4.5　Cluster

Cluster refers to a group of words gathered or occurring closely together.（Scott 1999）It is also called lexical bund le（Biber＆Barbieri 2007）or lexical chunk（Lewis1993）.Using WordSmith Tools，clusters are extracted from OETT and TETT respectively.The list of clusters is of critical importance for the study of collocation and colligation.Both collocation and colligation tell about the co-occurrence or co-selection.Firth（1957：182－183）sees collocation as“actual words in habitual company”，and colligation as“the interrelation of grammatical categories in syntactical structure”.In this sense，clusters can be taken as a way to find the lexical patterns and test the idiomaticity of language use.What's more，as clusters are“imm ediate constituents”of“constructions”（Gleason 1961：132－133），they contribute for the syntactic and textual analysis.Certainly，“lexical bund les provide a kind of pragmatic head for larger phrases and clauses”（Biber＆Barbieri 2007）.

Four-word-c lusters that occur more than three tim es are extracted by Wordsmith Tools.The statistics is filtered and classified manually before data mining.As we aim to discuss lexical patterns and the idiomaticity of language use but not themes，proper names such as“the Em pire State Building”，“Great Wall of China”and“the UNESCO world heritage”are exclusive.The filtered data is listed in Table 4.15；overall clusters are listed in AppendixⅢ.

Table 4.15　Classification and Frequency of Clusters

(https://www.daowen.com)

（To be continued）

As a whole，there is a significant difference in cluster use between OETT and TETT（LL－40.29，P＜0.001）.TETT has a big lead over OETT in number（968＆708），but they show different priorities.OETT highlights evaluation，time and suggestion but TETT strengthens background，description and location.The details will be further discussed in the following.