RE: [lojban] RE: set of answers.

["And Rosta" <a.rosta@dtn.ntl.com>]

pc:
> OK, so suppose we get away from sets and stick with the predicates we
> actually have.  I take it that a direct question, {ma broda}, is covertly of
> the form {ko xusra lo du'u makau broda}.  Suppose the directee says {la b
> broda}.  Has he answered the question,

To me it seems as though there are two aspects to answers that need to
be disentangled. On the one hand, an answer -- call it an 'illocutionary
answer' is any information that is as relevant as the answerer's knowledge
allows. On the other hand, an answer -- call it a 'logical answer' --
is a specification of the extension of a category (or so I think).
Sometimes a non-l-answer can nevertheless be an i-answer, e.g. "lo ninmu
(cu klama)" as an answer to "ma klama", in a context where, say, the
answer has no more relevant information, or where this information is
sufficient to satisfy the questioner's needs.

But my feeling is that interrogatives and qkau involve only logical
answers -- illocutionary answers are a red herring.

> i.e., is it the case that {la'e lu la
> b broda li'u du'u makau broda}, that is  is
> {le (better {to'u}) du'u la b broda cu du'u makau broda}?

I don't accept that lo'i du'u ma kau broda is the set of answers.

> The answer to this
> is not transparent.  As noted, it often fails to be the case (though not
> always by any means) that {da broda} and {noda broda} fail, and {lo broda cu
> broda} almost always fails.  In a given case, others may also fail (stones if
> the questions supposes an agent, long dead folks if the questions supposes a
> contemporary, and so on).
> But now at least, like And, I have {makau} universal -- though only the ones
> that actually fit the property are significant.  Note that this is still not
> {ce'u} for the property in question still has {makau} in its description, is
> still a property of expressions, not of things yet, unlike the {ce'u} cases
> (so far at least).
> In at least some cases we can carry out the elimination of indirect questions
> pretty thoroughly:  {la dubias frica la tclsys le du'u maka mamta ce'u}
> amounts to (by extensionality) {da zo'u le (or {to'u}) du'u da mamta la dubia
> cu frica le du'u da mamta la tclsys le ka ceu jetnu} which means {da zo'u
> gonai da mamta la dubias gi da mamta la tclsys} which amounts eventually to
> just {le mamta be la dubias na du le mamta be la tclsys} from which (euclid's
> law) it follows in fact that {la dubias na du la tclsys}.  Other cases behave
> similarly.

Of course this way of eliminating qkau is both obvious and correct, but I don't
think that for our purposes it counts as eliminating qkau, for the same reason
as other extensional formulations fail.

I think I am now able to offer a halfway decent analysis:

no da ro de poi ke'a cmima la dybiyb ce la tcelsik [-- or cmima of whatever
class of differers --] zo'u
da -extension-of tu'odu'u ce'u mamta de

= D frica C tu'odu'u ma kau mamta ce'u
= Dubya and Chelsea differ in who their mothers are

Now that can be done more simply as:

no da ro de poi ke'a cmima la dybiyb ce la tcelsik zo'u da mamta de

or indeed

no da mamta ge la dybiyb gi la tcelsik

But the longerwinded method comes into its own in cases like:

  X and Y differ in who gave them what
= ... frica tu'odu'u ma kau dunda ma kau ce'u
= ... da -extension of tu'odu'u ce'u dunda ce'u de

Admittedly, this "halfway decent analysis" does not use {frica}, but there
was no guarantee that {frica} is logically sound, and hence no guarantee
that frica could be used in a logically explicit formulation.

> Roughly, to take on the final case, {roda zo'u ganai da nu makau se citka fau
> le raljysanmi  gi ge de nu makau nenri le lenkytanxe gi de rodytcini da}

I won't comment on this, as, not being qkauless, it is part of an analytical
programme that I'm not participating in.

--And.