I don't know of any nice way to do this directly with lout files (and
I've looked and I've tried!). Lout has no BNF and is a nightmare to
parse.
The solution I've now switched to is to use my own ViewPDF program. This
is just a very simple PDF viewer but it highlights spelling mistakes and
a couple of other possible problems (e.g., double words). It is free and
open source and easy to build on Linux once you have installed all the
-dev libraries it needs:
http://www.qtrac.eu/viewpdf.html
(I'd be interested to know how you get on with it.)
BTW At the moment if you add words to your own list (words.lst) they are
handled case-insensitively.