Corpus-Informed Literary Translation

Machine translation and corpus tools have been largely accepted in applied and technical translation, but for many years computer-assisted translation technology has been considered to be of little relevance in the translation of texts which do not rely on a standard format, phraseology, and terminology. In literary texts, it is not only the content but also the form that is of relevance, and the subtleties and formal intricacy of literary style were considered to pose an insurmountable obstacle to the automated operations of these new tools. However, with the wide range of software now available, it is possible to undertake the quantitative analysis of aspects of translation such as author or translator style, the validity of translation universals, and pre- and post-translation variants in literary texts, as well as large-scale studies of translation across time, space, and languages. Software such as Sketch Engine, Paraconc, Antconc, CATMA, and Wordsmith provides insights not possible in manual analysis of source and target texts. The increasing sophistication of computer technologies, the use of large real-text corpora and, in particular, neural networking systems have allowed far more complex applications, even permitting the use of machine-translation of literary texts where the human translator functions as a post-editor. Translation technology allows comparative literary analysis from multiple perspectives. Three case studies illustrate the uses of computer-assisted textual analysis in 1) literary translation practice at all levels of the translation process, 2) a large-scale close and distant reading project analysing and mapping translations of Jane Eyre into multiple languages worldwide, and 3) a case study of Chinese–English crime fiction investigating to what extent translated texts maintain their generic markers or adapt to the target environment's demarcations of genre.