More on the single origin of agriculture
- I apologize for the long quotes. There was no way I could cut
further in them.
Laurent Sagart, Roger Blench, Alicia Sanchez-Mazas (eds.)
'The peopling of East Asia'
George van Driem
Tibeto-Burman vs. Indo-Chinese: implications for population
geneticists, archaeologists and prehistorians
Three arguments support the identification of Sìchua:n as the Tibeto-
The first is the centre of gravity argument based on the present and
historically attested geographical distribution of TB language
communities. Sichuan encompasses the area where the upper courses of
the Brahmaputra, Salween, Mekong and Yangtze run parallel to each
other within a corridor just 500 km in breadth.
The second argument is that archaeologists identify the Indian
Eastern Neolithic, associated with the indigenous TB populations of
northeastern India and the Indo-Burmese borderlands, as a Neolithic
cultural complex which originated in Sichuan and spread into Assam
and the surrounding hill tracts of Arunachal Pradesh, the Meghalaya,
Tripura, the Mizoram, Manipur, Nagaland and Chittagong before the
third millennium BC (Dani 1960; Sharma 1967, 1981, 1989; Thapar
1985; Wheeler 1959).
Archaeologists have estimated the Indian Eastern Neolithic to date
from between 10,000 and 5,000 BC (Sharma 1989; Thapar 1985). If
these estimates are taken at face value, it would mean that
northeastern India had shouldered adzes at least three millennia
before they appeared in Southeast Asia.
Whilst some archaeologists may give younger estimates for the Indian
Eastern Neolithic, a solid stratigraphy and calibrated radiocarbon
dates are still unavailable for this major South Asian cultural
The Indian Eastern Neolithic appears intrusively in the northeast of
the Subcontinent and represents a tradition wholly distinct from the
other Neolithic assemblages attested in India.
Assuming that the Indian Eastern Neolithic was borne to the
Subcontinent by ancient Tibeto-Burmans, then if the younger
estimates for this cultural assemblage can be substantiated by solid
dating, the linguistic fracturing of subgroups would have to have
occurred earlier in Sichuan before the migrations, as I had
suggested previously (1998, 2001).
The third argument for a TB homeland in Sichuan is that
archaeologists have argued that southwestern China would be a
potentially promising place to look for the precursors of the
Neolithic civilisations which later took root in the Yellow River
Valley (Chang 1965, 1977, 1986, 1992; Cheng 1957).
The Dàdìwa:n culture in Ga:nsù and Shanxi:, and the contiguous and
contemporaneous Peílígang-Císha:n assemblage along the middle course
of the Yellow River share common patterns of habitation and burial,
and employed common technologies, such as hand-formed tripod pottery
with short firing times, highly worked chipped stone tools and non-
perforated semi-polished stone axes.
The Dàdìwa:n and Péilígang-Císha:n assemblages, despite several
points of divergence, were closely related cultural complexes, and
the people behind these civilisations shared the same preference for
settlements on plains along the river or on high terraces at
Me: Note 'settlements on plains along the river'
Whereas the Sichuan Neolithic represented the continuation of local
Mesolithic cultural traditions, the first Neolithic agriculturalists
of the Dàdìwa:n and Péilígang-Císha:n cultures may tentatively
be identified with innovators who migrated from Sichuan to the
fertile loess plains of the Yellow River basin. The technological
gap between the earlier local microlithic cultures and the highly
advanced Neolithic civilisations which subsequently come into flower
in the Yellow River basin remains striking. Yet a weakness in this
third argument lies in the archaeological state of the art.
Just as it is difficult to argue for a possible precursor in Sichuan
in face of a lack of compelling archaeological evidence, neither can
the inadequate state of the art in Neolithic archaeology in
southwestern China serve as an argument for the absence of such a
precursor. Moreover, agricultural dispersals and linguistic
intrusions may be distinct issues altogether.
The concentration within a contiguous geographical region of all
major high-order TB subgroups other than Tujia: and Sinitic
constitutes a linguistic argument for an early TB linguistic
intrusion into the area that today is northern China.
If the Dàdìwa:n culture in Ga:nsù and Shanxi:, and the contiguous
Péilígang-Císha:n assemblage along the middle course of the Yellow
River are indeed primary Neolithic civilisations, then the eccentric
location of Sinitic and Tujia: may even trace the route of the early
migration out of the TB homeland to the affluent and more
technologically advanced agricultural societies
in the Yellow River basin. In other words, since the linguistic
evidence puts the TB heartland in southwestern China
and northeastern India, an archaeological precursor in Sichuan for
the Dàdìwa:n and Péilígang-Císha:n cultures would fit the hypothesis
that the displacement of Sinitic to northern China was the result of
an early TB archaeological dispersal.
The absence of any such precursor in Sichuan would fit a theory of
early migration from the northern end of the ancient TB dialect
continuum to the affluent areas of pre-TB agricultural civilisations
along the Yellow River. I collectively refer to the ancient TB
populations, who either bore with them from Sichuan to the
loess plateau the technologies of polished stone tools and cord-
marked pottery or were enticed to the loess plateau by the affluence
of the technologically more advanced agricultural civilisations
there, as 'Northern Tibeto-Burmans'. I identify these Northern
Tibeto-Burmans as the likely linguistic ancestors of the Sino-Bodic
groups. Subsequent technological developments were both innovated
and introduced comparatively rapidly in the North, whereas
relatively egalitarian small-scale agricultural societies persisted
in southwestern China until the Bronze Age. This hypothesis places
the split between Northern and Southern TB in the seventh millennium
BC, just before the dawn of the Dàdìwa:n and Péilígang-Císha:n
civilisations. I identify the spread of Bodic groups from Ga:nsù
with the dispersal of the Majia:yáo and Yangsháo Neolithic cultures
and the cultivars broomcorn millet (Panicum miliaceum) and
foxtail millet (Setaria italica), first domesticated on the North
China Plain, into the Himalayan region in the third millennium BC.
Sino-Bodic would have split up into Sinitic and Bodic before this
An alternative proposal to a TB homeland in Sichuan would be to
identify the earliest
Neolithic cultures along the Yellow River basin and on the North
China Plain with the TB homeland.
However, if the TB homeland were to have lain in the Yellow River
then we would be hard pressed to find a plausible archaeological
correlate for the spread of
Brahmaputran language communities, which once extended beyond Assam
and the Meghalaya and
formerly covered much of the area that is now Bangladesh and West
Furthermore, it must be kept in mind that the early Neolithic
civilisation on the Yellow River is
distinct from the cultural assemblages of the middle Yangtze basin,
the succeeding stages of which
ultimately spread as far afield as Oceania in the course of
Both the Yellow River and the middle Yangtze civilisations represent
societies as old as those of the Fertile Crescent.
So it's Tibeto-Butman (in Sichuan) > Sinitic.
Here comes the next step
An intriguing theory involving a remote linguistic relationship with
TB is the Sino-Austronesian theory proposed by Laurent Sagart (1994,
2001 and this volume) connecting TB with AN. Because Sagart
initially recognised possible Sino-Austronesian correspondences
in Chinese material more than in TB, he was originally inclined to
identify the Sino-Austronesian unity with the Lóngsha:n cultural
horizon. However, there is an alternative way of viewing the Sino-
Austronesian evidence and the archaeological record.
The Lóngsha:n coastal interaction ensued upon a northward expansion
of PAN or Austro-Tai culture from its ancient homeland in southern
and southeastern China, and this northward expansion of early
Austronesians would have brought them into contact with early
Northern Tibeto-Burmans. The ensuing contact situations between AN
and the Sino-Bodic branch of TB could have involved the ancient
exchange of vocabulary between the two language families.
The way to test this would be to determine whether items shared by
AN and TB are indeed limited to the Sino-Bodic branch of TB,
including rice terms such as
Malay beras and Tibetan hbras,
a correspondence already pointed out by Hendrik Kern in 1889.
There was that *(H-)bh/p-/r/l- word again!
(in case somewhere someone hadn't heard about them ;-)
The Lóngsha:n interaction sphere is an obvious candidate in terms of
time and place for early contacts between ancient Austronesians and
ancient Tibeto-Burmans, particularly the Dàwènkou Neolithic of
Shandong with its well-established ties both with the other coastal
cultures of the Lóngsha:n interaction sphere as well as with the
ancient Northern TB Yangsháo Neolithic civilisation.
In other words Tibeto-Burman (in Sichuan) > Sino-Bodic (in Northern
China) > (loans?) Austronesian.
I've wondered a long time what the r-suffix of my *(H-)bh/p- and
*(H-)bh/p-r/l- was about. Voilà Sagart to the rescue:
-ar- distributed action; distributed object
This infix was inserted between the root initial and the first vowel
of a stem. Attached to verbs of action it indicated that the action
was distributed in time (occurring over several discrete occasions),
or in space (involving several agcnts/patients/locations);
attached to nouns it indicated a referent distributed in space,
that is having double or multiple structure.
The reflex of this infix in the AN languages is -ar-, marking verbs
of distributed action and nouns of distributed object, including
names of paired or multiple body parts.
Aha. 'having double or multiple structure', 'names of paired or
multiple body parts'
Infixation is often, but not always, in the first of two
Paiwan k-ar-akim 'to search everywhere' (kim 'search')
k-ar-apkap-an 'sole of foot'
Puyuma D-ar-ukap 'palm of hand'
Bunun d-al-apa 'sole of foot'
(PAN *dapa 'palm of hand')
Amis p-ar-okpok 'to gallop'
t-ar-odo' 'fingers, toes'
Tagalog d-al-akdak 'sowing of rice seeds or seedlings
(dakdak 'driving in of sharp end
of stakes into soil')
Malay ketap 'to bite teeth':
k-er-etap 'to bite teeth repeatedly'
Other AN languages show an infix -aR- with similar functions (not
According to the sound correspondences presented above, both -r-
and -R- correspond to OC -r-.
Although no living TB language has -r- infixation as a living
process, paired nouns and verbs with what appears to be an infix -r-
show up here and there, with similar semantics as in Chinese:
Burm. pok 'a drop (of liquid)':
prok 'speckled, spotted'
pwak 'to boil up and break, as
khwe2 'curve, coil' :
khrwe2- 'to surround, attend'
Kachin hpun 'of pimples, to appear on the body' :
hprun 'pimples, on the body;
to appear on the body, of pimples'
Chepang -r- pop, prop 'the lungs'
brok 'be partly white, grey, streaked'
(of hair); compare
TB bok 'white'.
I first identified the Chinese -r- distributed action/object infix
from minimal pairs in Old Chinese (Sagart 1993).
Later on, I described some infixed pairs in modern dialects
where the infix showed up as the regular modern reflex -1-,
preceded either with a schwa or with a full or partial copy of the
syllable's rime (Sagart 1994, 2001).
Here are some examples of infixed nouns and verbs from Yimeng,
a Jin dialect of Inner Mongolia, where the infixed string is -&?1-
p-&?l-ai3 'to swing, oscillate'
p-&?l-&n1 'to run on all sides'
xu-&?l-a4 'to scribble'
t-&?l-&u1 'cluster(s) of fruit hanging from branches'
khu-&?l-u3 'wheel(s) of a car'
But there's my r-suffix! Check
for senses like 'speckled, spotted', 'to boil up and break, as
boiling liquid', 'pimples', 'partly white, grey, streaked', and even
TB bok 'white'.
Now, since all those cognates of *(H-)bh/p-r/l- in IE and
AfroAsiatic that designate a cultivar (eg. Latin far) do _not_ mean
rice, it might be wiser to identify the path of those words as going
directly westwards from the millet-growers of Sichuan, instead of
taking the detour over Austronesian, as I've done earlier.
End of story. Now that was that problem solved.
But isn't it tempting to analyse out *beR- in (eg.) Proto-
Austronesian *beRek "pig" and *beRas "rice", as standing for
something agriculture-related, whatever that is (river?)?
>"hanging from branches'
> p-&?l-ai3 'to swing, oscillate'
> p-&?l-&n1 'to run on all sides'
> xu-&?l-a4 'to scribble'
> t-&?l-&u1 'cluster(s) of fruit
> khu-&?l-u3 'wheel(s) of a car'even
> But there's my r-suffix! Check
> for senses like 'speckled, spotted', 'to boil up and break, as
> boiling liquid', 'pimples', 'partly white, grey, streaked', and
> http://www.angelfire.com/rant/tgpedersen/bHA.html (light),Cf. IE: *pad-/*plad- "flat (area)"?
> TB bok 'white'.
If affixes are imported from that far afield (no pun intended),
internal reconstruction in PIE might be risky business.
Note also the glosses fitting into KVR- "circle, encompass" (where K
is a velar or labiovelar and L is l or r), also occurring in the
words for "wheel": gilgal, *kWekWel. It seems to indicate
the 'Neolithic agricultural cultural assemblage" included some idea
of the circle.