Abstract: The present study examined two major models on text comprehension (i.e., resonance and scenario-based). The stimuli included passages about two people in which the characteristic of the supporting role was inconsistent with the subsequent target action carried out by the protagonist. Participants read passages line by line at their own pace and made judgments about probe sentences that were consistent with the characteristics of the supporting role. The probe sentences were presented at three possible locations: after the description of the supporting role’s characteristic, after the filler paragraph but before the first target sentence, and after the first target sentence. In Experiment 1, with the presence of the supporting role in the filler paragraph, the two characters both were the threads of the story. The result showed that subjects recognized the probe sentences following the target sentences faster than those preceding the target sentences, which means that when subjects read a sentence about the protagonist engaging in some action, early characteristics that described the supporting role were reactivated. In Experiment 2, without the presence of the supporting role in the filler paragraph, only the protagonist became the thread of the story. The result showed that early characteristics that described the supporting role were not reactivated by the target sentences. Based on the results of two experiments, we suggested that though we can use resonance to explain the accessing of backgrounded information in text comprehension in some way, but the resonance does not seem to be unlimited. The semantic overlap only offers a possible way to access backgrounded information, whether the information can be activated or not is likely to be constrained by reader’s representation or situation model on earlier portions of text.

Key words: text comprehension, resonance model, scenario-based model, situation-constrained