A Study Of Collocation From The Perspective Of Network Of Cognitive Models

Lexical collocation is a critical benchmark of comprehensive mastery and use of lexical knowledge. Because of its own uniqueness it has been one of the biggest headaches for EFL learners, especially when making language production. Both home and abroad, over the last decades, lexical collocation has been drawing linguists'and educators'attention. Some analyses and researches have been done on it, however, the subject seems to have been dealt with quite sporadically. Most of these studies seem to focus on such superficial characters as part of speech, meaning; moreover, influenced by the constraints of Transformational-Generative grammar, collocations have always been regarded to be untypical or arbitrary word combinations. This makes it difficult to accommodate the study of collocation into a reasonable theoretical framework. The present thesis, by putting forward network of cognitive models in the field of cognitive linguistics as the theoretical framework of collocation study, aims at a more rational explanation of the essence of collocation phenomenon and revealing collocation learning difficulties confronted by Chinese EFL learners. At the same time, the causes of collocation's overuse, avoidance and incorrect use are discussed.From the aspect of theoretical assumption: an assumption is presented that the collocation is the linguistic expression of network of cognitive models (NCM), and the formation of collocation is selective. The formation is subject to factors such as word frequency, part of speech and so on. As NCM itself is universal, it is assumed that the collocation of different languages expresses the same NCM and their differences only come from the different lexical ways of expression. As to Chinese EFL learners, their English collocation is not in the English lexical way to express NCM, but in the Chinese lexical way through the parasitic strategy of"Translation Equivalence".From the aspect of methodology: on the one hand, by the use of FLOB corpus and CLEC corpus, 925 collocations of commonly used and randomly chosen 100 words are retrieved, including their frequency, part of speech, strength of collocation, in order to...