Re: [tied] Re: Indo-European /a/

elmeras2000 wrote:

>
> --- In cybalist@yahoogroups.com, Andrew Jarrette <anjarrette@...>
> wrote:
>
>
>>Is the sound /a/ considered to be an original phoneme of Proto-
>
> Indo-European? [...]
>
> Everything is considered and accepted by some. If you mean, do
> individual list-members consider /a/ an original PIE phoneme, the
> answer for me is yes and no.
>
> Yes, I must accept /a/ as a PIE phoneme, since it sometimes occurs
> in environments where it would cause even greater embarrassment to
> posit /H2e/ or /eH2/. Some examples are *yag^- 'sacrifice', *sal-
> 'salt', *na:s- 'nose', but there are not many.

[For Andrew, since Jens knows it all very well:] Kurylowicz (1956) and
Wyatt (1970, in the booklet mentioned by Daniel) draw checklists of
roots reconstructed with "primary" *a (up to thirty-odd items long). The
most radical rejection of PIE *a can be found in Lubotsky's 1989 article
entitled, bluntly, "Against a PIE phoneme *a".

I agree with Jens that there are roots where it would be ridiculous
_not_ to reconstruct independent *a without an adjacent *h2. I don't
pretend I know what to make of *nas- and *sal-. The latter _could_
contain an *h2 if somehow connected with *saus-/*sus- 'dry' < *s(a)h2us-
(?), but unlike Lubotsky I find it impossible to accept Kortlandt's
analysis of *g^Hans- 'goose' (onomatopoeic?) and *nas- as, respectively,
*g^Heh2ns- and *neh2s-, or Lubotsky's own idea of forcing a laryngeal
into *bHag- 'divide' just in order to make his point. In some cases
coloration by a *k, *g or *gH seems likely (as in *twak- or *bHag-), but
cases of "unexplained" *a (as in *k^aso- 'hare') should probably be
accepted at face value.

> And No, I do not consider /a/ a very old member of the PIE phonemic
> system. It does not reveal any complementary distribution over and
> against the immensely common /e/, so it seems that items with truly
> independent /a/ (i.e. /a/ not from /e/ by influence from *H2 or
> plain velars) are simply secondary additions to the vocabulary
> introduced after a preceding clash of whatever older vowels there
> have been into the monotonous /e/.

As an alternative, one might consider the possibility that early *a was
relegated to the peripheries of the system since it did not participate
in productive vowel alternations and so had no grammatical function to
play. This is more or less Kurylowicz's position: he points out that *a
occurs mostly in isolated "concrete" lexemes.

Piotr

> The vowel system indicated by the PIE lexicon is one of a single
> vowel phoneme (later surviving as /e/ where not changed by special
> rules). That has been considered reason enough to reject it by many.
> While it is true and remarkable that no language is really known to
> make do with a single vowel phoneme, it is not very remarkable that
> no language has been recorded in precisely the stage when it has
> just merged all its vowels and has not yet moved on to create new
> ones. Old Indo-Iranian (Sanskrit, Avestan, Old Persian) comes close,
> but by the time of the oldest records there have been changes that
> demand some diversity to be posited for the phonemic system. But if
> the period before the emergence of the earliest differentiated
> vowels was a short one, that is only to be expected.