Corpus-based parser generator?

Does anyone know of any research on corpus-based parser generation?
By 'corpus-based', I mean an approach that analyses a large body of
existing, known valid code, as compared to the traditional approach
which starts by defining the syntactic/semantic rules.


Basically I want to take a whole bunch of programs in language X,
throw these at a tool which will then generate a parser for X. The
end-result might not be a compiler as such, perhaps just a tool
that can say 'Yes, that looks like an X program, except for this
and this, which are wrong'.


--
Pieter A. Hintjens
[Sounds well-nigh impossible to me, but it'd be fun to try. -John]
--

Return to the comp.compilers page.
Search the comp.compilers archives again.