Pangram isn't useful for single documents. Too many false positives. It only gives something approaching actionable results if you use it on an entire corpus.