Reasoning over Grammar: Can Synthetic Linguistic Reasoning Traces Enhance Low-Resource Machine Translation?
Summary
Large language models can improve translation for low-resource languages through structured linguistic reasoning traces, with the most significant benefits occurring during inference rather than training.
View Cached Full Text
Cached at: 06/09/26, 12:41 PM
Paper page - Reasoning over Grammar: Can Synthetic Linguistic Reasoning Traces Enhance Low-Resource Machine Translation?
Source: https://huggingface.co/papers/2606.03782
Abstract
Large language models can improve translation for low-resource languages through structured linguistic reasoning traces, with the most significant benefits occurring during inference rather than training.
Large language models(LLMs) offer a promising approach tomachine translation(MT) for extremelylow-resource languagesby incorporating linguistic resources throughin-context learning. However, LLMs often struggle to apply grammatical information effectively during translation. Inspired by recent progress inchain-of-thought reasoning, we investigate whether low-resource MT can benefit from structured intermediate steps of linguistic analysis and grammatical reasoning. We propose a pipeline for automatically generating step-by-steplinguistic reasoning tracesfromUniversal Dependencies treebanks, dictionaries, and grammar-rule banks. We evaluate these traces in three settings:in-context learning(ICL),supervised fine-tuning(SFT), andreinforcement fine-tuning(RFT), on Xibe and Chintang as test cases. Our results show thatlinguistic reasoning tracesare most effective as inference-time guidance: in ICL, reliable sentence-specific traces substantially improve translation performance across most models, languages, and metrics. In contrast, using thelinguistic reasoning tracesas training data yields smaller and less consistent gains, as models learn the trace format but often generate erroneous content. These findings suggest that LLMs can leverage grammatical information for low-resource MT when given reliable linguistic analyses, while learning to generate such analyses remains a major bottleneck.
View arXiv pageView PDFAdd to collection
Get this paper in your agent:
hf papers read 2606\.03782
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.03782 in a model README.md to link it from this page.
Datasets citing this paper1
#### OLAResearchX/LingReason Viewer• Updated5 days ago • 4.58k • 33
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.03782 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
The Price of Reasoning: Cost-Quality Tradeoffs in Reinforcement Learning for Neural Machine Translation
Investigates the importance of reasoning traces in reinforcement learning for neural machine translation, showing that including reasoning during inference improves quality at the cost of increased computational demands.
Large Reasoning Models Are (Not Yet) Multilingual Latent Reasoners
This paper investigates multilingual latent reasoning in large reasoning models across 11 languages, revealing that while latent reasoning capabilities exist, they are unevenly distributed—stronger in resource-rich languages and weaker in low-resource ones. The study finds that despite surface-level differences, the internal reasoning mechanisms are largely aligned with an English-centered pathway.
Reasoning Before Translation: Enhancing Legal Machine Translation with Structured Reasoning
This paper compares various approaches to enhance neural machine translation in the legal domain, including supervised fine-tuning and reinforcement learning with verifiable rewards. It focuses on Swiss legal system translations and shows that small language models can be greatly enhanced, with RL surpassing SFT performance.
A Factorial Study of Synthetic Data Generation for Low-Resource Machine Translation using Grammar Books
This paper introduces a pipeline that uses large language models to extract grammatical rules, example sentences, and lexicons from grammar books to generate synthetic parallel corpora for fine-tuning machine translation models, achieving ChrF++ gains of up to +8.8 on three low-resource languages.
Thinking in a Low-Resource Language: What SFT Builds, What RL Fixes, What Accuracy Cannot See
This paper investigates fine-tuning MoE models to reason in Greek, revealing that accuracy metrics are noisy, while supervised fine-tuning and reinforcement learning improve language-specific reasoning and fix behavioral defects, with proposed evaluation instruments.