Selecting a translation revision procedure: do common sense and statistics agree?

AbstractTranslators employ a wide variety of revision procedures in their day-to-day working practice. This paper empirically investigates the impact of these revision procedures on both the revision product and the revision process. Sixteen professional revisers revised four comparable translations, each time using a different procedure: (1) monolingual, (2) bilingual, (3) bilingual followed by monolingual, and (4) monolingual followed by bilingual. To analyse the data, a product-based revision analysis was combined with process-based analyses, i.e. using keystroke logs and think aloud protocols. Although common sense and research-based results seem to coincide at first sight with respect to the selection of a suitable revision procedure, some results deviated from the expected outcomes. To conclude, the paper formulates some practical advice on selecting revision procedures, taking quality and time into account.Keywords: translation revisionrevision proceduresrevision qualityrevision processmultilevel analysis Notes on contributorsIsabelle Robert has a Master's in Translation (1993, University of Mons, Belgium) and a PhD in Translations Studies (2012, University of Antwerp, Belgium). She is a lecturer of French at the University of Antwerp, Belgium (Department of Applied Linguistics). She is a member of the Editorial Board of Antverpiensia, New Series – Themes in Translation Studies (LANS – TTS, https://lans-tts.uantwerpen.be/index.php/LANS-TTS) and the Treasurer of the European Society of Translation Studies (EST). Her research interests include translation and revision processes, didactics of translation, and audiovisual translation.Luuk Van Waes is professor in Business and Technical Communication at the University of Antwerp, Belgium (Department of Management). Since 1985 he has been involved in different types of writing research, focusing on the influence of digital media on the organisation of writing processes. In order to observe the writing process in ethnographic settings he co-designed – together with Mariëlle Leijten – the logging program Inputlog (www.inputlog.net). Luuk Van Waes has published in several international peer-reviewed journals and he is also the editor of the Journal of Writing Research (ww.jowr.org).Notes1. The term 'loyal' has been chosen to echo the general theoretical framework used in the study, i.e. the functionalist vision of translation of Nord (Citation2005, p. 126), with its two pillars: 'loyalty' and 'functionality'. The term 'functional' has been chosen for the same reason. In the context of the study, 'loyal' must be understood as accurate and complete. It encompasses Mossop's revision parameters 'content' (logic and facts) and 'transfer' (accuracy and completeness) (Mossop, Citation2007b).2. Analysis of variance tests (ANOVA) are used to reveal differences between means. Since the assumptions related to parametric tests were not always met, the non-parametric Friedman test of analysis of variance had to be used. However, when the test was significant, the parametric variant (one-way ANOVA for repeated measures) was performed as well, because it allows for pairwise comparisons called post hoc tests.3. It should be noted that a third type of multilevel tests (three levels instead of two) was applied, but it will not be reported here, since it does not change the final findings but simply confirms them.4. Although there seems to be a general effect for the error detection potential (p=0.047), there does not seem to be a difference between procedures, as will be shown hereafter.