The Reasoning Lingua Franca: A Double-Edged Sword for Multilingual AI

TL;DR

This paper investigates multilingual reasoning in large reasoning models, revealing that English reasoning often outperforms native language reasoning but introduces translation errors, impacting interpretability and accuracy.

Contribution

It systematically compares reasoning in English versus native languages in LRMs, highlighting strengths and vulnerabilities of multilingual reasoning strategies.

Findings

English reasoning traces show more cognitive behaviors.

Reasoning in English yields higher accuracy, especially on complex tasks.

Translation errors can cause significant reasoning failures.

Abstract

Large Reasoning Models (LRMs) achieve strong performance on mathematical, scientific, and other question-answering tasks, but their multilingual reasoning abilities remain underexplored. When presented with non-English questions, LRMs often default to reasoning in English, raising concerns about interpretability and the handling of linguistic and cultural nuances. We systematically compare an LRM's reasoning in English versus the language of the question. Our evaluation spans two tasks: MGSM and GPQA Diamond. Beyond measuring answer accuracy, we also analyze cognitive attributes in the reasoning traces. We find that English reasoning traces exhibit a substantially higher presence of these cognitive behaviors, and that reasoning in English generally yields higher final-answer accuracy, with the performance gap increasing as tasks become more complex. However, this English-centric strategy is susceptible to a key failure mode - getting "Lost in Translation," where translation steps lead to errors that would have been avoided by reasoning in the language of the question.