English routing reduces accuracy in multilingual multi-agent systems
Translating the Translator: Decomposing the Cost of English-Forced Inter-Agent Communication
Computation and LanguageArtificial IntelligenceMultiagent Systems
Summary
The paper studies how relying on English for communication between AI agents affects performance when the user's language is not English. The authors compare two setups: one where agents communicate directly in the user’s native language, and one where all communication is forced through English with translation steps. They find that forcing communication through English lowers accuracy significantly, especially for languages that are very different from English. This suggests that letting agents talk in the user’s own language can improve results by avoiding translation errors.
What this means in practice
- •For multilingual chatbot developers: Improve chatbot accuracy by enabling agent communication directly in the user’s language rather than routing through English.
- •For global customer support teams: Reduce errors in automated multilingual support workflows by avoiding intermediate English translations between processing modules.
Authors
Kushagra Agrawal, Yuming Feng, Man-Fai Leung
Abstract
Multi-agent LLM architectures, such as LangChain and AutoGen, largely assume English as the lingua franca for internal inter-agent communication, even when the end-user task is non-English. We fill this gap by evaluating a two-agent extraction-answer core, with an additional back-translation agent in the English-forced condition, across four typologically diverse languages (Hindi, Chinese, Spanish, Arabic; n = 300 per language) using the Aya-23-8B model. We compare a native-language pipeline to an English-forced one (which incorporates a final back-translation step from English to the user's language). We discover a statistically significant English-Forcing Tax (surviving a strict Bonferroni correction) that isolates the cost of English routing from general multi-agent orchestration overhead. Forcing inter-agent communication through English reduces Exact Match accuracy by 13.0 percentage points (Spanish) up to 30.6 percentage points (Hindi) compared to native-language multi-agent execution. Using chrF scores as a diagnostic measure of English-reference lexical overlap, we find that lower overlap is strongly associated with pipeline failure, consistent with translation loss being an important contributor to the observed performance drop. These findings suggest a compelling case for native-language routing in agent frameworks when the source and target languages are typologically distant, reducing a compounding translation tax.