Re: THEORY: small inventories

Dirk:
> I don't think that there is any a priori reason why some languages
> should have small segmental inventories. What is more curious to me is
> why languages like Shoshone should derive such large surface
inventories
> from their relatively impoverished underlying stock (Gosiute has, by
my
> count, between 37 and 40 surface phones; over three times the number
in
> its underlying inventory), while the inventories of languages like
> Hawaiian show such stability and resistance to phonological
"diddling."

I suspect that what Mathias had in mind was not so much (I) the size of
the phoneme inventory but  (II) the number of linguistically-distinct
phonetic
segments in a given language. Without intending to patronize anybody,
it is quite common for nonspecialists to understand (I) as equivalent to
(II). (And indeed, the extrafictional history of Livagian phonology is
one
where (I) started out as equivalent to (II), but has since shrunk, while
(II)
has remained pretty constant or maybe even grown.)

Given the redefinition of the question, one does notice, say, that the
places
where speakers have to be phonetic acrobats or decathletes are rather
inaccessible, like the Caucasus or Namibia, or, indeed, to Scungria or,
to a much lesser extent, Livagia.

 --And.