I can think of one fine use of a large body of texts on line: historical
linguistics. The study of language change used to be done by some
scholar poring over centuries of texts armed with a pencil and a stack
of index cards. You would note down every instance you found of some
interesting language feature, and note every instance where you would expect
it to appear and it did not. Correlate all of the cards
with time, place and historical events and look for trends. This took
*forever* to do.
For an on line collection of texts to help it needs to
a) Span a long period of time
b) be ruthlessly accurate (preserve all the apparent typos and
c) contain some pretty mundane stuff: diary entries, love letters,
shopping lists... whatever.
So: probably no help for linguists here, either.
-- Judy Grass, CNRI