John Cowan eats crow about {co}

Well, it turns out that I have been preaching a heretical gospel about {co}
tanru for some time.  lojbab finally straightened me out on the phone, after
Jorge pointed out some difficulties and discrepancies in my previous account
that I had been sweeping under the rug.

Here is the one word of truth about {co} conversion, as I now understand it
(terminology: "seltanru" is the modifier, or left-hand, member of a tanru;
"tertanru" is the modificand, or right-hand, member.):

In normal tanru, the seltanru always precedes the tertanru.  By inserting
the particle {co}, however, the order can be changed.  Examples 1 and 2
mean exactly the same thing:

1)	ta blanu zdani
	That is-a-blue type-of-house.
	That is a blue house.

2)	ta zdani co blanu
	That is-a-house of-type blue.
	That is a blue house.

This change is called "tanru inversion".  The meaning, and more specifically,
the place structure, of a tanru is not affected by inversion: the place
structure of "zdani co blanu" is still that of "zdani".

However, the existence of inversion in a selbri has a very special effect on
any sumti which follow that selbri.  Instead of being interpreted as filling
places of the selbri, they actually fill the places (starting with x2) of the
seltanru.  Normally, filling interior places is done with "be...bei...be'o",
and in fact Examples 3 and 4 have the same meaning:

3)	mi klama be le zarci bei le zdani be'o troci
	I am-a-(goer to the market from the house) type-of trier.
	I try to go to the market from the house.

4)	mi troci co klama le zarci le zdani
	I am-a-trier of-type (goer to-the market from-the house).
	I try to go to the market from the house.

Example 4 requires fewer cmavo and less embedding, but means exactly the
same thing.

Sumti which precede a selbri with an inverted tanru fill the places of the
selbri (i.e., the places of the tertanru) in the ordinary way.  In Example 4,
"mi" fills the x1 place of "troci co klama", which is the x1 place of "troci".
The other places of the selbri remain unfilled.  The trailing sumti "le zarci"
and "le zdani" do not occupy selbri places, despite appearances.

As a result, the regular mechanisms (involving selma'o VOhA and GOhI, explained
elsewhere) for referring to individual sumti of a bridi cannot refer to any
of the trailing places of Example 4, because they are not really "sumti of
the bridi" at all.

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