@inproceedings{borchert-etal-2025-language,
    title = "Language Fusion for Parameter-Efficient Cross-lingual Transfer",
    author = "Borchert, Philipp  and
      Vuli{\'c}, Ivan  and
      Moens, Marie-Francine  and
      De Weerdt, Jochen",
    editor = "Che, Wanxiang  and
      Nabende, Joyce  and
      Shutova, Ekaterina  and
      Pilehvar, Mohammad Taher",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingestion-acl-25/2025.acl-long.1255/",
    pages = "25848--25868",
    ISBN = "979-8-89176-251-0",
    abstract = "Limited availability of multilingual text corpora for training language models often leads to poor performance on downstream tasks due to undertrained representation spaces for languages other than English. This `under-representation' has motivated recent cross-lingual transfer methods to leverage the English representation space by e.g. mixing English and `non-English' tokens at the input level or extending model parameters to accommodate new languages. However, these approaches often come at the cost of increased computational complexity. We propose Fusion for Language Representations (FLARE) in adapters, a novel method that enhances representation quality and downstream performance for languages other than English while maintaining parameter efficiency. FLARE integrates source and target language representations within low-rank (LoRA) adapters using lightweight linear transformations, maintaining parameter efficiency while improving transfer performance. A series of experiments across representative cross-lingual natural language understanding tasks, including natural language inference, question-answering and sentiment analysis, demonstrate FLARE{'}s effectiveness. FLARE achieves performance improvements of 4.9{\%} for Llama 3.1 and 2.2{\%} for Gemma 2 compared to standard LoRA fine-tuning on question-answering tasks, as measured by the exact match metric."
}