| The Connectionism has been the leading theory in cognition psychology since the 1980s later period. It is widely applied to each domain includes the linguistics. The distributional representation of connectionism and the rationality to explain studying mechanism are widely accepted by the language acquisition researchers. At present, the connectionism already became one of the leading schools that interpret the language acquisition mechanism.As for the connectionism theory and the artificial neural network model are mature,and the connectionism successfully introduced into language acquisition researcher in the overseas ,we introduce distributional representation theory,semantic extracting method and self-organizing clustering method into the research of Chinese as second language acquisition . We would try to use the new research view points ,and in the same time renew method.In the article ,the superiority of connectionism regarding the language acquisition was introduced .We have used the connectionism automatically withdraws semantic model WCD to do a series of automatically extracts semantic experiment from the modern Chinese corpus and the intermediary language corpus ,and then we do self-organization clustering analysis to the semantics, and take the result as the mental lexicon of foreign student and people of Chinese as mother tongue referencing the connectionism, inspect their similarities and differences and the developing process. The experimental study discovery, the primary level foreign student can not identify the synonymy and isomorphism of compound words. Along with the level enhanced, the compound words of same family of foreign student's grasping slowly increases , their semantic system gradually develops. To the high-level level, the foreign students have been able to recognise the synonymy and isomorphism of compound words of same family well, but still inferior to Chinese mother tongue user's semantic system. As to the reason ,it lies in that the compound words of same family of the foreign student to grasp is few, the frequency of use of partial glossaries is low, simultaneously the usage unavoidably is unitary. It is thought that , in order to truly grasp the compound words ,their grammar, semantics and pragmatics characteristic must be comprehensively grasped. The experimental result also indicated that, this way processing language materials with connectionism can process the problem that traditional corpus research process difficultly by the frequency analysis, the example retrieval analysis, and can discover some questions that traditional researches do difficultly. For example the model can carry on the comprehensive synthesis computation to the semantics and pragmatics of the language materials, and collect context information which the person is very difficult to induce. Moreover, in the article the discussion which the compound word semantic transparency automatic quota evaluates had made, the semantic transparency automatic appraisal formula supposition was proposed. The experimental result indicated, the supposition of computation of semantics distance on semantic representation withdraws by the model, and the supposition of the semantic transparency automatic appraisal are certainly rational and feasible.In this article,we uses the connectionism model to process the intermediary language material, and probed one method that completely differs from the traditional analysis to language material.We do self- organization clustering analysis and suppose it as the semantics mental lexicon to discusses that the foreign student study compound words . It is also different from traditional way . We hoped that this kind of discussion is beneficial. Simultaneously we also hoped the method in the experiment proposed to improve model training,the supposition of computation of semantics distance and the supposition of computation of semantic transparency have certain value. |
PDF Full Text Request