@inproceedings{hu-etal-2026-reinforcement,
title = "Reinforcement Learning Elicits Contextual Learning of Unseen Language Translation",
author = "Hu, Hanxu and
Snajdr, Zdenek and
Chen, Pinzhen and
Vamvas, Jannis and
Sennrich, Rico",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.2096/",
pages = "44643--44657",
ISBN = "979-8-89176-462-0",
abstract = "Prior work has shown that large language models (LLMs) can translate unseen or low-resource languages by undergoing continued training or even by encoding a grammar book in their context. However, both methods typically treat specific languages, with limited zero-shot transfer at test time. To translate extremely low-resource languages at scale, we argue that a model should be able to use in-context linguistic knowledge regardless of language, rather than memorize specific training languages. In this paper, we propose a reinforcement learning (RL) approach to unseen language translation given rich linguistic context, using a surface-level translation metric as the reward. Empirically, despite the lightweight reward, our RL-trained models effectively extract and apply relevant linguistic information from the provided context, leading to better translations on completely unseen languages than in-context learning or supervised fine-tuning. Our analyses suggest that outcome-based RL can extend beyond conventional reasoning tasks like math and coding to serve as a recipe for low-resource translation from linguistic context."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hu-etal-2026-reinforcement">
<titleInfo>
<title>Reinforcement Learning Elicits Contextual Learning of Unseen Language Translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hanxu</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zdenek</namePart>
<namePart type="family">Snajdr</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pinzhen</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jannis</namePart>
<namePart type="family">Vamvas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rico</namePart>
<namePart type="family">Sennrich</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Prior work has shown that large language models (LLMs) can translate unseen or low-resource languages by undergoing continued training or even by encoding a grammar book in their context. However, both methods typically treat specific languages, with limited zero-shot transfer at test time. To translate extremely low-resource languages at scale, we argue that a model should be able to use in-context linguistic knowledge regardless of language, rather than memorize specific training languages. In this paper, we propose a reinforcement learning (RL) approach to unseen language translation given rich linguistic context, using a surface-level translation metric as the reward. Empirically, despite the lightweight reward, our RL-trained models effectively extract and apply relevant linguistic information from the provided context, leading to better translations on completely unseen languages than in-context learning or supervised fine-tuning. Our analyses suggest that outcome-based RL can extend beyond conventional reasoning tasks like math and coding to serve as a recipe for low-resource translation from linguistic context.</abstract>
<identifier type="citekey">hu-etal-2026-reinforcement</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2096/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>44643</start>
<end>44657</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Reinforcement Learning Elicits Contextual Learning of Unseen Language Translation
%A Hu, Hanxu
%A Snajdr, Zdenek
%A Chen, Pinzhen
%A Vamvas, Jannis
%A Sennrich, Rico
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F hu-etal-2026-reinforcement
%X Prior work has shown that large language models (LLMs) can translate unseen or low-resource languages by undergoing continued training or even by encoding a grammar book in their context. However, both methods typically treat specific languages, with limited zero-shot transfer at test time. To translate extremely low-resource languages at scale, we argue that a model should be able to use in-context linguistic knowledge regardless of language, rather than memorize specific training languages. In this paper, we propose a reinforcement learning (RL) approach to unseen language translation given rich linguistic context, using a surface-level translation metric as the reward. Empirically, despite the lightweight reward, our RL-trained models effectively extract and apply relevant linguistic information from the provided context, leading to better translations on completely unseen languages than in-context learning or supervised fine-tuning. Our analyses suggest that outcome-based RL can extend beyond conventional reasoning tasks like math and coding to serve as a recipe for low-resource translation from linguistic context.
%U https://aclanthology.org/2026.findings-emnlp.2096/
%P 44643-44657
Markdown (Informal)
[Reinforcement Learning Elicits Contextual Learning of Unseen Language Translation](https://aclanthology.org/2026.findings-emnlp.2096/) (Hu et al., Findings 2026)
ACL