Abstract: The boundary effect refers to the phenomenon of the reading time for the first sentence of a new discourse unit being longer than the reading time for any other sentence. This phenomenon was first discovered by Haberland and his colleagues (Haberland K F, Berian C & Sandson J, 1980). Other studies have also found evidences for this effect (Lorch R F, Lorch E P & Matthews, 1985; Bestgen Y & Vonk W, 2000). When reading, readers first try to integrate the new information with preceding information that has just been read and that still remains in the working memory. When a reader experiences difficulty in integrating new information with the preceding information, he/she will conclude that the topic has been changed and that a new mental representation needs to be constructed to lay an integral foundation for the incoming information. To maintain continuity, readers have to reinstate the preceding concept or make coherent inferences and each of these processes increases their cognitive load; at the same time, construction of a new mental representation also needs extra cognitive resources. Therefore, reading speed slows down, and reading time increases. Bestgen and his colleagues explored whether segmentation markers could eliminate this effect. Their results showed that a segmentation marker, for example, “around two o’clock,” could successfully eliminate the boundary effect. However, the segmentation marker in their study did not show the temporal relationship between the markers and the event preceding. In our opinion, when the markers are within the duration time of the event preceding them, boundary effect will be observed; however, when they are not within the duration time of the event preceding them, the boundary effect will be eliminated. This phenomenon is referred to as the event duration effect. In the present study, three experiments were conducted to examine whether the boundary and event duration effects are observed in Chinese text reading.
A moving window technique was used in the study. Experiment 1 examined whether the boundary effect could be observed in Chinese text reading. A 2(topic continuous/topic shift) × 2(event duration long/short) design was used. Forty-four participants were asked to read 16 topic continuous or topic shift passages, and the reading times for the target sentences were recorded. Experiments 2 and 3 were designed to test whether segmentation markers could eliminate boundary effect. The segmentation marker “half an hour” was located before the target sentences. A 2(topic continuous/topic shift) × 2(event duration long/short) × 2(with/without segmentation marker) design was used. In Experiment 2, 48 subjects were asked to read the passages with or without the segmentation marker. The reading times for the target sentences in the topic continuous condition and topic shift condition were compared. In Experiment 3, the temporal relationship between the segmentation marker and the event preceding was controlled to further test the event duration effect.
Experiment 1 showed that the reading time for the topic shift condition was significantly longer than that for the topic continuous condition irrespective of the duration of the event. Experiment 2 demonstrated that the reading time for the topic continuous condition did not differ reliably from that for the topic shift condition if there was a segmentation marker before the target sentence. However, the reading time for the topic shift condition was significantly longer than that for the topic continuous condition if there was no segmentation marker. Experiment 3 further replicated the results of Experiment 2 under strict control.
The results implied that the impact of temporal segmentation markers on the boundary effect depended on the temporal relationship between the markers and the event preceding, a phenomenon referred to as the event duration effect. In other words, when the markers are within the duration time of the event preceding, the boundary effect can still be observed, however, when they are not within the duration time of the event preceding them, the boundary effect is not observed. This shows that segmentation markers reduce the amount of processing required for the part of the sentence that is topic discontinuous

Key words: boundary effect, temporal segmentation marker, event-duration effect, Chinese text reading