Theiling Online    Sitemap    Conlang Mailing List HQ   

THEORY: Underspecification

From:Ed Heil <edheil@...>
Date:Friday, December 10, 1999, 19:06
Underspecification is a phenomenon which is probably universal in
human semiotics.  Letters, even in a very phonetic alphabet, are never
sufficient to encode everything about the pronunciation of a text.
The cutoff for specification varies, but something is always left
unwritten, a blank to be filled in by inference by the speaker or
reader.  In the Latin alphabet, fairly phonetic, vowel length was
unspecified.  In the Greek alphabet, vowel length was unspecified only
for some vowels.  In both of those, punctuation marks and word
divisions were often left unspecified.

But this takes place on higher levels as well.  The meaning of a
sentence uttered in context is more than the sum of the meanings of
the words and the syntax, just as the sound of a word uttered in
context is more than the sum of the sounds of the letters.

The book _The User Illusion_ is largely about this fact -- the fact
that we put into our words, and get out of them, far more than one can
ever say is "in" them.  It's (superficially) like a very effective
compression algorithm -- megs and megs of data may be transmitted as a
 100K JPEG image.  Because there's something intelligent on one end
that compresses it and something intelligent on the other end that
decompresses it.  (Always distrust computer metaphors for human
phenomena, though. :)

That's just one of my favorite cogsci/linguistics topics.  The way
that language is not so much an encoding of meaning as a device
intended to elicit meaning in a suitably prepared brain.  It doesn't
"contain" meaning any more than a rider's spurs "contain" a horse's


Roland Hoensch wrote:

> Three letters granted. That leaves English with 30-35 seperate > sounds and 26 letters. I won't say that's good or bad... but three > letters when about a good 10-12 were needed is a small start > if anything.