Dravidian and Australian languages

[dɮ the phoneme]

Ok, so, everyone here has probably noticed at some point that, generally speaking, the basic phonology of Dravidian languages looks a lot like that of the indigenous languages of Australia. I know very little about Dravidian, but judging by Wikipedia's articles on some of the older Dravidian languages and Proto-Dravidian, these resemblances only seem to grow stronger the farther back one goes. This fact has always intrigued me. That said, trying to research the topic farther has mostly turned up baseless speculation and obvious cranks, with the exception of this article from 2013 that seems... at least suggestive of a possible genuine connection. Anyway, I'm curious what others here think of this, or if anyone has any baseless speculation theories of their own.

[Travis B.]

Ok, so, everyone here has probably noticed at some point that, generally speaking, the basic phonology of Dravidian languages looks a lot like that of the indigenous languages of Australia. I know very little about Dravidian, but judging by Wikipedia's articles on some of the older Dravidian languages and Proto-Dravidian, these resemblances only seem to grow stronger the farther back one goes. This fact has always intrigued me. That said, trying to research the topic farther has mostly turned up baseless speculation and obvious cranks, with the exception of this article from 2013 that seems... at least suggestive of a possible genuine connection. Anyway, I'm curious what others here think of this, or if anyone has any baseless speculation theories of their own.

The problem is that the time depth of the original settlement of Australia is sufficiently far back that it is practically impossible for us to find a connection with any outside languages using the methods of legitimate linguistic inquiry.

[dɮ the phoneme]

The problem is that the time depth of the original settlement of Australia is sufficiently far back that it is practically impossible for us to find a connection with any outside languages using the methods of legitimate linguistic inquiry.

The article I linked (a genetic study that, to my completely untrained eyes, seems reputable) suggests that there was migration to Australia by sea from South India around 4000 years ago. This is certainly within the range of linguistic reconstructability, especially given how far back the Dravidian languages are attested.

[Richard W]

From squinting at the actual paper, it seems that the best match to the immigrant group is the Santal people, who speak a Munda language. Disappointingly, Munda languages seem not to have the 3-away contrast dental v. alveolar v. retroflex, but just alveolar v. retroflex. I wonder if the 'European' component of the Dravidian languages is actually 'North Indian', so not a feature subsequent to the Indian influx into Australia.

[Ares Land]

I've always been intrigued by the similarities myself.

A glance through Dravidian and Pama-Nyungan Swadesh lists is fairly disappointing, though.

The WALS Article on 'Absence of Common Consonants' is interesting:

The relatively frequent absence of fricatives from Papuan languages as well (for example in Yimas, Gadsup, Wahgi and Yelî Dnye) establishes a degree of typological similarity between language groupings which are not known to be historically related and have certainly not been in recent contact, but between which an ancient connection is not implausible.

It may be telling that they quote a Great Andamanese has having no fricatives, either.

So maybe we're looking at a very ancient connection?

[Znex]

The problem is that the time depth of the original settlement of Australia is sufficiently far back that it is practically impossible for us to find a connection with any outside languages using the methods of legitimate linguistic inquiry.

The article I linked (a genetic study that, to my completely untrained eyes, seems reputable) suggests that there was migration to Australia by sea from South India around 4000 years ago. This is certainly within the range of linguistic reconstructability, especially given how far back the Dravidian languages are attested.

Plus, while the very early settlement of Australia is unquestionable, the relatively low diversity of the Pama-Nyungan languages for instance suggests a similarly late dispersal. It's quite possible that certain older languages were displaced at some point with a new wave of cultural change, while the original peoples themselves were not.

[bradrn]

The similarities between Papuan and Australian languages, particularly northern Australian languages, are actually quite striking. Aside from fricativeless, the most obvious similarity is grammatical: many Papuan languages have a small closed class of inflecting verbs, a trait shared by many northern Australian languages. (And I really do mean ‘small’: Komnzo is on the large side by having ~380 verbs, Kalam has only ~100, and Jingulu famously has three!) In both areas, this is frequently accompanied by an open set of ‘coverbs’ or ‘uninflecting verbs’. I’m sure there are more similarities, though I can’t think of any just right now.

As for Dravidian/Australian… well, of course I’ve wondered about a connection for a while, but I can’t say I’m convinced of any. If it does exist, then it would probably have to include Great Andamanese at least, if not other families. (On the Name That Language thread I once posted an Andamanese text… it was first confused with Dravidian, then Australian.) But I now suspect that fricativeless is just an areal feature (albeit a fairly sporadic one) of the whole region between India and Australia. Certainly, this is one of the only areas where fricativeless is at all common (the only one being South America).

The problem is that the time depth of the original settlement of Australia is sufficiently far back that it is practically impossible for us to find a connection with any outside languages using the methods of legitimate linguistic inquiry.

The article I linked (a genetic study that, to my completely untrained eyes, seems reputable) suggests that there was migration to Australia by sea from South India around 4000 years ago. This is certainly within the range of linguistic reconstructability, especially given how far back the Dravidian languages are attested.

Plus, while the very early settlement of Australia is unquestionable, the relatively low diversity of the Pama-Nyungan languages for instance suggests a similarly late dispersal. It's quite possible that certain older languages were displaced at some point with a new wave of cultural change, while the original peoples themselves were not.

My personal hypothesis is that Australia originally had a diversity of languages comparable to New Guinea. At some point, Sprachbund developed in northern Australia, with traits involving fricativeless, head-marking, ergativity etc. This area included Proto–Pama-Nyungan, which then expanded over the rest of Australia, replacing and/or assimilating the previous languages. (Of course, this is all just speculation.)

[dɮ the phoneme]

Dravidian has more in common with Australian languages, phonologically, than just fricativelessness. In particular, it's lack of fricatives + a three-way dental/alveolar/retroflex contrast among coronals + nasals at every stop POA + lack of voicing contrast (at least for Proto-Dravidian) + a preponderance of liquids that really make the similarities seem striking.

[WeepingElf]

Of course, this is because Dravidian and Australian languages descend from Lemurian!!!

No, just kidding. I have been thinking about this some time, and I don't know what to make of it. If, say, Dravidian and Pama-Nyungan had a common ancestor about 4,000 years ago, that would surely be reconstructible by means of the comparative method, and would have been established decades ago. One possibility that came to my mind is that Proto-Pama-Nyungan indeed came from India, but descends from a lineage that has since died out on the subcontinent (perhaps clobbered by Indo-Aryan, or by Dravidian languages that were themselves pushed southward by Indo-Aryan) and, while not phyletically related to Dravidian, stood in an areal connection with it and therefore had a very similar consonant inventory (the vowel systems are not particularly similar anyway).

[Moose-tache]

So it seems to me this whole thing is based on phonemic similarities, yes? Most reconstructions of Proto-Dravidian share four key features with most Australian languages:
1) Plosives do not have a voice contrast.
2) Most of them have a corresponding nasal.
3) There is a 3-way distinction between coronal consonants.
4) There are no phonemic fricatives.
The first two are basically nothing, and the last two don't seem like an impossible coincidence to me, especially since we are dealing with a reconstruction that cannot be determined with 100% accuracy. But the main problem is that if these phonologies really were similar due to genetic affiliation, it should be trivially easy to find cognates. Unless the phonemes have all played musical chairs and ended up in different spots, most vocabulary in proto-Dravidian and proto-Pama-Nyungan should be nearly identical. And of course we're ignoring the vowels completely, because that's basically a rule for these games. A look at the grammar gives us a hint, because there we do not see our nice coincidental similarities.

The fact is, there is no reason to think it's not a coincidence.

[Richard W]

Dravidian has more in common with Australian languages, phonologically, than just fricativelessness. In particular, it's lack of fricatives + a three-way dental/alveolar/retroflex contrast among coronals + nasals at every stop POA + lack of voicing contrast (at least for Proto-Dravidian) + a preponderance of liquids that really make the similarities seem striking.

Lack of Proto-Dravidian voicing contrast is dubious. And remember that even actual Tamil has initial voiced stops. Some people believe that initial voiced stops should be reconstructed for Proto-Dravidian, even though they're not very common.

[Travis B.]

Of course, this is because Dravidian and Australian languages descend from Lemurian!!!

No, just kidding. I have been thinking about this some time, and I don't know what to make of it. If, say, Dravidian and Pama-Nyungan had a common ancestor about 4,000 years ago, that would surely be reconstructible by means of the comparative method, and would have been established decades ago.

That is my thought too - if Pama-Nyungan were descended from a relatively recent common ancestor with Dravidian, surely this would be a settled question by now. The fact that it isn't points away from such a relationship.

[Richard W]

Václav Blažek has a notion that the Dravidian numerals 1-4 are connected with Australian numerals. However, I can't think of a plausible story to connect them that doesn't go back tens of thousands of years.

[anteallach]

Of course, this is because Dravidian and Australian languages descend from Lemurian!!!

No, just kidding. I have been thinking about this some time, and I don't know what to make of it. If, say, Dravidian and Pama-Nyungan had a common ancestor about 4,000 years ago, that would surely be reconstructible by means of the comparative method, and would have been established decades ago.

That is my thought too - if Pama-Nyungan were descended from a relatively recent common ancestor with Dravidian, surely this would be a settled question by now. The fact that it isn't points away from such a relationship.

I'd assumed that the idea wasn't that there was an actual genetic relationship between Dravidian and Pama-Nyungan (or indeed any other particular Australian family), rather that Proto-Dravidian phonology might show some evidence of having been part of some precursor to the Australian Sprachbund.

But presumably an Australian-like phoneme inventory is capable of developing independently...

[So Haleza Grise]

The current thinking is that Pama-Nyungan's closest relatives are - the other Australian languages, and there has been tentative work done to create "proto-Australian." Contra earlier views of proto-Australian, which were heavily biased towards PN languages, it seems that Proto-Australian was typologically closer to the so-called "northern" type - that is, dependent-marking, making use of prefixes, with four or five noun classes and really complicated verbs - and PN is the typologically divergent branch.

It's very difficult to say much about the time depth of Pama-Nyungan. There was a recent paper that suggested Proto-Australian was 10,000 years old, but I can't read it. There was lots of news coverage about it though.

As far as phonology goes, I'm not sure how strong the resemblances actually are? Australian languages and Dravidian languages both typically have retroflex stops, fine - but Australian languages often have three coronal series and I'm not sure if Dravidian languages normally do. The retroflexes in some Australian languages might have been innovated from medial [r]+stop. I think there's still debate about how many stop series proto-Pama-Nyungan or proto-Australian had, although I'm admittedly a bit out of date.

I've read suggestions somewhere that the typical Australian consonant inventory is a reflection of certain kinds of congenital deafness in Australian Aboriginal populations, that would make recognising voicing contrasts difficult. I have no idea whether that's true. It's true that some Australian languages do have voicing contrasts, but they are usually the ones that are likely to have been in contact with Papuan languages.

[Atom]

As far as phonology goes, I'm not sure how strong the resemblances actually are? Australian languages and Dravidian languages both typically have retroflex stops, fine - but Australian languages often have three coronal series and I'm not sure if Dravidian languages normally do. The retroflexes in some Australian languages might have been innovated from medial [r]+stop. I think there's still debate about how many stop series proto-Pama-Nyungan or proto-Australian had, although I'm admittedly a bit out of date.

Reconstructions of proto-Dravidian have 3 coronal series: dental, alveolar, and retroflex. Today only a small number of Dravidian languages keep all 3, although all keep the two way dental/retroflex contrast: Toda, Kodava, and Malayalam if I recall correctly. Old Tamil kept the distinction and the Tamil script distinguishes dental and alveolar nasals, but my understanding is that the current spoken language has lost the distinction. I read in "The Dravidian Languages" by Bhadriraju Krishnamurti one theory for the loss: most Dravidian languages have seen massive borrowings from Indo-Aryan languages which have a four way stop contrast (±voice, ±aspiration) but no dental/alveolar contrast so you end up with an unbalanced stop inventory like

Code: Select all

p  t̪  t  ʈ  c  k
b  d̪     ɖ  ɟ  g
pʰ t̪ʰ    ʈʰ cʰ kʰ
bʰ d̪ʰ    ɖʰ ɟʰ gʰ

so /t/ gets removed to preserve symmetry.

[bradrn]

The current thinking is that Pama-Nyungan's closest relatives are - the other Australian languages, and there has been tentative work done to create "proto-Australian." Contra earlier views of proto-Australian, which were heavily biased towards PN languages, it seems that Proto-Australian was typologically closer to the so-called "northern" type - that is, dependent-marking, making use of prefixes, with four or five noun classes and really complicated verbs - and PN is the typologically divergent branch.

I’m not sure that anyone accepts a unified ‘proto-Australian’ these days. (I believe Dixon once did some work to reconstruct it, but these days he doesn’t even accept Pama-Nyungan as a family.) Personally, I’m very sympathetic to some sort of relationship between Australian and Papuan languages on typological grounds (I mentioned some similarities above); I’d say it’s not even a given that Australian is monophyletic.

[Estav]

Dravidian has more in common with Australian languages, phonologically, than just fricativelessness. In particular, it's lack of fricatives + a three-way dental/alveolar/retroflex contrast among coronals + nasals at every stop POA + lack of voicing contrast (at least for Proto-Dravidian) + a preponderance of liquids that really make the similarities seem striking.

Lack of Proto-Dravidian voicing contrast is dubious. And remember that even actual Tamil has initial voiced stops. Some people believe that initial voiced stops should be reconstructed for Proto-Dravidian, even though they're not very common.

Which examples in Tamil are you referring to? The only thing I found that mentioned voiced initial stops in native Tamil vocabulary is Tamil expressives with initial voiced stops, which seems a little dubious to me as evidence for an systemic original voicing contrast.

[Richard W]

Which examples in Tamil are you referring to? The only thing I found that mentioned voiced initial stops in native Tamil vocabulary is Tamil expressives with initial voiced stops, which seems a little dubious to me as evidence for an systemic original voicing contrast.

I believe you'll find that words beginning with ஜ, such as ஜகத் "[d͡ʑʌɡʌt̪]", begin with a voiced stop. The scare quotes are because I think the vowel is a central vowel - the IPA vowel symbol has been misinterpreted as 'as in English butt'.

Voiced initial stops in expressives as a pan-Dravidian feature seems fairly persuasive evidence for voiced stops being present in proto-Dravidian.

Alexis Manaster Ramer produced a note "Not Even a Single Stem" on the topic of proto-Dravidian initial voiced consonants; it's available at www.academia.edu. He favours their existence. Sergei Starostin has some scathing remarks on the blanket denial of such consonants. I will concede that it has not escaped my attention that the first example, Kannada/Tulu/Telugu/Gondi guḍi 'hut, temple' resembles Pali kuṭi 'hut'. (For the Dravidian semantics, I envisage a flimsy wayside shrine.)

[Richard W]

I’m not sure that anyone accepts a unified ‘proto-Australian’ these days. (I believe Dixon once did some work to reconstruct it, but these days he doesn’t even accept Pama-Nyungan as a family.) Personally, I’m very sympathetic to some sort of relationship between Australian and Papuan languages on typological grounds (I mentioned some similarities above); I’d say it’s not even a given that Australian is monophyletic.

Robert Mailhammer certainly does, and there seems to be an opinion about that Dixon flipped in more than one way. The problem is that it's hard work. Given the time-depth, I'm not surprised.