关键词: 文本阅读, 先行词, 回指词, 回指推理

Abstract: This paper explores the influence that three text factors: the referential distance between anaphor and antecedent, the elaboration of distractor and the typicality of distractor to anaphor. O’Brien et al (1997) used a rapid serial visual presentation paradigm at a rate of 250 ms/word and found that there is no facilitation when there was a substantial distance between anaphor and its referent. Levine et al (2000) found that it was the elaboration of distractor led to the failure of anaphoric inference, not the anaphoric distance. however, they didn’t separate the elaboration and the typicality of the distractor. Basing on Minimal hypothesis, the semantic interference would influence the accessibility of the antecedent: the match between the cue and target. Our hypothesis is that the typicality of the distractor is the main influence factor to anaphoric inference . By manipulating the elaboration of distractor and the typicality of distractor, we studied their effects on the anaphoric inference. Two experiments were administered.

Method
Ss were 164 university students. The on-line window display technique was used. The time of responding to the probe words in different conditions was analyzed. There were respectively 15,12,14 and 16 passage in experiment 1a,1b,2a and 2b. All experiments were within-subject design. In experiment 1, by using a line-by-line reading paradigm and recognition probe measures, the study explored the effect of distractor’s elaboration on anaphoric inference. In experiment 2, by using the same test measures as experiment 1’s, and adopting an 16 category typicality assessing method the study explored the effect of the relative change of the distractor’s typicality and the antecedent’s typicality to anaphoric inference . Statistical analysis was done by SPSS.

Results
Experiment 1 shows that antecedents (eg. “cake”) were fastly accessed when an anaphor (eg. “dessert”) was read on the background of the extensive elaboration of same-category low distractor (eg. “bread”). Low distractor’s elaboration didn’t hinder the accessibility of antecedent thus consequently led to failure of anaphor resolution. Meanwhile, the effect of referential distance was tested independently and we found that inference distance was not sufficient to eliminate anaphor resolution. This results support the finding of Levine et al (2000). Experiment 1 also showed that elaboration of distractor was not a main influence factor to anaphor resolution.
Experiment 2a shows that anaphor was not resolved on the condition of high-typicality of the distractor. High-typicality in long-distance condition and in short-distance condition was not significantly different to the anaphor resolution. Experiment 2b also shows that high-typicality was the main factor to the anaphoric inference. the results of Experiment 2 is the change of antecedent typicality led to the effect of anaphor resolution.
Conclusion
Basing on experiment 1 and 2, we found that the main factor was the typicality of distractor— the semantic overlap between the anaphor and antecedent.
The study results suggest that increasing the typicality of same-category distractor will cause anaphoric inference resolution failure. This findings can’t be explained by O’Brien’s(1998) Resonance Model. According to Resonance Model, all concepts currently in short-term memory serve as cue can be activated when anaphor is read. To explain the findings above, one reason is the anaphor needn’t be resolved. For example, it can either be used anaphorically or to introduce a new concept. Another reason is Mckoon and Ratcliff’s(1997) Minimalist hypothesis. It appears that readers didn’t draw anaphoric inference because the antecedent was not easily available and the inference was not necessary from comprehension

Key words: text reading, antecedent, anaphor, anaphoric inference