A Tulu Resource for Machine Translation
This addresses the problem of machine translation development for low-resource languages like Tulu, though it is incremental as it applies existing transfer learning methods to a new language.
The authors tackled machine translation for the low-resource Tulu language by creating the first English-Tulu parallel dataset and developing a model using transfer learning from related languages, which outperformed Google Translate by 19 BLEU points without using parallel English-Tulu data.
We present the first parallel dataset for English-Tulu translation. Tulu, classified within the South Dravidian linguistic family branch, is predominantly spoken by approximately 2.5 million individuals in southwestern India. Our dataset is constructed by integrating human translations into the multilingual machine translation resource FLORES-200. Furthermore, we use this dataset for evaluation purposes in developing our English-Tulu machine translation model. For the model's training, we leverage resources available for related South Dravidian languages. We adopt a transfer learning approach that exploits similarities between high-resource and low-resource languages. This method enables the training of a machine translation system even in the absence of parallel data between the source and target language, thereby overcoming a significant obstacle in machine translation development for low-resource languages. Our English-Tulu system, trained without using parallel English-Tulu data, outperforms Google Translate by 19 BLEU points (in September 2023). The dataset and code are available here: https://github.com/manunarayanan/Tulu-NMT.