I've been reading up on Dravidian
phonology lately, and I had a question to myself about the retroflex continuant
(Tamil voiced retroflex fricative, Malayalam retroflex approximant).
Though it is commonly translated as l-underscore, implying an alveolar lateral,
it is really a retroflex rhotic and should be represented as one of the
alternatives, r with two dots below. So at least South Dravidian (except
modern Kannada) has three r's.
Hindi and other I-A languages, on the other
hand, have three letters commonly shown as rhotics: the "real r" and the plain
and aspirated (murmured?) retroflex flaps that come from Skt d. and
d.h.
My guess is that Sanskrit /r/ is an
alveolar tap, less likely a trill.
~DaW~
[Piotr wrote:]
I also agree that Old Indo-Aryan /r/ is not likely to
have been fully retroflex in the technical sense; I only propose that it had a
more retracted and more vowel-like articulation in non-prevocalic positions.
By the way, Dravidian languages often have more than one "rhotic" phonemes,
and in many modern Indo-Aryan languages there are two (an alveolar
tap/trill contrasting with a retroflex flap) -- so the peninsula offers more
than the "ordinary" rhotic :)