This study aims to retrieve the most frequent 4-word clusters in a self-built corpus of academic papers from SCI written by Chinese scholars, i.e. SCI-CH Corpus and to analyze the structural characteristics of these clusters from the perspective of both types and tokens. Besides, this paper makes contrast with native scholars the use of the frequent 4-word clusters in the four sub-corpora of the SCI-CH Corpus. It is found that the most frequently used clusters on the whole are those that incorporate prepositional and verb phrase fragments; the most frequently used clusters in each sub-corpus by Chinese scholars differ significantly in tokens from those used by native scholars, especially clusters from noun phrase with of-phrase fragment, prepositional phrase with of-phrase fragment, the active and passive structures and adverbial clause fragment. Therefore, this study can provide some guidance for Chinese scholars to write academic papers at higher levels.