This doesn't meet your criteria exactly, but might be a helpful lead.
Landauer's Latent Semantic Analysis website describes one of their
corpora, tasa, as texts broken out by grade level using a readability
score. You would have to contact the people that lent the corpus to
them. LSA is at http://lsa.colorado.edu/, and the corpus description is
under Information -> Semantic Spaces
--Clai Rice
This archive was generated by hypermail 2b29 : Fri Aug 04 2006 - 19:16:40 MET DST