Match evaluation and over-editing in a translation memory environment
Computer-assisted translation (CAT) has been touted as a means of increasing translator productivity and improving translation quality while decreasing the amount of effort required to complete certain translation and localization tasks. Translation memory (TM) tools are among the most prevalent in non-literary translation projects and potentially make the translation process more complex, since their use alters the task environment. To the typical comprehension-transfer-production sub-tasks of translation without a TM are added cross-language evaluation, acceptability decision-making (matching), and possible post-editing. Drawing on Angelone’s (2010) notion of uncertainty management in translation, we explore how professional translators behave when presented with translations proposed by a translation memory system. We argue that the editing behavior observed during an experimental task highlights a mismatch between the proffered TM segment and the participant’s internal conception of what an optimal translation (a match) should be. This mismatch results in a tendency to over-edit.