I think desperate means just what was said: "comparing corpora of different sizes" instead of "comparing the sizes of different corpora". When corpora of markedly different sizes are compared, the raw frequencies must be normalised to a common base, for per one million words, perl 100K words, per 1000 words etc. But the common base must be appropriate for the corpora under comparision. See unit 6.2 in the following document for discussion of normalisation.
http://www.corpus4u.com/upload/forum/2005052307351613.pdf
Re tutorials, here are some useful links:
http://bowland-files.lancs.ac.uk/courses/ahaw-nscl/clc_top.htm
http://bowland-files.lancs.ac.uk/monkey/ihe/linguistics/contents.htm