Emerging trends: translationese
Natural Language Processing · 2025-05 · 31 (3) · 965-981
DOI: 10.1017/nlp.2025.1 ↗Kurz gefasst
Fachartikel in der Zeitschrift Natural Language Processing, online erschienen am 14.05.2025. Verfasst von Kenneth Church und 4 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref). Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Audits of multilingual resources are reporting shockingly poor quality: “less than 50% … acceptable quality.” There is too much translationese in too many of our multilingual resources, e.g., Wikipedia, XNLI, FLORES, WordNet. We view translationese as a form of noise that makes it hard to generalize from a benchmark based on translation to a real task of interest that does not involve translation. Worse, too much of this translationese is in the “wrong” direction. Directionality matters. Professional translators translate from their weaker language into their stronger language. Unfortunately, many of our resources translate in the other direction, from a stronger (higher-resource) language into a weaker (lower-resource) language. In Wikipedia, for example, there is more translation out of English than into English. We recommend more investments in high-quality data, and less in translation, especially in the “wrong” direction.
Zitieren
APA
Church, K., Li, B., Vickers, P., Dudy, S., & Yue, R. (2025). Emerging trends: translationese. Natural Language Processing, 31(3), 965-981. https://doi.org/10.1017/nlp.2025.1
Vancouver
Church K, Li B, Vickers P, Dudy S, Yue R. Emerging trends: translationese. Natural Language Processing. 2025;31(3):965-981. doi:10.1017/nlp.2025.1
BibTeX
@article{Church2025,
author = {Church, Kenneth and Li, Boyang and Vickers, Peter and Dudy, Shiran and Yue, Richard},
title = {Emerging trends: translationese},
journal = {Natural Language Processing},
year = {2025},
volume = {31},
number = {3},
pages = {965--981},
doi = {10.1017/nlp.2025.1}
}
EUVIMED