Interactive Heatmaps as an Improved Means of Analysing Complex Socio-dialectal Patterns: German Loans in Silesian

ABSTRACT This paper presents an application of interactive cluster heatmaps in sociolinguistics, a method hitherto scarcely employed in the field. To that end, we developed a statistical workflow to illustrate the method and analyse large-scale Silesian questionnaire data. In our quantitative-linguistic study we demonstrate how heatmaps can uncover information about complex patterns of regional variation, thereby highlighting the added value of the method relative to standard statistical procedures. Specifically, we show (i) how differences in language use between two regions can be determined by deploying the heatmap method but not with traditional significance/hypothesis testing statistical procedures or summary statistics, (ii) and how differences in cohesion and tightness in clusters can be examined via heatmaps but not using cluster analysis. We conclude that heatmaps are a valuable tool for assessing why and how certain word-items group together because of the regional distribution of their usage. A major advantage of the heatmap method is that it can handle two dimensions with hundreds of instantiations and illustrate their interrelations, which would pose problems for traditional statistical techniques. Heatmaps provide a novel and accessible way of exploring large-scale sociolinguistic data in their entirety and of generating further hypotheses.