Re: Logical?

Mike S.
> And Rosta <a-rosta@...> wrote:
>
> >Mike:
>
> >> A conlang is logical if all well-formed statements (though
> >> not questions and commands) are logical expressions.
> >
> >I don't think you need to exclude questions and commands. Rather,
> >you need to enrich the vocabulary of logical expressions with a
> >set of illocutionary operators.
>
> In other words, a command like "Have a slice as pizza" could
> be interpreted as "I am inviting you to have a slice a pizza".
> The latter could be true or false depending on whether the
> speaker was making a sincere invitation.  Is this what you
> have in mind?  If so, then *I* agree with you that every
> sentence in a language can be made logical.
It's almost what I have in mind. I think that all sentences
involve an illocutionary operator that encode what the sentence
is (statement, question, command), but they have the status of
'conventional implicature', i.e. they are linguistically encoded
but outside the truth conditions of the sentence. It is the
propositional content of the sentence that has truth conditions.

It follows from that view either that sentences themselves are
not true or false or that, say, "Go!" is true iff the addressee
goes, and "Did she go?" is true iff she did go.

> I made the weaker claim in anticipation that someone might
> not accept that view.
I have noticed how adeptly emollient an interlocutor you are.

> >> An expression is logical if it evaluates to either true or
> >> false, but not both.  In plainer language, every expression
> >> in a loglang should *unambiguously* convey to the listener
> >> an idea about the way the world would have to be in order
> >> for the expression to be true.
> >
> >This is too high and too inappropriate demand. For example,
> >the truth of a sentence containing a referential expression
> >cannot be determined until the referent is determined, but
> >the language itself does not determine the referent; when
> >I say "I saw him/the man score", there is nothing in the
> >sentence -- in the linguistically encoded/determined meaning
> >-- that tells you "him/the man" refers to David Beckham.
> >Rather, the reference is determined pragmatically, and the
> >sentence encodes an *incomplete/underspecified* logical
> >formula.
>
> This gave me some food for thought.  There are two
> possibilities here.
>
> If "him/the man" is an anaphor for David Beckham, previously
> established in discourse, then we have a bound variable, and
> the sentence is totally logical.  I know this is not what
> you had in mind, but I mention it to be thorough.
This is true only if your loglang grammar defines binding
relations over the domain of the entire discourse and not
just the current sentence. For English, prior textual
context has the same grammatically-invisible status as
the rest of the context. (Indeed, for English and probably
all natlangs, referential phrases must be unbound.)

> If "him/the man" is being newly introduced into the discourse,
> we have what *appears* to be a free variable.  However,
> I think this only holds if one insists there is no discernable
> existential proposition inherent in the statement.  I am
> inclined to argue that there is.  What we have is:
>
> Ex [(x is a man) & (I saw x score)]
>
> Whether or not the speaker has a much more specific referent
> in mind than a simple existential does not, IMO, degrade
> the logicality of the statement as given.  The speaker might
> simply choose not to mention that data, just as he may
> choose not to mention what color shirt he was wearing.
> IMO, the sentence has provided enough information to convey
> to the listener an *idea* about the way the world would
> have to be in order for the expression to be true, or false.
But English, Lojban (le/lei) and Livagian, to cite just three
estimable languages, allow for what are called 'specific' or
'referential' expressions', where the encoded meaning is
precisely that the truth conditions of the sentence cannot be
determined until the referent has been identified. Thus I may
go into a bookshop and say "I'm looking for a (certain) book"
and be claiming not "Ex x is a book I'm looking for" but "I'm
looking for War & Peace".

> Incidentally, one might claim a non-veridical interpretation
> for "the man" (as Lojban might); nevertheless, the sentence
> implies that a score took place, and that "I" saw the x1 that
> did it;  IMO that is still satisfactory as a logical statement.
But "mi cilre le cukta" ("I read a certain book") is not
necessarily true if I read _A la recherche du temps perdu_;
it's true only if I read the referent of "le cukta".

> BTW, I think that Lojban descriptors are its greatest feature;
> they turn examples like the one you give into explicitly
> logical propositions.
They're an improvement on English, but I prefer the 134 Livagian
determiners...

[...]
> >As for 'unambiguous pragmatics', much as I attribute it to
> >Livagian culture, it would be a cultural rather than a
> >strictly linguistic phenomenon, since a fully logical
> >language may nevertheless be used illogically.
>
> On reflection, I might have overstated the need for unambiguous
> pragmatics.  To use a not-so-great example, if the English
> phrase "kick the bucket" is so commonplace for "die" that
> it is essentially rendered a lexical entry, then of course
> the speaker and listener will have an understanding and
> logicality is served.  This seems perilous though.  How
> do you express veridically when a bucket really is kicked?
And more generally, if you don't say what you mean, and
trust your addressee to infer what you mean from what
you say, how can you ever guarantee that your addressee
correctly understands you?

So yes, to get full mileage from a logical language you do
need a 'logical pragmatics' (a la Commander Data, Seven
of Nine, Spock), but my point was that this logical pragmatics
is a property not of the language but of the culture of
the community using the language. It is beyond the reach
of loglang design.

--And.