Language networks and reconstructed protolanguages

Through application of the linguistic comparative method historical linguists are able to posit "protolanguages", hypothetical ancestor languages which explain the variation in the observed descendants. These protolanguages have some special properties distinguishing them from attested natural languages. Some of these properties derive from processes of horizontal transfer within earlier branches of the family, as demonstrated by Nelson-Sathi et al.(2010), and others are artifacts of other evolutionary dynamics of the system, such as processes of semantic change accompanying lexical diversification. In this paper I will present a revised phylogeny of the Indo-European family of languages, built using Bayesian phylogenetic analysis on the basis of a substantially corrected and expanded version of the Dyen et al. (1992) Swadesh lists. I will test to what extent phylogenetic uncertainty corresponds to network structure in the family, and show how and to what extent the linguistic comparative method reconstruction of proto-Indo-European relates to this tree structure