An Exploration of Native Speakers' Eye Fixations in Reading Chinese Text

We collected the locations of eye fixations of Chinese native speakers when they read four Chinese articles, and attempted to analyze how the contextual linguistic and personal information influence the landing positions within the landing sites. In addition, we employed machine learning techniques to build models for the prediction of the landing positions. The models performed well for the closed tests, achieving 78% in accuracy in predicting whether a reader's eyes landed on the first or the second character within a word that contained two characters. Unfortunately, the accuracy for the same task in the 10-fold cross validations dropped to 60%, indicating the necessity of more future work.