Breaking Language Preference in Multilingual RAG via Language-Controllable Retrieval and Language-Agnostic Reasoning

Retrieval-Augmented Generation (RAG) significantly improves the factual accuracy and generation quality of large language models by incorporating external knowledge. However, in multilingual settings, RAG systems suffer from severe language preference. On the one hand, the retrieval stage is sensitive to the query language: semantically equivalent queries expressed in different languages often lead to substantially different retrieval results. On the other hand, when retrieved documents contain knowledge written in multiple languages, large language models tend to be influenced by surface-level language forms, rather than reasoning solely based on semantic relevance to the query. To address these challenges, we propose a unified optimization framework that explicitly disentangles multilingual RAG into language-controllable retrieval and language-agnostic reasoning. Our framework allows LLM to adaptively select retrieval languages while enforcing cross-lingual consistency during reasoning, thereby mitigating language bias without modifying existing retrievers or translators. Experimental results demonstrate that our approach effectively reduces language bias in multilingual RAG and consistently outperforms baselines across multiple multilingual benchmarks . Query: [Japan] 西遊記の 孫悟空 は どこで 生まれました か? (What is the birthplace of the Sun Wukong in Journey to the West?) Query (Japan) Query (Chinese) Doc -4 [ENG] Doc -5 [KOR] Doc -6 [JPN] Doc -1 [ENG] Doc -2 [ZHO] Doc -3 [JPN] multilingual retrieve 惑星ベジータ (Vegeta Planet) Language preference influences how documents are leveraged during answer generation Wo wurde der König der Affen geboren? What is the birthplace of the Monkey King in Journey to the West?