Thursday, January 7, 2010

278. The Baseline Scenarios -- 54: Questions

(Continued from previous post)
The map I was just referring to, map B in Figure 4, as displayed in the previous post, is a bit confusing and I must admit, after reading a comment by Maju, that I am now uncertain as to whether or not to take it seriously as evidence for a discontinuous early migration. Maju writes:

Check the rectangular plots in that same figure for true alleles and not just catch-all simplifications. There is no "blue allele" but an array of non-red alleles. The only thing that all "blue" carriers share is a negative one: they don't share the red allele, which is what the maps are focused on.
I'm not sure what to make of Maju's observation because frankly I don't understand exactly what the rectangular plots on the right refer to. It does seem clear that there is a positive-negative relation between the red and blue pie charts in the maps, because, as it seems, all humans carry some form of the gene, either ancestral or derived. So anyone not carrying red is carrying blue. And the caption for the figure identifies the blue alleles as "ancestral" and the red as "derived." In that case, I'm not sure whether it matters whether we are looking at the blue dots or the red dots, since both can be understood as representing migrations, as far as I can see. One is just the negative image of the other.

If I'm wrong about this, then Dediu and Ladd are also mistaken in attempting to correlate the distribution of the ancestral forms of ASPM and Microcephalin with the distribution of tone language, in the paper of theirs to which I've already referred: Linguistic tone is related to the population frequency of the adaptive haplogroups of two brain size genes, ASPM and Microcephalin. The data they are using appears to be drawn from the same source as the data used in the Coop et al. paper, since the layout of their maps is very similar. According to Dediu and Ladd, "Those areas of the world where the new alleles are relatively rare also tend to be the areas where tone languages are common." So clearly, as far as they are concerned, the distribution of the rare occurences is as important as the distribution of the frequent ones. And I see no problem with that aspect of their paper (though I do have a serious problem when they attempt to assign a causal relation between the derived allele and the absence of tone language).

Nevertheless, I have to admit I'm stumped by what Maju has to say about the haplotype data displayed in the rectangular plots, which means that I don't fully understand what the research is all about. So for now, I'm going to withhold judgement on this "evidence" until I understand it better.

Another problem with the same evidence, also noted by Maju, is the absence of any data from the southern or eastern portions of South Asia. South Asia is referred to in the text, but on the map most of it is blank. So even on that score, these results are problemmatic.

The map of tone languages presented in the WALS website seems much more straightforward and the apparent correlation between the distribution of tone languages, in Africa, SE Asia and Melanesia, and the distribution of P/B along the southern route looks impressive, especially when we can point to essentially the same gap in both distributions. But there are many questions regarding this evidence as well, since it's not clear how much of the very dense distribution of tone languages in SE Asia is due to the influence of tone languages from the north, especially China. I see evidence of what might be called "primordial" tone languages in this area as well, but the picture is not clear, so again I would prefer to withhold judgement on this matter until I have a chance to research it more fully.

All in all, and on balance, I do believe I see a pattern of evidence that looks very promising with respect to the hypothesis I've developed. I do see strong evidence of a gap in the same place Oppenheimer sees it, only not necessarily due to Toba. But I also see problems with some of the supporting evidence I've been presenting, so for now I'll simply repeat that all remains: hypothetical.

6 comments:

Maju said...

I have of course extended my comment in the previous post, so I won't be redundant.

But anyhow:

"Another problem with the same evidence, also noted by Maju, is the absence of any data from the southern or eastern portions of South Asia. South Asia is referred to in the text, but on the map most of it is blank. So even on that score, these results are problemmatic".

It is indeed a problem because the blank may hide relevant data. I believe this blank is caused by stringent gene-related laws in the Indian federation that forbid genetic material to be exported and/or demand collaboration with local researchers.

Whatever the case, there are at least three critical regions that are usually undersampled (or not sampled at all) in genetic studies, these are Sudan and the Horn of Africa (probably crucial for understanding the pre-OoA scene and much of African genetics too), South Asia (excepted Pakistan) and, oddly enough, France (that held most of the Paleolithic population of Europe). I can understand more or less why the other two but the French case is really puzzling.

Anyhow a lot of geneticists have used Rosenberg's data of 2005, which is publicly available. It is a valid (or at least understandable) approach but will unavoidably fail to represent properly the diversity at two critical OoA junctions: NE Africa and South Asia. It is of course the case of this paper as well.

German Dziebel said...

"The map of tone languages presented in the WALS website seems much more straightforward and the apparent correlation between the distribution of tone languages, in Africa, SE Asia and Melanesia, and the distribution of P/B along the southern route looks impressive, especially when we can point to essentially the same gap in both distributions. But there are many questions regarding this evidence as well, since it's not clear how much of the very dense distribution of tone languages in SE Asia is due to the influence of tone languages from the north, especially China. I see evidence of what might be called "primordial" tone languages in this area as well, but the picture is not clear, so again I would prefer to withhold judgement on this matter until I have a chance to research it more fully."

To reiterate, a real contribution that a musicologist can make to typological linguistics would be to show functional dependencies between tones and musical styles (as in glottalization and yodeling, or, IMO, between ingression in clicks and in throat singing). If such a correlation indeed exists, then your distributional observations are not due to chance. Also, in this case you could predict the directionality of diffusion of tones (population A has tones and a corresponding musical style, while population B has only tones, hence diffusion of tones from A to B) and may even be able to identify ancient tonal systems in Asia and Africa by the degree of their integration with corresponding musical styles.

Manju Edangam said...

Another problem with the same evidence, also noted by Maju, is the absence of any data from the southern or eastern portions of South Asia. South Asia is referred to in the text, but on the map most of it is blank. So even on that score, these results are problemmatic

The frequency for Tamils SLC24A5 derived allele is 30%.

Soejima et al. 2007

DocG said...

Maju: "Whatever the case, there are at least three critical regions that are usually undersampled (or not sampled at all) in genetic studies, these are Sudan and the Horn of Africa (probably crucial for understanding the pre-OoA scene and much of African genetics too)"

I agree completely that there are too many holes in the samplings and agree especially about the importance of Ethiopia. The Omotic speakers of SW Ethiopia are especially important imo judging from their music. When I see references to "Ethiopia" in so many of these papers, I sigh, because there is a huge variety in Ethiopia and it means very little when we read that such and such a marker is found at some percentage in "Ethiopia." This is a common problem throughout the entire field. Over time this should improve and when it does we'll be in a much better position to evaluate the meaning of all this evifdence.

DocG said...

German: "To reiterate, a real contribution that a musicologist can make to typological linguistics would be to show functional dependencies between tones and musical styles (as in glottalization and yodeling, or, IMO, between ingression in clicks and in throat singing)."

I see no such dependencies so how can I "show" them? What I see is a possible correlation between the distribution of tone languages and the "African signature," based on a history common to both. What I also see is a possible link between the original development of language and the development of music based on a common involvement with both tone and rhythm. The tonal differences that give meaning even to the simplest melody are based on essentially the same type of differences as the differences that make tonemes (tonal phonemes) possible.

German Dziebel said...

"What I also see is a possible link between the original development of language and the development of music based on a common involvement with both tone and rhythm."

Let me begin by saying that I support your moving away from the idea that original human language had clicks. Geneticists out of Stanford such as Knight and Mountain latched on the parallelism between earliest branches of human mtDNA and Y-DNA trees and click languages to argue that clicks were archaic phonemes only preserved in Khoisan languages. Most linguists don't buy into it. It also constitutes an example of false parallelism, or better to say right parallelism (clicks are unqiue to Khoisans as much as Khoisan lineages are unique to Khoisan because both of them are result of local evolution in Africa) but completely wrong interpretation of this parallelism. Nobody, as far as I know, brought up tones in conjunction with out of Africa. And in Africa tones are old, as there are no known paths for their development and diffusion (unlike Asia). I guess you could say that tones and polyphony distributions may represent the same kind of "common history" legacy without any functional dependency between them. However, as there are many more typological linguists out there than linguists interested in deep history, they may ignore it because 1) tones are known to emerge multiple times in human history; 2) tonogenesis is a complicated phonetic process, which means if it occurs multiple times, chances are slim that will ever be able to ascertain a "kinship" between African tones and Asian tones, not to mention Amerindian tones. If you do establish a functional connection between tones and music, you may skip the need to prove that Asian and African tones are genetically related and attract a large number of typological linguists.