The term "key clusters" is the same concept as that of keywords: statistically significant words extracted from a target corpus by using a log-likelihood test compared to a reference corpus (Baker, 2006). The only difference is that key clusters are expressed as multi-word sequences (Jhang & Lee, 2013, p.200). The purpose of this research is to compare clusters and key clusters and to examine the characteristics of key clusters in the English education corpora consisting of English Education Textbook Corpus, Foreign writers' research article corpus, and Korean writers' research article corpus. Four word lexical bundles were extracted by using Wordsmith 6.0 tools. For the analysis of key clusters, BNC baby was used as a reference corpus. The extracted key clusters were analyzed in terms of structural and functional types.
