Faster is not always better, but what if it’s not worse? A comparative study of human translation and post-edited machine translation quality

Publication type
Publication status
Daems, J, & Macken, L.
View in Biblio
(externe link)


According to TranslateMedia , it is more cost-effective to translate from scratch than to correct machine-translated (MT) output, but is this really the case? Research has shown that for some text types, post-editing (PE) MT leads to an increase in both productivity and quality (Guerberof, 2009; Plitt & Masselot, 2010; Tatsumi, 2010). However, research dealing with general texts and focusing on typical translation problems for human translation (HT) and MT+PE is still missing.

As part of a larger research project designed to address these issues, we conducted a pilot study with Master’s students in translation. Participants performed both a post-editing and a ‘regular’ translation task from English to Dutch. Translation products were assessed in terms of adequacy and acceptability (Toury, 1995).

Results show that MT followed by PE is faster than HT and of comparable quality to HT. There was, however, a notable difference in the type of errors detected for both types of translation. For example, almost ten per cent of all PE-problems consisted of word sense shifts, compared to four per cent of word sense shifts for HT.

This investigation supports research on PE-quality and productivity gain, while at the same time providing a deeper understanding of PE-problems, which may prove insightful for translation trainers