{"ID":2866373,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.19770","arxiv_id":"2509.19770","title":"EnAnchored-X2X: English-Anchored Optimization for Many-to-Many Translation","abstract":"Large language models (LLMs) have demonstrated strong machine translation capabilities for English-centric language pairs but underperform in direct non-English (x2x) translation. This work addresses this limitation through a synthetic data generation framework that leverages models' established English-to-x (en2x) capabilities. By extending English parallel corpora into omnidirectional datasets and developing an English-referenced quality evaluation proxy, we enable effective collection of high-quality x2x training data. Combined with preference-based optimization, our method achieves significant improvement across 72 x2x directions for widely used LLMs, while generalizing to enhance en2x performance. The results demonstrate that strategic exploitation of English-centric strengths can bootstrap comprehensive multilingual translation capabilities in LLMs. We release codes, datasets, and model checkpoints at https://github.com/NJUNLP/EAX","short_abstract":"Large language models (LLMs) have demonstrated strong machine translation capabilities for English-centric language pairs but underperform in direct non-English (x2x) translation. This work addresses this limitation through a synthetic data generation framework that leverages models' established English-to-x (en2x) cap...","url_abs":"https://arxiv.org/abs/2509.19770","url_pdf":"https://arxiv.org/pdf/2509.19770v1","authors":"[\"Sen Yang\",\"Yu Bao\",\"Yu Lu\",\"Jiajun Chen\",\"Shujian Huang\",\"Shanbo Cheng\"]","published":"2025-09-24T05:41:30Z","proceeding":"cs.CL","tasks":"[\"cs.CL\"]","methods":"[\"Large Language Model\",\"Language Model\"]","has_code":false,"code_links":[{"ID":609368,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2866373,"paper_url":"https://arxiv.org/abs/2509.19770","paper_title":"EnAnchored-X2X: English-Anchored Optimization for Many-to-Many Translation","repo_url":"https://github.com/NJUNLP/EAX","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}