Really interesting work and I'd like to read the original paper. The timing of divergence is on the short end, given that occupancy of huamns in Australia is dated up to 80 000 years.

@Svar #39 - "Abo" is a derogatory racial slur in Australia. It grates - I suggest you don't use that term.

It is a very interesting piece of work. I'd imagine the next step would be running similar analyses with additional data and more variation in the parameter space of the genetic algorithm. It would be rather interesting if the number of likely trees could be whittled down further. There is obviously a lot of noise here, but the situation is pretty similar to other analogous situations in cladistics (for example see the endless arguments surrounding the phylogenetic trees for many major groups of higher animals or plants).

In the long run it will either hold up to further analysis or not, and either way it is a good piece of science.

God gave humans language at 4pm 6th August 5497 BC.

Two things that bother me about this. 1) If in fact African languages have a wide range of phonemes, wouldn't that put them at the centre of a web of languages based on phonemes simply due to a high number of interconnections? This doesn't seem to necessarily imply origin to me, just a high hit rate.

2) Perhaps smaller languages, older ones, have more phonemes just due to fewer speakers. Pidgins tend to be simpler languages to facilitate intercommunication. Perhaps modern languages, which spread to wide areas, were pared down to allow easier adoption.

Good old NZ media. Don't pay any attention to science news in our back yard until it has been covered by international media - ooh, look, it must a story if the NY Times and the Economist have covered it, so quick (a week later), let's do some coverage. And, let's not make our coverage a real INTERVIEW with the man (which would involve picking up the phone), let's just repeat what everyone else has said about the research. Sigh.

The idea that the number of phonemes should decay with evolution sounds very weird to linguists. For instance French has many more than Latin (latin has 5 vowels, french a bit less than 15) and a few more consonants.

So first of all you should consider only consonnants. South African languages have more consonnants, especially clicks. But check Kabardian language (North Caucasian) for instance (http://en.wikipedia.org/wiki/Kabardian_language) which has more than 40 consonants, many of which don't show up in African languages.

I'm afraid the biggest caveat for this theory is it crucially relies on clicks. Without those you can almost divide by 2 the number of phonemes in Khoisan languages, and they no longer have anything exceptionnal.

See the example of Chinese: with the apparition of tones they lost the distinction between voiced an unvoiced consonnants (p/b, t/d, k/g...) thus losing... half their consonnant phonemes.

Of course now I should read it, but I don't have the money for the moment ^^

One more thing: phonemes are not sounds, but the mental representations of sounds. You can't just say all humans share them because we have the same mouth, lungs and tongue.

Okay smart pants. Quick question, how many phonemes in Chinese or dialects of Chinese? You can do a quick estimate, roughly approximating number of alphabet * number of tones. It will range from about 100 to well over 200... (e.g. the sound in Chinese that we approximate with the letter "a" can have 4-9 different pronounciations which go with different words, it's not the same phoneme, at least according to any definition of phoneme that I can find).

Svar: Depending on level of isolation, it may well be *virtually* unchanged. Words are added and removed all the time in every language. What matters is the structure: This implies that the most ancient structure of which all other langauges hold homage to is this one. But the modern version is likely the closest we can get.

Its like how all humans come from africa, but those original groups that spawned our species likely do not exist in a form that would be recognizable, but they do exist.

Or like how a third of the people in the world can be traced back to the ghengis khan fellow, though its doubtful many would use this to demand royal legitimacy.

Or like how we dont know what Jesus looks like, but we do know that he's neither white nor black, because the people living in that region have been living there long enough to form a continual genetic chain, and those people generally looked a lot like the way most people from the mideast look.

Svar said "It’s a bit crazy to think that hominids suddenly stood up on their hind legs, developed functional language and then spread out from the Savannah in one go…. c'mon. "

1. Our ancestors were on their hind legs LONG before they had language. 2. The development of language gives a HUGE advantage. Not surprising that if it only developed once - it would spread like wildfire.

There is always some push-back when a revolutionary new theory is presented, especially one that has the audacity to suggest that black people had any influence on the superior art and culture of white people, or more generally that of Western civilization. Not completely unlike Bent's views on Great Zimbabwe.