Normalizing Early English Letters to Present-day English Spelling

What it is about

Early English personal letters, from the 15th to the 19th century, are full of non-standard spellings. The paper normalises them to present-day spelling with five methods: VARD2-style rewrite rules, contextual edit distance, statistical and neural machine translation (with and without period information), and a weighted vote. Each method alone gets a minority of the words right; the best single method is the neural model. The generic test set reaches 67% when any one method is correct. The vote weighted by each method's accuracy gives 45% on the generic set, 56% on the 15th-century set and 22% on the 18th-century set, while a Markov chain over the candidates gives 36%, 40% and 24%. The authors conclude that the period, which they used as a variable, is not a good enough grouping for spelling variation, and that the methods make complementary errors.

Ideas

• I1. languages/en-x-emode/02-spelling-and-tokens.md: write the rewrite rules as an ordered table in a deterministic normaliser, and combine them with a dictionary lookup in en/src/emode.rs. Weight each rule by its accuracy on dev data, as the paper's weighted vote does. Measure exact match on held-out letters, run twice.
• I2. Report the "at least one method correct" figure as an upper bound before any change, stratified by period. It shows how much a better selector could gain.
• I3. Negative control: shuffle the period labels in the period-aware table and check that the score falls. The paper's period-aware SMT did not help, so the control is worth running.

For Mova

The chapters are languages/en-x-emode/02-spelling-and-tokens.md and languages/en-x-emode/13-improvements.md. The code is en/src/emode.rs (the closed table of Early Modern forms) and en/src/morph.rs. Measure: exact-match and lemma accuracy on a held-out letter sample, with the vote weighted and unweighted.

Cite

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

@inproceedings{hamalainen2018normalizing,
  title = {{Normalizing Early English Letters to Present-day English Spelling}},
  author = {Hämäläinen, Mika and Säily, Tanja and Rueter, Jack and Tiedemann, Jörg and Mäkelä, Eetu},
  booktitle = {Proceedings of the Second Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature},
  year = {2018},
  url = {https://aclanthology.org/W18-4510/}
}

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. Marilisa Amoia and Jose Manuel Martinez. 2013. Using comparable collections of historical texts for building a diachronic dictionary for spelling normalization. In Proceedings of the 7th workshop on language technology for cultural heritage, social sciences, and humanities, pages 84–89. find
2. Alistair Baron and Paul Rayson. 2008. VARD2: a tool for dealing with spelling variation in historical corpora. In Postgraduate Conference in Corpus Linguistics. Aston University, Birmingham. find
3. Kenneth R. Beesley and Lauri Karttunen, 2003. Finite-State Morphology, pages 451–454. Stanford, CA: CSLI Publications. find
4. Steven Bird, Ewan Klein, and Edward Loper. 2009. Natural Language Processing with Python. O’Reilly Media. find
5. Marcel Bollmann and Anders Søgaard. 2016. Improving historical spelling normalization with bi-directional LSTMs and multi-task learning. CoRR, abs/1610.07844. ↗ our card
6. Philip R Burns. 2013. MorphAdorner v2: A Java library for the morphological adornment of English language texts. Northwestern University, Evanston, IL. find
7. Kenneth Heafield, Ivan Pouzyrevsky, Jonathan H. Clark, and Philipp Koehn. 2013. Scalable Modified Kneser-Ney Language Model Estimation. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 690–696, Sofia, Bulgaria, August. find our card
8. Guillaume Klein, Yoon Kim, Yuntian Deng, Jean Senellart, and Alexander M. Rush. 2017. OpenNMT: Open-Source Toolkit for Neural Machine Translation. In Proc. ACL. find our card
9. Philipp Koehn, Hieu Hoang, Alexandra Birch, Chris Callison-Burch, Marcello Federico, Nicola Bertoldi, Brooke Cowan, Wade Shen, Christine Moran, Richard Zens, et al. 2007. Moses: Open source toolkit for statistical machine translation. In Proceedings of the 45th annual meeting of the ACL on interactive poster and demonstration sessions, pages 177–180. Association for Computational Linguistics. find
10. Krister Lindén, Erik Axelson, Senka Drobac, Sam Hardwick, Juha Kuokkala, Jyrki Niemi, Tommi A. Pirinen, and Miikka Silfverberg. 2013. HFST — A System for Creating NLP Tools. In Cerstin Mahlow and Michael Piotrowski, editors, Systems and Frameworks for Computational Morphology, pages 53–71, Berlin, Heidelberg. Springer Berlin Heidelberg. find
11. MED. n.d. Middle English Dictionary. University of Michigan. https://quod.lib.umich.edu/m/med/. Terttu Nevalainen, Helena Raumolin-Brunberg, Jukka Keränen, Minna Nevala, Arja Nurmi, Minna Palander-Collin, Samuli Kaislaniemi, Mikko Laitinen, Tanja Säily, and Anni Sairio. 1998–2006. CEEC, Corpora of Early English Correspondence. Department of Modern Languages, University of Helsinki. http://www.helsinki.fi/varieng/CoRD/corpora/CEEC/. ↗
12. Franz Josef Och and Hermann Ney. 2003. A Systematic Comparison of Various Statistical Alignment Models. Computational Linguistics, 29(1):19–51. find
14. Eva Pettersson, Beáta Megyesi, and Jörg Tiedemann. 2013. An SMT approach to automatic annotation of historical text. In Proceedings of the workshop on computational historical linguistics at NODALIDA 2013; May 22-24; 2013; Oslo; Norway. NEALT Proceedings Series 18, number 087, pages 54–69. Linköping University Electronic Press. find
15. Tanja Säily. 2018. Test set for https://doi.org/10.5281/zenodo.1300332. Normalization of Historical English in CEEC, June. ↗
16. Tanja Samardzic, Yves Scherrer, and Elvira Glaser, 2015. Normalising orthographic and dialectal variants for the automatic processing of Swiss German. Proceedings of the 7th Language and Technology Conference. ID: unige:82397. find
17. The BNC Consortium. 2007. The British National Corpus, version 3 (BNC XML Edition). Distributed by Bodleian Libraries, University of Oxford, on behalf of the BNC Consortium. http://www.natcorp.ox.ac.uk/. ↗