Lexical bundles in engineering research articles : materials development for Thai engineering students

Sanmuang, Kamonchanok and Tusting, Karin (2025) Lexical bundles in engineering research articles : materials development for Thai engineering students. PhD thesis, Lancaster University.

Text (2025Sanmuangphd)
2025Sanmuangphd.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (11MB)

Abstract

The aim of this study is to investigate the potential and impact of using analyses of lexical bundles to support Thai engineering students in learning to write research articles, as well as to develop a sample of materials for engineering students to use when writing research articles. This research focuses on 200 3-word and 4-word bundles extracted from a 400,000-word corpus compiled from a collection of 100 research articles on engineering. The Top 50 and Bottom 50 3-word and 4-word bundles were analysed to identify their structures and functions. Frequency and dispersion range were used as the main criteria in lexical bundle identification (Conrad & Biber, 2005). Additionally, inter-rater reliability was used to enhance validity and reliability. The study uses the structural and functional classification of lexical bundles in the Longman Grammar of Spoken and Written English from Biber, Finegan et al. (1999). The structures of the top 50 lexical bundles were categorised into Noun Phrase-based (NPbased), Preposition Phrase-based (PP-based), Verb Phrase-based (VP-based), Clausal fragments and other expressions. Preposition Phrase-based (33%), VP-based (28%) and NP-based (26%) appeared to be the most frequent lexical bundles. Other expressions (that did not fall into the previous categorisation) (7%) and clausal fragments (6%) occurred less than the three groups mentioned. Regarding the structures of the less frequent lexical bundles (the bottom 50), the study found that bundles were in three main structural categories: incomplete noun phrase fragments (65%), incomplete dependent clauses (30%), and other expressions (5%). Regarding discourse functions, referential expressions (49%) and discourse organisers (34%) of the top 50 lexical bundles were more frequent than stance (16%) and others (1%). The study employs structural and classification results to formulate/frame the materials. The results of the bottom 50 revealed similar proportions to the top 50: referential (39%), discourse organisers (29%), others (27%), and stance (5%). The inter-rater reliability of the structural categorisation of the top 50 and bottom 50 was 97%. The inter-rater reliability of the functional classification of the top 50 and bottom 50 was also 97%. For instructional purposes, five of these 200 bundles from the top 50 were selected for teaching 13 electrical engineering students. With the aid of AntConc software, the students worked with lexical bundles at their own pace, with help from the researcher whenever necessary. A pre-test and a post-test along with student interviews, students’ notes, classroom observation and a teacher’s diary were used as the main research instruments to assess the effectiveness of the process of using materials for using lexical bundles developed in the study and classroom instruction in lexical bundles. The post-test scores for the writing task were significantly higher than those in a pre-test. The results demonstrated that the students showed a more positive attitude towards learning English in general and revealed more awareness regarding their use of lexical bundles employed in the engineering field. The findings from the lexical bundle analyses shed light on the methodology for analysing lexical bundles, which pays more attention to not only frequent lexical bundles but also less frequent lexical bundles that are more relevant to the subject domain. Pedagogically, the experiment suggested that there is a need to incorporate lexical bundles and data-driven learning in teaching subject-specific writing. This study could be replicated with other groups of students in subject-specific areas.