RE: [lojban] Re: Loglish

["Ben Goertzel" <ben@goertzel.org>]

> > Regarding your point about whose argument the
> > fi'o specifies -- I believe
> > this is unproblematic,
> > becaues the fi'o clearly tells whose argument
> > "lo pliers" fits into, and "lo
> > pliers" is clearly
> > fitting into *some* argument of "murder."  To
> > have "lo pliers" fit into some
> > argument of "chicken"
> > one would have to use "be" or some other
> > similar mechanism, I believe.
> 
> Quite right.  But that means that, from the point
> of the parser, the FrameNet bits are unnecessary
> and that discussion is irrelevant at that point. 
> At some point in interpretation (and, yes, I
> think the two are --or ought to be -- separate as
> much as possible, which is 100%) we learn that a
> weapon is an instrument or means, but  that only
> because we already know that it is a peripheral
> argument of "murder."

Yes, the FrameNet lookup process is not necessary for
syntax parsing, but is necessary for semantic processing.

In some cases, though, if there is some uncertainty in
which sense should be assigned to a word, the correct
choice of sense might depend on how much sense the likeliest
FrameNet-based argument assignments make for each word.

However, this is also a matter of semantics, not syntax.

In general, Loglish is essentially *syntactically* identical
to Lojban; the difference is almost entirely semantic,
except for the very simple syntax of the new Loglish cmavo
(qui and quu).  The Loglish syntax for fi'o is the same as the
Lojban syntax for fi'o, even though the semantics is arguably
a little different.

-- Ben

-- Ben