Not possible currently. I'd like to find a source of "how common is this word" and add some kind of quantifiable number to each word. Perhaps I can scan existing open source text and just figure it out. On the backlog.