In a message dated 12/3/2001 7:33:00 PM Central Standard Time, thinkit8@lycos.com writes:. In computers, you compile a source code to machine This sounds like direct semantic representations, i.e., the last linguistcalike representation of propositions/thoughts. In that case, the usual thought about its nature would be a central concept with the other factors in the the proposition dnagling from it by uniquely specified hooks. That is, a variety of binary hooks, one end of which is always the central concept, the other the argument, if you will, and the hook specifying the type of connection. Supposing for the moment that "give" is a simple concept (rather than, say, "transfer of possession without compensation" which might have more detailed structure and ways for hoooks to attach). The it could have a variety of hooks: agent, recipient, specification of what possessed, purpose, and so on. But would need no one of these, giving rise to a number of different thoughts from the general idea of giving to a completely detailed description to a particular act of giving. Only one type of hook can be unspecified (and even that is probably a mistake); all the rest have particular functions. The alternative is to build the hooks into the central concept -- but then you always have some left over that you need to add in some cases. And for general semantics (no caps!) the more analytic approach is best, since different langauges collapse hooks into concepts in different ways. |