The conventions of narratological categories direct our expectations and interpretation as we read. But when these conventions are problematized, our interpretation becomes more theoretical, forcing us to contemplate why an author would choose to experiment with the category of, for example, narration. In Paul Auster’s
City of glass(1985), we have what appears to be a heterodiegetic, omniscient narrator who is mostly, but not fully, unproblematic. By the end, however, we discover that this narrator is (also?) homodiegetic. Auster’s breaking of the rules of traditional narration by having a narrator who is both hetero- and homodiegetic not only leaves readers in a quandary as to how to interpret the text, it also makes us realize how much we rely on the narrator for meaning. I propose to analyze the novel in order to explore the rhetorical strategy of Auster’s refusal to maintain a stable narrator. This analysis will illustrate how the category of narration prompts an unexamined trust in the teller and is therefore essential for our understanding of truth and meaning in narrative. I contend, in other words, that Auster’s experimentation with the category of narration in City of glassis key to the text’s insistence on a revised understanding of truth.