Re: terminators and bilingualism

[Ivan A Derzhanski <iad@MATH.ACAD.BG>]

And Rosta wrote:
> But first could we agree among ourselves what would count as a
> terminator/initiator?
>
> As a first go, without crafting the formulation, I would suggest:
>
>    [X  Y  Z]
>
> where either X, Y or Z is the head of the phrase, and X must
> be the first word in the phrase and Z must be last word, though
> they needn't be obligatory.
>
> Does that get to the essence of it?

Sounds good to me.  It seems to describe circumpositions, in any case,
and those are the closest thing to Lojban's bracketed constructions
that I've seen in natlangs.

> However, it occurs to me that if an XP, which is head-initial,
> systematically selected a YP, which is head final, then you
> would get [X [[Z] Y]].

True.  As in Chinese (where there seem to be many such constructions):
_[zai4 [NP shang4]]_ `above NP' (lit. `(be(ing))_at NP('s) top').

> Such an analysis is available for Lojban terminators, and perhaps what
> really marks out Lojban is the pervasiveness of the construction type.

And also the fact that when such constructions are nested, the repeated
brackets aren't readily elided.

--Ivan