Cognitive Load In Sight Translation From Chinese To English,French And Japanese:A Comparative Eye-tracking Study

This study compares and contrasts the eye-tracking behaviors of student interpreters with English,French and Japanese as target languages in sight translation and adopts a triangulation paradigm of eye-tracking,audio recording,and retrospective interview.Twenty subjects,all majors of Translation and Interpreting,spoke Chinese as their first language,six of them spoke English,five of them spoke French and nine of them spoke Japanese as their second language.Chinese text difficulty was controlled by sentence length.Tobii TX300,a remote eye tracker,was used to record the subjects’ behavior during the sight translation process and a modified version of the Translation Difficulty Scale（NASA-TLX）was applied to collect subjects’ subjective assessments of the difficulty of the task to cross-check the experimental results.The study also assessed the quality of the transcribed translation using scoring criteria of translation in Test for English Major 8,Test for French Major 8 and Test for Japanese Major 8 respectively.Based on the analysis of CL and eye-tracking technology,a quantitative model of CL for multilingual sight translation based on eye-tracking technology was constructed by combining three eye-movement indicators,namely interest area first fixation duration,interest area,interest area max fix pupil size,interest area min fix pupil size,as well as grade.The results of the study showed that:（1）The target language had a significant effect on the CL in the sight translation process;the CL was in descending order for Japanese,French and English.（2）The CL tended to be higher when the Chinese text was more difficult than when it was less difficult,or when the sentences were longer than when they were shorter,and it was significant.（3）The use of simplification and syntactic linearity strategies by the subjects could reduce the CL and the phenomena of DF.（4）The Chinese text difficulty had a significant effect on the quality of the translation,but the output quality of the more difficult text was not necessarily lower than that of the less difficult one.（5）A multiple regression model（MRM）was statistically deduced to include relevant data of three target languages.（6）The high CL of Japanese is due to less proficiency of the subjects.