Newsletter of the Association for the Study of Language In Prehistory

Issue 25 · Summer 1995

A Comment on the Austric Prehistory

Blust's theory of Austric pre-history strikes me as highly sophisticated, in the Oceanist tradition. Still it is similar to those of Renfrew, Cavalli-Sforza, and Ruhlen on Indo- European. The reasoning is profoundly ecological and archeological, but seriously linguistic too, i.e., informed by Sapir and Swadesh! Yet it does seem that he had a Procustean Bed -- the probable Neolithic revolution in southeast Asia -- into which he molded the sub-groups of the Ur-superphylum, their homelands, their movements, and the dates for all of these. Maybe 'molded' is too strong a verb?

As concerns the dispersal of the various sub-groups and the testimony they bear about the ultimate homeland, I suggest that Robert take another look at J.P. Mallory's analysis of Indo-European. Since taxonomy makes such a big difference to dispersal theory, he ought also consider Gérard Diffloth's last remarks in Mother Tongue, worrying about the divergence of Munda. Such would force PAA much closer to India, it seems.

More importantly, if he accepts AT and Miao-Yao is half of AT, residing in eastern China as he says, then AT's homeland might lie smack in the middle of the Neolithic unveiled recently in east China. Of course, Miao-Yao has been a problem for every taxonomist in the region, so Robert has our sympathy here.

On the question of dates we have a case of archeologically probable dates and a big 'epistemic correlation'; leaping to their linkage with proto-languages. Since that is what we do all the time, the real question is where his linguistic dates come from. The archeology? So the archeology is the source of the archeological dates and the linguistic dates? As one can see the circularity of that reasoning quite easily, once it is pointed out, let us point it out! Blust unfortunately seems not to trust linguistic dating other than Paul Thieme style, which we might call 'the logic of reconstruction'. If PAN had a word for rice and settled villages, then the date of rice's domestication in settled villages gives the date of the proto-language. This is the Afrasian problem with Natufian all over again; the linguistic date is chosen because it fits the archeology well. But said linguistic date has no integrity of its own.

If we use glottochronological estimates, a percentage as low as 7% argues for 13+k years of separation, as Kruskal, Dyen and Black see it, or between 6k and 8.5k depending on the use of 80% vs 84% (using "t = logC _ 2logr") in 'standard formulations' or by Greenberg's calculations. (8500 is Joos-based). So PAN may be as young as 6000 years or as old as 13,000 years. Given the greater realism of the Joos formula, I would guess that PAN is 8500 years old -- at least. It may be that linguistic dating usually fails in Melanesia because of severe borrowing problems with Papuan but that does not necessarily mean that it fails in the empty Pacific, especially Micronesia where we can also get down to 7% or thereabouts.

Just as a hunch, seeing that Benedict has established that just about no common vocabulary exists between AN and AA, I would guess that lexical retention between any two samples from either phylum would show percentages down around zero. Since Afrasian gets almost that low, but usually not quite, then I would also guess that PAU is older than Ur-Afrasian. If you will recall Swadesh's calculation in an early issue of Mother Tongue, note that 1% (one percent) retention should yield 22,000 years of separation. That is by the inventor's standard calculations. Actually by Joos, Greenberg only gets 20,000 years at 1%.

Even if these calculations are wrong by a quarter, say 5000 years, still the result of subtracting that from 20-22,000 would be 15-17,000 years. Surely that is much older than any Neolithic we know of. And surely one must be more wary of 'epistemic correlations' (F.S.C. Northrop, 195-. Original date and definition lost to memory. It is like 'circumstantial evidence', only more refined.)

Finally, quibbling aside, I think Blust's paper is a powerful boon to our endeavours. By moving Austric out of the traffic jam and by proposing matching prehistory, he will keep lots of scholars busy checking it all out. Assuming for the nonce that Miao-Yao was supposed to be included in his Austric, we can say that Robert has enormously simplified the linguistic picture in a crucial part of the world. Oceania and its mainland now have no less than three major superphyla, Austric, Indo-Pacific and Australian. Plus the major phylum (Sino-Tibetan) which may or may not belong to an outside super-phylum, Dene- Caucasic. Congratulations!

Mother Tongue Contents

(c) 1996 Association for the Study of Language In Prehistory