Trees, languages and genes: A historical path

The “tree metaphor” has been used in linguistics since the nineteenth century. The first to use it was Schleicher (Die Darwinische Theorie und die Sprachwissenschaft. Weimar: Böhlau, 1873) who, with his Stammbaumtheorie, supposed that we can build a family tree for languages for which we have exact knowledge as Darwin did for plants and animals. Schmidt (Die verwantschaftsverhaltnisse der Indogermanische sprachen. Weimar: Herman Bohlau, 1872), a Schleicher’s student, opposed the Stammbaum model and proposed a “wave model” of language change: he supposed that language change spread in waves emanating from some epicenter. This model bears an obvious resemblance to the demic diffusion model used more recently in biology (Menozzi et al. Science, 201, 786-792, 1978). Phylogenetic trees, where taxa separate only on a cladistic basis, have been put into discussion both in linguistics and in biology and a “web model” has been proposed instead (Heggarty et al. Diachronica, 27(2), 301-324, 2010). In a landmark paper published in 1988, Cavalli Sforza et al. reported a figure directly comparing a human genetic and linguistic tree based on Greenberg (Language in the Americas. Stanford: Stanford University Press, 1987). Although extremely controversial, this paper highlighted the similarities between processes of historical inference in population genetics and linguistics, as well as the potential importance of linguistic data for inferences about human population history. While biologists have embraced computational phylogenetic methods, quantitative approaches in linguistics, such as Swadesh’s (Proceedings of the American Philosophical Society, 96(4), 453-463, 1952) “lexicostatistics” or Greenberg’s “mass comparison”, were heavily criticized (cfr. Campbell. Historical linguistics: An introduction (2nd ed.). Edinburgh: Edinburgh University Press, 2004). Quantitative approaches have been recently reconsidered. Today, researchers using computational methods in evolutionary biology and historical linguistics (such as Network and Neighbornet) aim to answer similar questions and hence face similar challenges.