Possible sound sequences in Sanskrit

[Vidhyanath Rao]

The short answer is that h and m can, in theory be followed by any
resonant. But mm and mv seem not to occur as they (always? usually?)
become nm and nv. But all other combinations of m/h followed by
n, m, y, r, l, v occur, as examples already given or the dhaatuko"sa
show. [h followed by a stop, or m followed by a non-labial stop
would presumably be ruled out by IE constraints and assimilation
rules of Sanskrit. But are there any onomotopoeic exceptions?]

And s can be followed by k or p even at pada boundary: namaskaara
or paraspara (to give very common examples). In mantras, of course,
this can even occur at ``word'' boundaries (that is between
independent words, not compounded together). This reminds of
a question I have been meaning to ask sometime: Is there any
evidence of `s-preserving' Prakrits? By `s-preserving', I mean
Prakrits in which word final s/.s remains a sibilant in
conditions where Sanskrit changes them to r, <nil> or changes
as to o.

Determining whether h is visarga, aspiration, or voiced breathing
is mechanical if you use the modern standardized spelling conventions.
But manuscripts can have tad{}hi, and if you aim to transcribe
them faithfully, you will have even more fun :-) [Incidentally,
tad{}hi is most emphatically not a mistake. Panini and Rkpraati"saakhya
both allow such sequences, and possibly other praati"sakhyas as well.
I don't understand the modern penchant for `correcting' such ``mistakes''.]