diphthongs

[Michael Witzel]

Since I dropped the hint some time ago, but did not get around to detail,
I add a few words here:

On Tue, 2 Jun 1998, Jacob Baltuch wrote:

> There was some (entirely incidental) talk some
> time ago from (at least two) people who knew
> the phonetic realization of e/ai/o/au in
> (or at some stage of) Vedic. Hence my question:
> what is generally accepted as certain/probable
> in that respect? (E.g. when -- i.e. in relative
> chronology, relative to the texts that is -- did
> the first component of ai/au become short? when
> did e/o become phonetically pure vowels?)


I am not sure there is a way to show all of this, but there are, as
always, incidental obervations which can be expanded to more general
statements, open to proof / or disproof.

 long e < ai

probably the best case scenario.

We know from IA_Mitanni that eka was pronounced aika. We know the same
from innumerable correspodences in Avestan (Av daEuua [daeva] : Ved. deva)
But our recitation (no correct tape ecording here, perhaps) has -e-

Metrics do not help, as ai and e are both counted as long syllables.

(abhinihita  e ', o' needs a long special discussion, involving Panini 8).
(Sandhi rules may also help, later!)

Yet, we have an incidental case:
words such as zreSTha are counted a having 3 syllables, in the Rgveda.
This does not mean that metrics suddenly works here, just for this -e- as
it does not in deva.
Rather, we know that zreSTHa  is from the root zrI, from *IIr zriH,
with laryngeal lengthening short i > I, long i.

Thus, we get:
         *zraiH-iSTha > zrai'iSTha (with hiatus? or glide, cf.
jna-p-ayati)          > zreSTha,
with e in manuscripts & recitation, but still counted as if 3 syllables!


At some moment in the *post*-Rgvedic period thus, after the "RSis" who
still made their verses using zreSTha having 3 syllables, the monophtong e
developed. (nothing special about that, again in Pali, etc...)

As for the exact date,I cannot now think of a way to determine it.
Somewhere in the Pratisakhyas or in Grammar maybe, if the morae of a
syllable are mentioned, `a la Panini :  short-long-pluti (extra long ) u
in his rule about vowel length.

As I said, his rules about Abhinihita or Sandhi can / may help, but that
gets too long. Maybe another message. If here is interest. Note that
interest is rather specific, atomistic and/or selective. No answers on the
VERY interesting Q. about the development of Benares/Kashi so far...

What do the ZiSTa-s think?


> I'm more interested in how you can know such
> things [*] (the methodological aspect of the
> question, to use a word which has recently made
> a comeback) than in what the precise answers
> are.

In this case, obviously, a mixture of Vedic philology, metrical
information and historiacl linguistics, also general Indian (Skt.)
grammar.


PS: the old, Whitneyan, and partly still current, American method to write
all ai, au  with macron seems inappropriate for the post_RV period when ai
> e ...  (and au > o; by that time aai > ai and aau > au, I guess)



 ==========================================================================
Michael Witzel                       witzel at fas.harvard.edu
                                     www.fas.harvard.edu/~witzel/mwpage.htm

phone: 1- 617 - 495 3295 (voice & messages), 496 8570, fax 617 - 496 8571
---------------------------------------------------------------------------
my direct line (also for messages) :  617- 496 2990