From English-Corpora.org (Brigham Young University), the TV Corpus and The Movie Corpus together contain over 525 million words of data, and are a vital resource for looking at informal language. All of the 75,000 TV episodes, and 25,000+ movies, are tied in to their IMDB entry. Both Corpora allow you to look at variation over time (1950s-1970s to 1990s-2010s) and variation between dialects (e.g. American and British English).