CRITT poster at AMLaP in Turku

Arnt Lykke Jakobsen & Kristian T.H. Jensen will present Eye-to-IT research results at the AMLaP conference on 27 August 2007.

Eye movements were tracked, keystrokes logged, and voice was audio recorded as each of twelve translators performed two reading tasks and two translation tasks, one written and one spoken. The key variable in the reading tasks was reading purpose. One text was read for comprehension and one with a view to being translated. Both translation tasks were based on written source texts presented on screen. The key variable studied here was the difference in modality (reading-writing versus reading-speaking) in the two tasks. The language pair involved was English-Danish.

Overall a gradual increase gaze time and fixation count was observed across the four tasks. Striking differences in eye movements appeared as a result of the text alignment process in translation tasks. In the oral translation tasks, translators’ eye movements showed an increased number of regressions slowing down reading, but very few in comparison with what was the case in written translation. Here the translator’s eyes have to constantly keep track of two texts, a process which disrupts fluent reading, tends to disorient the translator, and slows down production very considerably.