Is there a corpus of press releases, etc. from lots of big (tech) companies? If so, has anyone done basic comparative word/n-gram frequency analysis of it?