Translating by post-editing: is it the way forward?

Abstract

Translation memory tools now offer the translator to insert post-edited machine translation segments for which no match is found in the databases. The Google Translator Toolkit does this by default, advising in its Settings window: "Most users should not modify this". Post-editing of no matches appears to work on engines trained with specific bilingual data on a source written under controlled language constraints. Would this, however, work for any type of task as Google's advice implies? We have tested this by carrying out experiments with English-Chinese trainees, using the Toolkit to translate from the source text (the control group) and by post-editing (the experimental group). Results show that post-editing gains in productivity are marginal. With regard to quality, however, post-editing produces significantly better statistical results compared to translating manually. These gains in quality are observed independently of language direction, text difficulty or translator's level of performance. In light of these findings, we discuss whether translators should consider post-editing as a viable alternative to conventional translation.

Translation Methods and Experience: A Comparative Analysis of Human Translation and Post-editing with Students and Professional Translators

“…There seems to be no statistically significant difference in quality between human translation and post-editing either, which confirms previous findings that post-editing can produce texts that are at least as good as human translations (Garcia 2011).…”

“…Carl, Dragsted, Elming, et al (2011) established that post-edited sentences were usually ranked as being better than sentences translated from scratch. Comparable results were obtained by Garcia (2011), who found that post-edited texts received better grades than texts translated from scratch. Guerberof (2009) compared human translation with translation from MT and translation from TM, and found that translation from MT led to better final quality than translation from TM, although regular human translation still outperformed translation from MT.…”

“…Findings for more general text types, however, were not always as convincing. There were indications of post-editing being faster, though not always significantly so (Carl, Dragsted, Elming, et al 2011;Garcia 2011).…”

Productivity and quality when editing machine translation and translation memory outputs: an empirical analysis of English to Welsh translation

“…As these studies informed the deductive hypotheses related to productivity measured in this study, all will now be briefly described below and the number of translators and type of source text will be given (if noted in the original publication). Only studies involving translation professionals have been reviewed below, but further studies that have used bilinguals or translation students when comparing translation speed are Koehn (2009), Daems et al (2013), De Sousa et al (2011), Lee and Liao (2011), García (2011), Yamada (2012, Vázquez et al (2013), Läubli et al (2013) and Depraetere et al (2014). These also report greater efficiency when using TM and MT.…”

Data-driven annotation of binary MT quality estimation corpora based on human post-editions

“…The result is a gradual shift from the original notion of professional translation (fully manual, hence slow and costly) to the concept of "translation as post-editing" , i.e. the revision of raw machine translated output (Guerberof 2009;Garcia 2011;Carl et al 2011;Zhechev 2012;Federico et al 2012;Green et al 2013;Läubli et al 2013).…”