A powerful alternative application of the collapsed text approach is to offer students texts that come from a wide range of text types. Is it possible to isolate the key words, and the key patterns of repetition, that identify text types? Does it work even when the text has been re-sorted into alphabetical or random order?