Effect Of Predictability On Parafoveal Processing In Chinese Reading: An Eye Movement Study

We conducted two experiments,four sub-experiments,to explore predictability the during the Chinese normal reading with the boundary technology in two experiments.In each experiment,the preview of a character was either correct or was visually dissimilar and meaningless but resembling a Chinese character;the predict of the target was either high or low.We define the predictability in Chinese as the local constraints and global constraints.In experiment 1,to avoid the effect of the local constraints,we used single-character words in Chinese,which can be only predicted by the preceding context.Experiment 1a manipulated the predictable and the preview type of the unfixed parafoveal characters in normal reading.We also set up the position of an invisible boundary before the target word.Results indicate that word predictable increased the parafoveal preview benefit slightly,but there is no parafoveal-on-foveal effect in skipping rates for the high prediction.This finding can be explained by models of eye-movement control during reading that N and N+1 can be could be processed in serial.Experiment 1b,we set up the position of an invisible boundary before the pretarget word,The results showed no N+1 preview effect,but N+2 preview.We hypothesized that the high prediction in Chinese single-character is processed serial.In experiment 2,we added the local constraint as adjoint variable.In experiment 2a,using the boundary technique,an invisible boundary was set between the two constituents.We observed higher preview effect in high prediction.Moreover,we found parafoveal-on-foveal effect for in the high prediction but not found in low prediction.In experiment 2b,we set up the position of an invisible boundary before the first constituent.We observed higher preview effect in high prediction.Moreover,we found N+2 preview benefit effect for in the high prediction but not found in low prediction.The results showed the compound words of the high prediction could be processed in parallel.This finding can be explained the reason that the second constituent be affected by two types of the prediction: local constraints and global constraints.