-s. vs. (-)t-

[Jens Elmegaard Rasmussen]

On Fri, 9 Apr 1999, Miguel Carrasquer Vidal wrote:

> Jens Elmegaard Rasmussen <jer at cphling.dk> wrote:

> >Now, it is also a fact that s-stems have alternants with
> >stem-final /t/: *nem-os/-es- 'worship', *nem-eto-s 'holy'; *lewk-ot/-es-
> >'daylight', and the eternally troublesome *meH1-not-/*meH1-ns- 'month' and
> >the ptc. in *-wot-/-us-.

> I note that at least the last two examples seem to show an
> inverted "reading rule".  We have meH1not(s), Gen. meH1n(e)sos;
> and Skt. nom/acc. n. sg. -vat (*-wot), vs. fem. -us.i: (*-wsiH2)
> etc.  In other words, these look like cases of -t word-finally
> and -s- medially.  What to make of them?

You are right, and I have adjusted the "reading rule" in a separate mail.
It seems that the morphophoneme concerned, which I would call //c// for
practical purposes, surfaces as /s/ before some boundary which is a bit
hard to define, viz. word-finally (Vedic vocative -vas, 2sg *-s, ntr.
-vat then being analogical on masc. *-wo:t-s) and before "weak" case
endings (i.e. all except nom., voc., acc.), and before SOME, but not all,
derivational suffixes: fem. *-us-iH2, but *nem-et-o-s. Glen Gordon's
generalization of -s- to intervocalic position (in a separate mail) does
not explain its occurrence word-finally

> As to *meH1- itself, it is interesting to note the variant *met-
> (Pokorny 2. me:-, me-t- "mow" (*H2meH1-/*H2met-) and 3. me:-,
> me-t- "measure", forms with *met-, like Slav. meto~ referenced
> there).  That looks like *metV ~ *meH1/*meH1C-.

Certainly worth considering. I have toyed with the idea of
"preaspiration", i.e. *H2meH1-ti > *H2methi -> *H2meti -> *H2met-e-ti
involving absorption of the aspiarting laryngeal in to following
consonant, then restoration of the unaspirated ending and, finally,
resegmentation drawing the desinential /t/ to the root and productive
inflection. But it is nice if can be as simple as phonetically regular.

Jens