[tied] Re: ka and k^a

--- In cybalist@yahoogroups.com, Miguel Carrasquer <mcv@...> wrote:

> On Mon, 26 Sep 2005 15:00:34 -0500, Patrick Ryan
> <proto-language@...> wrote:
>
>
> > But, since you follow these matters more closely than I, is the
> > prevailing theory at this moment that *H2 does not color *o when
> > before or after it?
> >
> >And, if that is the prevailing view, why would that be?
>
> Why would it be the prevailing view? Because it's what best
> fits what can be observed.
>
> Why would */o/ be immune to laryngeal colouring? Well, I
> think it's significant that */e:/ is also immune to it. If
> */o/ was originally a long vowel, that would explain it.

Good point! I hadn't thought of that.

> I reconstruct a 2x3 vowel system for pre-PIE. Under the
> stress, the normal developments were:
>
> *a > *e
> *i > *e
> *u > *e
>
> *a: > *o:
> *i: > *e:
> *u: > *o:

Are there any attested languages in the world that reflect such
changes?

It seems to me that: 1) any language would preserve the quality of
stressed vowels more faithfully than unstressed; and 2) an originally
triangular vowel system would tend to develop lowered and/or laxed
allophones in unstressed (or closed) syllables. However, some Arabic
dialects have a partial merger of */a/ and */i/ (realized as [e] or
[E]).

I wonder if any given language must have at least *one* of two
contrasts (height and/or frontness), but that one is not more basic
than the other?

> Because *o: had no short counterpart, the length was not
> contrastive and could be lost (but *ó is still long in open
> syllables in Indo-Iranian, and does not get reduced to /ä/
> in Tocharian [as are /e/, /i/ an /u/]).

Unstressed */o/ is also long in InIr, though: e.g. Vedic _ma:náyati_
< *monéyeti.

- Rob