We propose a method to represent the narrative and cognitive structures of a type of story that incorporate variations in its different versions without using abstractions such as narrative grammars. The method is based on motifs and a network model, the nodes of which are motifs, to represent the story progression. We modeled variations of Cinderella tales using our method to demonstrate its efficacy, and defined meta-knowledge about folktales that captures the progression of narratives and allows reconstruction of narrative and cognitive structures based on the network model. We claim that our research provides an effective approach to model narrative structures that makes it possible to generate new variations from the model.