Analyzing the accuracy of Machine translation in literary works : a study on the appropriateness of Google Translate output in comparison to the human translation of Long Walk to Freedom: The autobiography of Nelson Mandela

Abstract

Over the centuries, improper translation has been the cause of discomfort, miscommunication and great distress. Nowadays, plenty of machine translation software made by state-of-the-art technology and very prominent companies are available for free on the internet, but inaccurate translation remains a problem to a certain level. The purpose of this article is to compare the appropriateness in translation of both the machine translation (MT) and human translation (HT) versions of a literary piece, with the aim of assessing the inaccuracies in translation, as to the reasons behind such, why it is possible to observe these errors in the machine version and also highlighting the importance of human translators to operate translation software. In order to achieve such objective, the analysis of the translation “errors” is performed in light of the qualitative scope, using the works of Galve (1996) on translation appropriateness, and PUCHAŁA-LADZIŃSKA (2016) on the effect of machine translation on human translators. The corpus of this article consists of a chapter of the literary work, Long walk to freedom: the autobiography of Nelson Mandela, which provides a number of examples on which the analysis is based. A conclusion is then drawn on whether the need for human translators will become obsolete in the future.