Genes & the Mother of all Clicks

[Jacques Cinq-Mars]

One more for the “clicks”?

Isn’t this getting to be really exciting? Another piece of “palaeoanthropological hype” from the BBC, in which the author (one Paul Rincon) and, likely, some nameless Editor, didn’t see fit to present the interested readers with the actual reference to the paper they are reporting on. There is no doubt that many of the “palaeo-whatever” aficionados are already talking about it in “les chaumières”, as we say in French.

I don’t know about some of you, but, to me, it adds very little – unless it is a growing sense of confusion and frustration -- to my understanding of what this is all supposed to mean from a true interdisciplinary point-of-view, i.e., one in which a real effort is be made to arrive, at the very least, at some sensible integration of the various data sets emanating from different palaeoanthropological disciplines, including Palaeolinguistics. This effort at interdisciplinary integration/harmonization should come from all the parties involved, but I have the sneaky feeling (and I don’t know why!) that, so far, it has been (intentionally or not) pretty unidirectional.

But, once again, this is all based on my reading of a BBC digest that, most likely, is not necessarily a fair account of Sarah Tishkoff’s work.

Jacques Cinq-Mars

BBC - Last Updated:  Wednesday, 2 April, 2003, 12:00 GMT 13:00 UK

Tanzania, Ethiopia origin for humans

New DNA evidence suggests "African Eve", the 150,000-year-old female ancestor of every person on Earth, may have lived in Tanzania or Ethiopia

By Paul Rincon.

A genetic study has shown that the oldest known human DNA lineages are those of East Africans. The most ancient populations include the Sandawe, Burunge, Gorowaa and Datog people who live in Tanzania.

Researchers found a very high amount of genetic variation, or diversity, between the mitochondrial DNA of different individuals in these populations.

Mitochondrial DNA is passed down exclusively through the maternal line. The longer a population has existed, the more variation accumulates in its DNA lineages.

"They are showing really deep, old lineages with lots of diversity. They appear to be the oldest lineages identified in Africa to date," said Dr Sarah Tishkoff, of the University of Maryland, US, who led the research.

Great resource

The so-called African Eve represents the ancestral mitochondrial genome that gave rise to all the different types seen in people today.

Several of the ethnic groups sampled in the study also live in countries surrounding Tanzania.

"It's entirely consistent with what we expected," said Dr Spencer Wells, a geneticist and author. "All the evidence is pointing to East Africa as the cradle of humanity."

Dr Wells added that the data ties in well with archaeological evidence of a long occupation of East Africa by modern humans and hominids.

But Professor Ulf Gyllensten, a molecular biologist at the University of Uppsala, Sweden, was cautious about claims that the oldest DNA lineages were confined to East Africa.

"I wouldn't be surprised if Dr Tishkoff has found old lineages there, but I think we're just skimming the surface," he said.

"Too little research has been done in Africa to get a clear picture. I don't know why, because it's clear there is a great resource of genetic diversity there," added Professor Gyllensten.

'Click' language

Dr Tishkoff's team have collected mitochondrial DNA samples from 1,000 Tanzanians since they began their research in 2001.

Although the data comes from groups living in Tanzania, the Burunge and Gorowaa migrated to Tanzania from Ethiopia within the last 5,000 years.

Dr Tishkoff said Ethiopia was also a good candidate for the region where modern humans evolved.

One of the populations sampled in the study, the Sandawe, speak a "click" language like that of Khoisan people from southern Africa.

The Khoisan were previously thought to possess the oldest DNA lineages, but those of the Sandawe are older. This suggests southern Khoisan originated in East Africa, according to Dr Tishkoff.

"That is surprising, because it has been presumed that the oldest populations were in the south," said Professor Gyllensten. Some of the oldest modern human archaeological sites in Africa are in the south of the continent.

Dr Tishkoff said she planned to carry out further research to narrow down the most ancient East African lineages.

[colin]

Yes, I'm confused.
How, in any meaningful sense,  can it be true to say that one mt DNA lineage is older than another? The article even mentions the mtDNA Eve story, the  whole point of which is to say that everyone's Mt DNA lineage can be traced back 150,000 years to this mythical figure in Africa, beyond which by definition we would  all have a shared lineage.

Or am I missing something?
Cheers
Colin  

[Scott MacEachern]

One interesting thing about this as well is that according to traditional linguistic taxonomies, one of these languages is Khoisan (Sandawe), two are Afroasiatic (Gorowa and Burunge -- both Southern Cushitic) and one is Nilo-Saharan (Datog[a] -- Southern Nilotic). Evidence from historical linguistics suggest that these three language groups originated in separate areas of the continent. Given the primordial relationship that's supposed to exist between Khoisan clicks and genes, one wonders how this coincides with an origin of modern humans in specifically East Africa.

I'd like to see the work that this is based upon as well.

Scott

[lagarvelho]

Scott and all:

Well, it's obvious that them click speakers got around!(just trying to bring a smile to the generally serious nature of this list!).  More seriously, however, if anybody sees such a paper, maybe they should put it in Links or Bookyard or some other appropriate venue.  I wouldn't mind seeing it too.
Anne G

[lagarvelho]

Colin:

No, I don't think you were missing anything.  The story as written doesn't "compute", as they say.  If "click" languages are the oldest, by this logic, they can't be more than 150 kyr or so old.  But most linguists(the way I understand it), don't think you can trace language lineages back more than 8,000 years.  Which leaves an *awfully* big gap, IMHO.
Anne G

[Dale Hoogeveen]

What has me scratching my head is the jump from mtDNA chronologies to linguistic ones, as though they are interchangeable.

Dutch

[John Goodrum]

How, in any meaningful sense,  can it be true to say that one mt DNA lineage is older than another? The article even mentions the mtDNA Eve story, the  whole point of which is to say that everyone's Mt DNA lineage can be traced back 150,000 years to this mythical figure in Africa, beyond which by definition we would  all have a shared lineage.

When geneticists say "ancient lineage" I think it's just shorthand for "anciently diverged lineage," by which they mean an allelic or genomic (e.g., mtDNA) lineage whose origin is near the root (most recent common ancestor) of existing variation for that piece of DNA.  Some African populations such as the !Kung have high frequencies of haplotypes that are "old" in this sense, with respect to the mtDNA lineages that most other human populations belong to.  In other words, you might have to go back the full ~150,000 years on both lineages to find the MRCA of !Kunk mtDNA and the mtDNA of another African, European or whomever, whereas a German and Englishman (or two Nigerians) might have to go back only a thousand years to find their mitochondrial MRCA.  From the German's point of view then, the lineage of the Englishman is "younger" than the !Kung lineage.

Hope that makes sense.

JG

[colin]

Thanks for trying to make sense of this for me John, but......
If we are saying that the iKung mt DNA lineage has been more or less isolated from the lineage that led to French, Germans and most of the rest for a very long time, it still means that both lineages are equally old, in that each lineage has been seperated from the other for the same length of time.
Perhaps it is as misleading to speak of the ages of DNA lineages as Larry Trask argued it is to do the same with language lineages - that is, unless a new language, or mt DNA, is invented from scratch, then all lineages are of equal antiquity.
If, however, one is arguing that iKung mt DNA is more like the ancestral form of long ago than other mt DNA, then is it not strange that the evidence adduced for this is  that greater diversity (ie more mutations have occurred) is said to be found in the iKung mt DNA!  More mutations would suggest it has moved further from the ancestral form rather than the reverse.
So no, I still cant buy it.
Cheers
Colin    

[John Goodrum]

If we are saying that the iKung mt DNA lineage has been more or less isolated from the lineage that led to French, Germans and most of the rest for a very long time, it still means that both lineages are equally old, in that each lineage has been seperated from the other for the same length of time.  Perhaps it is as misleading to speak of the ages of DNA lineages as Larry Trask argued it is to do the same with language lineages - that is, unless a new language, or mt DNA, is invented from scratch, then all lineages are of equal antiquity.

Yes, I agree.

If, however, one is arguing that iKung mt DNA is more like the ancestral form of long ago than other mt DNA, then is it not strange that the evidence adduced for this is  that greater diversity (ie more mutations have occurred) is said to be found in the iKung mt DNA!  More mutations would suggest it has moved further from the ancestral form rather than the reverse.

More mutations can also mean that the population in question had a larger effective size over some part (maybe the early part) of the last 150,000 years.  In this case they would have retained more of the polymorphisms of the ancestral population - polymorphisms (and hence diversity) that have been mostly lost due to drift in non-African populations, who have had smaller population sizes sometime between then and now.  Smaller population means greater drift, which results in greater loss of alleles and lower diversity.

JG

[rlass]

I'd like to make some general remarks on some of the very few issues in this discussion that were not covered by Larry Trask and a few others, or not covered perhaps in enough depth.

1. Antiquity. The time-scale of language change is not, from a 'strict' biological point of view, sufficiently slow to allow anything except the most basic neurally encoded properties of language (or rather underlying 'facilitators') to survive. Those of us whose lives are devoted to working with the reconstruction of the detail of ancient languages and language families generally find that something like 8-10,000 years is about the limit beyond which you can reconstruct anything, or to put it another way, beyond which anything recogniseable as *particular* rather than a property of language-in-general will survive in recogniseable form.

This means that *if* in fact clicks were a property of Proto-World or some other silly and evidenceless invention, they'd be gone by now, because they are precisely the kind of thing that goes fast: superficial phonetic details. If you look at two languages that are very closely related, and whose separation time is measurable in a couple of millennia, like Dutch and English, you find that at that level of resolution, though certain sound *types* remain recogniseable and can be used to demonstrate relationship, there is probably not one consonant or vowel in any variety of Dutch that a compentent phonetician would say is 'the same as' any in English --except broadly in type.

2. Clicks are rare. So what. I could name dozens of sound types that are as rare or almost. Most important, however, the mechanism for making clicks is entirely outside the entire mechanism that has been developed for speech: they are, despite their utility, 'excrescences'. What I mean is this: insofar as one can find anatomical specialisations in humans for language, they involve the respiratory system. All the famous stigmata, like the low position of the hyoid, the lowering of the larynx from its position in the neonate pressed against the soft palate in other mammals, the feeding of voluntary cortical information from the motor strip to the larynx via the nucleus ambiguus so that there 'choices' of vocalisation are natural: all of these cluster around the upper respiratory tract and the use of what phoneticians call a pulmonic egressive airstream: speech is a function of the movement of air from the lungs, through the trachea, and out the oral and nasal cavities.

All languages use pulmonic egressive airstreams, though some use others; but that's the basic one, and the only universal one, which already does suggest that its connection with language is a significant one. The mechanism for clicks on the other hand is not unique to humans, has nothing to do with any human adaptation for speech, and is common to all mammals: it is, to be as simple as possible, the mechanism that all mammals use in suckling. (And we also use in kissing, by the way: a kiss is a bilabial click.) In a click a closure is made some where in the *supralaryngeal* tract, in fact necessarily anterior to the soft palate and the tongue root; a secondary closure is made posteriorly, sealing the 'velopharyngeal port', i.e. the potential opening into the trachea. This secondary closure is then moved sharply backwards, increasing the volume of the oral cavity, and hence lowering the air pressure; this lowered air pressure creates a differential with the external ambient pressure, and air rushes in breaking the anterior closure. To get a feel for this, kiss, or make a closure behind the upper incisors and produce 'tsk, tsk', or make a closure there and another one with the side of the back of tongue against the lower molars, and make the sound calling horses.

OK, so the click mechanism, as opposed to what's used in normal speech, is pre-primate, marginal, and in most languages that use clicks not applicable to most of the consonant inventory. It so happens that there is a striking areal clustering of languages using clicks (which should not be called 'click languages' because they're not different in any other ways, but just ordinary human languages with an uncommon frill). And this place happens to coincide with an important ancient location. But this says nothing historical: all the normal types of evidence from language change would suggest that clicks are a relatively recent idea, certainly nothing ancient.

However languages that do use clicks may make quite a meal of them, as often happens with novelties. The numbers given in the original journalistic account are rather odd. The commonest number of clicks in non-Khoe or San languages (i.e. in those Bantu languages that have borrowed them) is around 15: three places of articulation x 5 different coordinating articulations or types of airstream release (at least this is so for Xhosa, Zulu and their relatives). Some of the Khoe and San languages may have numbers in the 40s.

For information on these languages, the best source is the chapter by Anthony Traill in Raj Mesthrie (ed.), Language in South Africa [scholar fails to remember title because book is in his office, even though he has a chapter in it], Cambridge University Press 2002.

Roger Lass

[rlass]

A comment on the problem of the 'age of linguistic lineages'. Of course it is true that in one sense all language lineages are of the same age if (as seems likely) language is monogenetic. That's virtually a tautology.

But within a given lineage (say Indo-European), it is certainly not the case that all lineages are of equal age, since a language family is the result of branching, and a cladistic break point has a particular age. So within Indo-European for instance, it seems clear that Germanic is a 'new' lineage in comparison say with Indic (the group that contains Sanskrit and its descendants), or Greek or Italic (the group containing Latin and its descendants). The evidence that we have suggests that Germanic originated in late antiquity, much later than Italic; and within Germanic we can more or less date the emergence of new sublineages at quite late dates. E.g. while the earliest Germanic inscriptions date from around the 2nd century AD, the differentiation into the modern sublineages like the one containing German and English and their close relatives, or the Scandinavian languages, are considerably later.  So in this sense it is quite proper to speak of 'old' and 'young' lineages at a certain level of resolution.

So the existence of the clade {Frisian, English, Dutch, Low German} is a quite late development; another branch off the same parent lineage that leads to the (relatively) monotypic clade containing (High) German. And this lineage had branched earlier from one whose other branch is the modern Scandinavian languages. Etc.

Depends on level of resolution.

[rlass]

There is an important difference as everybody knows between correlation and causation. In linguistic prehistory there is just as important a one between similarity and common ancestry. And another important one between the reconstructable and the non-reconstructable.

In this particular case the correlation of clicks with antiquity of a genetic split is vacuous. It is so for the simple reason that as far as anybody knows *no* linguistic feature has this kind of shelf-life. In respectable professional reconstruction, there is a kind of horizon at about 10k years that represents the absolute limit of our being able to reconstruct protolanguages or establish filiations. The likelihood of the presence of clicks in these two languages as being due to anything but very recent (on the linguistic geological time scale) common ancestry is vanishingly small, to put it charitably.

The only features that would be likely to survive for the kind of time suggested here are those that all natural languages would have anyhow: e.g. verbs, referring expressions, some kind of syntax. And these features being universal would not be criterial for anything except characterisation as a natural language, and this is not in doubt anyhow.

The correlation of clicks and locale and antiquity is a seductive one, but it is not of any historical interest.

[lagarvelho]

Scott:

Where can you get "Historical Linguistics"?  I'm not familiar with the title. But t hen I'm not a linguist.
Anne G