Re: diversity

[Jorge Llambias <jorge@PHYAST.PITT.EDU>]

And:
> I suspect that opaque references arise from some element in the meaning
> of djica & co.

No doubt.

> It is better to make the definition & sumti structure
> of djica & co accurately reflect its true meaning (such that it may
> give rise to opacity) rather than leave the definition & sumti structure
> confused & patch it up by an opacity marker.

What is a true meaning? Why would the sumti structure be confused if
it accepts an object as well as an event?

I agree that the meaning is not precisely the same with an object or with
an event, but that is true for any predicate that takes both. If djica
can't take an object (transparently), then no predicate should be able
to accept both objects and events.

> (I am in favour of a kind of opacity marker that means "the following
> sumti can't be exported to the prenex out of the abstraction containing
> the bridi the sumti is sumti of".)

That's {tu'a}. It already exists.

> If this is throwing out the baby with the bathwater, it is justified
> in that the baby is as defective as the bathwater - in fact the problem
> is with the baby more than with the bathwater. (Where baby = definition
> & sumti structure of djica et al, & bathwater = opacity problem.)

Then to be consistent we'd have to identify all those et als, and forbid
any lujvo with similar characteristics. And whatever for?

> As I have just argued in a previous message, "want" can't meaningfully
> accept objects as its x2 unless it is defined so that it can accept
> objects as its x2, in which case reference must and should be transparent.

I agree. And that's how I'd like to define it.

> If an apple and an event can both be x2 I don't know what "djica" would
> mean. The relation between x1 and x2 in "I want an apple" and "I
> want you to go" is different. You must be more explicit about what
> you think "djica" should mean.

Do you know what any of these mean:

skicu     ski      describe
x1 tells about/describes x2 (object/event/state) to audience x3
with description x4 (property)

spuda     spu      reply                'respond'
x1 answers/replies to/responds to person/object/event/situation/stimulus
x2 with response x3

vajni vaj     vai  important            'significant'
x1 (object/event) is important/significant to x2 (person/event)
in aspect/for reason x3 (nu/ka)

xamgu xag     xau  good
x1 (object/event) is good/beneficial/nice/[acceptable] for x2 by standard x3

zanru zar     zau  approve
x1 approves of/gives favor to plan/action x2 (object/event)

These are a few of the gismu that have object/event places. I don't think
that the change in meaning when going from object to event in any of these
is much different from the same for djica.

> > I predict that it will be used as I propose, i.e. someone looking
> > for their umbrella will simply say {mi sisku le mi santa}.
>
> I think you predict rightly. In that case, the present x2 of sisku should be
> changed. The revised x2 would & should be transparent. To get an opque
> reference, use "troci lo siho cohe".

Suppose it is revised and made transparent. What would {mi sisku lo'e tanxe}
mean? Doesn't it mean something very close to "I'm looking for a box"?


> > > I wonder whether the meanings of "want" and "seek" are in fact
> > > more complex than you believe them to be.
> >
> > I have no doubt about it, but so are the meanings of any other verb
> > that you care to mention.
>
> Surely not. Most of the gismu are quite straightforward. "gerku", "pinji",
> "ckafe" etc.

I did say verb, not gismu. I agree that those that come from nouns
are usually pretty straightforward.

> Your solution would "solve" the problem at one stroke, by providing
> us with a way to express opaque reference. But it would be antilogical,
> unlojbanic.

Of course you can call it unlojbanic, I've done that a few times with
things I don't like, but what's antilogical about it? Does it lead
to any logical inconsistency?

> There is a conflict between my desire for the syntax to
> faifhfully reflect the semantics, and your desire for brevity.

I wouldn't sacrifice logicality to brevity, but I don't see that
problem in this case.

> The
> most I would be contented with is for both solutions to be adopted.

My solution in some sense includes yours, although I'm still unsure
about mixing du'u-things with nu-things. (For many predicates it is
explicitly allowed though, so it's nothing new.)

> But I do feel that my solution should come first, with yours adopted
> only if the abbreviatory device is generally felt to be needed.

Speaking in practical terms, I doubt very much that we'll be able to
avoid the natural language solution (let context determine when the
reference is opaque and when it isn't), but we can try.

Jorge