Posted by Steve Farmer on the Indo-Eurasian Yahoo list (9th September 2009)

****

Permission is given beforehand to repost this message on other Lists.

Brian Hole writes:

> Dilip Chakrabarti weighs in on the side of Rao in the Indus debate
> on indianexpress.com today: [tiny.cc] .....
> I'm not surprised that Chakrabarti holds such views, as he espouses
> them freely in his books - but it is quite disappointing to see
> someone who holds an academic post such as his to be actively
> associating himself with the right wing in the media.

It's really an old story, Brian. Chakrabarti has been Hindutva even
longer than I've worked in the field. I first started work on the so-
called Indus script back during the Rajaram "horse seal" fiasco, in
2000, when I first noticed how odd Indology as a whole is: I don't
know of any other premodern field that has to deal with rightwing stuff
like this on a daily basis. Chakrabarti' s _Colonial Indology_, which is
far to the right, was published three years before that.

I found two lines in his press article especially odd and worthy of
comment, also depressing, starting with the first. Chakrabarti writes:

> Professor K. P.N.Rao and his associates assert, on the basis of
> their recently published computer studies on the Indus script, that
> this script has statistical regularities which are in line with
> other natural languages. Thus, the various signs of the Indus script
> cannot be explained away as only symbols of different sorts.

Two things here: first, the phrasing of the last line is really curious.
He's clearly thinking here of our studies (although he is careful not to
mention those studies by name: that might lead someone to actually
go find them and read them). But it is odd to imply that the symbols
are "explained away" in our papers. We in fact emphasize in all our
works that recognizing what they are opens a powerful new window
on Indus studies. But only half or so of our 2004 paper deals with that
issue, and they never read the paper through so they never get that
key part of our argument.

Second, it is easy to demonstrate that statistical regularity in sign
positions, etc., is true of nearly ALL human symbol systems, linguistic
and nonlinguistic alike. We in fact already discussed that issue at
the beginning of our 2004 paper, anticipating arguments like this long
before Rao et al. "rediscovered" them. (I say "rediscovered, " since the
same arguments were made in the 1960s by the Soviets and Finns
just using different computational tools -- long forgotten.)

We write in our paper 5 years before the claims of Rao et al. (p. 20):

[www.safarmer.com]

> ...statistical regularities in sign positions show up in early all
> symbol systems, not just those that encode speech; moreover, third-
> millennium scripts typically omitted so much phonetic, grammatical,
> and semantic data, and used the signs in so many varied (or
> 'polyvalent' ways, that even when we are certain that a body of
> signs encoded speech, it is impossible to identify the underlying
> language solely from such positional data.

We gave a number of concrete illustrations of this in our paper. After
Rao et al. published their first article on conditional entropy and the
so-called script a few months ago we then illustrated this in a plot
(see the second page of this downloadable document) I showed in Kyoto
and Sproat showed in Singapore (the latter in his keynote address at the
most important annual world conference of computational linguists, none
of whom I've talked to who have anything good to say about Rao's work):

[www.safarmer.com]

End of the "conditional entropy" argument. Now of course that "Markov
processes" have entered the scene, we have to take out time to
show the same thing using them and nonlinguistic symbols. We're
currently
writing our paper on this which we expect to publish later this
month. And that will be the end of the "Markov processes" argument.
That should also be the end of Rao et al., since the group will soon
run out of bullets. You can only "prove" the obvious -- that there is
rough structure of some sort in Indus signs, just as in all other bodies
of signs -- before people start getting bored with your claim.

The strangest thing about this whole new fiasco is that no one has
pointed out that computer demonstrations of regularities in sign
positions in the so-called script have been known since GR Hunter's
dissertation was published in 1929. They were also previously
"discovered"
by Soviet and Finnish "decipherment teams" (also discussed in our 2004
paper at length) using a wide range of computational methods
over 40 years ago!

Amusing too: the announcements made by the Finns and Soviets in the
1960s about this new "discovery" (which Hunter had already shown with
hand-drawn charts in 1929) too created an international sensation and
were accompanied (as in Rao's case) by public anticipation that the
"script" might just be "deciphered" soon thanks to the miracles of
computers.

There is something odd going on here, including a failure of
collective memory. We *again* discuss this issue in our 2004 paper,
again anticipating the Rao claims by over five years. Read our paper:
nothing really needs to be added to it:

[www.safarmer.com]

The other odd thing in Chakrabarti' s paper worth commenting on comes
in the line following:

> the rarity with which many of these signs occur is almost a certain
> indication of the fact that much of the textual corpus of the Indus
> civilisation was written, on the analogy of the Indian tradition
> which continued down to the end of the nineteenth century, on
> perishable materials like palm and birch leaves.

Easily refuted, and discussed again on page after page of our 2004
paper and in our earlier refutation of Rao's "conditional entropy"
paper:

[www.safarmer.com] (see the end).

Making the "rarity" of Indus signs an argument for the use of
perishable writing materials is certainly a novel argument. But, in any
event, the "lost perishable manuscript" argument got a free ride from
1929 to 2004, but can hardly be asserted five years after our paper
was published. The claim is clearly undercut by massive comparative
evidence. In brief, we know of literally *hundreds* of premodern
literate
societies. Not one of these is known that wrote extensively on
perishable
materials and didn't leave long texts behind on durable materials
as well.

The big problem with Rao et al. and guys like Chakrabarti is that you
can't expect to *win* arguments with them, since they just ignore
those arguments. Same for the press, which these days will print just
about anything to get attention. (So far as the Indian press goes,
strike out "just about".) But you can win arguments in the scientific
community, and in the long run that's what counts. It takes a lot of
patience, though, and much wasted time repeating the same obvious
arguments again and again.

Well, it's interesting what a massive reaction our work has generated,
at least. :^)

Stay tuned for papers we're currently writing on all this (although
I'd much rather be finishing up my book: there is *nothing*
interesting to me anymore in the ludicrous "Indus script" story;
but you can hardly quit on the issue either, since oddly
enough these fantasies about ancient Indian history play a big modern
political role in helping maintain the caste system and the
far-right wing in India in general.)

Regards,
Steve

***

Hermione
Director/Moderator - The Hall of Ma'at

Rules and Guidelines

hallofmaatforum@proton.me



Edited 2 time(s). Last edit at 09/10/2009 11:55AM by Hermione.