Dylan, On 24/06/2008 22:41:
> I don't dive in too often but ...
>
> On Saturday 21 June 2008, And Rosta wrote:
>> How great a constraint on the properties of a formal system does
>> implementability in wetware impose?
>
> Surely the "implementability in wetware" is paramount if the formal system is
> to have any hope of providing insights into the /reality/ of our linguistic
> (or general cognitive) capabilities? If this is not the case, we may as well
> still argue over whether (human) language is a BNF grammar or a purely
> instinctive stimulus-response system.
>
> If we are to learn or discover anything significant in this enterprise, we are
> forced to ground our theorising in precisely the constraints which the brain
> imposes simply because IMHO the formal characteristics of the system *arise*
> from those constraints (and their complex interactions.)
I'm not persuaded that the formal characteristics of the system arise from constraints imposed by the brain. The system might be of sufficient generality/abstractness that lots of different kinds of thinking machines could learn it. Lots of knowledge, I contend, is platform-independent.
But even setting that aside, take WG as an example of theory that gives a lot of thought to implementability in wetware. I reckon you could fairly straightforwardly replace current WG syntax with, say, HPSG syntax. Are there, I wonder, any cognitivist grounds for preferring, say, standard WG to HPSG-flavoured WG?
--And.
|