Claude, you're exactly right. When I used the term "grammar," I was using it in a narrow, sentence-diagramming sense, and unless that's made clear, the statement that you quoted from me rings hollow.
Unfortunately, the broader brush you're using to describe the set of rules that humans use to speak to and understand each other is just the kind of tool that we haven't developed yet for computers (which is the context in which I originally thought the "he and only he" sentence to be at all interesting).
For an entity (natural or artificial) to have a full mastery of "grammar," in the comprehensive sense you've identified, it seemingly must have a non-rules-based understanding of the world, "world" defined as all the things that might be an object of a thought expressed through language. For example, without understanding that "he" refers to a human, and that a human can be in only one place at a time, and that no two humans can safely occupy the exact same place at one time, an entity will have a tough time interpreting the "he and only he" sentence correctly.
In other words, to try to put it more succinctly, description requires an object to be described, meaning that descriptive grammar requires not just an agreement between the speaker and listener to use a common set of grammar rules, but also a common understanding of the world in which they live to serve as the objects of descriptive grammar. That common understanding must be a nonverbal mental representation of everything in the world.
From the low-level computer programming perspective I have, it absolutely blows my mind to think that we'll ever be able to model that nonverbal mental representation within a computer to enable it to understand natural language. Yet I don't see how we can do it without that model.
Go ASND! |
