Theory and fact in comparative Dravidian: the case of -y-

[Sudalaimuthu Palaniappan]

The case of -y- offers some interesting insight into how facts and theory play
out in the field of Comparative Dravidian.  Let us consider DEDR 5352 and DEDR
5354.

DEDR 5352 Ta. vAy mouth, beak of birds, mouth as of cup, bag, ulcer, etc.,
mouthful, lip, edge, rim, edge as of knife, word, speech, hole, orifice. Ma.
vAy, vA  mouth, opening, juncture, edge of a sword. Ko. va.y (mouth of living
thing, cup, bag, pot), mouthful, rim, brink, edge (as of knife); a.ke.r va.y
other bank, other side (of road). To. po.y mouth; po.neRf- to fill to brim.
Ka. bAy(i) mouth, mouth of a vessel, bag drain, etc., head of a drum, edge of
any cutting instrument. Kod. ba.y mouth. Tu. bAyi mouth, edge of a knife,
sword, etc., opening, speech, utterance; ....Te. vAyi mouth, face, edge of any
cutting instrument; vA-konu to speak, utter, say, cry out, vAdara edge of
sword; (K) vAya blade� sharpness.....Pa. (S) vAy mouth of pot. Ga. (P.) vAy
edge of knife; (Oll.) vAsi lip...Kur. baI mouth (of man and animal), aperture
of a vessel, mouthful....Br. bA mouth, aperture, edge of a knife.

What is the reconstruction of the original form? According to an earlier
posting by Dr. Krishnamurti, " PD *c and *y fell together in most of the
languages but the distinciton is preserved by some members (particularly
Kannada). where Ka also has an intervocalic -y- (bayal) it goes to PD *y. So
Kannada holds the clue for PD reconstriction (Ta. poy = Ka. pusi <*posi 'a
lie'; PD *poc-)."

If one were to apply this principle, one would reconstruct *vAy- for DEDR 5352
rather than *vAc-. Indeed in Telugu Verbal Bases, p. 487, Krishnamurti
proposes *vAy-/*vA and not *vAc for mouth.

DEDR 5354 Ta. vAyil, vAyal, vAcal doorway, entrance, gate, place, king�s
court; vAytal doorway, entrance. Ma. vAtil id., gateway. Ko. va.l entrance
door; va.gl entrance; kava.l ground in front of house (for ka-, ? see 1376).
To. po.s entrance;  po.s-a.r (obl. po._s-a._t-) id., doorway (see 405 Ta.
Aru); poxol entrance, in song unit....Ka. bAgil, bAgal, bAkal, (inscr., Gai
221) vAkkil gate, doorway. Tu. bAkilu door, gate; bAdhaLa door. Te. vAkili
door, doorway. Kol. va.kal outside. Nk. vAkal id. Pa. vAl id., menses. Ga.
(P.) vAl outside..

In Telugu Verbal Bases, p. 33, Krishnamurti states:

"Te. -g- < PDr. *-y-; sometimes -y- itself is a weakening of older *-c-; this
development is also shared by Ta. Ma. Ka. and Tulu." What is clearly implied
here is that -g- of Telugu can be ultimately derived from PDr. *-y- in some
cases and PDr. *-c- in some cases. Not all -g-/-k- are traced to *-c-.

To show that -y- passed through the stage of -k- before becoming -g-, in
Telugu Verbal Bases, p. 34, he states:

"..-g- proceeds from an older *-y-; that -y- passed through the stage of -k-
is evident from Ta. vici (*viki) corresponding to Te. Ka. bigi < *viki <
*viyi; another striking proof of this can be found in the following etymology:
Te. vAkili doorway; Tu. bAkil; Ka. bAgil; Ta. vAyil, vAcal; Ma. vAtil (Ma. -t-
< *-c-)"

In other words, both of the following processes are possible:

(1) Te. -g-/-k- < PDr. *-y-
(2) Te. -g-/-k- < *-y- < PDr. *-c-

Krishnamurti did not offer a PDr. reconstruction for Te. vAkil in this
discussion. However, one can state that the reconstructed form should be *vAy-
or that DEDR 5354 is derived from DEDR 5352. There is ample evidence in old
Tamil literature to show that vAy not only referred to the aperture but also
door-like mechanisms like lids, and gates. In fact, Tamil Lexicon includes the
meaning "vAcal" (gate) for "vAy".

The following examples from Tamil texts clearly show the linkage between "vAy"
and "vAyil"..

 matu294   cENOn2 akaznta maTi vAy payampin2            (maturaikkAJci 294)

This can be translated as the "ensnaring pit with its covered opening, which
was dug by the person from the mountain."

AyiramkaNNOn2 arugkalac ceppu
vAy tiRantan2n2a matil aka varaippil    (cilappatikAram  14.68-69)

This can be translated as "in the area inside the rampart which looked as if
the lid of the precious container of the thousand-eyed one (Indra) had
opened".

Even more direct connection between vAy, vAyil, and vAcal are provided by the
following.

palar pukat tiRanta paku vAy vAyil                      (maNimEkalai 7.92)

This is translated as "the entrance (vAyil) with split/open door (vAy) through
which many can enter". One cannot ask for a more explicit link between vAy and
vAyil. In another text, tirumantiram, in trying to portray the human body as a
temple of worship for god, tirumUlar says,

uLLam perum kOyil Un2 uTampu Alayam
vaLLal pirAn2Arkku vAy kOpura vAcal             (tirumantiram 1823.1-2)

This can be translated as "the heart/mind is the sanctum sanctorum, body is
the temple and mouth (vAy) is the temple tower entrance (vAcal)".

That is why, traditional Tamil scholars and T. Burrow have said that vAyil is
derived from vAy. In the book, "Collected Papers on Dravidian Linguistics",
1968, p.79, he said, "Ta. Ma. vAy "mouth; opening of a sack, etc.", O, Ovu
"door (of a sluice)"| Ka. bAy "mouth; the mouth of a vessel, bag, etc."| Tu.
bAyi "mouth; opening"| Tod. bOyi "mouth"--Te. vAyi "id."| Kur. bai "mouth,
aperture of a vessel"| Brah. bA "mouth"|| A derivative of this word is found
in the following; Ta. vAyil "gate, doorway, entrance" | Ma. vAtil| Ka. bAgil|
Tu. bAkil| Te. vAkili "id.""

This means that having derived vAyil from vAy, we would expect the
reconstruction for both to be *vAy-.

But Kamil Zvelebil in Comparative Dravidian Phonology, p. 121, apparently by
not properly considering cases where *-y- can be the proto-Dravidian
reconstruction as discussed by Krishnamurti in Telugu Verbal Bases p.34 and
p.50, reconstructs the following for Ta. vAyil, etc.

Ta. vAyil, vAyal/vAcal, Ma. vAtil (<*vAcil), To. pO_s �entrance�: te.vAkili,
Tu. bAkilu, Ka.bAgil, bAgal �doorway�, Kol. Nk. vAkal �outside�: <
*vAyil/*vAyal<*vAc& -l/*vAc &-al."

He footnotes, "According to a kind of folk-etymology, followed by some Tamil
scholars, Ta. vAyil < *ilvAy (by metathesis) "mouth of the house", i.e.,
entrance, door�."

Actually the Tamil scholars� position is better than Zvelebil�s. At least they
get the base of the word right if one accepts the views of Krishnamurti  (for
*vAy) and Burrow (for the relationship between vAy and vAyil). Even though the
Tamil scholars� attempt to interpret the suffix -il as based on the word "il"
meaning "house" may not conform to modern linguistics, it is not very
different from that of Caldwell and Gundert. Krishnamurti (Telugu Verbal
Basesp. 145) says, "Caldwell thinks with Gundert that derivative -il and -uL
are identical with free forms il house, and uL to be (p.211). This sort of
explanation is outmoded in modern linguistics unless it can be demonstrated
that  all forms with the -uL suffix have some feature of distribution
(phonemic or morphemic) which can be explained only with reference to the free
form uL- to be."  (Tamil scholars base their interpretation on medieval
commentaries of the grammar nan2n2Ul, where a similar interpretation is
offered for the word munRil.) On the other hand Zvelebil totally misses the
basic semantics of "vAyil" in spite of virtually dedicating his scholarly life
to the study of Tamil.

How does all this relate to the -y->-c-/-s- alternation in Dravidian? Since
Tamil has some cases of intervocalic k>c before a front vowel, let us assume
one might argue that the form vAcal may be derived as vAcal<*vAcil<*vAkil<
*vAyil, and that this case really does not prove anything about -y- > -c-/-s-.
But, consider Toda words po.y for mouth and po.s for entrance. Here, s in po.s
is obviously  derived from an original -y-. Similarly, the cognates Ga. (P.)
vAy edge of knife; (Oll.) vAsi lip also show -y- > -s-.

I think Zvelebil gets a glimpse of the right solution when he talks about the
Tamil development of -y- > -c- as seen in medieval inscriptions. He says "The
view of sound change as a special case of total dialect borrowing, operating
through sound substitution, sound replacement (the phone representing -y- in
one or more sub-dialects was more similar, in the other sub-dialect(s), to
-c-, than to the phone representing -y- there), seems to be particularly
applicable in this case. We should certainly like to know much more, though,
about the isoglosses of the -y- > -c- replacement in Ta. dialects." But based
on the discussion of DEDR 5352 and 5354, we see that this type of alternation
is possible in other languages in South and Central Dravidian also. So, we
should have -y-/-c- alternation in PDr which is what G. Sambasiva Rao  says.
In fact, G. Sambasiva Rao�s etymology given in "A Comparative Study of
Dravidian Noun Derivatives" (published in 1991) is as follows.

"Ta. vAy-il, ma. vAtil, Ko. vAl, To. pOs_ �doorway�, pox-ol �entrance�, Ka.
bAkk-il, Tu. bAk-ilu, Te. vAk-il-i, Kol. vAk-al, Pa. vA-l �doorway�
*vAy(kk)-Vl"

In his article, "Proto-Dravidian *c- and Its Developments", M. B. Emeneau
comments on the position of Zvelebil regarding a different problem: "Zvelebil
(1970, p.106, fn. 23) offers an explanation for �the c-/t- alternation� which
is of this same nature (�loss of one of the components�), but finds difficulty
in the statement, since the Prague School of that period insisted that �a
simplex [t�] phoneme� not being �a complex [t+�] cannot lose a component,
rigidity of theory here overriding fact."

I think, the case of -y- discussed here also seems to be a case of theory
overriding fact. P. S. Subrahmanyam, who published his work on Dravidian
Phonology in 1983, seems to have accepted Zvelebil�s views in toto and has not
allowed for the possibility that  *-y- may have reflexes other than the -j- of
Kui.

In a separate posting, I shall show how even Kannada exhibits -y- > -c-/-s-
change.

Regards

S. Palaniappan