Hallo!

On Thu, 3 Jul 2008 21:12:55 -0400, Herman Miller wrote:

> Even inside those semantic domains, problems can come up. Take plant 
> classification: you've got hundreds of unfamiliar groups, and scattered 
> among those are a relatively few groups that contain the more familiar 
> plants. Also, these hierarchies can be wildly unbalanced. Ginkgo is a 
> whole division of the plant kingdom all by itself, while three levels 
> deeper on the hierarchy, the family Asteraceae has some 23,000 species 
> according to Wikipedia.

Yes.  The taxonomy of living beings is unbalanced because nature
is unbalanced, too.  A taxonomy that mirrors these unbalances may
be good for scientific purposes, but not for everyday usage.  Also,
there are often more useful criteria than phylogenesis.  I mean,
how important is the fact that apples and pears belong to the same
family (Rosaceae) as roses in everyday life?  The distinction
between 'fruits', 'vegetables', 'flowers' etc. is more useful, even
though it cuts across the scientific taxonomy.

> Yahoo used to have a classified index of web pages (and perhaps still 
> does; I haven't used it in a long time). I found it useful for finding 
> pages when I was trying to learn Japanese. Wikipedia's categorical index 
> can also be useful, although it can be haphazard in places. I still use 
> categorical word lists for my recent languages, which can be useful for 
> later clarifying the intended meanings of words if you weren't careful 
> enough in the first place, or finding words of similar meaning.

Sure, classifications (at least if done well) _are_ useful.
But classifying documents automatically throws up all sorts of
problems, and there are always things a classification misses.

On Fri, 4 Jul 2008 11:24:39 +0200, Benct Philip Jonsson wrote:

> Lars Finsen scripsit:
> [...]
>  > I also made use of some classification in the
>  > early stages of developing Urianian. In fact, I
>  > bought my English Thesaurus just for that
>  > purpose (no Norwegian thesaurus existed then).
>  > Thesauruses provide rather detailed
>  > classifications of words. However I found this
>  > one messy, so I made my own. Can't refer to it
>  > in detail now, as I'm not at home. But will
>  > later, if it's of any interest to the list.
> 
> IME even a buggy and mostly arbitrary
> classification like Roget's can be useful in
> vocabulary building if only because it is readily
> available to everyone. BTW I think Roget put more
> thought than e.g. dalgarno or Wilkins into his
> classification. To have a rather limited and broad
> scheme was one Good Thing. Having a limited and
> well-defined purpose was another. Not to claim
> that he got *the* final and perfect classification
> of everything was a third.

Classifications and thesauri are indeed very useful to
conlangers for vocabulary building.  Keeping your words
in a thematic dictionary like this one:

http://www.zompist.com/thematic.htm

is a great way of coming up with a well-rounded vocabulary.
With an alphabetically sorted word list, it is difficult
to see whether there are gaps in some fields of discourse.
A thematic dictionary, in contrast, gives you an instant
overview on which lexical fields still need more work.
Of course, if you forget something in your classification,
you will end up with a blank spot in your language's
vocabulary.

... brought to you by the Weeping Elf