Re: [tied] Re: Nominative: A hybrid view

From: Jens Elmegaard Rasmussen
Message: 22483
Date: 2003-06-02

You are confirming the very point I made: Sanskrit is monovocalic in
exactly the same sense as PIE is sometimes said to be. Also PIE has its
intricate set of problems with clustering semivowels and sonants, and
Sievers' Law does not always run as smoothly as desired. Still, it is not
entirely without foundation in fact if one associates [i] with [y], [u]
with [w], etc.

In most cases I do actually think there is one and only one realization of
a given sequence of semivowels in Sanskrit: dyut- never *divt-, etc. So we
have [i:] when there is no adjacent vowel, while /yy/ is [iy] in /tyya/
and actually hardly ever [yi] (still /rayym/ is [rayim], what else could
it be?).

On Mon, 2 Jun 2003, nathrao wrote:

> --- In cybalist@yahoogroups.com, Jens Elmegaard Rasmussen <jer@...>
> wrote:
> > Tee-hee, as I have said a number of times already, Sanskrit *is*
> > monovocalic in exactly the sense of the most extremist view of PIE.
>
> I am trying to pick this thread up in the middle, so please bear
> with me.
>
> What are the rules for semi-vowel clusters? In particular, how do
> we tell i: apart from yy or yi or iy? And, incidentally which
> stage of Sanskrit? RV? late Brahmanan prose/Panini, Classical
> (as found in Kavyas etc)? It makes a difference (for how Siever's
> law operates, which is surely relevant to this question).
> Desideratives can produce sequences like yiy: yiyapsati,
> wants to f*ck, actually occurs I believe.

Well, if that is decent Sanskrit, but who knows? And what did PIE have
here, if the root was *yebh-, and the desiderative was reduplicated with
-i-? Did it realize [yiy-], or did it change it into something else? These
cases are of marginal status, and they concern PIE and Sanskrit alike. The
two are really working the same here. I also believe we have PIE *n&1tr-
in the word for 'adder', as opposed to *n.-h1dti- in Gk. ne:~stis
'non-eating, fast', the first based on levelling in a paradigm that used
to contain forms with *neH1t- also. Based on such instances I would go so
far as to set up, say, /&1/ ("Schwa one") and /h1/ as different phonemes.
To be fully consistent I should then also posit /i/ and /y/, etc. as
different phonemes of Sanskrit (even /r/ and /r./ may be opposed to each
other), but the functional load these oppositions carry is close to zero.
In fact they are more inconsistencies in the use of the system than real
oppositions designed to convey a meaningful contrast.

To make it really work in Sanskrit, one needs a "silent consonant" to make
odd vocalizations predictable. In some Sanskrit forms, laryngeals are
alive right below the surface.

> Incidentally, note that actual pronunciations (and Vedic manuscripts)
> may have yy where printed editions (following >one< branch of
> tradition) have y. We must be careful not to reify Whitney's
> Grammar into an actual langauge.

That weakens your objections to the analysis just as much.

Jens