Inscriptions and Dravidian sound changes "y" > "c" and "y" > "t"

[Palaniappa]

Regarding the Dravidian y/c/s alternation, at 01:41 19/01/98 EST, Dr. Bh.
Krishnamurti wrote:

<<Coming to kuyawa~kucawa, an intervocalic consonant is weakened and not
strengthened by a natural phonetic process. -c- [-s-] thus gets weakened to
-y- and not the other way round.>>

But this is not as universal a process as it is made out to be. According to
the conventional weakening hierarchy of sound change as shown by Hans H. Hock
in Principles of Historical Linguistics, 1991, p.83, the voiceless stops and
voiced stops exemplified by "t" and "d" are higher in the weakening hierarchy
than "y". If this hierarchy were to apply strictly and universally, proto-
Dravidian "-y-" cannot have a reflex "-j-" which is a voiced stop. But this is
precisely what the comparative Dravidianists posit in the case of Kui which
Dr. Krishnamurti seems to accept (Pr. Dr. *kay > Kui kaju). Given this, there
seemed to be no reason to believe in the one-way change proposed by
Krishnamurti and Subrahmanyam. A two-way alternation as suggested by G.
Sambasiva Rao seemed to be the correct one.

Indeed, my examination of inscriptional evidence showed irrefutable evidence
for y > c. To prove my case, I have to show that a "y" reconstructable to
Proto-Dravidian should change to "c" in a medial position.

Consider the following case.
matacAn2ai (SII, vol. 5, no. 431 ) < matayAn2ai

The compound "matayAn2ai" is composed of "mata"+"yAn2ai". It means rutting
elephant. "mata" (DEDR 4687) means "to be furious as by must" and "yAn2ai"
(DEDR 5161) means elephant. From the inscription it is obvious that it was
treated as a compound. There is no word break between "mata" and "yAn2ai". In
effect, "y" in "matayAnai" is inter-vocalic.

There is no doubt that the word for elephant "yAn2ai" begins with yA-.
According to P. S. Subrahmanyam's "Dravidian Comparative Phonology" published
in 1983, this initial yA- can be  assumed for Proto-Dravidian (p. 388-390).

This means we have here a clear proof for y > c.

It is strange that the comparative Dravidianists have not used the data in
"Study of The Dialects in Inscriptional Tamil" by A. Velu Pillai, Publication
no. 21 of Dravidian Linguistics Association, published in 1976 six years
before the publication of P. S. Subrahmanyam's book.

Velu Pillai lists several cases where y > c. One of these is the word
cecuviccOm (p. 370).

cecuviccOm  (T.A.S. Vol. IV, p.118/119, Part II) < ceyvittom (we caused to do)

The causative form of Ta. cey (to do) is given as "ceyvi" (to cause to do)  by
P. S. Subrahmanyam in his book "Dravidian Verb Morphology" (p. 7).  Here "cey"
takes on an epenthetic vowel, "u". Along with the palatalization of the first
person plural past tense suffixes, this leads to the following process

ceyvittOm > ceyuvittom > cecuviccOm

In this case also, there can be no doubt that y > c.

Some more words with y > c are:
ucar < uyar (DEDR 646)
vacakkal < vayakkal (DEDR 5258)
vAcal < vAyil (DEDR 5352)
vicalUr < viyalUr = viyal (DEDR 5404) +Ur

P. S. Subrahmanyam notes that intervocalic *-c- can change to -d- (p.331-332).
What is interesting is the change y > c can also proceed further resulting in
y >t. For instance, we find:
utarattu (S.I.I. vol. III, p.191, no. 47, Part II) < uyarattu (of the height).

Regarding the position of L. V. Ramaswami Aiyar regarding y > c changes,
Subrahmanyam noted, " Ramaswami Aiyar (1932b) identified a number of such
cases but he argued that y is original and -s'-, -j- (Tulu) and -s- are
secondary developments. He based his argument on the fact that in Kannada and
Telugu -s- occurs in words with derivatives. He (1932b.21) says that y>s is
"ultimately due to the incorporation of a strong breath current to mark off
the medial syllable of derivatives." He seems to have been too much influenced
by the facts that (i) y is found in greater number of languages than c and
that (ii) literary Tamil y is often replaced by s'/s in the colloquial, e.g.
L. Ta. muyal > colloq. Ta. mosali 'hare'." While I do not have access to
Aiyar's paper, based on the evidence presented here, it rather looks like
Krishnamurti and Subrahmanyam are the ones in error who failed to take into
account very pertinent and relevant Tamil data. If that is done, many proto-
Dravidian reconstructions of *-c- will have to be replaced with *-y-.

Regards

S. Palaniappan