Kinds of knowledge was (RE: An elegant distinction (was Re: brz, or Plan B revisited (LONG)))

On Sun, 25 Sep 2005 Paul Bennett wrote:
> On Sun, 25 Sep 2005 12:45:59 -0400, Jrg Rhiemeier
> <joerg_rhiemeier@...> wrote:
>
> >> Jrg Rhiemeier <joerg_rhiemeier@...> writes:
> >> >...
> >> > > For evaluation purposes, I do find the post-fix order easier (too
> >> much
> >> > > Reverse Polish, no doubt  :)
> >> >
> >> > My personal taste goes more towards a prefix order.  But at any rate
> >> > NOT SVO!  That's unelegant, and the notions of "subject" and "object"
> >> > doesn't apply well to loglangs anyway.
> >>
> >> Just curious: why 'unelegant'?  I can understand that for more than
> >> two arguments, you get an asymmetry that might create the unelegance
> >> feeling.
> >
> > Yes.  I was referring to Lojban-style languages where predicates
> > can have any number of arguments, and surely "X1 P X2 X3 X4" as
> > found (to my meager knowledge) in Lojban is not very elegant,
> > is it?
>
> I think I see the problem. You guys seem to be conflating or confusing the
> artistic sense of "elegant" as "attractive" with the computer science
> sense of "elegant" as "well-designed".
>
> I think. Maybe. Certainly, it's a concept that needs to be more clearly
> defined when talking of loglangs and engelangs.
>
> It's an interesting usage point, I think, and kinda leads into the
> question:
>
> In what ways does your conlang partition semantic space differently to
> your native language? I'm most interested in the subtle, non-obvious or
> non-trivial ways this happens in different contexts or fields of
> discussion, but anything goes.
>
> I suspect this may require, in some cases, discussion of the ways your
> native language partitions semantic space differently to English, but I
> think I wouldn't mind that at all.
>
> Paul

Hi Paul, Jörg and others,

Possible answers to this question interest me greatly.  I see them
as perhaps contributing to a better understanding of this rather
more abstract issue:
      In what ways can any language partition semantic space?

Still, I wonder whether it's possible to answer such a question
in any natural language.   My doubts arise from the analogy of
natural languages to formal systems, where the well-known
theorem of Gödel tells us that no system can describe itself.
(Or at least I think it does - that may be TOO loose a paraphrase.)

Given, firstly, the great effort that so many people have seen fit
to expend on creating artificial languages of all kinds, and for so
many purposes; given, secondly, that a large proportion of those
conlangs are, it seems, deemed deficient in major ways by the
cognoscenti on this forum; and given, finally, that I myself have
much less time and energy than I would wish available to spend on
creating a conlang of my own, I feel it would be wasteful to spend
it in possibly recreating past errors.  For example, I have no wish
to simply "relexify" (lovely word!) any of the languages with which
I am familiar.  Equally, I do not wish to create something of less
expressive power than any natlang, nor of greater verbosity.
Ideally, my conlang would be novel in its organisation, and offer
possibilities of expressing meanings not yet given by any other
language of any origin.  In the light of this ideal, I would find it
very useful to have some kind of universal catalog of the semantic
capabilities of pre-existent languages.

Because of this, I've found the recent discussions of expressing
different modes of perception and knowledge, eg "I know it
because I saw it" vs "I know it because a friend told me" vs etc,
quite interesting and relevant to this aim.  They struck a chord
in me, because I once wrote an academic paper (1971, unpublished)
on using a multi-valued logic system to express much more than
truth or falsity.  The "thepfi" system proposed, as an example of
such a system, a hierarchy of states of knowledge that could be
posited of assertions.  These included such things as -

o  the truth of the assertion,
o  whether the truth of the assertion could (in principle) be known,
o  whether the truth of the assertion could (in practice) be known,
o  whether the predicate could be postulated of the subject,
and even
o  whether the assertion could be thought of.

Rather than being _merely_ an academic speculation, I believed
then - and still do! - that such an approach might, by giving us
better ways to frame statements of our knowledge, give us a
better handle on understanding the limits of that knowledge.
The system sketched then was not intended to be a final model,
nor do I think it adequate to describe reality.  In particular, I
very much doubt that these states of knowledge fall into a
readily discernible hierarchy, and that a network model may
well describe reality better.

However, the paper was able to show that a hierarchical system
of kinds of knowledge was amenable to a mathematical calculus,
of a form that directly generalises the everyday two-valued
Boolean logic.  (The mathematical foundations are due to Rosser
and Turquette.)

Without losing contact with the natural world in which natlangs
have their ground, I would like to create a conlang that encodes
various kinds of knowledge in an eminently logical, perhaps even
calculable way.  Doing so would lead to making certain kinds of
implication quite direct and natural to such a language.  I believe
this to be possible, but the primary question must be:
   What kinds of knowledge would it be most useful to encode?

I welcome your thoughts!

Regards,
Yahya


--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.344 / Virus Database: 267.11.6/111 - Release Date: 23/9/05

Reply