mlsanityData to 14 Oct 2026

Reinforcement Learning Elicits Contextual Learning of Unseen Language Translation

arXiv 2606.06428 ↗First posted 4 Jun 2026cs.CLcited by 0

On the map: Deep reinforcement learning methods → RL training of language and agent models

Abstract (arXiv)

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 overfit specific languages, with limited zero-shot transfer at test time. To translate extremely low-resource languages at scale, we argue that LLMs must acquire the meta-skill of utilizing in-context linguistic knowledge rather than memorizing specific 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 (chrF) 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 language learning from context.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in RL training of language and agent models, one of 15 directions in Deep reinforcement learning methods. Builds on 28 of its 43 references that are in the map. First posted on arXiv 4 Jun 2026.

Builds on

28 of 43 references.

Closest ancestors · 2

Same direction, newest first

  1. 2026Improving Low-Resource Machine Translation via Round-Trip Reinforcement Learning18 Jan
  2. 2025SSR-Zero: Simple Self-Rewarding Reinforcement Learning for Machine Translation22 May

Other references · 26

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  3. 2017Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networkscited by 4,897
  4. 2019ZeRO: Memory Optimizations Toward Training Trillion Parameter Modelscited by 1,301
  5. 2024Tulu 3: Pushing Frontiers in Open Language Model Post-Trainingcited by 1,053
  6. 2020The State and Fate of Linguistic Diversity and Inclusion in the NLP Worldcited by 798
  7. 2022What Can Transformers Learn In-Context? A Case Study of Simple Function Classescited by 715
  8. 2021MetaICL: Learning to Learn In Contextcited by 454
  9. 2021Meta-learning via Language Model In-context Tuningcited by 164
  10. 2018Meta-Learning for Low-Resource Neural Machine Translationcited by 105
  11. 2022BLOOM+1: Adding Language Support to BLOOM for Zero-Shot Promptingcited by 90
  12. 2023A Benchmark for Learning to Translate a New Language from One Grammar Bookcited by 75
14 more
  1. 2023Dictionary-based Phrase-level Prompting of Large Language Models for Machine Translationcited by 74
  2. 2025MT-R1-Zero: Advancing LLM-based Machine Translation via R1-Zero-like Reinforcement Learningcited by 48
  3. 2024Improving Machine Translation with Human Feedback: An Exploration of Quality Estimation as a Reward Modelcited by 33
  4. 2024Teaching Large Language Models an Unseen Language on the Flycited by 30
  5. 2024LINGOLY: A Benchmark of Olympiad-Level Linguistic Reasoning Puzzles in Low-Resource and Extinct Languagescited by 28
  6. 2024Can LLMs Really Learn to Translate a Low-Resource Language from One Grammar Book?cited by 23
  7. 2025Understanding In-Context Machine Translation for Low-Resource Languages: A Case Study on Manchucited by 13
  8. 2024Back to School: Translation Using Grammar Bookscited by 11
  9. 2024LLM-Assisted Rule Based Machine Translation for Low/No-Resource Languagescited by 6
  10. 2025Explicit Learning and the LLM in Machine Translationcited by 6
  11. 2025Expanding the WMT24++ Benchmark with Rumantsch Grischun, Sursilvan, Sutsilvan, Surmiran, Puter, and Valladercited by 4
  12. 2026Translation Asymmetry in LLMs as a Data Augmentation Factor: A Case Study for 6 Romansh Language Varietiescited by 2
  13. 2025Improving Low-Resource Translation with Dictionary-Guided Fine-Tuning and RL: A Spanish-to-Wayuunaiki Studycited by 1
  14. 2026RUMLEM: A Dictionary-Based Lemmatizer for Romanshcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Translate-R1: Cost-Aware Translation Tool Use via Reinforcement Learning
  2. 2026Mending the Holes: Mitigating Reward Hacking in Reinforcement Learning for Multilingual Translation
  3. 2026Reinforcement Learning with Semantic Rewards Enables Low-Resource Language Expansion without Alignment Tax
  4. 2024Advancing Translation Preference Modeling with RLHF: A Step Towards Cost-Effective Solution
  5. 2026PEGRL: Improving Machine Translation by Post-Editing Guided Reinforcement Learning
  6. 2026Rewrite to Translate, Translate to Reward: Reinforcement Learning for Source Rewriting in Machine Translation
  7. 2026Learning When to Translate for Multilingual Reasoning
  8. 2024Improving Sample Efficiency of Reinforcement Learning with Background Knowledge from Large Language Models