Re: Hypersimple & Dreadfully Unnatural Grammars
From: | Orjan Johansen <oerjan@...> |
Date: | Wednesday, March 17, 1999, 22:02 |
On Wed, 17 Mar 1999, Edward Heil wrote:
> I've been thinking about this, and while at first I thought of these
> RPN-ish grammars as dreadfully unnatural, they seem to be variations on
> the following theme:
>=20
> As symbols come in through your ears, you're sitting there building
> conceptions. You try to fuse the meanings of those symbols together as
> they come in. Sometimes a given meaning has several possible places
> where other meanings could fit into it -- multivalent verbs are example=
s
> of this. That's where you need things like case marking, morphological
> or syntactic.
>=20
> It strikes me that natlangs are probably nothing more than examples of
> this process, optimized according to parameters we don't understand to
> maximize communicative value and ease of use.
This reminds me of some thoughts I had during the great Center Embedding
debacle.
Basically, I thought that the reason why humans could not parse highly
convoluted word orders such as these, is that not only are there a limite=
d
number of specialized slots available, but each slot can only hold one
item at a time. So, for example, after parsing a subject that subject
would have to be collected into a higher order slot (e.g. for a whole
sentence) before another subject could be parsed.
RPN and enter embedding word orders both suffer from problems under such
restrictions.
Greetings,
=D8rjan.