Morphosyntactic dialect variation, once a neglected area of dialect research, has recently witnessed a large growth in interest. Various methods from geospatial data analysis have been applied to morphosyntactic data. To date, the focus has largely been on analysing the distribution of stable patterns of variation. This article extends this work to examine patterns of ongoing change. It uses a body of data from the Syntactic Atlas of Welsh Dialects and the Siarad Corpus of spoken Welsh to examine the innovation and diffusion of a new secondperson singular pronoun, chdi, testing the usefulness of Geographically Weighted Regression (GWR) as a method for identifying and modelling patterns of ongoing syntactic change. It is shown that GWR provides plausible models of the diachronic development of changes that are still in progress. Furthermore, it allows us to test whether rates of change are constant across geographical space, allowing us to test whether the Constant Rate Hypothesis (that diffusion of change proceeds at the same rate in different environments) holds between dialects.