> The biggest difficulty I see with the proposed etymology
> is 'Rule 14 of Krishnamurti (2003) (Palatalization of
> velars) k c occurs under different environments in
> different Dravidian languages. However such change is rare
> when ‘k’ is before vowel ‘a’ and is more difficult before
> long vowel ‘A’.' To me that says the closest we can get to
> the Sanskrit is *karSapa.
Thank you, Richard, for a careful reading of my post and for
your comments.
I am sure that if we can get to *karSapa we can go a step
further. I would like to state two reasons for that.
(1) Linguists have acknowledged that irregular changes do
occur. This could be an irregular change.
However earlier I have argued that the word ‘zarkara’
(=sugar/brown sugar) is derived from ‘karkara’ via
‘carkara’. This is another example for similar change – ka >
ca.
I can quote a third example for such a change:
(DED 1265) karaTu (Tamil) roughness, unevenness, karu
(Malayalam) rough, gari (Telugu) hardness, stiffness,
sharpness, karcnā (Kurux) to be tough, khara (Sanskrit)
hard, harsh, rough, sharp-edged
(DED 1260) garasu, garusu (Kannada) gravel garusu, (Telugu)
gravel
(DED 1298) kal (kaR-, kaN-) (Tamil) stone,
(DED 2354) caral, caraḷ, caraḷai (Tamil) gravel, laterite
The forms garasu and caral indicates the original initial
consonant as k (k > c and k > g).
This is a conclusive proof that ka > ca occurs in an
irregular manner.
(2) Indo Aryan need not have borrowed sarSapa from Tamil. It
could have been from a lost branch Dravidian in Punjab or
Gangatic plain in which the change ka > ca was regular.
When we speak of substrate language we are looking more to
such a language than to presently existing languages.
Therefore I suggest the following change
*karsapa > carsapa > sarsapa > sarSapa
-Jyothi