Date: Tue, 14 Feb 1995 23:23:47 +0000
From: ucleaar <ucleaar@UCL.AC.UK>
Subject: Re: replies re. ka & mamta be ma
In-reply-to: (Your message of Mon, 13 Feb 95 20:06:24 EST.)
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
To: Veijo Vilva <veion@XIRON.PC.HELSINKI.FI>
Reply-to: ucleaar <ucleaar@UCL.AC.UK>
Message-id: <01HN410GDPRA000OW3@FIPORT.FUNET.FI>
Content-transfer-encoding: 7BIT
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Content-Length: 8822
Lines: 183

Jorge:
> So you are saying that in order to model cognition, a word for "1" is
> indispensable. You wouldn't accept a description like {le namcu pe
> le solri}. What makes the description a worse choice to express that
> idea? (Other than the length of it, which could even be shortened with
> an appropriate lujvo.)
 
I'm suggesting that if there isn't a single primitive word for 1,
and if lg (iconically) models cognition, then such a lg implies
that 1 is not a basic concept. Furthermore, the sense of {le namcu
pe le solri} is not 1; it just happens to refer to 1. But I concede
that the sense of {namcu zei solri} could be 1.
 
> > It is legitimate to hope for a reasonable match between one's
> > world view and that modelled by the language one is using.
> Does this mean that there are ideas expressable in some language
> and unexpressable in another?
 
I believe so, though not very strongly, & don't want to argue the
point. What I would claim is that if 2 lgs can express the same idea,
but 1 lg takes 1 word to do so & the other takes 10000 words, then
the pragmatic difference between the two expressions is so great that
they don't feel like the same idea: one feels simple & the other
complicated. Ideally there is a fairly iconic relationship between
sentences and mental representations of their meaning.
This is speculative.
 
> > For example, if Lojban had a gismu such that its x1 are
> > either feet or alphabets, I would find it strange that (i)
> > feet & alphabets constitute an intuitive category, and (ii)
> > this category is relatively useful or basic (as gismu-hood
> > implies).
> You mean homonyms? I would find it strange too, but I don't see how
> that would affect my world view (whatever that is).
 
Not homonyms. One unambiguous word, such that if x is a broda then
x is an alphabet or x is a foot. I'm not claiming this would *affect*
my world view; I am claiming it would not *reflect* my world view,
and I wd rather it did.
 
> I would argue that {kau} is in the same boat with {pa}. Your
> argument against works just as well for {pa}:
> > Here is my case against: Given a sentence S,
> > from S can be derived a proposition, P, which constitutes part
> > of the meaning of S. P can be represented using pred calc, & pred
> > calc, in matters such as scope, quantification, variables, etc.,
> > is a reasonable model of P as it is in our minds. {kau} (in its
> > "indirect interrogative" use is merely a syntactic variant of
> > {kau}-less circumlocution, since both variants come out the same
> > when the S is translated into P done in prec calc.
> Replace the last sentence with:
> {pa} is merely a syntactic variant of a {pa}-less circumlocution,
> since both variants come out the same when the S is translated into P
> done in pred calc.
 
If you're thinking of {le namcu pe le solri}, this is not true. "1"
refers to 1 whatever the state of the universe is. {le n pe le s}
refers to 1 only if there is only one sun - in a world with two suns
it wd not refer to 1. Moreover, I don't see how or why {kau} is in
the same boat.
 
> > I'd been assuming we were talking about syntax/grammar. If you
> > have two alternative syntactic devices with same meaning then one
> > is redundant (e.g. "X kissed Y"/"X was kissed by" "long book"/
> > "book that is long").
> Those are simple cases, and even then, they are not really "syntactic
> devices with the same meaning". If you replace X with "everybody" and
> Y with "someone" you get two different meanings.
 
That they are simple cases doesn't affect my point. It even strengthens
it. As for whether passive changes meaning, there are other reasons
beyond the one you cite for thinking that it does affect meaning, but
they are subtle so I ignored them for the purpose of using it as an
example. As for the change of meaning you mention, it results not from
the passive construction but from word order. You can see this from
the following sentences, where favoured Q-scope follows word order &
is oblivious to whether the verb is passive or active.
   Everywhere was to be found a poster.
   Somewhere was to be found every poster.
   Into every room ran a mouse.
   Into a room ran every mouse.
 
> In any case, that doesn't show that you've reduced the ways of expressing
> an idea to one. Suppose you make one of "long book" and "book that is long"
> ungrammatical. There is still "book of great length", and who knows how
> many others.
 
Junk those too, till only "book that is long" remains.
 
> If the language is really a language for human comunication,
> then there will be many ways, and the more complex the idea, the more ways
> there will probably be to express it.
 
How do you mean there *will* be many ways? I'm sure that what you say
is true of any lg that ever gets used much. I don't think my hypothetical
lg would be practical or convenient. Just adequate for expressing
everything.
 
> Even if for such simple constructions
> you succeed in making only one way grammatical, as you increase the
> complexity of the concept, surely there will be many ways to say the same
> thing.
 
But not necessarily employing different inventories of constructions.
 
> > I'm saying that in principle you can restrict the grammar-lexicon
> > so that identical expressions of a given idea invoke the same set
> > of rules of grammar-lexicon.
> What do you mean by "identical expressions"? If they are truly identical,
> then obviously they invoke the same rules.
 
With identical truth-conditions.
 
> > I am assuming that sentences "expressing the same
> > idea" are truth-conditionally equivalent.
> Do you consider {ko'a zmadu ko'e} and {ko'e mleca ko'a} truth-conditionally
> equivalent?
 
In principle, they could be. In practice, it depends on exactly what
{zmadu} and {mleca} end up meaning when the lg gets known properly.
 
> Do you allow semantics to determine truth-conditionality?
 
I'm not sure I understand. I'd have thought it is precisely
semantics, & only semantics, that determines truth-conditionality.
 
> > > I'm not sure whether you are
> > > imposing the condition that humans should be able to handle it.
> > I would impose this condition, but under the heading of "necessary
> > conveniences" rather than "necessary for expressive power".
> Ok, but {kau} and {pa} both fall in that category.
 
{pa} also counts as "desirable on grounds of fidelity to cognition",
as discussed above. I am far from persuaded that {kau} counts as
desirable in that way; I'm inclined to think otherwise. And {Q-kau}
is a convenience, agreed, but its necessity is not blindingly
obvious - it saves a word or two and avoids the need for forethought
quantification.
 
> > Going back to {Q-kau}, I don't believe we understand it by magic:
> > rather we unconsciously translate it into a logical structure (however
> > difficult it may be to fathom consciously) and that constitutes our
> > understanding of Q-kau.
> That may or may not be the underlying mechanism of understanding,
> but how do you know that it is so for {kau} and not for {pa}?
 
I don't know it; I believe it. It's very hard to prove. That's why
semantics (that assumes psychological reality) is so difficult. On
the whole, progress seems to be made chiefly by emerging consensus
on shared intuitions. It's easy to be skeptical about that.
 
My reasons would be that singularity is an easy notion to grasp,
and difficult to break down into more primitive concepts (yes,
you can define 1 as 2/2, or 3-2, but this is circular, not
simplificatory), whereas Q-kau can be broken down  into more
primitive concepts (in ways we've previously discussed).
 
> > I'll demonstrate my thesis: Most kinds of attributive adjective are
> > semantically equivalent to a construction where the adj is used
> > predicatively in a relative clause - long book/book that is long.
> > These types of attrib adj could be got rid of, and we would still
> > be able to express the idea of long books - by "books that are long".
> That only shows that you can dispense with one syntactic construction.
> (Overlooking the use of "most".) That doesn't mean that you are left with
> only one way of expressing attributes in English.
 
You give me the list of constructions used for expressing attributes
& I'll cross off all but one, & use the surviving construction in the
stead of the deleted.
 
> Dismissing {kau} affects the grammar because [...]
> (b) the lexeme-specific
> and complicated rules for deriving semantic structure from syntactic
> structure containg {kau} could be scrapped.
> As for (b), if scrapping the semantico-syntactic rules relevant to {kau}
> simplifies the grammar, so would doing the same thing with respect to {pa}.
 
There aren't any semantico-syntactic rules specific to {pa}, apart
from the one teenyweeny rule that says {pa} means "1".
 
---
And