Typology before decipherment?

[ECOLING at aol.com]

Another example of the early stages in deciphering
a yet-undeciphered writing system is Mayan.

Until 1960, it was believed that Mayan writing
was not phonetic, and the only thing that was well
understood was the dates, a "long count" with
five positional values,
400 x 360 days
20 x 360 days
360 days
20 days
day
and many ancillary cycles, 20 x 13 and 365
(which combine to make a 52 x 365 cycle),
lunar cycles, venus cycles (65 x 584 = 104 x 365 etc.).

In 1960, Tatiana Proskouriakoff observed that at the site of
Piedras Negras, there were temples, each with several stelae
on them, and the dates on those stelae spanned approximately
a human lifetime.  Further, there was a particular glyph which
recurred with on of the earliest in each set of dates,
another glyph with a middle date,
and another with one of the latest in each set of dates.
She inferred that these were verbs, specifically
birth, accession, and death, and it turned out she was right.
This was the seminar paper in the decipherment of Mayan
writing as history.

Was typology involved?   One can argue yes and no.
The important point I want to make with the above example
is that one can indeed "decipher" a script partly using only
internal evidence.

In later stages of the decipherment of Mayan,
typology has indeed been used,
in quite a number of small ways,
including noting mostly verb-initial
word order (with date or other introductory element possibly
preceding the verb).
It is not always labeled as "typological" reasoning,
simply because it is taken for granted as legitimate.

That is of course AFTER the language(s) were more or less known,
though Mayan was assumed on iconographic and ethnographic
evidence from before any content except dating was deciphered,
in this case.

But in the last year or so, the understanding of what the languages
were has shifted from a mixture of Yucatec and Chol
to an ancestor very closely related to the existing Ch'orti
and its ancestor Ch'olti (and yes there was a change /l/ to /r/
in some contexts).  Typological reasoning might be considered
involved in a part of that more exact identification of language.

In cases such as the work of Proskouriakoff,
it is indeed possible to identify Nouns and Verbs
based simply on patterning within an undeciphered script.
Luck and persistence are of course involved,
and an attitude that does not attempt to push a pet theory,
but is willing to take the risk of being wrong,
by using what evidence is known,
with plausible enlightened common sense.

Best wishes,
Lloyd Anderson
Ecological Linguistics