Recent studies using the disappearing text paradigm have shown that reading in alphabetic writing systems is not impaired substantially if words remain visible for only a short period when they are fixated (e.g., English: Rayner et al., 2006; Finnish: Blythe et al., 2011).There are, however, important developmental differences in performance on this task, and findings indicate that developing child readers often must fixate more often on words, especially long words, whereas skilled adult readers do not (Blythe et al., 2011). The present study used the disappearing text paradigm in two experiments to investigate if there is a similar developmental difference in performance on the disappearing text paradigm when reading Chinese. These experiments recorded the eye movements of native Chinese developing child readers and skilled adult readers. Participants read sentences formed from regular sequences of two-character words and each sentence was between 7 and 8 words long. These sentences were either presented normally or in a disappearing text paradigm in which, as each word was fixated, the word remained visible only for a short period before disappearing. There were four disappearing text conditions: in Experiment 1, each word remained visible for 20ms, 40ms, 60ms, or 80ms following fixation onset; and in Experiment 2, each word remained visible for 20ms, 60ms, 100ms, or 140ms following fixation onset. Both experiments showed that there was no overall cost to reading times for either age group except in the 20ms display condition, which was longer compared to when sentences were shown normally. This was consistent with the findings from studies in alphabetic languages. However, both the child and adults readers produced different patterns of eye movement behaviour in the disappearing text conditions compared to when sentences were shown normally. First, there was a trade-off between refixation probability and fixation duration, such that both age groups of readers made fewer but longer fixations on words. In addition, both groups of readers were more likely to make regressions back to words in the disappearing text condition compared to when sentences were shown normally, indicating that both age groups were more likely to refixate words, to facilitate word identification, when words were shown only briefly. However, there was also a clear developmental difference in the use of this strategy and the developing child readers made more regressions back to words in the disappearing text paradigm, as compared to when text was shown normally, than did the skilled adult readers. This showed that, as in previous studies using alphabetic languages, developing child readers had more difficulty identifying words when these were visible for only short periods following fixation and often made a regression in order to reinspect words. The indication, therefore, is that developing child readers identify words more slowly than skilled adult readers. Finally, both age groups made initial fixations on words which were closer to the center of the two-character words in disappearing text conditions than when sentences were shown normally, suggesting that the disappearing text manipulation cued readers to the regularity in the construction of the sentences. To conclude, skilled readers of Chinese can identify words extremely quickly during reading. This is consistent with the findings from alphabetic writing systems. For less skilled developing Chinese readers, however, more time is needed to encode words and so these readers often have to reinspect words that are available only very briefly when reading disappearing text. In other words, the findings show that developing child readers' encoding of words is slower than that of skilled adult readers.
