Re: On {lo} and existence

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

Jorge:
> > > What is a proposition? Is it derived from a sentence or from an
> > > utterance?
> > From an utterance. Because of things like deictics and +specific
> > references, many sentences will not yield complete propositions.

To clarify, it is has to be from an utterance, because deictics refer
to utterances.

> Ok, now to probe further. Does every utterance (use of a well formed
> sentence) determine a unique proposition? Or are there other factors
> that intervene?

Every utterance not containing a +specific reference determines a
unique proposition. Utterances containing a +specific reference
determine a unique but incomplete proposition. (I may have been
wrong about deictics as a source of incompleteness.) But see also below
re empty tersumti.

Propositions not determined by the grammar can be derived from
these complete or incomplete propositions that are determined by
the grammar.

> What kind of proposition does something like {ai mi klama} determine?

I'm not sure whether the focus of your question is on {ai} or on {mi},
or on the empty tersumti of {klama}.
(i) {mi}: Assuming deictics are -specific, we have:
       Ex utter(x,this-utterance)
(ii) empty tersumti: since these are not necessarily -specific, the
proposition determined by the utterance is incomplete.
(iii) {ai}: I think this means that once we have decided how to fill
the empty places of the predicate Klama, the resulting predication is
claimed to be true of the world of your intentions:
   true(proposition345,world-of-Jorge's-intentions,1)
And this predication is in turn claimed to be true of the universe
of discourse.

> > But note that the proposition derived by grammatical rules (augmented
> > by reference assignment) from an utterance is not necessarily the only
> > proposition communicated. This is especially obvious with irony.
> Does the meaning of predicates enter into the grammatical rules?
> If I remember correctly, you consider the equivalence or not of
> for example {mleca} and {se zmadu} to be determined grammatically,
> so the meaning of predicates would have to be a part of the grammar
> rules.

Yes.

> Or are the rules independent of meaning, so that we only need to
> know the selmaho of each word to determine the proposition?

No.

> If meaning enters, why couldn't a more sophisticated grammatical
> analysis detect irony as well?

Because irony isn't determined by grammatical rules. Grammatical
analysis detects that which grammatical rules have determined.

> > > If, on the other hand, it is derived from an utterance, the context is
> > > already determined, and then its truth value is also determined. Then
> > > there is no crime in saying that an utterance has a truth value, it is
> > > well defined, it's the truth value of the proposition associated with it.
> > First, even if we have determined the context, and can therefore
> > establish the x3 of true(proposition3234,world2,?), the resulting
> > truth value doesn't inhere to proposition3234.
> It doesn't matter. The utterance still has a well defined truth value,
> even if it is not inherent to its proposition, it is inherent to the
> utterance.

I still don't think this is so. If the grammatically-determined meaning
doesn't include a specification of which world fills x2 of True, then
the speaker(s) and every hearer of the utterance are free to pick
a different world as x2, with resulting variation in what the truth
value in x3 will be.

> > Second, the context is not determined. The comprehender recursively
> > expands the context until a relevant interpretation of the utterance
> > is found.
> At which point, we have our proposition. Then context is determined
> by the comprehender.

But there can be many comprehenders of a single utterance. They can
disagree.

> The truth value can only be assigned to the relevant interpretation,
> I suppose. When we talk of *the* truth value of a given proposition
> in a given world, we mean the one assigned to the relevant
> interpretation of the "typical" comprehender. Or what?
> We need a context before we can assign a truth value.

I'll discuss the following example:

   A: Do you want a cigarette?
   B: I don't smoke.

Step 1: Grammatical rules will derive from B's utterance the incomplete
proposition "not smoke(B,?)" [let's assume "omitted sumti" in English
work as in Lojban].
Step 2: The incomplete proposition must be completed by anyone
comprehending it. E.g. "not Ex smoke(B,x)".
Step 3: The comprehender decides which world to associate the completed
proposition to. E.g. R, the real world.
Step 4: The comprehender draws inferences until some proposition
relevant to the context is hit upon. E.g. "If B doesn't smoke then
B won't want a cigarette. B doesn't smoke. Therefore B doesn't want
a cigarette." "B doesn't want a cigarette" is the relevant proposition.

These steps form a logical though not a procedural sequence.

Assignment of a truth value can come only after Step 3. In principle
the truth value a proposition gets is an objective matter, independent
of the comprehender or of further interpretations made in Step 4.

> > Thus, whoever interprets an utterance chooses their own
> > context. Two people hearing the same utterance may choose different
> > contexts.
> I agree, and they may therefore end up giving it a different truth
> value. *The* truth value, if there is such a thing, would have to
> be a pondered average of those given by each comprehender.

They don't give utterances truth values, because utterances don't
have truth values. Propositions have truth values, but that doesn't
mean that entertaining a proposition necessitates computing its
truth value.

I think we can talk about *the* truth value of a proposition if this
value is an infinite set of pairs matching worlds to values between 0
and 1.
By this definition of truth value, we could then say that only Step 2
can result in two people hearing the same utterance coming up with
different completed propositions which may then have different truth
values. So if Step 2 can be skipped, if the utterance contains
only overt -specific sumti, then it becomes possible to associate
such an utterance with a truth value.

> > Fine. This is what I've been advocating.
> >     {nu} without explicit {dahinai} is associated with implicit {dahi}.
> >     All other predicates without explicit {dahi} have implicit {dahinai}
> > {dahinai} = real. {dahi} = real or imaginary. "Real" means "real in the
> > universe of discourse". "Imaginary" means "not real in the universe of
> > discourse".
> I think I see now why you say that {nu} would be an exception. You want
>        ro da'inai nu broda cu fasnu
> to be necessarily true. I don't. If {nu} is only a potential event,
> then {da'inai nu} is still only a potential event. In your language,
> I would say it is a potential event in this world, while presumably
> {da'i nu} would be a potential event in some other world.
> My position now is that I would like that {ro nu broda cu fasnu}.
> Otherwise, if it is decided that {ro nu broda na fasnu}, then

= {naku ro nu broda cu fasnu}, yes?

> that should be equivalent to {ro da'inai nu broda na fasnu}, so that
> there is no exception in the use of {da'i}.

You understand me right. I won't accept your view, because a potential
event is indistinguishable from a dahi event. If {nu} is inherently
{dahi}, then using an overt {dahinai} will not override the inherent
{dahi}, thus ruling out a way to restrict {nu} only to actually
happening events.

---
And