Re: teach your AI to sign

From: Bryan Moss (
Date: Mon Feb 12 2001 - 19:31:36 MST

Damien Broderick wrote:

> The argument, fairly well established in linguistics and
> child development studies, is that single words uttered by
> infants are often holophrastic - they encode an entire
> sentence or action, they are not simple indexical or
> ostensive in the sense of `naming' the thing pointed at.

So they progress from a simple mapping of a word to an action
to a complex multi-word description. Has anyone looked at the
in-between stages? Sounds like a strategy where you first
make a simple connection and then add exception clauses.
(This is actually how people tend to express logic, they say
"if A then C unless B" whereas formal logic would have them
say "if A and not B then C".) Are the holophrastic words
arbitrary? If not it seems they'd be very restricted.


This archive was generated by hypermail 2b30 : Mon May 28 2001 - 09:56:39 MDT