MET: Theory-Grounded and Culture-Aware Multilingual Moral Reasoning

Hugging Face Daily Papers Papers
multilingual moral-reasoning culture-aware benchmark distillation nlp ethics

Summary

This paper introduces MCLASH, a multilingual moral decision-making benchmark, and MET, a theory-grounded prompting method for culture-aware moral reasoning, along with MET-D, a self-distillation training method that improves multilingual moral reasoning across different model families and sizes.

Language models are increasingly used for moral decision-making across diverse linguistic and cultural contexts, yet existing work overlooks multilinguality on three aspects: 1) multilingual evaluation benchmarks use direct translation, failing to adapt culture-specific items; 2) inference-time methods for moral reasoning rely on static, English-centric scaffolds and lack grounding in moral theory; 3) training methods for moral decision-making typically require expensive supervision from stronger models or human annotators. We address these gaps with three contributions. First, we introduce MCLASH, a multilingual moral decision-making benchmark to capture culturally situated moral intuitions and social norms across languages. Second, we propose MET (Multilingual Ethics with Theory-grounded reasoning), a two-step prompting method built on expert-curated, theory-based grounds drawn from psychology and philosophy: the model first selects situation- and culture-specific grounds, then reasons over them in the native language of the user. Third, we introduce MET-D (MET-Distillation), which enhances the second step through a self-distillation training stage that requires no external supervision. MET-D improves macro-F1 over the base model on all three models of different sizes and families (Qwen3-4B, Qwen3-8B, Gemma3-4B), by an average of 3.71 points on MCLASH and 4.23 on MMoralExceptQA, with a peak MCLASH gain of 12.94 points for Malay on Qwen3-8B. We further reveal that MET-D increases native-language reasoning by 62.13 points on average, and that beneficial grounds differ systematically across cultures. Together, these contributions open the path for culture-aligned, theory-grounded multilingual moral reasoning.
Original Article
View Cached Full Text

Cached at: 07/14/26, 08:17 PM

Paper page - MET: Theory-Grounded and Culture-Aware Multilingual Moral Reasoning

Source: https://huggingface.co/papers/2607.11736

Abstract

Languagemodelsareincreasinglyusedformoraldecision-makingacrossdiverselinguisticandculturalcontexts,yetexistingworkoverlooksmultilingualityonthreeaspects:1)multilingualevaluationbenchmarksusedirecttranslation,failingtoadaptculture-specificitems;2)inference-timemethodsformoralreasoningrelyonstatic,English-centricscaffoldsandlackgroundinginmoraltheory;3)trainingmethodsformoraldecision-makingtypicallyrequireexpensivesupervisionfromstrongermodelsorhumanannotators.Weaddressthesegapswiththreecontributions.First,weintroduceMCLASH,amultilingualmoraldecision-makingbenchmarktocaptureculturallysituatedmoralintuitionsandsocialnormsacrosslanguages.Second,weproposeMET(MultilingualEthicswithTheory-groundedreasoning),atwo-steppromptingmethodbuiltonexpert-curated,theory-basedgroundsdrawnfrompsychologyandphilosophy:themodelfirstselectssituation-andculture-specificgrounds,thenreasonsovertheminthenativelanguageoftheuser.Third,weintroduceMET-D(MET-Distillation),whichenhancesthesecondstepthroughaself-distillationtrainingstagethatrequiresnoexternalsupervision.MET-Dimprovesmacro-F1overthebasemodelonallthreemodelsofdifferentsizesandfamilies(Qwen3-4B,Qwen3-8B,Gemma3-4B),byanaverageof3.71pointsonMCLASHand4.23onMMoralExceptQA,withapeakMCLASHgainof12.94pointsforMalayonQwen3-8B.WefurtherrevealthatMET-Dincreasesnative-languagereasoningby62.13pointsonaverage,andthatbeneficialgroundsdiffersystematicallyacrosscultures.Together,thesecontributionsopenthepathforculture-aligned,theory-groundedmultilingualmoralreasoning.

View arXiv pageView PDFProject pageAdd to collection

Models citing this paper11

#### launch/MET-D-Qwen3-4B Text Generation• 4B• Updatedabout 4 hours ago • 759 #### launch/MET-D-Qwen3-8B Text Generation• 8B• Updatedabout 4 hours ago • 836 #### launch/MET-D-Gemma3-4B Text Generation• 4B• Updatedabout 4 hours ago • 859 #### launch/MET-D-Qwen3-4B-en-only Text Generation• 4B• Updatedabout 4 hours ago • 775 Browse 11 models citing this paper## Datasets citing this paper1

#### launch/MCLASH Viewer• Updatedabout 4 hours ago • 2.61k • 103

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.11736 in a Space README.md to link it from this page.

Collections including this paper1

Similar Articles

Think Multilingual, Not Harder: A Data-Efficient Framework for Teaching Reasoning Models to Code-Switch

arXiv cs.CL

This paper introduces a data-efficient fine-tuning framework for teaching reasoning models to code-switch (mix languages) effectively, demonstrating that strategic code-switching can improve reasoning capabilities for lower-resource languages. The work analyzes code-switching behaviors in large language models across diverse languages, tasks, and domains, then develops interventions to promote beneficial code-switching patterns.