Disclaimer: I know this post is on a paper released over a year ago; however, I’m still going to write about it for three reasons: 1) I did a presentation about it earlier this week (20/01/08); 2) I think it relates to a recent buzz around gene-culture co-evolution; and, 3) It’s a bloody awesome paper.
So, what is the paper called? Okay, once you read this title, do not yawn, go to another website or… Linguistic tone is related to the population frequency of the adaptive haplogroups of the two brain size genes, ASPM and Mircocephalin. See, now we’ve got the hard part out of the way, I can begin to discuss exactly what the authors, Dan Dediu and Robert ‘Bob’ Ladd, found and why it’s important to our understanding of linguistics, genetics and evolution. It’s really interesting, honestly.
ASPM and Mircocephalin: the latest pokemon, right?
No — If you’d read the title, then it’s fairly obvious that we’re discussing genes involved in brain development (anyway, Pikachu has plenty of pokemon with which to contend). Quickly explained: ASPM (Abnormal Spindle-like, Microcephaly-associated) and Microcephalin (MCPH1) are two genes showing signs of recent positive selection in some, though not all, modern human populations (for more information, read John Hawks comprehensive overview of Bruce Lahn and colleagues’ findings).
These derived hapolgroups appeared approximately 5,800 (ASPM-D) and 37,000 (MCPH1-D) years ago, giving us a clear indication that they underwent positive selection at different points within the human lineage. Additionally, deleterious mutations on these genes (among several others) are commonly associate with primary microcephaly, a neurodevelopmental disorder. Following the discovery of these gene variations, various suggestions emerged as to their functional role, including: variability in brain size, intelligence (specifically IQ), head circumference and schizophrenia. The basis for these hypotheses seemed reasonable enough, especially variation in brain size, as microcephalics demonstrate, “in which the circumference of the head is more than two standard deviations smaller than average for the person’s age and sex.” (wikipedia.org/microcephaly). As it turns out, no such correlation was found in these areas, leading some to suggest that the initial observations of positive selection were incorrect. I think this is wrong, and thankfully so did Dediu and Ladd.
Nonsuprious correlations between genetic and linguistic diversities
Adopting a new approach, Dediu and Ladd hypothesised “a relationship between the population frequency of these two alleles and the presence of linguistic tone”. Linguistic tone? Okay, I’ll explain: tonal languages use variances in pitch, alongside vowels and consonants, to discriminate between words and grammatical categories (see this website for a good example of the Chinese tone difference). Non-tonal languages such as english do not use tone in this way; instead, we use to tone for emphasis or to express emotion. The important point to take away is that in non-tonal languages tone inflection does not change the meaning of words.
Very interesting, but what does tonal languages have to do with genes?
Dediu and Ladd noticed a nonsuprious correlation between low levels of ASPM-D and MCPH1-D (D = derived allele) and the distribution of tonal languages:
(You might notice one major omission from the above maps, specifically the entire North and South American continents. One reason is the prevalence of genetic admixture in this region due to recent contact with people of European ancestry, and the other is “the fact that the Americas are very diverse linguistically.“.) To help shed some light on these observations, the authors then set about confirming the correlation by testing their hypothesis against 49 populations from the old world (to see a far better explanation, the authors themselves add further information, which also discusses the stats. In addition, Mark Liberman over at the language log offers his own critique, plus a response post by Bob Ladd.)
First, they consider each derived allele separately to calculate their relationship with tone within the tested populations; finding a highly significant correlation that suggests the finding is “probably not due to chance.” Second, they probe the significance of these correlations by gathering frequency data for 983 alleles and values for 26 typological features, confirming the distribution of ASPM-D/MCPH-D and tone to be “unusual when compared to other correlations between genes and typological features… [and turns out] stronger than 98.6% of the thousands of gene-language correlations we tested.”
Third, both the derived alleles are considered together and then compared with tone using logistic regression. Again, they considered this relationship in the context of genes and language and get a significant result, suggesting that “linguistic tone is predicted by the population frequency of the two “derived” alleles together.” Finally, both geography and history are controlled for by performing a mantel test, which computes the correlation between two distances. Considering these other demographic processes allowed Dediu and Ladd to make sure their original results are a valid explanation for the distribution of genetic and typological features.
Just to confirm…
… this paper finds a negative correlation between tone and population frequency of the derived haplogroups of ASPM and Microcephalin. This infers a causal link between ASPM-D/MCPH1-D and non-tonal languages. Though as Daniel Nettle points out in his commentary: the paper is hypothesis-generating, not hypothesis-testing. Still, the authors do offer their own thoughts as to what is going on. Specifically, they argue that a cognitive bias exists, which is further expounded upon in two related papers (Ladd et al. 2008; Dediu, 2008), and broadly consists of three components:
(i) from interindivual genetic differences to differences in brain structure and function,
(ii) from differences in brain structure and function to interindividual differences in language-related capacities, and
(iii) from these to typological differences between languages.
Essentially, these three components are summed up as cultural transmission + small individual biases = the trajectory of language change. In the case of ASPM-D and MCPH-D, these act as genetic bases that slightly bias an individual towards non-tonal languages — amplified by iterated cultural transmission. It’s important to state, Dediu and Ladd are not saying there is necessarily a direct connection between language and genes. Nor are they implying specific languages have selective advantages. Plus, languages themselves are not acting as selecting mechanisms on ASPM-D/MCPH-D. Rather, they clarify this relationship as being:
[…] only a very indirect and probabilistic one; we certainly are not suggesting that there are ‘genes for Chinese’. But we believe that the broad outlines of an explanation based on the interaction of bias and cultural transmission are very plausible indeed (Ladd et al. 2008).
Even if Dediu and Ladd turn out to be wrong, perhaps the most prevalent aspect of this paper is its application of linguistic and genetic features in the context of demography. Obviously the next move is to provide experimental data either confirming or disproving their hypothesis, with Patick Wong and his colleagues offering one possible line of inquiry:
[…] they [Wong et al.] have shown that some monolingual adults find it much harder than others to learn an artificial language vocabularly that makes use of tone or pitch distinctions, and that the difference between these groups show up in subtle differences of brain structure as well. If we could show that these differences also reflect differences in genetic make-up, it would go some way to showing that the correlation we have found is based on a real causal link. (Dediu & Ladd, 2007, further information.)
D. Dediu, D. R. Ladd (2007). From the Cover: Linguistic tone is related to the population frequency of the adaptive haplogroups of two brain size genes, ASPM and Microcephalin Proceedings of the National Academy of Sciences, 104 (26), 10944-10949 DOI: 10.1073/pnas.0610848104