Language of the Olmecs

[ECOLING at aol.com]

Peter Selverstone writes today,
concerning the hypotheses put forward by John Justeson and Terrence Kaufman,
that the language of the La Mojarra Stele is Proto-Zoquean,
and my comments on those hypotheses.  Here some rejoinders:

[I think this discussion rightly belongs on the AZTLAN email list,
rather than on the Nahaut-l list,
and am transferring a copy of this there, hoping we will continue it
there.  I responded on Nahuat-l to an assumption expressed here,
and should have attempted to transfer lists earlier.  Those not interested
can of course delete this message.]

>Unlike Lloyd, I am not a linguist, but my understanding is that the
>identification of the language of the Epi-Olmec writing system is
>based on the results of a complete grammatical analysis of the
>texts which conforms to the structure of pre-proto-Zoquean and to
>no other reconstructable language.

For the reading of an undeciphered text, the readings of specific details
can be hypothesized in such a manner as to reflect the conclusion
one started with.  Up to a point, at which we should observe either
an avalanche of things easily falling-into-place, or a failure.

Since the above quotation is Mr. Selverston's conclusion not the evidence,
and since Mr. Selverstone held this same opinion as of a year or so ago
when he told me he had not yet really read
"The Writing System of La Mojarra" (see below),
an evaluation of the conclusion must rest on the facts.
The opinion just quoted does not reflect the facts.

I believe that the review of these issues which HAS appeared,
by Dr. Steven Houston, which I cited in my previous post,
is perhaps too severe.  Dr. Houston concludes that this text
cannot be deciphered because it is too short and there is no bilingual.
I agree that makes it more difficult, but with luck and hard work
(for example, if the reasoning of Kaufman and Campbell about
the language origins of high-culture Mesoamerican vocabulary
does lead to the correct conclusion about the language)
it may be possible.  There is no reason to exclude it a priori.

>One would not expect things to suddenly fall into place for a
>writing system in which only four texts are known, however,
>as reported in the more recent article in Science, an unexpected
>discovery did provide a compelling test of the work.

There has been no "compelling test" of any hypotheses put forward,
neither Proto-Zoque nor Mayan nor any other, in the sense that
no new texts which have been found are more than barely
readable with lots of good will and suspension of disbelief.

This claim of a "test" of the hypothesis
was made once before by these same authors,
in a presentation at Austin, Texas.  That previous claim
referred to the "O'Boyle Mask", a very short and damaged
text, which the authors "read" at that time as concerning bean plants.
It involved a few new glyphs, many read only with difficulty,
and enough degrees of freedom that it could not possibly
constitute a "test".  No such text concerning "bean plants"
is known elsewhere in Mesoamerica, which is not a counter-proof,
but certainly does not strengthen the hypothesis.

The most recent claim of a "compelling test" involves a single
column of glyphs on one of the narrow sides of the stela,
discovered only with great difficulty, almost completely unreadable.
According to Justeson and Kaufman, whom I have no reason
to doubt on this point, what would have been vertical lines in the
drawings of glyphs are at least mostly not readable,
because the grain of the stone goes
vertically there and they disappear into the grain.  Only some lines
crossing that grain are readable, and even those only with great
difficulty, as it is very worn.  I have personally examined that section
of the stone, and I can have only great admiration for the stamina and
abilities of the authors Justeson and Kaufman, who spent many long
hours trying to record that column.  I could see essentially nothing there.

I choose to assume they have correctly drawn what was there based
on the traces remaining, and admire the careful statement they made,
that in doubtful cases they drew a conclusion ON THE ASSUMPTION
that the glyphs present there were glyphs already known from the long
front text of the stela.  That is an entirely reasonable procedure,
minimizing the likelihood of error, but does not remove the difficulty
that the glyphs are almost completely destroyed.

Given the nature of this almost disappeared text,
it could not constitute a "compelling test" of anything.

But it does not constitute a compelling test for a more important reason.
Justeson and Kaufman conclude (correctly, I believe,
as I assume their drawings are accurate) that the phrases
they record from that single column are structurally very like phrases
found on the long front text of the stela.  This does serve to confirm
the analysis of the text into such phrases, since they are now found
on a text which was not known when the phrases were proposed.

But MOST OF THE PHRASES of the text were in fact discovered
by quite a number of investigators years ago,
working quite independently of each other  so far as I am aware.
Crucially, MOST OF THOSE PHRASES DO NOT DEPEND ON
THE LANGUAGE HYPOTHESIZED as the basis of the text.
So what is really confirmed is the work in identifying repeating
phrases.  The hypothesis of a particular language is not confirmed
by any pattern which was predicted by nearly all students of the
patterning of the text, no matter what the language.

There are to be sure a few points where a particular hypothesis
for the phonetic reading of a glyph works to create a certain word
reading (for Justeson and Kaufman, in Proto-Zoquean).
But unless that word reading can be independently confirmed,
as by a highly plausible hypothesis of the meaning conveyed by the text,
preferably a meaning which can be confirmed and locked in place
by some literary or historical or astronomical context, there is
no confirmation of the readings.

Mr. Selverstone writes:

>Unless
>someone demonstrates problems with either the evidence or
>the reasoning or proposes another language consistent with
>the regularities of the very long text of Stela 1, I'm happy to
>accept the conclusion presented in that paper:

The only test available so far is the careful analysis of the hypotheses,
how some of the hypotheses depend on others, how the logic flows,
and whether the results are plausible or not.
That test, or the portion which is possible now, has indeed been
carried out, and is displayed in great detail and in a maximally
accessible manner in the cited publication "The Writing System of
La Mojarra".  Since neither full translations nor even full information
on sentence boundaries has been provided by Justeson and Kaufman,
only a partial test is possible.

Despite those limitations, the book "The Writing System of La Mojarra"
does demonstrate the near-circularities of the reasoning lying behind
many of the specific hypotheses of how to read the text, and thus
beyind the hypothesis of Proto-Zoque language;
that new subsidiary hypotheses are added when needed
to shore up mis-matches between what their
initial starting point led them to (that initial starting point being a few
presumed logogram+phonetic-complement readings), for example
hypothesizing a word order which is NOT that of previously reconstructed
Proto-Zoquean (etc.).  Justeson and Kaufman's
work does not even really consider another language possibility,
and did not consider any other possibility than Mixe-Zoquean from
the very beginning, because they thought they knew it had to be
Mixe-Zoquean.

Those interested should actually read that review of their work
in its full detail, and deal with specifics if they with to draw conclusions
either privately or publicly.

Mr. Selverstone quotes from their paper:

>     This study shows that a previously unknown
>     segment of text can be read and
>     understood in terms of the same model for
>     language structure, sign values, and spelling
>     conventions that were developed in the
>     previously achieved decipherment of the
>     epi-Olmec script, and shows that the segment's
>     content is well integrated with the
>     previously read portion of the same text.
>     Conversely, there are no phenomena in this
>     stretch of text that challenge the model in
>     any way. It is difficult to imagine that this
>     model would yield a complete, coherent,
>     and grammatical text if these portions of
>     the decipherment-language structure, sign
>     values, and spelling conventions-were not
>     essentially correct. In our view, the data
>     confirm the results obtained in the first two
>     of our by now six years of our work on the
>     decipherment of epi-Olmec writing.

This statement has been answered above.

Mr. Selverstone writes:

>When a testable hypothesis is presented in detail in
>peer-reviewed journals, critics should make an effort
>to understand the hypothesis, cite the basis of their
>reservations, and submit their work to the peer review
>process.

This has been done, with the publication of
The Writing System of La Mojarra.
That book itself has been reviewed,
as pointed out in my previous message.
The many details of the alternative sets of hypotheses could certainly use
a more thorough review by someone interested enough to take the time
to do so, but that has not yet happened.

>It appears to me that the only serious review
>and testing of the epi-Olmec work has been done by
>the authors themselves.

Mr. Selverstone formed his conclusions without actually
reading "The Writing System of La Mojarra",
at least not in any detail.

>In my (amateur) opinion,
>this reflects badly on the state of scholarship in
>Mesoamerican writing systems.

I am not so sure.  It may be normal in any field that there
are too few people really interested and willing and able
to spend the time to do this.  Even harder to find one who
is objective and non-partisan.

I do believe that the initial Science article by Justeson and
Kaufman should have been reviewed by some specialist in
methods of decipherment, one who was not even involved
in Pre-Columbian matters, simply as a check on whether the
structure of the evidence and argument could in principle
count as a plausible claim of decipherment.  *Science* should
not have published the article until that had occurred.
I know of one such, who did read the article after it was published
and considered the claim of a decipherment to be unsupportable
based on the lack of the kind of
evidence that could constitute evidence of decipherment.
But since I have commercial dealings with this other person,
they might be biased in favor of my point of view.

So how do we get more non-partisan attention to matters like this?

I wish at the end to reiterate what I have said many times,
that I have the highest respect for the abilities of both John
Justeson and Terrence Kaufman, and believe they have put in long
hours and done much careful work in proposing their hypotheses
of Proto-Zoque language for the text of La Mojarra.

But as a linguist familiar with how decipherments normally work,
even granting that there can be fortuitious discoveries and that every
one is different, I judge that nothing like a decipherment has been
achieved.  We have only a CLAIM of decipherment,
and a very competent attempt to read the text as some language of
a particular language family.
I do not think we have a particularly successful proposal for reading,
however, because the meaning content of the text, as fragments of it
have been presented in various lectures and a few publications,
is rather odd from the point of view of Mesoamerican texts.
So is La Mojarra simply different in what its rulers chose to record?
(After all, La Mojarra is not simply classic Maya or classic Mixtec
or classic Mixe or classic Zapotec or whatever, it is considerably older.)
Or is the hypothesis of language wrong?  Or are some of the specific
hypotheses of phonetic readings of glyphs wrong, and the hypothesis
of language either right or wrong?

I do not presume to know.
What I am sure of is that this text needs more work and more discoveries,
and that the many central and ancillary hypotheses need to be carefully
examined by experts in decipherment methods and by experts in
various language families other than the authors who alone propose
that they know what language it is.  I have not claimed that the language
is Mayan or any other language, I have simply urged that we should actually
do the job of considering a range of possible languages in detail,
rather than assuming we know in advance what language it has to be.
IT IS ENTIRELY LEGITIMATE to proceed as Justeson and Kaufman
did, assuming a language or language family to see how well it works.
But if one starts with an assumption,
then the conclusion can only be an evaluation of how well that assumption
pans out, not of how well that assumption works compared to other assumptions
which were not even tried.

They MAY have made the right choice of assumption,
though my conclusion, seeing the results so far, is that either they did not,
or else that some of their other hypotheses of specific phonetic readings are
not right,
because I find the meaning of the text they propose unconvincing,
and because I think their readings violate a part of what I see as the
repeated phrasal patternings in the text.

I will be glad when this text is deciphered,
INCLUDING if Justeson and Kaufman succeed in deciphering it,
or if anyone else does.

What I did with "The Writing System of La Mojarra"
was to provide some tools in the venerable tradition of decipherments,
listings of all recurrences of patterned phrases, clauses,
sentences, to the degree these may emerge from the text
without hypothesizing a particular language,
adding any conclusions one might tentatively draw from calendrics
or other circumstances tending to constrain possible meanings,
and to provide at least two interpretations based on the assumption of
two different languages, so they may be compared in a systematic fashion.

Best wishes,
Lloyd Anderson
Ecological Linguistics