The effect on upper-level university students of humorous versus serious instruction was studied by presenting students with sentences having introductory modifiers and asking them to edit incorrect or recognize incorrect and correct sentences. Sentences varied in the main clause, a third having active verbs, a third having passive verbs, and a third having the implied subject of the introductory modifier a possessive modifying the main clause's subject. Uninstructed groups were controls. To determine the relationship between treatment effectiveness and student English background, units of Iowa State English and grade point in that English were examined. All groups were given a pretest during the term's first week, instruction (if any) and a posttest the next day, and the posttest as a followup three months later. The editing instructed group was asked to rewrite 30 incorrent sentences; the recognition instructed group was given the same 30 incorrect sentences and 30 correct sentences to recognize as correct or incorrect. In both editing and recognition experiments, instructed groups were more effective, and type of sentence made a difference. The instructed groups best recognized correct sentences with active or passive main clauses and found the distracting possessive more difficult. Yet student edited passives best and worked equally well with actives and possessives. The recognition group was most effective with actives and passives if instructed and more effective with possessives if taught with serious rather than humorous examples. In the editing study, there was no effect of varied instruction on performance with differing types of sentences. Background variables, alone, had no effect in either experiment. In the editing study, those with many units of Iowa State English performed better than the low group with active sentences; unexpectedly, the low group performed better with possessives and best with passives. The low grade point editing group was more effective if taught with humor, whereas the high grade point group was more effective if taught with serious examples. The high grade point editing group was least effective with possessives if taught with humor, yet the low group was most effective with possessives if taught with humor.