[Corpora-List] Is a complete grammar possible (beyond thecorpus itself)?

[Ken Litkowski]

In the parser I use, developed by the inventor of syntax-driven 
compiling/parsing (Ned Irons, in 1960), errors are part of the system. 
Positive points are given to every parse that is carried forward (the 
number is user-definable) for each anomalous seeming construction.  At 
the end of each round (token), those with the highest number of points 
or above a certain level (also user-definable) are discarded.  So, I see 
no reason why a typology of errors shouldn't be part of what we're 
doing.  I'll remain agnostic on whether this should be construed as part 
of a complete grammar.

maxwell at umiacs.umd.edu wrote:
> Rob Freeman wrote:
>> If you've got another 50 years to waste there are!
> 
> Well, I wouldn't call what has been learned about language in the last 50
> years by formal linguists a waste; I would say we know a *lot* more as a
> result.  But your mileage may vary :-).
> 
>> I'm just saying we should at least explore
>> the possibility formal grammars are "necessarily incomplete"
>> descriptions of corpora
> 
> I don't think most generativists would disagree about the possibility of
> completely describing *corpora* by a grammar; but that would be because
> they (I'm including myself) regard corpora as inherently errorful, and
> it's silly to expect a formal grammar to generate all (and maybe only) the
> attested errors (in addition to all and maybe only the attested
> non-errorful parts).
> 
> (Well, you can describe all the errors by listing them, but that's hardly
> insightful.  What might be an insightful treatment of errors would be
> trying to come up with generalizations for classes of errors, like slips
> of the tongue.  But these generalizations are not going to be part of the
> *grammar*; rather, they're going to be things like processing limitations,
> or interference by the first language in production in the second
> language, or noise in the telephone system, or some such.)
> 
> Of course that's precisely why most generativists focus on (the elusive)
> competence, rather than performance.
> 
>    Mike Maxwell
>    CASL/ U MD
> 
> 
> _______________________________________________
> Corpora mailing list
> Corpora at uib.no
> http://mailman.uib.no/listinfo/corpora
> 
> 

-- 
Ken Litkowski                     TEL.: 301-482-0237
CL Research                       EMAIL: ken at clres.com
9208 Gue Road
Damascus, MD 20872-1025 USA       Home Page: http://www.clres.com

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora