Abstract: Situational model is one of the hottest issues in researches on discourse comprehension. At present, it is widely agreed that narrative comprehension involves setting up a mental representation of the states of affairs described by the text. According to the event-indexing model, which accounts for situational model in details, events are the core units of the mental representation, and they are indexed on five situational dimensions: time, space, protagonist, causality, intentionality. Time is an important dimension for constructing situational model, and a number of previous studies demonstrated readers tracked time information while reading texts. Some of them provided indirect evidence supporting that readers segmented the consecutive narrative by time into a series of discrete events in order to understand it. Speer and Zacks (2005) directly proved it with explicit event segmentation paradigm. Specifically, participants were asked to segment the points where they believed one meaningful unit ended and another began while they read some narrative texts describing everyday events. Their experiment results suggested that participants segmented event boundaries based on time adverbials. However, analyzing carefully with their experimental materials, we found that their demonstration failed to distinguish between event factor and time factor, because a new event appeared at object – time sentence. According to the event-indexing model and event segmentation theory, the whole event unit played original role in situation updating. Given the above consideration, in the present study, event segmentation paradigm was adopted to investigate event segmentation in narrative comprehension in Chinese narrative comprehension in three experiments. Participants were required to segment each narrative no less than three times. Moreover, they were required to list a title to each narrative in order to make sure that they were reading carefully. Experiment 1 was to repeat Speer and Zacks’ research results in 2005 in Chinese. Experiment 2 was to explore how readers segmented narratives without temporal adverbials, Experiment 3 further explored how readers segmented narratives in the condition of dissociating the time factor and the whole event unit, namely inserting temporal adverbials into the whole event unit. The results in experiment 1 showed that participants were more likely to segment them according to temporal adverbials; Furthermore, participants preferred segmenting them before disconnected temporal adverbials（“an hour later”）to before connected temporal adverbials(“a moment later”), which repeated the results of Speer and Zacks in 2005. However, the results in experiment 2 showed the participants more frequently segmented them according to the whole event unit, indicating that they mainly segmented the narratives according to the whole event unit. More important, the results in experiment 3a revealed that readers remained more to segment them according to the whole event unit, and there were not significant differences between disconnected and connected temporal adverbials The results in experiment 3b further replicated the results in experiment 3a. Taken together with experiment 3a and 3b, the results showed the temporal adverbials also facilitated readers’ event segmentation to some extent. The present findings above indicate that the whole event units are the core units of the mental representation constructed for a narrative; time shifts lead to situation model shift only in the condition of the time indexed the event shifts.

Key words: Narrative, Situation model, Event-indexing model, Time Shift, Event Segmentation

摘要： 采用事件分割范式探讨了中文条件下时间维度在记叙文理解中的作用。实验1在中文条件下重复了Speer和Zacks(2005)年的研究，结果发现读者更多将时间短语所在位置作为分割边界，表明时间转换在情景模型更新中发挥了重要作用。实验2探讨了在没有时间短语条件下读者对文本分割的情况，结果发现读者主要是按照事件单元进行分割；实验3a和3b在文本的事件单元中插入时间短语，形成时间转换与事件转换分离的条件，探讨读者在这种时间转换与事件单元转换不一致的条件下所采用的分割策略，结果表明，读者更多仍然按照事件单元转换进行分割，而且对事件单元转换的依赖强于时间转换，但同时时间短语仍对文本的分割有一定影响。上述结果表明，事件单元作为建构记叙文心理表征的核心单元，时间维度作为事件单元转换的线索，只有在标识事件转换的情况下，才能引发读者情景模型的更新。

关键词: 记叙文, 情景模型, 事件指针模型, 时间转换, 事件转换