Are parallel translation tasks parallel in difficulty? an eye-tracking study

Abstract

ABSTRACT This study explores translation task equivalence by comparing parallel translation test tasks in levels of difficulty, using eye-tracking, as well as NASA-TLX and retrospection. It also examines the effectiveness of measures of translation difficulty when differences in translation difficulty may be minor. Fifty-two participants completed two Chinese-to-English parallel translation test tasks, ‘Airplane’ and ‘Subway’, in a randomly assigned, counter-balanced order, with their eyes tracked. Upon completion of each task, participants were asked questions regarding translation difficulty. They were also required to fill in the NASA-TLX questionnaire. Paired samples t-tests showed significant differences in eye-tracking metrics, temporal measures, subjective ratings, and translation quality, indicating that ‘Airplane’ was more difficult. The retrospective interviews confirmed these differences. Temporal measures were found more sensitive to task change than eye-tracking measures. The target text and source text eye-tracking measures were similar in sensitivity. The effectiveness of the measures is explained in relation to what is actually measured or elicited, translation directionality, translators, and task. This study has implications for validating parallel translation tests, translation task development, understanding differences and similarities in translation difficulty, and measuring translation difficulty.