265. The Baseline Scenarios -- 41: The Gap

An interesting paper presenting an alternative interpretation of Old World tone language distribution was published two years ago by PNAS, and got a good deal of attention at the time: Linguistic tone is related to the population frequency of the adaptive haplogroups of two brain size genes, ASPM and Microcephalin, by Dan Dediu and D. Robert Ladd. ASPM and Microcephalin are

two genes involved in brain growth and development. Deleterious mutations of both ASPM and Microcephalin are involved in recessive primary microcephaly, together with at least four other loci identified to date (p. 10945).

As the authors explain, both genes exist in two forms, original and "derived," with estimates for the origin of the derived states, designated with the suffix "-D," as 5.8 thousand years ago for ASPM-D and 37 thousand years ago for Microcephalin-D.

According to the theory presented in this paper, there is a relationship between

the distribution of tone languages and the geographical structure of ASPM-D and MCPH-D. Those areas of the world where the new alleles are relatively rare also tend to be the areas where tone languages are common. As previously discussed, the effects of ASPM-D and MCPH-D on brain structure and functioning remain largely hypothetical, but it is entirely plausible that they influence the cognitive capacities involved in processing phonological structures and thereby lead to linguistic biases of the type suggested above (ibid. -- my emphasis).

The distributions in question are presented in the following maps:

Microcephalin-D

ASPM-D

Tone Languages (in blue)

Theoretically, the Old World regions with the fewest instances of both genes are where the greatest number of tone languages are to be found. Or, to put it another way, there would appear to be a correlation between the frequency of the original forms of both genes and the frequency of tone language. And what Dediu and Ladd are claiming is a cause and effect relationship:

The relationship between genetic and linguistic diversity in this case may be causal: certain alleles can bias language acquisition or processing and thereby influence the trajectory of language change through iterated cultural transmission.

What is being claimed, in other words, is that the presence of the newer, derived forms of the genes, will tend, over time, to influence the evolution of language such that tone languages will eventually become non-tonal.

On its face, these results would look to me like a perfect example of the logical fallacy called "cum hoc ergo propter hoc" (Latin for "with this, therefore because of this"). In other words, correlation does not imply causation. It would seem more likely that both the genetic distributions and the linguistic distribution are due to underlying historical factors, possibly the same historical factors we've already been considering in the last few posts.

The authors are aware of the problem, however, and claim to have controlled for both geographical and historical factors:

To control for the effects of geography and shared linguistic history on our results, we compared geographic, genetic, typological linguistic, and historical linguistic distances between all pairs of populations in the sample (p. 10946).

Their conclusions are based on mathematical models that seem rather abstruse, to say the least. For example:

Individually, the Mantel correlation with geography for tone is
r
= 0.169, P = 0.015; for ASPM-D, r = 0.074, P = 1.000 (because of Holm’s multiple comparisons correction; ref. 56); and for MCPH-D, r = 0.543, P less than 0.001. Each of tone [sic], ASPM-D, and MCPH-D have low but significant spatial autocorrelations (62): Moran’s I (63) is 0.178, 0.164, and 0.121, and Geary’s c (64) is 0.634, 0.438 and 0.718, respectively, P less than 0.001 for all, suggesting that, potentially, geographical factors might explain the observed relationship. However, the (partial) Mantel correlation between tone and the pair ASPM-D/MCPH-D is r =0.333, P less than 0.001, and, when controlling for geography, it decreases only slightly and still remains highly significant, r = 0.291, P
=0.003, showing that geography is not a good explanation for our empirical findings (ibid.).

While the above goes well beyond my present math skills (as a high school student I was a math whiz, but most of what I learned then has been long forgotten, sad to say), it seems clear that any attempt to assess the distribution of tone languages in either geographical or historical terms is going to be seriously affected by the very same gap that I've been pointing to over and over -- and trying to understand.

It might seem reasonable, in view of the author's analysis, and especially in view of their very impressive mathematics, to forget about this gap and simply accept that the distribution of tone languages must be due to the influence of genetic factors having nothing to do with either geography or history. But there is a serious problem with their reasoning, a problem which should serve as a warning to anyone relying too heavily on mathematical models and too little on common sense:

While it might be possible to attribute the distribution of tone languages to an evolutionary process determined by certain genes, what is it that has determined the distribution of the genes? On this question, as far as I can tell, they have nothing whatsoever to say. If the dates for the origin of each genetic marker have been determined, as they claim has been done, then clearly each must also have had a very specific place of origin as well. And must therefore have migrated since then along a very specific path or paths from that point. Which tells us that when considering the distribution of both the genes and any effects those genes may have had, on language or anything else, neither geography nor history can be ignored.

Much as we might want to explain it away with fancy schmancy genetic and mathematical models, the gap remains.

[Added at 10PM: I've been thinking a lot about this article and the questionable reasoning behind it, and wondering more and more why it was even accepted for publication and why the results have been so widely and uncritically reported. Meanwhile Dediu and Ladd are taking these results and running with them, writing followups that have also been published in respectable academic venues. Their work is now being cited in the work of other cognitive scientists as a beautiful example of how cultural evolution is determined by brain biology, and, unless someone with some influence (I have zilch) rings a bell, will soon take its place as part of the accepted wisdom of the day, among cognitive scientists, linguists and, no doubt, anthropologists.

For this reason, it is all the more important that I make myself as clear as possible regarding the thinking behind this paper. It's not only that they are clearly wrong in claiming that the gapped distribution of tone languages in two very different regions of the Old World is due to the effects of ASPM and Microcephalin, but there is also no reason to assume any sort of cause and effect relationship between the presence or absence of these genes and the existence of tone language, even within either of these two regions. The genes and the tone languages are, in all likelihood, totally unrelated. Even if there were a perfect one to one correspondence between the three distributions (and there is not), this would most likely be due to the influence of more fundamental evolutionary factors underlying the history of all three, factors totally independent of any possible relation between genes and language other than the simple fact that both the genes and the languages were carried from their points of origin to their present locations by the same migrating populations.

While I'm reluctant to attach too much importance to the genetic distributions reported in this particular paper, since the authors were working with seriously incomplete evidence (Australia and New Guinea were deliberately omitted from consideration due to lack of reliable genetic data), it's worth mentioning that the distribution they appear to have found for the original (non-derived) form of ASPM might provide additional genetic support for the gap I've been making so much of. I'm wondering whether other, completelly unrelated, genetic markers might eventually be found with more or less the same geographical distribution, due to essentially the same population history, marked by the very same gap.]