Whit's the Richt Pairt o Speech: PoS tagging for Scots

What it is about

Scots is closely related to English and has no annotated data, so an English tagger does not transfer to it well. The authors hand-tag a small Scots set of about 1,040 tokens, from several dialects and domains, and use it both for evaluation and for training. They compare zero-shot transfer from two English treebanks (GUM and EWT) with a Scots-only model and with several ways of combining the two sets: concatenation, mixed training and fine-tuning. Training on the 1,040 Scots tokens alone gives 60.5% accuracy, which is better than zero-shot transfer from much larger English data. Concatenation of English and Scots data is the best method. It improves the GUM zero-shot model by 16.9 to 23.1 points and the EWT zero-shot model by 5.2 to 9.9 points. The typical errors are auxiliaries tagged as adverbs and interjections missed.

Ideas

• I1. en/src/ptag.rs: a small hand-tagged dialect sample (about 1,000 tokens) as a lexical override layer. Learn word-to-tag counts from it and let them win over the English counts, in BTreeMap order. Measure tag accuracy on a held-out part of the sample, run twice.
• I2. The paper's concatenation method, in counts: merge English tag counts with dialect counts with a weight, and report the result per tag (precision and recall), as the paper does, so that the gain is visible.
• I3. The error list is a rule list: modal-like forms (micht, 've) must not get ADV, and sentence-initial conjunctions must not get PROPN. Write these as constraints in en/src/udreg.rs checks and measure how many errors each rule removes.
• I4. Negative control: a dialect set with the tags shuffled inside each sentence must score clearly worse than the true set.

For Mova

The chapters are 10. Word classes (tag sets) and 37. Register and variation. The code is en/src/ptag.rs (the perceptron tagger), en/src/dict.rs and the tag checks in en/src/udreg.rs. Measure per-tag precision and recall on a held-out dialect sample, with an English-only control. The neural model is not the target; the data design and the concatenation idea are.

Topics: Part-of-speech tagging

Cite

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

@inproceedings{lameris2021whit,
  title = {{Whit's the Richt Pairt o Speech: PoS tagging for Scots}},
  author = {Lameris, Harm and Stymne, Sara},
  booktitle = {Proceedings of the Eighth Workshop on NLP for Similar Languages, Varieties and Dialects},
  year = {2021},
  url = {https://aclanthology.org/2021.vardial-1.5/}
}

References

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

1. Željko Agić, Dirk Hovy, and Anders Søgaard. 2015. If all you have is a bit of the Bible: Learning PoS taggers for truly low-resource languages. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 268–272. find
2. Alan Akbik, Duncan Blythe, and Roland Vollgraf. 2018. Contextual string embeddings for sequence labeling. In COLING 2018, 27th International Conference on Computational Linguistics, pages 1638– 1649. find
3. Rami Al-Rfou, Bryan Perozzi, and Steven Skiena. 2013. Polyglot: Distributed word representations for multilingual NLP. In Proceedings of the Seventeenth Conference on Computational Natural Language Learning, pages 183–192, Sofia, Bulgaria. Association for Computational Linguistics. find
4. Alasdair Allan, Andy Eagle, John Law, J Derrick Mc-Clure, George Philp, Iseabail Macleod, Liz Niven, David Purves, and John Tait. 1998. Report & recommends o the Scots spellin comatee. find
5. Jean Anderson, Dave Beavan, and Christian Kay. 2007. Scots: Scottish corpus of texts and speech. In Joan C. Beal, Karen P. Corrigan, and Hermann L. Moisl, editors, Creating and Digitizing Language Corpora: Volume 1: Synchronic Databases, pages 17–34. Palgrave Macmillan UK, London. find
6. Daniel Andor, Chris Alberti, David Weiss, Aliaksei Severyn, Alessandro Presta, Kuzman Ganchev, Slav Petrov, and Michael Collins. 2016. Globally normalized transition-based neural networks. In Proceedings of the 54th Annual Meeting of the Association Armand Joulin, Edouard Grave, Piotr Bojanowski, Matthijs Douze, Hérve Jégou, and Tomas Mikolov. 2016. Fasttext.zip: Compressing text classification models. arXiv preprint arXiv:1612.03651. for Computational Linguistics (Volume 1: Long Papers), pages 2442–2452, Berlin, Germany. Association for Computational Linguistics. ↗
7. Marcel Bollmann. 2013. POS tagging for historical texts with sparse training data. In Proceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse, pages 11–18, Sofia, Bulgaria. Association for Computational Linguistics. find our card
8. Katharina Kann, Johannes Bjerva, Isabelle Augenstein, Barbara Plank, and Anders Søgaard. 2018. Character-level supervision for low-resource PoS tagging. In Proceedings of the Workshop on Deep Learning Approaches for Low-Resource NLP, pages 1–11. find
9. Chenhui Chu and Rui Wang. 2018. A survey of domain adaptation for neural machine translation. In Proceedings of the 27th International Conference on Computational Linguistics, pages 1304–1319. find
10. Billy Kay. 1988. The mither tongue. Grafton, London, United Kingdom. find
11. Thomas Clark, Alistair Heather, James McDonald, Jamie Smith, Elizabeth Thoumire, and Antonia Uri. 2020. Mak Forrit. Blog: https://www.makforrit.scot . ↗
12. Raj Dabre, Atsushi Fujita, and Chenhui Chu. 2019. Exploiting multilingualism through multistage fine-tuning for low-resource neural machine translation. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 1410– 1416. find
13. Joo-Kyung Kim, Young-Bum Kim, Ruhi Sarikaya, and Eric Fosler-Lussier. 2017. Cross-lingual transfer learning for PoS tagging without cross-lingual resources. In Proceedings of the 2017 conference on empirical methods in natural language processing, pages 2832–2838. find
14. Eleanor Lawson, Jane Stuart-Smith, and Lydia Rodger. 2019. A comparison of acoustic and articulatory parameters for the GOOSE vowel across British Isles Englishes. The Journal of the Acoustical Society of America, 146(6):4363–4381. find
15. Viktoriia Lemeshchenko-Lagoda. 2019. What is Scots? In Halyna Matiukha and Maria Karpinska, editors, Novi perspektyvy v angliys’kiy filolohiyi ta navchannya angliys’kiy movi. TOB, Kolor Print, Melitopol. find
16. Amitava Das. 2011. PsychoSentiWordNet. In Proceedings of the ACL 2011 Student Session, pages 52–57, Portland, OR, USA. Association for Computational Linguistics. find
17. Yu-Hsiang Lin, Chian-Yu Chen, Jean Lee, Zirui Li, Yuyan Zhang, Mengzhou Xia, Shruti Rijhwani, Junxian He, Zhisong Zhang, Xuezhe Ma, Antonios Anastasopoulos, Patrick Littell, and Graham Neubig. 2019. Choosing transfer languages for cross-lingual learning. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3125–3135, Florence, Italy. Association for Computational Linguistics. find
18. Meng Fang and Trevor Cohn. 2017. Model transfer for tagging low-resource languages using a bilingual dictionary. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 587–593. find
19. Dan Garrette, Jason Mielens, and Jason Baldridge. 2013. Real-world semi-supervised learning of PoS-taggers for low-resource languages. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 583–592. find
20. Pierre Magistry, Anne-Laure Ligozat, and Sophie Rosset. 2019. Exploiting languages proximity for Part-of-Speech tagging of three French regional languages. Language Resources & Evaluation, 53:865– 888. find
21. Tao Gui, Qi Zhang, Haoran Huang, Minlong Peng, and Xuan-Jing Huang. 2017. Part-of-Speech tagging for twitter with adversarial neural networks. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2411–2420. find
22. 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 (EMNLP-CoNLL), pages 390–399. find our card
23. Erhard Hinrichs, Marie Hinrichs, Sandra Kübler, and Thorsten Trippel. 2019. Language technology for digital humanities: introduction to the special issue. Language Resources & Evaluation, 53:559–563. find
24. Barbara Plank, Anders Søgaard, and Yoav Goldberg. 2016. Multilingual Part-of-Speech tagging with bidirectional long short-term memory models and auxiliary loss. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 412–418, Berlin, Germany. Association for Computational Linguistics. find
25. Anna Jørgensen, Dirk Hovy, and Anders Søgaard. 2016. Learning a PoS tagger for AAVE-like language. In Proceedings of the 2016 conference of the North American chapter of the association for computational linguistics: human language technologies, pages 1115–1120. find our card
26. Mark Sebba. 2019. Named into being? language questions and the politics of Scots in the 2011 census in Scotland. Language Policy, 18(3):339–362. find
27. Philippa Shoemark, James Kirby, and Sharon Goldwater. 2017a. Topic and audience effects on distinctively scottish vocabulary usage in twitter data. In Proceedings of the Workshop on Stylistic Variation, pages 59–68. find
28. Philippa Shoemark, Debnil Sur, Luke Shrimpton, Iain Murray, and Sharon Goldwater. 2017b. Aye or naw, whit dae ye hink? Scottish independence and linguistic identity on social media. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 1, Long Papers, pages 1239–1248, Valencia, Spain. Association for Computational Linguistics. find
29. Natalia Silveira, Timothy Dozat, Marie-Catherine de Marneffe, Samuel Bowman, Miriam Connor, John Bauer, and Christopher D. Manning. 2014. A gold standard dependency corpus for English. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC-2014). find
30. Zhilin Yang, Ruslan Salakhutdinov, and William W Cohen. 2016. Transfer learning for sequence tagging with hierarchical recurrent networks. arXiv preprint arXiv: 1703.06345v1. ↗
31. David Yarowsky, Grace Ngai, and Richard Wicentowski. 2001. Inducing multilingual text analysis tools via robust projection across aligned corpora. Technical report, Johns Hopkins University Baltimore, MD Dept. of Compute Science. find
32. Amir Zeldes. 2017. The GUM corpus: Creating multilayer resources in the classroom. Language Resources and Evaluation, 51(3):581–612. ↗
33. Daniel Zeman, Joakim Nivre, Mitchell Abrams, et al. 2019. Universal dependencies 2.5. LINDAT/CLARIAH-CZ digital library at the Institute of Formal and Applied Linguistics (ÚFAL), Faculty of Mathematics and Physics, Charles University. find
34. Katja Zupan, Nikola Ljubešić, and Tomaž Erjavec. 2019. How to tag non-standard language: Normalisation versus domain adaptation for Slovene historical and user-generated texts. Natural Language Engineering, 25(5):651–674. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/lameris-2021-whit-s-richt-pairt-o/