This chapter gives a general overview of the research topic, the motivation for the study, and its general theoretical and methodological embedding. It articulates the research questions of the study and provides an overview of the following chapters.

This chapter retraces the main analytical steps in the book and the theoretical insights gained from the analysis and provides an outlook for future research conducted in a similar spirit.

This chapter discusses the data and method applied in the present study. It gives a concise overview of the ICE corpus project and the ten national sub-corpora that enter the analysis as well as describing how the Twitter (TwICE) corpus was sampled. Following this, the catalog of 236 linguistic features extracted from the corpus data is introduced. The chapter concludes with a description of the individual steps and parameter settings of the statistical procedure and a bird’s eye view of the resulting space of variation.

This chapter presents the first three dimensions of linguistic variation developed in the statistical analysis. The interpretation of these dimensions can largely be derived from differences in modality and attendant communicative-situational properties.

This chapter discusses the remaining four dimensions, which are less easily subsumed under a logic of either modality or conventions of a narrowly defined register. For each of these, explanations are provided that have to do with drifts in discourse conventions, cultural differences, and grammatical peculiarities across varieties of English.

This chapter introduces perspectives on the analysis and quantification of linguistic variation, comparing variationist sociolinguistics, corpus-based text linguistics, and the multifeature, aggregational approach associated with dialectometry and MD analysis. The choice of the MD framework is justified and the central steps involved in such an analysis are outlined.

This chapter presents the next three dimensions. These differ from those presented in Chapter 5 in that they are dominated by a single register, i.e. they show particularly high values for one specific kind of corpus text.