Abstract
Languages, like molecules, document evolutionary history. Darwin(1) observed that evolutionary change in languages greatly resembled the processes of biological evolution: inheritance from a common ancestor and convergent evolution operate in both. Despite many suggestions(2-4), few attempts have been made to apply the phylogenetic methods used in biology to linguistic data. Here we report a parsimony analysis of a large language data set. We use this analysis to test competing hypotheses - the "express-train''(5) and the "entangled-bank''(6,7) models - for the colonization of the Pacific by Austronesian-speaking peoples. The parsimony analysis of a matrix of 77 Austronesian languages with 5,185 lexical items produced a single most-parsimonious tree. The express-train model was converted into an ordered geographical character and mapped onto the language tree. We found that the topology of the language tree was highly compatible with the express-train model.
