In todays “Ask Marilyn”, Marilyn vos Savant claimed that "humans have been able to speak since roughly 100,000 B.C., but did not go on to explain how we know that. Is she correct, and if so, how do we know that?
We don’t know that. We can’t disprove it, either. Marilyn is safe either way.
Well, a cave painting in the south of France that says “Kilroy was here” has been carbon-dated to 97,842 BC…
:::: ducking ::::
There are specific morphological changes in the throat and palate of modern humans that are considered to be associated with the ability to produce speech. Theoretically, by tracing the evolution of these features we can get an idea of when speech evolved (or at least came under significant selection). Here’s a nontechnical overview.
JRB
It is generally accepted that our species is about 200k years old. It is reasonable to assume that we’ve been able to speak for at least that long, although there is another hypothesis that we weren’t capable of “fully articulate speech” until more recently-- perhaps only 60k years or so. This is based on a lack of objects representing symbolic thought in the fossil record until about that time. But that date gets pushed back when new fossil evidence emerges like these beads that date back 75k years.
I’m guessing that she got the 100k year number from the linguist Joseph Greenberg who claimed to have traced all languages back to common tongue spoken about 100k years ago. But that theory is very controversial, and really can’t be considered to be the scientific consensus on the subject. If she stated it as fact, she doesn’t know what she’s talking about.
I just read a book called Before the Dawn: Recovering the Lost History of Our Ancestors, that puts the date much later, at about 50,000 years ago.
The issue is not settled, though. Other estimates range from 200,000 years ago to 50,000 years ago.
We know that language must have arisen after the split between human ancestors and chimpanzee ancestors, which happened about 5 million years ago. (This human-chimp ancestor must not have had language, or chimps would speak.) And all human races speak equally well, so modern language must have arisen before the ancestral human population spread out from Africa about 50,000 years ago.
Many archaeological finds are thought to favor the later date. There is a dramatic change in the appearance of stone tools found in the Upper Paleolithic, which began about 50,000 years ago. Art appears in the archaeological record at about the same time. It is thought that both of these advances were associated with the rise of language.
The latest piece of evidence, however, is the discovery of a possible gene associated with language. Researchers found a family with a defective version of this gene, most of whom have severe language deficiencies. (Their speech is hard to understand, and they have difficulty with language comprehension.) This modern human version of this gene, called FOXP2, is thought to have arisen and rapidly spread through the human population at some point within the last 200,000 years.
Putting all of the evidence together, the best estimate appears to be at the more recent end of the ranges mentioned. Language is likely to have been the spur to the changes seen in the archaeological record, putting the likely date at about 50,000 years ago.
I just wanted to add that more research, especially genetic research, may help narrow the range of uncertainty.
That’s a good book-- I read it last year. But I think if you go back and re-read it, you’ll find that there is a split between paleo-anthropologists and archaeologists on the subject. The former tend to favor a much older start date for language (possibly going back to H. erectus) and the latter tend to favor the ~50k year date.
Except that recent DNA analysis of Neanderthal bones show that they (or at least the specimen looked at) had identical FOXP2 genes. That would push its origin back to around 500k years (which is roughly when their line diverged from ours). Assuming, of course, that they didn’t get that gene from interbreeding with us.
I just heard a talk by John McWhorter in which he discusses the dissemination of languages as a window into human migration over the ages. In it, he advances the idea that maybe one of the oldest of all language groups is the Koisan (sp?) language group - what are often called the “click” languages from Africa - e.g. Miriam Makeeba’s language. He didn’t give a date for them, but provides some reasons that they might be considered to be among the oldest.
Yes. The hypothesis (and it’s just a hypothesis at this point) is that the speakers of those click languages (like the San Bushmen) represent the oldest lines of any living population and that all the other non-click languages lost their clicks at some point. The Hadza language also has clicks, and the people who speak it also seem to have one of the oldest genetic lines still extant. There are clicks in some Bantu languages, but linguists seem to consider them to be loan sounds from one or more of the Khoisan click languages.
Yep - and one of the more compelling arguments (to me) that he makes is that it’s almost impossible to think about the clicking as an addition to a language. It’s very unlikely that such a sound, which is relatively difficult to make in speech, would find its way into a language that didn’t have it. Borrowed occasionally in one word or another, yes, but not imported into a language wholesale. So his thinking is that the Khoisan languages must be old because the clicking has disappeared from just about all the other languages. Of course, it does beg the question of how it occurred in the first place, but that’s a different question. xo, C.
Could you qualify what you mean by “oldest”? If, say, English and Khoisan both derive from a common ancestor, wouldn’t they be equally old? Or do you mean that Khoisan split off from the rest of the human languages first, or that it is the most “primitive” (that is, the least changed from the proto-language), or something else?
Funny you should ask. It was actually CC who called it the oldest language, not me:
And I was going to correct that by saying it’s more accurate to say the click languages retain some older features rather than to call it the oldest language. As you note, they’re all the same age (or so we think, assuming language only came about once in our history). Similarly, Lithuanian is often called the “oldest” I-E language, but really it just retains some features that most other languages have not-- it’s not any “older” than the others.
Whether the Khoisan languages are thought by linguists to retain other older features, I don’t know.
'Course I’m out over my skis here, but the retention of sounds was only one of the reasons that he contended that the click languages were among the oldest; there were several others. And I don’t know when those languages were to have developed compared to Indo-European, for example.
I’m not familiar with John McWhorter’s work, and it’s entirely possible that he has a compelling argument that was being simplified for the purposes of his talk, but if his position is really “Khoisan has to be old because it has rich click phonetics, which are hard to say,” then I’d want to take issue with it.
First off, I should mention that the typology of African languages is still a little fuzzy: many linguists, including McWhorter, tend to base their analysis on Joseph Greenberg’s typology, but I’m one of the linguists who are extremely uncomfortable with his system. In any event, depending on whose typology you use, the click languages we see in Africa may or may not all fall within the same family. Regardless, while they are typologically rare, clicks aren’t that unreasonably difficult: there are a bunch of equally troublesome phonemes (speech sounds, basically) which occur in other families and spread throughout a region quite easily. When it comes to little kids who, as new language learners, are going to be responsible for cementing a lot of typological changes into a language’s grammar, the relative difficulty of a sound shouldn’t be a significant obstacle. Kids are language learning machines, and as such you can’t base a theory of historical linguistics on learnability issues without considering other phenomena.
Now there may have been, and probably were other things that McWhorter used to build his argument, but I just wanted to mention that “this sound is hard to say” is a very weak piece of evidence on its own.
Well, I dunno. If someone had a time machine and went back 50K or even more years back and found the earliest humans speaking A language, I’d expect that. But if they went back even 20K ago, and found someone speaking a language we could clearly understand now (clicks or no clicks) I’d be shocked. (Yes, I would not be suprised if a linguist found root words and similarities, but not a clearly understood language. In fact, my WAG is if you went back even 4K years no one today would be able to clearly understand any of the then extant languages, although there’d be quite a few roots and similariites. Languages change fairly fast.)
But the question then is- why did every other language lose the click?
I would be surprised if McWhorter was the guy who developed the idea that clicks were part of the original language from which all others descended. And I think that hypothesis is put forward based as much on genetics, not just linguistics.
I haven’t seen Johanna around lately; she had a wealth of information on the hypothetical ancestry of Proto-Indo-European and the other Proto-language-stock constructs: Nostratic, what’s probably not called Proto-Eurasian but ought to be, and Proto-World. Any of you other linguistics mavens have some information on those hypotheses and how they’re regarded (highly skeptically, IIRC)?
Most of those hypotheses are regarded with skepticism, mainly because we cannot reconstruct a language much past the 10,000 year mark; we’re fuzzy enough with Proto-Indo-European as it is. Even if there were a Nostratic and Proto-World language, we could not say anything meaningful about it, because they existed too far in the past for any significant reconstruction to be done on them. Linguistic drift is a bitch, and the relevant daughter languages (well, every language spoken on Earth in the case of Proto-World) are simply too far-flung and too divergent to use the comparative method effectively.
As for the Khoi-san clicks, I read something a few years ago (in a popular magazine; Time, maybe) that suggested that the clicks were incorporated into the Khoi-san languages because they allowed humans to communicate vocally during the hunt — animals weren’t scared off by clicks as much as they were by normal human vocalizations. My bullshit-meter went off on that one; not every word in Khoi-san has a click in it, for one. And if that were such an advantageous trait, then why wouldn’t other hunter-gatherer populations in other regions retain such features? Utterly implausible, I say. As far as I know (as a know-it-all anthropology undergrad), most hunters don’t bother with vocal communication on the hunt, when hand signals and pre-planned attack patterns suffice. Still an interesting idea.
The speed of language change really varies from period to period, and more importantly from language to language. Icelandic, for instance, has undergone hilariously little typological change from the first recovered texts (which I seem to recall originate from the 1100s) to the modern period.
And yea, language typology can be a deep pit from which many good linguists never escape… there are a lot of good theories, but there are also a lot of theorists who’re wackier than a barrel of monkeys.