methods of resolving mismatches between place structures and number of overt sumti
> Jorge Llamb?as scripsit:
>
> > I would very much like to do a
> rationalization of place
> > structures. Unfortunately (or fortunately
> rather) it is
> > not only up to me.
>
> There is certainly no reason why you can't
> create a proposal with
> place structure changes and the reasons for
> them under the byfy
> aegis.
>
> > That's one of the reasons I proposed at some
> point to
> > re-interpret blank in the prescription as
> {zi'o} rather than
> > as {zo'e}, to bring the prescription more in
> line with usage.
>
> This evades the issue, though, because it
> brings in novel semantics
> under the table. If you want new semantics for
> gismu, figure
> out what they should be, rather than claiming
> that "da broda"
> and "da broda de" have different possible
> values for da. Doing
> that splits each n-place gismu into 2^n-1 new
> predicates, only
> one of which has a formal definition. (It also
> forces observatives
> to be treated as even more of a special case
> than they already are.)
>
To summarize the situation "briefly." Many
(most?) predicates have places that are often
(usually?, always?) left unfilled in use. This
includes use in forming lujvo, where some places
never appear in the resulting new predicate.
>From this it might be inferred that such places
are not really essential to the concepts intended
for the predicates — or, rather, that the
concepts that should have been intended for the
predicates were ones that did not involve these
places. This latter claim is reinforced by (and
reinforces) notions about what are basic and
useful concepts to have, in light of which it is
possible to say that some predicates are for
concpts that are narrower, less basic, than other
concepts closely related to the given ones and
not represented (typically generic concpts to
which the given are specific).
The straightforward response to this perceived
problem would be to propose redefinition of some
(probably, if everyone got into the act,
virtually all) predicates, eliminating
"unnecessary" places and moving to more general
terms where "appropriate." For a variety of
reasons, this has not seemed to be the way to
proceed (it involves a lot of work on each narrow
individual case and so is much harder than making
sweeping changes in broad general areas). So the
discussion has turned to finding some way to
achieve the desired results within the existing
defintions of predicates. The obvious solution
-- within the existing frame — for solving the
"extra" place problem is to use {zi'o}, which is
designed for just that purpose. However, since
the "bloated" predicates are so numerous and the
"extra" places even more so, this would result in
an overabundance of {zi'o}s in the langauge,
several in almost every sentence — at least an
esthetic (and practical) disaster. So the
suggestion comes along to make the omission of a
sumti in a place the equivalent of the occurrence
of {zi'o}. (There appears to be another
suggestion in the &-n discussion, but I can't
figure it out.)
Unfortunately, omission of sumti is already a
common occurrence (again in virtually every
sentence — this is what started the process) and
it is clear that it does not always mean {zi'o}
-- although it is equally clear that it often
does or at least might well. So, to change
omissions to {zi'o} the suggestion arises that
other omissions should instead be replaced by
overt (unclear) {zo'e} or some other appropriate
marker (the only one currently available is the
vague {zu'i} but more could be added if need be).
This, of course, means that we are relieved of a
plethora of {zi'o} only at the cost of a plethora
of {zo'e}s; not an obvious gain. Indeed, a loss
(according to Cowan) since we now have unmarked a
change in the meaning of a sentence which is also
in some sense uncontrolled — we don't know what
is the result of removing one place from a
predicate, especially if the removal allows that
that place might have no filler in the larger
predicate. (Of course, we very often do have in
mind a particular new predicate for that new
expression, but there is no guarantee that that
is what we will get — at least in the eyes of
our conversational partners.)
If we look at actual omissions in use, we find
that (so far as we can reconstruct or remember or
notice them) they arise from a variety of
reasons: those represented by {zo'e} (whatever
they may be) as well as {zu'i} and even {zi'o}
and several others besides. The blank = {zi'o}
proposal would require a massive reworking of the
existing text — much more than redefining
predicates, if the claim that certain places are
never used is correct. Formulating and
formalizing the existing practice might be
simpler, if less satisfying in some intellectual
sense. So far as I can tell, the reasons for
omission can be divided into three broad
categories (to match the three available markers
-- each category has obvious subdivisons): a)
place filler significant, speaker knows what
fills it, hearer can supply the filler correctly
(several senses of {zo'e}and {zu'i} too), b)
place filler insignificant, sppeaker doesn't care
what, if anything, goes there, hearer shouldn't
bother trying to find out ({zi'o} plus a bit) c)
place significant but speaker is not going to say
-- can't or won't — what goes there, hearer can
not certainly identify the filler (ignorance,
secrets, nastiness and so on). If we move {zu'i}
to the last of these, we can briefly explain away
any challenged omission, though in the blank =
{zi'o}'s defense a hearer who cannot identify the
filler immediately is probably wisest to assume
it is insignificant.