SUMMARY
The purposes of this corpus-driven study were to compare the use of four-word lexical bundles between native English and Thai dissertation writings. Two language corpora, roughly 1,000,000 words apiece, were gathered from dissertations in the field of English Language Teaching written by both groups of writers. Each corpus was subdivided into three sub-corpora, namely, the Introduction, the Methodology, and the Results and Discussions sub-corpora. Two frameworks employed for the structural and functional analysis of the four-word lexical bundles were Salazar's (2011) adaptations of Biber et al.’s (1999) and Hyland’s (2008a). The analysis of lexical bundles was performed using concordance software AntConc. The results showed that Thai writers overused lexical bundles in comparison with that of English speaking writers in each part of the dissertations, especially in the Results and Discussion section, which could result from institutional factors such as expectation and practice of Thai universities that expect Thai Ph.D. students to be more critical of the findings and to offer more implication. The structural analysis revealed the overuse of verb-structured lexical bundles throughout the three sub-corpora of Thai writers, which was likely to stem from the non-native speaker's failure to employ noun- and preposition-structured lexical bundles effectively. The proportion of functions of lexical bundles in each section of dissertations written by both groups of writers shared a relatively similar trend, indicating that English speakers and Thai writers conformed to the same convention of dissertation writing.