Re: Folk Psychology and Social Convention

Neil W Rickert <[EMAIL PROTECTED]> wrote in message news:<[EMAIL PROTECTED]>...
> >From The Free On-line Dictionary of Computing (17 May 2003) [foldoc]:
>  
> >  extensional
>  
> >     Extensional properties, e.g. extensional equality, relate to
> >     the "black-box" behaviour of an object, i.e. how its output
> >     depends on its input.  The opposite is intensional which
> >     concerns how the object is implemented.
> 
> That might be a computer science usage.  But it isn't what is
> being used when discussing natural language.

I think they are the same abstract idea. However, the rest of my
argument does not depend on FOLDOC definition, of course. You can take
the ESGS definition which is quite correct unlike the output of David
Longley's weak mental processing.
 
> >Yet, the syntax of natural language and programming languages are not
> >Turing-complete generally. Rather, we could say that the "syntax of
> >semantics" of those languages are Turing complete!!!
> 
> The division (with computer languages) into syntax and semantics is
> somewhat arbitrary.  We often treat some syntactic requirements as
> restrictions on formal semantics, and pretend that they are not part
> of the syntax.  That allows us to treat the computer language as
> context free, and allows us to use well known parsing methods for
> context free grammars.

I agree. That is exactly what I am saying above.

> 
> >Neil, by the way, do you know anybody has argued for or against that
> >general Turing-complete syntax can give rise to semantics easily? What
> >do you recommend on that subject?
> 
> If you are talking about natural language semantics, then in my
> opinion the question is on a wrong track.  A robot might be able have
> semantics, but not purely on the basis of syntax.  It would depend on
> the physical interfaces of the hardware.

Of course it should apply to natural language semantics if it is an
argument about general semantics.

You have made the "let's get physical" argument again, but I am unable
to fit this into the current discussion.

I think it demands a deeper analysis but let me try to clarify my
problem with this idea. I think semantics depends on the existence of
well-defined interactions and processing, ie. computation.

So, could you give an example for a robot that has semantics but not
only on the basis of syntax (ie. motor planning)? Do you mean
something like physical properties (material, tension, etc.) being a
necessary condition for a certain semantic property irreducible to the
abstract specification of the robot?

Regards,

--
Eray Ozkural