Harappan Deciphered?!

[Michael Witzel]

Granted that all below is very preliminary since we only have K.Elst's
brief report and the book announcement by N. Jha and Rajaram, a few points
stick out,  like the proverbial sore thumbs.

Following the major points of Elst's summary:

>The Jha/Rajaram  ... in a chronological framework putting the
>Rg-Veda before 3000 BC  ...  correspond contentswise with
>the late-Vedic/Sutra period, ... read with a close eye on the cultural or
>"literary"
> background...

Let's simply forget this as it has little bearing on the reading of a
script which was used from 3300/2600 BCE- c. 1900 BCE by people(s) speaking
(an) unknown language(s).
All of the above  is interpretation of data that would first have to be
shown to have been deciphered correctly.

>Jha and Rajaram claim that ... seals were not commercial documents
>but were religious in nature, hence ... contentswise they have a lot in
>common with the largely > religious Vedic corpus.

Again, a presuppostion that is not allowed. Some complex seal impressions
(36 times) were made out of a combination of several seals; some such
impressions have cloth marks on the back side. Obviously, to mark [goods]
enveloped in cloth. Second, (large) vessels stamped or written on: is that
religious? Looks more like administration in any comparative 3 millennium
framework.

All of this aside, let's look at the script itself:

>Leaving aside the remaining logographic signs,

-- that would certainly mean some 3/4 of the signs which appear only 1x...

>the phonetic mainstream of
>the script would be most comparable with the Semitic scripts.

The theoretical question is where to draw the line for a (Semitic)
alphabet.  Signs used more than 10 x through the Indus  Civ. at all times
come to c. 100, those used more than 20 x come to c. 80 signs. Nice, for
an Indian style alphabet?
Nope.
Even the ten-fifteen most common signs  include some that have clear
"diacritics". Let's use Chinese style nick names for them. From the most
common one down:

855x found : rimmed jar
322 : small Roman ' ' (there also are full length signs such  as ||, see
below )
194 :  simple vessel  V
187 : simple fish sign with two downward fins
179 : fish with two horns on head
159 : arrow sign, with bottom of arrow head closed
149 : 5-pronged comb, prongs turned right
140 : Roman | | |
127 : fish with circumflex roof ^  on head
127 :  rounded off rhombus with v inserted in upper part
125 : six-spoked 'wheel'  (more rhombus-like!)
125:  `/  type, closer to a Y
117:  fish with  - stroke through body
116:  Trishula   (Y with  inserted ' )
115:  Roman | |

Note that no.s 187x, 179x, 127x, 117x are "diacritical" versions of each
other. (More cases among the signs with frequencies a little lower than
these).

We now apply our knowledge of Skt., Vedic or other, and  if the IV language
were any form of Sanskrit, as with Rajaram/Jha, these signs would be,
based on rounded off Skt. sound frequencies, roughly:

855 would be:  a   (20%)
322  = aa               =8%
194  = t                =7%
187 =  r                =5%
179  = v                =5%
159  = i                =4.8%
149  = n                =4.8%
140 =  m                =4.3%
127 =  y                =4.2%
127 =  s                =3.5%
125  = d                =2.8%
125  = u                =2.6%
117  = p                =2.4%
116  = k        =2%
115 =  bh       =1.3%

(This is just frequency order. The Skt. %s obviously do not make for a
close fit with the actual %, just with the frequency  of the  Indus signs.
First hurdle.  -- Anyhow, now,  start reading!!!)

Result :  187 = r,  179 v ,  127  y ,  117  k  would be  phonetically
related????
(or try any other permutation r, R, RR, L, LL, ...???)  Not possible at all.

Or read the Dholavira inscription....
http://sarasvati.simplenet.com/html/Corpusmain.htm
(scroll down just once) which has:

crab - spoked wheel (d)  -  spoked wheel  (d) -  '  -  ^  - empty rhombus -
spoked wheel (d) - pipal leaf with projections -  square on pole - spoked
wheel (d).

These are *not* the most common characters, -- for the reading of which one
would have to marshal the rest of the Skt. alphabet not covered by the
above 15 signs... (e.g.,  all e, o, R vowels. None of the most common, Skt.
a, aa). Sounds good?  Nope.

Or, you can read, with a slight reversal of above list, if you like it
better:
a Paninean succession of  u  | uu  | uuu.
In other words, early grammar. That is certainly what I would put on my
city wall! Much better than the recently broadcast version (I forget by
whom, information overload, no one can keep up any more with all these
decipherments!), which ran roughly:

 "I have driven off all horse thieves!"

As for non-alphabetic signs, such as the svastika character: it is
omnipresent in many parts of the Old World. I like the interpretation given
to it in Ghana best:

"monkeys feet"!

So much for  reading any unknown symbols, characters and signs. BB Lal
(1997: 214) tells us:

 "Let us continue to be optimistic."  --  Rather, I say:  lasciate ogne
speranza!

{{ I overlooked until just now that S. Kak, Cryptologica XII, 3 , 1988,
129-143, has taken  a statistical approach, similar to mine above,  but
then taking a completely different turn, identifying some of the signs with
those of the Brahmi script, while on the other hand, taking some of the
most common signs as case endings (arrow, rimmed vessel etc.). One would
guess that the wide margin allowed here would result in any desired
interpretation, and indeed:  no surprise! These are  Skt. inscriptions...
Of course, the table just given above has no resemblence to Kak's. Note the
interview of Kak at: http://www.rediff.com/news/1999/nov/18inter.htm }}

The procedure shows, as always, that changing one assumption will lead to a
completely different "decipherment".  And then, as always, comes the leap
of faith plus  secondary assumptions to make things work out... In other
words: No inscription has been read yet.
(Of course, I have my own "decipherment" :  not Norse or English, but
--you guessed it-- Munda.  Such as, on one tablet: "to be given to the
chief queen!"  Do you think I will publish it? Naaah!

Continuing with Elst's summary:

>There is an initial vowel sign, like the Aleph, which can represent any
>vowel: this is
>the omnipresent jar sign.

Not very practical in any alphabet, unless the initial vowel represents one
preceded by glottal stop (found in certain languages before any vowel but
not in Skt.!), --  which then, by necessity, has to be followed by a
separate vowel sign. Otherwise, even in English (which has such stops), a
phonetically written  sentence such as:
        'think'lovthi'bavcaen'nlibiirigardid'zmakingnosens
is not immediately decipherable. Especially, when words are not separated.
If separated,  this would read:
        ' think 'l ov thi 'bav caen 'nli  bii rigardid 'z making no sens.
Got it this time?

> Vowels following consonants are generally not
>written, so one consonant signs can be read as ka, kaa, ke, ki etc.

Wonderful, so the sentence used above would be:
        'thnk'lvth'bvcn'nlbrgrdd'zmkngnsns.

The Lord of Dholavira or Harappa would have made the inventor of this
alphabet work in the
copper mines of the Aravallis for the rest of his rather short natural
life. Or, if a Brahmin, have him sent beyond Kaala PaaNi, to Punt.

Of course, Rajaram's vowel rule is, again, guesswork. Based on much later
Indian scripts but worse than, say, modern Hindi  (unspoken a, without
halanta marking), or the ambiguities of Urdu script, for that matter.
Rajaram's system allows insertion of any vowel, to fit the "decipherment".

>Thus, the crab is [ma],

this occurs only 33 x (crab turned right)  in all of the Indus, too low for
Skt.  ma = c. 4%.

> the 5-stroke standing man is [ra] or [R]

only 47 x , too low for r   (= 5%) (vowel R has 0.74%)

>the standing  fish is [sha] (also "100", sha-tam),

but see above 187x;  r should be much lower than this freqency, at 1.5 %,
and  not at position 187x.

> the same standing fish with a little roof overhead is [shri].

Now, that's cute!
But what do do with all the other common fishes, 187 = r,  179 v ,  127  y
,  117  k ?  See above.  (Is this  = zrImatI, zrImant:  Mr., Mrs., Ms.,
young master,...? zrI zrI for kings, zrI zrI zrI  for gods, as in medieval
mss???)

>An obvious weak point is that several sounds have
>more than one sign representing them, though care was taken to avoid cases
>where two such "allographs" appear on the same seal.

Now that's already the *third* major ambiguity, added to those  pointed out
above.
And the one which finally destroys the "system" altogether.


IN SUM:

Since we have (at least) some 80 or 100 frequent signs at our disposal, a
really bad writing system! Brahmi/Gupta/Nagari does it with c. 50 signs
plus the c. one dozen of 'added vowel' signs (always depending on the
particular variety of alphabet used at what time, therefore: "circa").

>Jha and Rajaram also claim that in many cases, the picture and the text on a
>seal correspond directly or obliquely, somewhat like in heraldry where the
>motto and the picture sometimes match

Oh well, but there are equally many cases where the *same* written message
occurs with another picture and vice versa !!

And, typically, all of the above leaves out the likely possibility that
other language(s) than Skt. are represented by the IVS ...  (see my earlier
message on Rgveda loan words).

A few notes on the press release separately.

Case closed.

Next case:    Munda tribal writings:)
http://www.engr.mun.ca/~asharan/bihar/indus/indus~3.htm

Enjoy!
==================================
========================================================
Michael Witzel
Department of Sanskrit & Indian Studies, Harvard University
2 Divinity Avenue, Cambridge MA 02138, USA

ph. 1- 617-496 2990 (also messages)
home page:  http://www.fas.harvard.edu/~witzel/mwpage.htm

Elect. Journ. of Vedic Studies:  http://www1.shore.net/~india/ejvs
-------------- next part --------------
A non-text attachment was scrubbed...
Name: attachment.bin
Type: text/enriched
Size: 10337 bytes
Desc: not available
URL: <https://list.indology.info/pipermail/indology/attachments/20000708/51dd1470/attachment.bin>