This should be good for a laugh
anal...@hotmail.com
Franz's Magadelenian is better thought out and mroe credible than
these studies.
benl...@ihug.co.nz
wrote:
> http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
>
> Franz's Magadelenian is better thought out and mroe credible than
> these studies.
Those of us whose interest in linguistics extends beyond scoffing will
probably wait to see the original papers before starting the laugh-
track. The Economist (whose readers apparently know what a "Venn
diagram" is, but have to have "cognate set" explained) manages to give
such a murky account of Dunn's paper that I can't even imagine what he
is actually doing, never mind the implications for Chomsky (another
allusion Economist readers apparently need help with). Atkinson's
phoneme-distance correlation (the only data supplied) looks pretty
feeble, and the phoneme/size correlation they claim has been "known
for a while" is AFAIK pretty weak in itself. Still, I would rather
read Atkinson himself than assume The Economist has got it right. The
fact that you compare these unfavourably to Franz's Fancies, however,
gives us a good indication of where your vantage point lies.
Harlan Messinger
I had come across the New York Times article on this earlier today:
http://www.nytimes.com/2011/04/15/science/15language.html?_r=1&hp
I wasn't keen on the report of Atkinsons reliance on the genetic
*metaphor* for language as a *model* of phonemic/phonetic evolution.
Harlan Messinger
I wonder if he tested a variety of metrics (number of vowel phonemes,
number of consonant phonemes, ratio of consonant phonemes to vowel
phonemes, number of consonant sequences, number of affricates, etc.) and
chose number of phonemes because it happened to give him what he felt
was a neat result.
Athel Cornish-Bowden
> On Apr 15, 11:11 am, "analys...@hotmail.com" <analys...@hotmail.com>
> wrote:
>> http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
>>
>> Franz's Magadelenian is better thought out and mroe credible than
>> these studies.
>
> Those of us whose interest in linguistics extends beyond scoffing will
> probably wait to see the original papers before starting the laugh-
> track. The Economist (whose readers apparently know what a "Venn
> diagram" is, but have to have "cognate set" explained)
That seems reasonable to me. I suspect that if you drew a Venn diagram
with two regions, one containing all the people who know what a Venn
diagram is, the other containing all the people who know what a cognate
set is, you'd find the latter very largely contained within the former.
You might get a different result if you replaced "people" by
"linguistics professionals".
--
athel
benl...@ihug.co.nz
wrote:
Maybe. I seem to remember seeing and understanding Venn diagrams in
high school, but I don't believe I knew a name for them until
university mathematics classes. Maybe they're more widely talked about
in schools these days, or used in economics, sociology, or other
things readers of The Economist would know about?
Athel Cornish-Bowden
I've looked at both papers (in Science and Nature as well the
Economist) and I see nothing to laugh at. They are both serious
studies, of which Atkinson's paper is the easier to read for a
non-specialist. That doesn't mean that their conclusions are correct,
of course, and I'll be interested to know what the experts think.
Atkinson is in effect looking at the second principal component,
recognizing that population size is a more important variable, and I
didn't think he sufficiently explained how he corrected for that in
studying the distance from Africa.
The idea that Franz's fantasies are better thought out is, of course,
absurd. Anyone who wants to repeat Atkinson's study can do it without
too much difficulty. Repeating Franz's is impossible, because he just
asserts conclusions without ever explaining the methods used to arrive
at them.
--
athel
Ruud Harmsen
<anal...@hotmail.com>: in sci.lang:
>http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
===
It has been known for a while that the less widely spoken a language
is, the fewer the phonemes it has.
/===
That is not my observation. For example, if I look at English, German,
Hungarian and Frisian, I see a comparable range of phonemes, but also
a marked difference in geographic spreading, from world-wide to quite
local.
--
Ruud Harmsen, http://rudhar.com/new
Ruud Harmsen
>http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
When looking at the diagram of "Phonetic complexity and geographical
location of languages", I don't find it convincing at all. What I see
is that the complexity varies a lot and there is no correlation
whatsoever with distances to Africa.
The only obvious thing is that some Polynesian languages (all in a
common family) have rather small phoneme inventories.
As I understand it, they tried the same thing with hundred of other
hypotheses about the origin of language per se. If this is the best
result they got, when assuming Africa, it means the other results
showed even more randomness!
Ruud Harmsen
>http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
The last 5 paragraphs are interesting though, and not funny at all
(not in the sense of rediculous, but amusing they are).
António Marques
> I seem to remember seeing and understanding Venn diagrams in
> high school, but I don't believe I knew a name for them until
> university mathematics classes. Maybe they're more widely talked about
> in schools these days, or used in economics, sociology, or other
> things readers of The Economist would know about?
Odd as it is, we were introduced to the thing along with its name in primary
school. Of course, nobody really called them Venn diagrams, rather 'sets',
but the name may have remained in the memory of some. The episode stuck
precisely because even 7 year olds noticed the incomprehensibleness of there
being (and we being taught) such a formal name for what amounted to
circles/'sets'.
António Marques
And to the point, even The Economist's article is prima facie credible - to
people who know little of the matter. Not so with Franz's word dumps.
Ruud Harmsen
Athel Cornish-Bowden
Yes, I should have noted that. It seems a very surprising statement. I
don't speak either Chinese or Georgian, but my impression is that
Chinese has rather few phonemes, whereas Georgian has lots. As a
Portuguese speaker you can perhaps correct my impression (or António
can) that Portuguese has more phonemes than Spanish. Of course, they
will doubtless argue that their claim is true only in a statistical
sense, but if so they need an "in general" in there somewhere, or
"tends to have" rather than "has".
--
athel
Ruud Harmsen
<athe...@yahoo.co.uk>: in sci.lang:
>> ===
>> It has been known for a while that the less widely spoken a language
>> is, the fewer the phonemes it has.
>> /===
>>
>> That is not my observation. For example, if I look at English, German,
>> Hungarian and Frisian, I see a comparable range of phonemes, but also
>> a marked difference in geographic spreading, from world-wide to quite
>> local.
>
>Yes, I should have noted that. It seems a very surprising statement.
Then again, the spreading of English is over a time scale of maybe 400
years (or some 1700 when including the jump to the British Isles),
whereas IE is maybe 6k-10k and the investigation discussed looks at
50k -100k years.
>Chinese has rather few phonemes, whereas Georgian has lots. As a
>Portuguese speaker you can perhaps correct my impression (or António
>can) that Portuguese has more phonemes than Spanish.
It has many more indeed.
>Of course, they
>will doubtless argue that their claim is true only in a statistical
>sense, but if so they need an "in general" in there somewhere, or
>"tends to have" rather than "has".
Click languages versus Hawaiian is a lot more convincing.
Athel Cornish-Bowden
> Thu, 14 Apr 2011 16:11:57 -0700 (PDT): "anal...@hotmail.com"
> <anal...@hotmail.com>: in sci.lang:
>
>> http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
>
> When looking at the diagram of "Phonetic complexity and geographical
> location of languages", I don't find it convincing at all. What I see
> is that the complexity varies a lot and there is no correlation
> whatsoever with distances to Africa.
>
> The only obvious thing is that some Polynesian languages (all in a
> common family) have rather small phoneme inventories.
That, of course, is a statistical trap that is too easy to fall into.
Just last week I was at a meeting at which a speaker presented some
data for a microbial model, in which two variables were compared for
different cultures, of which about half were aerobic (in the presence
of air) and half were anaerobic. They gave a correlation coefficient of
about 0.95. However, they were clearly in two clumps, and if you looked
just at the aerobic group they gave a correlation coefficient of at
best about 0.55, and similarly for the anaerobic group. (I should say
that the speaker in this case was perfectly well aware of the problem,
and said that he did it that way because the referees of his paper
insisted.) As it happens there were 18 men in the room when he said it,
and one woman, and I was thinking that it would be easy to choose two
variables that would give an excellent correlation if one included the
woman in the sample and no significant correlation at all if one
didn't. Looking at Atkinson's paper I felt that there was an element of
this, but I wouldn't go so far as to say that there was "no correlation
whatsoever with distance from Africa".
--
athel
António Marques
I'm dense today, could you elaborate? (Also on the microbes.)
> Looking at Atkinson's paper I felt that there was an element
> of this, but I wouldn't go so far as to say that there was "no correlation
> whatsoever with distance from Africa".
With the possibility of closely related languages having very dissimilar
metrics, I'm not sure his data are that meaningful. But even if they were,
there may be better correlations than [fewer phonemes <-> further from
Africa] which nobody cared to analyze.
Joachim Pense
>
> That seems reasonable to me. I suspect that if you drew a Venn diagram
> with two regions, one containing all the people who know what a Venn
> diagram is, the other containing all the people who know what a cognate
> set is, you'd find the latter very largely contained within the former.
> You might get a different result if you replaced "people" by
> "linguistics professionals".
>
Is the set of Venn diagram knowers roughly equal to the generation that
got "new math" at school?
Joachim
Joachim Pense
>
> Odd as it is, we were introduced to the thing along with its name in
> primary school. Of course, nobody really called them Venn diagrams,
> rather 'sets', but the name may have remained in the memory of some. The
> episode stuck precisely because even 7 year olds noticed the
> incomprehensibleness of there being (and we being taught) such a formal
> name for what amounted to circles/'sets'.
Today, sets seem to be "out" at school (at least in Germany). My
children (11 to 16) never heard of them, while for us they were just
normal. I remember a Maths Prof who started the first lecture for the
beginner students with "Theorem 1: Everybody knows what a set is" and
proceeded without any definition. He could not do that today.
Joachim
Athel Cornish-Bowden
I'll try. (It'll look better if you display it in a fixed-width font
like Courier.)
Suppose you choose two variables for both of which men tend to resemble
other men more than they resemble women, and women tend to resemble
other women more than they resemble men, but which are not particularly
well correlated between men or between women. Maybe liking for
pornography and frequency of attending football matches might do, but
with a bit of thought one could doubtless choose better variables. Then
if you plotted them against one another you might get something like
this (where M = man, W = woman):
M M
M
M M M
M M
M M M
M M
M M
M M
M
W
If you calculate a correlation coefficient just for the points M you'd
probably find a value that was greater than zero, but not very
impressive. If you add point W you'd get a much higher value.
For the microbial cultures (A and B) it was more like this:
A
A
A A
A
A
A
A
A
B B B
B
B B
B
B
with not much correlation within either of the two populations, but an
apparently good correlation if you forget that they are two populations
and treat them as one.
--
athel
Peter T. Daniels
My class (fortunately) was the last one that didn't get "new math" (we
would have liked to play with the Cuisinart rods too, but we
couldn't), but we had Venn diagrams (by name), surely at least by 7th
grade if not before.
Trond Engen
Much about sets in maths in my earliest schooldays in late seventies'
Norway too. I think it was the misguided idea that the right way to
proceed is from generalized, theorethical concepts towards mere
practical applications like arithmetics. Still, sadly, we never used
formal set operations, which probably would have made it a lot more
interesting (and useful). And I didn't encounter a Venn diagram (by that
name) until university.
My children (9 and 12) have hardly seen sets in maths class. But they do
use Venn diagrams in other classes as part of the contemporary oddity of
"learning strategies", where they're supposed to apply methods like that
to the solution of a task.
--
Trond Engen
Athel Cornish-Bowden
Up to a point, yes, if you include the parents of children who met Venn
diagrams at school and wanted their parents to explain. In those cases
the parents probably wondered why something as obvious and trivial
needed a fancy name -- I was "using" them a long time before I'd "heard
of" them!
--
athel
Harlan Messinger
It has always puzzled me that Sesame Street has always had skits that
pound away at oppositions like "big" versus "little" or "inside" versus
"outside", as though these are tough concepts and there are children who
struggle later in school for failure to have grasped these concepts
early. Likewise, it has always surprised me that the basics of sets were
*taught* when I was in primary school, as though such concepts as
"Johnny is in our class and Sarah is not in our class, and our class and
Sarah's class have nobody in common" eluded us. The only things that
really needed to be taught were the terminology ("member", "subset",
"intersection", "union", "complement") and the associated symbols, but
it isn't clear what the point of that is at age 10 when no one in the
class would have further use for any of those terms or symbols for
another six to ten or more years, if ever.
António Marques
Our first 4 years of school used to hammer away at such logical operations,
even going as far as pointing the isomorphism with logical ones
(conjunction, disjunction, though xor only came later). Not using those
terms, the only fancy one was 'Venn diagram'. As to its usefulness, I have
no idea. Looking around I don't see any geniuses, math or otherwise, but...
António Marques
> Suppose you choose two variables for both of which men tend to resemble
> other men more than they resemble women, and women tend to resemble other
> women more than they resemble men, but which are not particularly well
> correlated between men or between women. Maybe liking for pornography and
> frequency of attending football matches might do, but with a bit of thought
> one could doubtless choose better variables. Then if you plotted them
> against one another you might get something like this (where M = man, W =
> woman):
>
> (snip graph my composer refuses not to break)
>
> If you calculate a correlation coefficient just for the points M you'd
> probably find a value that was greater than zero, but not very impressive.
> If you add point W you'd get a much higher value.
>
> For the microbial cultures (A and B) it was more like this:
>
> (snip graph my composer refuses not to break)
>
> with not much correlation within either of the two populations, but an
> apparently good correlation if you forget that they are two populations and
> treat them as one.
Very nice and clear, thanks.
(I should add it seems that women like porn even more than men, it's just
that they're only learning about it now. At least based on an internet
article I've read t'other day!)
Brian M. Scott
[...]
> Today, sets seem to be "out" at school (at least in
> Germany). My children (11 to 16) never heard of them, while
> for us they were just normal. I remember a Maths Prof who
> started the first lecture for the beginner students with
> "Theorem 1: Everybody knows what a set is" and proceeded
> without any definition. He could not do that today.
If he's being mathematically honest, he'd *have* to do that
today: 'set' is a primitive. It's defined only implicitly,
by the axioms that one adopts (e.g., ZF, ZFC, GB, etc.). At
most he could give an intuitive description.
Brian
Franz Gnaedinger
> On Apr 15, 11:11 am, "analys...@hotmail.com" <analys...@hotmail.com>
> wrote:
>
>
> > Franz's Magadelenian is better thought out and mroe credible than
> > these studies.
>
> probably wait to see the original papers before starting the laugh-
> track. The Economist (whose readers apparently know what a "Venn
> such a murky account of Dunn's paper that I can't even imagine what he
> is actually doing, never mind the implications for Chomsky (another
> allusion Economist readers apparently need help with). Atkinson's
> phoneme-distance correlation (the only data supplied) looks pretty
> feeble, and the phoneme/size correlation they claim has been "known
> for a while" is AFAIK pretty weak in itself. Still, I would rather
> read Atkinson himself than assume The Economist has got it right. The
> fact that you compare these unfavourably to Franz's Fancies, however,
> gives us a good indication of where your vantage point lies.
Ross Clark did not go for my two Magdalenian
test cases. I repeated the first one extra for him,
because he was too lazy to go through the long
thread on the question of the etymology of bear
started by Analyst in the summer of 2008: bear
as the furry one versus bear as the brown one;
nor did he go for my second Magdalenian test case:
deus theos / Zeus vs. deus Zeus / theos. Calling
my work Fancies makes me call him a Doctrinary
who considers present textbook wisdom eternal
truth (we have several textbook fetishists and one
veritable textbook fashist among us), and the look
backward in time as the only valid one, a priori
superior to the experimental approach that
assumes a position in the past and then goes
forward in time. The Economist article - good
science in my opinion, not laughable -, suffers
from the same exlusively backward oriented
look, top down approach, not knowing what
language is. My approach is based on my
definition of language from 1974/75:
Language is the means of getting help,
support and understanding from those
we depend upon in one way or another,
and any means of getting help, support
and understanding may be called language,
on whatever level of life it occurs ... This goes
along well with the present understanding of
humin language that is believed to have
emerged some two million years ago,
embedded in gestures. If you think from
there, you can easily see that word language
originated several times - as eyes originated
seventeen times in the course of evolution -,
but all the different languages are still one,
embedded in body language, gestures,
mimics, a smile has the same meaning
everywhere, among all peoples, element of
the universal humin language that comes
before language (I just read the latest novel
by T. C. Boyle, he speaks about flesh to flesh,
the language before language, I like it very
much and use the formulation here).
Harlan Messinger
> On Apr 15, 4:09 am, "benli...@ihug.co.nz"<benli...@ihug.co.nz> wrote:
>> On Apr 15, 11:11 am, "analys...@hotmail.com"<analys...@hotmail.com>
>> wrote:
>>
>>> http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
>>
>>> Franz's Magadelenian is better thought out and mroe credible than
>>> these studies.
>>
>> Those of us whose interest in linguistics extends beyond scoffing will
>> probably wait to see the original papers before starting the laugh-
>> track. The Economist (whose readers apparently know what a "Venn
>> diagram" is, but have to have "cognate set" explained) manages to give
>> such a murky account of Dunn's paper that I can't even imagine what he
>> is actually doing, never mind the implications for Chomsky (another
>> allusion Economist readers apparently need help with). Atkinson's
>> phoneme-distance correlation (the only data supplied) looks pretty
>> feeble, and the phoneme/size correlation they claim has been "known
>> for a while" is AFAIK pretty weak in itself. Still, I would rather
>> read Atkinson himself than assume The Economist has got it right. The
>> fact that you compare these unfavourably to Franz's Fancies, however,
>> gives us a good indication of where your vantage point lies.
>
> Ross Clark did not go for my two Magdalenian
> test cases.
So? Come to think of it, I still haven't figured out what you think it
would mean to "go for" your "test cases" (which aren't test cases).
yangg
wrote:
> http://www.economist.com/node/18557572?story_id=18557572&fsrc=rss
>
> Franz's Magadelenian is better thought out and mroe credible than
> these studies.
***
As far as a good laugh is concerned here's two reviews:
http://diachronica.pagesperso-orange.fr/TMCJ_vol_2.1_Fournet_Review_of_Talageri.pdf
http://diachronica.pagesperso-orange.fr/TMCJ_vol_2.1_Fournet_Review_of_APUA5_2010.pdf
A.
Franz Gnaedinger
On BBC I heard a brief review of Atkinson's paper
in Science: 'modern language' originated 160,000
or even 200,000 years ago in subsaharian Africa,
where the richest variety of phonemes is found,
and, not coincidentally, also the richest genetic
variety in humans. Early modern language was
the same as recent language - what Peter T.
Daniels keeps telling me. I guess a genetic shift
and recombination of a rather small but important
part of the human genome turned a former Homo
into Homo sapiens sapiens, lending us the fine
motoric abilities in moving our fingers and lips
and tongue, so that our linguistic abilities could
be filled with language, modern word language.
How did it sound? I guess this can be found out
by means of the experimental approach: form
words by onomatopoesis and look what you can
say with them - not writing computer manuals,
of course, but express what was useful for life
in the Stone Age.
Harlan Messinger
>
> On BBC I heard a brief review of Atkinson's paper
> in Science: 'modern language' originated 160,000
> or even 200,000 years ago in subsaharian Africa,
> where the richest variety of phonemes is found,
> and, not coincidentally, also the richest genetic
> variety in humans. Early modern language was
> the same as recent language - what Peter T.
> Daniels keeps telling me. I guess a genetic shift
> and recombination of a rather small but important
> part of the human genome turned a former Homo
> into Homo sapiens sapiens, lending us the fine
> motoric abilities in moving our fingers and lips
> and tongue, so that our linguistic abilities could
> be filled with language, modern word language.
> How did it sound? I guess this can be found out
> by means of the experimental approach: form
> words by onomatopoesis and look what you can
> say with them - not writing computer manuals,
> of course, but express what was useful for life
> in the Stone Age.
Making it up yourself is not "experimental" and doesn't determine how it
actually happened.
Franz Gnaedinger
Having read more on the article by Quentin
Atkinson in Science (but not yet the paper itself)
I dare say that the genetic shift and recombination
allowed for a sound language - testified to by the
189 (hundred-eighty-nine) phonemes of the San
who are genetically closest to early Homo sapiens
sapiens -, which then, slowly, accompanying the
technological process, turned into word language,
fewer phonemes but more permutations, the idea
of permutations reaching its peak in Ice Age
Eurasia, from the Aurignacian to the Magdalenian.
Franz Gnaedinger
> Having read more on the article by Quentin
> Atkinson in Science (but not yet the paper itself)
> I dare say that the genetic shift and recombination
> allowed for a sound language - testified to by the
> 189 (hundred-eighty-nine) phonemes of the San
> who are genetically closest to early Homo sapiens
> sapiens -, which then, slowly, accompanying the
> technological process, turned into word language,
> fewer phonemes but more permutations, the idea
> of permutations reaching its peak in Ice Age
> Eurasia, from the Aurignacian to the Magdalenian.
Sorry for the typo, I meant to say technological
progress.