Using fuzzy clustering to reveal recurring spatial patterns in corpora of dialect maps

Abstract

In this article, a new method to identify groups of spatially similar dialect maps is presented. This is done by comparing statistical properties ofthe maps: the empirical covariance is measured for every map in a corpus of dialect maps. Then, the Fuzzy C-Means clustering method is applied to these covariance data. Thereby, one is able to detect and measure gradual similarities between maps. By employing the method on lexical data from the dialect atlas Sprachatlas von Bayerisch-Schwaben, it can be shown that clusters of spatially similar maps also share semantic similarities. This method can thus be used for grouping maps based on spatial similarities while at the same time indicating patterns of semantic relationships between spatially related variables.