The project of Richard Gerrig's Experiencing Narrative Worlds (1993) is, as the cover blurb proclaims, to "integrate... insights from cognitive psychology and from research in linguistics, philosophy, and literary criticism to offer new analyses of some classic problems in the study of narrative." The opening gambit, which organizes the book as a whole, is the observation that people often talk about reading in terms of being "transported" elsewhere, to the "world" suggested and depicted by a narrative text. And so he wants to take these metaphors and run with them, to use them as the organizing structure for his extended discussion of what the insights of psycholinguistic language-processing research reveal about literary appreciation and the nature of narrative.

In his first chapter, Gerrig unpacks these metaphors, outlining in loving detail the conceptual mappings that undergird them, explaining quite properly along the way, by reference to the work of George Lakoff and Mark Johnson (1980), that metaphors are neither trivial nor arbitrary. It's quite true -- metaphors are not just a fancy way of speaking. They reflect extensive conceptual mappings between domains, and attention to what metaphors people use in talking about one domain can be extraordinarily useful in revealing all kinds of things about the way that domain is structured, used, and experienced in their minds. And I can appreciate that one of the things this observation about metaphors licenses you to do is to take metaphors seriously as a useful (and indeed, an unavoidable) way to talk about your subject.

So Gerrig uses this introduction to explain the motivation for his decision to talk about narrative in terms of "worlds" and reading in terms of "transport." And the metaphors are indeed useful in highlighting some of the elements of language processing that he wants to emphasize; they further his project of moving our focus from the structural analysis of what an example of narrative language would consist of (cf. Labov's (1972 and elsewhere) definition of narrative) to the way narrative manifests in meaning construction. As Fauconnier (1997: 65-66, in passim) points out, language typically underspecifies, acting as a relatively bare prompt that can be fleshed out cognitively with a variety of potential conceptual configurations that may be constrained to a greater or lesser degree by various pragmatic considerations. Utterances that are not patently narrative in structure can nonetheless be narrative in experience, in that they prompt the hearer or viewer or reader to mentally construct a narrative as part of the process of understanding and/or appreciating that utterance.

I should stop and note that I really, really like Gerrig's work. But now here's the part where I quibble: I found myself wishing that this whole metaphor discussion had gone a different way, more along the lines of the Reddy (1979) article about the conduit metaphor (which I desperately wish were online so that I could link to it here; Lakoff and Johnson's summary of his argument is about halfway through this selection of excerpts). I wanted to see as much emphasis on what these metaphors mask and distort as on what they illuminate. Gerrig does, indeed, explain ways in which the metaphor of the so-called "narrative world" -- which has, after all, been used in lit crit for a long time, and explicitly identified with the theoretical construct of the "possible world" from predicate logic -- maps poorly to the current state of our knowledge of how meaning construction works. But these explanations are presented as caveats, rather than the main subject of the chapter.

This bothered me because for all the usefulness of Gerrig's organizing structure, and as much as I (a) like organizing conceits and think they can be genuinely illuminating, and (b) talk about "the world of the text" to my own students all the time, I still felt uncomfortable about the way the "narrative world" metaphor occludes what I consider one of the most useful insights of cognitive linguistics in the last 20 years: You don't need to postulate logically complete possible worlds to account for the way people construct meaning. Possible worlds are metaphysical constructs, not cognitive ones. In fact, at least if you're convinced by the arguments for mental spaces theory in Fauconnier 1985 and 1997, as I am, a great many otherwise difficult-to-account-for linguistic phenomena can be explained quite clearly and directly under a theory in which people create mental models in response to linguistic prompts, but ones that are importantly unlike possible worlds. These representations are both partial and underspecified. (Actually, recent research in psycholinguistics suggests that mental models generated in reponse to linguistic prompts may be even less elaborated than Fauconnier suggests.) And the trouble is that an approach that ultimately works to embrace these metaphors rather than to problematize them leads to some unfortunate oversimplifications down the line.

Gerrig is careful and subtle in his discussion of the experimental evidence for ways in which reading seems to induce a disconnection between "real world" knowledge and the reader's apprehension of and reasoning about the "narrative world." But it makes me acutely unhappy when I see this area of discussion summarized in Poetics Today¹ thus:

One might formulate as follows a provisional criterion for a theory of fiction: it must accounct for the role played by all three components in the reader's relocation from his or her current deictic center...

[Okay, though I'm already uncomfortable with the breeziness of calling this a "relocation" of the reader rather than a projection of her viewpoint, say, but sure.]

...to an alternative possible world.

[Here's the bad part:]

That alternative world will count as actual for as long as the reader remains immersed, phenomenologically speaking, in the fiction itself.

What? Might we narrow down just a bit what on earth "count as actual" is supposed to mean here? Count how? To whom? In what regards? To what degree? Nope, sorry, no specifics today. Try again tomorrow. If you read this review generously, with a good knowledge of the nuances of what we do and don't know about the cognitive processes the author is writing about, it might work as an acceptable shorthand for the kinds of experimental phenomena Gerrig reviews throughout his book. But as a proposal for "a provisional criterion for a theory of fiction," it stinks. Because the interesting thing about all these phenomena is not merely that they exist -- that reading is immersive, that there is a level on which people are demonstrably simulating the experience of witnessing things from the domain of the narrative, sometimes even overriding their direct experience of their own surroundings and experience. It's that they are immersive, but also non-immersive. They do their trick partially, incompletely, in such a way that the reader is not generally deluded about the distinction between representation and reality.² Skipping out on either half of this equation is a mistake.

Finally, a side note: Is there much on this earth more annoying than reading slapdash reviews of books you know well and respect? (Answer: Well, sure, because surely when it is a book you wrote yourself, the experience is multiplied a hundredfold.) It seems that they are bound to include little dashed-off remarks designed to make my hackles leap off my... wherever it is that hackles are supposed to be. A little discursive footnote in a review of something else³ just made me see red: "Fictionality is a crucial category for literature; but judging from the conspicuous omission or slighting of fictionality in most cognitive studies of reading (for instance Gerrig (1993), who conflates 'narrative' and 'fictional'), cognitivism is not ready to cope with this complex issue and our response to it." Oh, snap! Meanwhile, as it happens, Gerrig is not in fact blithely conflating "narrative" and "fictional," but making explicit claims about phenomena that people have treated as obtaining only for fictional narratives, but which his research indicates hold true for any kind of representation of narratives, whether fictional or not. Harrumph. I suppose I can look forward to a long career of inadvertently doing this same thing to other people, and thus the great cycle of life continues.