Optimization Of Collocates In English Collocation Dictionaries

To select collocates for a lemma in a collocation dictionary is an important step in editing the dictionary. When it comes to small- and medium-sized collocation dictionaries, both dictionary makers and researchers agree that collocates for a lemma should be selected on the basis of their frequencies in a corpus. However, as to how to select collocates for a lemma in large-sized collocation dictionaries, there are no studies on this issue. Therefore, the present study attempts to establish a model for selecting collocates for a lemma in large collocation dictionaries so as to help editor of this type of dictionaries to supplement more collocates that are useful to users. Firstly, the author made a survey of collocational information in the entries of all the 14 fruit words in The Kenkyusha Dictionary of English Collocation, and found that the collocational information provided in different fourteen entries is very imbalanced, and that the information of collocates for the same colligation pattern in different entries is also imbalanced. To solve these problem, the present author, in the light of Langacker’s theory of cognitive domain, analyzed the collocates of each of the 14 fruit words revealed by British National Corpus and established its matrix of domains. On the basis of 14 matrixes, a universal matrix of a fruit word was formulated. Then this universal matrix was used to select collocates for two lemmas to test its effectiveness. Then model for selecting collocates for fruit lemmas was established. The entries of two lemmas underpinned by the model proves that this model could provide a wider range of collocates for the editor of dictionary so that it to some extent could overcome the limitations of the method of completely depending on the principles of frequency to select collocates, and restraints imposed by the capacity of a corpus on selecting collocates.