Re: replies mainly re "ka"

[ucleaar <ucleaar@ucl.ac.uk>]

Jorge:
> la and cusku di'e
> > > > (b) It's not clear to me that your example does involve a property.
> > > Can two things differ in anything but their properties?
> > No.
> > If x1 differs from x2 in property x3, whose property is x3's supposed
> > to be? x1's or x2's? If it's a property they both have, then it's not
> > one by which they differ.
> Do you consider "size" to be a property? Can two things have that property
> and differ in that property? i.e. can two things differ in size?

We must distinguish "features" from "feature values":
      feature      feature value
(I)   colour       red, blue, green
      sex          male, female
      species      human, dog
(II)  age          3, 42
      height       5', 6'
      mother       Sue, Sophy
      hobby        origami, moral philosophy

"Differ" needs a feature (e.g. size) as x3, not a feature value.

I've distinguished 2 types of feature. In (I) the value of the feature
is a category, so if the colour of X is red, then X is a red (a red
thing). The values of a type (III) feature of X is not a category
containing X.

I have been assuming that "property" means "red, blue, green, male,
female, human, dog". Quite possibly I've misunderstood the intention.

> > > > ("Ka" works like the
> > > > predicate "x1 is species of x2".) So it doesn't make sense
> > > > to have a full bridi (i.e. having a truth value) as complement
> > > > of ka.
> > > Does a full bridi for {li'i} make sense to you?
> > Yes. "Mi se lihi [zohe] carvi" makes sense to me as "I experience rain".
> But that is {mi se li'i carvi ke'a} or can you experience it by just
> seeing it? But then what is the difference between experiencing and
> observing?

Observing is one way of experiencing. There are other ways too.

> > But "I experience rain" can be "Mi lifri lo carvi", and the tricky
> > thing to say is "I experience being a man". So I would like to see
> > ka and lihi parallel, as you suggest. In that case, "Mi se lihi [zohe]
> > carvi" ought ideally to be ungrammatical, and "Mi se lihi keha nanmu"
> > means "I experience being a man".
> I'm not sure I get the problem. I don't much like {mi lifri lo carvi}
> instead of {mi lifri le nu carvi}. It is sumti raising.

I can experience rain by seeing it, or feeling its wetness.

> I think that the only difference between {ka} and {li'i} is that for
> {li'i} the one having the property has to be sentient and conscious
> of the property.

This is the view I came to. So we agree on lihi.

> > I am here using your keha for purposes of clarity only. I believe
> > the need for it shows that ka, and, I am persuaded, lihi, are lodged
> > in the wrong selmaho.
> I don't agree. As long as we have a lambda variable, it can be made
> perfectly clear what is the slot in question, and when it is obvious,
> omitting the lambda variable causes no problem.

I'm not saying that because ka and lihi are in the wrong selmaho
there are certain things we are unable to say. Rather, I am saying
that the logical structure expressed by ka and lihi is the logical
structure of words in a different selmaho from NU.

> > > You would say: {lo ka ke'a nelci do kei be mi} = "the property of
> > > liking you which I exhibit" or {lo ka mi nelci ke'a kei be do} = "the
> > > property of being liked by me which you exhibit".
> > > If ke'a is replaced in these cases by the corresponding sumti, it still
> > > makes sense, even if it's not perfectly logical to do so.
> > If keha is replaced in these cases by a non-corresponding sumti, then
> > you get syntactically fine garbage that is semantic garbage (real junk,
> > not just something surreal [=pragmatic "nonsense"]).
> I agree, but syntactically correct Lojban is not guaranteed to be free
> of semantic garbage.

Well that strikes me as a bug in the language. Every syntactically
well-formed Lojban sentence ought to be translatable into some form
of predicate calculus, in principle (though sufficiently rich and
standard forms of pred calc may not have been devised yet).

> There is no syntactic rule that constrains {ke'a} to be inside a
> relative clause anyway, so there you have another nice source of
> semantic garbage.

This is easily rectified by a rule that defines what keha means when
it is not within a relative clause (e.g. it is equivalent to a zohe).

> Why should you want to syntactically eliminate the possibility of
> garbage in {ka} when there is a lot of possibility everywhere else.

I should wish to eliminate the possibility of garbage everywhere.
Even if some is ineliminable, I don't see why we should therefore
give up trying to eliminate it where we can.

> > You advocate a grammar change to allow use of keha.
> It depends what you mean by grammar change. I don't want to make
> any change in syntax, since {ke'a} is already syntactically
> allowed, being an ordinary sumti.

I include within the grammar the rules that derive logical structures
from sentences, so in this sense you are proposing a grammar change.
This is a reasonable (as well as standard) notion of grammar: it's
what a learner (or a Lojban-speaking computer) will have to learn.

> > I instead advocate a grammar change whereby ka and lihi move out of
> > NU. Ka could move into LE, but a more conservative move would be
> > to set up a new selmaho for ka and lihi, such that:
> >    as a syntactic head it behaves like LE
> >    is a syntactic dependent it is a selbri
> I don't see how that would work. If you want it to be like LE you will
> need to use lots of {be}s and you lose the flexibility that {ke'a} gives
> you because you are forced to bring the slot in question to the x1 place.

You do see how it would work. Indeed my proposals would lose flexibility
and require {be}s. Exactly the same criticism can be levelled at LE,
which would be more flexible and require no {be}s if it took as
complement a full bridi containing a keha. I think that syntactic-
semantic homomorphism, or at least consistency in syntax-semantics
correspondences is more valuable than flexibility and concision.

> > This better maintains the pleasing homomorphism between syntax and
> > semantics in Lojban.
>
> There is no such homomorphism. Semantic garbage abounds in syntactic
> space.

What are examples? If there are any, I doubt if they are to be found
in the foundations of Lojban's/Loglan's design.

> > > > Since I think ka yields a singleton category, the choice of gadri
> > > > (lo vs. lohe) doesn't matter.
> > > That's why I always prefer {le ka}. I agree that ka yields a singleton,
> > > and to me, {le} is the best gadri for such things, but I agree that
> > > {lo} is also acceptable. (Same thing with {le du'u})
> > That's interesting. I mean about duhu. I reckon ni is the same.
> I had forgotten about {ni} in my classification! I never really
> understood it, I think it's probably expressable as {le ka ...xokau... }, but
> anyway, I'm not sure it has to be a singleton.

When wouldn't it be a singleton? Are you thinking of "the amounts by
which John and Mary are tall"? I think if you meant two separate
amounts that should be expanded to "the amount by which John is tall
and the amount by which Mary is tall", just as if you want to say
"the propositions John is tall and Mary is tall" you need to
expand to "The duhu John is tall and the duhu Mary is tall".

> > But not lihi or siho or nu.
> {li'i} and {si'o} would seem to be the personal equivalents of {ka} and
> {du'u}, and I suppose you could say that they are not singletons because
> they are subjective. I agree too that {nu} is not a singleton.

Right.

> This is how I'm seeing the 'abstractions' now:
> nu   (mu'e, pu'u, za'i, zu'o)
> du'u (si'o)  (su'u = du'u taimakau...)
> ka   (li'i)  (ni = ka ...xokau...)
> jei (not really an abstraction, but often misused as jei = du'u xukau)

I agree with the analysis, I think, except for the rendering of ni, since
I'm opposed to ka being in NU. And "duhu taimakau" is more specific than
(I presume) suhu is intended to be.

> > > It is not the property that is a process. It is the acquiring of it.
> > No I did mean the property is a process. But I see now for the first
> > time a problem.
> > This is what I meant: A property is a category, which is basically
> > like a set. The set is a process: it comes into being in stages.
> I don't think sets can be processes either.

Do you see how a set can come into being in stages? If so, you can see
how a set can be a process.

> > > To me, na'o means "typically" in the sense of "for most of the time", not
> > > "for most cases". Your sentence says that for each Scotsman the property
> > > is inherited most of the time, but not constantly.
> > That looks like ro scoping over naho.
> > What I want is "In general, if you examine the world you will find that
> > each scotsman inherits the property", "the world is generally such that
> > each scotsman inherits the property", "it is generally the case that
> > each scotsman ....".
> And I think that {na'o} has nothing to do with that, because it is a time
> tense, a modifier of a time interval. Am I wrong?
> Maybe {su'a} would help. It means something like "in general".

{su'a} is a discursive: what is needed here is something that actually
modifies the proposition expressed by the bridi, so not something in
UI. I'm still not convinced that naho doesn't mean what I thought it
did.

----
And