Eighth grade students, in response to a pictorial stimulus (either an intact picture, or a series), were instructed to write either a story (which would be enjoyable for another eighth grader) or a description (which would help another eighth grader visualize the picture). Thirty responses from each of the four task conditions were then analyzed. For the inductive or holistic analysis, raters assigned the stories to the top, middle, or bottom third of a scale, based upon specified (primary trait) criteria. The second method of inductive analysis determined the syntactic complexity of the stories, based upon: (1) length of T-units--minimal units capable of being punctuated as a sentence; (2) length of dependent clauses; (3) number of dependent clauses per T-unit; and (4) length--number of T-units. Responses were also classified, deductively, as poetic or transactional; further judgments regarding story quality and grammar made according to this classification. Regression analyses revealed that the variables contributing to the primary trait (holistic) rating were different for each of the task conditions. Except for the primary trait rating, the evaluation variables were not found to relate to one another in any consistent manner. (GDC)