Elamur Research⌕
☰
home / papers / 2023 / A Stochastic Analysis of the Linguistic Provenance of English Place Names

A Stochastic Analysis of the Linguistic Provenance of English Place Names

Paper · 2023 · Michael Dalvean

Summary

Etymologists often explain an English place name by matching its parts to landscape words, personal names or settlement terms in some influencing language. The trouble is choosing which language to start from, because equally plausible readings can come from different starting languages. Dalvean does not attempt to recover meanings. Instead he estimates each name's likely linguistic source. He sets 18,799 English place names against 84,687 names drawn from ten comparison sources: the Celtic regions of Wales, Scotland and Ireland; the Scandinavian countries of Sweden, Norway and Denmark; the continental neighbours France, Germany and the Netherlands; and Ancient Rome. Every name becomes 263 mostly "atomic" features, such as which letter sits at each position from the end, plus ratios like vowels per length. Ten separate England-versus-country classifiers then score how English each name looks against each rival set. The scores rank the English names. Harlington comes out as the most typically English and Anna as the least. About 97% of names look more English than foreign. The method gives etymologists an objective first step that can suggest a source language or rule one out.

Key points

  • The data are 18,799 English place names and 84,687 names from ten European comparison sets, one of which is Ancient Roman.
  • Names are broken into 263 mainly letter-position features. The model is never told about known elements like "-ton". It nevertheless learns that names ending in n, o, t (read from the end) are strongly English.
  • Harlington ranks as the most archetypally English name and Anna as the least. Roughly 97% of English names resemble English names more than foreign ones. A few hundred names resist classification as English; the paper gives 475 in one place and 471 in another.
  • Classifier accuracy serves as a distance measure. The Roman classifier is the most accurate, so Roman names are the most unlike English ones. The Scottish classifier is the least accurate, so Scottish names are the closest.
  • Correlating scores across classifiers gives further findings:
    • Norwegian names look most like the other nine sets and Welsh names least.
    • Roman scores correlate only modestly with the rest, averaging 0.44. The figure is 0.36 with Germany, 0.37 with Wales and 0.45 with Scotland, but the link with France is high.
    • The author reads this as conquest shaping place names more than cultural prestige does.

Interesting ideas

  • I1. The model scores letters by their position counted from the word's end, each independently of the others, instead of using hand-written suffix rules. This finds known endings and also uncovers patterns no one listed, such as vowel density signalling Roman origin.
  • I2. Each name is judged with a set of pairwise "home versus rival" classifiers instead of one multi-class model. The result is a readable profile per item showing which alternative it resembles.
  • I3. The paper treats how easily two classes can be told apart, measured as classifier accuracy, as a proxy for how different they are.
  • I4. Correlating the score outputs of different classifiers shows which reference groups behave alike, without any extra labels.
  • I5. The system's job is framed as narrowing the hypothesis space, here the language to start from, before expert analysis. It does not try to deliver the final answer.

For Mova

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

  • I1 (tokens): position-from-end letter features could back an explainable guesser for the tag or origin of unknown proper nouns, where each decision traces to specific letters.
  • I2 (world): named entities could carry a small origin profile, such as likely language or region, that feeds entity typing and stays inspectable rather than opaque.
  • I3 (learn): the same trick could measure how distinct treebank dialects or text genres are before conversion. Low separability would signal that merging them is safe.
  • I4 (learn): correlating the errors or confidence scores of witness parsers could reveal which witnesses are redundant when building silver data.
  • I5 (world): ranked priors over candidate sources could steer lookups into WordNet, Wikidata or the knowledge graph, with the ranking itself kept as an explanation.

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

🎓The University of Western AustraliaPerth · Australia8 works · 1,707 citations

 Fields, topics and concepts

📚 Shelves 3The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksLiterature and Literary Theory · 519 works

🧭 Topics 1Research subjects and languages the work is about.English · 2,263 works

Cite

Cite as Michael Dalvean. 2023. A Stochastic Analysis of the Linguistic Provenance of English Place Names. arXiv:2312.12850.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2312.12850 · arXiv 2312.12850

@misc{dalvean2023stochastic,
  title = {{A Stochastic Analysis of the Linguistic Provenance of English Place Names}},
  author = {Michael Dalvean},
  year = {2023},
  doi = {10.48550/arxiv.2312.12850},
  eprint = {2312.12850},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2312.12850}
}

References

25 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Jean-Gabriel Gigot. Notes sur le toponyme «londres»(hérault). in memoriam paul marres (oct. 1974). Revue internationale d’onomastique, 26(4):284–292, 1974. find
  2. Richard Coates. A new explanation of the name of London. Transactions of the Philological Society, 96(2):203–229, 1998. ↗
  3. David Mills. A dictionary of British place-names. Oxford University Press, 2011. ↗
  4. Margaret Scott, Andrew Clark, et al. Directions in English place-name studies: An invitation to debate, with a case study of Salford Quays. Nomina, 34:27–50, 2011. find
  5. Carole Hough. Place-name evidence relating to the interpretation of Old English legal terminology. Leeds Studies in English, pages 19–48, 1996. find
  6. Oliver JT O’grady. Judicial assembly sites in Scotland: archaeological and place-name evidence of the Scottish court hill. Medieval Archaeology, 58(1):104–135, 2014. English Place Names ↗
  7. Katherine ME Cuthbertson. The use of name evidence in lexicography: comparative analysis of onomastic data for historical and contemporary Scots. PhD thesis, University of Glasgow, 2018. find
  8. Tamás Fekete. Anglo-Scandinavian code-mixing in English place-names. 2015. ↗
  9. Ivre Marjorie R Machado, Rafael Odon de Alencar, Roberto de Oliveira Campos, and Clodoveu A Davis. An ontological gazetteer and its application for place name disambiguation in text. Journal of the Brazilian Computer Society, 17(4):267–279, 2011. ↗
  10. Cillian Berragan, Alex Singleton, Alessia Calafiore, and Jeremy Morley. Transformer based named entity recognition for place name extraction from unstructured text. International Journal of Geographical Information Science, pages 1–20, 2022. ↗
  11. Xuke Hu, Hussein S Al-Olimat, Jens Kersten, Matti Wiegmann, Friederike Klan, Yeran Sun, and Hongchao Fan. Gazpne: annotation-free deep learning for place name extraction from microblogs leveraging gazetteer and synthetic data by rules. International Journal of Geographical Information Science, 36(2):310–337, 2022. ↗
  12. Imene Bensalem and Mohamed K Kholladi. Toponym disambiguation by arborescent relationships. Journal of Computer Science, 6(6):653, 2010. ↗
  13. Mark J Nigrini and Linda J Mittermaier. The use of Benford’s law as an aid in analytical procedures. Auditing, 16(2):52, 1997. ↗
  14. Xiaoyong Yan, Seong-Gyu Yang, Beom Jun Kim, and Petter Minnhagen. Benford’s law and first letter of words. Physica A: Statistical Mechanics and its Applications, 512:305–315, 2018. ↗
  15. Jayne Carroll. Identifying migrants in medieval England: the possibilities and limitations of place-name evidence. Oxford University Press (OUP), 2020. ↗
  16. Open Geography portal. Index of Place names in Great Britain (September 2019). https://geoportal.statistics.gov.uk/datasets/ index-of-place-names-in-great-britain-september-2019/. Last accessed on October 27, 2022. ↗
  17. Insee. Code officiel géographique au 1er janvier 2022téléchargement des fichiers. https://www.insee.fr/fr/information/6051727/. Last accessed on October 27, 2022. ↗
  18. Statistisches Bundesamt. Gemeindeverzeichnis. https://www.destatis.d e /E N /T h e m e s /C o u n t r i e s- R e g i o n s /R e g i o n a l - S t a t i s t i c s/ OnlineListMunicipalities/. Last accessed on October 27, 2022. English Place Names ↗
  19. Kobenhavns Universitet. Autoriserede stednavne i Danmark. https://www.stednavneudvalget.ku.dk/autoriserede_stednavne/. Last accessed on November 30, 2023. ↗
  20. National Geospatial-Intelligence Agency. Geographic names server. https://geonames.nga.mil/gns/html/namefiles.html/. Last accessed on October 27, 2022. ↗
  21. Pleiades. Data for download. https://pleiades.stoa.org/downloads/. Last accessed on October 27, 2022. ↗
  22. Gustavo EAPA Batista, Ronaldo C Prati, and Maria Carolina Monard. A study of the behavior of several methods for balancing machine learning training data. ACM SIGKDD explorations newsletter, 6(1):20–29, 2004. ↗
  23. Reyhane Askari Hemmat and Abdelhakim Hafid. SLA violation prediction in cloud computing: A machine learning perspective. arXiv preprint arXiv:1611.10338, 2016. ↗
  24. Institute for Name-Studies at the University of Nottingham. Key to English Place-Names. https://halogen.le.ac.uk/query/kepn/. Last accessed on December 16, 2023. ↗
  25. Frank Merry Stenton. Presidential address: The historical bearing of placename studies: The Danish settlement of Eastern England. Transactions of the Royal Historical Society, 24:1–24, 1942. English Place Names find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/dalvean-2023-stochastic-analysis-linguistic-provenance-english/