aiqbench analysis
This research demonstrates that Large Language Model-based translation effectively preserves subtle moral nuances across languages, enabling the expansion of ethical AI datasets beyond English. By validating this approach on Polish social media data, the study provides a cost-effective pathway to building more inclusive and globally representative models for automated moral classification. The findings confirm that cross-lingual machine learning can bridge gaps in under-resourced domains without significant loss of semantic fidelity or predictive accuracy.
Moral Semantics Survive Machine Translation: Cross-Lingual Evidence from Moral Foundations Corpora
arXiv:2605.22660v2 Announce Type: replace Abstract: Moral language is subtle and culturally variable, making it difficult to translate faithfully across languages. Idiomatic expressions, slang, and cultural references introduce hard-to-avoid translation artifacts. Yet automated moral classification depends on language-specific annotated corpora that exist almost exclusively in English. We investigate whether LLM-based translation can bridge this gap, taking Polish as a test case. Using $\sim$50k morally annotated social media posts from a diverse range of topics, we apply a principled four-method validation pipeline: LaBSE cross-lingual embedding similarity, Centred Kernel Alignment (CKA), LLM-as-judge evaluation, and deep learning classifier parity tests. We show that despite shortcomings in handling slang, vulgarity, and culturally loaded expressions, direct translation preserves subtle moral cues well enough to be harvested by cross-lingual machine learning --- with a mean cosine similarity of 0.89 and classification accuracy gaps of 0.01--0.02 AUROC across foundations. These results demonstrate that machine translation is a practical and cost-effective path to moral values research in languages currently under-resourced in this domain. We demonstrate this for Polish as a representative Slavic language, with expected generalization to related languages.
Abstract:Moral language is subtle and culturally variable, making it difficult to translate faithfully across languages. Idiomatic expressions, slang, and cultural references introduce hard-to-avoid translation artifacts. Yet automated moral classification depends on language-specific annotated corpora that exist almost exclusively in English. We investigate whether LLM-based translation can bridge this gap, taking Polish as a test case. Using $\sim$50k morally annotated social media posts from a diverse range of topics, we apply a principled four-method validation pipeline: LaBSE cross-lingual embedding similarity, Centred Kernel Alignment (CKA), LLM-as-judge evaluation, and deep learning classifier parity tests. We show that despite shortcomings in handling slang, vulgarity, and culturally loaded expressions, direct translation preserves subtle moral cues well enough to be harvested by cross-lingual machine learning --- with a mean cosine similarity of 0.89 and classification accuracy gaps of 0.01--0.02 AUROC across foundations. These results demonstrate that machine translation is a practical and cost-effective path to moral values research in languages currently under-resourced in this domain. We demonstrate this for Polish as a representative Slavic language, with expected generalization to related languages.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
I.2.7; I.2.6; J.4
arXiv:2605.22660 [cs.CL]
(or arXiv:2605.22660v2 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2605.22660 Focus to learn more arXiv-issued DOI via DataCite
https://doi.org/10.1145/3794786.3830756 Focus to learn more DOI(s) linking to related resources
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 284 words
Related coverage
LuxInstruct: A Cross-Lingual Instruction Tuning Dataset For Luxembourgish
arXiv · Jul 28, 2026
The Alignment Journal: Organization, Personnel, and Scope
AI Alignment Forum · Sep 1, 2026
BETA-Labeling for Multilingual Dataset Construction in Low-Resource IR
arXiv · Sep 9, 2026
"I am an AI Safety Researcher"
LessWrong · Sep 23, 2026