[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
Lojbab:
> At 12:23 PM 1/6/03 +0000, And Rosta wrote:
> >Lojbab:
> > > At 03:58 PM 1/5/03 +0000, And Rosta wrote:
> > > > > >If you drink a glass of water, then there is none left once you
> > > > > >have drunk it. What qualifies as 'none left' is determined by
> > > > > >ordinary criteria of relevance
> > > > >
> > > > > Correct, but LOGIC is orthogonal to pragmatics. When one is being
> > > > > hyperlogical, one is ignoring pragmatics. All means ALL, not almost
> > > > > all
> > > >
> > > >This is a warped notion of hyperlogicality. All means all, not almost
> > > >all -- that is a matter of logic/semantics. But whether "all of"
> > > >can describe something short of "every molecule of" is a matter of
> > > >pragmatics. This sort of confused inability to distinguish logic
> > > >and pragmatics has hampered Lojban for too long and led to silly
> > > >notions like the idea that logical precision adds complexity and
> > > >hence should add verbosity
> > >
> > > Backwards
> >
> >By what criteria? Natural language works as I describe
>
> Natural language isn't logical,
Many linguists (including me) think it is, and some (including me) think
logic is just an abstraction from natural language.
> and the logic breakdown seems to be
> especially egregious in precisely the areas that concern you these days
What you must mean is that natural language isn't logically unambiguous.
That is true.
> > > Logical precision is overly simplistic and what people want to
> > > say seldom matches what is easy to express logically
> >
> >I think you misunderstand what logic is. To take your example of
> >the drum with petrol in, there is no logical impediment to specifying
> >that no relevant amount of petrol is in the drum, or to specifying
> >what criteria determine a relevant amount. The impediment is the
> >verbosity that is required, and the excessive effort, in specifying
> >these things
>
> Being logically precise about the specification is verbose
In what sense is the precision *logical*? This is the crux.
> > > (hence what happens
> > > when we try to explicate "only" or "just")
> >
> >I don't know what you mean by these examples
>
> The logical expansion of English "only" is very long-winded, and I'm not
> sure we ever agreed on the logical expansion of English "just". Thus we
> added po'o as an abbreviation for the logical long-windedness
A good logical language is one that is not only logically precise but
also concise. It is positively desirable that lexical form be shorter
than and hence not homomorphous with logical form, but that lexical
form should map unambiguously to logical form. Hence it is not true
to say that the logical form is long-winded, because the language
designer can choose how long-winded the make the lexical form that
encodes it.
{po'o} is not an abbreviation for logical 'only': it is a vaguely-
defined UI that may also cover the ground of, say, 'merely'. It
is a vaguer alternative to logical 'only', not any sort of
abbreviation.
> > > Therefore we add short forms
> > > for what people want to say as abbreviations for logical formulations,
> > > leaving the pure logical forms there (though seldom used in
> > > conversation)
> >
> >I'm not sure what you mean by "abbreviation" versus "pure logical
> >form". If the abbreviation is equivalent to the unabbreviated
> >counterpart, then the abbreviation is as much a pure logical form
> >as the unabbreviated. If the abbreviation is not equivalent, then
> >in what sense is it an *abbreviation*?
>
> It may be a pure logical form on its own, but it hides possibly-key logical
> structure, which could cause problems when things start interacting
>
> For example, we have a logical expansion for po'o.
We don't. We know what logical 'only' means, but to the best of my
recollection {po'o} has never meant logical 'only'.
> I am not sure what
> happens to that meaning when within a negation. I am quite sure that
> people won't consider what happens to that meaning, until someone does the
> analysis and calls them on it
It is easy to say what happens to logical 'only' within negation.
We can't say what happens to {po'o}, because that has no logical
definition, and nor is there any general account of how UI that
might encode logical meaning interact with other logical elements
in the bridi.
> >In that case, i am all in favour
> >of vagueness as an option, but I object to the mistaken notion
> >that brevity requires vagueness and logical precision requires
> >verbosity
>
> So you reject the Cowan doctrine about the price of infinite precision
> being infinite verbosity?
No, not at all. I reject the widespread tacit belief that logical
precision is infinite. Logical precision is finite and be made as
concise as the language designers choose. So I reject the
erroneous belief that the Cowan doctrine applies to logical
precision. So on the one hand we have the deluded newbies, to whom
Cowan's dictum is addressed, who come to a logical language
misguidedly seeking general precision beyond what a natlang can
offer. And on the other hand we have the deluded oldtimers who
have taken Cowan's dictum to heart, but fail to understand that
it doesn't apply to logical precision (because of its very limited
and finite nature).
> > > Thus, people do NOT say "all" unless they really mean
> > > logical "all", without a single exception
> >
> >The words _every_ and _ro_ mean logical all, but when uttering a
> >sentence that contains them, the speaker may mean anything,
> >constrained only by Gricean principles (which rule out no
> >interpretation tout court)
>
> In Lojban, if ro doesn't always mean logical quantifier ro, then we'll have
> trouble with Lojban being a logical language. Grice might get
> communication to proceed anyway, but it won't be logical communication
ro always means logical quantifier ro. But sentences in general
(or ro in particular) is a tool used to get the message from
speaker to hearer; it is not the message itself. I have no idea
what 'logical communication' might be, as distinct from ordinary
communication. I like the idea of a logically unambiguous language
because it makes more feasible a protocol whereby the sentence
meaning is as close as possible to the message itself, but Cowan's
Doctrine means that in certain nonlogical respects there must
always be a gap between sentence and message.
> > > >The general opinion, which I share, is that the standard {pa} default
> > > >is useful for {da'a} (e.g. giving 'penultimate'), while {ro} is the
> > > >most sensible default for {me'i}
> > >
> > > It is not what is a "sensible" default that matters. "Default" refers to
> > > the value that pragmatically will occur most often when the number is used
> > > elliptically
> >
> >Can you think of cases where these word give different defaults?
> >Since {me'i ro} is the most sensible reading for {me'i mo'e zo'e}
> >(i.e. bare {me'i}, it stands to reason that usage unbiased by any
> >silly prescribed defaults would tend to interpret {me'i mo'e zo'e}
> >as {me'i ro}
>
> I haven't thought about it, and am not awake enough to do so now
>
> > "lo" is an operator on broda that produces a phrase "lo broda". The
> > > meaning of "lo broda" should be determined from "broda" in the same way
> > for
> > > all values of broda. It is not different if broda is nounlike, verblike,
> > > countable, always singular
> >
> >"lo xunre" is red, "lo blanu" is blue. Those are predicate-specific
> >properties -- they follow from the definitions of xunre and blanu
> >Likewise, if a predicate means "drink all of" or "touch some of",
> >then "all of" and "some of" are predicate-specific properties
>
> I don't believe we have any predicates with quantifiers built into their
> meanings. Not to say that we couldn't..
>
> > > >"pa lo re si'e" would be a continuous half, such as is found in a
> > > >tin of apricot halves
> > >
> > > Huh? "pa lo re si'e means nothing to me, since I have trouble
> > > contemplating a si'e greater than 1
> >
> >Read the ma'oste (entry for si'e) and you will understand. In learning
> >Lojban it is important that you should read the ma'oste
>
> I wrote it.
I know you wrote it. But you obviously had forgotten what you had
written. Read the entry for si'e and you will see that resi'e
means 1/2.
> Read the entry for piresi'e
That is the entry for piresi'e. If you read through the ma'oste you
will see that 1/2 can be said as resi'e or pimusi'e. You may opine
that the ma'oste is mabla, but at least you would know what resi'e
means, since ma'oste is baselined.
> pa lo re si'e is not an example of a si'e phrase, but is a sumti including
> the si'e phrase "resi'e"
>
> resi'e if meaningful would have the place structure
> x1 is a 2(-some) portion of mass x2
> lo resi'e refers to the x1 of that construct
> pa lo resi'e selects one out of however many such 2-some portions of the
> unstated x2 as may exist
>
> This is nonsense
resi'e = x1 is a 2th portion of mass x2
lo resi'e = at least one thing that is a countable half of something
That said, though, fractions are intrinsically in the penumbra of
countability, since they have a definite size even though they
don't have definite boundaries. So to talk about a continuous half
one would have to use a selbri that specifically encodes the notion
of continuity and/or intrinsic boundaries.
> > > >What are the default quantifiers for li, lo'i, le'i, lu, zo?
> > >
> > > Not a meaningful question for some of these, as there is no grammatical
> > > form "mu li tu'o"
> >
> >It's still a meaningful question, because a sentence has to mean
> >something -- it's not a meaningless string of words
>
> In Lojban, we do not presume that all legal strings of words have defined
> meaning
Are you saying that a default quantifier can be something that turns
the sentence into a legal but meaningless string of words?
> >So the question
> >is, when you interpret a sentence containing "li", what is by default
> >the quantifier on li? The answer, I would hope, is that it is unquantified
>
> John corrects me that we allow a quantifier on li. It is meaningless to
> say that it is "unquantified". If a quantifier can go there, then it is
> quantified. He says he specifies what the default quantifier is in CLL
So how do we talk about numbers without quantification? That is, how
do you talk about 2, rather than a 2 or every 2? That is certainly not
meaningless; it is absolutely normal, and it is rather irksome that
Lojban does not offer a ready means to do it.
> > > Since I've never actually seen an
> > > outside-quantifier usage on lo'i, I can't say which of these holds more
> > > meaning, which makes the decision rather arbitrary
> > >
> > > If John is right and we can use set descriptions for collections, the the
> > > answer may be different. I would be inclined to use "piro" on lo'i and ro
> > > on le'i, but if something other than this was said in CLL, I wouldn't
> > > likely argue
> >
> >Why would you not use {mo'e zi'o}? Why do you insist on quantification?
>
> I don't. But you can in fact put a quantifier there. So the question is
> what that quantifier is when it is ellipsized.
You don't have to assume it is elliptized. You could take the veiw that
the absence of an overt quantifier means the absence of a covert one.
Even if you had some objection to that, you could make the elliptized
quantifier mo'ezi'o.
> These sorts of arguments on
> default quantifiers came from just the same sorts of efforts by earlier
> jboskeists to try to assign logical meanings to strings that were
> grammatical but not logically explicit. Some of those logical meanings
> were useful ones, so that we can say "lo broda" without having to say "su'o
> lo broda".
This one is not a very good example, because we can say "su'o broda";
{lo} is a comparatively useless gadri. But "le broda" as short for
"ro le broda" is useful.
--And.