It's a silly test to begin with. It doesn't take into account shortened URLs which are a big part of the Twitter discussion. It can't take into account phrases like "Democratization of News" or "New Media". It doesn't even take into account stemming ("tweeet" has a dramatically different graph than "tweeting")

It just doesn't prove a thing