earticle

영어

Almost a decade after Pym’s (2013) prediction that the rapid development of machine translation (MT) would soon turn all translators into post-editors, post-editing has indeed become an essential skill for every translator to master, even if it has not completely replaced all forms of human translation. It is by now obvious that every comprehensive translator training programme must include instruction in post-editing. Additionally, to keep pace with market demand, it is vital that students simultaneously develop their skills both as translators and as post-editors, as they could not otherwise compete effectively upon graduation. However, this kind of simultaneous training entails a number of difficulties. Firstly, students who have not yet fully mastered human translation skills face particular challenges in post-editing – unlike those professionals who acquired their human translation skills first. Secondly, the great disparity in the quality of MT output between language pairs, as well as the constantly changing nature of neural MT engines make it very difficult to develop exercises and curricula for effective post-editing instruction since training materials can quickly become outdated or irrelevant. Training in post-editing therefore needs to gain a broad understanding of the pitfalls of student translator behaviour in post-editing tasks. This paper presents a small-scale pilot study of English to Hungarian student post-editing. The study presents the results of a qualitative and quantitative examination of three post-editing tasks based on DeepL’s MT output carried out by 30 Hungarian BA students reading English at Károli Gáspár University. The aim of the study is to identify and quantify the typical problems and issues trainee translators struggle with in post-editing. According to the results, for trainee translators, post-editing and translation problems are not strictly separated. Furthermore, high quality MT output can in and of itself become a hazard as students scrutinize it less than they do lower quality MT. This paper aims to offer insights and suggestions that can be applied to post-editing training regardless of the particular language pair.