Search

This article introduces a text mining tool that can automatically extract information using Hyland's analysis model as a theoretical framework to analyse the use and characteristics of metadiscourse in large quantities of academic texts. To verify its validity, we present the results obtained using this tool on various bachelor's degree theses with a particular focus on the field of engineering. Our results on a 6.9 million-word corpus extracted from 680 bachelor's theses available online show that interactive metadiscourse markers are prevalent in engineering bachelor's theses as well as in the authors' metadiscourse patterns. In addition, we compared our results with previous research on metadiscourse markers. Our study can be used to identify the usage of various types of metadiscourse markers during the production of texts and for the development of software applications involving quantitative linguistic methods for the production of academic texts. [ABSTRACT FROM AUTHOR]