The Indo-European Jihad (was:IE SEMITIC relations) - sci.lang

Viesti keskustelusta The Indo-European Jihad (was:IE SEMITIC relations)

Brian M. Scott

Lisäasetukset 19 joulu 1997, 10:00

Uutisryhmät: sci.lang

Lähettäjä: sc...@math.csuohio.edu (Brian M. Scott)

Päivämäärä: 1997/12/19

Aihe: Re: The Indo-European Jihad (was:IE SEMITIC relations)

  Välitä | Tulosta | Näytä samanaiheinen viestijoukko
  | Näytä alkuperäinen
  | Raportoi tästä viestistä | Etsi tämän kirjoittajan viestit
  

On 17 Dec 1997 22:32:16 -0500, hu...@pegasus.montclair.edu (H. M.

Hubey) wrote:
>This problem of finding out what is due to chance and what not
>is not difficult. It has been done. Here it is

It's extremely difficult, like virtually all serious problems in
mathematical modelling. You can make unreasonable assumptions and
keep the mathematics tractable (in this case trivial), or you can try
for more realistic assumptions. In this case it's by no means
self-evident what reasonable hypotheses would be.

[snip]

>Here is the concise formulation of the problem by
>Donald Knuth:
>A group of N fans of the winning football team throw their
>hats high into the air. The hats come back randomly, one hat
>to each of the N fans How many ways h(n,k) are there for
>exactly k fans to get their own hats back?

So you're assuming that the phonemes of a word are independent,
uniformly distributed random variables; this is plainly unreasonable.
And you're insisting on exact matches, which under a sufficiently
powerful semantic microscope probably *never* occur, despite the fact
that those who argue for relationships on the basis of superficial
resemblance invariably allow themselves considerable leeway in both
sound and meaning. On the strict interpretation that lets your model
come anywhere near the facts, it's essentially irrelevant anyway.

At the very least you probably have to treat semantic space as a
measure space with some reasonable associated metric. With very mild
assumptions this leads to the conclusion that the expected number of
matches is a non-decreasing function of the allowable semantic
distance. Here, however, I'm still making the unreasonable assumption
that the languages have the same phoneme inventory, a statement that
I'm not sure is even meaningful. One should really do the same thing
to phonetic space that I did to semantic space, but then matters get
substantially more complicated.

>Here is the solution:

[snip of irrelevant elementary combinatorics]

>It is time to stop looking for lots of accidental matches.
>There are only a few that all of the world's linguists have
>been able to find, and they can be found in books, treasured
>like valuable heirlooms. But pointing out that such words
>exist does not mean that one can draw the conclusion that
>where there is one, there must be hundreds more. There aren't!

Since I know that you've been reading sci.lang longer than I have, I
can only conclude that you haven't been paying attention. Hardly a
month goes by without someone posting a new set of superficial
matches.

>Laws of probability theory take precedence over simple
>heuristic rules of 19th century historical linguists.

And evidence takes precedence over probability theory. But even
without the evidence it should be obvious that the cap problem is a
poor model of the linguistic reality.

Brian M. Scott

Välitä