A Pittsburgh software engineer studied the language of 2.6 million tweets to discover the words which can determine if someone is geeky or nerdy.