[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >


[Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [jboske] grammar & pseudogrammar



#>>> cowan@hidden.email 04/29/03 03:06am >>>
#And Rosta scripsit:
#
#> In linguistics, a 'formal grammar'
#> is just that, a formal grammar, and not a pseudogrammar like
#> Lojban's.
#
#Under the definition of "formal theory" used by logicians, the set of
#linguistic theories that instantiate it is precisely zero.

Partly because it's such a tall order to produce one that fits the facts
of any natural language.  Many linguistic theories would have
adequate formal machinery to define the grammar of an invented
language, though inventing the grammar in the first place is far
from trivial if it endeavours to add to formal logic the functionality
of a natural language.

#> As we have said before on this list, the thing that
#> Lojban (and computer languages, I gather) call a 'grammar', is
#> (or at least was) in linguistics called a 'grammaticality checker'.
#
#It's rather more than that, And.  It also produces a parse tree,
#if not necessarily a deep semantic one.

Yes, but it was the consensus -- and if it is no longer the consensus
then it is still the default & majority position -- that the parse tree is
largely meaningless (predicate--argument structure being the
obvious albeit murky exception), and hence is a mere by-product of 
the grammaticality checker. You yourself endorsed this view last time
it was discussed. (Specifically, the view you endorsed is that the
parse does not serve as a constraint or determinant of the meaning
of a phrase.)

--And.