From snark!cowan Fri Aug  9 16:40:37 1991
Return-Path: <snark!cowan>
Message-Id: <m0k8dcg-0002VSC@snark.thyrsus.com>
Subject: Re: Definition of lujvo
To: cbmvax!uunet!math.ucla.edu!jimc
Date: Fri, 9 Aug 91 16:38:31 EDT
From: John Cowan <cbmvax!uunet!marob.masa.com!cowan>
In-Reply-To: <9108091907.AA28799@luna.math.ucla.edu>; from "math.ucla.edu!jimc" at Aug 9, 91 12:07 pm
X-Mailer: ELM [version 2.3 PL11]
Sender: cowan
Status: RO


My remei fepni on lujvo, creative vs. algorithmic:

I don't think there's a genuine dispute that regularities exist that make
the interpretation of lujvo straightforward.  The question is, what
regularities are they?  This leads to a meta-question:  should the
regularities be prescribed in advance of usage (as in -gua!spi) or
figured out on the basis of usage?  If the latter, there is simply not
enough usage yet.

-gua!spi lujvo rules (or tanru rules; lujvo and tanru are the same thing
in that language) were figured out on the basis of jimc's usage.  He then
chose to prescribe them for all future usage, adding mechanisms to the
language to unambiguously determine which of his three compounding rules
is to be employed at any point.

For Lojban, however, the view taken by lojbab (and others at Lojban
Central) is that it is far too early to prescribe compounding rules.
Instead, lots of lujvo need to be made and actually used in text (not
just in a word-list).  Then the compounding rules can be inferred from
that data base.  This is more or less the approach taken by Kalocsay
to Esperanto lujvo.  la nitcion.  has recently associated himself with
this approach, as have I.

I think, therefore, that jimc's "diklujvo" rules would better be interpreted
as a set of lujvo heuristics, something like the following:

1) Some two-part lujvo A-B can be interpreted as "B be FA lo nu A".
(FA means a member of lexeme FA specifying some place of B.)
This interpretation is plausible when B has a place which can be filled
most naturally by an event; the lujvo typically has the places of B, except
that the event place is replaced by the places of A.  Example:

	ko'a caryri'a ko'e
	X turn-causes Y
	ko'a rinka lo nu ko'e carna
	X causes the event-of Y turns

2) Some two-part lujvo A-B can be interpeted as "A je B".
This interpretation is plausible when A has only one place, but possible
even with more than one place.  The lujvo typically has the places of B.
If A has more than one place, a determination needs to be made whether
any of A's places logically coincide with those of B; if so, they are
not repeated.  Any remaining places of A are typically placed at the
end of the lujvo place structure.  Example:

	ko'a blazdani ko'e
	X is-a-blue-house-for Y
	ko'a blanu je zdani ko'e
	X is-blue and a-house-for Y

3) Some two-part lujvo A-B can be interpreted as "A be FA lo B".
(FA means the same as before.)  This interpretation is plausible when
"lo B" could plausibly fill one of the places of A.  The lujvo typically
has the places of B, except that the place being filled is removed and
replaced by the places of A other than the first.  Example:

	ko'a mi'ife'o ko'e
	X machine-sews Y
	ko'a fenso be fo lo minji ko'e
	X sews with-tool a machine Y

Note that here the second place of "minji" (the purpose) is dropped from
the lujvo because it is obvious (the purpose is sewing).

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