This paper is concerned with sketching future directions for corpus-based dialectology. We advocate a holistic approach to the study of geographically conditioned linguistic variability, and we present a suitable methodology, 'corpusbased dialectometry', in exactly this spirit. Specifically, we argue that in order to live up to the potential of the corpus-based method, practitioners need to (i) abandon their exclusive focus on individual linguistic features in favor of the study of feature aggregates, (ii) draw on computationally advanced multivariate analysis techniques (such as multidimensional scaling, cluster analysis, and principal component analysis), and (iii) aid interpretation of empirical results by marshalling state-of-the-art data visualization techniques. To exemplify this line of analysis, we present a case study which explores joint frequency variability of 57 morphosyntax features in 34 dialects all over Great Britain.
corpus-based dialectology; holistic approach; corpus-based dialectometry; feature aggregates; multivariate analysis; visualization techniques
