resolving the lo'ei debate? (was: RE: RE: Llamban

["And Rosta" <a.rosta@lycos.co.uk>]

xorxes:
> la and cusku di'e
> 
> >It's not clear to me that djica means "want x2 to happen". If it
> >does, then it conforms to my preferred model 
> 
> The gi'uste says x2 is an event/state, it doesn't say x1 wants
> it to _happen_, but what else could one want an even for?

Okay, but -- incredibly -- {nitcu} is not parallel, and I am not
sure I haven't seen {djica le plise}, etc.

Anyway, I'll accept that {djica} is as I would wish it to be.

> > > My preferred definition follows the
> > > {kalte} model, as you know
> >
> >I didn't know that. How would it apply to djica?
> 
> It would mean that you can want things as well as events 

But would {mi xorxesian zei djica le plise} mean {mi po'edji
le plise} or {mi djica tu'o du'u co'e le plise}? If the
former, then it doesn't make much sense to po'edji an
event.

But I guess you mean the latter. "xorxesian zei djica" would
mean "have desires about".

> > > tu'o ka ce'u goi ko'a ce'u goi ko'e zo'u
> > > ko'a po'edji ko'e
> > > cu du tu'o ka ce'u goi ko'a ce'u goi ko'e zo'u
> > > ko'a djica lo nu ko'a ponse ko'e zi'o kei zi'o
> >
> > > The new predicate takes an object in x2. Is this new predicate
> > > somehow ill-defined?
> >
> >I can't work out (I can't think it through) whether your definition
> >perforce excludes ko'e from being quantified within the nu bridi 
> 
> Yes, they are excluded. {po'edji} has no event to quantify within 

Okay.

> >If it does exclude it, then we exclude the very cases we're
> >interested in 
> 
> But they can be recovered through {lo'e} 

At least I understand now that this is your intention. More below.

> > > I can say {mi po'edji ta} for "I want that"
> >
> >sure
> >
> > > and {mi po'edji lo'e karce} for "I want a car" 
> >
> >If this is the logical conclusion of a chain of reasoning, then
> >I don't yet see it 
> 
> {mi po'edji lo'e karce} is {mi kairpo'edji tu'o ka ce'u karce},
> where {kairpo'edji} is parallel to {sisku} 

Okay. 

mi sisku tu'o ka ce'u broda
= mi troci tu'o du'u sisku co'e lo broda

So

mi po'edji lo'e karce
= mi kairpo'edji tu'o ka ce'u karce
= mi troci tu'o du'u (kair)po'edji co'e lo karce
= "I try to bring it about that there is a car that has the
sort of property associated with wanting-to-have"

That equation seems to me to fail. Maybe this one instead:

mi po'edji lo'e karce
= mi kairpo'edji tu'o ka ce'u karce
= mi (kair)po'edji co'e tu'o du'u (kair)po'edji co'e lo karce

? I can see how that will generalize to both {nitcu lo'ei tanxe}
and {tinbygau lo'ei cinfo}.

> >Sorry, but I'm not following you. The key problem with the
> >definition of po'edji is that it loses the world-shifting/intensional/
> >irrealis element provided by the embedded bridi in the djica
> >version. So I don't see how the definition can work 
> 
> That's where {lo'e} comes in 
> 
> >I can see
> >how "if x po'edji y then x djica lo nu x ponse y" will work 
> >But I don't see how the reverse -- "if x djica lo nu x ponse y
> >then x po'edji y" can work 
> 
> The reverse doesn't work with an individual argument. To get
> precisely that meaning I use {lo'e} 

je'e

> > > But I want to use {lo'e} with normal predicates, not with
> > > predicates that have been "fixed" 
> >
> >But most predicates don't need fixing, since most predicates
> >don't have these intensional contexts 
> 
> My point is that those predicates are not broken in the first
> place. Even if the gismu list is carefully purged of any
> predicates like "x1 needs object x2", they can easily be
> reintroduced as lujvo or fu'ivla and there is nothing at
> all wrong with them. It is useful to be able to say "I want
> a car" using the same lujvo that one would use for "I want
> that". Making every potentially intentional predicate
> of the gi'uste take an abstraction as argument still leaves
> infinitely many other predicates not in the gi'uste that take
> an object and that can make good use of {lo'e}. Predicates
> like "x1 wants object x2", "x1 needs object x2", "x1 looks
> for object x2", etc 

Okay. But in jboske we make use of different sorts of arguments,
including:

I. "We need for there to be a way to express X"
II. "Logic and other guiding principles entail that Y must be
a property of Lojban"
III. "It is useful for there to be a convenient way to express X"

Your case for {lo'ei} appears to be of Type III -- a utilitarian
argument. Which is fair enough, but it helps if we all recognize
this.

> >I understand that you want to use lo'ei more generally, but
> >I don't see how to extrapolate from sisku to other ordinary
> >predicates 
> 
> Pick one, let's say {dunda} 
> 
> Then {dunda lo'e xrula} means {kairdunda tu'o ka ce'u xrula},
> "giving flowers" 
> 
> This will normally be true together with {dunda lo xrula}, but
> not necessarily so. Logically they are different:
> 
> dunda lo'e xrula = kairdunda tu'o ka ce'u xrula
> dunda lo xrula = da poi xrula zo'u kairdunda tu'o ka ce'u du da

= dunda co'e tu'o du'u dunda co'e lo xrula

If you accept that, then I think we can all rest content with
the definition of {lo'ei}.

I guess the equivalences would be:

lo'ei broda cu brode
= brode co'e tu'o du'u lo cmima be lo'i broda cu brode co'e

le'ei broda cu brode
= brode co'e tu'o du'u lo cmima be le'i broda cu brode co'e

> And {dunda lo'e xrula} is orders of magnitude easier to
> manipulate. For example {la djan dunda lo'e xrula ca ro nu
> dy nelci lo ninmu}, "John gives flowers every time he likes
> a woman". To say it with {lo} you have to change the order,
> else he ends up giving always the same flowers. And it is
> really pointless to have some flowers as part of the
> relationship here, we are only interested in the predicate
> "gives flowers", not in the fact that there are some flowers
> that he gives. If we could use a brivla that means "x1 gives
> flowers" wouldn't we prefer it, and forget about all the
> quantifier issues?

I accept your case for {lo'ei}.

{la djan dunda lo'e xrula ca ro nu dy nelci lo ninmu}
= {la djan dunda co'e ca ro nu dy nelci lo ninmu kei tu'o du'u 
la djan dunda co'e ca ro nu dy nelci lo ninmu kei lo cmima
be lo'i xrula}

In effect the strategy is: When it is cumbersome and time-wasting
to spell out an explicit logical form, when most of that form
could be glorked from context, we can instead use a device,
{lo'ei}, that makes the logical form much vaguer and more
underspecified, but leaves the sentence much easier to say.

I have no problem with that. I think {lo'ei} and {le'ei} are
definitely worth having. (Not as the meanings of {lo'e} and
{le'e}, but that's a different discussion.)

> >I do know where you would use lo'ei, and I think it is the
> >case that
> > ko'a broda lo'ei brode
> >is variously equivalent to (I) or (II), (I) being the buska/
> >kalte-like case, and (II) being the more general case 
> >
> >I. ko'a xxxx zei broda tu'o du'u co'e lo brode
> >II. ko'a co'e tu'o du'u ko'a broda lo broda
> 
> I can't tell what difference you're trying to make 

nitcu lo'e tanxe
= nitcu co'e tu'o du'u nitcu co'e lo tanxe

tinbygau lo'e cinfo [the more general case]
= co'e tu'o du'u tinbygau lo cinfo

But I am now satisfied that these both fall under the general
equations I gave above, repeated here:

lo'ei broda cu brode
= brode co'e tu'o du'u lo cmima be lo'i broda cu brode co'e

le'ei broda cu brode
= brode co'e tu'o du'u lo cmima be le'i broda cu brode co'e

> >An example of (II) would be "man-eater", which, when not
> >meaning "citka lo remna", means "x has disposition such that
> >x citka lo remna", so wherever lo'ei can't be paraphrased by
> >lo there is some sort of intensional element lurking that
> >could be made explicit 
> 
> I would say that {citka lo'e remna} is {kaircitka tu'o ka
> ce'u remna}, where {kaircitka} is parallel to {sisku} 
> "eats things with property x2" without making a claim
> that there is a given thing such that x1 eats it 

Okay. But spelt out formally and explicitly, it means

kakne tu'o du'u citka lo remna

The truthconditions of lo'ei or kair- will always reduce to
"tu'o du'u lo".

> >Would (II) not suffice as a definition of lo'ei?
> 
> I'm not sure what you mean by (II). How do you deal with
> another sumti in x3 for example. It is not the disposition
> of x1 in particular that is relevant. It is a relationship
> among several arguments, one of which is replaced by a
> property so that the original argument in that position
> enters only indirectly 
> 
> >Yes. Wanting/needing/seeking objects is experientially more
> >basic but logically less basic than wanting/needing/seeking
> >events 
> 
> Even if it is logically less basic, the relationships are
> just as valid, and {lo'e} is useful to deal with those
> relationships 

Agreed. 

> >Whichever set get the gismu and whichever get the lujvo,
> >though, objects quantified within the intensional context
> >can only be rendered by means of the the version with the
> >event x2 
> 
> Or with my {lo'e} in the version with object x2 

Agreed.

> > > tu'o ka ce'u goi ko'a ce'u goi ko'e zo'u
> > > ko'a buska ko'e
> > > cu du tu'o ka ce'u goi ko'a ce'u goi ko'e zo'u
> > > ko'a troci tu'o du'u co'e ko'e
> > >
> > > The important thing to notice is that there is no "intensional
> > > context" passed on to {buska}
> >
> >I have certainly noticed this, and, as I say above, this is the
> >stumbling block in my understanding. How can
> >
> > [troci [Ex [co'e x]]]
> >
> >translate into
> >
> > [buska x]
> >?
> 
> It doesn't. You need {buska lo'e ...} to get the first meaning 
> 
> >-- The variable is unbound. Okay, we allow unbound variables
> >-- ce'u is such a one -- but since
> >
> > [troci [Ex [co'e x]]]
> > [troci [Ax [co'e x]]]
> >
> >mean different things, how can they each mean the same as
> >[buska x]?
> 
> Neither of them means [buska x]. The Ex case is
> {buska lo'e broda}. I hadn't thought before how to get
> the Ax case. It seems difficult to get in terms of
> buska 

It would be overkill to go for the Ax case, I think. The
essential rationale for lo'ei is that you trade vagueness/
inexplicitness for convenience. Once you start wanting
something more explicit than ordinary lo'ei gives you,
the obvious thing is just to be explicit and not use lo'ei.

--And.