unseen-language-translation
Filtering by topic unseen-language-translation(1)Clear all filters
- PaperarXiv — Language & NLP (cs.CL)4 Jun 2026
Reinforcement Learning Elicits Contextual Learning of Unseen Language Translation
Hanxu Hu, Zdeněk Šnajdr, Pinzhen Chen, Jannis Vamvas et al.
Reinforcement learning enables large language models to learn to translate completely unseen languages by using in-context linguistic information, outperforming in-context learning and supervised fine-tuning. The approach uses a surface-level translation metric as reward, suggesting that outcome-based RL can generalize beyond reasoning tasks to language learning from context.
Original 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 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.