arithmography (fuit: Re: The philosophical language fallacy)
|From:||Alex Fink <000024@...>|
|Date:||Tuesday, July 8, 2008, 1:23|
On Mon, 7 Jul 2008 22:16:32 +0200, Jörg Rhiemeier <joerg_rhiemeier@...>
>Perhaps the (as I call it) "arithmographic" approach of Leibniz
>works better than a Wilkins-style taxonomic vocabulary.
This arithmographic approach is interesting. Do you have a pointer to
Leibniz' discussion of such a language? What did he call the approach? Did
he actually undertake the design of one to any significant extent? For that
matter, do you have anything presentable about X-5 yet?
A potential source of difficulty that suggests itself to me, if primes are
going to correspond give or take to morphemes, is commutativity and
associativity. Associativity is already a problem with concatenative
morphology (_malsanulejo_; "pretty little girls' school"), but losing even
the order of the morphemes seems like it would make this fairly worse.
If you mean primes to correspond to semantic atoms of some sort instead of
morphemes, this _could_ be less of a problem -- but I'd really have to see
it done to be convinced. The claim that it makes sense to treat sememes
this way, so they're all orthogonal and freely combinable without
sensitivity to order, seems to me at least as strong a claim about semantic
space as traditional hierarchical taxonomies make.
I'm doubtless entirely ruining this as a puzzle by posting it in this
thread, but, in a cryptography course I once took, it was posed as a bonus
problem to decode a text something like this one:
1306835 46657 290191 297690715 15549 1439383 1342 25073257 582567818284265
212005876221425 [but rather longer]
which encoding scheme has, for just these reasons, potential problems with
lossiness: for instance, what tense is 1562 supposed to be -- or is it in
fact a noun?