Re: [tied] The phonetic value of PIE *h3 and the 'drink' root.

On Wed, 24 Jul 2002, Miguel Carrasquer wrote:

> On Wed, 24 Jul 2002 03:38:59 +0200 (MET DST), Jens Elmegaard Rasmussen
> <jer@...> wrote:
>
> >The instrumental morpheme *-bhi is certainly founded on the plural
> *-bhis,
> >so -eaw looks *very much* as a formation based on instr.pl. -eawk'. That
> >strips -eaw of its probative power and leaves -o- visible in
> >the singular only, and -a- visible in the plural (with its accessory)
> >only. And that makes it *very* likely that the -o- of the loc.sg. -oj^
> is
> >a stem-forming *-o- and nothing else. I completely fail to see the
> wisdom
> >in taking the instrumental in -eaw to be based on the nominative. Is
> that
> >ever done in this language?
>
> I should have said: "based on the nominative-accusative".  In knaw
> (I.pl. kanambk`).  Arguably in the o-stems (-ow < -o + -bhi), where
> the PIE instrumental plural was *-o:is.

I admit I had overlooked cases of incongruence in stem-formation between
the instr.sg. and teh instr.pl. which do exist, if only sporadically. For
-ov it seems forced to disregard -ovk' which is a fact even if it is not a
direct continuation of the PIE protoform. My basic objection to your whole
line of reasoning is precisely that "arguably" is not enough. It takes
positive evidence to introduce new PIE morphological types as you are
doing. That's why I have called it wishful thinking.

>
> >I am seriously challenged if I am to see a form in -oj^ which, for all
> the
> >versatility you would like to ascribe to it, is nowhere an instrumental,
> >as evidence for how the specific form of the IE instrumental was. Is
> there
> >an easy way to get "pre-Armenian [...] I. -oj^" to give up its
> >instrumental function? I notice that you are not suggesting widening of
> >the function of the instrumental (as in Gothic, where the functions of
> the
> >old instrumental are also carried by the socalled dative), but a
> complete
> >displacement of the instrumental form to serve some other case functions
> >and only those. I find that most unsatisfactory.
>
> The Armenian ins.sg. nowhere continues the PIE ins.sg., so a complete
> displacement of the instrumental form to serve some other case
> functions is perfectly possible.

Again, then there is no evidence, so why assume something special?

>
> >I regard the correspondence between Skt. -aya: and OCS -ojoN as the
> joint
> >reflex of what would be *-aya: in Brugmannian terms (the BSl. nasal is
> >secondary, from rudimentary *-mi from the other stam classes).
>
> What's rudimentary *-mi?  *-aya: + *-mi would have given *-oyamI in
> Slavic.  I would rather compare the Skt. Loc. in -a:ya:m, where we
> have the same *-m as in Slavic.

The story appears to be repeated in the OCS 1sg prs. in -oN which would be
to Skt. -a:mi exactly as -ojoN is to Skt. aya: + -mI of other stems. Also
Lith. instr. -a was earlier nasalized, and is still so marked in the
definite adjective sausáN-ja.

>
> >If the
> >instr.sg. was always end-stressed in mobile paradigms, i.e. ended in
> >full-grade -VH1 even in PD words, the devi: type would have *-iH2-VH1.
> >Now, if that ending were to influence the old form in *-aH2-H1 (which
> >survives as a variant in Vedic and elsewhere), how could it produce a
> form
> >in *-aya: ? I'd say it could do just that by adding *-iH2VH1 after the
> >*-a- of the old form
>
> But why -a- and not -ah2-?

Because I expected *taH2iH2aH1 to end up a trisyllable, IIr. *ta.i.(y)a:,
from which I would get Skt. *teya:, not táya:. But perhaps the full form
*could* work, if the form was properly a compound (or acted as one).
Forms like DsgM tásmai, DsgF tásyai *must* be old compounds, something
like 'this one'.

Jens