The real distinction between Chatman, Kozloff, Levinson and others who find the concept of the film narrator to be useful and those who like Bordwell do not may derive not from the nature of film narrative, but from the nature of narrative in general. Shlomith Rimmon-Kenan remarks that the derivation of a similar fabula/story from different versions—"the abstraction of the same story from different media"—should not be ascribed to some abstract, latent entity that precedes its manifestation in any medium. Instead, that derivation results from "a commonality of potentialities and constraints among [those media], thus stressing the dependence rather than the independence of the story upon the medium" ("How the Model Neglects the Medium" 160). Rimmon-Kenan goes on to suggest that "a succession of events becomes a story only when it is verbalized" and that "narratological studies of ‘story’ are in fact studies of verbal accounts (by the narratologist) of (sometimes) non-verbal events." Chronology and causality, she continues, "are not ‘things in the world’ but concepts we impose on the world in order to apprehend it, and how do we impose such linkages if not through language?" (160-161).

Rimmon-Kenan’s comments are brief and not meant to be definitive, but they do suggest that the verbalization of narrative through language—what critics do when they talk about film—entails the introduction of communicative concepts, such as "narrator" and even "author," that are not necessary or even relevant to all media. Such concepts are part of the schemata that we bring to the understanding of narrative, whether in literature or some other medium. Similarly Mieke Bal contends (opposite to Levinson’s view) that "By definition, a ‘third-person’ narrator does not exist: any time there is narrating, there is a narrating subject, one that to all intents and purposes is always in the ‘first person.’ The ‘person’ of the narrator . . . can be distinguished only in terms of his presence or absence in the narrative at the level in question" (237, my emphasis) [9].

If we take Bal’s comments to apply to print narrative, and if she and Rimmon-Kenan are correct, then the specific forms of narration—which may or may not include narrators as such—are functions of their media. As a function of the text, narrators are not actual agents that bring the narrative events to the reader—or to the perceiver in other media. They are an aspect of causality as one of those concepts that—to paraphrase Rimmon-Kenan—we impose on the narrative world in order to apprehend it. Presented with a narrative, we will presume—even need—the presence of a teller (or even an implied author) as much as we need to assume that time moves in one direction (whatever detours it may take in a text) and that events are linked by cause and effect. A fair amount of "experimental" and avant-garde fiction actually tries to call such assumptions into question by removing cues or offering ambiguous or contradictory cues to our presumptions. In film, for example, Un Chien Andalou continually offers hints that it is actually conveying a narrative but undercuts those hints with various contradictions of time, space, causality and narration (for instance, the intertitles that offer meaningless time and date cues).

In this sense, Bordwell is correct in contending that to posit an overall narrator for a film or to look for a single creative source will not necessarily tell us much of importance about the film itself—it may indeed lead us down narrowly defined paths that ignore the substance of the medium and the actual film. However, if we take "narrative" to be inclusive of both diegesis and mimesis—of "telling" and "showing"—then the presence or lack of a narrator is a relative, not an absolute, criterion. The lack of a narrator in most theater is simply a function of that medium, just as the presence of a narrator in print fiction is of that medium. Film—which has the capacity to "show" and "tell"—may sometimes function as though it has one or more narrators but does not always need to do so. As spectators, though, we sometimes impose the concepts of "narrator" and "implied filmmaker" on the movie in order to make sense of it (as when we talk about how "they" made this or that mistake when we criticize a bad film). We may also be prompted by the film itself, which offers cues of one kind or another because its (real) makers also proceed from such assumptions of causality or because they want to toy with the audience’s expectations. And, of course, we may see cues where none were actually "intended." Regardless of actual "intention," though, it is more meaningful to see when, how and why a film evokes a sense of its own agency than to look for those intentional agents from the outset. Some examples of such evocation will be examined below.

This conception of agency is also consistent with Edward Branigan’s definition of narrative in film: "the overall regulation and distribution of knowledge which determines how and when the spectator acquires knowledge, that is, how the spectator is able to know what he or she comes to know in a narrative. A typical description of the spectator’s ‘position’ of knowledge includes the invention of (sometimes tacit) speakers, presenters, listeners, and watchers who are in a (spatial and temporal) position to know, and to make use of one or more disparities of knowledge. Such ‘persons’ are convenient fictions which serve to mark how the field of knowledge is being divided at a particular time" (Narrative Comprehension and Film 76). Branigan’s concept of agents as "convenient fictions" is also consistent with Bordwell’s claim that film is not communication as such but can mimic it.

To summarize, there are two different answers to Chatman’s vexing question, Who is responsible for the "regulation and distribution of knowledge" in a film narrative? The first answer is, the creator(s) of the narrative text. The second is, the text itself, through the reader’s activation of cues within the text that may or may not be the deliberate result of the creators’ efforts. The first answer is practical, simple, and obvious. The problem with it is that we rarely, if ever, have full access to the creative process that the creator(s) engaged in, whether that process is relevant or not. Sometimes—maybe always on some level—even the narrative’s creators may be unaware of that process. The creators’ intentions then must be assumed by the perceiver—based on the cues offered by the text, leading to the second answer. That answer, though, still begs the question of how readers know what cues to respond to. Some agency—presumably the implied author and/or narrator, depending on how those agents are defined—must have created those cues to provoke particular reactions. Yet, as we have seen, both "implied author" and "narrator" may actually be functions of the very texts that they are supposed to create.

How can this paradox be resolved? The final answer has to be that there is no single answer, except to the extent that both answers above entail the active engagement of the perceiver, who actively seeks cues in the text as he or she seeks them in the world. Whatever the actual circumstances of the narrative’s creation, a perceiver will—to one extent or another—assume the involvement of some agency in its creation and diffusion. If we often try to make life and the world meaningful by seeking cues to its intelligibility (or even proof of its non-intelligibility), we almost always expect fictional texts to be intelligible. As Edward Branigan puts it, "The apprehension of a system of intelligibility in a specific text is the apprehension of a human quality in the text" (Point of View in the Cinema 66, Branigan’s emphasis). "With this approach," Branigan continues, "it is possible to speak of coherence or intelligibility without taking the next step and implying a consciousness—a corporeal though invisible body blended with the text which must be searched out and reconstituted" (67). In fact, an apparent lack of coherence will itself serve as a cue to the perceiver—of the need to search more diligently for a system of intelligibility, of the artistic failure of the work (and perhaps of the implied author), or of the need to seek criteria other than intelligibility (as we have to do with Un Chien Andalou). The "implied author" and the "film narrator" are important to the viewer only to the degree that those schemata seem to be cued by the text as part of a "system of intelligibility."

Such an approach also avoids appeals to a "vulgar" intentionality. As V.F. Perkins remarks, "If the relationships established in a film are significant, it makes no difference to the spectator how they came, or were brought about, or to what extent their significance was intended. Provided that a film has its own unity, it seems unimportant whether the unity was evolved through cooperation and compromise with the production team or conceived by one man and imposed on his collaborators" (173). It is of historical and social importance, perhaps, that The Cabinet of Dr. Caligari had a frame narrative superimposed on it after its primary filming. That knowledge has no effect on the comprehension of the film narrative itself, even though it may have profound implications for the ultimate affective or psychological or social or aesthetic reaction to the film. In fact, deconstruction, feminist and psychoanalytical criticism, and other forms of reading "against the grain" are ways of demonstrating at what cost the "unity" of a film may be perceived.