Abstract: Text reading is one of the most complex and unique cognitive activities of human and is an important way of getting information and knowledge. The mechanism of information processing in text reading has been a hot topic in experimental psychology, and several models have been proposed. According to the memory-based text processing view, readers activate the previous information and integrate it with the current information by resonance in text reading. If the previous information has local-incoherence with the current information, coordinating integration could then be occurred. The key issue involves the mechanism of the coordinating integration in text reading. In our view, the coordinating integration could occur only when the previous information has event-relatedness with the current information. Furthermore, coordinating integration is hypothesized to be an automatic process so that readers are not aware of the integration process during their reading. The main purpose of the present study was to investigate the mechanism of the coordinating integration in text reading.
Three experiments were conducted using a self-paced, line-by-line, reading paradigm. In experiment 1, 30 university students were asked to read 12 narrative passages. The independent variable involves the relationship between the elaboration on a characteristic of the first protagonist and two subsequent target action respectively carried out by two protagonists. The independent variable had two levels-consistent version and qualified version. The dependent variable were the reading times of the first target sentence describing the actions of the second protagonist and the second target sentence describing the actions of the first protagonist. In experiment 2, 28 university students were asked to read 12 narrative passages and to judge whether a probe word appeared in the previous section of the text. The relationship between the elaboration on a characteristic of the first protagonist and subsequent target action carried out by the second protagonist was the first independent variable, which had two levels-consistent version and qualified version. The position of the probe word was the second independent variable, which also had two levels- appeared before or after the target sentence. The reaction time and the accuracy of the probe word were the dependent variables. In experiment 3, 66 university students were asked to read 1 narrative passage, and then give answers to two questions. The relationship between the elaboration on a characteristic of the protagonist and subsequent target action carried out by the protagonist was the independent variable which had three levels -consistent version, inconsistent version and qualified version. The dependent variables were the reading times of target sentence and control sentence, the accuracy of recollection of key information in the text. In addition, the participants’ self-report was also analyzed.
In experiment 1, for the first target sentence, there was no difference between the consistent and qualified conditions, and for the second target sentence, the reading time in the qualified condition was longer than that in the consistent condition. In experiment 2, the reaction time for the probe word after the target sentence was shorter than that before the target sentence in both the consistent and qualified conditions. The accuracy of the probe word after the target sentence was higher than that before the target sentence in both the consistent and qualified conditions. In experiment 3, the reading time of target sentence in the inconsistent condition was longer than that in qualified condition and consistent condition, the reading time of target sentence in qualified condition was longer than that in consistent condition, and for the control sentence, there was no difference among consistent, inconsistent and qualified conditions. In addition, the accuracy of recollection of key information in the inconsistent condition was higher than those in qualified condition and consistent condition, while there was no difference between the consistent and qualified conditions. More important, the participants in the inconsistent condition reported inconsistent information, while the participants in the consistent and qualified conditions reported no inconsistent information.
The present results showed that previous information could be activated when related with current information semantically; however it was integrated with current information only if it is related in events. Furthermore, coordinating integration is an automatic process and readers are thus not aware of the integration process during reading

Key words: text reading, information processing, coordinating integration, mechanism