The Routledge handbook of historical linguistics ed. by Claire Bowern and Bethwyn Evans (review)

Numerous collaborative handbooks of particular subfields of linguistics have appeared in the past couple of decades, and all are useful.¹ This one is exceptionally useful because of its unusually wide coverage. Among its welcome innovations are discussions of: language change in terms of compositionality, modern computational phylogenetics, the stability of typological features of grammars, syntactic change and reconstruction (from more than one viewpoint), change in signed languages, the relation between language acquisition and language change, and several other topics. The contributors have typically illustrated their discussions with a wide range of examples, many of them unfamiliar to most linguists—a welcome change from the old-fashioned reliance on familiar textbook examples.

I have neither the expertise nor the space to discuss all thirty-four contributions individually; I thus discuss those that I have some hope of evaluating.

Paul Kiparsky’s overview (Ch. 2) should be required reading for anyone currently working in historical linguistics. Kiparsky manages to discuss or mention an astonishing range of current lines of work in historical linguistics. Many lines of work are reported without comment or evaluation, but the comments that Kiparsky does make are thought-provoking; a typical example is his observation that the transfer of Turkish verb morphology into two dialects of Cappadocian Greek almost certainly occurred when the dialects were moribund and can be accounted for by that fact (p. 87, n. 8).

The other overviews are interesting because they are unconventional. Roger Lass’s discussion of the history of historical linguistics shows that tracing a single thread that leads to modern theory and practice does not have to be tendentious. Nigel Vincent focuses on the intersection of language change and compositionality, a topic that, as he notes, has been almost entirely overlooked.

Because the comparative method (CM) is central to the methodology of historical linguistics, every handbook needs a chapter that explains it in terms accessible to nonspecialists and illustrates it with a broad range of examples. Michael Weiss’s contribution meets that need admirably. Mark Hale’s complementary chapter does not call the fundamentals into question; it does show, however, how a naive understanding of the CM, or of uniformitarianism, can lead to apparent paradoxes and other problems.

Chs. 6 through 8 on the processes of language diversification are extensive, and rightly so. Alexandre François discusses models of diversification in great detail, especially in light of Malcolm Ross’s seminal work, and illustrates his discussion with interesting original research. He points out that networks are a better model of language diversification than trees in a large majority of instances, but I think he misses the one advantage of the Stammbaum model: it is clearly falsifiable, which makes it a good initial scientific hypothesis, even though one soon comes to expect that it will in fact be falsified more often than not. It should also be pointed out that not all types of network can be converted into phylogenies without further information; in particular, NeighborNet and other ‘data display’ networks do not provide any direct information about evolutionary history (cf. Morrison 2011:47).² Michael Dunn provides a good overview of computational approaches to the recovery of linguistic phylogenies, noting the strengths and weaknesses of each. I think that he too misses a methodological point. The usual objection to distance-based methods is that the conversion of a character matrix into a distance matrix discards information about how each character partitions the tree; some consideration of that criticism in his discussion of ‘When to use distance, and when not to use it’ (195–96) would have been welcome. Finally, SØren Wichmann reviews work on the stability of typological characteristics of languages, much of it very recent. Perhaps his most interesting conclusion is that ‘structural features do not preserve more ancient phylogenetic signals than does the basic vocabulary; … they are more prone...