From cbmvax!uunet!cuvmb.bitnet!LOJBAN Thu Aug 13 21:17:24 1992
Return-Path: <cbmvax!uunet!cuvmb.bitnet!LOJBAN>
Date: Thu Aug 13 21:17:24 1992
Message-Id: <9208130628.AA24568@relay2.UU.NET>
Reply-To: cbmvax!uunet!oasis.icl.co.uk!I.Alexander.bra0122
Sender: Lojban list <cbmvax!uunet!pucc.princeton.edu!LOJBAN>
Comments:     <Parser> W: Field "Resent-To:/To:" duplicated. Last occurrence
              was retained.
From: cbmvax!uunet!oasis.icl.co.uk!I.Alexander.bra0122
Subject:      RE: proposals regarding abstractors
X-To:         shoulson@ctr.columbia.edu
To: John Cowan <cowan@snark.thyrsus.com>
Status: RO

Mark:
> 'Course, that's still a tanru, and I started thinking about how de-tanru
> it.  {le ni/ka sutra poi mi bajra sekai ke'a} (or maybe {la'u ke'a}, but I
> don't think that's as good) is a good start.  Any other suggestions?

The problem with picking a specific example is that you always
seem to find 57 better ways of expressing it.  Some more ideas:

        le ni sutra voi su'u mi bajra
        The amount-of speed which-is-described-as how I run

(I think {su'u} works quite well in this particular example,
though not necessarily in general) or

        [le] ni mi sutra lenu bajra
        [The] amount I am-quick-at-doing the-action (I) run

> (not sure which of {ni} and {ka} works better)

I think it depends on the context, which we haven't supplied here.
{lo ka sutra} is "quality-of speed", being fast, the opposite of
{lo ka masno}, whereas {lo ni sutra} is a quantity - so many miles per hour.

> The example that Iain had, {lexu'u nizmapti la cicac. xu'u tcika mi'o
> penmi}, points up a weakness in that there's that need for the {niz-} rafsi
> in {nizmapti}; I suspect that without better definition of the semantics,
> just about everything using it is going to have to have a {ni} in it
> somewhere, (and if not lujvo'd, it'll be an abstractor as part of an
> abstractor, which is a level of complexity we should avoid, if possible).

Yes, a lot of the things you want to abstract are _specific_ kinds
of amount.  I suppose the x2 place of the abstractor could be used
to specify this:
        le ni mi bajra kei be le ka sutra
        The amount-of (I run) on-the-scale-of the property speed
but this seems a bit disjointed, and reopens the question of
{ni} vs. {ka}.


John:
> Under Change 15, you do not need a separate cmavo:  "le du'u broda" is the
> claim that broda, and "le se du'u broda" is the assertion that broda.
> (Without Change 15, you need "le se ke du'u broda".)  We have never exploited
> the x2 places of the abstractions before, but it is now easy to do so.

Say again?  I'm afraid the distinction between "claim" and "assertion"
is lost on me.

> x1 is the predication [bridi] expressed in sentence x2

This isn't much clearer to me.
But from the way I've seen {du'u} used, {le du'u broda}
must be the "(putative) fact", whereas {le se du'u broda}
is the text, the combination of words.  Does this make
{le du'u broda} the same as {la'e le se du'u broda}?

And what does this mean for {ko'a cusku lu broda li'u}?

Iain.