Message-Id: <m0jdQKi-00030fC@snark.thyrsus.com>
From: cowan (John Cowan)
Subject: Re: More Chemelem stuff
To: lojban-list
Date: Wed, 15 May 91 14:12:19 EDT
In-Reply-To: <9105150157.23130@mullian.ee.mu.OZ.AU>; from "mullian.ee.mu.OZ.AU!nsn" at May 15, 91 11:57 am
Status: RO

la nitcion. cusku di'e:
> What I would like (and what would finally put the MFC/Chemelem/flora/fauna 
> dispute to rest) is for these items to leave the gismu list and be recognised
> as our first official Type 4 le'avla. The semantic silliness we've con-
> structed around them would die off, and given that all le'avla can be used as
> rafsi and modifiers, the criterion for Type 4 vs. Type 3 usage becomes simply
> one of Zipfism or neutrality, rather than semantics. This changes only the
> clasification of words, not necessarily the words themselves; so such a change
> need not be unnecessarily disruptive (as outright zapping the words' current
> forms might be). As a byproduct, the gismu move much closer to being a set
> of semantic primitives.

There is a difference between Type 4 le'avla and gismu, though, and that's
the size of the rafsi.  Given a (hypothetical) gismu "fubra", the rafsi are
	fubry		(for sure)
	fub or fur	(possible)
	fu'a		(possible)
	fra or bra	(possible)

However, if "fubra" is a lea'vla, the only rafsi form is "fubra,iy", longer than
the lea'vla itself.  This is a significant difference:  "lojbau" (the
predicate form of "lojban.") is a lot more palatable than "lojbo,iybau".

> Re other things: the diklujvo po'e la djim. are a start for discussion, but
> not the solution for lojban, of course. His first class is impossible in
> lojban: a lujvo MUST have the place structure of its last rafsi (we shouldn't
> be playing games with place structures like that).

On the contrary.  The place structure of a tanru is indeed dictated by the
place structure of the last element, but this rule does NOT hold for lujvo.
It is frequently the case that lujvo either need places supplied from the
nonfinal elements, or else have some of the places of the final element
automatically filled in.

> It is not essential to
> resolve between parallel and transitive, or place structure - leave it to
> context. that's what we have {ki'a} for.

The whole purpose of lujvo, as opposed to tanru, is to "freeze" some of these
decisions so they do not have to be thought out on the fly.  "brivla", for
example, is a lujvo, but is no longer felt as such (at least by me): it is
simply the Lojban term for one of its fundamental word classes, distinguished
from cmavo and cmene.


-- 
cowan@snark.thyrsus.com		...!uunet!cbmvax!snark!cowan
		e'osai ko sarji la lojban