关键词: 文本阅读, 回指解决, 类别距离, 几何距离, 情境模型

Abstract: Most researches agree that readers can build situation models of the information described by texts. The spatial dimension of situation model has been explored most often. Anaphora resolution is an important way to explore the representation of spatial distance in text reading. First, participants learn the layout of a building which contains several objects in different rooms, and then read narratives describing motions of the protagonist in this building. A motion sentence tells how the protagonist moves from one room to another. A target sentence including a definite noun phrase follows the motion sentence. The definite noun phrase refers to an object in the building, and this anaphor-like phrase will cause a memory search. Researchers have found that anaphora resolution depended on the spatial distance between the reader and the object. But there are two kinds of spatial distances: category distance and metric distance. The former is the number of rooms and the latter is the length of them. Most of the past studies used the number of rooms to represent distance, and did not take the length of them into consideration. Which distance caused the spatial gradient in reading times? Several studies had been conducted to explore this problem, but all of them could not give a satisfactory explanation.
In this article, anaphora resolution was taken as a technique model to explore the representations of spatial distance in text reading. Moving window method was used in our study. We hypothesized that both the category distance and metric distance could be represented by readers during reading, and they might produce independent influences on anaphora resolution. Sixty-six university students took part in these two experiments. All of the participants learned the layout for about twenty minutes first, and then drew it by memory. If there were more than three mistakes, they had to learn the layout again. At last, we asked them three questions about the layout. If all of these questions were answered correctly, they read narratives that described how the protagonist moved through the layout sentence by sentence on a computer monitor. They could read at their own pace by pressing the key. Reading time of target sentences was recorded and then analyzed by the computer.
Experiment 1 separated category distance from metric distance. The results showed that when the metric distance was the same, the category distance influenced the time of anaphora resolution. It took more time to find the referent when there were two rooms between source room and goal room. Experiment 2 further explored the effect of metric distance. It showed that when there was only one path room between source room and goal room, but the metric distance of the path room was different, readers established a representation of the metric distance. It influenced the time of anaphora resolution. The longer the metric distance of the path room, the more time it took to complete the anaphora resolution.
These two experiments showed that both category distance and metric distance had independent influence on anaphora resolution. Researchers should pay attention not only to the number of rooms but also to the length of them when explore the spatial distance of situation models. How these two kinds of distances interact is another important question for further discussion

Key words: text reading, anaphora resolution, category distance, metric distance, situation model