Eye movement control during reading: fixation measures reflect foveal but not parafoveal processing difficulty.

The main purpose of this study was to determine whether, during natural reading, the difficulty of the upcoming parafoveal word affects eye movement behaviour on the currently fixated word. A model in which visual attention is allocated in parallel over both the fixated and the upcoming parafoveal word predicts such as effect, while a sequential attention allocation model in which attention is directed first to the fixated word and then to the upcoming parafoveal word, does not. The data reported here show that neither the frequency nor the combined length, frequency and class of the upcoming word affect eye movement behaviour on the current word. These data support the sequential attention--parallel programming model of eye movement control in reading.