Re: *old response to And on fuzzy proposals

[ucleaar <ucleaar@UCL.AC.UK>]

Lojbab:
> And on fuzzy logic proposals:
> >A. The proposals are to express 4 things:
> >(1a) degrees of {jaa} & degrees of {na}
> >(1b) degrees of {jea} & degrees of {nae}
> >(2a) degrees intermediate between {jaa} & {na}
> >(2b) degrees intermediate between {jea} & {nae}
> >B. The current proposals for ways of expressing them are as follows:
> >(1a) jaa + CAI & na + CAI
> >(1b) jea + CAI & nae + CAI
> >(2a) i.  number expression + xoi
> >     ii. jaa + xi + number expression
> >(2b) i.  number expression + xio
> >     ii. jea + xi + number expression
> >Of these, only (2a/b.ii) require no change to the syntactic portion of
> >the grammar; they're ugly quick fix solutions that (I think) John is to
> >adopt in the refgrammar, in order to avoid a grammar change (in the
> >lojbo sense of "grammar change").
> To which I'll add
>    sei li {quantifier} ninjetnu [se'u]
> which can be attached to ja'a, je'a, na, na'e, and a bunch of other
> things as needed, requires no grammar change, but does require agreement
> on the appropriate lujvo (which need no be ninjetnu).

... and has completely the wrong semantics, so is not what is wanted at
all. {sei} adds metalinguistic comment. It does not override semantics
of jaa, nae or whatever. It does not work in subordinate bridi.

> Being fond of cu'o, which I certainly expected/intended to have
> application to fuzzy logic expression,

then how come the maoste says it does probabilities? Isn't probability
to do with degrees of certainty, while fuzz is to do with degrees of
truth or of appropriacy of categorization?

> this could also be done with the shorter though more elliptical "sei
> {quantifier} cu'o [se'u]"

No, for same reasons as above.

> Then there is the indicator/discursive je'u which has only 3 degrees of
> positive or negative strength - not as flexible as what Steven has in
> mind, but certainly useful for many fuzzy expressions.

But not even a decent partial solution, for same reasons as above.

> Somewhat less flexible than a free modifier, but more flexible than the
> above proposals would be to use the quantity modal at the bridi level:
> "sela'i li {quantifier}" for the value on the scale, possibly with a
> fraction slash and scale maximum.

I don't understand. {sela'i}?

> NONE of these require new grammar or cmavo.  If they are not ideal for
> some reason, they are still at least as expressive and flexible as any
> of the proposals, and barely more wordy.

They (the ones I understand) are of no use.

> >But with NAhE you can predict what the modified selbri will mean,
> >whereas with tanru you can't, or at least not nearly as much.
> Not really.  You do not know what scale is implied when NAhE is used,
> and the vaguenesses that are possible (as opposed to plausible) are as
> nebulous as tanru.  For example does "na'e dzukla" mean that the x1 is
> running, or that he is walking-in-place, or that he is performing some
> other action involving all expressed sumti which contrasts in some
> "other" way with "walk-going" (maybe he is telephoning the
> "destinations", and his "fingers are walking through the "Yellow Pages"
> - a reference to American commercials for business phone listings - and
> the contrast with actual "walking" seems cutely apropo).
> Any of these could be implied by na'e.

But at least, I think, one can be sure that {lo,i nae broda} is not a
subset of {lo,i broda}. That is something. And that is why it is apter
to do fuzzy membership by something nae-like thhan something brivla-like
as part of a tanru.

> >I'm willing to accept that something in MOI (I guess it would be glossed
> >"extent selbri" or "degree selbri") *might* do for (1b) and (2b), but
> >this won't do for (1a) & (2a), because (1a) & (2a) the require something
> >with the syntax of NA.  As for whether {cuo} would do for (1b) & (2b), I
> >reject the notion that degrees of truth are degrees of probability.
> Since there is NO usage history for "cu'o" and I put it in based on my,
> umm, less than thorough understanding of what might be needed for fuzzy
> logic, we can define what "cu'o" means by what is is useful for.  If a
> quantifier+MOI construct would be useful in a fuzzy logic context, I
> wouldn't have any qualms with making it clearer that cu'o was intended
> for just that context.

I now think we need something nae-like, not a brivla. MOI is just an
abbreviation for brivla - e.g. {x1 x2 3 MOI} = {x1 x2 li 3 BRIVLA}.

> Steven answered among other things with the problem:
> >"The "thing" is fuzzily 2 of 5 on our cow - oak tree scale."
> >"On a 0 to 5 fuzzy ordinal scale where 0 is a cow and 5 is an oak tree,
> >the thing is a 2 out of 5."  (That is, the thing is more cowish than oak
> >treeish, but it has good helpings of both).
> And this could be done using intervals and my use of cu'o as:
> le tarmi  cu bakni bi'o sei refi'umucu'o se'u cindu

Ignore the {fiu}, which isn't quite right. Ignore {sei ... seu}, as
it's metalinguistic. That leaves {le tarmi cu bakni bi'o cindu}. At
least that provides a neat way to do weird belknapian scales. (But can
{bio} conjoin selbri?)

coo, mie and