« The arming of the hero | Orals guidelines »
Today I was reading Anna Kavan's Ice (which reminds me interestingly of both -- I want a single word that means "both," but for more than two items -- John Fowles' The Magus, Haruki Murakami's A Wild Sheep Chase, and Samuel Delany) thinking about a discussion that came up in the Grammar of Narrative seminar I took last year. Brian Richardson came in as a guest lecturer and talked about narratology and the problem of postmodern novels that feature deeply unresolvable fabulae, or which place the very concept of the fabula in question.
Now, the fabula, as every student of narratology knows, is the fancy term for the apparent "real" thread of events that underlies a narrative. It's distinct from the narrated events, and its events and elements don't have to be explicitly denoted or directly referenced in a text, though they may be. For example, when Stephen Dedalus, in the Proteus chapter of Ulysses, embarks on an extended riff in which he goes to stay with his aunt and uncle, it seems at first as if he is really going there, but the reader soon realizes that he is "really" just thinking about what would have happened if he had. The sequence of events in which Stephen walks along the beach, contemplating the ineluctable modality of the visible, is the fabula of the scene.
But not all fabulae are so easily delineated. In some cases, there may seem to be a stable fabula -- and the reader has the experience of trying to figure out what it is -- but incomplete evidence for all the details of that fabula in the text. The narrator of Pale Fire, for example, tells us a story in which Shade's titular poem references the narrator's own circumstances, interests, and pet concerns. The reader will quickly recognize that this version of events is not the "right" one; but what is the correct interpretation is hazy in many respects (and debates over how best to solve it make a nice little cottage industry for Nabokov scholars). Cases like this one don't present a serious problem for narratologists à la Genette; the concept of the fabula is in fact key to explaining the nature of the puzzle posed for readers of Pale Fire, which rests on the distinction between what they are told and what they are trying to deduce about the unnarrated events of the fictional world.
Brian wanted to talk about more difficult cases, in which the artifice of the sujet overwhelms any representational coherence of the fabula. I can't remember which postmodern novel he was using as his example -- I need to remember to ask him next time I talk to him -- but it had a fabula that was profoundly unstable, and even, one might possibly claim, the notion of its having a fabula at all was basically meaningless. Different versions of events superceded one another and were presented in a way that suggested that they never could be resolved, that the artificiality and unknowability went all the way down. Invisible Cities is another fine example -- the locally coherent sections fit together poetically, but no fabula never resolves and frankly, why worry your pretty little head about it?
That's a fairly standard postmodern gambit, of course, the idea that a narrative need not or cannot add up to an ultimately coherent whole on that level, though the early nineteenth century classic and favorite source material for hippies The Saragossa Manuscript gestures toward the same thing, a century and a half ahead of time. A simpler case of an unresolvable fabula might be something like The French Lieutenant's Woman. And in the absurdist tradition of, say, Beckett and Pinter, the fabula seems not only unstable or unsolvable, but even fundamentally beside the point. In any case, Brian was bringing up this class of narratives as something that presents a problem for classical narratology, because of the way it doesn't rely on that fundamental reading strategy of searching for a coherent fabula, instead expecting and encouraging the reader to suspend, abandon, or never engage that conceptual framework.
Our class, freshly steeped in Herb Clark and other interactional ways of thinking about discourse and narrative, tended to take this kind of narrative as fundamentally derivative (a stance for which there is a fairly unassailable case to be made on both the historical and experiential evidence), and its reading thus best explained as involving the experience of trying to contemplate a fabula and eventually coming to the point where one is led to suspend that search. And furthermore, the reader is even then presumably still aware of the author implied by the text, and able to contemplate and rely upon the meta-fabula of the writer's laboring to create an evocative and self-contradicting liteary artifact. There's something to be said for this view of things, I still think, though see my entry on Jerry Hobbes' Literature and Cognition for an argument against the assumption that readers necessarily construct an author-interlocutor persona in their construction/understanding of a text.
But today I was moved to think about these unresolvable fabula in conjunction with a different area of research, the status of predictive inferences in discourse comprehension. The degree to which readers actually understand discourse (including written literary narratives) by means of appeal to mental models as they proceed, though often taken for granted in some branches of cognitive science, is a hotly contested question in psycholinguistics, particularly among researchers on discourse processing. In their overview of research on inferencing in discourse processing, Graesser, Singer and Trabasso (1994: 383-385) identify six broad theoretical positions regarding the online generation of inferences during narrative comprehension, arranged along a spectrum:
- Explicit Textbase. In this view, language users construct no knowledge-based inferences at all during comprehension. This position is a straw man, in the estimation of Graesser et al, but interesting to consider conceptually and to a certain extent implicitly endorsed by very early theories of discourse processing, in that they focused exclusively on explicit text in their discussions of the representation of meaning.
- The Minimalist Hypothesis. This hypothesis is proposed in McKoon and Ratcliff (1992). It covers only those inferences that are generated and encoded automatically during comprehension, and holds that the only inferences that are encoded in this way are those based on information easily available in working memory and those necessary for establishing local text coherence: resolving reference, case structure role assignment, and causal antecedents.
- Current-State Selection Strategy. Like the minimalist hypothesis, this model proposes that representations are generated solely when needed for local coherence. However, its definition of local coherence is predicated on causality, and so in addition to predicting that people generate online inferences about reference, case structure roles, and causal antecedents, CSS predicts that people make online inferences about superordinate goals.
- The Constructionist Theory. In the constructionist theory advocated by Graesser et al, only a subset of inferences are taken to be made online. However, unlike the other theories outlined by Graesser et al, these inferences are assumed to be generated based on the principle of search after meaning: the reader constructs a meaning representation in accord with the reader's goals, she attempts to construct meaning that is coherent globally, as well as locally, and she makes use of na´ve theories of both physical and psychological causality to attempt to explain why states, events, and actions are mentioned in the discourse.
- Prediction-Substantiation. This model predicts that reading is driven not only by the search for explanation, but also by the generation of expectations. Expectations are automatically formulated whenever a higher-order knowledge structure is activated. So if a story activates a REVENGE theme, for example, and specifies that character A hurts character B, the reader will form broad expectations about what character B will do to character A.
- Promiscuous Inference Generation. The straw-man position on the other end of the spectrum from the explicit textbase position: that all kinds of inferences are generated online, resulting in a meaning representation that is complete, life-like, and exhaustive. Graesser et al claim (and I agree) that McKoon and Ratcliff (1992) suggests unfairly that a constructionist theory would embrace such a hypothesis.
An insight worth further thought?
Graesser, A. C., Singer, M., and Trabasso, T. (1994). Constructing inferences during narrative text comprehension. Psychological Review 101: 371-395.
McKoon, G. and Ratcliff, R. (1992) Inference During Reading. Psychological Review 99 (3): 440-466.
all about the log | discussion | genre references | language | literature | narrative gaps | planning |
Movable Type 3.2