A fixed law, an established rule: that is what immobilizes narrative. (Todorov 165)
Rereading [is] an operation contrary to the commercial and ideological habits of our society, which would have us "throw away" the story once it has been consumed ("devoured") .[ldots] [Rereading] is tolerated only in certain marginal categories of readers (children, old people, and professors) .[ldots] (Barthes 15-16)
Narratives, it seems, move toward closure. This is an impulse of both the narrative itself (which must finally come to an end at a certain page number) and of the reader (who must eventually close the book, put it down, and begin something else). Even texts that attempt to keep meaning in motion, to present multiple possible endings for their plots, are subject to this totalizing pressure. In a society which considers rereading to be a kind of marginal activity (as Barthes implies), when confronted with a perplexing ending most readers will simply pick one scenario and consider the case closed. For example, Toni Morrison's Song of Solomon (1977) ends with an extremely ambiguous moment in which the protagonist, Milkman Dead, leaps into the arms of his friend Guitar, who has been pursuing him in order to kill him. Although Morrison crafts this moment so that it is subject to multiple interpretations, many readers feel obligated to take a stand as to how the plot of the novel ends. But Morrison's point in creat ing this ending, and in much of her fiction, is to keep meaning in motion, to keep the story going on and on in the reader's mind and heart. The reader might turn back to the first page of the novel to begin rereading it for clues about what the ending means--but only if the reader has been able to resist the totalizing impulse, the desire to close the book, resolve the story, find an ending that sutures over uncertainties in favor of a unified and unambiguous conclusion.
The search to find narrative methods that resist the totalizing impulse of narrative and of readers themselves is a central aspect of Morrison's fictional technique, and is certainly connected to her investment in an oral, African American tradition of storytelling, of the Griot. Beloved (1987) marks the height of Morrison's achievement, for it is a narrative that resists closure in numerous ways. I have found that for this reason teaching Beloved is always a new experience--no class reacts to it the same way, as it generates multiple ambiguities that cannot easily be sutured over. Yet in teaching this book, I am always surprised by how ready students are to resolve the issue of Beloved's status in this novel, to decide unambiguously that she is a ghost--in fact, the ghost of the child Sethe killed eighteen years earlier. In my mind, however, the text balances between realistic explanations of Beloved's presence (she is an escaped slave woman who has been sexually abused by a white man) and supernatural ones (she is Sethe's dead child come back to haunt her), and is therefore an excellent example of what Tzvetan Todorov has called the fantastic. Why do students ignore the text's balance between the realistic and the marvelous? And even more puzzling, why has this tendency to fix on a particular meaning for Beloved been replicated by literary scholars, most of whom view Beloved as a ghost? 
This essay uses structuralist, post-structuralist, and reader-response theories of textuality to argue that the narration in Beloved creates a too close identification between the main characters' points of view and the point of view of the reader; the ultimate result is that many readers finish the text believing, like Sethe and Denver, that Beloved was the ghost of the dead child. From this point of view, the text's meaning is closed, totalized, finalized; and although the narrative voice reasserts itself to raise troubling questions about what Beloved really was, these questions are often ignored. But this essay also argues that Morrison is aware of this tendency, and that she puts the meaning of Beloved back into circulation with her next novel, Jazz (1992). …