A study required children to tell and to dictate stories that were real and make believe, all about the same basic topic, as part of things people do when they "write a story." A further purpose was to explore the reading knowledge of children who are just beginning to read. Children's reading attempts for these stories were used to evaluate their emergent reading ability in order to test whether there was a relationship between emergent reading ability and the structural "well-formedness" of the stories. Three language productions were obtained: a told story, a dictated story, and a handwritten story. The transcriptions were scored according to (1) the completeness of context (degree of specificity), (2) adapted production story grammar (structural complexity), and (3) emergent reading ability. The contextual analysis gave support to the claim that children will produce real stories that are more scriptlike and predictable, whereas they will elaborate more upon make-believe stories. There was also some indication that the make-believe stories contained more of the elements of a well-formed story. Furthermore, these elements were more highly specified, or more understandable to a nonpresent audience. When overall structural complexity was judged, however, there were no differences for topic (real versus make-believe) or for mode (told versus dictated), nor was there a regression of emergent reading abilities upon structure. (HOD)