Elamur Research⌕
☰
home / papers / 2021 / Tracking Semantic Change in Cognate Sets for English and Romance Languages

Tracking Semantic Change in Cognate Sets for English and Romance Languages

Paper · 2021 · Ana Sabina Uban, Alina Maria Cristea, Anca Dinu et al.

Summary

The paper looks at how words with a shared Latin origin have drifted apart in meaning across English and five Romance languages. Diachronic corpora are the usual tool for this. The authors take a different approach: they treat the present-day meanings of cognates in different languages as snapshots of separate histories that grew from one etymon. They started from an existing dictionary-derived list of Romance cognate sets and added English. They then categorised and manually fixed errors from the automatic cognate matching. Each English item was labelled as an old borrowing (arrived via Old French) or a recent borrowing (taken directly from Latin). Cross-lingual word embeddings placed all languages in one vector space, which let the authors measure semantic similarity between cognates alongside frequency and polysemy. The expected effect of borrowing age did not show up in average similarity. However, English similarity scores fell into two distinct clusters, which the authors explain through transmission channels and semantic fields. The curated, etymologically annotated dataset is released as a resource for work on cognates, borrowings and false friends.

Key points

  • The source data is the Ciobanu and Dinu (2014) list of 3,218 Latin-derived cognate sets, each with a member in all five of Romanian, French, Italian, Spanish and Portuguese, extended here with English.
  • English cognates are tagged by period of entry:
    • "old borrowings" came through Old French before roughly the first half of the 15th century;
    • "recent borrowings" came directly from Latin afterwards.
  • Similarity distributions for English words are bimodal. Mean similarity did not support the old-vs-recent hypothesis.
  • Case studies show the effect of the entry channel:
    • powder (via Old French) scores about 0.75 against French but about 0.5 against other Romance languages;
    • camp, borrowed as a military term, scores around 0.1 against Romance "field" words.
  • Concrete referents (animals, materials, seasons) and narrow technical or abstract terms stay highly similar across languages, averaging around 0.75. The authors link this to Hamilton et al.'s laws of conformity and innovation.

Interesting ideas

  • I1. Instead of tracking one word through time, the paper compares related words across languages today, treating each language as an independent "run" of the same word's history. This is useful wherever diachronic data is scarce.
  • I2. The route a word took into a language (for example, via Old French or straight from Latin) predicts which sister language its meaning stays closest to. Provenance thus works as an explanatory feature for meaning.
  • I3. A hypothesis that failed on averages still revealed structure once the authors looked at the full distribution, which turned out to be bimodal. The lesson is to inspect distributions before concluding there is no effect.
  • I4. Before any analysis, the authors catalogued the error types produced by automatic extraction and corrected them by hand. This made the downstream statistics trustworthy.
  • I5. Concrete and narrowly technical words keep stable meanings, while polysemous everyday words drift. This gives a cheap prior on where meaning ambiguity concentrates.

For Mova

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

  • I1 (back): when translating through the meta-world, cognate pairs could be checked against their senses in the graph to catch false friends, rather than trusting surface similarity.
  • I2 (world): the compiled lexicon could store borrowing path or domain-of-entry notes (such as a military sense of camp) to help choose WordNet senses explainably.
  • I3 (learn): evaluation reports could plot per-item score distributions for parsers and readers, so that hidden subpopulations of failures are not masked by mean accuracy.
  • I4 (learn): silver data agreed by witness parsers could get a similar error typology, and its categories could seed self-induced repair rules.
  • I5 (world): word sense disambiguation effort and confidence could be weighted by concreteness and polysemy, so the system spends explanation budget on drift-prone words.

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

🎓Universitat Politècnica de ValènciaValencia · Spain21 works · 3,275 citations🎓University of BucharestBucharest · Romania5 works · 363 citations

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 works

🧭 Topics 2Research subjects and languages the work is about.English · 1,373 worksHistorical linguistics & language change · 159 works

Cite

Cite as Ana Sabina Uban, Alina Maria Cristea, Anca Dinu, Liviu P. Dinu, Simona Georgescu and Laurentiu Zoicas. 2021. Tracking Semantic Change in Cognate Sets for English and Romance Languages. Proceedings of the 2nd International Workshop on Computational Approaches to Historical Language Change 2021.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2021.lchange-1.9

@inproceedings{uban2021tracking,
  title = {{Tracking Semantic Change in Cognate Sets for English and Romance Languages}},
  author = {Ana Sabina Uban, Alina Maria Cristea, Anca Dinu, Liviu P. Dinu, Simona Georgescu and Laurentiu Zoicas},
  booktitle = {Proceedings of the 2nd International Workshop on Computational Approaches to Historical Language Change 2021},
  year = {2021},
  doi = {10.18653/v1/2021.lchange-1.9},
  url = {https://doi.org/10.18653/v1/2021.lchange-1.9}
}

References

45 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. William L. Hamilton, Jure Leskovec, and Dan Jurafsky. 2016. Diachronic Word Embeddings Reveal Statistical Laws of Semantic Change. In Proceedings of ACL 2016, pages 1489–1501.
  2. Piotr Bojanowski, Edouard Grave, Armand Joulin, and Tomas Mikolov. 2016. Enriching Word Vectors with Subword Information. arXiv preprint arXiv:1607.04606. ↗
  3. Paul Heggarty. 2012. Beyond Lexicostatistics: How to Get More out of ”Word List” Comparisons. In Quantitative Approaches to Linguistic Diversity: Commemorating the Centenary of the Birth of Morris Swadesh, pages 113–137. Benjamins. find
  4. Francis Bond and Kyonghee Paik. 2012. A Survey of WordNets and their Licenses. Small, 8(4):5. find
  5. Lyle Campbell. 1998. Historical Linguistics. An Introduction. MIT Press. find
  6. Diana Inkpen, Oana Frunza, and Grzegorz Kondrak. 2005. Automatic identification of cognates and false friends in french and english. In Proceedings of RANLP 2005, volume 9, pages 251–257. find
  7. Santiago Castro, Jairo Bonanata, and Aiala Rosá. 2018. A High Coverage Method for Automatic False Friends Detection for Spanish and Portuguese. In Proceedings of the Fifth Workshop on NLP for Similar Languages, Varieties and Dialects (VarDial 2018), pages 29–36. find
  8. Andrey Kutuzov, Lilja Øvrelid, Terrence Szymanski, and Erik Velldal. 2018. Diachronic word embeddings and semantic shifts: a survey. arXiv preprint arXiv:1806.03537. ↗
  9. Yanqing Chen and Steven Skiena. 2016. False-friend detection and entity matching via unsupervised transliteration. arXiv preprint arXiv:1611.06722. ↗
  10. Vladimir I. Levenshtein. 1965. Binary Codes Capable of Correcting Deletions, Insertions, and Reversals. Soviet Physics Doklady, 10:707–710. find
  11. Alina Maria Ciobanu and Liviu P. Dinu. 2014. Building a Dataset of Multilingual Cognates for the Romanian Lexicon. In LREC 2014, pages 1038–1043. find
  12. April McMahon, Paul Heggarty, Robert McMahon, and Natalia Slaska. 2005. Swadesh Sublists and the Benefits of Borrowing: an Andean Case Study. Transactions of the Philological Society, 103(2):147–170. ↗
  13. Alina Maria Ciobanu and Liviu P. Dinu. 2015. Automatic Discrimination between Cognates and Borrowings. In Proceedings of ACL 2015, Volume 2: Short Papers, pages 431–437. find
  14. Antoine Meillet. 1906. Comment les Mots Changent de Sens. In Linguistique historique et linguistique générale. Champion, Paris. find
  15. Alina Maria Ciobanu and Liviu P. Dinu. 2019. Automatic Identification and Production of Related Words for Historical Linguistics. Computational Linguistics, 45(4):667–704. ↗
  16. James W. Minett and William S.-Y. Wang. 2003. On Detecting Borrowing: Distance-based and Character-based Approaches. Diachronica, 20(2):289–331. find
  17. Alina Maria Ciobanu and Liviu Petrisor Dinu. 2013. A Dictionary-Based Approach for Evaluating Orthographic Methods in Cognates Identification. In Proceedings of RANLP 2013, pages 141–147. find
  18. Marı́a Moliner. 2007. Diccionario de Uso del Español. Gredos, Madrid. find
  19. Alexandru Ciorănescu. 2002. Dicţionarul etimologic al limbii române. Saeculum, Bucharest. find
  20. Svetlin Nakov, Preslav Nakov, and Elena Paskaleva. 2009. Unsupervised extraction of false friends from parallel bi-texts using the web as a corpus. In Proceedings of RANLP 2009, pages 292–298. find
  21. Eugenio Coseriu. 1958. Sincronı́a, diacronı́a e historia. find
  22. Charles Talbut Onions, George Washington Salisbury Friedrichsen, and Robert William Burchfield. 1994. The Oxford Dictionary of English Etymology. Clarendon Press, Oxford. find
  23. Mihai Dinu. 1996. Personalitatea Limbii Romane. Cartea Romaneasca. find
  24. Haim Dubossarsky, Yulia Tsvetkov, Chris Dyer, and Eitan Grossman. 2015. A bottom up approach to category mapping and meaning change. In Net-WordS, pages 66–70. find
  25. Alexander Onysko and Esme Winter-Froemel. 2011. Necessary loans–luxury loans? Exploring the pragmatic dimension of borrowing. Journal of pragmatics, 43(6):1550–1567. find
  26. Haim Dubossarsky, Daphna Weinshall, and Eitan Grossman. 2017. Outta control: Laws of semantic change and inherent biases in word representation models. In Proceedings of EMNLP 2017, pages 1136–1145.
  27. Sanda Reinheimer Ripeanu. 2001. Lingvistica Romanica: Lexic, Morfologie, Fonetica. . BIC ALL, Bucuresti. find
  28. Sanda Reinheimer Ripeanu. 2004. Les emprunts latins dans les langues romanes. Editura Universităţii din Bucureşti. find
  29. Robert Anderson Hall. 1960. Linguistics and Your Language. Doubleday, New York. find
  30. Ana Sabina Uban, Alina Maria Ciobanu, and Liviu P Dinu. 2019b. Studying Laws of Semantic Divergence across Languages using Cognate Sets. In Proceedings of the 1st International Workshop on Computational Approaches to Historical Language Change, pages 161–166. find
  31. Alain Rey. 2011. Dictionnaire historique de la langue française. Le Robert, Paris. find
  32. André Rousseau. 2000. L’évolution lexico-sémantique: explications traditionnelles et propositions nouvelles. Théories contemporaines du changement sémantique, pages 11–30. find
  33. Ana Sabina Uban and Liviu P Dinu. 2020. Automatically Building a Multilingual Lexicon of False Friends With No Supervision. In Proceedings of LREC 2020, pages 3001–3007. find
  34. Wolfgang Schweickard. 2008. Le sirene degli eti-mologi nel mare onomastico. le reinterpretazioni paretimologiche. D’Achille, Paolo/Caffarelli, Enzo (edd.), Lessicografia e onomastica, 2:83–95. find
  35. Yang Xu and Charles Kemp. 2015. A Computational Evaluation of Two Laws of Semantic Change. In CogSci. find
  36. Samuel L Smith, David HP Turban, Steven Hamblin, and Nils Y Hammerla. 2017. Offline bilingual word vectors, orthogonal transformations and the inverted softmax. arXiv preprint arXiv:1702.03859. ↗
  37. Robyn Speer, Joshua Chin, Andrew Lin, Sara Jewett, and Lance Nathan. 2018. LuminosoIn-sight/wordfreq: v2.2. find
  38. Adam St Arnaud, David Beck, and Grzegorz Kondrak. 2017. Identifying cognate sets across dictionaries of related languages. In Proceedings of EMNLP 2017, pages 2519–2528. ↗
  39. Morris Swadesh. 1954. Perspectives and Problems of Amerindian Comparative Linguistics. WORD, 10(2-3):306–332. ↗
  40. Uri Tadmor. 2009. Loanwords in the world’s languages: Findings and results. In Martin Haspelmath and Uri Tadmor, editors, Loanwords in the World’s Languages, chapter 3, pages 55–75. De Gruyter Mouton, Berlin. find
  41. Nina Tahmasebi, Lars Borin, and Adam Jatowt. 2018. Survey of computational approaches to diachronic conceptual change. arXiv preprint arXiv:1811.06278. ↗
  42. Lianet Sepúlveda Torres and Sandra Maria Aluı́sio. 2011. Using machine learning methods to avoid the pitfall of cognates and false friends in Spanish-Portuguese word pairs. In Proceedings of the 8th Brazilian Symposium in Information and Human Language Technology. find
  43. Yulia Tsvetkov, Waleed Ammar, and Chris Dyer. 2015. Constraint-Based Models of Lexical Borrowing. In Proceedings of NAACL-HLT 2015, pages 598–608. find
  44. Ana Sabina Uban, Alina Ciobanu, and Liviu Dinu. 2019a. A Computational Approach to Measuring the Semantic Divergence of Cognates. In Proceedings of CICLing 2019. In press. find
  45. Ana Sabina Uban, Alina Ciobanu, and Liviu P Dinu. 2021. Cross-lingual Laws of Semantic Change. In Nina Tahmasebi, Lars Borin, Adam Jatowt, Yang Xu, and Simon Hengchen, editors, Computational Approaches to Semantic Change (In press). Berlin: Language Science Press. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/uban-2021-tracking-semantic-change-cognate-sets/