BAYESIAN PHYLOGENETICS FOR LANGUAGE PREHISTORY-AND ARCHAEOLOGY

Among various approaches to correlating archaeology and language(s), Bayesian language phylogenetics is highly promising in principle—yet controversial in practice. It can provide a framework of a language family's divergence phases, and aspires to estimate their chronology, to help identify which processes in the archaeological record best correspond. Results have been dogged by inconsistency and artefacts, however, so some cautionary tales here first identify past failings, and new solutions. We review the whole pipeline from raw language data through encoding to Bayesian phylogenetic analysis and results. We then focus on how to interpret those results specifically against archaeology. We assess what splits in the trees actually mean; the pros and cons of ancient and proto-languages; and signal on deep prehistory. Throughout, we frankly assess the state of the art, and explore how Bayesian phylogenetic models might be more closely tailored to how descent with modification applies specifically to language families.