Points: A Theory of the Structure of Stories in Memory
Robert Wilensky University of California, Berkeley
Story comprehension has recently received a great deal of attention from cognitive psychologists and researchers in artificial intelligence (for example, see Bower, 1976; Kintsch & van Dijk, 1975; Mandler & Johnson, 1977; Rumelhart, 1975; and Charniak, 1972; Cullingford, 1978; Schank & Abelson, 1977; and Wilensky, 1978). However, the complaint has been made repeatedly that most of the work on story understanding has little to do with stories. Rather, what is being studied both in Al and psychology are coherent texts. The difference is that not all coherent texts are stories, and that a theory of stories per se is still lacking.
One ostensible exception to this criticism is a formalism known as a story grammar. Using a grammar to try to capture the notion of "storyness" seems first to have occurred to Rumelhart ( 1975). Since then, the notion of story grammars has been expanded theoretically by a number of researchers, and even used as the basis for a number of empirical studies (e.g., Mandler & Johnson, 1977; Stein & Glenn, 1979; and Thorndyke, 1977).
Unfortunately, the story-grammar concept is lacking in a number of ways that make it inadequate for its intended purpose. The chief problem is that story grammars purport to capture the idea of what a story is by trying to express the structure of a story text. My claim is that a theory of stories must be much more concerned with the content of a text than with its form. Moreover, when story grammars are examined closely, most of the story structure they aim to capture dissolves away, and they end up saying little more than that story is a coherent set of sentences.
A detailed critique of story grammars is found in Black and Wilensky ( 1979),