Re: On {lo} and existence

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

Jorge:
> > If S is a sentence containing {lo broda}, and D is the set of propositions
> > derivable from S (by grammatical rules), then it is always possible
> > to find in D a proposition that is true in this world, and a proposition
> > that is false in this world. The same goes for sentences containing
> > {le broda}.
> I think this is probably true with any interpretation of {lo}.
> If I understand correctly, a sentence is any string of words that makes
> the parser happy. I agree with you that sentences defined like that can't
> be said to have truth values. An utterance is an instance of use of a
> sentence. This is what can have a truth value.
> I'm not sure about the relationship between the set of propositions
> and the utterances, though.

Take a sentence and run it through the semanticosyntactic rules, and
see which proposition(s) you end up with. The sentence {da broda}
yields a single proposition [Ex broda(x)]. By contrast, the sentence
{koha broda} yields an infinitude of truth-evaluable propositions.
This is because {koha} can refer to an infinitude of individuals.
Put another way, {koha broda} yields an underspecified proposition,
which is not truth-evaluable until reference of {koha} has been
assigned.

(Reference assignment (and syntactic disambiguation, for lgs where
that exists) is logically the first step in utterance interpretation.
This step is skipped when all referents are non-specific.)

The crucial difference is between sentences involving constants (specific
reference) and sentences involving only quantified variables (non-
specific). I was suggesting that {lo}, unlike {da poi}, involves
a specific reference (to a certain universe), and was wallowing
rather muddily in the logicophilosophical ramifications of that.

> > However, even if many propositions can be derived from a given sentence,
> > when we utter it there is a guarantee that all but one of these
> > propositions is to be rejected.
> Ok, but what about in the other direction. Is it possible that more
> than one utterance map to the same proposition?

Absolutely it is possible. There is no proposition such that only one
utterance maps to it. To prove that, just find the sentence uttered,
and utter it again. Furthermore, there are sentences (containing only
quantified variables) every utterance of which maps to the same
proposition.

> If yes, then propositions are only a rough coarsening of what we
> transmit by utterances, and in that case I don't agree that there
> is an objective way of assigning a proposition to a given utterance.
> Either the assignment will be precise but arbitrary, or not so
> arbitrary but then quite imprecise. In the last case in borderline
> cases you can't guarantee the rejection of all but one proposition
> for a given utterance.

I'm not sure what you have in mind. {da de gerku} yields
"Ex Ey gerku(x,y)". That seems both precise and non-arbitray.
{mi do gerku} yields "gerku(A.R.,J.L.)".
{koha kohe gerku} yields, out of context, "gerku(X,Y)", resolved
in context to  "gerku(individual38938473,individual78222326)".

> > Consider {le cukta cu blanu}. From
> > this we can derive an infinitude of propositions, differentiated
> > by which individual(s) {le cukta} denotes.
> Yes, and even for the same individual, we still have an infinitude
> of utterances differentiated by the speaker, the audience, the time
> and place of the utterance, etc. These also can have different truth
> values, so I suppose that they correspond to different propositions.

I don't see how different utterances of {da de zehe gerku} correspond
to different propositions. I am talking about propositions derivable
by grammatical rule, not by ad hoc inference.

> > Exactly the same applies to {mi terxra lo cukta}, where {lo cukta}
> > adds "in universe U x is a book". From this too we can derive an
> > infinitude of propositions, differentiated by which universe universe U
> > is. And similarly, if I utter that sentence you know that all but one
> > of those propositions has been discarded; this is because the identity
> > of U gets fixed. And, like every proposition, this undiscarded one
> > must be either true or false.
> Right, but why don't you call that nonveridicality? Just as in
> {le cukta} it doesn't really help us to know what the predicate {cukta}
> means, then in {lo cukta} it doesn't either. We must go inside the
> speaker's head to know what is being refered as {le cukta}, or to know
> what things satisfy {cukta} in that imaginary universe.

I don't call this nonveridicality because it is specificity rather
than nonveridicality that make {lo} and {le} alike.

> You may argue that the meaning of {cukta} in that universe will be
> much related to the meaning of {cukta} that we know, otherwise why not
> use some other word, since in principle in that universe we could
> redefine every selbri. But exactly the same applies to {le cukta}.
> There is a reason why the speaker chose to say {le cukta} over
> {le panka}.

I am immune to arguments from pragmatics. They concern language-as-it-
is-really-used, not language as the abstract idealized (idolized)
object I like to spend all my days pondering.

> >   Ex in This World I drew x & in Middle Earth x is a balrog
> (As an aside, do you agree with me that we only make believe that
> we can write down propositions? What you wrote down is a sentence,
> but because it is a bit more precise than ordinary, we accept to
> call it a "proposition".)

I don't think we can write down propositions, no. But using alternative
notations to ordinary language can be helpful.

> > I maintain [without necessarily believing] that that can't mean
> > "I drew a tulip", in the sense that there is a tulip that the picture
> > resembles.
> Then you should add to your "proposition", & in This World x is not
> a tulip, otherwise how can we tell that the Middle Earth balrog isn't
> a tulip here? What you drew could be anything of this world, as long as
> it is a balrog in M.E. If all balrogs of M.E. are tulips here and
> viceversa, then it means exactly the same as "I drew a tulip".

You are right. I thought some more after I wrote it and thought
of a solution that might salvage things.

(a) A picture or description or story or suchlike, creates (a
fragment of) an imaginary world, I. We form a new universe, N,
from the "union" of this world, W, and I, such that anything
existing in either W or I exists in N. Then we can say:
  In universe I, Ex balrog(x) & Ey is-model-for-picture(x,y).
  "There is a picture of a balrog"

Same solution goes for "She told a story about a bloke with 3
heads", etc.

> > > (The concept of "imaginary" is already a predicate, I don't see
> > > the need to make it a metapredicate.)
> > How would that work? The set of imaginary wings is not formed from
> > the intersection of the set of wings and the set of imaginaries.
> Of course not. If you like, it is the intersection of things imaginary
> and things that look like wings, although it's not quite that either.

It's not clear to me how you'd do "I told a story about a man with
3 heads".

> > Or would you take the view that the set of wings includes imaginary
> > wings, but when we speak of {lo nalci} there is a presumption that
> > we are quantifying not over the set of wings but over the intersection
> > of the set of wings and the set of reals?
> No, I don't take that view. The quantifiers quantify over all things
> that we can predicate about, and these are often not real. All they
> need to do is satisfy some predicate, and we can quantify over them.

Ah, I see. But to decide whether a proposition is true, we do need
to know whether things predicated about are real. So to make statements
that aren't truth-conditionally vacuous we need a way to distinguish
the real from the imaginary.

> > > > You may be right. How would you unraise draw-a-pic-of?
> > > You have to explain the predicate better. "Take a photograph of" is
> > > clearly not a problem, the object is a real object. For drawing
> > > a picture, the case is the same if the x2 is the model. If the
> > > model x2 is not a really-is broda but a generic or somesuch, then
> > > don't use {lo}.
> > The x2 of the predicate I had in mind is not the model but the
> > "subject-matter".
> I don't see the difference. What would be an example of subject matter?
> If I draw a picture of my cat (if I had one) would the subject matter
> be anything other than my cat? That's what I meant by model.

The subject matter and the model would be your cat. But I could draw
(from imagination, not from life) a three-headed cat: this would be
subject-matter but not model. Or, put another way, it is possible for
the model to exist only in an imaginary world created by the picture.

> > > _Every_ {nu <bridi>} would be a potential event.
> > I see no difference between your definition of {nu} and my observation
> > that {lo nu} defaults to {lo dahi nu}. Maybe you see no difference
> > either.
> The difference is that you make {da'i} come from {lo}, and I make it
> come from {nu}. {da poi nu <bridi>} would still be a potential event
> for me, but presumably a happening event for you.

I see. My intention was actually the same as yours. All predicates
default to {dahinai}, except for {nu}, which defaults to {dahi}.

I must seem very confusing, because I've been espousing a number of
equally acceptable but incompatible positions.

> > It's exceptional because it's not verifiable. All other selbri can
> > be verified/falsified by inspecting the world. Put another way:
> > however you restrict the category, it always remains non-empty.
> Yes, that's why I don't like it, I think events should be non-vacuously
> quantifiable, like all other space-time enduring objects.

Well, even if {nu} defaults to {dahi}, one could always use {dahinai}
in order to avoid vacuity.

> But it wouldn't be an exception, at least not the only one. Du'u works
> like that too. {lo'i du'u <bridi>} is never empty.

Ah, but the cardinality of {lohi nu <bridi>} is always infinite, while
the cardinality of {lohi duhu <bridi>} is always exactly one.

> > ({lohi nu broda gihe na broda} would fairly clearly be non-empty,
> > but lo dahi god knows whether {lohi nu broda kei gihe na nu broda}
> > is non-empty.)
> It is empty, but it is not of the form {lo'i nu <bridi>}.

Oh? I thought it should yield the intersection of {lohi nu <bridi>}
and {lohi na nu <bridi>}.

> In any case, I like it less and less the more I think about it.

Even if the remedy of an explicit {dahinai} is available?

> > > > Any of these three is okay by me.
> > > My first choice would be that {nu broda} be a real event of brodaing.
> > > I can accept it being a potential event of brodaing, but I'd hate to
> > > let {lo broda} be anything other than a this-world broda.
> > > ("This world" being the world in which the discourse takes place.)
> > I think that of those who've expressed a view, only Lojbab doesn't
> > go along with your first or second choice,
> But I doubt he accepts your third proposal. That would mean accepting
> that {lo ninmu cu nanmu} can be true in this ordinary world. How are
> we going to teach the nonveridicality of {le} if that is so?  :)

Yes, I'm sure Lojbab would be horrified to see how I've extrapolated
from his position, though he did, I think clearly articulate the view
that {lo} makes a weaker existential claim than {da poi}. I guess life
would be easiest for all concerned if we agreed that by default
   <selbri> = dahinai <selbri>
   nu = dahi nu
and that the inconsistency is an accident of history. And that {lo}
= {da poi}.

I think that this is the likeliest to find a consensus.

> > though of the remainder
> > I don't know whether it is your first or your second choice that
> > they prefer. John & pc said they agree with you, but the status
> > of {nu} wasn't mentioned.
> I'm becoming convinced that the only way to have opaque contexts
> (without tu'a, xe'e, lo'e, etc) is to use {du'u}, and not {nu}.
> I suspect that with {nu} used for really-happen events, its
> quantifiers should always be exportable. And I believe that this
> may have to do with {du'u} meaning "x1 is what we make believe
> to be the proposition associated with the sentence:<bridi>". The
> quantifiers belong to the proposition, and so we can't take them
> out, but they don't belong to really happening events.

I'm glad, because this is very much what I thought until we embarked
on this thread. Since then I have concluded that at least some
opaques are best marked by {dahi} (e.g. stories about lo dahi 3-headed
cats). But {dahi} and {duhu} are the only two instruments for
handling opacity.

---
And