Iser has an interesting moment on page 1676, a few pages into the essay. He's describing something like what Schklovsky called defamiliarization, considering it as a basic factor in literature, quoting Virginia Woolf on Jane Austen, to support. He gives a statement that could fall right into Russian formalism except for its emphasis on the readers' position:
What is missing from the apparently trivial scenes [in Jane Austen's reserved prose], the gaps arising out of the dialogue -- this is what stimulates the reader into filling the blanks with projections" (1676).
No surprises here and none to follow, but he does articulate the situation admirably:
What is concealed spurs the reader into action, but this action is also controlled by what is revealed; the explicit in its turn is transformed when the implicit has been brought to light" (1676).
Now, what's really really big here is that there's a tendency for many late-formalists to priviledge the "gaps" or the "open" text in discussion, particularly discussion of hypertext and end-of-millenium American poetry.
Now, Iser's explicit deserves examination. This could be what the text signifies directly, but that's subject to the criticisms Derrida levelled at Searle over the issue of constative and performative language. But since that explicit must be a re-creation that the reader carries in mind on the fly during reading, the explicit or elements thereof must contain something like what Kristeva describes as chora in Revolution in Poetic Language. That is, the system here would seem to be that the reader approaches with ideas relatively fixed by experience, and re-interprets the text in accord with these. The text draws the reader into activities in which the reader projects his or her own ideas into unforseen situations, evaluates the results, and eventually revaluates the original ideas based on the results.
The essential insight here for formalists would be that discourse must be partially new, partially old; partially open, partially closed.
The convergence of many different semioticians and fellow travellers on this kind of point reinforces my confidence in it greatly. There is another aspect to integrate, however. All these folks seem to treat semantics and WORD atomistically, and that cannot work. There has to be a range of syntactic responsiveness or variability, a range of meaning, within semantics that corresponds to the need for flexibility presented by syntax.
The best model I have for that is the one that has come out here in discussions of Chomsky and Deleuze and Guattari. The word is not defined, but is a cluster of spatio-temporal and electro-physical associations that fire sympathetically with associations as created by syntax. Subjectively, this amounts to this: a word summons various "associations," but these are not cleanly distinct, and the summons is not at all digital. In other words, as the call-for-meaning firing happens, the general area fires sympathetically. Strength of association may be determined by physical proximity, or by some other factor of pattern similarity. Given the kinds of slips we make, probably both are true. The mix of association available to linguistic impulse must be extremely broad, since we talk about most anything, and we do so more easily than we represent, say, sights as sounds or smells as sights. (I should qualify this last because there is a strong correspondence between sight, sound, and kinetics that allows for very fluid transfer and re-representation; I suspect that this will be because each is rigged to represent spatiotemporal relations and our sort of personal GPS).
Continuing. His referential field is close to what I mean by frame, a term I in part took from Joe Byron's description of actual physical frames while discussing Apocalypse Now. But I mean the data grouped to combine for meaning. Ongoing consideration means re-grouping this in various ways recursively. There is also in some sense a group of groupings, but this is probably done in almost the same way.
So let's to Iser's discussion of this.
As the reader's wandering viewpoint travels between all these segments [POV's of narrators, characters/focalizers, plot, and fictitious reader], its constant switching during the time flow of reading intertwines them, thus bringing forth a network of perspectives, withi which each perspective opens a view not onoly of others, but also of the intended imaginary object" (1677-1678).
OK, let's call imagary object something like the old structuralist sujet, since it's clearly not the fabula or the language-object.
He grants the blank three functions:
The first structural quality of the blank, then, is that it makes possible the oganization of a referential field of interacting textual segments projecting themselves one upon another
Alright, we can tell that neither the blank nor any element of text does this. So Iser must mean that the reader does this in interaction with the text. However, that also means that the actual activity is not the spatial rearrangement described here, so Iser's indulging in a metaphor, consciously or not. What gives? This appears to resemble what Chomsky describes in his recursive structures, only he has chosen to focus on subsentence-level combinations. One might assume that the process, which every commentator describes as recursive, recurses at sentence, paragraph, and other narrative levels as well, and that the levels play into one another as Deleuze describes as well.
So the referential field is the combination of elements that shall be referred to one another, the parts of the assembled pattern. And in the recursion, the codified-abbreviated results of the previously assembled pattern become one element in the following pattern-field.
Unfortunately, Iser drops his parallel structure and does not straightforwardly describe a second function. However, in the cluster of metaphors that follow, he does leave some pretty good clues.
. . . the fact that they [segments present in the field] are brought together highlights their affinities and their differences.
This is still the first function, as the reader may observe. But Iser continues.
This relationship gives rise to a tension taht has to be resolved, for, as Arnheim has observed in a more general context: 'It is one of the functions of the third dimension to come to the rescue when things get uncomfortable on the second" [cites Toward a Psychology of Art. Berkeley, '67]. The third dimension comes about when the segents of the referential field are given a common framework, which allows the reader to relate affinities and differences and so to grasp the patterns underlying the connections" (1678).
Iser's dimensions are those of comparison, not of space, though human brains do seem to use spatial metaphors to describe these kinds of things, and there may be some biological basis for it. (If so, that probably involves streams of related neurons or dendrites selected from neurons, planes of related neural columns, then spaces. But this will be more complex than a 3d comparison will easily allow, since it has to involve proximities of firings, and presumably some crossing of data for comparison.
So what we have is an element that relies, in its nature, on a previous comparison; these elements are compared and combined on another level of comparison. Iser presents this as though the act retained all of the original data. It probably does not, but re-represents it in a more economical form, as a single pattern-unit, not completely discrete from associations.
This last qualifier, not completely discrete from associations is crucial and new to my thinking. There's a tendency to treat all of these things as discrete units, and it's inaccurate. The unit does not have crisp boundaries, but remains suggestive. So we have not completely lost the various components implicit in the assembled pattern-unit, but they are not all equally under attention in the abbreviated, assembled pattern unit; more precisely, they are not all summoned at once, not all firing at once.
Now we come to the third and most decisive function of the blank. Once the segments have been connected and determinate relationship established,a referential field i formed which constitutes a particular reading moment, and which in turn has a discernable structure. The grouping of segments within the referential field comes about, as we have seen , by making the viewpoint switch between teh persepective segments" (1678).
OK, again Iser insists that this kind of combination is only of POV's as opposed to universal for various pattern-units, but I think we can fairly ignore that and retain something very like Iser's superstructure.
I'd like to make some still unwarranted guesses about this process. First, that we would indeed have comparisons built of comparisons and patterns built of patterns. But also, we should beware of assuming that a new comparison means a new level of pattern as suggested by Iser's metaphor of dimension. Principles of biologial economy suggest that patterning devices will be re-used to the greatest extent consistent with processing, and that processing will even be compromised to some extent. After all, the human skull kills and cripples mothers and children in childbirth, makes adults slow, gives us back and neck injuries, and so forth. So a lot of times, the next "dimension" must be accomplished by iterating the comparison in ways somewhat analogous to the ways dimensions might be expressed in binary code. (That is, string gives one value, string gives another, string says how to assemble).
Now, I should note that later on he describes his second level of blank as a vacancy, and relates that "blanks refer to suspended connectability in the text, vacancies refer to nonthematic segments within the referential field of the wandering viewpoint" (1679). So he's trying to fix these cognitive events on aspects of the text, probably more than a little too categorically. After all, why would the reader not simply compare other things? Also, there is as yet no specific mention of the reader's experience, which of course is a constant feed of data in this, since it constitutes all semantic information, among other things.
Among other things of interest here, it's very clear in Iser's construction that the process he's talking about could not happen were the reader to determine story events. The process would be considerably different. I strongly suspect that an investigation of those differences (which I don't think I can manage right now!) would reveal something of the differences between a reading and a game.