Item

Abstract

Languages are evolving systems, and so phylogenetic comparative methods allow us to produce probabilistic reconstructions of past states of languages. These reconstructions make it possible to measure dependencies in the evolution of language structures, that is, the statistical "coupling" of evolutionary changes in one linguistic feature with changes in another. The existence of correlated evolution of aspects of language structure is unsurprising; the novelty of this way of thinking about language change is that it allows independent inference of so-called language universals from lineage-internal data (data from a single language family). Phylogenetic comparative methods also reveal details of the processes of change in dependent evolutionary structures, e.g. directionality of change, causation. The startling outcome (to me at least) of pilot studies has been how differently putative language universals pan out in different lineages. Language variation does not seem to be governed by universal systems constraints reflected in universal dependencies. Rather, variation is the product of cultural evolution, constrained by affordances particular to the lineage, so that regularities in evolutionary outcomes are predicted by complex interactions within related linguistic systems.