Translation Revision: revision procedures and their impact on revision product and process

Abstract

Since 2006, European translation agencies wishing to work in accordance with the European standard EN 15038 relating to translation services must incorporate revision into their process. Yet, the standard remains unclear on the way in which the revision must be accomplished. Therefore, this research focuses on the impact of the translation revision procedure on revision product and process.

As far as the product is concerned, the aim is to determine whether the revision procedure has an effect on revision quality, and if so, whether there is a difference in terms of efficiency between the following four procedures that were selected on the basis of an exploratory study (Robert, 2008) : procedure M (monolingual proofreading), procedure B (bilingual proofreading), procedure BM (bilingual proofreading followed by a monolingual proofreading), and finally procedure MB (monolingual proofreading followed by a bilingual proofreading).

As far as the process is concerned, the same research question has been formulated, but this time focusing on the process duration and what we have called the ‘error detection potential’, i.e. the capacity to detect an error in a translation with a particular revision procedure. Additional research questions concern specific revision interventions that were not used as quality indicators, the logical focus of attention of revisers when using procedures BM and MB, and possible ‘best revision practices’.

The general hypothesis verification strategy is quasi-experimental: the experiments that were conducted were aimed at observing the revision product and process (dependent variables) of sixteen professional revisers when the independent variable varies under four modalities. Four data collection instruments were used: revision product analysis, Think Aloud Protocols, and the keystroke logging software Inputlog, and short retrospective interviews.

Statistical tests (analysis of variance and multilevel analysis) confirmed the hypothesis of an effect of the procedure on product and process and also showed that procedures B, BM and MB were more efficient than procedure M as far as revision quality and error detection potential are concerned. However, when it comes to duration, procedure M showed to be more efficient than BM and MB, but not more efficient than B. Additionally, no significant difference was observed between B, BM and MB (product and process). Consequently, we were able to formulate some advice taking quality and time into account.