[INDOLOGY] Second-syllable rhyme in Dravidian

[Kevin M. Ryan]

A few more points in tentative favor of a southern/Dravidian origin of 
second-syllable rhyme (SSR). I don't think any of these is clinching, 
but they add to the suggestive evidence.

(1) SSR is already present -- pervasive even -- in the oldest attested 
Dravidian literature. I don't think anyone in this thread has made this 
explicit. The discussion of Dravidian cases so far seems to be focused 
on Middle Tamil (and Kannada and Telugu). But SSR is very frequent in 
Sangam texts and in the Tolkāppiyam. It's true, it's not as strict as it 
is in some later texts, but its presence as a systematic poetic device 
(far beyond mere chance or occasional ornamentation) is clearly 
established from that earliest literature.

(2) I don't know if anyone has previously noted this connection, but the 
phonological peculiarities of SSR bear an uncanny resemblance to another 
linguistic phenomenon that can be reconstructed securely (e.g. 
Krishnamurti 2003: 487) to Proto-South Dravidian and perhaps even 
further back, namely, echo reduplication of the type puli-kili [gili] 
"tigers and such". Not only does the span of correspondence in such 
doublets begin with the second syllable (or, more properly, with the 
consonant immediately following the first vowel; see below), but 
tellingly, just as in SSR, the initial vowels must agree in quantity, 
while being free to disagree in quality (e.g. pāmpu-kīmpu "snakes and 
such"; NB. *pāmpu-kimpu is out even though both initial syllables are 
heavy; thus, the restriction is about vowel length per se and not 
syllable weight, at least not in the usual metrical sense.) As such, 
such doublets form rhyming pairs, and I would find it surprising if the 
two systems, both with the same peculiar treatment of vowel length, 
arose independently.

(3) This is perhaps a more conjectural point, but as a general 
phonologist, SSR strikes me as the sort of rhyme system likely to 
originate only in a language (group) with initial accent, which again 
would put it at home in ancient Dravidian. Across the world's languages, 
a strong (though perhaps not exceptionless) generalization about rhyme 
spans is that they tend to be tied to the syllable of greatest 
prominence in the word. In English full-rhyme, for example, the rhyme 
span normally begins with the primary stressed syllable, wherever in the 
word that syllable occurs, e.g. expectátion ~ reificátion ~ vacátion ~ 
státion; glée ~ Tennesée; etc. Same goes for French, Arabic, Old Norse, 
Tagalog, Georgian, etc. Dravidian is usually taken to have (if anything) 
initial accent, nowadays usually realized phonetically (as with many 
non-Dravidian languages in the Sprachbund as well) as an L*+H contour, 
i.e., a dip followed by a rise.

I think this initial accent jibes nicely with line-initial SSR because 
"second-syllable rhyme" is a misnomer; the span is really anchored by 
the first syllable in SSR. First, initial vowels must correspond in 
length (as a function of the rhyme, not the meter); thus, quantitative 
correspondence begins with the first syllable. And then the proper 
generalization about the consonants is not that they must be in the 
second syllable, but rather that they are the first consonants following 
the initial vowels, regardless of their syllabic position. Thus, 
puT.pi.Ra rhymes with muT.ka.ran, but not with, say, kup.pai.

An unrelated tradition that developed this same kind of rhyme phonology 
is Old Norse (in its so-called "half-rhyme"). It works essentially 
identically: the accented (normally initial) syllables are unregulated 
for quality (in half-rhyme) but have a tendency to be of the same 
length. And then the consonant rhyme span begins immediately following 
the vowel, regardless of whether the consonants are onsets, codas, or 
both. Thus Snorri Sturluson: gramr ~ fremri; Sjalfr ~ Elfar; jo,rD ~ 
fyrDum; etc. But this third argument is weak since (a) it's based on 
typological expectations and (b) Middle Indic languages could've 
developed initial accent as well, so it doesn't necessarily exclude a 
Middle Indic origin anyway. But put together with other evidence it's at 
least reassuring that SSR makes good phonological sense for a language 
like ancient Tamil.

Kevin Ryan