Statistical Language Identification of Short Texts