Tourism Translation: from Corpus to Machine Translation (and back)

Abstract

Tourism language is characterised by features which make it distinct from other sector-based or technical languages. There are many examples of corpus-based studies and corpus-driven translation in the tourism sector but little regarding machine translation. Fewer still are the case studies or research papers dedicated to a comparison between machine-translated and corpus-based translated tourism texts. This paper aims to fill this gap by investigating whether, and to what extent, Google machine translation (from Italian into English) of a variety of tourism texts can be considered reliable or, at least, acceptable. To this end, it compares machine translations of tourism texts to their respective corpus-based translations. The paper’s findings uncover issues which mostly concern lexical and collocational choices, as well as a neglect of certain English writing conventions, such as those relating to clause structures, ego-targeting and figurative language. MT appears to perform well with informative and descriptive tourism texts, where sentences are simpler and no vivid language is involved. These, however, could hardly be considered representative of tourism texts, as a whole. The paper calls for advancements in MT algorithms in order to address certain lexical and collocational issues. Moreover, it is the opinion of the authors that MT in the tourism field is best left to translators capable of discerning accurate word usage in context.