Although linguists have thoroughly studied the formation of language areas for given dialectal phenomena, little quantitative research has been conducted on how these areas relate to each other, and how the transition between these dominance areas of dialectal variants can be modelled. We propose gradient estimation methods used in GIScience to answer the key question to the analysis of such dialectal boundaries: to what extent we can find crisp boundaries in a dialectal landscape (termed ‘isoglosses’ in linguistics) or whether the transitions are rather gradual. Our methods are also aimed at the comparison of these boundaries. We apply trend surface analysis and regression analysis to Swiss German syntax data and test our methods on dialect phenomena with typical variation exhibiting the spatial competition of two variants. We conclude that subdividing the linguistic landscape of the given competing variants into three subregions (two dominance zones for each of the two variants and a transition zone between them) and calculating regression models in these subregions lets us quantitatively compare their relationships to each other and to other linguistic phenomena.