So I think your i- nouns would indeed have to have another object marker
before them, and that from the very get-go, not even as some eventual
reanalysis once the origins were no longer transparent. They would also
appeared in non-object positions from the very beginning.

Yeah, I think what happened is I confused myself thinking about
"another" (which is still interesting, but a different phenomenon).

No, the order of events is the syntax.

Ah, I see.  So the fact that we get played instead of edplay is also
the syntax?

If you can do this with morphology...

{X [V] <-> Xed [V (past)]}

...i.e., suggesting that that is a pattern of the language, which  
invalidates other patterns (like "edplay"), then why can't you do
the same with the order of words?

{X [V] <-> should X [V (oblig. Z)] <-> should be Xing [V (oblig.  
progr. Z)]}

The patterns tell you what the order is and isn't.  Further, there's
nothing that prevents a deviation except popular opinion, which
itself is derived from, among other things, similarity to already
instantiated patterns.  Where just about every theory of syntax
got it wrong was assuming there were rules.  They come up with
the rules, make predictions, and then find data that should be
illegal.  Then they either try to quickly patch up their syntax
(non-Chomskyan syntacticians) or claim that the data themselves
are wrong or insignificant (Chomskyan syntacticians).  In the end,
they're doing nothing more than describing their theory, as
opposed to language.  As I'm more interested in the latter, I figure
we can do just fine without the conlang that is syntax.

"A male love inevivi i'ala'i oku i ue pokulu'ume o heki a."
"No eternal reward will forgive us now for wasting the dawn."

-Jim Morrison