, Lennart Elmevik, and Gun Widmark. 2005. The Nordic Languages, volume 2. De Gruyter, Berlin,Boston. find
5. Bruce Mitchell and Fred C. Robinson. 2012. A guide to Old English. Eighth edition. John Wiley & Sons, Malden, Oxford. find
6. James E. Cross and Thomas D. Hill. 1982. The Prose Solomon and Saturn and Adrian and Ritheus. University of Toronto Press, Toronto. find
7. Rafał Molencki. 2017. Syntax. In Laurel J. Brinton and Alexander Bergs, editors, The History of English. Old English, volume 2, chapter 5, pages 100–124. De Gruyter Mouton, Berlin, Boston. find
8. Miryam de Lhoneux, Yan Shao, Ali Basirat, Eliyahu Kiperwasser, Sara Stymne, Yoav Goldberg, and Joakim Nivre. 2017a. From raw text to Universal Dependencies - look, no tags! In Proceedings of the CoNLL 2017 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 207–217, Vancouver, Canada. Association for Computational Linguistics. find our card
9. Kadri Muischnek, Kaili Müürisep, Tiina Puolakainen, Eleri Aedmaa, Riin Kirt, and Dage Särg. 2014. Estonian Dependency Treebank and its annotation scheme. In Proceedings of the 13th Workshop on Treebanks and Linguistic Theories (TLT13), pages 285–291, Tübingen, Germany. find
10. Miryam de Lhoneux, Sara Stymne, and Joakim Nivre. 2017b. Arc-hybrid non-projective dependency parsing with a static-dynamic oracle. In Proceedings of the The 15th International Conference on Parsing Technologies (IWPT)., Pisa, Italy. find
11. Joakim Nivre and Beata Megyesi. 2007. Bootstrapping a swedish treebank using cross-corpus harmonization and annotation projection. In Proceedings of the 6th international workshop on treebanks and linguistic theories, pages 97–102. Association for Computational Linguistics Pennsylvania, PA. find
12. Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre, and Daniel Zeman. 2021. Universal Dependencies. Computational Linguistics, 47(2):255–308. find our card
13. John C. Pope. 1968. Homilies of Ælfric: a Supplementary Collection. Early English Society, Oxford University Press, London. find
14. Hubert Haider. 2010. The Syntax of German. Cambridge University Press, Cambridge. find
15. Peng Qi, Yuhao Zhang, Yuhui Zhang, Jason Bolton, and Christopher D. Manning. 2020. Stanza: A Python natural language processing toolkit for many human languages. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations. ↗ our card
16. Dag T. T. Haug and Marius L. Jøhndal. 2008. Creating a parallel treebank of the old indo-european bible translations. In Proceedings of the Second Workshop on Language Technology for Cultural Heritage Data (LaTeCH 2008), pages 27–34. find
17. Milan Straka. 2018. UDPipe 2.0 prototype at CoNLL 2018 UD shared task. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 197–207, Brussels, Belgium. Association for Computational Linguistics. find our card
18. Antonia Karamolegkou and Sara Stymne. 2021. Investigation of transfer languages for parsing Latin: Italic branch vs. Hellenic branch. In Proceedings of the 23rd Nordic Conference on Computational Linguistics (NoDaLiDa), pages 315–320, Reykjavik, Iceland (Online). Linköping University Electronic Press, Sweden. find
19. Ferdinand von Mengden. 2017a. Morphology. In Laurel J. Brinton and Alexander Bergs, editors, The History of English. Old English, volume 2, chapter 5, pages 73–99. De Gruyter Mouton, Berlin, Boston. find
20. Eliyahu Kiperwasser and Yoav Goldberg. 2016. Simple and accurate dependency parsing using bidirectional LSTM feature representations. TACL, 4:313–327. find our card
21. Ferdinand von Mengden. 2017b. Old english: Overview. In Laurel J. Brinton and Alexander Bergs, editors, The History of English. Old English, volume 2, chapter 3, pages 32–49. De Gruyter Mouton, Berlin, Boston. find
22. Ryan McDonald, Joakim Nivre, Yvonne Quirmbach-Brundage, Yoav Goldberg, Dipanjan Das, Kuzman Ganchev, Keith Hall, Slav Petrov, Hao Zhang, Oscar Täckström, Claudia Bedini, Núria Bertomeu Castelló, and Jungmee Lee. 2013. Universal Dependency annotation for multilingual parsing. In Proceedings of the 51st Annual Meeting of the Association for find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/villa-2023-modern-languages-parse-ancient-ones/