Are gaze shifts a key to a translator’s text segmentation?

Keystroke logging has demonstrated that a translator’s text production can be broken down into units separated by pause boundaries (Dragsted 2004, 2005, 2010). Reading research has not identified analogous boundaries, as the only interruptions in a reader’s visual attention to a text are often only blinks. However, in an experimental setup with tracking of a translator’s gaze movements across a screen showing the source text and (emerging) target text, gaze data show the translator’s shifts of visual attention between the two texts. Can such shifts be seen as an index of content processing units? And do such shifts give us more accurate information about segmentation or more information than keystroke intervals? Using a rather poorly calibrated recording of just one translator’s translation of a single sentence (within a longer task) for illustration, the paper seeks to tentatively explore the feasibility of identifying segments, understood as processing units, on the basis of gaze shifts, and to inquire into what motivates gaze shifts. It also seeks to illustrate how much our interpretation of gaze representations, not least suboptimal representations, depend on a theory of reading.