Comparing forum data post-editing performance using translation memory and machine translation output: a pilot study

In this paper we present the results from a pilot study undertaken with translation students to compare community forum content post-editing performance based on suggestions from different translation systems. Output from both Translation Memory (TM) and Statistical Machine Translation (SMT) was presented to participants in the ACCEPT online postediting environment, where they needed to perform a translation task with and without translation proposals. Observed data showed that post-edited MT output obtained higher results on each of the variables measured: the amount of time needed to complete the task, the participants’ keystroke movements and the quality of the resulting translations.