Cross-Genre Learning for Old English Poetry POS Tagging

What it is about

Old English poetry differs from Old English prose in its syntax and its tag distribution, and a tagger trained on one genre loses accuracy on the other. The authors take two annotated corpora, one of prose (YCOE) and one of poetry (YCOEP, 6,299 sentences), and tag both with a detailed set of over 200 tags and with a mapping to 17 UPOS tags. A tagger trained on prose only and tested on poetry falls from 90% to 76% accuracy on the detailed tag set, and its F1 drops from 70% to 46%. Adding more prose does not close the gap, but a small amount of poetry helps more than none. The main finding is that when the genres diverge, more support data from the other genre does not necessarily raise accuracy.

Ideas

• I1. en/src/ptag.rs: give the averaged perceptron a genre feature taken from en/src/ctx.rs (genre of the document). Measure tag accuracy by genre on EWT test and on the GUM test set, with GUM used for test only.
• I2. Before adding any silver training data to the tagger, run the paper's test: train on a large support genre, then a small target sample, and plot accuracy against the amount of each. Report it with en tag-eval split by genre, so that more data is not added by default.
• I3. Learn the archaic forms of 09. Early Modern and archaic forms (thou, hast, doth) from a few sentences of the Gutenberg tales in tools/gutenberg; measure on held-out tale sentences whether 20 to 60 target sentences beat zero, as the paper finds for poetry.

For Mova

The chapter is 09. Early Modern and archaic forms, with 10. Word classes for the tag set. The code is en/src/ptag.rs and en/src/ctx.rs, and the command is en tag-eval. The paper's corpora are Old English, not modern English, so no number from them is a Mova number. Measure first on EWT test, with a two-fold check (en folds).

Topics: Part-of-speech tagging

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.acl-srw.46

@inproceedings{miani2025cross,
  title = {{Cross-Genre Learning for Old English Poetry POS Tagging}},
  author = {Miani, Irene and Stymne, Sara and Darwin, Gregory R.},
  booktitle = {Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop)},
  year = {2025},
  doi = {10.18653/v1/2025.acl-srw.46},
  url = {https://doi.org/10.18653/v1/2025.acl-srw.46}
}

References

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

1. Hirona Jacqueline Arai. 2021. Optimizing an automatic part of speech tagger for poetry text using data augmentation. Undergraduate thesis, Middlebury College, Computer Science Department. find
2. Kristin Bech and Kristine Eide. 2014. The iswoc corpus. Department of Literature, Area Studies and European Languages, University of Oslo. Accessed: 2025-04-22. find
3. Steven Bird, Edward Loper, and Ewan Klein. 2009. Natural Language Processing with Python. O’Reilly Media Inc., Sebastopol, CA. find
4. Marie-Catherine de Marneffe, Christopher D. Manning, Joakim Nivre, and Daniel Zeman. 2021. Universal dependencies. Computational Linguistics, 47(2):255–308. find our card
5. Kyle P. Johnson, Patrick J. Burns, John Stewart, Todd Cook, Clément Besnier, and William J. B. Mattingly. 2021. The Classical Language Toolkit: An NLP framework for pre-modern languages. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing: System Demonstrations, pages 20–29, Online. Association for Computational Linguistics. find
6. Yi Yang and Jacob Eisenstein. 2016. Part-of-speech tagging for historical english. In Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT 2016). find our card
7. Raoul Karimov. 2018. Combined machine-learning approach to pos-tagging of middle english corpora. Crossroads. A Journal of English Studies, 21:42–52. find
8. Bruce Mitchell. 1985. Old English Syntax, volume 1. Clarendon Press, Oxford. find
9. Taesun Moon and Jason Baldridge. 2007. Part-of-speech tagging for middle english through alignment and projection of parallel diachronic texts. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 390–399. Association for Computational Linguistics. find our card
10. Khalil Hassan Nofal. 2011. Syntactic aspects of poetry: A pragmatic perspective. The Buckingham Journal of Language and Linguistics, 4. find
11. Paul Rayson, Dawn Archer, Alistar Baron, Jonathan Culpeper, and Nicholas Smith. 2007. Tagging the bard: Evaluating the accuracy of a modern pos tagger on early modern english corpora. In Proceedings of the Corpus Linguistics Conference: CL2007. UCREL, University of Birmingham. find
12. Sara Stymne, Miryam de Lhoneux, Aaron Smith, and Joakim Nivre. 2018. Parser training with heterogeneous treebanks. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 619– 625, Melbourne, Australia. Association for Computational Linguistics. find
13. University of Oxford. 2001. The york-helsinki parsed corpus of old english poetry (YCOEP). Oxford Text Archive. find
14. University of Oxford. 2003. The york-toronto-helsinki parsed corpus of old english prose (YCOE). Oxford Text Archive. find
15. Rob van der Goot, Ahmet Üstün, Alan Ramponi, Ibrahim Sharaf, and Barbara Plank. 2021. Massive choice, ample tasks (MaChAmp): A toolkit for multi-task learning in NLP. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, pages 176–197, Online. Association for Computational Linguistics. ↗
16. Elly van Gelderen. 2014. A History of the English Language. John Benjamins Publishing Company, Amsterdam. Casalini ID: 5001619. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2025-miani-cross-genre-learning-old-english/