An Experimental Study Of Scalar Implicature Processing In Chinese

This paper reports experimental study of cognitive processing of scalar implicature in Chinese language. The present experiments are intended to test the two processing models:the Default model proposed by the neo-Gricean theorists and the Context-Driven model supported by the post-Griceans, had long been debated in the field of pragmatic research. The study consists of three on-line experiments:(1) scalar implicature generation in neutral contexts;(2) the picture-sentence verification;(3) and the on-line experiment of scalar implicature in Upper-Bound and Lower-Bound context. They were conducted in sequence in order to testify the two models in details by using the psychological experimental software DMDX at the Laboratory for Pragmatics and Cognitive Linguistics of the School of Foreign Languages and Literatures, Lanzhou University.Before the experiments, the validity of the Chinese language materials were verified with questionnaires and interviews, thus the reliability of the experiment should be guaranteed. At first, the experiment of Brehenyet al.(2006) was replicated in order to investigate whether scalar implicature can be generated in neutral contexts. The method of self-paced reading was adopted. By manipulating the scalar trigger "Some of the Xs (yixie)" in sentence initial and sentence final conditions, the reading times on the target phrase "the rest/the others (qiyude/qitade)" were supposed to be facilitated in sentence initial, because the topic position can create certain "contextual issues" as old information.Totally27native speakers of Chinese participated in the experiment. Results showed no main effect of interaction effect of the two factors. However, a main effect of sentence position was revealed in the reading time on the segments following the target phrase, which might be explained by the reason that scalar implicature was generated in sentence-initial condition. So the Context-Driven model was supported.Then a picture-sentence verification of under-informative utterance was performed in order to test the cost of generating scalar implicatures. The pictures created scenes about "informativity". Two factors:Quantifiers <all, some> and Picture-Match/NoMatch were combined to form a2*2within subjects design. Participants were asked to judge "yes/no" to the sentences.36native speakers of Chinese participated in the on-line experiment. Results showed significant effect of the two factors and interaction effect. In QSPA condition, response time of "No" was significantly longer than "Yes" response. On the assumption of longer response time indicating more processing effort, the Context-Driven model was supported in terms of processing cost. That is, the generating of the scalar implicature is more effortful. The split yes/no responses show that native speakers of Chinese are sensitive to scalar implicature.As an innovation, the third experiment was designed to test the processing of scalar implicature in Upper-Bound and Lower-Bound context. Participants were asked to decide whether a reward should be granted to the speaker according to certain scenarios. The Context-Driven model predicated longer response time in the Upper-Bound context. While the Default model were right in contrast with this.25students from the Lanzhou University participated in this experiment. Results were in favor of the Context-Driven model, since the longer response time in the Upper-Bound context than that in the Lower-Bound context and control items. Although no interaction effect was confirmed, the main effect of the relationship could be verified. In other words, the social and situational factor might affect the processing of the scalar implicature, which was confirmed by the post-experiment interviews. So the purpose, background, situation of the communication and the relationship of the speaker and hearer were all constraint factors for scalar implicature processing.Above all, results of the three experiments are mostly in agreement with the Context-Driven model with a little variation. The processing of scalar implicature was effortful and costly. Scalar impliplicature was generated in Upper-Bound context but not in the Lower-Bound. Inferred from those results and discussion, the processing of scalar implicature is rather a dynamic and complex processing mechanism in which the interaction of psychological, and reality factors, and language forms, is involved. The interpretation of scalar implicature depends on the contexts, i.e., background knowledge, world knowledge, previous experience and mental or epistemic state.