Quality and reformulation in simultaneous-with-text
This study aimed, first, to test the widely held belief that meaning-based processing (‘deverbalisation’) results in better-quality simultaneous interpretation (SI), by the indirect method of correlating users’ quality judgments with linguistic reformulation and elaboration, and additionally, with transcript-based measures of accuracy, style and fluency. Twenty-four French SI-with-text versions of two English speeches by experts and novices were played to a panel of users from international organizations, and the transcripts independently scored by a panel of interpreters for these specific parameters. Analysis showed major errors to be a significant predictor of user quality judgments (although users did not have access to the original speeches), as well as some positive effect of lexical and cohesive elaboration, but none for restructuring per se.
Secondly, since interpreting quality evaluation may vary between peers (the traditional judges), users, and other product-based methods, the study also explored ways of triangulating different sources — user judgments, transcript scores, and computer analysis of output text features — to learn more about the determinants of interpreting quality.