A tagged and aligned corpus for the study of Proper Names in translation
Résumé
In this paper, we propose the creation of a tagged and aligned corpus for the study of a linguistic phenomenon, the translation of proper names. We try to modify the hypothesis according to which proper names cannot be translated and should therefore appear as borrowings in a targetlanguage. To do so, we introduce a parallel multilingual corpus made of eleven versions in ten different languages of a novel. One of these versions, the French one, which appears to be the source-text, undergoes named entity extraction so as to localize more easily the phenomenon we try to study. We focus on the tools used for the creation of our corpus and present some results refuting the idea that proper names are not translatable.