Readability Analysis of Textual Content Using Eye Tracking

Characterization of silent reading involves a joint analysis of the reading material and reader’s capacity to assimilate the content. The amount of mental workload, to understand a content, reflects the overall cognitive load which is commonly measured using electrophysiological signals. Eye movement provides a direct means of evaluating one’s reading characteristics in much finer details. In this paper, we use a commercially available low-cost eye-tracking device, EyeTribe. We analyze the eye movement behavior to compare readability of textual contents and also characterize an individual’s reading profile. The experiment is performed using two types of textual contents significantly separated in difficulty level as evaluated using standard readability indices. The Flesch-Kincaid Grade Level for one type varies between 3 and 6 and the other between 13 and 16. Eye gaze analysis is done on features related to fixation and saccade using both global and local features. Results indicate the difference in content is reflected in global features and individual level variations, for a given type of content, are observed in the entropy derived from local features.