sandhi of Skt. -as [was Re: "syllabicity"]

[Ralf-Stefan Georg]

>>Since the ante-voiced sandhi form of -as is -o Pat is right in finding -e
>>simply incredible.

>The sandhi rule in question is that -as becomes -e before voiced dentals, and
>only there.  It applies in internal as well as external sandhi, cf. _edhi_ <
>*as-dhi (_as-_ "be" + imperative).

>This is not, as it happens, my analysis, but what I was taught nearly 25 years
>ago in my first Sanskrit class...

I'm afraid, I have to stick to what I said before. In external and internal
Sandhi the regular result is -o, and not -e, both before voiced dentals,
and before voiced non-dentals.

To wit take RV III,59,1: Mitro da:dha:ra pRthivi:m uta dya:m (pada-text:
MitraH da:dha:ra pRthivi:m uta dya:m) "Mitra supports earth and heaven".
Or, to take one of the most popular Vedic hymns, RV II, 12, 2 (from the
Indra-hymn): yo dya:m astabhna:t; sa, jana:sa, IndraH (pada-text: yaH dya:m
...) "who supported heaven; he, O men, is Indra". Now this may leave us
still in doubt who it is, who actually does support heaven, but not about
the Sandhi behaviour of -as.

However, the form you quote is correct, and it does lead to the assumption
that *at some early stage* some instances of -as ( > -az) before voiced
dentals went to -e. I understand that this behaviour is still not well
understood, but I may be not really up-to-date. Anyway, it is not a normal,
predictable Sandhi.

By this I don't doubt your analysis of dive-dive; only that it may be due
to predictable Sandhi.  I'm unaware of how many examples for the process
seen in /edhi/ exist, but /divedive/ may be another one, but then for an
early sound-law, involving early univerbation of *divas + *dive not a
productive rule still operative in historical times.

Stefan Georg
Heerstrasse 7
D-53111 Bonn
FRG
+49-228-69-13-32