I hope that this will give me some positive answer.
Thanks for your time Stefano
Stefano Federici ------------------------------------------------- Università degli Studi di Cagliari Facoltà di Scienze della Formazione Dipartimento di Scienze Pedagogiche e Filosofiche Via Is Mirrionis 1, 09123 Cagliari, Italia ------------------------------------------------- Tel: +39 349 818 1955 Fax: +39 070 937 1870
----- Messaggio originale ----- Da: John F. Sowa <sowa at bestweb.net> A: stefano federici <s_federici at yahoo.com> Cc: corpora at uib.no Inviato: Giovedì 10 aprile 2008, 13:14:23 Oggetto: Re: [Corpora-List] Sorry, my mistake, I need a Context-DEPENDENT grammar simulator
Stefano,
You are asking for somebody to hand you the Holy Grail:
> What I need is a context-DEPENDENT grammar simulator.
Chomsky's original definition of context sensitive grammar rules was very general, but not very practical for dealing with the kinds of context dependencies in natural languages (or even the much, much simpler programming languages).
The number of different kinds of context dependencies in NLs is enormous, and many different branches of linguistics (both theoretical and computational) have evolved to deal with various aspects in an open-ended variety of complex ways.
At one extreme, any background knowledge that people have in their heads (or computers have in their knowledge bases) may be necessary for disambiguating a sentence in context.
At the other extreme, a table of names, references, symbols, etc., is a highly efficient mechanism for keeping track of anaphoric references in NLs and symbol declarations in programming languages.
A complete list of all the kinds of context dependencies and how to describe, formalize, and implement them would touch on essentially every aspect of linguistics, logic, and knowledge representation.
In between, there are interconnections of the many different levels, such as semantic variations that may cause a puff of white noise to be interpreted as different phonemes, depending on what the listener was expecting.
I think that everybody who subscribes to this list would like to see a simulator of all that. If anyone has such, please let us know.
John Sowa -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.uib.no/mailman/public/corpora/attachments/20080410/b0e2f66e/attachment.html