mlsanityData to 7 Oct 2026

When Does Reasoning Help in Machine Translation? A Hierarchical Analysis of LRM Reasoning Traces

arXiv 2609.21247 ↗First posted 18 Sep 2026cs.CLcited by 0

On the map: LLM reasoning → Reasoning benchmarks and evaluation

Abstract (arXiv)

Large Reasoning Models increasingly use intermediate traces for machine translation, but it remains unclear when such reasoning helps or hurts. We analyze reasoning traces across models, languages, domains, and datasets, focusing on reasoning language, length, and structure. We find that the best reasoning language is model-specific, reasoning length has a non-monotonic relationship with quality, and traces exhibit recurring functional patterns. To uncover these patterns, we introduce Hierarchical Meta-Summarization (HMS), a scalable framework that induces coarse- and fine-grained reasoning structures without predefined taxonomies. HMS reveals a shared organization--understanding/planning, translating/drafting, and refining/verifying--alongside domain-specific variation. Our results suggest that MT reasoning should be controlled in a model-aware, length-aware, and pattern-aware manner rather than uniformly encouraged.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 27 of its 32 references that are in the map. First posted on arXiv 18 Sep 2026.

Builds on

27 of 32 references.

Closest ancestors · 11

Same direction, newest first

  1. 2025Cross-lingual Collapse: How Language-Centric Foundation Models Shape Reasoning in Large Language Models6 Jun
  2. 2025When Models Reason in Your Language: Controlling Thinking Language Comes at the Cost of Accuracy28 May
  3. 2025How Well Do Large Reasoning Models Translate? A Comprehensive Evaluation for Multi-Domain Machine Translation26 May
  4. 2025Multilingual Test-Time Scaling via Initial Thought Transfer21 May
  5. 2025Language Mixing in Reasoning Language Models: Patterns, Impact, and Internal Causes20 May
  6. 2025Crosslingual Reasoning through Test-Time Scaling8 May
  7. 2025DeepSeek-R1 Thoughtology: Let's think about LLM Reasoning2 Apr
  8. 2025New Trends for Modern Machine Translation with Large Reasoning Models13 Mar
  9. 2025R1-T1: Fully Incentivizing Translation Capability in LLMs via Reasoning Learning27 Feb
  10. 2025Evaluating o1-Like LLMs: Unlocking Reasoning for Translation through Comprehensive Analysis17 Feb
  11. 2024DRT: Deep Reasoning Translation via Long Chain-of-Thought23 Dec

Other references · 16

  1. 2025Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilitiescited by 3,655
  2. 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMscited by 564
  3. 2025DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Modelscited by 154
  4. 2025Between Underthinking and Overthinking: An Empirical Study of Reasoning Length and correctness in LLMscited by 115
  5. 2024MetricX-24: The Google Submission to the WMT 2024 Metrics Shared Taskcited by 92
  6. 2025WMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & Dialectscited by 79
  7. 2023Benchmarking Machine Translation with Cultural Awarenesscited by 53
  8. 2025PolyMath: Evaluating Mathematical Reasoning in Multilingual Contextscited by 52
  9. 2025Understanding Aha Moments: from External Observations to Internal Mechanismscited by 48
  10. 2025Beyond Semantics: The Unreasonable Effectiveness of Reasonless Intermediate Tokenscited by 41
  11. 2025The relationship between reasoning and performance in large language models--o3 (mini) thinks harder, not longercited by 39
  12. 2024Translating Step-by-Step: Decomposing the Translation Process for Improved Translation Quality of Long-Form Textscited by 31
4 more
  1. 2025Mapping the Minds of LLMs: A Graph-Based Analysis of Reasoning LLMcited by 21
  2. 2025LIMOPro: Reasoning Refinement for Efficient and Effective Test-time Scalingcited by 15
  3. 2025Why Distillation can Outperform Zero-RL: The Role of Flexible Reasoningcited by 15
  4. 2025Exploring the Translation Mechanism of Large Language Modelscited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Unlocking Reasoning Capability on Machine Translation in Large Language Models
  2. 2025LLM Reasoning for Machine Translation: Synthetic Data Generation over Thinking Tokens
  3. 2026Reasoning over Grammar: Can Synthetic Linguistic Reasoning Traces Enhance Low-Resource Machine Translation?
  4. 2026Reasoning about In-Context Samples for Machine-Translation
  5. 2025Test-Time Scaling of Reasoning Models for Machine Translation
  6. 2025Are Large Reasoning Models Good Translation Evaluators? Analysis and Performance Boost
  7. 2026Should We be Pedantic About Reasoning Errors in Machine Translation?
  8. 2026TRACES: Tagging Reasoning Steps for Adaptive Cost-Efficient Early-Stopping