Translation Asymmetry in LLMs as a Data Augmentation Factor: A Case Study for 6 Romansh Language Varieties
This solves a domain-specific problem for low-resource language translation, with incremental improvements in data augmentation direction.
The paper tackled the problem of low-resource machine translation for Romansh by addressing LLM confusion across its 6 varieties, achieving a 23 BLEU improvement over Gemini 3 Pro in the lowest-resource variety and producing the first fluent translations in individual varieties.
Recent strategies for low-resource machine translation rely on LLMs to generate synthetic data from higher-resource languages. We find that this method fails for Romansh, because LLMs tend to confuse its 6 distinct language varieties. Our experiments show that instead, the direction of data augmentation should be aligned with the resource gradient between source and target language. This approach surpasses Gemini 3 Pro in the lowest-resource variety of Romansh by 23 BLEU. A human evaluation confirms that our experiments yield the first model that generates fluent translations in the individual Romansh varieties.