SCIENTIFIC CORRESPONDENCE colonization of Australia more than groups) on the same graph as groups specific analysis to within-species varia­ 40,000 years ago as the earliest indica­ when looking at the relationship be­ tion. In considering the relationship be­ tion of language-based skills. Recent tween two sorts of variation is bound to tween linguistic and genetic variation results appear to put the event before produce an apparent correlation. In this there are no clear-cut clades, even at the 5 6 53,000 years · . Theoretical and empiric­ case, if one eliminates the four super­ terminal twigs of any branching model , al evidence shows that radiocarbon groups from the graph, the apparent so that even a plot of independent data substantially underestimates a~es even correlation is entirely eliminated. points, as opposed to the hierarchically 7 beyond tree-ring calibrations · . Paired JEREMY J. D. GREENWOOD organized ones that were used, is unlike­ radiocarbon and other radiometric ages British Trust for Ornithology, ly to be truly independent. The purpose (see figure) show a consistent trend, The Nunnery, Nunnery Place. of the plot was not to read too much into predicting that conventional radiocarbon Thetford, Norfolk IP24 2PU. UK the significance levels but to examine at estimates of 40,000 years for first a broad level the comparative rates of colonization of Australia approach the FOLEY REPLIES - One of the clearest genetic and linguistic divergence. current claim, by thermoluminescence, points made by Noble and Davidson in Whether there is a linear relationship in of 53 ,000 years. their paper2 was that the presence of such data is a matter of probability, not is the only means of language should be directly rather than inevitability, which affects the inter­ calibrating rates of hominid evolution. indirectly inferred, hence their stress on pretation of the statistics but not the There are no independent criteria for the 40,000-year datum. Their arguments validity of the general observation. As I 1 estimating the time gap between modern relating to Australian colonization seem stated in News and Views , there are human morphology and modern human at variance with this position. Either good historical and geographical reasons behaviour. We need a clear perception Noble and Davidson are recoupling why more 'recent' clades should show of what lies on either side of the gap. behavioural and anatomical evolution, greater variation. Over time these may lAIN DAVIDSON arguing that because we know only ana­ be evened out. This effect can be seen Department of Archaeology and tomically modern humans reached Au­ by plotting the limited data on Renetic 1 13 Palaeoanthropology, stralia, and that this event was at least distance and language diversity · at WILLIAM NOBLE 53,000 years ago, therefore language higher (continental) through to lower Department of Psychology, must be older still; or else that com­ (regional) levels; the higher the level, University of New England, Armidale, munication is essential for such a jour­ the greater the relationship, the lower New South Wales 2351, Australia ney, which is special pleading for hu­ the greater the scatter. Regrettably, of mans given the occurrence of coloniza­ course, the higher the level the smaller SIR - There is a striking correspond­ tion events such as playrrhine monkeys the sample size and hence the lower the ence between 'phylogenetic' trees of in the New World in the Oligocene. significance. There is, though, some evi­ human linguistics and those of human Such special pleading is closer to the dence in microevolutionary studies14 for 11 genetics . This has been used in a News traditional approaches that they were at such a positive relationship to exist. and Views article1 to support the view pains to resist. In the end we are left with the that both trees reflect the historical di­ Personally, I am perfectly happy to observation that there is some rela­ vergence of human populations. Leaving believe that hominids have been making tionship between genetic and linguistic aside the many general problems associ­ both remarkable and unremarkable divergence. To determine whether this ated with the hennigian reconstruction of utterances for the past 100 ,000 years. relationship is the product of common 12 phylogeny , there is a particular prob­ Definitions of language that are not ancestry as I implied or subsequent lem here: the method assumes that there simply tautologically self-referential to population interaction, as Greenwood is no interchange between branches of humans remain a problem, but such favours, requires more formal models the tree once they have separated, yet definitional nuances do not circumvent that might show how these two inter­ human populations that are very diffe­ the problem of what earlier and contem­ pretations would differ empirically. In rent both genetically and linguistically porary () hominids with very the absence of such formal models all we may exchange both genes and language high encephalization quotients were have at the moment are the observations components. Knowing insufficient about doing with their brains. My suggestion and some ad hoc generalizations (my language to be sure, I wonder whether was that distinguishing language as the own included). such exchange may not be a major basis for thought from language as com­ ROBERT FOLEY reason why the so-called phylogenetic munication might provide a way forward Department of Biological , trees are so similar. that is consistent with both ethological , But I am quite sure that the graph in and archaeological data. Far from bypas­ Cambridge CB2 3DZ, UK Foley's article 1 used to illustrate an sing the issue of proto-language, this aspect of the correspondence between implies that the structure of what be­ 1. Foley. R. A. Nature 353. 114-115 (1991). 2. Noble. W. & Davidson. I. Man J. R. Anthrop. lnst. 26. language and genetics presents nothing came communicated language evolved as 223-253 (1991). more than a statistical artefact. It shows systems of thought that were then 3. Cheney. D. L. & Seyfarth. R. M. How Monkeys See the World (Chicago Univ. Press. 1990). a strong correlation between the number grafted on to animal communication sys­ 4. Lieberman. P. The Biology and Evolution of Language of languages in human groups and the tems through the evolution of speech. (Harvard Univ. Press. Cambridge. 1984). Turning to Greenwood's letter, it is 5. Roberts. R. G.. Jones. R. C. & Smith. M.A. Nature 345, between-subgroup genetic diversity with­ 153 (1990). in those groups. But only four of the perhaps a truism to state that the prob­ 6. Hiscock. P. Archaeol. Oceania 25. 122-4 (1990). eight groups presented are statistically lem in investigating the biology of hu­ 7. Barbetti. M. & Flude. K. Nature 279. 202-205 (1979). 8. Stuiver. M. eta/. Quat. Res . 35. 1-24 (1991). independent; the others are formed by man evolution is that there is only one 9. Bell. W.T. Archaeometry 33. 43-50 (1991). successive hierarchical grouping of these species. This imposes major methodolo­ 10. Bard . E.. Hamelin. B. . Fairbanks, R. G. & Zindler. A. Nature 345. 405-410 (1990). (and of a ninth group for which no data gical problems, particularly when we 11. Cavalli·Sforza L. L. Piazza . A.. Menozzi . P. & Mountain. are shown). Unless two groups do not attempt to extend analyses into extra­ J. Proc. natn. Acad. sci. U.S.A. 85, 6002- 6006 somatic patterns such as language, and (1988). differ in the variation within them, a 12. Benton. J. J. Trends Ecol. Eva!. 5, 393-394 (1990). supergroup formed by combining them Greenwood rightly draws attention to 13. Ruhlen. M. A Guide to the World's Languages (Stanford will show more variation than either of some of these in my News and Views University Press. 1987). 14. Neel. J. v .. Layriss. M & Salzano. F. M. in Population them in all characteristics, so the inclu­ article. In particular they arise from Structure and Human Variation (ed. Harrison, G. A.} sion of supergroups (and supersuper- applying methods developed for inter- 109-142 (Cambridge University Press. 1977). 404 NATURE· VOL 355 · 30 JANUARY 1992 © 1992 Nature Publishing Group