Retroflexes (continued)

Vidhyanath Rao vidynath at
Wed Dec 11 15:16:45 UTC 1996

[Some of the references I need to look up have gone missing from the
libraries more accessible to me. These relate only to an alternate
theory of Dravidian influence on Indo-Aryan. So it seems better to
complete my objection to `retroflexes are due to Dravidians mispronouncing
OIA' theory, deferring the formulation of an alternate theory to a
later time.]

How did the Dravidians manage to preserve phonemic value of aspiration,
even though aspiration is not found in proto-Dravidian?

Colin Masica, `Indo-Aryan languages' states that aspiration is not
phonemic in Sinhalese and Maldivian. Given the geographic situation,
these are likely to be most subject to Dravidian influence. On the
other hand, Pali shows little evidence of loss of aspiration.

So how did Dravidians, who had such a hard time with alveolars because
they did not exist in their mother tongue, and ended up changing them into
retroflexes, manage to preserve aspirated stops?


Finally, I would like to address the argument that North Dravidian may
not have preserved alveolar stops at the time of the arrival of
Indo-Aryans because modern North Dravidian languages do not have them.
This seems to enjoy some vogue, so it seems to worthwhile to address it
in a public post.

I have already alluded to the weakness of this argument: North
Dravidian (and most of the Central Dravidian) languages are said
to have merged n/_n and .n into one phoneme, given as `n' in DEDR.
Apparently, the retroflex pronunciation of .n is preserved only
in the cluster .n.t (pronounced as .n.d).

If we wish to conclude that proto North Dravidian did not have alveolars
because they are not present in modern dialects, we must, a fortiori
conclude that proto North Dravidian did not preserve retroflex nasal .n.
This is fatal to the theory that retroflexes in IA are due to
Dravidian substratum. After all, in Sanskrit, retroflexion of nasal n
occurs in more contexts than retroflexion of non-nasals.

Incidentally, the fact that North Dravidian (and much of Central Dravidian)
merges n/_n and .n throws an unexpected light on the statements in
Aitreya Aranyaka 3.2.6 about .n and .s. It may well have been the North
Dravidians who were changing the original .n of rks to n! [Incidentally,
.Rkpraati"saakhya 14.19 mentions mistakes in pronunciation of aspitates.
But this never did catch on. Why?]

There are a few other minor points: North Dravidian languages show
different outcomes of -t-, -_t- and -.t-. So no two of them could
have merged at any point. [The change -_t_t- => -tt- is found in
South Dravidian also: it is common in colloquial Tamil.] There are
Central Dravidian languages which have a common outcome for -.t-
and -_t-, namely Gondi, Pengo, Konda, Manda and Kuwi, with partial merging
in Kui. But the outcome is palatalized in Pengo, and Manda, and
partially palatalized in Kuwi. Kui palatalizes -.t-, and partially
palatalizes -_t-. Konda changes -.t- and -_t- to -_r-, the outcome of
-_t- in Tamil and Malayalam also. Gondi has several varieties of
r as the outcomes of -.t- and -_t-. [By palatal, I mean such as c, j, y
or a further change to s, and not apico-palatals.]

We would need to consider which are more likely to palatalized,
retroflexed or unretroflexed apicals. It seems to me that the latter would
be more likely. In that case, we need to consider the possibility of
loss of intervocalic >retroflex< stops in Dravidian dialects, due to being
merged with with alveolars into a single non-retroflexed apical series.

Given these considerations, simple assertion that North Dravidian had
retroflexes but not alveolars is unconvincing. Such an argument needs
systematic consideration of historical evolution of the phonology of
Dravidian languages, paying attention to the whole phonemic system.
To talk about outcomes of -_t- without also considering what
happened to dentals and retroflexes and the nasals is methodologically

Furthermore, if we want to assert that retroflexes are due to Dravidian
substratum, we cannot use that conclusion to date the changes in
Dravidian phonology. That simply begs the question.

Reports of the loss or nonuse of alveolars are also incoorect:
A minimal pair distinguishing -_n- and -.n- is `pa_ni', dew vs `',
service/work, to bow to (`' in the sense of speak, give orders, does
not seem occur in Modern Tamil). [As Telegu merges, -n/_n- and -.n-, it is
possible that -_n- is missing in the northernmost dialects of Tamil. But it
is certainly present in Madurai area, and I think in Trichi area as well.]

Tamil also uses _n to transcribe intervocalic n in English words.
Thus Germany would be written as `jerma_ni' (`cerma_ni' is the writer
wishes to stick to the `pure Tamil' alphabet) and not as `'.

It is true that Alveolar t and d are transcribed as .t. But this is
because Tamil does not preserve -_t- as a stop. It is a two-flap r
(explaining the transcription R in ITRANS) in the `learned' pronunciation,
and a simple r in colloquial pronunciation. -_t_t- is _tR in learned
pronunciation and tt in colloquial. In nt and nd, the required
assimilation rules then force retroflexes. But, apparently, not
for everyone.

Just out of curiosity, I checked Tamil novels that I found at the Columbus
Public Library and selected one (by its cover :-) that I thought was most
likely to contain trasncriptions of English sentences and words. This was
`varappirasaatam' by Hemaa Aa_nandatiirta_n. Here is a sample (using
Anthony Stone's suggestions for transliteration; in particular, :e/:o are
the *short* vowels):
    yuu ke_n.t sii truu ta .t:elipo_n
      (you can't see through the telephone). [p7]
    paarti mai_nas pip.tii_n iikv:els .tv:e_n.tipaiv
      (forty minus fifteen equals twenty-five). [p9]
     :o_n mi_ni.t
       (one minute) [p11].
Not only are word final and intervocalic n of English transcribed by _n,
even nt is being transcribed by _n.t! (But for some reason, the use
of the digraph _kp is avoided for f).

Interesting side-note: Some varieties of Jaffna Tamil are said to
pronounce -_t- as a stop. Speakers of these dialects apparently
prefer to transcribe t and d with _t. A post in the Usenet group
soc.culture.tamil, that mentioned this gave the example of Tolstoy,
which was written (in ASCII graphics, which I will not reproduce) as
`RaalsRaay' (or perhaps `RaalsRai'?). It sounds funny to my ears,
but then I have never been to Jaffna.


Finally, there seems to be a belief in a `mystical connection' between
retroflexes and South Asia.

But there is no such mystical connection. By now, most Indologists
know that some Swedish dialects have retroflex stops. What may not be so
well known is the fact that retroflexes are also found in some Sicilian
dialects and in some American Indian languages. The later also
distinguish dental and retroflex sounds at the phonemic level.
(`Phonetics', B Malmberg, Dover 1963, pp.41--42.)


To sum up, the objections are

(1) The history of Dravidian phonology is either is ignored or used in
a piecemeal fashion, and without full attention to factual accuracy.

(2) No explanation is given for the preservation of aspirates by
the Dravidians who  are said to have been unable to pronounce the apicals,
even though aspiration too is not found in proto-Dravidian

(3) Appeal is made to some sort of `mystical' connection between
retroflexes and South Asia, even though retroflexes are found in
Europe and the New World.

Nath Rao (nathrao+ at		614-366-9341

More information about the INDOLOGY mailing list