This book is in two parts, the first of which presents papers on unified anthropological themes and the second, coevolutionary approaches to some anthropological questions, involving the integration of findings from more than one anthropological field.
Three chapters are case studies in linguistic phylogeny. Despite what the back cover copy says, none of the contributors seems to be a linguist per se (though several worked on projects involving linguistic material), and it shows. Simon Greenhill & Russell Gray's “Dating population dispersal hypotheses: Pacific settlement, phylogenetic trees and Austronesian languages” uses lexical material from Robert Blust's as yet unpublished Austronesian comparative dictionary and combines it with distance-based phylogenetic computer packages such as NeighborNet, in order to test the probability of the accuracy of five hypotheses so far advanced regarding the settlement of the Pacific by waves of speakers of Austronesian languages. Their Austronesian trees make several grave errors; for instance, they link Chamorro and Palauan together, but separate Niuean from Tongan, bundling Tongan together with Samoan, which they separate from its Nuclear Polynesian sister Rennellese, and they fail to recognize that Cebuano and Hiligaynon are more closely related to one another within Bisayan than either is to Kapampangan. This article provides three major Austronesian trees, all different, and all full of historical improbabilities.
Matters improve in “Comparison of maximum parsimony and Bayesian Bantu language trees” by Clare J. Holden, Andrew Meade & Mark Pagel. This chapter presents (broadly similar) trees compiled according to the methodologies outlined in the title, and relies on lexical evidence for 95 Bantu and Bantoid languages. But even here several languages, such as Mambwe, move around considerably within the tree branches depending upon whether Bayesian or other techniques are used, and whether morphological as well as lexical criteria are invoked.
“Untangling our past: Languages, splits, trees and networks” by David Bryant, Flavia Filimon & Russell D. Gray focuses on the modern Indo-European languages. Their data source comprises the Swadesh 200-word lists compared in Isidore Dyen, Joseph Kruskal & Paul Black's An Indoeuropean classification of 1992 (Philadelphia: Transactions of the American Philosophical Society, 82[5]), covering material from more than 80 modern Indo-European languages. Some findings of the 1992 study were startling: Sets of pairwise percentages showed English and the Surinamese creole Sranan Tongo (nonexistent before c. 1650!) as earliest offshoots from Germanic; Aromunian alone rather than it and Daco-Romanian as first offshoot of the Romance languages; and no especially close historical relationship between Indic and Iranian languages. All of these findings, reproduced here on p. 82, demonstrably are ahistorical nonsense.
My overall conclusion is that phylogenetic software designed to construct family trees can produce sets of different-looking and mutually incompatible trees. Unfortunately, many software packages used here consistently fail to identify sets of exclusively shared innovations, which are the gold nuggets of linguistic phylogenetic work, and thereby misidentify the closest relatives of many languages examined. Certainly the character-based trees in various chapters of this book give much more solid results than the often wildly inaccurate (and inexplicably trendy) distance-based approaches. This book should provoke plenty of controversy.
