Retroflex Consonants in P.I.E. (was [tied] Re: A New language tree)

From: david_russell_watson
Message: 37785
Date: 2005-05-10

--- In cybalist@yahoogroups.com, "Patrick Ryan" <proto-language@...>
wrote:
>
> Thank you for your comments and valuable observations.

You're welcome.

> I am afraid I have to touch on Nostratic to even explain my
> reasons so please bear with me.

> In my own Nostratic reconstructions, I thought I observed
> a pattern that where an apical (or coronal) was followed by
> Nostratic *o, Semitic responded with a retroflex (emphatic).
> The same situation seemed to obtain in Proto-Dravidian. Just
> as I found, I think, that dorsals before Nostratic *e were
> palatalized in IE.
>
> If one has a method to determine Nostratic vowel qualities,
> this seems to work out fairly well.

Well I know little about Nostratic besides the fact that
Nostraticists must work backwards from P.I.E. and other
properly reconstructed proto-languages, and not the other
way around. Therefore one can't demonstrate the existence
of retroflexes in P.I.E. solely on the basis of their
being useful to one or another Nostratic theory.

Actually though, even if it were properly demonstrated
that P.I.E. descended from a language that had retroflexes,
the evidence from its daughters will still impel us to
believe that such retroflexes had completely merged with
the plain dentals prior to the P.I.E. stage. So no matter
how we look at it, there seems to be no evidence for
retroflexes in P.I.E. itself.

> Now, before proceeding, I do not understand how alveolar *s
> and *z can be said to be backed if the "backed" position is
> palatal or pre-palatal. I could easily accept "post-aveolar".

Well the pre-palatal and palatal positions are further back
in the mouth than the alveolar position, and so shifting to
either of the former from the latter constitutes backing.
Or are you instead objecting to the writing of sounds that
far back with 's' and 'z'?

> We all know that retroflexion is a notable characteristic
> of Dravidian so that any retroflexion in Sanskrit was probably
> viewed somewhat suspiciously as a mark of low register.

I believe the current view is that Munda or Munda-related
Austroasiatic substrates were the most likely source of
retroflexion in Indo-Aryan. If I remember correctly, it
has been suggested that they were the original source of
retroflexes in Dravidian as well.

> That could account for a gradual elimination of originally
> inherited retroflexes - if their was such a phenomenon.

Such a process wouldn't discriminate between retroflexes
arising after RUKI, and retroflexes supposedly inherited
from P.I.E., but we see no such pattern. Where the rules
that I described require a retroflex, one almost invariably
appears. In the few instances where it does not, there
is usually a very convincing explanation why it does not.
For one thing, some are next to the morpheme boundary
following a prefix ending in one of RUKI. On the other
hand, where a retroflex occurs outside the context of
the RUKI change, there is usually a clear Dravidian or
Munda source.

I'm sure a list of all of the unexplained retroflexes has
been made - though I don't know offhand where you might
locate it - the items on which you could compare to those
which you suppose had retroflexes in P.I.E., and see if
any pattern emerges.

> By the way, why not call it RUCKI? I would bet the Slavicists
> on the list would appreciate that.

The term was borrowed from the ancient Indian grammarians,
for whom the result of *c^ was 's^', and of the cluster
*c^s was 'kS' (S = retroflex 's') arising out of an older
*TS (T = retroflex 't'). Therefore they had no reason to
suspect the existence of an older *c^.

David