I don’t think it can be proved, just that the idea seems reasonable to me. I suspect any real linguist would admit this. I see no way to prove any of this kind of stuff that is pre writing. Again I am not a linguist.
Capt
If we were, in fact, anatomically modern as a species 200,000 years ago, why would we NOT be speaking? There is a school of thought in linguistics (which believe it or not I actually have a degree in from way back when) that suggests that speech predates H. sapiens sapiens. We do know that Pan troglodytes have discernable vocalizations and regional variations of such. Once our species evolved to have our rather strange nasopharynx and vocal cords, then speech seems to be a given, at least to that school of thought.
This, of course, leads to the discussion as to whether we can have innominate thoughts or whether words somehow drive our thought processes. But, that is probably a discussion for another thread.
I think almost certainly we were, and our anatomy evolved from an earlier time based on that. Speech and language of some sort must of predated our modern form. IMHO.
There must be continuity over that time period, because the alternative is complete lexical replacement. The controversy is over how far back you can safely go before the attested evidence. Most linguists are quite comfortable with a range of a few thousand years, which is already pretty amazing. There’s no question that languages go back further than that, and that the existing families are probably somehow related, but it’s a question of whether the evidence can be read.
Look at it this way: you and I are probably biologically related. We can each trace our family trees back a few generations, and a professional genealogist would be able to push it back further. We can also give DNA samples. Chances are that they would be reasonably satisfied that we were related, but at a complete loss to prove how. “Okay, the two of you are both part English [or whatever], so you probably have a common ancestor somewhere in England in the Middle Ages, give or take.” Is that proven? Is that even useful?
So, that Indo-European is related to Uralic: the conventional view is “probably, but maybe not. No way to prove it.” Their view is “OMG! Look! Superficial similarities in some of the modern languages and some math and science! We totally proved this!” (ETA: At least as the press is reporting it.)
Dr. Drake thank you for saying clearly what I was trying to say.
Seems to me that we, as a population, would tend to keep the same words for the same things until some outside force caused us to change our usage, like an invasion followed by assimilation from a different language group. How do you find what has changed and what has not, that is the question. The author sounds logical and reasonable, but proof?
Capt
septimus writes:
> To a very very very very() crude approximation, one can imagine that the
> replacement of vocabulary items is similar to the disintegration of radium atoms.
> If the half-life of radium is 1600 years, that does NOT mean that all the radium
> will be gone in 3200 years! Instead 3/4 of it will be gone in 3200 years. After
> 9000 years 2% of the radium will still be intact.
>
> A comment like “The traditional view is that words can’t survive for more than
> 8000 to 9000 years” shows ignorance of the statistical principle just described
> (via the example of radium half-life). If the “half-life” of a basic vocabulary item
> is 6000 years(), then one would indeed expect 12.5% of that vocabulary to
> persist after 18,000 years
6,000 years is notl the usual estimate for the half-life of even very stable words. The usual estimate is that after 1,000 years, 80% of the 200 most stable words will be preserved in a language. Even if you restrict yourself to the 100 most stable words, only 86% will be preserved:
So the following is true:
For the 200 most stable words:
After 1,000 years, 80% will be preserved.
After 2,000 years, 64% will be preserved.
After 3,000 years, 51% will be preserved.
After 4,000 years, 41% will be preserved.
After 5,000 years, 33% will be preserved.
After 6,000 years, 26% will be preserved.
For the 100 most stable words:
After 1,000 years, 86% will be preserved.
After 2,000 years, 74% will be preserved.
After 3,000 years, 64% will be preserved.
After 4,000 years, 55% will be preserved.
After 5,000 years, 47% will be preserved.
After 6,000 years, 40% will be preserved.
The half-life of the 200 most stable is thus a bit more than 3,000 years and the half-life of the 100 most stable is thus somewhat more than 4,000 years. And after 15,000 years, 4% of the 200 most stable words will be preserved and 10% of the 100 most stable words will be preserved. The claim of this paper is that after 15,000 years, a group of 23 words will be preserved.
A good place to check when you read a news story about language is the website Language Log:
http://languagelog.ldc.upenn.edu/nll/
This is written by a group of linguistics professors and is a very good way to validate the usefulness of such news stories. Here’s what they have to say:
http://languagelog.ldc.upenn.edu/nll/?p=4612
They are very dubious.
Capt Kirk writes:
> Seems to me that we, as a population, would tend to keep the same words for
> the same things until some outside force caused us to change our usage, like
> an invasion followed by assimilation from a different language group.
Sounds reasonable, but it isn’t true. Regardless of how stable society is, language changes at a steady pace.
I don’t try to defend the cited paper, which is by non-linguists, including one whose clading diagram for Indo-European was criticized at SDMB a few months ago. I have read writings by Starostin, Ruhlen, etc. (respected linguists) and find them more convincing than their critics. Caveat: I have no linguistic expertise. However much of the debate is about statistics, logic, and puzzle-solving.
Some comments.
-
The penultimate sentence in this excerpt implies that 8 words from Swadesh-200 will be preserved, of which 10 come from the Swadesh-100 subset! This is not a criticism of Mr. Wagner who I’m sure has applied the formulae “correctly”, just a reminder that elementary math is not a strongpoint of some of these historical linguists!
-
Sorry about the very crude 6000 year estimate. A better estimate for Swadesh-100 would be 5000 years. But please note that these 100 words are not equally stable. Just as the Swadesh-100 are more stable than the Swadesh-200, a list of 22 especially stable words can be found in some text. I won’t hunt down a cite, but the top words according to the source are: I, two, you, what/who, tongue, name, eye, heart, tooth, … Their “half-life” will be longer than the Swadesh-100 average.
-
Wiki’s Glottochronology article has a subsection on Starostin’s method. It treats borrowing differently, does not have a simple half-life form and leads to retention estimates somewhat higher than the usual formula.
I don’t think it’s important whether a formula predicts 23 preserved words after 15,000 years, or just 15 words (Starostin’s formula). Such estimates will be very very crude anyway.
If one wishes to pursue this topic, I’d focus on South American languages. Ethnologue shows only four indigenous language families in Africa, but dozens in South America, despite that Joseph Greenberg, who defined the four African families in the first place, considers the South American languages to be in single family, better supported than 3 of the four African families! Since humans arrived in South America a little less than 15,000 years ago probably, likely speaking languages from a single family, it would be a good case study. Ruhlen has written extensively on it.
Ruhlen presents evidence of genetic linkage among the Amerindian languages which, to this layman, seems strong. And whether you take Ruhlen’s side or not, the shrill, almost vitriolic, opposition to his work does surprise.
P.S. I think much of the debate over Amerindian can be found on-line. If we do choose to debate I’d prefer NOT to rely on a simple headcount. (“3 out of 5 professors of linguistics think Ruhlen is wrong.”)
I’d prefer instead to focus on the details of the argument. Ruhlen presents clear examples of cognates across dozens of languages. His critics ridicule these with counterexamples that are patently vapid and obtuse.
I don’t see any point trying to debate, in this thread, the validity of Ruhlen’s reconstruction of proto-Amerindian or of proto-World, just as I don’t see any point in trying to debate the article cited in the OP about the possibility of the reconstruction of even 23 words in proto-Eurasiatic. My point in posting on this thread was to show that historical linguists are not impressed by the arguments made by the authors of that study. This is why I cited Language Log. (Really, read some of the older articles in Language Log. It gives lots of examples of bad news stories about language.) It’s annoying how often papers about language that are not generally accepted are talked about in news stories as if they were generally accepted. Frankly, journalists do this because headlines like “Language from 15,000 years ago has been reconstructed” sell better than “Reconstructing languages more than 10,000 years old generally considered impossible”.
I don’t want to debate it here because it would take writing entire books worth of evidence to prove anything. It would also take relearning immense amounts of linguistics, a field that I’ve been out of for over 35 years. What you call “vitriolic” replies to Ruhlen I would characterize as exhausted linguists being forced to write long rebuttals to poorly argued statements. I’ll pass on putting in the many long hours it would take to properly debate Ruhlen here.
Please do not let’s pretend to use the strawman of OP’s paper by non-linguists to rebut esteemed linguists like Greenberg and Ruhlen. :rolleyes:
Here’s a paper by Bengtson and Ruhlen laying out etymologies and answering their critics. Link to a good refutation if there is one. Let Dopers draw their own conclusions.
I’m getting all verklempt–talk amongst yourselves. [Hand wave]
Gagundathar writes:
> If we were, in fact, anatomically modern as a species 200,000 years ago, why
> would we NOT be speaking?
This is a “It sure sounds reasonable to me, so it must be true” type of argument. I don’t have any argument for or against this statement. I don’t think you have any strong argument for it either.
I do ask Dopers to read the Bengtson-Ruhlen paper and this rebuttal. But frankly if this is the best rebuttal out there, I think Ruhlen’s side can claim victory.
With as many words as the Bengtson-Ruhlen paper cites, it’s no surprise there are some errors. But to show a word meaning “son, man child” as “boy” ? How egregious a flaw!! :dubious:
Bengtson-Ruhlen show apparent cognates for ‘Kuna’ = ‘woman’ in over a hundred languages. Is the egregious substitution of ‘kwan-iswa’ for ‘kwaaniswa’ truly unforgivable?
Do note that these “errors” were highlighted with bullets in the very short rebuttal. I don’t want to trivialize unfairly the arguments against Ruhlen, and I’ve excrpted two of the sillier parts of the very brief paper, but …
This topic is interesting; I was sorry to see discussion die. I’d hoped discussion would focus on general validity, not on the deficits of one particular paper (especially one by non-linguists). Bengtson and Ruhlen are linguists, though controversial. Here’s a paper by Bengtson and Ruhlen on Global Etymologies. Better might be to focus on the much narrower Amerindian Hypothesis discussed in many on-line articles including the 1992 Scientific American.
The only real rebuttal to “global etymology” posted upthread chose to go into special detail on “five types of error” related to Algonquin. Three of the five types were each illustrated with a single example:
Blackfoot and Arapaho are both Algonquin languages, so that mistake, however egregious, was unlikely to be a major factor in an argument about global etymologies. Ruhlen has never claimed to be an Amerindian specialist, but spent hundreds of hours scouring Greenberg’s raw data. If these were the worst errors I think he did a fine job. To consider “boy” and “son, man child” to be different glosses seems to me to go well beyond normal pedantry.
There are good reasons to guess that the Amerindian Hypothesis is very likely. Because of geographic bottlenecks at the Bering Passage and the Isthmus of Panama, the earliest humans in South America probably spoke a single language or languages from a single family. I don’t think “mainstream” linguists even dispute this. The dispute is about whether inherited cognates among the diverse Amerindian languages would still be detectable 12,000(?) years after they diverged.
Linguists have measured empirically the rates at which words are replaced. The number of cognates that show up in long-distance etymologies is about what one would expect given observed retention rates. Obviously there is much fuzziness: for example Icelandic (isolated from languages it could borrow from) has changed more slowly than the average. (South America is also isolated from non-Amerindian languages.) Different types of word are replaced at different rates; the pronouns ‘I’ and ‘you’ are especially resilient and they have the form /na/ and /ma/ respectively throughout Amerindian language families and nowhere else. About this one of the most respected historical linguists of all wrote:
I am quite aware that most Amerindian specialists reject the Amerindian hypothesis. (Note that they would have important career-related motives for doing so.) I am aware that many eminent linguists ridicule Greenberg, Ruhlen, and Bengston vehemently – too vehemently by half if you ask me. Here’s a review by one of the most eminent historical linguists about another.
I hope Doper linguists will take a stand in this debate. Non-linguists should feel free also; much of the debate revolves around common-sense. But please PLEASE don’t just repeat what we already know: A majority of linguists reject the Amerindian Hypothesis.
(Although I do find this topic interesting, let me admit upfront that I post here to set up a reference for a possible future post or thread unrelated to linguistics that looks at controversy in general.)
I wonder how long “Okay” is going to last.
That’s a good question. On the one hand, words that mean “no” are more important than words that mean “acceptable.” On the other, among negation words, only “not” makes the list of the 200 most conservative words.
On yet another, technology may change everything so we head back toward a unified language. “OK” is understood by more than half the world already. I wonder if a unification of language would slow the changes in languages. The only reason I could think it would is that you would have fewer substitute words from other languages.
One of the problems with word reconstruction is that, as far as I know, all attempts so far to artificially reconstruct a mother-of-all-languages have resulted in a startling number of false positives. That is, out of the list of “reconstructed words”, the words the programs derived these words from frequently use words that we are absolutely, 100% certain are in no way directly derived proto-Troglyditese. Imagine if someone tried to prove to you that two languages are very similar because of a few cute false cognates. It’s like that but more subtle and systemic an error.
It’s not that some of the reconstructions aren’t plausible, or even likely, so much as the fact that it generates so much garbage it throws the validity of the plausible reconstructions into doubt. Kind of like if I told you the answer was “4” for any math problem you gave me. Sure, there’s an infinite number of problems where 4 is the right answer, but you’d be right to question my math knowledge when that’s all I say no matter how often it happened to be correct.
septimus writes:
> I am quite aware that most Amerindian specialists reject the Amerindian
> hypothesis. (Note that they would have important career-related motives for
> doing so.)
Oh, come on. You can’t do any better than “There’s obviously a conspiracy among the professionals in this field”? You’re claiming that the linguistics professors of the world who specialize in American Indian languages are a tightly-knit group who regularly get together to control the world? They are the ones responsible for wars, depressions, and, just incidentally, declarations that anyone who claims to have reconstructed the earliest forms of American Indian ancestral languages is a crank. Why did I ever leave the field thirty-six years ago? If I had stayed, you would all be bowing before me in terror.