How do we make sense of stories? There's an excellent cognitive analysis of story comprehension in terms of frames by Manfred Jahn (University of Cologne, Germany) in the Winter 1997 issue of Poetics Today ("Frames, preferences, and the reading of third-person narratives: towards a cognitive narratology" 18 (4):441-68). His focus is the cognitive modeling of how third-person narratives are read, and his main references are Minsky's theory of frames, Jackendoff's concept of preference rules, Perry's theory of literary dynamics, and Sternberg's Proteus Principle.
The concept of a frame is central to cognitive theory, one of those ideas that name what is so basic that we normally have no reason to pay attention to it. Artificial intelligence research stumbled on it in the 1960s (McCarthy & Hayes 1969); Dennett (1984) marvels that such a fundamental problem of cognition had not been discovered earlier. But it is not until the steps of thinking need to be formalized in a set of computer programming instructions that the astounding range of implicit assumptions in quotidian thinking becomes evident. Not that our thinking really is formalized in the way it must be for computers; the mind, after all, is not a computer, only similar to it in certain ways. Yet it, too, must build and maintain frames that provide the scaffolding for interpretation. Jahn shows convincingly that an analysis in terms of frames can be applied to narrative poetics, so let me pick a simple kind of story to illustrate some elements of his theoretical apparatus.
Jokes typically operate by imposing a novel frame on an initially defined situation. What, the joker asks, does Clinton tell his wife after making love? "I'll be home in twenty minutes, honey." How does the joke work?
Jahn utilizes Jackendoff's notion of preference rules as a way of understanding frames. In this case, the implicit preference rule for a conversation is that the two interlocutors are in the same place at the same time. This rule is reinforced by the conjunction of "wife" and "making love," where the default again is that a man makes love to his wife. The joke relies on the listener making these routine inferences, though of course in fact they are not explicitly stated. They define what is normal, what goes without saying, and the joke works by violating this norm.
"I'll be home in twenty minutes, honey" initially creates an interpretive puzzle: it is incongruous in the framed situation. But according to Sternberg's Proteus Principle, the same literal statement can function differently according to how it is framed, and there is in principle no limit to the number of possible frames (450). This gives language an infinite versatility – but how do we avoid getting swamped contemplating this endless range? Strategies that rely on computations of Saussurean differences or Derridean deferrals are impracticable. There is typically a single target meaning – in the case of the joke, we need to "get it" by quickly locating the unexpected but determinate frame – and the fact is we make a beeline for it. How do we locate the preferred meaning?
Jahn cites Jackendoff's conversion into preference rules of Grice's conversational implicatures. In this case, the relevant rules are "Prefer to assume that the narrator is saying something relevant" and "Prefer to assume that the narrator is giving the right amount of information" (p. 447). Again, we assume that the act of telling the joke must rely on the listener adopting these preference rules -- in fact, the listener must assume that the joke-teller tacitly intends her to adopt the embedded preference rule that Clinton is acting Gricean, an example of the kind of embedded frames Jahn discusses (453).
The task, then, is to create a frame where Clinton tells his wife information that would be relevant to her in an economical way. Here, there is also a default solution: Hillary is home at the White House and Bill is in the city, telling her on the phone he is on his way home. Since he does not tell his wife what he's been doing, the Gricean implicature is that this is not something she would be interested in knowing -- as a default, we assume she assumes he is giving the right amount of information -- namely, say, that he's had a meeting, doing his job, that (since he calls her "honey") he is keen to see her, and is affirming, however routinely, his commitment to her.
What we have, then, is the presence of two overlapping but incompatible interpretive frames, where "Bill is at home with his wife" is being challenged by "Hillary is at the other end of the phone line." The former is still present in our minds, as Gricean listeners, carried over from our initial attempt to model the situation. The puzzle posed by the joke -- "What does Clinton tell his wife after making love" -- supplies a set of cues we utilize to construct a scenario from default values, and this frame is then preferentially used to make sense of the rest of the story. Jahn, following Minsky, Jackendoff, and Perry, calls this tendency of the initial frame to persist the primacy preference rule (457); it is in effect a meta-preference rule. The second part -- "Honey, I'll be home in twenty minutes" -- is most easily (preferentially) interpreted according to a novel frame (the wife expecting the husband to come home) with its own default values. How do we deal with this new frame? As Fludernik puts it, "Once the reader has established a prevalent perspective, he tends to persevere with it as long as possible" (Jahn 459).
This sets up a tension in the mind's attempt to model the event. Jokes are a kind of 'garden-path stories' in that they induce the listener to construct one frame initially, and then, in the punch line, supply cues that are incompatible with this initial frame, forcing a reinterpretation. This is dubbed the recency preference rule (457): novel information has the power to dictate an expensive reframing of the whole scenario. (The same effect is apparent in garden-path sentences, analyzed in detail by Jackendoff, such as "The horse raced past the barn fell.") Clinton cannot be making love with his wife if he is not home and she is waiting for him; the inference (on the adoption of the preference rule "Prefer to assume that the narrator believes what he intends to convey" (447)) is that he must have been making love with someone else.
This simple example illustrates the more general case of narrative. As Perry writes,
So what is so funny? Certain kinds of recency reinterpretations appear to be experienced as enjoyable in themselves, as already Freud pointed out. Why this should be so is a fruitful area of investigation. In this specific instance, one factor is that the joke relies on providing surprising insider-information in an instance of cheating. Vilayanur Ramachandran (1996) proposes that in our evolutionary history, laughter arose as a signal of a false alarm. Since cheating in a social exchange situation is potentially serious, the fact that the story is framed as fiction allows us to see it as humorous: similar to or nearly a threatening situation, but not actually so.
Manfred Jahn's article is not itself hilarious, but I hope I have shown how the conceptual framework he presents and elaborates can be fruitfully utilized for the analysis of Hillary jokes, which are a short and concise form of narrative. In addition to the elements I have highlighted, the piece contains a great treatment of point of view. The whole issue of Poetics Today is devoted to Narrative Poetics, and contains a string of interesting articles and reviews.