Re: Why do you say glottochronology is crap?

On Sep 19, 8:06 am, "Brian M. Scott" <b.sc...@xxxxxxxxxxx> wrote:
On Tue, 18 Sep 2007 09:33:18 -0700, Darkstar
<darkstar...@xxxxxxxx> wrote in
<news:1190133198.531415.119680@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
in sci.lang:

On Sep 17, 2:36 am, "Brian M. Scott" <b.sc...@xxxxxxxxxxx> wrote:
On Sun, 16 Sep 2007 09:24:50 -0700, Darkstar
<darkstar...@xxxxxxxx> wrote in
<news:1189959890.464949.33860@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
in sci.lang:
I've been wondering about this *idea* ever since it has
been proclaimed in the western linguistic community and
elsewhere on the net. Here's a typical example of this
discussion:
http://itre.cis.upenn.edu/~myl/languagelog/archives/000210.html
Actually, that discussion is considerably less dismissive
than many.

Here you made an unmarked snip, which I've restored with the
correct number of quote chevrons:

So, the 8 samples from Lees' work seem to be rather
consistent. The 2 other samples (Armenian, Georgian)
show moderate deviations, but not drastically high, so
they can still be seen as statistically viable.
So 'statistically viable' means 'compatible with Darkstar's
pet theory'? Convenient, but hardly science. And your
judgement is questionable anyway, since the retention rate
reported for Armenian is comparable to the Icelandic rate
and very different from the rates calculated for the eight
samples that you like.
No, Icelandic showed just 4 out of 200. That's much lower.

Bullshit. The reported rates are 0.976 (rural Icelandic),
0.962 (urban Icelandic), and 0.940 (Armenian).

Now, most argumentation seems to be based on blindly
repeating Bergsland & Vogt that, hey, you know Icelandic
retains most of it, which is simply ignoring the fact
that it is located on a _distant island_ with little
outside contacts, so the low loanword percetange and high
retentions rates are to be expected.
And what excuse are you going to find to exclude the almost
equally divergent Armenian data?
So Armenian is the only problem?

In the very limited data that you were using Armenian is the
obvious problem that you were too stupid or too dishonest to
notice.

[...]

The data presented on that web page are adequate to show
that there is great variation in lexical retention rates,
but they are far too meagre to give any real picture of the
distribution of those rates, let alone to allow any serious
attempt to correlate retention rates with geography or
history.
It's 7 good against 2-3 bad and 1 (East Greenlandic) that is
completely off-the-wall...

Bzzzzt!! You don't get to categorize data as good, bad, and
off-the-wall depending on how well they fit your preferred
hypothesis.

[...]
the one that is likely to have had an extensive bilingual
influence from Danish in the past which always brings a
huge stream of new words into a language.
Greenlandic contains four loanwords from medieval
Norse; from the colonial period after 1721 there have
been surprisingly few borrowings until the mid-20th
century.
Let's not discuss some remote and forloren languages whose
Swadesh lists are hard to obtain. Aren't there enough
well-known languages in the world?

You poor thing: you really haven't a clue. Even ignoring
the fact that 'well-known' is a matter of judgement, the set
of well-known languages obviously can't be assumed to be a
random sample.

<http://www.britannica.com/ebc/article-75301>
It's cited to have some mysterious tabooing of words or
something which is hardly or not at all attested in other
"normal" languages.
Rubbish. It's apparently unusually common in East
Greenlandic, but taboo avoidance is a well-known
phenomenon.
Are IE languages known to have that?
Look up the etymologies of English <bear> (the animal)
and Russian <medved'>.
That's the textbook example. But the truth is it's a
rather rare phenomenon.

You haven't the data to say how rare it is. It's certainly
not hard to find examples.

In Mexican Spanish <blanquillo> has largely replaced <huevo>
'egg' because of the prominent secondary sense, 'testicle',
of <huevo>. Similarly, in American English <rooster> has
largely displaced <cock>. The normal French word for 'fox'
is <renard>, from a masculine personal name; it has
completely replaced OFr <golpil>, from Vulg. Lat.
*vulpiculum, a diminutive of <vulpes>. The Turkic languages
have a word for 'wolf', <börü> (Kirghiz, Turkish dialect),
<böri> (Kazakh, Uyghur, Middle Turkic), etc., that I believe
in Turkish and some other Turkic languages has been replaced
by <kurt> 'worm' (Turkish). According to Larry Trask:

In all indigenous Australian languages, when a person
dies, his name becomes taboo, and not just his name,
but all words similar in sound to his name. For example,
when a man named _Djäyila_ died in 1975, the common
verb _djäl-_ 'want' became taboo in his community, and
was replaced by _duktuk-_, apparently borrowed from
a neighboring language (native Australians are
traditionally multilinguial).

And "bear" and other animal names are *not* part of the
Swadesh lists.

Naturally: the concepts included in the Swadesh lists were
supposed to be universal, and animals generally are
restricted to specific geographical regions. Who cares?
The Swadesh lists aren't magic; it's well known that they
don't even accomplish what Swadesh set out to do. It's
clear that the bear was extremely important in most of its
European range[*], so replacement of the IE term with
euphemistic terms in Germanic and Balto-Slavic is a
significant datum about language change.

[*] Presumably also in N. America, but I know virtually
nothing about the languages involved there.

(Why don't they cite Elfish as a counterpoint, anyway?)
What's even more ridiculous, "East Greenlandic" has
been studied for a period of just ONE GENERATION which
is a gross statistical error.
Presumably much of the evidence comes from comparison
with other dialects of Greenlandic and with closely
related languages. Statistics doesn't come into it at
all.
Greenlandic is still completely non sequitur/non compos.
That's an unverifiable experiment, as simple as that!
[...]

You're getting to be almost as ridiculous as Aggie.

[...]
So it's at least 10 rather good and well-attested,
statistically acceptable examples against 2 bad, hardly
verified or indescriminately selected.
More accurately, there are eight data points that support
your view, two more that badly strain your view but that you
choose to accept anyway, and two (actually three) that don't
and that you therefore find reasons, some clearly spurious,
to dismiss.
And curiously, the two bad instances somehow manage to
win over the other 10.
There's nothing curious about it at all. The data clearly
show that lexical retention rates vary over a broad range.
Have professional linguists ever heard about stuff like "normal
distrubution", "standard deviation" and "sigma rule"?

This is an obvious non sequitur, but I'll answer it anyway.
Some -- an increasing number -- are familiar with
statistics. Some even started out as mathematicians before
they turned to linguistics. There's also more and more
collaboration between historical linguists and people with
mathematical and statistical backgrounds. Many of the
mathematical tools, statistical and otherwise, are
considerably more sophisticated than you show any signs of
understanding.

[...]

The fact that I condescended to answer your last post hoping that you
start talking some sense, does not mean that I'm going to read your
gibberish blabbering. Learn to answer by factual argumentation first
before talking to me (and I don't mean stuff known to every school-
boy). Look at Ross Clark who's just presented some rather important
data on the An languages...

.

Relevant Pages

  • Re: Why do you say glottochronology is crap?
    ... that there is great variation in lexical retention rates, ... Swadesh lists are hard to obtain. ... well-known languages in the world? ... Statistics doesn't come into it at ...
    (sci.lang)
  • Re: Why do you say glottochronology is crap?
    ... reported for Armenian is comparable to the Icelandic rate ... sample of languages. ... that there is great variation in lexical retention rates, ... Statistics doesn't come into it at all. ...
    (sci.lang)
  • Re: Why do you say glottochronology is crap?
    ... reported for Armenian is comparable to the Icelandic rate ... sample of languages. ... that there is great variation in lexical retention rates, ...
    (sci.lang)
  • Re: Why do you say glottochronology is crap?
    ... that there is great variation in lexical retention rates, ... Danish documentary ... And very likely other Greenlanders ... Statistics doesn't come into it at all. ...
    (sci.lang)
  • Re: "STL from the Ground Up"
    ... Except that the vast majority of programmers already use more modern ... Gathering statistics from lots of different sources. ... Languages like C have strong staying power thanks to critical code like ... You can also analyze web statistics as TIOBE do: ...
    (comp.programming)