Re: TECH: zi'o: never decided formally

[Logical Language Group <lojbab@ACCESS.DIGEX.NET>]

la lojbab. cusku di'e

> Nora's opinion is that zi'o per se does not logically belong in the
> language for reasons similar to those [Kennaway] stated.  However it does
> belong effectively in the rafsi set, as a means to conventionally define
> new 'metaphorical' lujvo which have omitted a sumti from the
> relationship.  Exactly what it means to omit a particular sumti is, as
> you describe, somewhat difficult to determine in any degree of
> formality.  The key point is that a predicate with a zi'o place is a
> totally different predicate from the predicate with the place normally
> there.  Inferences between such predicates are problematical at best.

I mostly agree.  No inference is possible from the zi'o version to the
full version; only limited inference is possible from the full version to
the zi'o version.

> Cowan then said later, after several exchanges]
> >P(a,b,c,d,zi'o) is a four-place
> >predicate, such that P(a,b,c,d,e) implies P(a,b,c,d,zi'o), but not vice
>                                    ^^^^^^^
> >versa.  We can infer from "mi klama le zarci" that "mi klama le zarci
> >zi'o zi'o zi'o", but the relationships are not the same:  the first case
> >asserts "I go to the store from somewhere with-route something
> >with-means something" whereas the second asserts "I go to the store",
> >period.  I do >not< assert the existence of origin, route, or
> >destination -- which is not the same as denying they exist.
>
> whereas Nora assertion is that "implies" is too strong, because the
> proper notation should be P(a,b,c,d,e) and P'(a,b,c,d,zi'o)

I was using (a typographical variant of) Prolog notation, in which the
predicates "foo(x,y)" and "foo(x,y,z)" are not assumed to have anything
to do with one another without further information.  Prologists distinguish
the two predicates as foo/2 and foo/3.  (It is common, though, for foo/2
to be defined as some variant of foo/3 with some argument implicit, more
or less our zo'e.)

I continue to affirm that if "I go to the store from home" is true, then
"I go to the store" (without asserting anything about an origin) is true too.
This is the "implication" I wish to see part of the definition.

> This would suggest that zi'o should be covered in Nick's lujvo-making
> paper, if it is not already, in more detail than in any other paper
> Cowan is writing, since I would define zi'o in all its grammatical glory
> as a cmavo as a back-formation from the lujvo-making technique, with the
> semantics of a lujvo.  Thus, while it has the grammar of KOhA, it is
> really more akin to "zei", the lujvo-linking cmavo.

There is no doubt that "zi'o", despite its grammar, is a selbri-changing
particle like ZEI or SE or JAI.

> My opinion, therefore, is that there is sufficient consensus to add the
> zi'o artifact as a rafsi (which I think should come after a number if
> using a numerical convention to say what is deleted, since this leads to
> minimal hyphenation).

I still think it is better to use the "zi'o" rafsi first, before the number,
to avoid garden-pathing the listener, and because the number then looks sort of
like a subscript.  Note that "zi'o" as a rafsi is baselined for "dzipo",
and I see no reason to change a baselined rafsi for this small feature.
I prefer "zil-" instead.  Hyphenation is then minimal: "zilpav-", "zilrel-",
"zilcib-", etc.; at most a "y" rather than a more intrusive "r"/"l".

> I am not convinced that there is a consensus
> regarding the cmavo itself, but we have no rafsi in the language that do
> not map to a non-rafsi form, so this alone may be reason to add the
> word.

I have to agree with this.  A bound form with no free form is very unLojbanic,
even if the grammars differ wildly, as in KOhA and PA rafsi.

--
John Cowan              sharing account <lojbab@access.digex.net> for now
                e'osai ko sarji la lojban.