Figure 10.

Task Phases. Data collection began on a daily basis with calibrations of the EyeLink tracker and the AVR4 gaze subsystem (top row). EyeLink calibration was performed using their 13-point method which presents each point exactly once, but allows the experimenter to repeat the suite should there be a poor calibration. That initial calibration provides a highly accurate signal to the AVR4 software via a dedicated ethernet link. A subsequent calibration of the AVR4 gaze subsystem uses a 9-point method with repeated measurements at each point to provide consistent scaling and reduce any residual errors. Overall, both phases of calibration took no more than two or three minutes. Once successful calibration had been obtained, the experiment began (second row). A central fixation point was presented to the subject which the subject foveated in order to engage the trial. A brief while later, the stimulus was presented and the subject allowed an indefinite amount of time to read each sentence before indicating they had finished by looking at an exit point (Done) near the top center of the screen to proceed to the next trial of any remaining. A total of 40 sentences were presented in each session. For stimulus text presented in the clear, the reading period typically lasted only a few seconds, while for difficult conditions, the reading period could take 45 seconds. It was never the case that the subject took more than 2 minutes to read any sentence. Sessions lasted no more than 30 minutes, and, with the development of reading skill, were 15 minutes or less for later sessions with a given subject.