Elamur Research⌕
☰
home / papers / 2023 / Advances in the Automatic Lemmatization of Old English: Class IV Strong Verbs (L-Y)

Advances in the Automatic Lemmatization of Old English: Class IV Strong Verbs (L-Y)

Paper · 2023 · Roberto Torre

Summary

Old English is highly inflected, has no spelling standard, and lacks a lemmatized corpus. Standard data-driven lemmatizers and taggers therefore cannot be applied, and NLP work on the language stalls. Torre addresses one slice of this problem: class IV strong verbs whose lemmas begin with L through Y. Instead of learning from annotated data, he builds a rule-based lemmatizer within the Morphological Generation framework. Starting from each lemma, a set of algorithms produces the inflected forms, stem and ending variants caused by morphophonology and dialect, and derived (prefixed) counterparts. Each generated form is then checked against the YCOE (Taylor et al. 2003) and the DOEC (Healey et al. 2004). Forms found there count as attested and inherit the lemma that produced them. The process works on types only; disambiguating tokens in context is left for a later step. The result is a set of dependable form–lemma links plus explicitly flagged zones of ambiguity and mismatch for manual review. This matters for OE lexicography, where the DOE has so far reached only the letter I.

Key points

  • Scope: OE class IV strong verbs, lemmas L–Y. The pipeline generates forms, validates them against corpora, and assigns lemmas, all at the type level.
  • The generation algorithms model inflection, derivation, morphophonological change, and diatopic (dialectal) variation.
  • The reference corpora are large but unlemmatized: DOEC has about 3,000,000 words and YCOE about 1,500,000. Among the smaller ones are the Helsinki Corpus (300,000) and York-Helsinki's parsed collection of OE verse (70,000).
  • The work belongs to the Nerthus project. Its lemmatized ParCorOEv2 holds about 110,000 records, with a planned 300,000, and draws on the Knowledge Base of OE (KBOE).
  • The outcome is successful form–lemma associations for attested forms, alongside an inventory of ambiguous cases and mismatches.

Interesting ideas

  • I1. Generate-then-attest: rules expand every lemma into its full paradigm, and only forms that actually occur in a corpus are kept, so each surviving form carries its lemma by construction and overgeneration is filtered out cheaply.
  • I2. Separating type-level lemmatization from token-level disambiguation lets a lexicon offer all candidate lemmas first, while choosing among them is deferred to a context-aware step.
  • I3. Variation is handled as separate composable layers (inflection, sound/spelling change, dialect, derivation), and derived forms are built from the simplex paradigm rather than listed independently.
  • I4. Ambiguous and mismatched outputs are treated as a deliverable, a list of cases flagged for human review, rather than as noise to hide.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (tokens): Build the irregular-form and lemma tables in crate en by generating paradigms and keeping forms seen in treebank text, so every lemma can be traced to the rule that produced it.
  • I2 (tokens): Keep a type lexicon listing every candidate lemma for a surface form, and let the tagger or the UD tree pick one, so ambiguity stays visible rather than being silently resolved.
  • I3 (tokens): Derive prefixed verbs such as re- or out- forms from their base paradigms, and treat British/American spelling as its own variant layer instead of storing duplicates.
  • I4 (learn): Route generated forms that match several lemmas, or none, into a review queue, and reuse confirmed failures as negative controls when evaluating the lemmatizer.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Universidad de La RiojaLogroño · Spain24 works · 125 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksLiterature and Literary Theory · 519 works

🧭 Topics 3Research subjects and languages the work is about.English · 2,263 worksMorphology · 405 worksHistorical linguistics & language change · 249 works

Cite

Cite as Roberto Torre. 2023. Advances in the Automatic Lemmatization of Old English: Class IV Strong Verbs (L-Y). Atlantis 45.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.28914/atlantis-2023-45.2.02

@inproceedings{torre2023advances,
  title = {{Advances in the Automatic Lemmatization of Old English: Class IV Strong Verbs (L-Y)}},
  author = {Roberto Torre},
  booktitle = {Atlantis 45},
  year = {2023},
  doi = {10.28914/atlantis-2023-45.2.02},
  url = {https://doi.org/10.28914/atlantis-2023-45.2.02}
}

References

47 references, 2 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bauer, Renate and Ulrike Krischke, eds. 2011. More than Words: English Lexicography and Lexicology Past and Present. Bern: Peter Lang. find
  2. Bosworth, Joseph and Thomas N. Toller. (1898) 1973. An Anglo-Saxon Dictionary. Oxford: Oxford UP. find
  3. Brunner, Karl. 1965. Altenglische Grammatik, nach der Angelsachsischen Grammatik von Eduard Sievers. Tübingen: Max Niemeyer. find
  4. Campbell, Alistair. (1959) 1987. Old English Grammar. Oxford: Oxford UP. find
  5. Chambers, Raymond Wilson, Max Förster and Robin Flower, eds. 1933. The Exeter Book of Old English Poetry: Facsimile. London: Humphries. find
  6. Clark Hall, John Richard. (1894) 1996. A Concise Anglo-Saxon Dictionary. Supplement by Herbert D. Merritt. Cambridge: Cambridge UP. find
  7. Ferrés Daniel, Ahmed AbuRa’ed and Horacio Saggion. 2017. “Spanish Morphological Generation with Wide-Coverage Lexicons and Decision Trees.” Procesamiento del Lenguaje Natural 58: 109-16. find
  8. Forcada, Mikel et al. 2011. “Apertium: a Free/Open-Source Platform for Translation.” Machine Translation 25 (2): 127-44. find
  9. Förster, Max. 1933. “The Preliminary Matter of the Exeter Book.” In Chambers, Förster and Flower 1933, 44-54. 7 This research has been conducted with the support of Grant PID2020-119200GB-100 funded by MCIN/ AEI/10.13039/501100011033. ↗
  10. García Fernández, Laura. 2020. Lemmatising Old English on a Relational Database. Preterite-Present, Contracted, Anomalous and Strong VII Verbs. Munich: Utzverlag. ↗
  11. Grundy, George B. 1919. “The Saxon Land Charters of Wiltshire.” The Archaelogical Journal 76: 143-301. ↗
  12. Healey, Antonette di Paolo, ed. 2018. The Dictionary of Old English: A to I. Toronto: Centre for Medieval Studies, University of Toronto. —, John Price Wilkin and Xin Xiang, eds. 2004. The Dictionary of Old English Web Corpus. Toronto: Centre for Medieval Studies, University of Toronto. find
  13. Heidermanns, Frank. 1993. Etymologisches Wörterbuch der germanischen Priäradjective. Berlin: De Gruyter. find
  14. Hogg, Richard, ed. 1992. The Cambridge History of the English Language I: The Beginnings to 1066. Cambridge: Cambridge UP. — and Robert D. Fulk. 2011. A Grammar of Old English. Chichester: Wiley-Blackwell. find
  15. Kastovsky, Dieter. 1992. “Semantics and Vocabulary.” In Hogg 1992, 290-408. ↗
  16. Khemakhem, Aida et al. 2015. “ISO Standard Modeling of a Large Arabic Dictionary.” Journal of Natural Language Engineering 22 (6): 849-79. ↗
  17. Krygier, Marcin. 1994. The Disintegration of the English Strong Verb System. Bern: Peter Lang. find
  18. LangScape Project. 2008. “The Language of Landscape: Reading the Anglo-Saxon Countryside.” Version 0.9. [Accessed online on 2 August, 2021]. find
  19. Levin, Samuel. 1964. “A Reclassification of the Old English Strong Verbs.” Language 40 (2): 156-61. ↗
  20. Mailhammer, Robert. 2007. The Germanic Strong Verbs: Foundation and Development of a New System. Trends in Linguistics. Studies and Monographs 183. Berlin: De Gruyter. find
  21. Márquez, Lluis, Chris Callison-Burch and Jian Su, eds. 2015. Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing. Lisbon: Association for Computational Linguistics. find
  22. Martín Arista, Javier. 2012. “Lexical Database, Derivational Map and 3D Representation.” RESLA-Revista Española de Lingüística Aplicada (Extra 1): 119-44. —. “Nerthus. Lexical Database of Old English: From Word-Formation to Meaning Construction.” Research Seminar given at the School of English, Sheffield, 2013. find
  23. Martín Arista, Javier. “The Nerthus Project at the Crossroads: From Lexical Database to Parallel Corpus of Old English.” Lecture delivered at the 2017 International Conference of SELIM. find
  24. Málaga, September 2017. — et al., comp. 2021. ParCorOEv2. An Open Access Annotated Parallel Corpus Old English-English. Nerthus Project, Universidad de La Rioja. find
  25. Metola Rodríguez, Darío. 2015. “Lemmatisation of Old English Strong Verbs on a Lexical Database.” PhD diss., University of La Rioja. —. 2017. “Strong Verb Lemmas from a Corpus of Old English: Advances and Issues.” Revista de Lingüística y Lenguas Aplicadas 12: 65-76. find
  26. Müller, Thomas et al. 2015. “Joint Lemmatization and Morphological Tagging with Lemming.” In Márquez, Callison-Burch and Su 2015, 2264-74.
  27. Myers, Louis M. 1966. The Roots of Modern English. Boston: Little Brown. find
  28. Novo Urraca, Carmen and Ana Elvira Ojanguren López. 2018. “Lemmatising Treebanks: Corpus Annotation with Knowledge Bases.” RAEL-Revista Electrónica de Lingüística Aplicada 17 (1): 99-120. find
  29. Oflazer, Kemal and Murat Saraçlar, eds. 2018. Turkish Natural language Processing. Berlin: Springer. find
  30. Orel, Vladimir. 2003. A Handbook of Germanic Etymology. Leiden: Brill. find
  31. Pintzuk, Susan and Leendert Plug. 2001. The York-Helsinki Parsed Corpus of Old English Poetry. Department of Language and Linguistic Science, University of York. find
  32. Reiter, Ehud and Robert Dale. 1997. “Building Applied Natural Language Generation Systems.” Natural Language Engineering 3 (1): 57-87. ↗
  33. Rissanen, Matti et al., comp. 1991. The Helsinki Corpus of English Texts. Department of English, University of Helsinki. find
  34. Robertson, Agnes J. 1925. The Laws of the Kings of England from Edmund to Henry I. Cambridge: Cambridge UP. find
  35. Saggion, Horacio. 2017. Automatic Text Simplification. Berlin: Springer. ↗
  36. Seebold, Elmar. 1970. Vergleichendes und etymologisches Wörterbuch der germanischen starken Verben. Berlin: De Gruyter. ↗
  37. Sweet, Henry. (1896) 1976. The Student’s Dictionary of Anglo-Saxon. Cambridge: Cambridge UP. find
  38. Tapsai, Chalermpol, Herwig Unger and Phayung Meesad. 2021. Thai Natural Language Processing. Berlin: Springer. ↗
  39. Taylor, Ann et al. 2003. The York-Toronto-Helsinki Parsed Corpus of Old English Prose. York: U of York. find
  40. The Holy Bible. (1899) 1971. Translated from the Latin Vulgate (Douay Rheims Version). Tan Books. find
  41. Thorpe, Benjamin. 1861. The Anglo-Saxon Chronicle. Vol 2. London: Longman, Green and Roberts. find
  42. Tío Sáenz, Marta. 2019. “The Lemmatisation of Old English Weak Verbs of a Relational Database.” PhD diss., University of La Rioja. find
  43. Von Mengden, Ferdinand. 2011. “Ablaut or Transfixation? On the Old English Strong Verbs.” In Bauer and Krischke 2011, 123-39. find
  44. Wang, Alex et al. “GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding.” Paper delivered at the 2019 ICLR Conference, New Orleans, May 2019. ↗
  45. Zhang, Yu and Qiang Yang. 2018. “An Overview of Multi-Task Learning.” National Science Review 5: 30-43. ↗
  46. Received 27 August 2021 Roberto Torre Alonso is a Lecturer at the University of La Rioja, Spain. He is a member of the Functional Grammars Research Group at the same university and is currently participating in the development of ParCorOEv2. An Open Access Annotated Parallel Corpus Old English-English. His research interests include Old English morphology, the semantic-syntactic interrelation of Old English verbs and Old English verb lemmatization. find
  47. Revised version accepted 3 October 2022 find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/torre-2023-advances-automatic-lemmatization-old-english/