Splits or Waves? Trees or Webs?

Tree models are used in both linguistics and genetics for inferring population history. The trouble with them is that human populations do not really evolve (either genetically or culturally, as in language), tree-like, but rather exchange both genes and words.

Linguistic evolution has been mostly described in terms of tree models, but languages are not insulated from each other, and they interact after their initial differentiation. This interaction is facilitated by geographic proximity, and also by linguistic proximity.

Geographic proximity makes it possible for speakers of different languages to talk to each other, learn each other's languages, or develop hybrid languages or a lingua franca. Linguistic proximity facilitates communication: it is fairly easy, for example, for speakers of Germanic languages to interact, and much more difficult for those of, say, English and Chinese.

If speakers of a language become separated by distance or geographical barriers, then lateral exchange between different groups becomes minimal, and language evolution can be well-described by a tree model. If, on the other hand, there exists a language continuum across a wide area, effected by a common process (say, the spread of agriculture), then there is room for substantial cross-interaction of different emergent languages at the stage when they can be still thought as dialects of the parent language.

While the current paper's focus is on Germanic languages, the endgame seems to be on the much harder and more vigorously contested field of Indo-European studies.

The author has put up a nice supplementary page online on a first attempt of using NeighborNet with an Indo-European dataset, pictured on the right. A publication on the topic is listed as being in preparation:

The utility of Germanic as a case-study is that it provides a (reasonably) known external history against which to assess our methodological approaches. On the strength of the findings here, a similar logic can now be extended to probing the unknown of how the early divergence history of Indo-European unfolded. In the full exploration in Heggarty (in preparation a), it transpires that even the data underlying figures 1 and 2 here suggest an early divergence along the lines of a dialect continuum. And for all the purported analytical elegance of binary branches, as a real-world demographic scenario it is this Indo-European continuum that offers the more straightforward and economical explanation. A splits-then-borrowing scenario has instead to invoke not just a complex series of divergent migrations, but then later movements to attenuate this by bringing certain groups back into contact again. This in turn entails consequences for which of the main rival hypotheses—the migratory Kurgan ‘horse culture’, or the progressive demic diffusion of agriculture—best fits as the driving force that shaped the pattern of the earliest Indo-European expansion.

Phil. Trans. R. Soc. B 12 December 2010 vol. 365 no. 1559 3829-3843

Splits or waves? Trees or webs? How divergence measures and network analysis can unravel language histories

Paul Heggarty et al.

Linguists have traditionally represented patterns of divergence within a language family in terms of either a ‘splits’ model, corresponding to a branching family tree structure, or the wave model, resulting in a (dialect) continuum. Recent phylogenetic analyses, however, have tended to assume the former as a viable idealization also for the latter. But the contrast matters, for it typically reflects different processes in the real world: speaker populations either separated by migrations, or expanding over continuous territory. Since history often leaves a complex of both patterns within the same language family, ideally we need a single model to capture both, and tease apart the respective contributions of each. The ‘network’ type of phylogenetic method offers this, so we review recent applications to language data. Most have used lexical data, encoded as binary or multi-state characters. We look instead at continuous distance measures of divergence in phonetics. Our output networks combine branch- and continuum-like signals in ways that correspond well to known histories (illustrated for Germanic, and particularly English). We thus challenge the traditional insistence on shared innovations, setting out a new, principled explanation for why complex language histories can emerge correctly from distance measures, despite shared retentions and parallel innovations.

Link