--- In
cybalist@yahoogroups.com, Piotr Gasiorowski
<piotr.gasiorowski@...> wrote:
>
> ----- Original Message -----
> From: "tgpedersen" <tgpedersen@...>
>
> > Suppose a system p, p', b. On any substrate without glottalics
either p' > p or p' > b. This will "jam" the system apart. In the
first case you must exaggerate the p-ness of the p, giving ph and a
system ph, p, b.
>
> There are no "musts" in such cases. If you don't "exaggerate the p-
ness of the p", /p/ and /p'/ will simply merge (see below). In the
case of the labial series, at any rate, no harm would have been done
in an IE language, given the extremely low lexical incidence of *p'
(in the glottalic transcription).
>
> > In the second case you must exaggerate the b-ness of the b,
giving bh
> and a system p, b, bh (and further bh > ph in Greek, supplemented
> with ph in Indo-Aryan). I can't see there's any other way.
>
> Of course there are many other ways. For example, the original
system might resist substratal influence and survive, which under
your interpretation doesn't seem to have happened anywhere. Another
obvious possibility is a merger of two or even three rows: the latter
happened in Tocharian, the former at least in Celtic, Balto-Slavic,
Albanian and Iranian.
Alright! Tertium datur (no change). Et quartum et quintum (merger to
either side).
>
> Now isn't it curious that in all these groups the *d series merged
with the *dH series, not with the voiceless stops? The traditional
model of PIE phonology has an answer to this: *d and *dH were rather
similar, the former being voiced, the latter breathy-voiced. If the
original system was simplified, a merger involving the two of them
(and producing a {t, d} system) was more natural than the merger of
eithet of them with *t (sure enough, the latter didn't happen
anywhere). How do you explain the fact that, in terms of your version
of the glottalic theory, the unconditional merger of *t and *t' (to
the exclusion of *d) never occurred, while the falling together of
*t' and *d was so common?
I'll have to think about that.
> There's also another problem with your "substraticist" explanation.
If substratal influence is strong enough to undermine the original
system, you could expect a merger rather than a shift UNLESS the
substrate language also has three rows of stops. Otherwise, why give
up the ejectives only to create a new row of aspirates? Was Neolithic
Europe aswarm with languages having stop systems like {t, d, tH} or
{t, d, dH}? Why attribute to an unknown and unknowable substrate a
system that might just as well be PIE?
>
Don't get agitated now ;-). I thought you just pointed to the many
languages that _did_ merge? So, by your own admission, there were two-
stop substrates, and three-stop ones, the latter outnumbered.
Torsten