Simulating Studies Of EFL Learners' Chinese Orthographic Awareness Development Based On Self-Organizing Feature Map Network

It is a simulating study concerning the foreign student's orthographic awareness development based on the model of self-organizing feature map. The study can be summarized as following four aspects:The first aspect of the study is concerned with the statistic analysis of implicit knowledge of Chinese characters implicated in the text. The analysis indicates that the implicit knowledge has a potential influence on the recognition ability and the acquisition of Chinese orthography of CRCC model. Therefore the cognitive effects can be attributed to the influences from the implicit knowledge of characters in the simulating texts and the training samples.The second aspect of the study deals with the architecture and the training of the CRCC model we have built. The CRCC model is composed of Kohonen's SOM network and James and Miikkulainen's SARDNET. Based on the CRCC model three sub-models are built for representing the three levels of foreign language learners. The results of the model training indicated that the three sub-models successfully simulated human brain's accumulating process of knowledge due to the change of distribution of character knowledge structures and the creation of the frequency accumulating mechanism.The third aspect is involved with the simulating study on the model's cognitive effects of Chinese characters. The three sub-models show that there are orthographic and structural type effects when performing the recognition tasks of true and false characters and non-characters. In addition, we revealed the models' character recognition strategies based on analysis of errors the models made.The fourth aspect is concerning a discussion about the results of simulation. In accordance with the theoretical exploration several conclusions are made as follows: (1) implicit knowledge of characters plays an important role in acquiring the orthographic knowledge and developing the models' recognition ability. (2) the results of the CRCC model training indicate that the connecting mechanism of old and new knowledge and frequency accumulating mechanism are imperative mechanisms for the model's knowledge generation and accumulation. If the model is lack of the mechanisms the catastrophic consequence will occur. (3) the model's character recognition indicates that there are structural type effects of characters and the effects have an influence on the acquisition of orthographic rules. (4) the error analysis of characters suggests that the model's first cognitive strategy is component recognition, and the recognition of structural types is the second one. (5) there is consistency to a certain extent between the model and behavior experiments in cognitive effects of characters. That is to say, the model's cognitive effects are significant for Chinese character teaching. Hence, the suggestions for the teaching are proposed.