On 05-02-15 20:13, Miguel Carrasquer wrote:
> Of course I'm not denying that the root *pot- occurred with
> the "stative" suffix *-eh1(i)-. I just don't see how
> inserting a laryngeal makes the derivation of *potnih2 any
> easier, and the participal suffix -nt- (why was it reduced
> to -n-?) disconnects the feminine from the masculine form.
The reduction of *-nt- to -n- is hardly unusual in the oldest layer of
substantivised participles. We have quite a few agent nouns like
*tek^þ-o:n (excuse the thorn), where the *-on- can scarcely be anything
else but an allomorph of *-ont-. Note also the variation of *-n- and
*-nt- in the individualising suffix (whether or not ultimately related
to the pres.part.). Presumably the old pronunciation of //-ont-// +
nom.sg. //-s// was -o:n, with the whole obstruent part of the coda
deleted. I would therefore expect *póth1o:n --> *pot(h1)nih2 (influenced
by the surface form of the masculine word), just like *ték^þo:n >
*tek^þn[.]ih2 (Gk. tektaina, Skt. taks.ni:). The dropping of *h1 already
in PIE is perhaps irregular (one can't invoke Saussure's Law here), but
any conventional term of reverent address is likely to be so contracted
(obvious examples are aplenty).
> A form *pot-h1i-n(t)-ih2 would have given Skt. *path(i)ni:,
> not patni:. Same goes for a masc. *pot-h1i-s, where would
> then expect Av. *pa(i)Ti, not <paiti>. Cf. the Indo-Iranian
> forms of N. *póntoHs, A. *póntHm., G. *pn.tHós (Skt. N.
> pántha:s, A. pántha:m, obl. path-; Av. N. pantå:, A.
> panta,m, obl. paT-, OPers. A. paTim).
I'm not at all persuaded that *h1 (as opposed to *h2) causes aspiration
in Indo-Iranian. At the very least we could get oblique forms like
*poth1jos > *potjos already in PIE (via Pinault's Law) with their *t
spreading analogically, though I doubt if any such explanation is really
necessary.
As regards the origin of *póth1i-, I take it to be in the same relation
to *poth1ont- as the first element of Avestan tac^i-a:p- 'having flowing
water' is to *tekW-ent- 'flowing' or (perhaps) the second element of
Slavic *medv-e^dI (< *medHu-h1edi-) 'bear' to *h1ed-ont- 'eating'. I
don't want to take Caland's name in vain or to get sidetracked into
another discussion of that phenomenon, but any explanation which
accounts for such substitutions in general will work for *poth1on(t)- ~
*poth1i- as well. In other words, I regard *poth1i- as originally the
compositional counterpart of free-standing *poth1on(t)- (as in "rerum
omnium potens Iuppiter"), found in such frequently-used compounds as
*gHosti-poth1i-, *wik^-poth1i-, *dems-poth1i-, *swe-poth1i- etc. (the
full form of the pres.part. was consistently avoided in this environment).
> As is the case in the u-stems, the presence of original
> -(i)n- in what later became i-stems should not be a
> surprise. Not only do we have neuter i/n-stems in Vedic,
> but some forms originating in the old in-stems have
> permeated the i-stem paradigm (although to a much lesser
> degree than in the u-stems): Isg. -ina:, Gpl. -i:na:m, NA
> du. n. -ini:, NA pl. n. -i:ni. And there is of course
> patni:, which must stand in the same relationship to patis
> (mutatis mutandis) as genitive asthnas stands to asthi.
I'll address the question of i/n-neuters later (it's tied up with
something else I intend to discuss, time permitting). However, I can
already say I disagree with the last sentence. The word "must" would be
justified if the two cases were parallel, which they aren't. To my mind,
<patni:> is inseparable from <taks.ni:> and <ra:jn^i:>, which clearly
have nothing to do with oblique cases of neuter stems.
Piotr