Which Language Should a Skeleton Speak? Language Choices in Multilingual Reasoning

Key points

• Skeleton-based reasoning prompting is a promising training-free approach for structuring LLM reasoning, but prior work largely assumes an English-centric setting.1
• Across math benchmarks, model scales, and languages, we show that English skeletons yield a small positive tendency on average, most visible for smaller models and low-resource languages.1
• Combining greedy decoding, multi-rollout evaluation, translation ablation, and cross-benchmark validation, we further find three patterns of skeleton-language effects: directionally consistent, evaluation- and benchmark-dependent, and asymmetric negative.1
• Overall, skeleton language is a context-dependent design variable that requires multi-level exploration.1

Sources (1)

• [1]Which Language Should a Skeleton Speak? Language Choices in Multilingual Reasoning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 07:53 AM

  We propose the Language-Aware Skeleton Exploration Framework (LASEF) to study skeleton-language choice in multilingual mathematical reasoning.

  Skeleton-based reasoning prompting is a promising training-free approach for structuring LLM reasoning, but prior work largely assumes an English-centric setting.

Extractive summary: sentences quoted from the sources.