[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
Nick:
> Lojbanists, thanks to Bob's crucial insight (lo'e is a predicate), I
> think I've got a compromise formulation that will satisfy everybody
My comments are meant in a similar spirit: I limit myself to what I
am unable to agree with.
I think we might do better if we considered establishing an agreed
set of gadri meanings, and then have the BF assign forms to them,
rather than struggle to find some formulation of the meaning of
lo'e/le'e that will cover everyone's views.
> lo'e is a predicate relating a singular entity, a population, and a
> means of singularisation. As is typical of Lojban predicates, when the
> value of the singularisation is not made explicit, it is deemed
> underspecified. This means that without further specification, lo'e
> brode may be Median, Mode, or Average; any statement involving lo'e
> broda will be true in all three cases. The decision on which
> singularisation applies is left to pragmatics (but see on defaults
> below), and must be interrogable by the interlocutor --- again, as is
> characteristic of Lojban predicates
>
> Since there is no provision in Lojban for places of gadri, the
> singularisation is specified by {pe veju'o} after {lo'e/e'e broda}.
> Possible values include: {mupli}: exemplary, ---- {stereotype},
> {klusi'o}: prototype, {kantymidju}: median, {cnano}: average,
> {cmikantymau}: mode
If lo'e/le'e are predicates, then so should the other gadri be.
lV'e views lo'i in such a way as to see all manifestations of
the membership of lo'i as the same individual, while lV and perhaps
lVi view lo'i in such a way as to see some manifestations of
the membership of lo'i as different individuals.
Thus {pe veju'o} must generalize to all gadri.
> The distinction between lo'e and le'e, given past keywords and the
> parallels in the lo/le paradigm, can be based on the following features:
>
> +/- veridical
> +/- specific (whether the population is all of broda or a specific
> subset}
> +/- objective
> +/- representative
>
> I want +/- objective to be the criterion, and I hesitate to consider
> prototype to be objective (or even veridical --- if veridicality even
> applies to this construct).
I could never be part of a consensus on this. The rest of the o-gadri
and e-gadri consistently pattern by the +/-specific criterion, and
this criterion makes perfect sense for all the different conceptions
of genericity that have been proposed. I do realize that you go on
to say:
> However, this setup for lo'e readily allows
> the objectivity and representativeness to be discerned by interrogating
> the singularisation --- but does not allow the specificity, the
> membership of the sample population, to be interrogated in the same
> way. Therefore And wins: the distinction between lo'e and le'e is one
> of specifity
... so I'm not disagreeing with your final proposal!
So I'm highly sympathetic to having gadri that distinguish +/-objective
and +/-representative, but these distinctions cross-classify
the +/-specific distinction encoded by the o-gadri/e-gadri distinction.
This said, I'm not convinced that +/-objective and +/-representative
would be the best features to capture the different models of
genericity.
Veridicality does apply to this construct, IMO. But this can wait
until further discussion.
> While officially lo'e and le'e are by default underspecified as to
> singularisation, I believe there are sensible defaults. IMO only,
>
> lo'e: +objective +representative
> le'e: -objective -representative
>
> This allows le'e to remain stereotypical, and lo'e to remain typical
>
> My own opinion is that the prototype is -objective +representative
> Maximum is +objective -representative
I don't understand this enough to comment.
> Defaults should be settable in advance in a discourse ("All my
> singularisations shall be prototypes".) Moreover, I would be
> sympathetic to exp.cmavo of the form, say, lo'e'V disambiguating
> between the various singularisations (say lo'e'a for prototype, lo'e'e
> for median.) But lo'e itself should be officially underspecified as to
> singularisation
I would much rather the BF create two or three pairs of gadri (including
lo'e/le'e). Unless I am mistaken, those of us who have a clear
conception of how lo'e/le'e should work have converged to one of
two models, one quasistatistical and the other prototypical or
'antipluralizing' (= loi'e/lei'e).
--And.