Sandhis of word-final N

[Madhav Deshpande]

Thanks to Georg von Simson and Hans Hock for referring to the
onomatopoetic usages like bhaN iti found in late Vedic material.  For
Panini, such words are called imitations of inarticulate sounds
(avyaktaanukara.na), and are considered to be grammatical Padas which can
be subjected to regular sandhi rules.  In known usages, their sandhi
treatment is seen to be irregular, so jha.titi, and not jha.diti, but
consider usages like: pha.t pha.t phaa.d iti where the expression pha.t is
subjected to voicing before a vowel.  On the expression dvir va.sa.t
karoti in (Aitareya Br), Saaya.na says:  vau.sa.d vau.sa.d iti.  Here the
voicing of the final .t to .d occurs, indicating that these onomatopoetic
expressions are subject to the same sandhi rules as regular Padas.  I
suspect something like this was assumed by Panini for expressions like
bha.n iti.  Many more such onomatopoetic expressions ending in .n are
found in modern Marathi:  .ta.n .ta.n, .tu.n .tu.n, bhu.n bhu.n, ku.n
ku.n, ru.n jhu.n, gha.n gha.n, ta.n ta.n, da.n da.n, .tha.n .tha.n, pha.n
pha.n, va.n va.n, etc.  I suspect something like these did occur in
Panini's Sanskrit.
        What seems interesting is that the rule of doubling of final .n
after a short vowel and before another vowel does not extend to palatal ~n
and m.  Why does it apply only to the other three nasal consonants?  Any
suggestions?
                                        Madhav Deshpande