中文摘要：

通过三个实验探讨元理解监测精确性的延迟效应及其机制。结果显示：相对于即时写关键词和不写关键词，延迟写关键词可以显著提高元理解监测的精确性；延迟的时间内写关键词相对于读关键词更利于元理解监测精确性的提高；延迟的时间内前测自测相对于前测学习更利于元理解监测精确性的提高。这表明时间的延迟和延迟时间内的主动认知加工对元理解监测精确性的提高至关重要。

英文摘要：

Metacomprehension monitoring involves monitoring how much they understand a text and predicting future test performance on questions related to recently read texts. Metacomprehension monitoring plays an important role in reading comprehension Accurate metacomprehension monitoring is crucial to effective metacomprehension control. However, previous research has showed that metacomprehension monitoring is far from accurate, which is operationalized as the intra-individual correlation between a person＇s comprehension ratings and his or her test performance. Variables that affect the metacomprebeusion monitoring accuracy include the types of texts, the judgment of monitoring, the processes involved during and after reading that may provide cues for monitoring. Recent studies have shown that metacomprehension accuracy could be dramatically increased by having participants read texts and then perform certain generation tasks （summarizing or key word listing） that captured the essence of each text prior to judging comprehension for it. In this study, three experiments were conducted to investigate the delay effect of metacomprehension monitoring accuracy and its mechanism. 105 college students participated in Experiment 1. The participants were divided into delayed-generated-key word group, immediate-generated-key Word group or no-key word group. After reading all the six expository texts on a computer, they rated their level of comprehension for each text. Finally, all participants took a criterion test, through which their monitoring accuracy was measured. In the test, three questions assessed text-based knowledge, and three inferential questions assessed knowledge of a situation model. The results showed that there were significant main effects of key words on the magnitudes of JOL and bias score, the immediategenerated-key word group was higher than the delayed-generated-key word group and the control group. Moreover, metacomprehension accuracy was dramatically greater for the group of students that wrote key