ABSTRACT
In this paper, we analyze degrees of automatization in translations from German to English. In a translation experiment, we used German original sentences that include post-finite subjects of varying length and thus of varying informational load to investigate the preferred translation strategies for different informational structures. We hypothesized that the change to the order of subject and finite verb is the most automatic, cognitively least demanding translation and that less heavy German subjects will show a wider variety of translation strategies and an increase in cognitive effort during the translations, since their reduced informational value can be aligned more flexibly onto other grammatical functions in the target sentence. We triangulated keystroke logging and eye tracking data from twelve professional translators and tested the results using linear mixed regression modeling. Experiment participants were asked to translate ten paragraph-sized texts, presented in a pseudo-randomized order. Each text contained one stimulus, i.e. a heavy or non-heavy subject. The heaviness of the subject in the experiment did not have a significant effect on translation behavior, though the subject verb shift strategy was shown to be more automatic than the mapping of new semantic meaning onto the subject. Any other change to the constituent order in the translations was not significantly different from the assumed automatic translation in terms of cognitive effort.
