Re: [lojban] Toward a {ce'u} record

["Bob LeChevalier (lojbab)" <lojbab@lojban.org>]

At 11:25 AM 8/19/01 -0400, pycyn@aol.com wrote:
> "A propositional function [roughly, property or relation] is an incomplete
> object whose completion is a proposition" (Frege, loose trat). So, every
> {ka} insofar as it creates a propositional function, property or relation,
> contains at least one hole and that is marked by {ce'u}. Did it contain no
> holes, it would be a complete object and, asuming the original type was
> right, a proposition.
> I*think* everyone agrees thus far.

I think so, put this way. To me the archetypal ka has holes in ALL places 
that are unfilled, including all plausible modal places. Others seem to 
think that the archetype has one and only one hole which is filled by the 
ce'u and indicates what the property is "about".

I will agree that most places where a ka abstraction is used in a sumti are 
single-hole examples, but I don't want the assumption that this is always 
the case to hold because I'm pretty sure that exceptions can and will occur.

> So, the disagreement is about whether the {ce'u} must always be written in
> and, if not, where the implicit one is.

You use singular ce'u here, begging my issue.

> 1) Every {ce'u} must be explicit. . Any slot not filled by an overt marker
> is filled by {zo'e} or some such thing. At least one slot must be filled by
> {ce'u} (unless we collapse the distinction between {ka} and {du'u}, in which
> case, {du'u} are the {ce'u}-less {ka} -- or conversely). An easy rule and
> ambiguity-proof, but possibly verbose.
> 2) Not so -- some {ce'u} may be implicit, so long as there is a rule for
> identifying the place(s). The rule may now be somewhat more complex, but the
> results will be less verbose (generally).

How about 3) Like most aspects of the language, features which are implicit 
are optional with no guarantee of a "rule" that identifies the 
preference. After all there is no "rule" for what tense applies in a 
tenseless bridi, or what number applies in a non-quantified sumti. There 
are *conventions* for some cases but for the most part these conventions 
are understood to be less-than-binding over all usages.

Conventions, being non-binding, are best when they are descriptive of 
normal use and not prescription that contradict normal usage from before 
the convention.

> A) The implicit {ce'u} is always the first (x1) place. This runs into
> immediate conflict with the possibility (indeed, reality) of {ka} phrases in
> which the first place is filled with a content sumti. This is not
> ungrammatical, so it needs an interpretation.
> i) Assuming there are no explicit {ce'u} in the phrase, this is treated like
> other {ce'u}-less {ka} -- reduced to {du'u}. With an explicit {ce'u}
> elsewhere, it is taken as the propositional function defined by the explicit
> places, with no implicit ones (it was on a permission, after all).
> ii) The {ce'u} is always in the first place, even if something else is also
> there. The something else is
> a) an exemplary argument to which the function applies to produce a true
> proposition or a new propositional function (depending on whether there are
> explicit {ce'u}), but the {ka} phrase refers to this an all other such
> functions. This does not seem to really give the first-place phrase any role
> that would justify it being there, unless it is to suggest a range of values
> for {ce'u}, and that would better be done using explicit predicates.
> b) as in a), but now the {ka} phrase indicates just the function with the
> sumti in first place. This reduces to i).
> c) like a) in all respects except that the sumti in first place markes a
> special relationship between its referent and the function, which is
> different from (but may include) the application relationship a) assumes.
> {leka do xunre} may or may not mean that {do xunre} is true, but it indicates
> a special relationship between you and leka ce'u xunre. The nature of this
> relationship is not specified anywhere that I can find, and so the whole does
> not seem different from {ledo ka ce'u xunre} (to be on the safe side), which
> is also unexplained, but in the same way.

No. It is unexplained in a different way. leka do xunre specifically 
associates do with filling the x1 of xunre; ledo ka ce'u xunre says 
absolutely nothing about a relationship between do and x1 or with any other 
specific or modal place in the xunre predicate.

> B) The implicit {ce'u} is the first unfilled place in the bridi as written
> (if none then {du'u}). This comes, in a way that A) does not, into conflict
> with other Lojbanic habits, in particular, not filling uninteresting places,
> dropping {zo'e}. Using it correctly requires noticing that the place
> (assuming it is not the first, as it most often will be) is important, since
> it gets a {ce'u} and then dropping that {ce'u}. It thus is harder to use
> than A when it does not have the same effect as A and so harder than A
> altogether, and more likely to errors in what is said.

You are failing to distinguish between "interesting" and "important". All 
of the places of a predicate are "important" by definition or they would 
not be part of the place structure, but some are not interesting because 
common sense or context is sufficient to identify the value, or because, 
while we recognize that there has to be a value, we really aren't will to 
think about what it is.

ce'u needs to be stated explicitly when the listener isn't likely to figure 
it out from context.

> C) A) and B) save at most a couple of syllables, so could B be generalized
> to, say, all the unfilled spaces up to the first explicit {zo'e}. This is
> actually simpler than B) since we only have to decide that something is
> unimportant and put in a {zo'e}, not decide it is important and then leave
> out a {ce'u} we were going to put in. It also gives rather natural results,
> e.g. {le ka prami} is the love relationship, not either the property of being
> loved or of being a lover. It could be extended (but I doubt it is worth it)
> by returning to {ce'u} after an explicit one after a {zo'e}.
>
> The last example raises a general issue: each occurrence of {ce'u} is new,
> independent of others in the context (like {ma} and unlike {ke'a}). How,
> then, do we force two occurrences to be the same, as we can do with the
> lambda operator from which {ce'u} derives. How, for example, do we talk
> about self-love, leka prami with the two implicit {ce'u} identified. Notice,
> we can't do this with any identity predicate, since that just introduces two
> more {ce'u}, unconnected with the earlier ones. For this and general
> reasons, I suggest that {ce'u}, like KOhA generally, be taken as having
> implicit subscripts (starting with 0) assigned in left to right order. So,
> self -love is le ka ce'uxino prami ce'uxino, which might be shortened somehow
> (to, for example, {le ka prami ce'uxino}) but probably shouldn't be.

leka ce'u prami ri
--
lojbab lojbab@lojban.org
Bob LeChevalier, President, The Logical Language Group, Inc.
2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273
Artificial language Loglan/Lojban: http://www.lojban.org