Post-editing of machine translation output with and without source text
Matej Kovačević · Repozitorij Filozofskog fakulteta u Zagrebu' at University of Zagreb (University of Zagreb) · 2014
Post-editing of machine translation output is a practice which aims to speed up translation production and distribution of information.There is still no consensus regarding the question of whether post-editors should have access to the source text of the translations they are postediting.The aim of this paper was to see how access to source text influences post-editors' quality of work and their speed, which is directly related to productivity.An experiment was conducted among 22 graduate students of English, who post-edited two translations about the European Union produced by Google Translate.The subjects were divided into two groups and each had access to the source text for only one of the translations.In the experiment, it was measured how long it took to post-edit the texts and how many errors in the MT output the subjects were able to correct.The errors were analysed and divided into categories in order to get a more precise picture.Contrary to expectations, access to source text was found not to have significant impact on speed.As expected, it did have an impact on the quality of the final translation.