The n-gram extraction can be carried out on any %[attribute]%. The most useful are:

lemma – the different %[wordForm|word forms]% of the same %[lemma]% will be treated as the same thing

word – the different %[wordForm|word forms]% of the same lemma will be processed individually