Building Multilingual Parallel Corpora for Under-Resourced Languages Using Translated Fictional Texts
Résumé
In this paper, we present an ongoing research project which consists in collecting all the translations worldwide of one fictional text in order to build multilingual parallel corpora for a large number of under-resourced languages. Building such corpora is vital to help preserve and expand language and traditional knowledge diversity. These corpora will be useful to handle under-resourced languages in a number of interconnected research fields such as computational linguistics, translation studies and corpus linguistics. Our project taps into a wealth of translated versions of a single fictional text spanning a period of over a century. It consists in collecting, digitizing, transcribing and aligning translations of this text. Our data collection process is fluid and collaborative. It is based on volunteer work from the scientific and scholarly communities, the power of the crowd and national libraries and archives. Our first experiment was conducted on the world-famous and well-traveled American novel “Adventures of Huckleberry Finn” by the American author Mark Twain. This paper reports on 10 parallel corpus that are now chapter aligned pairing English with Arabic, Basque, Bengali, Bulgarian, Dutch, Hungarian, Polish, Russian, Turkish and Ukrainian processed out of a total of 20 collected translations.