FormosanMT: A Multilingual Parallel Corpus of the Formosan Language Family
Nearly 500,000 Formosan–Mandarin and Formosan–English sentence pairs, with extraction tools, a Traditional Chinese tokenizer, fine-tuning code, and trained NLLB and OpenNMT models.
Paper PDF DOI Recorded talk Conference page
Citation
Hunter Scheppat, Joshua K. Hartshorne, Sema Koc, Éric Le Ferrand, and Emily Prud’hommeaux. 2026. “FormosanMT: A Multilingual Parallel Corpus of the Formosan Language Family.” Proceedings of the Fifteenth Language Resources and Evaluation Conference, pp. 3445–3455.