Ordinal ROI: the very idea!

[John Cowan <cowan@ccil.org>]

Ever since Jorge proposed an ordinal analogue to ROI, such that

1)	mi pare'u klama le zarci
	I for-the-first-time go-to the store

was grammatically equivalent to

2)	mi paroi klama le zarci
	I once go-to the store

the idea has been bothering me, but I was never able to pin down just
why.  Now I understand the problem.  *(BTW, "re'u" was reserved for
this function but never made it onto the official cmavo list.)

"roi", like its vaguer equivalents in TAhE, specifies a property of the
time (or space) interval over which the event of a bridi stretches.
It is, in some sense, an ancillary claim:  Example 2 means that during
the (default, vague) time interval of the main claim, I go to the store
exactly once.

In Example 1, however, the effect of the "pare'u" is restrictive;
it picks out the first event within the main interval and makes it
what the whole bridi is about.  This makes "re'u" not operate as an
interval modifier with respect to subintervals.  Intervals subdivided
with "roi" can be sub-subdivided by using a ZAhO and/or more "roi"/TAhE,
for arbitrarily fine precision.  "re'u" doesn't fit into this system.

Instead, its function is really parallel to that of tanru modification:

3)	le nu mi klama le zarci cu krefu fi li pa
	The event-of my going to the store is-the-repetition numbered-1
		[of some stream of my goings to the store]

is really what is meant.  Indeed, a different x2 would give a different
stream; if I am juror #5, then my book about the trial may be "krefu
fi li mu" in the sequence of books about the trial.

So while the parallel between "roi" and "re'u" is tempting, I believe
that it is not necessary to have "re'u", nor does it fit well into the
ROI selma'o.

Comments are urgently solicited.

-- 
John Cowan						cowan@ccil.org
			e'osai ko sarji la lojban