Saved in:
Bibliographic Details
Main Authors: Kunz, Jenny, Jarochenko, Anja, Bollmann, Marcel
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.08450
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911499105075200
author Kunz, Jenny
Jarochenko, Anja
Bollmann, Marcel
author_facet Kunz, Jenny
Jarochenko, Anja
Bollmann, Marcel
contents Translations often carry traces of the source language, a phenomenon known as translationese. We introduce the first freely available English-to-Swedish dataset contrasting translationese sentences with idiomatic alternatives, designed to probe intrinsic preferences of language models. It includes error tags and descriptions of the problems in the original translations. In experiments evaluating smaller Swedish and multilingual LLMs with our dataset, we find that they often favor the translationese phrasing. Human alternatives are chosen more often when the English source sentence is omitted, indicating that exposure to the source biases models toward literal translations, although even without context models often prefer the translationese variant. Our dataset and findings provide a resource and benchmark for developing models that produce more natural, idiomatic output in non-English languages.
format Preprint
id arxiv_https___arxiv_org_abs_2603_08450
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle A Dataset for Probing Translationese Preferences in English-to-Swedish Translation
Kunz, Jenny
Jarochenko, Anja
Bollmann, Marcel
Computation and Language
Translations often carry traces of the source language, a phenomenon known as translationese. We introduce the first freely available English-to-Swedish dataset contrasting translationese sentences with idiomatic alternatives, designed to probe intrinsic preferences of language models. It includes error tags and descriptions of the problems in the original translations. In experiments evaluating smaller Swedish and multilingual LLMs with our dataset, we find that they often favor the translationese phrasing. Human alternatives are chosen more often when the English source sentence is omitted, indicating that exposure to the source biases models toward literal translations, although even without context models often prefer the translationese variant. Our dataset and findings provide a resource and benchmark for developing models that produce more natural, idiomatic output in non-English languages.
title A Dataset for Probing Translationese Preferences in English-to-Swedish Translation
topic Computation and Language
url https://arxiv.org/abs/2603.08450