Draft version: November 15th 1997
Abstract. Postmodern, liberationist claims of hypertext theorists -- mainly Bolter, Landow, and Moulthrop -- are assessed in relation to literary reading. It is argued that such claims are based on a misrepresentation of existing reading practices, and that to relocate reading to the spatialized, non-linear, unstable environment they celebrate jeopardizes the kind of reading that is characteristic of literary response.
Trivializing the Word: Hypertext, Postmodernism, and Reading
The textual world is no longer what it was. The new world of cyberspace and hypertext has freed text from the constraints of the pre-electronic medium of print. Stuart Moulthrop is one of several recent theorists who speaks for this new vision: what it means, he urges, is "that we must fundamentally re-think our position as subjects of electronic textuality" (Moulthrop, Edge). The enthronement of hypertext as the writing space of the future has been accomplished: now we must be prepared to accept and work through its implications. George Landow adopts a similar tone in the opening paragraph of Hypertext: "we must abandon conceptual systems founded upon ideas of center, margin, hierarchy, and linearity and replace them with ones of multilinearity, nodes, links, and networks" (Landow 2).
The admonitory tone suggests that we (I include myself) may not be yet quite at ease in the hypertextual world that its advocates urge upon us. The claim seems to be an appealing one. Associated with a liberationist and democratizing rhetoric that promises to realize the the true potential of text, hitherto locked within the prison-house of the book, it simultaneously announces that the abstractions of postmodern theory are about to be incarnated and tested on the electronic pulse. A vision as comprehensive as this has, not surprisingly, met resistance -- the work of Sven Birkerts (1994) or Myron C. Tuman (1992) comes to mind. In this article I want to articulate my own sense of resistance to hypertext advocacy by focusing in particular on problems inherent in the view of reading proffered in the hypertext literature. Although the hypertext medium may have much to offer, both aesthetically and pedagogically, the universalizing of the claims made on its behalf, and the polarization involved in deprecating (often explicitly) all previous forms of printed literature, obscures the central issue: what it means to read, and how far reading practices change in the context of a new medium such as hypertext.
Neither hypertext theorists nor the postmodern critics to whom they appeal have examined actual reading, except putatively or by extrapolation from their own reading processes (cf. Miall and Kuiken, in press), nor do the writings I will mention show signs of serious engagement with the empirical evaluation of hypertext by its users (e.g., the work reviewed by Jean-François Rouet and Jarmo J. Levonen, 1996). Thus accounts of reading in the hypertext literature cannot be relied upon to describe accurately either pre-electronic or post-electronic modes of reading. This shortcoming can be brought into focus through four specific issues which I will examine in detail. First, relocating text to "cyberspace" or the "writing space" makes it difficult to conceive of reading except in a two-dimensional spatial model. Second, electronic text is said to be inherently unstable, a claim that undermines the act of reading while also overlooking the revisionary practices of pre-electronic authors. Third, electronic culture may modify our subjectivity, but not to the extent that we change fundamentally as "subjects"; this obscures continuities in reading practices across media. And fourth, being freed to follow a network of linked texts is not a substitute for the process of individual literary reading.
More generally, advocacy statements such as Moulthrop's reproduce an assumption that was common -- even an article of faith -- among cognitive scientists of thirty years ago: that the mind is to be understood in terms of the computer, that is, as a set of information processing mechanisms, probably based on association (cf. the trenchant critique of this assumption by Dillon, 1996). As a working hypothesis this, in its time, inspired some valuable research, including the initiation of a tradition of discourse studies that has offered significant insights into the reading process (although not, so far, that of literary reading: Miall and Kuiken, 1994). At a time when newer models of the mind, notably connectionism and those originating within neuropsychology (Miall, 1995), are superseding the computer model, we might be sceptical about embracing hypertext as another model of mental functioning. I will argue that understanding of literary reading is rendered impossible within a framework restricted to the capacity of the present-day computer. The ability to read within the hypertext medium is a new phenomenon, and one with some potential advantages; but the claims being made for it suggest that it is antipathetic to literary reception.
Hypertext, it should be noted, takes many forms. It is frequently used as an online substitute for the printed text: text sections are arranged within a hierarchical structure analogous to a table of contents. The internet pages of commercial firms, universities, or government organizations are invariably structured in this way. Cross-links may enable the reader to jump between non-adjacent sections, but this is equivalent to cross-referencing in the printed text. To the advocates of hypertext such formalized structures are of little interest: the most important uses of the hypertext medium are not those built with a tree structure but those that are interlinked like a web or network with no superordinate structure or single, predetermined logical sequence. Once launched, the ordering of text sections is driven by reader choice, based on the links available from the current text. In contrast to the linear sequence presupposed by a hierarchical organization, a network hypertext is self-navigating. It is this kind of hypertext that will be examined below. It has been used for both literary and non-literary purposes.
Although hypertext theorists seem principally to contemplate hypertext fiction as exemplary of their claims, this article will not focus on the fiction as such (for discussions of its merits see the excellent bibliography of Michael Shumate): rather, my aim is to question the imperializing implications of the rhetoric of hypertext theory itself. I will focus primarily on the writings of three authors who have been prominent in such discussions: Jay David Bolter, George Landow, and Stuart Moulthrop.
* * *
An emphasis on the spatial characterizes all recent rhetoric about hypertext. Derived from the foregrounding of the visual components of writing by the computer screen, it is proposed as a site of opposition to the linear rigidity of the printed page. Thus it usually connotes a liberationist ideology of text (cf. Dillon's critique of this as a myth, 29-31). The change in the interaction between reader and text that it produces, however, appears to bring significant and possibly disabling constraints that call into question the liberationist premise. I say appears, since there is little reliable evidence to enable us to assess the significance of hypertext reading (cf. Douglas, 1997; studies of the pedagogical implications of hypertext are more common: see Rouet and Levonen 1996). The argument that follows is derived primarily from the claims of hypertext theorists themselves, and from insights that can be drawn from the few empirical studies of literary reading that are available -- narrowly focused though these are, perhaps necessarily so, given the very complex nature of the phenomenon being examined.
In what follows I elaborate the following arguments. First, I suggest that the iconicity of spatialized text is likely to restrict reading to shallow, associative forms of processing; second, that the foregrounding of links between textual sections or nodes encourages shallow reading, or "surfing"; third, that the choice of multiple pathways through hypertextual space provides only an illusion of reader emancipation; and, fourth, that the more the reader's attention is given to the spatial qualities of text, such as the three aspects just enumerated, the less feasible becomes any commitment to non-spatial dimensions of reading, such as literary context, personal imagery, feelings, or self-reference.
The appeal to spatial principles is shown by Landow's reference to Derrida's call for "a new form of hieroglyphic writing" which would escape "the constraints of linearity" (43). Similarly, Bolter notes that "Computer writing is primarily visual, rather than oral"; it "gives a renewed prominence to the long discredited art of writing with pictures" (Bolter 45-6). Bolter's discussion includes analysis of an Ojibwa pictogram of 1858 which, as he notes, provides no indication whether it should be read from right to left or left to right. In a basic sense this mirrors self-navigating hypertext itself in its most common instantiations: texts can be read in any order, links can be traced either forwards or backwards (that is, there is no forwards that is not purely relative), and no link has priority over any other link (cf. Rosenberg 280). In this framework text sections, I will argue, are forced towards the status of icons.
The icon is a common feature of our present-day graphical computer screen, whether Macintosh or Windows. The icon offers a miniature representation of a program or other function that can be executed by the user, but this representation is arbitrary (as a user I can change it for another): it cannot be operated on directly by the user, it is discontinuous with the program that it is used to launch, and its fixed pictogram has only a coincidental relationship with the processes I carry out through the program. For example, the word processor I am now using, Microsoft Word, is launched from an icon of the letter W. But it would be naive indeed to think that I could produce only Ws with the program. By placing texts within a hypertext system we invite the risk that they will be treated iconically.
Bolter's commitment to the iconic is shown by his account of the electronic page with its multiple windows. The simple printed page with its single block of text offers no distracting competition for the reader's attention. In contrast,
A magazine or newspaper is . . . closer in spirit to the topographic writing space of the computer . . . Larger units of text together with images can be isolated on the computer screen. The screen itself becomes a magazine page in which units even rearrange themselves to meet various needs. (69)
The magazine page as an example of reading is in one respect less challenging for the reader than a hypertext: a glance at each box on a magazine page enables us to anticipate whether directing attention to the text within it is likely to be worthwhile. But since in most hypertexts the nodes linked to the current text are off-screen, and little or no information about the nodes is offered, anticipation is disabled; thus the choice of which link to pursue becomes either problematic or arbitrary. Either I spend time attempting to reflect on the relationship between the current text and the one I have just left (but this places heavy demands on memory if the previous text is now off-screen), or I accept that the present text is no more meaningful than any other text to which I might have moved, so that the quality of attention I give to it is likely to be reduced. In either case, further links embedded in the present text beckon me on.
The problem of link navigation has been discussed by George Landow ("Rules," 1991), who formulated a set of "rules" for facilitating a reader's ability to navigate a hypertext knowledgeably and with anticipation, a rhetoric of arrival and departure. But to the extent that Landow's rules for hypertext linking are accepted, so the iconic or visual qualities of hypertext praised by Bolter are diluted. Where Landow's rules put back in place the inherent structure of the printed text by signalling the type and quality of the relationship between one text and another, the iconicity of texts linked without such information leaves the reader to infer a relationship. As there is no certainty that the reader will guess what the author had in mind (if anything), inferences made by the reader are likely to be more or less arbitrary and of short duration; moreover, the primary task now becomes that of reading the current text, not considering its relation to the previous one. Thus from the perspective of the reader the inherent tendency of hypertext is, paradoxically, to disconnect text sections, not to connect them (cf. Landow, Hypertext 54). The end point at which hypertext arrives is, in Stuart Moulthrop's term, "breakdown" (Moulthrop, 1995). The reader may continue to read; but the behavior that ensues we have learned to think of as "surfing."
The emphasis on the spatial (shown by the title of Bolter's book, Writing Space) originates with pre-computer literary theorists, notably Roland Barthes. In S/Z the writerly text is figured as a network; "we gain access to it by several entrances"; "the codes it mobilizes extend as far as the eye can reach, they are indeterminable" (Barthes 5-6). In the view of Bolter or Landow Barthes is a hypertextualist avant le lettre. Thus, Landow observes, "hypertext creates an almost embarrassingly literal embodiment of a principle that had seemed particularly abstract and difficult when read from the vantage point of print" (Landow 53). What is more embarrassing, perhaps, is the literalization of arguments that Barthes proposed figuratively. Where Barthes, with great subtlety, reveals what he terms the "writerly" text within Balzac, pointing to a remarkable array of reading operations to which we are invited in construing this narrative, the hypertext equivalent described by Bolter or Landow imprisons the reader within a predetermined set of operations that preempts the writerly response in favour of controlling the reader.
For Bolter, Derrida's Glas is an example of a hypertext in advance of its time. As the reader scans the different apparently unrelated blocks of text, "connections seem to be there, as words and sentence fragments refer the reader back and forth between Hegel and Genet" (116). Thus, he continues, "Glas belongs in the electronic medium, where . . . any relationships between textual elements can float to the surface; the computer invites the writer to reveal the inner structure in the appearance and the behavior of the text" (117). But this, precisely, is what Glas stops short of doing: once the links are specified, the array of potentially infinite connections implicit in Derrida's text is eliminated. To what extent the connections invoked by Glas could even be specified in the writing space is not considered by Bolter.
The computer framework thus places more limitations on writing (and reading) than does conventional printed text. Wordsworth's "spontaneous overflow of powerful feelings" would be rendered impossible by the computer, according to Bolter. The computer requires that text "must be planned and structured by the author" (153).
The pre-eminence given to the spatial in hypertext theory is thus at odds with the liberationist ideology, and in the end self-confuting. The choice of multiple pathways through hypertextual space provides only an illusion of reader emancipation. More seriously, it privileges the visual in the reader's response in ways that return us to outmoded eighteenth-century accounts of literature from Addison to Kames, with their elevation to a critical principle of Horace's ut pictura poesis (poetry tends towards painting; cf. John Tolva, 1997, who makes an analogous argument for hypertext). Both Edmund Burke (Enquiry 55-58) and the Romantic writers, such as Coleridge and Wordsworth, argued strenuously against this conception. Wordsworth, for example, spoke of a stage in his own approach to nature when "the eye was master of the heart," the visual organ, he added, being, "The most despotic of our senses" (Prelude 1805, XI.172-4). It must be asked whether the systematic discounting of non-spatial dimensions of reading, such as personal memory or literary context, is intrinsic to hypertext reading -- whether this is a necessary concomitant of the ludic, uncommitted ambience of hypertext in Bolter's account. I will return to this issue later when I consider the disembodied aspect of computer reading.
* * *
The lack of commitment to reading, however, appears to be driven by another aspect of hypertext writing that is singled out by hypertext theorists as a signifying property of the medium: electronic text is said to be inherently unstable. Bolter's statement is typical:
Electronic text is the first text in which the elements of meaning, of structure, and of visual display are fundamentally unstable. . . . All information, all data, in the computer world is a kind of controlled movement, and so the natural inclination of computer writing is to change, to grow, and finally to disappear. (31)
This is an odd claim in several respects. Bolter's ascription of a "natural inclination" to computer writing might seem a trivial overstatement, but similar claims are common among hypertext theorists. Richard Grusin (1996), who cites a score of additional examples, including several from Bolter's book, points out that after the much-heralded death of the author initiated by Foucault and Barthes, agency for writing is transferred to discourse. Translated into the electronic domain, this results in a kind of technological determinism. It also calls up fleetingly the organic analogy of "to change, to grow, and finally to disappear," which is at odds with Bolter's otherwise relentless rejection of Romantic poetics (as the example of Wordsworth I noted earlier shows).
The ascription of agency to electronic writing, and its resulting instability, appear to be an essential feature of Bolter's view: "Elements in the electronic writing space are not simply chaotic; they are instead in a perpetual state of reorganization. They form patterns, constellations, which are in constant danger of breaking down and combining into new patterns" (9). The claim is misleading in several respects. First, print text is not necessarily stable either. The "frozen structure of the printed page" (Bolter 21) is subject to the pressures of change, perhaps as much as any electronic text, as a superficial glance at the history of scholarly editing will show. The practice of many writers, similarly, will provide numerous examples of texts that were subjected to revision, both before and after they reached print. Stillinger, for example, in Coleridge and Textual Instability (1994) points to the ten different versions of Coleridge's poem "Frost at Midnight" produced by the author. Which of these is the "correct" one?
More important, if a text is in "a perpetual state of reorganization" (it is not clear where Bolter would have us look to see this process in action), then the question of agency is not a trivial one. A text that changed at random would hardly capture our attention for long. The quality of attention we give to, say, the seventh version of Coleridge's poem is a product of our engagement with the debate that poem offers us as readers. If Coleridge has revised a word, we care about this because it may impact on our understanding. His change of "dead calm" to "deep calm" after 1817, for example, significantly affects the meaning of the whole poem. If such words permutated by chance, like printer's errors, we would spend little time considering them.
In other words, the verbal stability of a text determines the quality of attention we give to it, principally because literary texts typically sustain a richness of interpretive possibilities at the semantic and affective level. If electronic text is inherently unstable at the level of the word then it will attract less serious attention from its readers. This will impact especially on the reception of writing that is designed to be "literary." Bolter welcomes this as the playfulness of electronic writing: "Electronic literature will remain a game, just as all computer programming is a game." He contrasts it with the "solemnity" of printed literature (130). But just as the fixedness of print does not transfer into our mode of reading it, neither need our response to electronic print be doomed to mere play. However, the characteristics of hypertext that I have considered so far seem likely to ensure that it will elicit less attention and less commitment from its readers. Perhaps a further important reason why this should be so lies in the disembodied nature of electronic reading.
* * *
Literary reading outside the classroom seems to depend in part on memory, feelings, or desire, such as the pull of narrative; it may call upon feelings not previously acknowledged or recognized by the reader, and this is likely to be at times a defamiliarizing, perspective-shifting experience. (I include in this the relations, explicit or not, that are elicited in the reader's mind to other literary texts, art works, etc.) This, at least, is the picture that is beginning to emerge from a number of empirical studies of actual readers, including our own (e.g., Miall, 1989; Miall and Kuiken, 1994). It is possible that electronic literary texts could be designed to evoke this engaged and transforming response. The way the electronic medium is understood by its theorists, however, seems to preclude this, as I will suggest.
The presence of the body during ordinary reading makes available types of response that are less available or absent from the electronic medium. We do not have to resort to nostalgia about the weight, the odour, or the binding of the printed book in order to see why. Our studies show that literary reading is predominantly an affective process, whereas the medium of hypertext tends to place emphasis on the discursive. Feelings in literary response draw upon bodily configurations of meaning, on personal resources of imagery and memory that implicate the reader's self concept to some degree. In contrast, if we endow the computer with bodily properties, we do so figuratively. As Pamela Gilbert (1997) remarks of our relationship to the internet, "We "finger" one another, "surf" the Net, "go to" a remote site, "get" files from one place, and "put" them somewhere else. The metaphor of the Net as space masks the disassociation of Netters from their bodies, masks the fact that the bodies are elsewhere, real, material -- invested with a responsible subjectivity."
As this suggests, for the internet and other computer domains we deploy physical or spatial metaphors to domesticate and familiarize. Literary texts in contrast are more likely to invoke the familiar in order to unsettle it. In this respect, it is our existing repertoire of concepts and schemata that prove inadequate (Miall, 1989). Literary reading involves a response beyond the cognitive or information processing systems that can be modeled by computer programs. As Van Peer (1989) showed, to the extent that the attributes of literature are confined to what can be represented by a computer, our understanding of literary reception will be shallow and trivialized. Literary reading requires a degree of absorption over time if new understanding is to evolve. Such absorption is not, as Bolter would have it, the "passive reading" which is "the goal of the naive reader or one who reads for entertainment" (Bolter 155). The interactive nature of literary reading requires readers to permeate the text with their own images, memories, and desires; but the text in turn refashions these and situates them within a new perspective. This interactivity is hard to see, as Marie-Laure Ryan (1994) suggests: it "has been obscured by the reader's proficiency in performing the necessary world-building operations." The postmodern or hypertext fiction, however, by drawing attention to its own fictionality, ironizes the constructive process, repeatedly decentering the reader and blocking participation in the fictional world. The text turns out to be worth less than our wishes to invest in it had suggested.
The "metafictional gesture," to borrow Ryan's term, intellectualizes the reading process and returns the reader to an awareness of the artifices of fictional construal. During literary reading, in contrast, reading is engaged by pointers in the text that signal a fuller meaning, inviting the reader to go beyond the expected or familiar framework of understanding. Such pointers may be provided by an unexpected turn in the narrative, by a discourse evaluation (Hunt and Vipond 1986), or by a moment of stylistic foregrounding. Since it is the schemata of understanding that are deficient (or overlearned, as Rand Spiro (1982) put it), it is the feelings evoked in the reader at such moments that provide the vehicle for resituating the unexpected.
Feeling locates the reader within an experiential world, with all its bodily and sensory concomitants (the shape of reading may often be registered within physical tensions, muscular dispositions, and impact the autonomic system, although we tend to remain unaware of this dimension of reading). A study of readers' responses to a Virginia Woolf short story that I carried out (Miall 1989) showed that when readers found their first assumptions about the story deficient, their feelings in response to passages of foregrounding provided a focal point for initiating an alternative conception of the story's meaning. Such feelings have an anticipatory quality: over the space of a short story (we have no empirical evidence for longer reading processes), a feeling evoked early in the story provides a framework for shaping the reader's understanding of the story as a whole.
This interactive process of reading is systematically disrupted within the hypertext medium. The linking of one text node to another tends to promote superordinate connections and elicit an analytical response more appropriate to expository prose than to literary response. The mechanical invocation of nodes through links will rarely correspond to the process of anticipation that a reader of a literary text experiences, since the need to choose from an array of multiple pathways at each step is unlikely to sustain the progressive unfolding of the reader's affective engagement with the text. This suggests, paradoxically, that the fixed form of the printed text may be more liberating for the reader than the constrained process of linking imposed by a hypertext, where the requirement to decide on which link to follow every few sentences seems likely to prevent the immersion characteristic of literary reading.
The failure of hypertext theorists to consider the reading process except superficially is apparent in the inconsistent treatment given to non-visual and other components of reading. Bolter, for example, attempts to reverse the priority of mind and body in favour of hypertext while preserving his poststructuralist view that text is all that the mind is. Thus, thinking of the role of the reader in the electronic fiction Afternoon by Michael Joyce (1987), he observes: "The computer gives the reader the opportunity to touch the text itself, an opportunity never available in print, where the text lies on a plane inaccessible to the reader. Readers of a printed book can write over or deface the text, but they cannot write in it" (144). Touch is, of course, mediated by the mouse cursor. While touch in this respect activates a link to another node, this form of touch is no more "real" than following a cross-reference in a printed text. It is minimal touch, indeed.
At the same time, Bolter attributes disembodiedness to "traditional" literature: "by ensuring that the reader cannot enter into the space that the text occupies, printing encouraged worshipful reading" (152). But Bolter's model of the reading mind is effectively a disembodied one. He discusses artificial intelligence as a process of "modeling the mind," but it becomes evident that it is a mind without physical connection to the world, and devoid of either senses or feeling: "Every computer program models the mind, as it reflects and reiterates the interplay of writer and writing surface" (175). Such a mind captures only an impoverished fragment of what the experience of literary reading contains.
* * *
If hypertext is the norm of textuality, as Bolter and Lanham propose, their discussions show that it is founded on an information processing model. Landow's account of reading Paradise Lost is typical. Read within a hypertext, he notes, "intratextual and intertextual connections between points of text (including images) become equivalent, thus bringing texts closer together and blurring the boundaries among them" (61). The reader of Milton he has in mind is the teacher or the student, whose fully linked electronic version of the poem provides immediate access to the Bible, Homer, Virgil, Dante, and Spenser (176). This reader, to follow the logic of Landow's description, will be able to click on a link every few lines to call up a different text; the text now on screen becomes equivalent to that of Milton and will (presumably) possess its own links to a multitude of further texts. While this may provide an appropriate environment for studying Milton's poem, it fails to capture the significance of reading it. The intertextual connections presupposed by the poem are caught up and transformed by Milton's master discourse. Until we see how Homer or Virgil are appropriated by Paradise Lost, we remain outside the poem as readers. Put another way, Homer is background as I read Paradise Lost, not equivalent, as the network model suggests.
Landow's interlinking model, moreover, can address only the most superficial relationships between texts. If Milton echoes Virgil in a line of Paradise Lost, the meaning of that connection cannot be modeled by clicking on a link to the relevant passage in the Æneid. Understanding the nature of Milton's reference requires an immersion in Roman culture and how Milton reads and appropriates it, for which a reading of the Æneid is only the beginning. A hypertext system is a valuable tool for alerting the student of Milton to the kind of work that needs to be done; but none of the more significant aspects of Milton's text, or any literary text, lies within the modeling capacity of hypertext.
Thus the network metaphor that Landow specifies, following postmodern theories of text, cannot account for the literary experience. It may suffice for mapping relationships between texts, or for explaining aspects of the reading experience post hoc, but it cannot underlie the reading experience itself. The hypertext which instantiates it is a device for explicating textual connections, not for experiencing them. Perhaps more importantly, hypertext cannot map the textual links that matter the most, where the text links to our existing schemata, feelings, and memories, including echoes of other texts that we have read. Hypertext makes explicit and concrete; in its insistently visual nature, it delimits and specifies. The rich array of resonances of an individual's reading of a literary text thus cannot ever be represented in hypertextual form. This might seem obvious, except that hypertext theorists such as Bolter or Moulthrop insist on representing hypertext as the ideal site of reading, as their accounts of textuality make clear.
While neither Landow nor Bolter engage with the experience of literary reading, their theoretical positions are derived from postmodern claims that embrace all texts. It is in specifying hypertext as the embodiment of such claims that the encounter with text is turned into information processing. Text is reduced to nodes or lexias, to morceaux, bits or morsels, in Derrida's term (Landow 9); these are then multiply interlinked. The logical end point, as Landow goes on to observe, is that of text as database, where a "search . . . permits the active reader to enter the author's text at any point and not at the point the author chose as the beginning" (Landow 75). While this atomistic view of text may be appropriate for sifting the data of literary (or scientific) scholarship, it cannot account for literary reading.
The hypertext network model, according to Landow, supports Derrida's notion of textual openness, "the irrelevance of distinctions between inside and outside a particular text" (Landow 8). This, he notes, is analogous to "the way that some chemicals destroy the cell membrane of an organism: destroying the cell membrane destroys the cell; it kills." And just as the individual section of text loses its distinctiveness, so is it "dispersed" into other texts (53). The biological analogy is instructive. In self-sustaining systems, the integrity of the different organs, from the cell upwards, is critical to survival. Landow's model of text suggests the metastatic stage of cancer, when individual cells from a given organ disperse and multiply elsewhere in the body. It suggests, in other words, a loss of functionality in the domain of text, and the end of reading. This view, however, is derived from a fallacy which is repeated insistently by several writers on hypertext.
Hypertext rhetoric is often derived from opposition to the book, seen as a closed, self-sufficient entity that necessarily falsifies the nature of text. The technology of the printed book, says Landow, "engenders certain notions of authorial property, authorial uniqueness, and a physically isolated text that hypertext makes untenable" (33). Thus in hypertext "the notion of an individual, discrete work becomes increasingly undermined and untenable within this form of information technology" (40). But to make his argument, Landow literalizes the book in a way that conflates function with form, or type with token. The instantiation of a text within a book does not determine how it is read. Reading is an interactive process: our studies show that at any given point during the reading of a literary text, a group of readers is likely to bring a varied range of assumptions, experiences, or feelings to the text. At the same time, certain textual devices appear to invite systematic connections and transformations within the reader's experience. These processes are consistent enough, at least across the short stories we have employed, for them to be statistically measurable.
The transformations that are typical of literary reading are robust, and their occurrence is clearly independent of the medium in which the text is presented. At the same time, they depend on the reader's acceptance of the text as a recognizable and discrete possible world, possessing its own integrity. The transformation processes we observe would almost certainly be aborted if the reader consistently disrupted the engagement with the current text by pursuing links to other texts. (The explicit study of a literary text is, of course, another matter.) What hypertext cannot do is replicate the pattern of meanings that an individual reader brings to bear on interpreting the text.
In another respect the network model of meaning offered by hypertext might be regarded as a formalizing of the reader's understanding. A part of the work of interpretation is, after all, to make connections across a text, from parallels between plot elements to extended metaphors, as well as beyond the text to other works of literature and beyond that again to the world of experience and history. The defamiliarizing process may impel us to situate a specific text element in relation to other elements that in linear terms are at a distance from it. Again, however, this process depends upon the personal resonances of the reader. To attempt to represent such connections explicitly in a network for all readers represents a premature formulation of the reading process; in effect, the network displaces the reader's own response. Only the more obvious allusions to other texts, or references to historical facts or concepts will be valid for all readers, but to model these in a hypertext is unlikely to include the most significant aspects of literary response for an individual reader.
Electronic media, as most critics agree, modify subjectivity and generate previously unrealized cultural formations. My argument against hypertext rhetoric is intended to make two primary points. First, that hypertext as a mode of reading does indeed change the nature of the reading process, and does so in ways that appear to be antipathetic to literary response. Second, that it is fallacious to claim that hypertext instantiates the "real" nature of reading by liberating it from the constraints of linearity. Cyberspace is a significant and powerful new medium, and one which our literary and academic culture must learn to accept and control. But resistance to the imperializing claims made on its behalf will be equally significant. The fate of reading is too important to be decided by hypertext theorists.
Works Cited
Barthes, Roland. S/Z, trans. Richard Miller. London: Jonathan Cape, 1975.
Birkerts, Sven. The Gutenberg Elegies: The Fate of Reading in an Electronic Age. New York: Fawcett Columbine, 1994.
Bolter, Jay David. Writing Space: The Computer, Hypertext, and the History of Writing. Hillsdale, NJ: Erlbaum, 1991.
Burke, Edmund. A Philosophical Enquiry into the Origins of our Ideas of the Sublime and the Beautiful. Ed. Adam Phillips. Oxford: Oxford UP, 1990.
Dillon, Andrew. "Myths, Misconceptions, and an Alternative Perspective on Information Usage and the Electronic Medium." Hypertext and Cognition. Ed. Jean-Francois Rouet, Jarmo J. Levonen, Andrew Dillon, and Rand J. Spiro. Mahwah, NJ: Lawrence Erlbaum, 1996. 25-42.
Douglas, J. Yellowlees. Gaps, Maps and Perception: What Hypertext Readers (Don't) Do. http://noel.pd.org/topos/perforations/perf3/douglas_p3.html (15 November, 1997).
Gilbert, Pamela. "On Space, Sex and Stalkers." Women and Performance 17: Sexuality and Cyberspace (15 November, 1997). http://www.echonyc.com/~women/Issue17/art-gilbert.html
Grusin, Richard. "What Is an Electronic Author? Theory and the Technological Fallacy." Virtual Realities and Their Discontents. Ed. Robert Markley. Baltimore: John Hopkins UP, 1996.
Hunt, Russell A., and Vipond, Douglas. "Evaluations in Literary Reading." Text 6 (1986): 53-71.
Joyce, Michael. afternoon, a story. Electronic text. Cambridge, MA: Eastgate Systems, 1987.
Landow, George P. "The Rhetoric of Hypermedia: Some Rules for Authors." Hypermedia and Literary Studies. Ed. George P. Landow. Cambridge, MA: MIT Press, 1991. 81-103.
Landow, George. Hypertext: The Convergence of Contemporary Critical Theory and Technology. Baltimore: John Hopkins UP, 1992.
Miall, David S. "Beyond the Schema Given: Affective Comprehension of Literary Narratives." Cognition and Emotion 3 (1989): 55-78.
Miall, David S. "Anticipation and Feeling in Literary Response: A Neuropsychological Perspective." Poetics 23 (1995): 275-298.
Miall, David S., and Kuiken, Don. "Beyond Text Theory: Understanding Literary Response." Discourse Processes 17 (1994): 337-352.
Miall, David S., and Kuiken, Don. "The Form of Reading: Empirical Studies of Literariness." Poetics (in press).
Moulthrop, Stuart. "Traveling in the Breakdown Lane: A Principle of Resistance for Hypertext." Mosaic 28 (1995): 55-77.
Moulthrop, Stuart. Getting over the Edge. http://raven.ubalt.edu/staff/moulthrop/essays/edge.html (15 November, 1997).
Rosenberg, Martin E. "Physics and Hypertext: Liberation and Complicity in Art and Pedagogy." Hyper / Text / Theory. Ed. George P. Landow Baltimore: John Hopkins UP, 1994. 268-98.
Rouet, Jean-François Rouet, and Levonen, Jarmo J. "Studying and Learning with Hypertext: Empirical Studies and Their Implications." Hypertext and Cognition. Ed. Jean-Francois Rouet, Jarmo J. Levonen, Andrew Dillon, and Rand J. Spiro. Mahwah, NJ: Lawrence Erlbaum, 1996. 9-23.
Ryan Marie-Laure. "Immersion vs. Interactivity: Virtual Reality and Literary Theory." Postmodern Culture 5:1 (1994). http://muse.jhu.edu/journals/postmodern_culture/v005/5.1ryan.html
Shumate Michael. Hyperizons: Theory and Technique of Hypertext Fiction (web bibliography). http://www.duke.edu/~mshumate/theory.html (15 November, 1997)
Spiro, Rand J. "Long-term Comprehension: Schema-based versus Experiential and Evaluative Understanding." Poetics 11 (1982): 77-86.
Stillinger, Jack. Coleridge and Textual Instability: The Multiple Versions of the Major Poems. New York: Oxford UP, 1994.
Tolva, John. "Ut Pictura Hyperpoesis: Spatial Form, Visuality, and the Digital Word." http://www.cs.unc.edu/~barman/HT96/P43/pictura.htm (15 November, 1997)
Tuman, Myron C. Word Perfect: Literacy in the Computer Age. Pittsburgh: University of Pittsburgh Press, 1992.
Van Peer, Willy. "Quantitative studies of literature. A critique and an outlook." Computers and the Humanities 23 (1989): 301-307.