Ukrainian-to-English folktale corpus: Parallel corpus creation and augmentation for machine translation in low-resource languages

Machine translation has tremendous potential in connecting people and cultures. The Ukrainian language has an extensive collection of myths, legends, proverbs, songs, and folktales. They all represent the emotions, beliefs, and world views of Ukrainians. In this paper, we focus on several widely known Ukrainian folktales, all of which are anonymous due to the nature of this genre. Furthermore, folktales are usually passed on from one generation to another, going back hundreds of years. Interestingly, many available translations are rather transcreations, in which stories are retold and adapted to the target language and culture. We believe machine translation can be a useful supplemental tool in translating Ukrainian folklore, creating opportunities for more research and knowledge transfer about the Ukrainian language and culture. The first step in improving the machine translation performance of Ukrainian folktales is the creation of a highquality corpus that addresses domain-specific nuances and challenges.