A Connectionist Approach To The Process Of Consecutive Interpreting

Based on connectionism, the present study, considering the complexity of human mind, attempts to construct a process model of Consecutive Interpreting (CI) which consists of a parallel distributed pattern of information processing with suggestions for improvement of connectionist theories. In order to show the whole processing model, three key layers concerning an effective and efficient CI model are elaborated. Firstly, emphasis is placed on the lexicon-grammar layer, arguing that both the lexical subsystem consisting of chunks and the grammar subsystem interact with each other showing a parallel distributed pattern for the formation of sense. This is the most creative point because it is a reaction away from the previous linear description of the process. Secondly, stress is put on the proposal of integrating the pragmatic layer into the lexicon-grammar layer and the de-verbalized sense layer to ensure idiomatic, natural and proper interpreting. The proposal is intended to improve connectionist theories in that the evolutionary nature of human beings should be taken into account. The third layer is the de-verbalized sense layer, which emphasizes the abstract feature of sense, and concepts as the base of sense to indicate the difference between CI and SI in terms of time allowed.Then, a seven-layer model is introduced and a working pattern is explained with the support of the feed-forward model and the back propagation algorithm. The hypothesis about the working pattern is that information and knowledge are processed and stored in the connection of neurons. The stimulation of related connection can ensure successful operation, while the inhibition of related connection would hinder operation, thus costing more effort to go through other connections to find clues, or would lead to the failure of operation. Meanwhile, in the discussion of the processing model, implications for CI teaching and training are also given.