Origin of retroflexes: answers to Hock's objections?

Vidhyanath Rao vidynath at math.ohio-state.edu
Wed Nov 20 15:48:37 EST 1996

Now seems to be a good time to ask this question, as this seems to be
the time for controversies in INDOLOGY. And now that Dominic himself
has broken the `two screenfuls' rule, I feel free to send this long
post. [And this does have something to do with `Aryan Invasion Theory'.]

I am trying locate any answers to Hock's objections to the often made
claim that retroflex consonants of IA are due to Dravidian speakers
`mispronouncing' Proto-Indo-Iranian words. (The only paper of Hock
I have read in this regard is the one in `Ideology and Status of
Sanskrit'. But I understand that the objection discussed below is older.
At least I was aware of this objection before reading that paper.)

IMHO, the most serious objection Hock raises is that proto-Dravidian
has a >three-way< contrast: dental vs alveolar vs retroflex (for
non-nasal stops). So why would a proto-Dravidian speaker confuse an
alveolar stop with a retroflex one? Many papers I have seen, which argue
that `Dravidians, with dental vs retroflex contrast, mispronounced
the undifferentiated dento-alveolars of PIIr, thus introducing retroflex
stops into IA', never address this objection. I would like pointers to
papers which do so.

Looking at the references usually quoted, I noticed an intriguing fact:
Most of the basic references date from 1972 or earlier, some (such as
the ones by Bloch and Kuiper) being much older. Kamil Zvelebil, in his
`Comparative Dravidian Phonology', published in 1970, goes into great
detail to argue that the `vallina ra' of Tamil etc, are reflexes of
alveolar in proto-Dravidian, suggesting that the realization that
proto-Dravidian had a three way contrast dates from the 70's. I am not
sure how to interpret this.

In the paper of Hock I mentioned, he does not address the issue of the
nasal. So I will describe some details which will bore the
Dravidianologist. I beg your indulgence.

Malayalam is said to have a three way contrast of dental (n) vs
alveolar (_n) vs retroflex (.n). Tamil has this in the writing system;
but there are really only two phonemes: `n' occurs (mostly?) in
word-initial position and before `t'; `_n' and `.n' occur in other
positions. Zvelebil mentions two words from Old Tamil literature
ending in `n'. But he proposes that originally these ended in `nt(u)'.
So I will ignore them. He also says that `n' and `nn' are in free
variation with `_n' and `_n_n'. But the examples he gives are borrowed
from Sanskrit where `n'/`nn' reflect the Sanskritic pronunciation
while `_n'/`_n_n' reflect the Tamil pronunciation. So I am not convinced
that n and _n are in free variation, instead of being dependent on
adjoining phonemes.

As I just hinted, alveolar _n and retroflex .n are separate phonemes
in Tamil. In my experience, words borrowed from Sanskrit are pronounced
with _n/_n_n intervocalically instead of n/nn. But never with .n/.n.n.
[I wonder if Malayalam has n vs _n contrast due different treatments
of n in Sanskrit words and n/_n in Dravidian words.]

Zvelebil constructs two phonemes, one a >dental-alveolar< and the
contrasting retroflex. DEDR lists them as `n' and .n, but does not
get into issues of pronunciation except to note that Tamil, Malayalam
etc have n or _n in place of their `n'. More interesting is the fact
that more northerly Dravidian languages generally collapse the two into
a single phoneme.

So, the evidence suggests that proto-Dravidian had two phonemes,
one retroflex, the other having a dental and an alveolar allophones.
This distinction is fully preserved in South Dravidian, but tends to
get lost in North Dravidian.

How do the proponents of `retroflexes are due to Dravidians mispronouncing
alveolars' explain the above? Dravidians may have converted word-initial
alveolar nasal to dental, but why would they convert alveolar nasal to
retroflex when they belong to distinct phonemes? And why is North
Dravidian going in the opposite direction, erasing the preexisting

There is another, touchier point, that I feel compelled to raise: 
Tamil converts intervocalic dental nasal to alveolar. But NIA speakers
tend to hear alveolars as retroflexes. So Sanskrit `janaka' will become
Tamil `ja_naka' which the North Indian will hear as `ja.naka'. This
would give rise to the stereotype that `Dravidians pronounce dentals
as retroflexes' among North Indians. Did this stereotype play a role
in the genesis of `Dravidians mispronounced dento-alveolars as
retroflexes' theory?

This would be most ironic, as the stereotype is due to the inability
of the >IA< speaker to distinguish alveolar from retroflex, while the
Tamilians must, can and do distinguish them.

I will break off here, and raise my third question at a latter time.

Nath Rao (nathrao+ at osu.edu)		614-366-9341

More information about the INDOLOGY mailing list