All Languages Matter: Understanding and Mitigating Language Bias in Multilingual RAG

arXiv:2604.20199v1 Announce Type: new Abstract: Multilingual Retrieval-Augmented Generation (mRAG) leverages cross-lingual evidence to ground Large Language Models (LLMs) in global knowledge. However, we show that current mRAG systems suffer from a language bias during reranking, systematically favoring English and the query's native language. By introducing an estimated oracle evidence analysis, we quantify a substantial performance gap between existing rerankers and the achievable upper bound. Further analysis reveals a critical distributional mismatch: while optimal predictions require evidence scattered across multiple languages, current systems systematically suppress such ``answer-critical'' documents, thereby limiting downstream generation performance. To bridge this gap, we propose extit{extbf{L}anguage-extbf{A}gnostic extbf{U}tility-driven extbf{R}eranker extbf{A}lignment (LAURA)}, which aligns multilingual evidence ranking with downstream generative utility. Experiments across diverse languages and generation models show that LAURA effectively mitigates language bias and consistently improves mRAG performance.

Related

• Language-Coupled Reinforcement Learning for Multilingual Retrieval-Augmented Generation
• Location Not Found: Exposing Implicit Local and Global Biases in Multilingual LLMs
• Do LLMs Use Cultural Knowledge Without Being Told? A Multilingual Evaluation of Implicit Pragmatic Adaptation

Source: arXiv cs.CL | 2026-04-23