Inferring sound changes using Bayesian MCMC

In this paper we analyze dialect phonetic data using Bayesian Monte Carlo Markov Chain inference (MCMC), in recent years one of the most powerful and the most successful methods in molecular phylogeny for inferring the relationships between species. This method enables us to infer the historic relationships between the language varieties, but also to explore historical accounts of the sound correspondences in the data set. The dialect divisions obtained by applying Bayesian MCMC inference are compared to the divisions described in traditional scholarship. In this experiment we also test three different models of vowel evolution and show which changes are most likely within the vowel space.