--- In
cybalist@yahoogroups.com, "Andrew Jarrette" <anjarrette@...>
wrote:
>
> (snip)
> Also, according to this theory, which
> are more original, centum or satem; if the centum velars are the more
> original sounds, why did all languages nearest to the urheimat
> (excluding Tocharian) participate in a shift velar>palatal; if the
> satem palatals are more original, why did all the most western
> languages participate in a shift palatal>velar, which I personally
> find rather implausible? Isn't it easier to suggest that the velars
> are the more original, and in like fashion the original homeland was
> (much) nearer than India to those languages that had the velars rather
> than the palatals?
>
> Andrew
>
I don't think there's any automatic positive correlation between
linguistic conservatism and geographic continuity. In fact, some
linguists have suggested that the opposite occurs, based on dialect
distribution of colonial languages (and maybe some other things).
NOTE: I see that I haven't actually said much that's relevent below,
but since I took a lot of trouble to write it, I'm leaving it in.
As for PIE dorsals, there are 3 views that I can recall.
(A) the traditional view, with 3 types: palatal, velar, and labiovelar.
Here, the satem languages merge the labiovelars with the velars (no big
problem), but the centum languages, which are historically
discontiguous in a big way, must merge the palatals with the velars,
which to me suggests a substrate (e.g. it's been proposed that the lack
of dorsal palatalization in Sardinian before front vowels is due to a
possibly Afroasiatic substrate).
(B) a modified view (previously discussed on CyBaLiSt) also with 3
types: velar, uvular, and labiovelar. The centum languages only have to
merge the uvulars with the velars. This also has the advantage of
making the less common set of phonemes correspond with the
more "marked" set of sounds.
(C) In this view, there were originally only velars and labiovelars.
Then the satem languages underwent an originally conditioned
palatalization which was obscured by analogy.
I don't think (A) is very tenable, regardless of the homeland's
location, but it would cause problems for the Indian homeland
hypothesis, in that the centum languages would have to all leave in a
single group, while Germanic has much in common with Balto-Slavic and
Greek with Indo_Iranian.
For (B) and (C), I don't see anything relevent to the question,
although I'll mention that I don't see the earmarks of unidirectional
expansion either.