Multilingual access to medieval literature requires retrieval systems to bridge historical and cross-lingual semantic variation. This article introduces HistRAG, a language-aware framework combining lexical retrieval, multilingual dense representations, historical adaptation, hybrid fusion, and cross-encoder reranking. HistRAG separates the effects of document modernization, query adaptation, and retrieval architecture within a unified experimental setting. The framework is evaluated on a heterogeneous collection of Medieval Catalan literature using equivalent queries in three contemporary languages. Results show that lexical retrieval is highly sensitive to query language, while multilingua... More >