[Corpora-List] PS:minimal changes in a paragraph (based on a corpus it appeared) ... (2nd attempt (after first one was deleted))

[John F. Sowa]

On 8/10/2011 10:07 PM, Albretch Mueller wrote:
> I do believe one day we will get text
> processing (all those issues like NLU, NLG (which to me is a form of
> NLU, ...) right and I also believe the way to go is using corpora and
> forget about the semantic web thing and all of that kind of stuff

I also believe that "one day we will get text processing (all those
issues like NLU, NLG ...) right" and that "using corpora" is necessary
to achieve that goal.

Unfortunately, I would also "forget about the semantic web thing"
as relevant to that goal.

But I don't believe that arguing about our terminology is useful.

The following slide from a recent talk summarizes my position.

John
____________________________________________________________________

Relating Psycholinguistics to Neuroscience

Neurocognitive networks are not a "Language of Thought":
  ● They link nodes in the sensory, motor, and association areas of the 
cortex.
  ● Some nodes are associated with images, sounds, feelings, and actions.
  ● Other nodes and patterns of nodes, which may be called concepts and
schemata, have indirect links to the sensory and motor nodes.
  ● Some of those nodes and patterns are associated with words.
  ● But they don't form a language that is distinct from a natural language.

Reasoning is a process of forming new patterns:
  ● The basic mechanisms form, strengthen, weaken, and inhibit links.
  ● The frontal lobes are actively involved in the process.
  ● But the new links or modified links may be anywhere in the cortex.
  ● The resulting patterns may be mapped to actions (including speech).

Syntax, semantics, and pragmatics are theoretical abstractions:
  ● They can be useful for analyzing and understanding neural patterns.
  ● But they are, at best, indirectly related to those patterns.

This is slide 21 of http://www.jfsowa.com/talks/ca4cs.pdf

The paper associated with the talk: http://www.jfsowa.com/pubs/ca4cs.pdf

_______________________________________________
UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora