Chinese Character Recognition Test, The Multiple Read-out Model

In various fields of Chinese cognition, Chinese characters recognition has been focused on in past 20 years. In the meantime, many researchers at home and abroad are proposed constructive model to explain the processing of word recognition. The series search model, logogen model, interactive activation model, parallel distribution process model, etc. are more influential. There are advantages and disadvantages when these models explained the study results of word recognition. Therefore, new models have been proposed by some researchers, hoping to explain the existing research conclusions better. In visual word recognition, Grainger and Jacobs(1996) proposed multiple read-out model because of the inconsistent results of frequency and amount influences on orthographic neighborhood effects and the model can account for the inconsistent results. So far, the applications and tests of the model are based on the alphabetic language, but Chinese characters are very different from alphabetic writings. So it takes opportunity for us to explore the uniqueness and universality between languages. If the assumption of the model is proved in Chinese visual word recognition, it shows that the processing mechanism assumed by the model is common in different languages and we will have a better understanding in Chinese visual word recognition.This study includes two experiments and each of them has two branch experiments. In the first experiment, the moderate wordlike nonwords are included and the two branch experiments have same materials. The word decision task is used in the two branch experiments but the different is task demand. In experiment 1A, a feedback will be given when subjects make a wrong decision so the accuracy is stressed. This experiment explores whether more inhibition from a masked high frequency neighbor will presence and shows that subjects make a decision based on the information from lexical access when accuracy is stressed. In experiment 1B, a reaction time feedback will be given after subjects make each decision. This experiment explores whether more facilitation from a masked high frequency neighbor will presence when speed is emphasized and show that in this condition subjects make a decision mainly based on information from global activation amount in mental lexicon. In the second experiment, different wordlike pseudoword(nonword) are used in two branch experiments in order to explore the task demand of influence on a shift in the response criteria when stimuli feature is different. Experiment 2A contains high wordlike pseudoword and subjects may be inclined to the information from lexical access when they make a decision. So this experiment explores whether more facilitation from a masked high frequency neighbor will still presence in the stress-on-speed condition. Experiment 2A contains low wordlike nonword and subjects may be inclined to the information from global activation when they make a decision. So this experiment explores whether more inhibition from a masked high frequency neighbor will still presence in the stress-on-accuracy condition.The results of this study mainly have two points:the first is the multiple criteria concept is valid and it can explain the inconsistent results in past studies. In those studies, some researchers observed facilitate neighborhood effects and some observed inhibit neighborhood effects. This phenomenon may be relative to the shift in the response criteria. The second is considering the stimuli feature, part of effects predicted by the multiple read-out model is absence because of the unique two-dimension structure of Chinese characters. Henceforth, researchers should focus on proposing some models which are fit for Chinese visual recognition.