Elamur Research⌕
☰
home / papers / 2026 / Degree Zero of Translation: Using Interlinear Baselines to Quantify Translator Intervention

Degree Zero of Translation: Using Interlinear Baselines to Quantify Translator Intervention

Summary

Computational tools for translation usually score quality (BLEU, COMET) or identify translators (stylometry, Burrows' Delta). None of them measures how far a human translator departs from the source. The authors address this by treating an interlinear translation, a strict word-for-word gloss that keeps the Greek word order, as a neutral "Degree Zero" reference. This draws on Barthes' notion of neutral writing. The difference between the embedding of a literary translation and the embedding of its interlinear counterpart is called the Intervention Vector, and its length indicates how much the translator intervened.

The method was tested on New Testament texts. Interlinear baselines were collected for five languages, and 74 post-1900 literary translations were drawn from the targum corpus. Each translation was labelled as Literal, Formal, Dynamic or Paraphrase, and texts were compared at chapter level.

Without any supervision, vector magnitude ordered the translations from literal to paraphrase. Dynamic and paraphrase versions produced clearly longer vectors than literal and formal ones. The approach needs no set of reference translations and can be applied to a single text.

Key points

  • The corpus has 5 interlinear baselines, each scraped from a different website: the English one from BibleHub, the French one from Editeur BPC, the Italian one from Altervista, the Polish one from Oblubienica and the Spanish one from Bibliatodo. These are paired with 74 translations: English 16, French 14, Italian 12, Polish 16 and Spanish 16. The translations were selected from the 657-translation targum corpus.
  • Texts are compared at chapter level, giving 260 units per translation. Only translations from 1900 onward were included, to keep diachronic language change out of the comparison.
  • In pairwise comparisons of individual translations, chapter-level distances differed significantly in about half of the pairs overall and in 90% of the English pairs.
  • PCA projections of the vectors look different in each language. Polish shows a clear orthogonal V-shape. English and French place high-intervention texts further from the origin. Italian and Spanish show no pattern linked to intervention magnitude.
  • The authors list several limitations:
    • The interlinear gloss itself involves word choices, so it is not a truly neutral baseline.
    • Translations may follow different critical editions (Textus Receptus vs. Nestle-Aland), so some distance may reflect textual variants rather than intervention.
    • Biblical text is heavily represented in model training data, so results may not carry over to other literature.
    • The scraped data contain formatting artifacts that the metric could mistake for intervention.
  • Code and data are available at github.com/mrapacz/sighum-interlinear-vector-baselines.

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

Fields: Linguistics & languages · History & humanities

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.latechclfl-1.22

@inproceedings{rapacz2026degree,
  title = {{Degree Zero of Translation: Using Interlinear Baselines to Quantify Translator Intervention}},
  author = {Rapacz, Maciej and Smywiński-Pohl, Aleksander},
  booktitle = {Proceedings of the 10th Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature 2026},
  year = {2026},
  doi = {10.18653/v1/2026.latechclfl-1.22},
  url = {https://doi.org/10.18653/v1/2026.latechclfl-1.22}
}

References

23 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mona Baker. 1993. Corpus Linguistics and Translation Studies: Implications and Applications. In Mona Baker, Gill Francis, and Elena Tognini-Bonelli, editors, Text and Technology: In Honour of John Sinclair, pages 233–250. John Benjamins. ↗
  2. Roland Barthes. 1953. Writing Degree Zero. Hill and Wang, New York. find
  3. Walter Benjamin. 1923. The Task of the Translator. In Lawrence Venuti, editor, The Translation Studies Reader, pages 15–25. Routledge, London. find
  4. John Burrows. 2002. ‘Delta’: a Measure of Stylistic Difference and a Guide to Likely Authorship. Literary and Linguistic Computing, 17(3):267–287. find
  5. Marina Fomicheva and Lucia Specia. 2016. Reference Bias in Monolingual Machine Translation Evaluation. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 77–82, Berlin, Germany. Association for Computational Linguistics. find
  6. William L. Hamilton, Jure Leskovec, and Dan Jurafsky. 2016. Diachronic Word Embeddings Reveal Statistical Laws of Semantic Change. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1489–1501, Berlin, Germany. Association for Computational Linguistics. find our card
  7. Theo Hermans, editor. 1985. The Manipulation of Literature: Studies in Literary Translation. Croom Helm, London. find
  8. Junghwan Kim, Kieun Park, Sohee Park, Hyunggug Kim, and Bongwon Suh. 2025. MAS-LitEval : Multi-Agent System for Literary Translation Quality Assessment. arXiv preprint. ArXiv:2506.14199 [cs] version: 1. ↗
  9. Andrey Kutuzov, Lilja Øvrelid, Terrence Szymanski, and Erik Velldal. 2018. Diachronic word embeddings and semantic shifts: a survey. In Proceedings of the 27th International Conference on Computational Linguistics, pages 1384–1397, Santa Fe, New Mexico, USA. Association for Computational Linguistics. find our card
  10. André Lefevere. 2016. Translation, rewriting, and the manipulation of literary fame. Routledge. find
  11. Emad Mohamed, Raheem Sarwar, and Sayed Mostafa. 2023. Translator attribution for Arabic using machine learning. Digital Scholarship in the Humanities, 38(2):658–666. ↗
  12. Jeremy Munday. 2013. Style and Ideology in Translation: Latin American Writing in English. Routledge, New York. ↗
  13. Shira Wein and Nathan Schneider. 2024. Lost in Translationese? Reducing Translation Effect Using Abstract Meaning Representation. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 753–765, St. Julian’s, Malta. Association for Computational Linguistics. find our card
  14. Si Wu, John Wieting, and David A. Smith. 2025. Multiple References with Meaningful Variations Improve Literary Machine Translation. arXiv preprint. ArXiv:2412.18707 [cs]. ↗
  15. Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, and Jingren Zhou. 2025. Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models. arXiv preprint. ArXiv:2506.05176 [cs]. ↗ our card
  16. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. Bleu: a Method for Automatic Evaluation of Machine Translation. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pages 311–318, Philadelphia, Pennsylvania, USA. Association for Computational Linguistics. ↗ our card
  17. Maciej Rapacz and Aleksander Smywiński-Pohl. 2026. Targum – A Multilingual New Testament Translation Corpus. arXiv preprint. ArXiv:2602.09724 [cs]. ↗
  18. Ricardo Rei, Craig Stewart, Ana C Farinha, and Alon Lavie. 2020. COMET: A Neural Framework for MT Evaluation. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 2685–2702, Online. Association for Computational Linguistics. find our card
  19. Jan Rybicki. 2012. The great mystery of the (almost) invisible translator: Stylometry in translation. In Michael P. Oakes and Meng Ji, editors, Quantitative Methods in Corpus-Based Translation Studies: A practical guide to descriptive translation research, pages 231–248. John Benjamins Publishing Company. find
  20. Jan Rybicki and Magda Heydel. 2013. The stylistics and stylometry of collaborative translation: Woolf’s Night and Day in Polish. Literary and Linguistic Computing, 28(4):708–717. find
  21. M. Shuttleworth and M. Cowie. 2014. Dictionary of translation studies. St. Jerome Publishing. ↗
  22. Vered Volansky, Noam Ordan, and Shuly Wintner. 2015. On the features of translationese. Digital Scholarship in the Humanities, 30(1):98–118. find
  23. Anna Wegmann, Marijn Schraagen, and Dong Nguyen. 2022. Same Author or Just Same Topic? Towards Content-Independent Style Representations. In Proceedings of the 7th Workshop on Representation Learning for NLP, pages 249–268, Dublin, Ireland. Association for Computational Linguistics. TODO. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/rapacz-2026-degree-zero-translation-interlinear-baselines/