Re: tech:opaque

[ucleaar <ucleaar@UCL.AC.UK>]

This message is probably worth reading if you're at all interested
in how lojban handles opacity. (It lays out two opposed positions
quite clearlyish, I hope.)

Replying to pc:

I think here we're really getting to the crux of the debate.

I think:
  The rules of the language are already clear enough to tell us
  that [on the whole] the only source of opacity is through bridi
  subordination, even if the community of users [barring Jorge] has
  not realized this so far.
You think:
  The community of users understand opacity from sources additional
  to bridi subordination; therefore we need some rules to regulate
  this understanding and some cmavo to signal it.

pc> > I think it would be very nice indeed if there were no gismu
pc> > that had opaque places except those that took full lenu (etc.) sumti.
ar> That is indeed, I believe, the current situation as far as the design
ar> goes, though not as far as usage goes.
pc:
> Did I miss a major revision?

No.

> My dictionary file still has _kalte_ as "x1 hunts x2," with no
> indication that x2 is transparent. To be sure, there is not indication
> that x2 is opaque, either, but, since it is in "hunt" in English, feel
> free to carry that over, else some other translation would be used or
> at least indicated.

(i) Gismu glosses are merely approximate indications of the actual sense
of the gismu. There is no guarantee that the gismu means whatever its
keyword gloss means. The actual sense of the gismu is partly determined
by its place structure, with the rest to be filled out by current
and future usage. Thus nothing can reliably be inferred from the
keyword "hunt".
(ii) The gismu list fails to consider issues of opacity, so nothing
can be inferred from the absence of any mention of whether x2 of
kalte is opaque or transparent.
(iii) My (& Jorge's, I believe) contention is that no selbri can
render any of its sumti opaque. All quantified sumti export to the
prenex of the localmost bridi, unless there is explicit indication
that they export to a higher prenex.
(iv) Only by subordinate bridi can we get the effects of opacity.

> But, truth to tell, I can't think of a translation that is reasonably
> like "hunt" but has a transparent x2.

If you could think of one, then that would make a better keyword.
But if we can't think of a better one, we must just bear in mind that
one must approach the english gloss with caution.

> It is part of the notion of hunting that I may not find and that,
> generally, I do not care what I find, so long as it is the right
> sort of thing, both of which militate against transparency of any
> kind of _lo/loi_ expression at x2. For, unlike the case of  "I
> will kill a deer," which is false if I don't and whose preposed
> existential is instantiated with the one I get if I do, "I will hunt a
> deer" may be true even if I never see one and have no one even in
> mind as I tramp the woods, so that "There is a deer I will hunt"
> turns out false.

I contend that you describe "hunting" correctly, but that your
description does not apply to "kalteing". {mi ba kalte lo mirli} means
only "There is a deer I will hunt".

> I can do English "hunt" in Lojban without _kalte_,

Good. Then there's no problem for Lojban then.

> but I cannot see how to do _kalte_ in English, if x2 is transparent.

There indeed is no single word that is equivalent to kalte. But
"There is a deer I will hunt" renders {mi ba kalte lo mirli}
reasonably accurately.

> Nor can I imagine why we would want such a word -- and a gismu at that!

No, I do not think them valuable. They were added before these opacity
issues were thought through, and now we're lumbered with them. Either
they're always transparent, or they get redefined to take a {duu}
argument.

> I don't know of a use for it, except those special cases (already
> covered by other means) when we want external reference in opaque slots,
> when there is a specific deer I am hunting (Ol' Snagglehorn) or I have
> a guarantee of success (past tense references to successful hunts, for
> example -- but how did we talk of even those hunts before we went out?).
> And this applies to all the old opaque words other than those that take
> full bridi arguments?  We can only paint pictures of real scenes? We can
> only look for things we are sure to find ( Jesus' assurance seems less
> inspirinig in Lojban)? Even are dream girl must be the one down the
> block?  Well, not quite, of course, because we can say the right thing
> eventually, but to get there we have to mow through a mass of things we
> never use except at the bottom of heaped _lenu_ clauses. Hawk (I omit
> the correct spelling here since it is the vilest insult in Klingon and
> I do not want to embarrass Nick) ptui!

Basically yes to all these, except for the picture painting. For the
picture painting, the scene can be non real if it is selected from an
expanded domain of the not-necessarily-real. The way I had thought to
do this is with {dahi}, as in {skicu lo dahi gerku} - where the dog
comes into existence in drawn form - but this can't be right, for
{dahi} is in UI and therefore metalinguistic, whereas the purpose for
which I had been attempting to put it to use very much changes truth
conditions. So something in - I suppose - NAhE is called for, to expand
the domain from the real to the not-necessarily-real.

The general solution is to create brivla that have a leduu/kuau x2.
E.g. instead of {kalte} use {troci kuau kavbu}.

My position requires a usage change, but not a grammar (in broad
sense) change. Your position approves of glico-influenced usage,
and requires a grammar change to make sense of that usage. I think
the onus is on you to spell out the grammar change in detail and
sell it to the decreers of change.

Lojbab:
> sisku was changed a while back to be "searching for property in set".
> There was further discussion in the endless "any" debate that MAY have
> concluded that this weird formulation was unnecessary as well as 'opaque'
> to the typical user %^), but no official decisions came out of said debate
> because no one has attempted to figure out what if anything was decided.
> So it is possible you are trying to talk about sisku and not kalte.

No, we must ignore sisku-as-currently defined, because it's an
aberration. Some time a fair bit prior to the great opacity debate
(in 1993 maybe?), the problem of opacity was discussed, but only
as it applied to {sisku} (where John's "Galadriel seeks a unicorn"
originated, & Mark of course spotted the <la> - so if you can grep
archives, try earliest uses of "Galadriel" or "gaLADri,el") - it
wasn't realized that the problem is in fact much more widespread.
And as you suggest, the new definition given to sisku hardly makes
a ton of sense.

Sisku should really go back to behaving like kalte & co.

coo, mie And