> A minor correction: you mean the putative substrate in Bangani, not
Bangani itself (which is an otherwise inconspicuous Modern Indo-Aryan
language).
Yes, sorry. <smacks head> If you all keep reminding me of this, some
day I'll type it correctly.
> That's already two (even if you intend to play down the
significance of Iranian in the scenario ;)). <
Um, how is it two? Not that the number really matters a lot as long
as its a small number! ;) ProtoPIE leaves India and goes to Central
Asia (one) where it behaves according to the standard theories about
PIE (depending on whose theory you happen to prefer.) OK, later I'll
admit to two or three -- but only one major one.
>However, you can't derive all the non-Indo-Iranian languages from a
single "sister" of Indo-Iranian in the IE family tree. <
Why do you want to make everything a "sister"? :-) It would be a
parent or grandparent. And why not? I assume all the reasons are
being covered in this thread but please mention any that are missed.
>The primary divide (and the earliest split) within IE is quite
clearly between Anatolian and "the Rest". <
This is disputed. Some will say Anatolian is explained by mixing with
other languages. Obviously any Indian homeland will have to go with
the anti-Indo-Hittite school of thinking, cf. Beekes "its simplicity
need not suggest antiquity, but could rather be owing to loss."
>Apart from the topology of the family tree, there are additional
problems if you assume an early separation of Indo-Iranian.<
If I understand his theory correctly, Indo-Iranian does not
necessarily separate early, per se. But the PIE that shares some
features with them separates later; IIr I think is a separate,
although closely related dialect. Does the discussion below cover all
of the additional problems?
> Typological and lexical isoglosses demonstrate that Proto-Indo-
Iranian must have formed an areal grouping with the laguages
ancestral to Greek, Armenian, Phrygian, and probably Albanian as
well, to the exclusion of the other branches (with the possible
exception of Thracian, about which little can be said).<
After PIE left India and settled in Central Asia, the first branch of
it would have split off to go south to evolve into the languages
above. I think the way its proposed is: Aryo-Greco-Armenian went
south of the Caspian into Anatolia and beyond. Balto-Slavo-Germanic
and Italo-Celtic went north of the Caspian and Black Sea into the
Balkans. Greek and Italo-Celtic may have met in the Balkans although
this is beyond his concern right now.
>It's easier to imagine such a grouping with Indo-Iranian located in
the Pontic steppes and then expanding southwards (while the rest of
the group remains in the vicinity of the Black Sea) than a concerted
migration of a cluster of distinct dialects miraculously preserving
their geographical configuration.<
This is an important concept to understand about the theory: there is
NO "cluster of distinct dialects miraculously preserving their
geographical configuration." There is only ONE migration of PIE and
it is very similar to the existing PIE except that it is
differentiated from Indo-Iranian. This ONE dialect then proceeds to
diverge from the BMAC or someplace farther west according to the
accepted theories. Well, OK only ONE major migration. Iranian is a
second but since its "within the family" its not emphasized. After
all, if you can locate even IIr in India, then getting Iranian out
isn't going to bother anyone after they've accepted the big one.
>Such a group migration would have had to be independent of the
migrations of the Balto-Slavic speakers and of the "Western" branches
(the major ones being Italic, Celtic and Germanic) -- which further
undermines the "one migration" scenario. I really can't see how one
could plausibly derive the IE languages from a homeland in India
without proposing _at least_ five or six separate movements.<
I hope I explained this above. If not, let me know where the problem
is. The movements happen after PrePIE leaves India and PrePIE is NOT
Vedic Sanskrit but an isogloss probably with centum forms. Why are
five or six movements required in this scenario?
> > This PIE would then have interacted with Uralic/Finno-Ugric since
it is defensible that the loans are only one-way into those languages
families. After it left India, you can take your pick about where it
might have settled to play the role of PIE.
>
> It isn't so simple. The loans in Finno-Ugric are from IE, but not
from PIE or any dialect "playing the role" of PIE; in particular, I
know of no evidence of contacts between Proto-Finno-Ugric and any
centum branch (which is a serious problem for your scenario).
1) "split" is based on idea of divergence from PIE in C Asia -- this
is a dialect that has the features of PIIr but didn't really split
from IA into two but rather split from Indo or prePIE some time back
and retained the IIr features. So instead of going: PIE > IIr >
IndoAryan and Iranian it goes prePIE > IIr and IndoAryan > Iranian
2) OK so maybe there were two (or three -- see below) waves.
Actually, you just clarified something that I didn't understand: he
said something like: "now I just have to figure out if IIr went W or
NW". I thought he was talking about Scythians but it turns out he
feels that the IIr left South Asia via one of two routes: E through
the Helmand into Central Afghanistan and onto the Iranian Plateau or
NW through the Khyber into the BMAC and then into Iran. Some of them
would also migrate further NW to interact with the Uralic family.
And they would separate Tocharian from its centum roots. This is all
quite a bit later than the PIE migration. At this time he prefers the
Helmand route for the Iranians. OK, I guess there is also a third
migration -- but again one within the family like the Iranians -- at
some point the Indo-Aryans who were probably also the BMAC (Sergent
and Parpola agree that BMAC is Indo-Aryan) invaded to create the
Kassite and Mitanni regimes -- again this is late compared to PIE.
>The loans (very likely all of them) are distinctly Indo-Iranian and
many of them can't be dated to the post-split period, since the most
archaic of them show features like unpalatalised velars before front
*e (still retained!), and satem stops still reflected as affricates.<
This would be the character of this so-called IIr but is really the
Iranian split.
> But the uniquely Indian names of plants and animals are non-IE in
terms of morpheme structure and phonotactics -- mind you, for
_structural_ reasons, not merely because they can't be found in the
other branches. They cannot be analysed or etymologised as normally
formed IE words, while they can often be traced back to various South
Asian sources. Therefore, they can't be PIE items lost by the other
branches; they are foreign words borrowed by the single branch that
reached India.
There is very little consensus on which of these words are IE and
which are not. Kuiper will say they are not IE, Mayrhofer will say
they are, Masica will question both of them. The few that are clearly
not IE can be explained by adstrate and especially through trade.
> Just for the record, I personally do not place PIE in Central Asia.>
Where do you place it?
>As for homogeneity or heterogeneity indicating the homeland, it
would be naive to formulate any hard and fast rules, but homogeneity
tends to indicate relatively recent expansion.<
Nichols disagrees.
> A homeland area will often show traces of _very old_ linguistic
boundaries rather than just a larger number of dialects. In the US,
for example, the East Coast and the old South show much more regional
differentiation (and historically older dialect divisions) than the
West.
It would be hard to find a culture that is more antithetical to
IndoIranian linguistic conservatism than the US. Innovation is a
matter of pride there and gives rise to lots of heterogeneity.
> > It is very clear that there were many Sanksritic and other IA
dialects in India that have been lost. Witzel has written several
papers on this. Also it seems that the modern IA languages aren't
directly connected to recorded Sanskrit -- there was another dialect
that gave rise to them. So we already have evidence that at some time
in the past the Indian languages became homogenized for some reason
perhaps due to Brahminical influence during the freezing of the
language in the Vedic texts.
>
> What we see is the normal level of dialectal variability in _Old
Indo-Aryan_.<
Yes but we also see evidence of many forms of Vedic that have been
completely lost.
> Vernacular varieties of Indo-Aryan have never become homogenised
or "frozen". Quite the opposite: they have differentiated into scores
of Modern Indo-Aryan languages and local dialects.<
I agree. What is proposed is that this differentiation was also very
common in Vedic times as well. But for some reason, Sanskrit became
frozen by Panini and a *different branch* of IA went on to form the
modern dialects.
> My original point was that while non-IE substrates are copiously
attested in India, we don't see any traces of old IE languages that
didn't belong to the Indo-Aryan group (as represented by Vedic and
languages or dialects different from Vedic but closely related to
it).<
The pre-PIE in question would have been in the NW -- those that
didn't leave were subsumed by other languages -- first IA during the
Paninian homogenization and the shift of the spiritual center of
gravity away from the Panjab and into the Ganga-Yamuna Doab then
later by the hordes of outside invaders.
>The source of the Bangani substrate is a possible exception, but
since it's peripheral and isolated, it's more likely intrusive than
autochthonous.<
There may be others to be found. But it seems that the status
associated with the "new" Vedic rituals and languages associated with
the shift to the Doab may have replaced many of them (cf. Witzel.)
> The Tocharian languages were _not_ spoken in the Indian region;
they were separated from it by the Himalayas, the Tibetan Plateau,
the Kunlun Shan and the Taklamakan Desert.<
I'm not suggesting that it was spoken in India but it and Bangani are
Asian centum (substrate!) languages. It had to come from somewhere --
why not the NW part of South Asia? That's at least as feasible as
some scholars theories of massive treks across Central Asia from
Europe, if not more feasible. And they may be *separated* by the
above mountains, etc., but more importantly they are *linked* by
mountain valleys that have been highways used for millenia for trade
and invasion. The Tocharians are not so far away.
>There was no Indian influence in those parts before the arrival of
Buddhist missionaries about the beginning of the common era.<
There is no *attested* evidence. Also it depends on who you consider
to be the source of cultures like BMAC. It's not really clear who the
Tocharians were before. If you have some references with more
information on this, I'd love to see it. But I haven't seen any.
>You demonise what you call "AMT thinking" by suggesting that it
makes scholars adopt a blinkered attitude.<
I think demonize is a bit strong. On the other hand, I can quote
several mainstream archaeologists who would be quite happy to
demonize the AMT. Shaffer refers to "linguistic tyranny";
Jarrige: "the processes are too complex to be attributed to the
arrival of invaders"; Kenoyer: "These changes were made by the
indigenous inhabitants, and were not the result of new people
streaming into the region."
>J. Koivulehto's conference paper "Varhaiset
indoeurooppalaiskontaktit: aika ja paikka lainasanojen valossa" (The
early IE contacts: time and place in the light of loanwords. Lammi,
1997), which I know only from an English-language summary.<
Do you have a reference for this english summary?
> Koivulehto, Jorma. 2000. "Finno-Ugric reflexes of North-West Indo-
European and early stages of Indo-Iranian". In: _Proceedings of the
11th Annual UCLA IE Conference_, pp. 21-43.<
Who is the editor for this? These journal articles are nearly
impossible to find -- even after you track down the rare library that
has them on the shelves.