A Cognitive Functional Model Of Possession Clauses

Based on the Neurocognitive Linguistics, and by referring to Functional Grammar and Frame Semantics, this present paper aims at constructing a cognitive functional model of possession clauses. This model is a relational network, which simulates the operation of neurons in the brain at the abstract level. The emphasis is laid on the operational validity and the neurological plausibility.To build the model, three steps are involved: analysis, catalysis, and testing. To analyze is to decompose a linguistic structure into different structural components. In this thesis, the analysis is based on the clauses of predication structures in the possession domain. Based on the evidence obtained from analysis, the model is constructed through catalysis. To catalyze is to build up. The model is built by virtue of a marking system created by Lamb. The result of catalysis is a model of relational network, which covers the conceptual stratum and the grammatical stratum of the linguistic system. To prove that the model is workable, testing is required. Computer programs are employed in testing, which mainly aims at simulating the relational network of the linguistic system. In the testing, the production is processed from the conceptual constructions to the grammatical constructions.Analysis, catalysis, and testing having been done, we are able to summarize the findings we've got. Firstly, in the possession domain, conceptually, the basic types of possession clauses are space structure, action structure and causation structure, among which the action structure is always conflated with the causation structure. Secondly, due to different choices of Topic, one conceptual construction may be realized by different grammatical constructions. Therefore, the model we constructed demonstrates the realizational relationships between the conceptual constructions and the grammatical constructions. Thirdly, the result of testing is satisfying: the input and the output of the model are supported by the linguistic phenomena, and the operation process of the model is not incompatible with that of the neural network. Therefore, the operational validity and the neurological plausibility are guaranteed.