# [On logic and fuzziness]

S.Kalyanaraman kalyan99 at NETSCAPE.NET
Sun Nov 8 04:51:48 UTC 1998

```[snip]> Employing probability theory as neutral referee (<<It is math, and its
> branch of probability theory is younger but is available for all those not>
too pompous.>>) may be only warranted by the kind of material one has at>
hand. Certainly, it is available - but not always helpful. (Although - >
ultimately, it might help someone suddenly discover their ability to speak> in
prose...)> > There are times when one is clearly better off by sticking to
good old> s(t)olid procedures.> > Such procedures, however, since they are the
product of pre-post-modernist> modes of thought, do not permit everything to
be connected with everything.>[snip] Artur Karp

I agree with Artur Karp. Fuzziness is, by definition, unavoidable, in dealing
with lexemes, for example. There are so many variables which impact on the
accuracy of the 'meanings' or 'phonetics' presented in the lexicons, a lot of
caution has to be exercised in testing language hypotheses. As a statistical
student, in my graduate studies, I was taught that there is something called
'spurious correlation' which a researcher should watch out for. A good example
is a statement to this effect: geographical areas which have struck petroleum
are likely to be of Islamic faith.

Another example which can be cited relates to the century-old problem of Indus
script decipherment. It is a statistical impossibility to decipher a  set of
about 3000 inscriptions each inscription having an average of 5 signs plus a
pictorial motif, the total number of signs, pictorials and ligatures of the
sign and pictorial stes, being more than 500. Cryptography will fail with such
a set; assuming that an entire sentence can be read from an inscription, and
assuming that each sign or pictorial is a syllable or an alphabet, one would
need at least an average of seven signs per inscription and the total sample
will have to be more than 10,000 for decoding the crypt with a reasonable
level of assurance that the phonemes assigned to each sign are 'probable'. Lot
of words...any statistician will reject this verbal jugglery and some may even
declare it illogical. This has not prevented the appearance of nearly hundred
claims of decipherment; every new claim of decipherment will render all the
decipherment claims equally suspect! Almost any language can be read into such
a limited sample. Fuzzy logic may help by introducing some additional control
variables...

Regards,
Kalyan

____________________________________________________________________
More than just email--Get your FREE Netscape WebMail account today at http://home.netscape.com/netcenter/mail

```