--- On Mon, 6/29/09, Brett Williams <[log in to unmask]> wrote:

> <deinx nxtxr><[log in to unmask]>
> wrote:
> >
> > I have a loglang that I've started.  It uses
> predicate logic too just like
> > Lojban but I've the is nothing with a valency beyond
> 2, subject and object.

> Two-part predicates is how I would think of doing it, if I
> were
> putting together a new loglang.  I would actually have
> exactly two
> places for each root, no more no less, to make it
> completely
> predictable.
It's possible to have complete predictability with six or seven places; it just requires proper design. And there are relationships that simply have more than two places. If you go somewhere X, you do so from a point of origin Y (and over a route Z); if you have a father, there was a mother as well--and so on. Nor is it unimportant whether you have all the places: if you try something moderately ambitious, such as changing the order of elements in an incomplete statement, you can wind up unambiguously asserting something you don't mean.

My currently mothballed loglang is thoroughly predictable and human-friendly, though it is a bit lengthy. I've been meaning to re-examine the phonology and morphology, since, as I mention in my notes, a loglang needn't have the restrictions an auxlang should.