VARTRA: A Comparable Corpus for Analysis of Translation Variation

This paper presents a comparable translation corpus created to investigate translation variation phenomena in terms of contrasts between languages, text types and translation methods (machine vs. computer-aided vs. human). These phenomena are reflected in linguistic features of translated texts belonging to different registers and produced with different translation methods. For their analysis, we combine methods derived from translation studies, language variation and machine translation, concentrating especially on textual and lexico-grammatical variation. To our knowledge, none of the existing corpora can provide comparable resources for a comprehensive analysis of variation across text types and translation methods. Therefore, the corpus resources created, as well as our analysis results will find application in different research areas, such as translation studies, machine translation, and others.