ABSTRACT
This chapter focuses on the use of statistical techniques in corpus-based discourse analysis. It outlines the connection between statistics and discourse analysis and proposes a simple three-step procedure for more rigorous discourse analysis; this procedure is designed to guide researchers through the analytical process, and focuses on important features of the scientific protocol. The chapter offers statistical techniques of calculating inter-rater agreement, which make the categorisation process more reliable. It also outlines the concept of meaning relations in discourse and introduces the technique of collocation and collocation networks. The chapter explains inferential statistical techniques such as statistical tests and confidence intervals and demonstrates their application in the analysis of discourse. Making informed statistical choices is an essential skill for a researcher, ensuring reliability and validity of a quantitative study. The basic statistical choices related to collocation involve the span, collocation statistic and statistical and frequency threshold.
