Re: Bob's message #3

[cowan (John Cowan)]

la atlstan. cusku di'e:
>     Add JAI+tag as another kind of conversion besides SE.  (JAI is a new
>     selma'o.)  This is usable only on selbri, not in the other places where
>     SE is legal.  The result is that the se tcita sumti comes to occupy the x1
>     place, and the original x1-x5 places are "pushed down" to x2-x6.

What I meant was that JAI+tag is >grammatically< equivalent to SE; it
appears as a grammar rule exactly parallel to that which supports SE in
selbri.  Nonetheless, this rewording is clearly better, and I adopt it.
(Note the correction of "tcita sumti" to "se tcita sumti":  the former
means "tagging sumti" rather than "tagged sumti".)

la lojbab. cusku di'e:
> But [Athelstan] further noted that the proposal (separated from a formal
> grammar change list) supports not only description use JAI+tag, but
> also main selbri use JAI+tag.

That's not a bug, it's a feature.  Note also that JAI+tag is short-scope,
like SE, and can be used in tanru:

	ma jai bau cusku cuxna vi le .ubuny
	What is-a (language of expression) chooser at the U.N.?
	Who chooses the language of expression at the U.N.?

To make it long-scope you need JAI+tag+ke...ke'e, just as for SE conversion.

> If JAI+tag is the equivalent of SE in non-description selbri like this,
> it will also work the same way:
> 
>     ti   jai ca         klama mi       le  zarci
>     This is-the-time-of going by me to the market.
> 
>     le  jai ca  klama be mi bei le  zarci
>     the-time-of going by me to  the market.
> 
> But if we are sumti-raising around a cleft place structure problem,
> and you are trying to replace the event by the agent, confusion will reign
> 
>             gau        mi                             galfi    le  bitmu
>             with-agent me (some-x1-event-unspecified) modifies the wall
> 
> must become as a description
> 
>      le  jai gau  galfi     be fi le bitmu
>                                --
>      The actor-in modifying       the wall.
> 
> Without the "fi", the wall is the former first place of galfi, the event
> that modifies, which is not what is intended.

This oddity results from the unusual format of the underlying bridi.
Here we have an observative (with elided x1 place) that has a se tcita sumti
preceding the selbri.  If we express the x1 place with a zo'e we get:

	gau mi zo'e galfi le bitmu
	le jai gau galfi be zo'e bei le bitmu

which is the right answer.

> The only feature of SE conversion not obtainable
> with FA is the specification of the x1 place accessible from a
> description.

Not quite true, due to the use of conversion within tanru.  The effect of

	mi se pilno zbasu
	I am a (something used) maker
	I am a toolmaker.

cannot be achieved by any manipulation of the place structures of
"pilno zbasu" (which would be "tool-user maker") using FA tags.  
However, I agree that compound SEs aren't needed in this use either.

> Athelstan's
> position [is] that 'jai+tag conversion' should indeed be identical with SE
> conversion, including the exchange of sumti.  But what is the x1 place
> exchanged with when we do a 'jai+tag' conversion.  Athelstan calls this
> an 'extradimensional place', one that lies outside the numerical
> sequence since a tagged sumti lies outside the numerical sequence of the
> place structure.  To access this place in a specified description, we
> need an 'extradimensional' member of FA.

Er, what is supposed to happen if you have two consecutive JAI+tag conversions?
For concreteness let us take the bridi:

	mi cusku bau la lojban. bai la lojbab.
	I express (something) in-language Lojban under-compulsion-of Lojbab.

If we convert this with "jai bau" under Athelstan's proposal (and using
"fai" to FA-mark the extradimensional place), we get:

	la lojban. jai bau cusku fai mi bai la lojbab.
	Lojban is-the-language-of-expression by-me under-compulsion-of Lojbab.

If we now (perversely, but legally) employ another jai-conversion, we get:

	la lojbab. jai bai jai bau cusku fai mi fai la lojban.
	Lojbab is-the-compeller of-the-language-of-expression by me as Lojban.

Here the previous x1 place (la lojban.) has been swapped to the extradimensional
slot, which was already filled by the previous operation.  We now have two sumti
in the same slot, which loses.


One perverse possibility would be to have the original x1 to truly exchange
with the se tcita sumti:  i.e. it gets marked with the relevant sumti
tcita, thus:

	mi klama le zarci ca le nicte
	I go to the market during the night.

	le nicte jai ca klama le zarci ca mi

This defies easy gloss, but "le nicte" and "mi" have really, truly been
swapped; "le nicte" is now in the positional x1 place, and "mi" is
following the "ca" tag!

These are deep waters:  on the whole, I think the original "pushdown"
proposal is safer.

--
cowan@snark.thyrsus.com		...!uunet!cbmvax!snark!cowan
		e'osai ko sarji la lojban