COMPUTERS and COMPOSITION 9(2), April 1992, pages 95-129

Bolter, Jay David. (1991). Writing Space: The Computer, Hypertext, and the History of Writing. Hillsdale, NJ: Lawrence Erlbaum and Associates. (Hypertext version available from Eastgate Press, Cambrdige, MA.)

Bolter, Jay David, John B. Smith, and Michael Joyce. (1990). STORYSPACE. [Computer program.] Cambridge, MA: Eastgate.

Teaching "writing" . . . is like teaching a child how to draw only straight lines. It will probably enable them to draw straight lines, but it doesn't open up the more interesting questions of curves, or spheres, or the constraints of dimensionality itself. . . .
--John McDaid, "Hypermedia as an Ecological Context for Writing" (p. 217)

The computer can be viewed as either an intermediate or final medium. Most uses of computers in writing fall into the intermediate category, in which the computer is used as a step between mind and page, an environment in which writers can perfect their words before printing out a final draft (Oren, 1990; Slatin, 1990; Smith and Smith, 1990; Kiefer, 1991). The second category, the computer as final medium, departs from a paper-based goal--writers and readers interact with texts on-line, where they might take advantage of what Jay Bolter terms the new "space" of writing and reading, a space in which students might ask questions about the constraints of the dimensionality of print text (McDaid, 1991). In this second, less common category belong works such as the Macintosh hypertext program STORYSPACE, by Jay David Bolter, Michael Joyce, and John B. Smith and the perspective of books such as Writing Space, Bolter's overview of hypertext and literacy.

Common claims for hypertext include the possibility for promoting associative thinking (Beeman, Anderson, Bader, Larkin, McClard, McQuillan, and Shields, 1987), collaborative learning (McDaid, 1991; Slatin, 1988), synthesis in writing from sources (Neuwirth, Kaufer, Chimera, and Gillespie, 1987; Neuwirth and Kaufer, 1989); distributing traditional authority in texts and classrooms (Moulthrop and Kaplan, in press; Nelson, 1982; Marchionini, 1988); and facilitating deconstructive reading and writing (Moulthrop,1989; Johnson-Eilola, 1991b). Nonetheless, computers and composition research on hypertext is still at a stage that Anne Dipardo and Mike Dipardo (1990), in an essay discussing their own work in hypertext and composition, describe as consisting largely of "enthusiastic speculation (and not a little visionary zeal)" but "for the most part unexplored" (p. 7). Sorting out these claims is particularly difficult because each specific hypertext program attempts to fulfill a different conception of what hypertext should be. In this respect, hypertext is in much the same situation as computers and writing research in as a whole: One possible reason that research on word processing is often contradictory is that different programs, although they were designed for the same general activity, may enforce very different types of behaviors in writers (Hawisher, 1989, p. 57). Many current hypertext programs for personal computers, in addition, are directed primarily at helping writers of technical or functional documents rather than the essays or narratives more typical of composition classes. This is not to say, of course, that either technical writing or HYPERCARD scripting are dull, uncreative activities--rather, that technical writing is only one type of writing and that HYPERCARD scripting may be a clumsy way of encouraging writers to engage in a full range of complex and recursive activities. But STORYSPACE and Writing Space seem particularly well suited for composition use because they are directed at writing in a broader and more general sense. The theories developed in Writing Space help to make the foundations and implications of STORYSPACE more clear to writers and writing teachers. In this essay/review, I find myself continually moving away from an operational review of the merits/problems of either the book or the program on its own and toward a broader, more speculative essay on the implications of these works for the composition discipline. (Readers preferring a more straightforward synopsis of Writing Space should refer to Bolter's (1990) "Topographic Writing," which summarizes much of his book, or to Joseph Amato's (1992) more traditional book review and response in Computers and Composition.) What I am aiming for now is not so much an overview or response to Writing Space and STORYSPACE as an extrapolation of them into the context of composition.

The value of hypertext-related works such as these is not in what they are but in what they point toward, a way to address problematic questions about the nature of print text production and use: How does the static nature of print text constrain writing and reading? How do we represent or encourage the process of deconstruction or reader-response or social construction in terms of a fixed, static product? How or should we translate print-text writing into an electronic medium? Questions such as these characterize what Bolter (1990) calls "the late age of print."

STORYSPACE and Writing Space approach the angst that characterizes the late age of print in a more direct manner than is possible with traditional, static media such as typewritten pages. Writing Space and STORYSPACE provide, in a sense, a bridge between two ages, a sort of Rosetta Stone of theory for the move from print to computer media. In Writing Space, Bolter develops a rich history of text literacy that provides a foundation for theories and concerns about the move from print text to hypertext. Similarly, STORYSPACE exists on both sides of the boundary between page and screen. The program can function either as a outlining program in which writers can produce print-like texts (either for paper or on-line) or as a hypertext program (the program's main focus).

Unlike many other hypertext programs (HYPERCARD, GUIDE, TOOLBOOK, etc.), STORYSPACE explicitly departs from--or as Bolter puts it, "cheerfully violates" (p. 240)--many of the conventions of print. For example, the guiding metaphor for the hypertext programs HYPERCARD and TOOLBOOK is a stack of index cards, while GUIDE relies primarily on the extension of the cross-referenced print text or encyclopedia. Certainly, these programs can be--and have been--explored in more innovative ways, but such uses are relatively rare and are not explicitly encouraged as they are in STORYSPACE. The most immediately distinctive feature of STORYSPACE is that way in which the program represents texts as graphic, multiply-connected networks of text. Considering STORYSPACE in light of the theories and observations outlined in Writing Space reveals more subtle but important differences. According to Bolter, hypertext programs such as STORYSPACE offer not only a new visual space in which readers and writers can work, but also new conceptual and metaphorical spaces.

In order to understand how STORYSPACE and Writing Space contribute to reconceptions of writing and reading, it is important to understand the current status of computer-based composition in terms of its relationship to print-based composition. Writers in computer environments position themselves somewhere near the edges of the late age of print--often without intending or even realizing their position. The manner in which computers are normally used in composition pedagogy--that is, to produce print texts--is a simultaneous celebration of the freedom from print and acceptance of the limitations of print.

Even computer-based writing tools that are aimed at the production of a final, print text are valued for their nontextual qualities. Although little conclusive evidence bears out any large, consistent effects of computers on writers and text (Hawisher, 1989 [1]), text on the computer clearly differs from text on the page--on the computer, the text is fluid and easily changed (Sommers, 1985; Spitzer, 1985; Smith and Smith, 1990). The virtual screen text mimics our mental text; according to philosopher Mark Poster (1990), writers working with text on-screen see "a representation that is similar in its spatial fragility and temporal simultaneity to the contents of the mind or the spoken word" (p. 111); John B. Smith and Catherine Smith (1990) predict that "At some point, thinking and using the system will merge symbiotically" (p. 131). To some users, writing comes to seem less like crafting an artifact and more like thinking and participating in conversation (McDaid, 1991; Yankelovich, Meyrowitz, and van Dam, 1985).

But computer-based writing tools also evoke nostalgia for the print. As Tim Oren (1990) points out, computers are traditionally tools for the act of producing content for other media, "instruments for storing and manipulating data that ultimately will be printed out" (p. 467). In many cases, even "on-line" texts bear the marks of print: linear/hierarchical, paragraphed arguments split into a vertical string of pages in word processing programs, for example [2]. We are in what Elizabeth Klem and Charles Moran (1991) described as "an amphibious stage, operating as we do partly in print, partly on-screen" (p. 132), or, as Stuart Moulthrop (1991b) puts it, "still orbit[ing] the aging star of Gutenberg technology" ( p. 150). Many composition instructors still view the computer as a transcription tool, a conception that may prevent us from using the computer as effectively as we might (Kiefer, 1991, p. 121). In using the computer as a print-oriented tool, writing teachers may be extending their process-based pedagogies while missing the possibility of also creating a process-based product.

The seeming contradiction in the term "process-based product" is not surprising: The current model of writing as a complex set of socially situated, recursive, overlapping processes attempts to overcome the concept of writing as a static product (Phelps, 1988, p. 132). When composition moved from a product-based to a process-based pedagogy, the idea of the physical text as a product became more pronounced, not less. In highlighting the positive qualities of process, composition teachers enforce a sharp division, a separation of the product from the mental activity of constructing the text--both physical and psychic--in reading and writing. The printed, typed or handwritten text comes to take on, according to Louise Phelps (1988), "all the negative associations of teaching based on texts" (p. 135). Text becomes the antithesis of the act of writing a text, the only visible effect of the more important process of writing.

This division results in a double-bind conflict for teachers and theorists: A great deal of the last few decades of research in composition and literary theory (reader-response, deconstruction, social construction, intertextuality) represents an attempt to either hide, negate, or overcome the contradictions inherent in the process/product dichotomy (or, in the case of deconstruction, to playfully celebrate those contradictions). Because process situates itself as the antithesis of product--rather than a complement or correlate of it--theorists and teachers must deny the text even while they interpret, analyze, deconstruct, and evaluate it. The print text becomes something that must--but can never be--ignored. It is difficult to follow Donald Murray's (1984) advice to "work with language in action" (p. 90) when that language is frozen, obviously inactive. Written language is active, it seems, only when the writer's first attempt at production was inadequate and the text must be re-drafted. The whole process of drafting comes to represent a linear, temporal movement from inadequate to adequate (if not perfect) text. Revision, even with word-processing programs, is commonly cast as an activity in which expert writers spend time "cutting and polishing. . . . [I]mprovement comes only with hard effort applied to recalcitrant material" (Sudol, 1991, p. 921). Writers, especially students, most commonly re-draft texts only if they are significantly flawed (and if revising those flaws will improve the grade on the paper or in the course). Many writers do not consider the idea that re-drafting might allow them to work with multiple, equally valid texts; for such writers (who are, I would argue, the most common type of student writers), Draft (n) of an essay is necessarily superceded by Draft (n + 1). The process is lost in the product.

As writers and teachers, we can obviously value both product and process. But we must ask ourselves what purposes are served by maintaining a distinction between the two.

The computer was introduced into this difficult situation as an additional way to support process-based pedagogy--a tool that allowed teachers to deny the print-text product (even while the goal of most computer writing is still print). What makes the computer a valuable tool for writing, in a process view, is the way in which the fluid characteristics of computer text separate "writing/reading" from "print." Unfortunately, the move from typewritten page to the more malleable computer memory/display often serves only to make the dichotomy between process and product more pronounced than when the intermediate product was pen and paper rather than virtual text: The computer, Mark Spitzer (1985) has observed, "encourages writers to play until their texts are properly sculpted" (p. 211). That is, the virtual, fluid computer text is never delivered because, in most cases, the text will be frozen into print as a final step of the sculpting.

The disappearance of the fluid text in the final step of writing represents a fundamental difficulty with the process paradigm in composition theory and practice: The process shift has not gone far enough yet; or, to phrase the problem more accurately, composition theory still requires a corresponding product shift. The process shift in composition might have been due to not only a realization of the importance of the process of writing, but also the angst of the late age of print, an unspoken dissatisfaction with the media of print itself: Print physically denies the existence of a dynamic, associational text (Moulthrop, 1991b). In addition, a cognitive-based process pedagogy stresses a largely isolated writer, with the rich social influences secondary if considered at all (Cooper, 1986). The isolation of author and text from world inherent in a product-oriented pedagogy requires a more radical movement than, as the process-shift encouraged, paying more attention to the writer's processes. The problem was--and is--not only a lack of attention to the activities of the writer but also in the technological environment in which that activity takes place. Print is static, and physically resists all of the things that the process shift would seem to encourage it to do. The general ideas behind the process shift in composition are valuable, but a re-analysis of the technologies of writing is in order.

According to Catherine Smith (1991), hypertext "promises to become a redefining technology, a tool that reshapes not only practices but also the understanding of the thinking, reading, and writing activities it supports" (p. 224). Such redefinition casts new light on research in computers and composition, even in non-hypertextual areas. For example, researchers comparing paper-based and screen-based writing and reading often report that writers and readers working with paper were more capable of locating information and, when planning for writing, more apt to plan longer before beginning to write (Haas and Hayes, 1986; Haas,1989a; Haas, 1989b). Haas (1989a) found as well that when "planning" was divided between conceptual (whole essay) and sequential (word- and sentence-level) planning, writers on computers devoted more time to sequential planning and less to conceptual planning than paper-based writers. Klemm and Moran (1991) interpret these results as perhaps indicating that writers may need to learn new strategies for reading screen text (p. 137). However, we might also consider the fact that the word-processing programs such as the ones these studies utilized are based on a print foundation. Hypertext programs such as STORYSPACE offer a markedly different way of representing text--and, perhaps more importantly, a different way of writing and reading text. In this context, STORYSPACE and Writing Space offer a more fruitful way of dealing with the process/product dichotomy. From a compositionist or literary theorist's point of view, hypertext is not so much a "revolution" as one way to do things that we've been trying to do to print for some time now. If the computer is viewed as a medium separate from print--affording different types of reading and writing activities--rather than a tool for producing old media, the noun and verb forms of writing might begin to collapse into each other more than they have so far. [3]

Characterizing the computer as a final medium rather than an intermediate tool sheds new light on our unexamined assumptions and preconceptions about the medium of print. We might analyze some of the current "discomfort" with print text and the dichotomy between process and product as one characteristic of what Jay Bolter calls the "writing space" afforded by print. Bolter separates his discussion of the spatiality of various writing media into three, interrelated areas: visual space, conceptual space, and metaphorical space. For print text, the visual space is fixed, monologic, an object. The visual, physical space of the print text, in turn, engenders a particular view of the relationships between writer, text, and reader: The visual, physical artifact of the book becomes synonymous with the text itself. As Bolter puts it, the conceptual space of the book is one in which "writing is stable, monumental, and controlled exclusively by the author" (p. 11). The visual and conceptual spaces, in turn, eventually generate a metaphorical space; in the case of the print text, the text becomes a metaphor for the mind. The book seems a stand-in for the author, an extension or reproduction of mind. None of these ideas, of course, are unheard of: The conceptual and metaphorical space that the book tries to connote is at the heart of the phrase "the late age of print." What is interesting is the conceptual system in which Bolter casts the problem and the ways in which his formulation relates to STORYSPACE and other hypertext programs.

To an important degree, all computer-based text changes the characteristics of the writing space in important ways; the degree to which the computer is seen as encouraging process can be viewed as an effect of the undoing of some of the fixity of print text. But, Bolter observes, traditional word-processing programs and computer-aided instruction programs are frequently "attempts to transfer previous techniques of writing into an electronic idiom" (p. 6)--the computer as tool. As mentioned above, many hypertext programs attempt to transfer characteristics of print space to the computer space. The interconnected network of text that can be written with HYPERCARD, for example, is similar to that of a cross-referenced book: Certainly, it is easier to follow a cross-reference in HYPERCARD than in a book but the network of text is only abstractly implied, not shown. The textual practices that HYPERCARD encourages the program often no more than an electronic page turner, a book on steroids. HYPERCARD's support for "process" is only slightly better than that of the book--and worse, in some ways, than that of a simple word-processing program. One of HYPERCARD's most telling limitations is the way in which the program handles text. Editable text in HYPERCARD cannot be easily chosen as a hypertextual link; if text is linked, then later revised, the invisible link and the text become spatially separated on the screen. Writer/programmers can, of course, overcome this limitation (see, for example, DiPardo and DiPardo, 1990; McDaid, 1988), but the task is often complex and difficult to achieve. [4]

STORYSPACE, on the other hand, represents an attempt to explicitly move beyond some crucial limitations of the spatiality of print, limitations that composition, literature, and literacy theorists have struggled with for years. STORYSPACE, like many hypertext programs, explicitly supports multiple (sometimes contradictory) readings, intertextuality. Like computer writing in general, STORYSPACE makes the text a fluid, open process. Moreover, STORYSPACE situates these characteristics in a way that is simultaneously verbal and visual; a text in STORYSPACE can be shown as a graphical network of multiple levels--inside individual texts or intertextually.

When we write, we do not think in a letter by letter manner only, but also in terms of topics: sentences, paragraphs, sections and subsections of text, as well as the relationships between these items (Hayes and Flower, 1980; Kintsch and van Dijk, 1978; van Dijk and Kintsch, 1983). The visual, physical space of the typewriter supports writing primarily at the lowest level, the individual letter [5]. These other levels exist in most texts, but they are placed there one character at a time and manipulated at that level; writers and readers must place the linear text into a hierarchical/associational structure by decoding cues such as headings, tables of contents, paragraphing, and their own knowledge of the environment in which the text is composed and/or read.

The use of computers in composition instruction, at least in the "early" days, was often driven by the recognition that text on a computer is "fluid," that writers could easily move about, add, or delete large chunks of text with little effort in comparison to their work with the typewriter (Marcus, 1984; Spitzer, 1985; Sommers, 1985). In this way, the computer begins to change the spatial relationships involved in the writing process: The typewritten page does not facilitate modification of text in the same way that the computer does. The typewriter enforces a fine level of "granularity" in writing: each character must be typed or individually. "Moving" one section of text to another place in the document can only be done at that fine level: the complete section (and usually many other sections) must be retyped, letter by letter. The only way around this limitation is to abandon the technology and revert to scissors and paste.

One way in which writers and teachers attempt to impose a new granularity onto the text is through structural maps of the text (see Figures 1 and 2). The outline or tree, for example, represent a theoretical organization that attempts to guide the letter-by-letter composition of the text (van Dijk and Kintsch, 1983). Mentally, readers and writers strive to integrate the hierarchy of the outline with the line of the text. To the degree that writers are able to suggest to readers (or to themselves) that the line of the text is in fact a hierarchy or network, writers have succeeded in overcoming the linear, static characteristics of print.

Figure 1. Outline View in STORYSPACE (screen image).  
 

Figure 2. Tree/Hierarchy View in STORYSPACE (screen image).  

Whatever the technology, Bolter points out that writers always attempt to deal with text spatially: "With or without the computer, whenever we write, we write topically. To write is to do things with topics--to add, delete, and arrange them" (p. 16). The differences between technologies, however, encourage certain habits and discourage others. Nancy Kaplan (1991) offers the example of the blackboard as a pedagogical technology for writing: Although the blackboard is useful in teaching revision because writers can easily erase and insert new sections of text, the blackboard's physical characteristics make the sort of revision it encourages low-level: It is relatively easy to erase one or two words and replace them with one or two others, but revision on the scale of full sentences, paragraphs, and sections are difficult to illustrate (p. 27).

On the computer screen, the ease with which large blocks of text (sentences, paragraphs, or sections) can be moved about begins to change the visual, physical space of writing. In outlining programs, the outline and text integrated even more fully, as hierarchy becomes a functional, dynamic feature of the text (see Figure 3). The introduction of the dynamic and integrate outline changes, according to Bolter, the topography of the writing space.

Figure 3. Outline View with One Associated Text Window (screen image).  

Where the high-powered typewriters that we call word-processing programs are mainly tools for writing the flattened outline of linear text, outline processors allow, as Bolter says, "writing at a different grain. . . . With a pen or typewriter, writing meant literally to form letters on a page, figuratively to create verbal structures. In an electronic writing system, the figurative process becomes a literal act" (p. 19).

In STORYSPACE, then, the outline or tree/hierarchy becomes an operational, integrated component of the text. STORYSPACE writers and readers open and close text boxes by clicking on the icons or headings in the outline or tree (the heading "w/p = print" in Figure 1, for example, opens to show the text block in Figure 3). As with some other computer-based outlining programs, the headings (and attached text and/or subordinate headings) can be easily moved to other places in the hierarchy by cutting and pasting or by dragging them from one place to another.

Still, the outline--even when dynamic--imposes a preferred, hierarchical (not associational) order on the text. Multiple outlines may exist, each representing a differently organized hierarchical text, but individual hierarchies still limit the arrangement and connection between topics in the text. The outline is a topical structure developed in relationship to the needs of print: It is a simple matter to translate the topics of an outline or hierarchy into a linear text by expanding each subject heading (with the hierarchical tree, expanding top to bottom, left to right). The outline or tree view in STORYSPACE, as with other outlining programs, helps to reveal to writers and readers the operation of that structure in linear text; a structure that, John McDaid (1991) argues, "has become so inculated as to have vanished in a Barthesian mythological sleight of hand" (p. 217).

But much writing is associational, not so easily translated into a line. Consider the unseen, multiple, mental texts theorized by, for example, reader response theory. Bolter points to the pre-writing exercise of building networks of topics (see, for example, Flower, 1985, p. 59). Because most writing is still directed toward print media, the associative network must be broken into a hierarchy at some point and, in turn, to a line so that the text may be printed (Smith, Weiss, and Ferguson, 1988). Normally, there is no turning back--the line cannot be reconnected to the network in most writing spaces. Because the goal of the writing is print, associative connections can only be implicitly shown during the late stages of writing. Even when the computer is used for associative work, that use is usually relegated to "pre-writing" just as associational or networked writing is in paper-based composition pedagogy (Rodrigues and Rodrigues, 1984, p. 35). Although composition teachers would admit that the writing process is actually a set of recursive, overlapping processes--that association should not be limited only to the temporally first stage of writing--most computer-based writing programs do not support this type of recursive activity.

Although there are obvious differences between the personal, somewhat incoherent type of writing that occurs at one extreme and the carefully considered, polished prose of a "final" draft--what Flower calls reader-based and writer-based prose--even in hypertext, the distinction between "pre-writing" and "writing" seem far too pronounced, an effect of the tradition of linear, print text. As Bolter says,

Association is not really prior to writing, as the term "pre-writing" suggests. Association is always present in any text: one word echoes another; one sentence or paragraph recalls others earlier in the text and looks forward to still others. A writer cannot help but write associatively: even if he or she begins with and remains faithful to an outline, the result is always a network of verbal elements. The hierarchy (in the form of paragraphs, sections, and chapters) is an attempt to impose order on verbal ideas that are always prone to subvert that order. (p. 22)

Because print text technology cannot easily represent these associations, Bolter adds, they must be ignored or subdued. The associational, graphical method of writing comes to represent "inferior" writing (Turkle and Papert, 1990; Selfe, Selfe, and Johnson-Eilola, in press). The same is true of computer programs that attempt to emulate print. As Bolter says, the computer does not completely replace other forms of text, but coexists with and often imitates them (p. 41); the computer is often called a "virtual" machine because it can be programmed to emulate almost any other type of machine--from calculator to piano to printing press. Although the word-processing program is a step away from print, and the outliner is a further step, each still recalls print in more fundamental ways than hypertext.

Writers in print have developed methods for representing the associational text: Print conventions such as footnotes, citations, and indices represent an important concession to the associational nature of even linear text. An index, for example, contains an alphabetical list of topics and pages associated with those topics. But while an index might be viewed as a topical organization to the text, print technology necessarily and traditionally separates and subordinates non-hierarchical organizations--the index at the end of the book, the footnotes at the bottom of the page or end of the chapter. Consider how many texts--even those that desperately need one--show little or no evidence of effort towards offering a comprehensive index. In hypertext, however, the associational network can become a primary feature of the text (as in Figure 4).

Figure 4. STORYSPACE Network View (screen image).  

Hypertext as well--especially in terms of programs such as STORYSPACE--removes some of the distinctions between the primary, linear text and the secondary "texts" of the outline, index, footnote, or cited source. The hypertextual link, according to Terrence Harpold (1990), "functions as a much more radical sign of intertextuality than does, for example, the footnote marker in a paper text" (p. 172). And as George Landow and Paul Delany (1990) observe,

electronic linking destroys the binary opposition of text and note that feeds the status relations that inhabit the printed book. . . . [T]he assignment of text and annotation to different "statuspheres" . . . becomes very difficult, and such text hierarchies tend to quickly collapse. (p. 11)

STORYSPACE is unique in the way in which it embraces all of these structures--linear, hierarchical, tree, and associational or network--recursively, at any point in the process of writing or reading a text. Figures 1-4 are all taken from STORYSPACE screens; all are different views of the same document. Writers and readers move from one view to another by clicking on an icon in the window's scrollbar. More importantly, the network view is not necessarily prior to the outline view, as in common pre-writing exercises. And although association is the guiding principle of almost any hypertext program, popular programs such as HYPERCARD and GUIDE only suggest this structure through buttons contained within discrete segments of text rather than among segments of text; a writer could, of course, choose to construct a graphical representation of the network by manually drawing the network, but the network is not automatically constructed nor easily linked, as it is in STORYSPACE; the network, tree, and outline are all inherent features of the medium rather than added onto the medium by an individual writer or reader (as is necessary in most other writing and reading environments). In an important way, the "links" available in many hypertext programs are not really located in any visible space, they exist off-stage. In STORYSPACE, the link is a visible, explicitly and dynamically manipulable element of the text.

The physical, visible spaces of writing technologies are strongly connected to corresponding conceptual spaces. The product/process dichotomy discussed earlier, for example, is a conceptual feature of print that relates to the visual space of print: The physical text seems closed and monolithic. The fluid visual space of hypertext programs such as STORYSPACE, on the other hand, can encourage a conceptual space that is open and polylogic. Many of important concerns of recent composition theory and pedagogy involve what Bolter calls the conceptual space of print text. Although print can be made to show intertextuality, it does not encourage it, given that books are individual, bound, objects separated from all else. Books also connote a misplaced sense of truth, because they can be held, analyzed, quantified (Fish, 1980, p. 43). The book is also a singular monologue (Ong, 1982, pp. 132-135), what Roland Barthes (1977a) calls "the 'message' of the Author-God" (p. 146) that should be not so much read as consumed (1977b, p. 162).

This, of course, was not always the case; books are not the "natural" way of communicating. Bolter points out that "Writing with pen and paper is no more natural, no less technological than writing at a computer screen" (p. 37). The characteristics of any writing space are a complex mixture of technological possibility and human (social and individual) intention and use. The papyrus scroll, for example, was poor at conceptually connoting closure or even physical presence: the scroll was an aid to memory for verbal performance. As Bolter notes, Homeric poets were not normally even literate (p. 85). The codex, on the other hand, allowed full works (such as the New Testament) to comprise both a closed mental and closed physical work. Even with the codex and manuscript, the physical text is both a verbal and visual element because pictures were completely integrated into text. Print texts however, soon displaced the visual in favor of the verbal. Illustrations began to be produced and placed separately from the verbal text: Consider, for example, the modern convention of the phrase (see Figure 1). The implication in this stock wording is that the reader, at that point, should abandon the verbal text to consult the visual.

The conceptual space of the book also engenders the idea of thematic unity in a single text. Where in codices or manuscripts a single physical work often contained a variety of texts--often unrelated except by their being bound together--printed books became separate entities, self-contained objects. If multiple texts were held in a single volume, they were related by more than chance. The text becomes an isolated authority. In this evolution, print text comes to represent the apex of intellectual thought. Homeric poems, Bolter points out, "were unbounded; they were fragments of a network of stories that could be extended indefinitely" (p. 85). Modern print text, however, tends to be closed and linear. The abstract, linear, logical form of modern intellectual thought has become so conventionalized and accepted that today it is given privilege over the non-linear (Turkle & Papert, 1990; McDaid, 1991; Selfe, Selfe, and Johnson-Eilola, in press). Associational text is "pre-writing," something that exists apart from and prior to (and, implicitly, inferior to) the "real" writing.

To writing students, these bound "texts" are what professional, expert authors write. The textbooks that they read in class (composition or otherwise) are normally bound volumes that admit no dispute or even questioning. Students may choose not to value a printed book, but they have a much harder time arguing against the book. The book attempts to be a closed artifact. Process pedagogy attempts to undo this closure through such activities as peer-critiquing, student-teacher conferences, and collaborative work. But even when students or teacher respond to the somewhat more "open" texts written in class, the responses--both figuratively and literally--are often marginalized. Student-written texts are still isolated and fixed, even if only at the final stage, when the word-processed text is printed out and handed in--in other words, student texts are judged by the degree to which they can impersonate the style of writing in published books, the effortless, authoritative, completely unified voice of the book. To students, the "perfect" essay would be one to which only agreement--no questioning--was written in the margins, an object signifying the end of the writing process.

The idea of the author as the person in complete control of the text also becomes problematic. In his explicitly Derridean reading of Linda Flower's cognitive models of writing, Robert Brooke (1989) points out that Flower has "privileged 'control' over 'out of control', and in so doing, has obscured the problematic relationship between them. 'Out of control' is always a part--even a primary part--of writing processes in her work" (p. 410). In this way, the fixity of print text ends up promoting the view that expert writers are continually sure of their purpose and method, even in exploratory pre-writing. As well, this idea indicates that there is, if the writer was "in control" of their text production, a true meaning that can be gleaned by expert readers from the text (Spivak, 1976, p. xi).

The conceptual space of the book thus becomes increasingly difficult to reconcile with modern composition pedagogy. Social construction, to take but one example, explicitly attempts to overturn the idea of the text as an authoritative monologue developed by a single, completely isolated writer. Such attempts, however, are continually negated by the force of print against the intangible, socially constructed portions of the text.

Hypertext--along with not only modern composition and literary theory but also other non-print-based applications such as computer-based conferencing--begins to undo the book's conceptual sense of closure and authority: A hypertext is relatively open, both conceptually and physically. In many cases, its boundaries are continually shifting as new texts and new links are added. The guiding metaphor for hypertext is not the bound and complete text but the networked intertext. "An electronic book," Bolter says, "is a structure that reaches out to other structures, not only metaphorically, as does a printed book, but operationally" (p. 87). While Bolter argues that Ted Nelson's "docuverse"--a networked computer text incorporating all the world's writing into one great hypertext--is probably not feasible for both economical and political reasons, Nelson's vision remains an important (if unreachable) standard for hypertext development. The STORYSPACE document in Figure 4, for example, might more profitably contain not only my numerous quotes from other writers, but also the full text of the cited sources, so that both I and other readers could read the quote in its full context, or read in more depth than I have been able to provide here. Furthermore, Figure 4's hypertext might offer new readers (including the original author) the opportunity to add their own commentaries and associations to the text. The intertextuality inherent in such a structure would lend itself to collaborative writing between students and teachers or responses by students to other texts contained in the document (see, for example, Moulthrop and Kaplan's (in press) discussion of students using STORYSPACE in an "Introduction to Literature" class). In fact, Bolter calls the national and international computer networks ARPANET and BITNET--forums in which geographically displaced colleagues post and respond to messages from their own local "nodes"--"the first collaborative hypertext" (p. 29). Although many current examples of hypertext are individual efforts, the ease with which the conventional parenthetical citation can become a hypertext link to a completely different text promotes an intertextual conceptual space.

In addition, hypertexts are "open" in another sense that begins to blur the line between "reader" and "writer" (Moulthrop, 1989; Joyce, 1988; Slatin, 1991; Johnson-Eilola, in press). In a hypertext, the "structure" of the text develops according to a mixture of an individual reader's choices in "navigating" the network (rather than line) of text constructed by the original author. For example, Michael Joyce's (1990) hypertext Afternoon, a story (see Figure 5), a short story written in STORYSPACE, contains a network of approximately five hundred "nodes" or discrete chunks of text connected into a network by almost one thousand links. Because Afternoon was saved in a STORYSPACE's simple "reader" format, readers of Afternoon do not see the complex (and often overwhelming) tangled web of connections between the nodes, a small subset of which is shown in Figure 6 [6]. Despite the seeming simplicity of the opening screen in Figure 5, normal Afternoon readers are offered twenty different choices in moving from the first screen to a second screen in the story; most subsequent screens, in addition, offer multiple navigational choices to the reader. By making choices in navigating each screen of the story, individual readers in effect construct their own personal readings of Afternoon. In an important sense, different texts coexist in the overall network of the story, to be realized during each individual reading. As Bolter says in discussing Joyce's hypertext, thinking of Afternoon as a text means that we have to consider it as a story that contains multiple, contradictory endings (p. 124-125). In a hypertext where readers are also offered the opportunity to be writers (as is the case with STORYSPACE), the distinction between reading and writing begins to break down, process and product blurring into each other.

Figure 5. Opening Screen from Michael Joyce's Afternoon, a story © 1990 Eastgate Systems (screen image).  
 

Figure 6. STORYSPACE View (normally hidden) of Structure Around Screen Shown in Figure 5. Highlighted "Begin" Box at Upper Left Contains Screen Shown in Figure 5 (screen image).  

The "openness" of the text to the reader at both of these levels--each reader chooses paths and, in some hypertexts, adds paths or texts to the collaborative hypertext--holds profound implications for our conceptions of text, of what it means to write and read text in our society.

The open qualities of hypertext exhibit characteristics of both print and spoken language--not on a line between spoken and written, but at one corner of a triangle, a written form of Walter Ong's "second orality." As Bolter points out, where Platonic dialogue "invites the reader to participate in a conversation and then denies him or her full participation" (p. 111), hypertexts react to the reader's decisions in a way reminiscent of (although still clearly different from) the way in which an oral storyteller modified his or her tale to the audience's feedback. STORYSPACE, for example, not only allows writers to fashion multiple paths through a text, but to also gather and respond to other types of feedback: A writer can add to their text mechanisms which ask the reader to respond "Yes" or "No" to a question (as shown at the bottom of the opening screen of Afternoon on the left side of Figure 5) or even base the structure that readers see on actions that the reader has taken earlier in reading the text (e.g., if the reader has not read a certain space, direct them to that space; if they have already read it, direct them to a new section of the text). In this way, hypertext moves beyond artificial platonic dialogues to something closer to oral dialogue, a collaborative rather than individual effort for both the writer and the reader.

Because the pseudo-dialogue of a hypertext cannot be completely directed (or, perhaps, should not be) by the writer, it becomes impossible for the structure of the text to follow the linear/hierarchical ordering of traditional texts; the network becomes an operational structure in both reading and writing. So, while the structure of the book changes from line/hierarchy to network, the concept of the book changes as well. The book is no longer the final, exact word of the author but a space in which, according to John McDaid (1991), truth is no longer something to be discovered in a text, but something to be constructed in a polylogue with text(s), with the other writers contributing to the hypertext (pp. 214-215).

Because hypertext forces the reader to metaphorically "inhabit" the space of the text in navigating it, hypertext also begins to enact in a visible way theoretical approaches that work against "worshipful reading," theoretical approaches such as deconstruction or reader response theory. In fact, hypertext fulfills the role of deconstructive technology so well that Bolter claims that deconstructive theory must be surpassed for hypertext [7]; deconstruction attacks the qualities of print text that hypertext attempts to completely remove from the writing and reading technology:

Electronic writing takes us beyond the paradox of deconstruction, because it accepts as strengths the very qualities--the play of signs, intertextuality, the lack of closure--that deconstruction poses as the ultimate limitations of literature and language. (p. 166)

In this movement, Bolter begins to recover some of the usefulness of deconstruction for composition pedagogy. One of the difficulties of deconstruction for writing classes is that the process seems to contradict the presence of the printed text. But the deconstructed text cannot be handled, pointed to, or even finished in any sense. The physical text can be pointed to, but the deconstructed text does not traditionally exist anywhere except in the mind of the deconstructor (Belsey, 1980, p. 145). [8] What deconstruction violates is the conceptual space of the book, a conceptual space that is closely related to our traditional sense of order and truth. For many people--students and teachers alike--deconstruction is a personal affront to their own culture (Paine, 1989, pp. 565-566; Swearingen, 1990). Such readers and writers, understandably, find it difficult to reconcile the print text as physical artifact with the endless play of signification in deconstruction. In hypertext, however, the visible artifact of the text--no longer printed or authorial, but open and polylogical--comes to represent the deconstructing text. According to Stuart Moulthrop (1989), hypertext makes deconstruction common practice, not only for literary texts, but also for any text--student essays, instructional manuals, corporate memos.

Hypertext can help to extend much theory that composition teachers are already applying to print. Hypertext is, almost by definition, intertextual and collaborative. Reading and writing in hypertext are constructive, personal activities as well as socially constructed. The visual space of STORYSPACE highlights these activities, with its dynamic, malleable, graphical display of textual structure. Meaning is no longer an object that students are encouraged to find in a fixed volume; on the contrary, readers are clearly unable to find a single, fixed meaning because a hypertext always contains multiple texts.

In the final section of Writing Space, Bolter begins an ambitious exploration of the relationship between text and mind. Once writing is invented and internalized, Bolter says, it is difficult to think of the mind without thinking of the text. Thinking and memory become "writing in the mind" (p. 207). In current composition theory, the activity of writing is often conceived of as not being the act of transcribing thought into text, but a form of the act of thinking (McCrimmon, 1984; Gage, 1987). From this viewpoint, what is at issue in the use of hypertext as a writing environment is the influence of that particular writing space on thought/writing processes. Current composition theory, however, takes little account of the influence of different writing spaces or writing technologies on the processes of writing (Sullivan, 1991, p. 45). In equating writing with thought, composition theorists and instructors relate "intelligent" writing with "intelligence" itself. [9] By not considering the characteristics of the writing space, compositionists may find it difficult to evaluate computer-based texts, especially those texts that depart from print-based formats.

Bolter illustrates his point by characterizing artificial intelligence programs as texts--complex, interactive texts, but still texts. The "text" of an AI program in this view is composed of a verbal text and a reactive text; the program "responds" to the environment by following rules contained in the reactive text. An intelligent text becomes synonymous with an intelligence. In the end, Bolter points out, the original "texts" were written by human programmers. As Bolter says, AI programmers have hit upon some key issues of written discourse: Where does the author reside? Who is the author of the text created within a hypertext or AI program? Or might we wish to say that the text of AI is socially constructed? If so, is the AI an active participant?

By allowing a more complex level of deferral between author/programmer and reader, Bolter's exploration of AI returns to semiotic theory. The textual network of hypertext is itself a visible play of signs; "the computer as a text that can seem to read and write itself also provides its own semiosis" (p. 196). Bolter's return to deconstruction is not surprising. If there is a banner movement to the late age of print, it is deconstruction. Just as deconstructive reading tends to broaden from a reading strategy to the worldview of postmodernism, hypertext comes to represent not only a reading and writing strategy, but a metaphor for mind and society.

Bolter's continued use of deconstruction places the theory in symptomatic light: deconstruction is a symptom of the late age of print, a dissatisfaction with the ability of print text to do the things readers and writers wish to do with it. Instead of the interminable, mental play of signs consuming all, as deconstructions tend to, Bolter works to make deconstruction more "manageable," or at least less terrifying. The current interest in hypertext in composition and literature can be seen, in this light, simultaneously as a symptom for the fragmentation of modern society as well as a way to deal with that fragmentation in a positive, non-dominating manner.

The hypertext may be finite, but the interconnectedness of the text (explicit interconnections) means that there is no endpoint, no transcendental signified to the signifying chain. Hypertext is a medium in which resistance to participation with the text is impossible (Moulthrop and Kaplan, in press). Students cannot resist interconnection because their writings--even when they express disagreement with the original text--are necessarily a portion of the new text.

In this way, hypertext programs such as STORYSPACE move beyond one of the most noted difficulties of deconstruction, what Bolter terms deconstruction's obsession with "an endless transit from one passage to another" (p. 204). Because deconstruction is interminable, it never moves out into the real world. Even deconstructive or postmodernist readings of the world share similar characteristics. But the deconstruction of a hypertext takes place visibly, infinitely but in a finite area (the physical limits of the computer). The deconstructed text is in the world.

Our culture, as well, is seen by Bolter as "a vast writing space" undergoing the same types of changes as text (and for many of the same reasons): "[J]ust as our culture is moving from the printed book to the computer, it is also in the final stages of the transition from a hierarchical social order to what we might call a 'network culture'" (232). In a society such as our own, Cultural Literacy and Great Books movements seem misplaced because they attempt to place a singular hierarchy on our fragmented society(ies). The guiding idea of the Great Books program is that individual writers were able to isolate their own ideas into a timeless place that can be viewed and appreciated (although not touched). In composition theory, at least, such programs ignore intertextuality, culture and change, and the personal construction of text by new readers. The anti-hierarchical tendencies of current student-centered composition and literature pedagogies are, according to Bolter, a natural parallel to the idea that print is too limiting, too dominating. In an electronic, networked society, "a stable canon of works and authors is meaningless" (p. 237). Hypertext offers a sort of fragmented unity, an explicit intertextuality on the level of society as well as text:

The computer provides the only kind of unity now possible in our culture: unity at the operational level. Hypertextual publication can accommodate all the mutually incomprehensible languages that the intellectual world now speaks, and this unification of technique must serve as the consolation for the lost unity of purpose. (p. 235)

New Media/New & Old Problems

The early stages of the movement from one media to another are obviously unsettling times: Writing Space and STORYSPACE exist at the beginnings of such a time, when there are many pertinent issues that have not been answered and many more yet to be even addressed. Despite the enthusiasm of writers about hypertext (including myself), moving from linear text to hypertext is not a simple solution (or even a necessary step in a solution) to the complex problems of teaching and learning writing, not to mention the complexities of the social situations in which computers are used. Hypertext is a new lens to apply to our current situation, a lens that does not give us "true" vision but different vision. And those differences are not always ones for which we might wish.

At the concrete, day-to-day level of using STORYSPACE, writers will discover that although the network view of STORYSPACE makes the writing space richer, it also makes it more complex. The screens of Figures 4 and 5 contain less information than that of Figure 6, but more information also means a greater burden on the reader. Michael Joyce rightly saw that readers new to hypertext and STORYSPACE would probably not be able to cope with the complex network, both because of their unfamiliarity with hypertext in general and their unfamiliarity with the "power tools" that STORYSPACE can offer advanced writers and readers. As Bolter notes, "In electronic text both the reader's eye and the writing surface are in motion" (p. 71). Similarly, John Slatin points out that for hypertext writing in general, because a hypertext contains a collection of possible texts to be realized by individual readers, hypertext writers must pay more attention than linear writers to both small details and overall structure of the text (1990, p. 113). Readers and writers in STORYSPACE must pay attention to not only the verbal text enclosed in the separate places/boxes on the screen, but also to where the box occurs on the screen, how the boxes are linked to each other, and much more; this burden requires writers and readers to master not only verbal, but also visual rhetoric. This is not necessarily a drawback--on one hand, composition and literature theorists and teachers have begun to call for pedagogy that includes elements of visual rhetoric (Bernhardt, 1986; Sullivan, 1988; Sullivan, 1991; Kaplan and Moulthrop, 1990; Fortune, 1989; and Wahlstrom, 1989). On the other hand, Marcia Halio (1990a; 1990b) theorizes that the attention paid to visual elements of the text--fonts and graphics--may detract from verbal-skill learning. Although Halio's methodology and results are questionable (Kaplan and Moulthrop, 1990), the increased burden that hypertext may place on writers and readers is something that definitely must be considered in classroom practice. Writing instructors and students can, in STORYSPACE, choose to work in a completely outline/linear form if attention to the visual elements of STORYSPACE prove to be an overloading. In addition, theorists such as Michael Heim (1987) and Ronald A. Sudol (1991) express concern that the fluidity of computer-based text may encourage writers to be more concerned with accumulation than contemplation. The danger of concentrating on adding new material (text or links) to a hypertext without pausing to consider the material already in the hypertext may result in the construction of hypertexts that are so large and complexly related that neither writer nor reader can navigate the tangled text (Johnson-Eilola, 1991a).

A more troubling possiblity for a multiply connected text relates to the way in which "multiplicity" may seem to connote the necessity of "truth" or "fairness." Although it may be true that opening up the text to many viewpoints and multiple relationships can give readers a fuller sense of an issue, it is also true that multiplicity can be overwhelming instead of enlightening (Moulthrop, 1991b)--either accidentally or by the conscious intention of the original author.

The functions of STORYSPACE as well may pose difficulties for some writers, at the level of adding and manipulating nodes of text and links. Some new STORYSPACE users--even if they are already highly "computer literate"--may be uncomfortable with the demands of the program. Their confusion, often mingled with excitement, is reminiscent of users new to word processing when they discover the new ways in which they can write. John Smith and Catherine Smith (1990) call this brand of discomfort "the confusion of unlearning, of forming new conceptions, particularly about the nature of text" (p. 123). Our intuitions are built up by the traditions of print text, the medium that STORYSPACE purposely violates in important ways. The most obvious differences are visual ones--writing into a graphical, dynamic network instead of a static (or even dynamic) hierarchy or line.

Writing Space itself suffers from a similar problem--or benefit, depending on how one looks at it. Where most books rely heavily on learned genres to help readers build a hierarchical structure of the text, Writing Space is somewhat like a hypertext: associational, interconnected, fragmented. The sections of the text are organized somewhat hierarchically (as is traditional), but important structural cues seem to be missing. Transitions are often non-existent, sections sometimes seem to end prematurely. Often, material from one section recalls observations from a section fifty or more pages (and a major section) away; these implicit connections are often not clear on the first reading, but only the second or third. Some readers of Writing Space have expressed problems trying to unite the text into one integrated whole. Joseph Amato (1991), in his review of Writing Space, theorizes that the text, even in print form, might have benefitted from being cast in a more hypertextual format. In my own reading of Writing Space, I've found that the text is almost impossible to summarize in any coherent fashion beyond: "This is an important book about hypertext, literacy, and society. And a lot more." For Bolter is trying to move beyond print text, even though the book itself is printed. Given Bolter's discussion of the role of hierarchy as one way in which the book works to dominate the world, we should not expect the book to be easily summarized. In the "Preface," he admits that "the argument kept trying to cast itself intertextually, or 'hypertextually'" (p. ix). Writing Space is organized into the traditional hierarchy/outline of print text, but the individual sections of the text do not fit comfortably into the traditional form. Bolter himself comments that hypertexts "deconstruct themselves"; in a limited sense, even Bolter's print text engages in a similar practice. As with other print texts about hypertext (See Nelson, 1987 for an extreme example, or Smith, 1991 for a more conservative attempt), Writing Space is ordered into consecutive, linear sections, but readers are somewhat free to move around in the text as they might in a hypertext.

The difficulties with Writing Space are not pronounced: Writing Space is surely not as explicitly anti-textual (and mind-bending) as Derrida's The Post Card or Deleuze and Guattari's A Thousand Plateaus, deconstructive and postmodern texts that purposely attempt to defy the tradition of "book" in ways that make reading almost painful. It is not even clear that Bolter intended the book to follow this slightly fragmented, associational structure; possibly, what we see here is an effect of Bolter's experience with hypertext manifesting itself in a non-hypertextual medium. For this reason, I am especially hesitant to call these characteristics of Writing Space "flaws" or "mistakes" on Bolter's part. It would be more productive to view the state of Bolter's text as another level of commentary on the expectations and processes ingrained into a traditionally print-based culture. In fact, for an additional $8, readers can purchase an abbreviated (although expanded in some areas) version of Writing Space in the form of a stand-alone STORYSPACE file; users can read a hypertext version of the text even if they do not own STORYSPACE. The interconnections of the hypertext version add coherence and texture to the text, although, unfortunately, the document exists as a "page-mode" STORYSPACE document: The hypertext does not illustrate the connections between elements of the text graphically, but in a footnote-like manner. [10] The visible structure that is so valuable in STORYSPACE is not accessible to readers. This situation is probably a concession to the print-based models that most readers will use--casual readers may not want to struggle with mastering a basic hypertext literacy.

In addition, even if readers do own STORYSPACE, they cannot open up the hypertext version of Writing Space without changing file attributes with a Macintosh resource-editing program such as RESEDIT or DISKTOP. This is a fairly simple process provided that the user has achieved an intermediate level geekdom; most users do not aspire to these heights. Because of this, the hypertext version of Writing Space remains largely closed. To many readers, a closed text such as this might seem normal; after reading the print Writing Space, it seems limiting. By omission, the hypertext version illustrates some of the problems when writers and readers attempt to move from print to hypertext.

Writing Space and STORYSPACE are similar to the "incunabular" forms that Tim Oren (1990) discusses--texts on the boundary between new media and old. Writing Space is sometimes difficult to read because the text is influenced by a new media although still written in the old. STORYSPACE may be difficult for some users to learn for the converse reason: Users influenced by the conventions of print media must learn new ways of interacting with text. Furthermore, where readers and writers of print text can learn and rely on relatively stable conventions for suggesting structure in text, hypertext readers and writers, while they can more easily suggest some structure visually, will find that there are few if any conventions for structuring hypertext. Changing media is not an easy task; many of us are still struggling to master "the multi-layered grammars of the computer" (Selfe, 1989); hypertext literacy is yet another layer, often in conflict with more deeply buried grammars. The difficulties in reading Writing Space and working with STORYSPACE are well worth the effort. At the very least, they are a window on one possible direction the textual media might take.

Bolter, I think, is correct in his analysis of the fragmentation of modern society: hypertext may provide at least some way of dealing productively with this situation. However, modern society still resists that fragmentation through various forms of hierarchical domination. In a sense, the originality and futurity of Writing Space are somewhat utopian, what Gail Hawisher and Cynthia Selfe (1991) would characterize as a vision of the way things should be rather than the way they are. While Bolter might, for example, posit a society in which the computer provides the only kind of unity now possible, only those with computers can be a part of that society. Still marginalized are the traditionally displaced: poor, minorities, women, etc. As Joseph Amato (in press) writes, Bolter's vision might become "a theme park with no admission," giving one the sense of control and integration while only being an "illusion." This is, of course, the danger in attributing to technologies a cause and effect relationship. [11] Technology is socially and individually situated, and its effects are not easily predicted. In the quote in the previous section in which Bolter offers hypertext as a way of dealing with societal fragmentation, he explicitly narrows the arena to "the intellectual world." The danger exists here--perhaps in a more real sense than when the same criticism is leveled at "traditional" deconstruction--of totalizing the world in terms of the text. The only way to enter to gain value into a hypertextual society is to enter into the hypertext: If one is not part of the text, one is nonexistent. Carolyn Handa (1990), for example, has noted that contributing one's voice to the hypertextual discourse means necessarily giving up some of one's sense of identity (see also, Moulthrop and Kaplan, in press; Guyer and Petry, 1991; Moulthrop, 1991c) [12]. It is not clear if there is a way around this problem; in a postmodern society, the subject is always, at best, suspect.

This criticism, of course, can be leveled at any text that discusses the positive effects of computers without discussing the portions of society which do not have access to (or are not encouraged to use) this technology. (Both the forum and style of my own writing exist within a somewhat isolated and exclusive discourse community.) I raise the issue here, however, because Bolter shows so well the connection between the arcane practices of deconstruction and the realities of postmodern society. The late age of print is not just about changes in the surface of text, but also about changes in an ideology of text, mind, and society. From Socrates to Ong, changes in media have been considered more than merely simple, incremental movements. As Bolter's text develops, it becomes clear that hypertext not only changes the appearance of text, but also holds important implications for how we think of books and authors, how we write, respond to, and critique texts, as well as how we think of the mind and society itself.

The author would like to thank Cynthia Selfe, Mark Bernstein, Jay Bolter, and three anonymous reviewers for Computers and Composition for their patient readings and insightful comments on this essay. This essay was written while the author was being supported by a generous fellowship provided by the Humanities Department of Michigan Technological University and The Ford Motor Company.

Johndan Johnson-Eilola is a Ph.D. candidate at Michigan Technological University.

1. Hawisher identifies three main generalizations in her most recent review of computers and writing: improved attitude of students toward writing, longer texts, and fewer mechanical errors. Various other claims for the effect of computers are frequently made, but they continue to be inconclusive and conflicting.
   
   
2. Most on-line documents, of course, are hybrid cases, bearing the marks of both the computer screen and the page. My point here is that many of the the print-based characteristics of on-line text remain largely unconsidered.
   
   
3. The terminology here is drawn from Martin Nystrand's "On Teaching Writing as a Verb Rather than as a Noun: Research on Writing for High School English Teachers," and from similar observations about the object/activity distinction in the terms "writing" or "communication" in Calvin Schrag (p. 21) and Louise Phelps, especially the chapter "The Dance of Discourse."
   
   
4. This is not to say that using HYPERCARD or GUIDE or TOOLBOOK as an extension of print metaphors is wrong or misguided; only that these types of hypertext do not take advantage of the writing space of the computer in the same way that STORYSPACE attempts to. Conversely, as I will later discuss, STORYSPACE and similar hypertext programs introduce a new level of complexity that must be learned and continually managed by writers and readers.
   
   
5. For an interesting example of a print text that uses the medium's inherent physical characteristics to suggest a higher level of organization, see Marvin Minsky's The Society of Mind. Minsky's text is composed of page-sized sections, each adding its own small contribution to Minsky's attempt to explain how high-level intelligence might conceivably develop out of much smaller, more simple components.
   
   
6. As Mark Bernstein of Eastgate Systems (the publisher of both Afternoon and STORYSPACE) pointed out to me, even writers do not normally see this view, relying instead on other features in STORYSPACE to manage the complexity: tracking nodes and paths by name, date, author, keyword, or even color. More commonly, beginning and intermediate users of STORYSPACE will construct simpler networks such as that of Figure 4-hybrid hierarchy/network structures that alleviate some of the potential confusion.
   
   
7. See also Mark Poster's (1990) postmodernist discussion of computer-based writing in general (especially p. 99-128).
   
   
8. As Nancy Kaplan (1991), drawing upon Terry Eagleton, points out, at best one can write a secondary, physically separate text that metaphorically deconstructs the first-such as Roland Barthes' re-working of Balzac in S/Z. But even these metaphorically deconstructed texts, Kaplan notes, are "constantly threatened by the material presence of the text on its page" ( p. 17). Barthes' text becomes as frozen as Balzac's.
   
   
9. For "intelligence" here one might insert any evaluative measure-competence, ability, effort, contribution, etc.
   
   
10. Eastgate Press, which has assumed the marketing of the hypertextual version of Writing Space, has tentatively planned to release a revised version of the hypertext which may alleviate some of the difficulties I point out here.
    
    
11. See, for example, Myron Tuman's (1983) critique of Walter Ong's Orality and Litearcy, in which Tuman observes that literacy is socially as well as technologically situated.
    
    
12. Moulthrop (1991c) and Guyer and Petry (1991) are portions of a special issue of Writing on the Edge which includes a disk of hypertexts written in STORYSPACE and printed responses by a number of hypertext theorists.
    

Amato, J. (1991). Review of Writing Space. Computers and Composition, 9(1).

Barthes, R. (1977a). The death of the author. In S. Heath (Ed.), Image-music-text (p. 142-148). New York: Hill and Wang.

Barthes, R. (1977b). From work to text. In S. Heath (Ed.), Image-music-text (p. 155-164). New York: Hill and Wang.

Beeman, W. O., Anderson, K. T., Bader, G., Larkin, J., McClard, A. P., McQuillan, P., & Shields, M. (1987). Hypertext and pluralism: From lineal to non-lineal thinking. In Hypertext '87 papers, (p. 57-88). Proceedings of the Hypertext '87 Conference. New York: Association for Computing Machinery.

Belsey, C. (1980). Critical practice. London: Methuen.

Bernhardt, S. A. (1986). Seeing the text. College Composition and Communication, 37, 66-78.

Bolter, J. D. (1990). Topographic writing: Hypertext and the electronic writing space. In G. P. Landow & P. Delany (Eds.), Hypermedia and literary studies (p. 105-118). Cambridge, MA: Massachusetts Institute of Technology Press.

Brooke, R. (1989). Control in writing: Flower, Derrida, and images of the writer. College English, 51, 405-417.

Cooper, M. M. (1986). The ecology of writing. College English, 48, 364-375.

Dipardo, A., & Dipardo, M. (1990). Towards the metapersonal essay: Exploring the potential of hypertext in the composition class. Computers and Composition, 7(3), 7-22.

Fish, S. (1980). Is there a text in this class? The authority of interpretive communities. Cambridge, MA: Harvard University Press.

Flower, L. (1985). Problem-solving strategies for writing, (2nd ed.). San Diego, CA: Harcourt Brace Jovanovich.

Fortune, R. (1989). Visual and verbal thinking: Drawing and word-processing software in writing instruction. In G. E. Hawisher & C. L. Selfe (Eds.), Critical perspectives on computers and composition instruction (p. 145-161). New York: Teachers College Press.

Gage, J. T. (1987). The shape of reason. New York: Macmillan Publishing.

Guyer, C. & Perty, M. (1991). "Notes for Izme pass expose´." Writing on the Edge, 2(2), 82-89.

Haas, C. (1989a). How the writing medium shapes the writing process: Effects of word processing on planning. Research in the Teaching of English, 23(2), 181-207.

Haas, C. (1989b). "Seeing it on the screen isn't really seeing it": Computer writers' reading problems. In G. E. Hawisher & C. L. Selfe (Eds.), Critical perspectives on computers and composition instruction (p. 16-29). New York: Teachers College Press.

Haas, C., & Hayes, J. R. (1986). What did I just say? Reading problems in writing with a machine. Research in the Teaching of English, 20(1), 22-35.

Halio, M. P. (1990). Computers and controversy: Maiming re-viewed. Computers & Composition, 7(3), 107-112.

Halio, M. P. (1990, January). Student writing: Can the machine maim the message? Academic Computing, p. 16+.

Handa, C. (1990). Politics, ideology, and the strange, slow death of the isolated composer or why we need community in the writing classroom. In C. Handa (Ed.), Computers and community: Teaching composition in the twenty-first century (p. 160-184). Portsmouth, NH: Boynton/Cook.

Harpold, T. (1990). Threnody: Psychoanalytic digressions on the subject of hypertexts. In P. Delany & G. P. Landow (Eds.), Hypermedia and literary studies. Cambridge, MA: Massachusetts Institute of Technology Press.

Hawisher, G. E. (1989). Research and recommendations for computers and composition. In G. E. Hawisher & C. L. Selfe (Eds.), Critical perspectives on computers and composition instruction (p. 44-74). New York: Teachers College Press.

Hawisher, G. E., & Selfe, C. L. (1991). The rhetoric of technology and the electronic writing class. College English, 42, 55-65.

Hayes, J. R., & Flower, L. S. (1980). Identifying the organization of the writing process. In L. W. Gregg & E. R. Steinberg (Eds.), Cognitive processes in writing (p. 3-30). Hillsdale, NJ: Lawrence Erlbaum Associates.

Heim, M. (1987). Electric language: A philosophical study of word processing. New Haven, CT: Yale University Press.

Johnson-Eilola, J. (1991). Click here . . . no here . . . maybe here: Anarchy and hypertext. Paper presented at the annual Conference on College Composition and Communication, Boston, MA, 1989. (ERIC Document Reproduction Service No. ED 331 060.)

Johnson-Eilola, J. (1991). "Trying to see the garden": Interdisciplinary perspectives on hypertext use in composition instruction. Writing on the Edge, 2(2), 92-111.

Johnson-Eilola, J. (In press). An overview of reading and writing in hypertext: Vertigo and euphoria. In S. J. Hilligoss & C. L. Selfe (Eds.), Computers and literacy. New York: Modern Language Association.

Joyce, M. (1988, November). Siren shapes: Exploratory and constructive hypertexts. Academic Computing, 10+.

Joyce, M. (1990). Afternoon, a story. Macintosh computer program. Cambridge, MA: Eastgate Press.

Kaplan, N., & Moulthrop, S. (1990). Computers and controversy: Other ways of seeing. Computers & Composition, 7(3), 93-106.

Kaplan, N. (1991). Ideology, technology, and the future of writing instruction. In G. E. Hawisher & C. L. Selfe. (Eds.), Evolving perspectives on computers and composition studies: Questions for the 1990s (p. 11-42). Houghton, MI/Urbana, IL: Computers and Composition Press/National Council of Teachers of English.

Kiefer, K. (1991). Computers and teacher education in the 1990s and beyond. In G. E. Hawisher & C. L. Selfe (Eds.), Evolving perpsectives on computers and composition studies: Questions for the 1990s (p. 117-131). Houghton, MI/Urbana, IL: Computers and Composition Press/National Council of Teachers of English.

Kintsch, W., & van Dijk, T. A. (1978). Toward a model of text comprehension and production. Psychological Review, 85, 363-394.

Klem, E., & Moran, C. (1991). Computers and instructional strategies in the teaching of writing. In G. E. Hawisher & C. L. Selfe. (Eds.), Evolving perspectives on computers and composition studies: Questions for the 1990s (p. 132-149). Houghton, MI/Urbana, IL: Computers and Composition Press/National Council of Teachers of English.

Landow, G. P., & Delany, P. (1990). Hypertext, hypermedia, and literary studies: The state of the art. In G. P. Landow & P. Delany (Eds.), Hypermedia and literary studies (p. 3-50). Cambridge, MA: Massachusetts Institute of Technology Press.

Marcus, S. (1984). Real-time gadgets with feedback: Special effects in computer-based writing. In W. Wresch (Ed.), The computer in composition instruction (p. 120-130). Urbana, IL: National Council of Teachers of English.

Marchionini, G. (1988, November). Hypermedia and learning: Freedom and chaos. Educational Technology, p. 8-12.

McCrimmon, J. M. (1984). Writing as a way of knowing. In R. L. Graves (Ed.), Rhetoric and composition: A sourcebook for teachers and writers (p. 3-11). Upper Montclair, NJ: Boynton/Cook.

McDaid, J. (1988). Uncle Buddy's phantom funhouse. Unpublished Macintosh Computer HYPERCARD stack. New York.

McDaid, J. (1991). Hypermedia as an ecological context for writing. In G. E. Hawisher & C. L. Selfe. (Eds.), Evolving perspectives on computers and composition studies: Questions for the 1990s (p. 203-223). Houghton, MI/Urbana, IL: Computers and Composition Press/National Council of Teachers of English.

Minsky, Marvin. (1985). The society of mind. New York: Simon and Schuster.

Moulthrop, S. (1989). Hypertext and "the hyperreal." In Hypertext '87 papers, (p. 259-267). Proceedings of the Hypertext '87 Conference. New York: Association for Computing Machinery.

Moulthrop, S. (1991a). The politics of hypertext. In G. E. Hawisher & C. L. Selfe. (Eds.), Evolving perspectives on computers and composition studies: Questions for the 1990s (p. 253-271). New York: Teachers College Press.

Moulthrop, S. (1991b). Polymers, paranoia, and the rhetoric of hypertext. Writing on the Edge, 2(2), 150-159.

Moulthrop, S. (1991c). Writing on the hypertextual edge. Writing on the Edge, 2(2), 79-81.

Moulthrop, S., & Kaplan, N. (In press). They became what they beheld: The futility of resistance in the space of hypertext writing. In S. J. Hilligoss and. C. L. Selfe (Eds.), Literacy and computers. New York: Modern Language Association.

Murray, D. M. (1984). Teach writing as a process not product. In R. L. Graves (Eds.), Rhetoric and composition: A sourcebook for teachers and writers (p. 98-92). Upper Montclair, NJ: Boynton/Cook.

Nelson, T. H. (1982, March). A new home for the mind. Datamation, p. 169-180.

Nelson, T. H. (1987). Computer lib/dream machines (2nd ed.). Redmond, WA: Microsoft Press.

Neuwirth, C. M., & Kaufer, D. S. (1989). The role of external representations in the writing process: Implications for the design of hypertext-based writing tools. In Hypertext '87 papers, (p. 319-341). Proceedings of the Hypertext '87 Conference. New York: Association for Computing Machinery.

Neuwirth, C., Kaufer, D., Chimera, R., & Gillespie, T. (1987, November). The notes program: A hypertext application for writing from source texts. In Hypertext '87 papers, Proceedings of the Hypertext '87 Conference. New York: Association for Computing Machinery.

Nystrand, Joel. (1990). On teaching writing as a verb rather than as a noun: Research on writing for high school English teachers. In G. E. Hawisher & A. O. Soter (Eds.) On literacy and its teaching. Albany, NY: State University of New York Press.

Ong, W. J. (1982). Orality and literacy: The technologizing of the word. New York: Methuen.

Oren, T. (1990). Designing a new medium. In B. Laurel (Ed.), The art of human-computer interface design (p. 467-479). Reading, MA: Addison-Wesley.

Paine, C. (1989). Relativism, radical pedagogy, and the ideology of paralysis. College English, 51, 557-570.

Phelps, L. W. (1988). Composition as a human science: Contributions to the self-understanding of a discipline. New York: Oxford University Press.

Poster, M. (1990). The mode of information: Poststructuralism and social context. Chicago: University of Chicago Press.

Rodrigues, D., & Rodrigues, R. J. (1984). Computer-based creative problem solving. In W. Wresch (Ed.), The computer in composition instruction (p. 34-46). Urbana, IL: National Council of Teachers of English.

Schrag, C. O. (1986). Communicative praxis and the space of subjectivity. Bloomington, IN: Indiana University Press.

Selfe, C. L. (1989). Redefining literacy: The multilayered grammars of computers. In G. E. Hawisher & C. L. Selfe (Eds.), Critical perspectives on computers and composition instruction (p. 3-15). New York: Teachers College Press.

Selfe, C. L., Selfe, R., & Johnson-Eilola, J. (In press). Our pal Penelope: Weaving and un-weaving models of theory, practice, and research for designing and operating computer-supported writing facilities. In L. Myers (Eds.), Approaches to computer classrooms: Learning from practical experience. New York: State University of New York Press.

Slatin, J. (1988). Hypertext and the teaching of writing. In E. Barrett (Ed.), Text, context, and hypertext: Writing with and for the computer (p. 111-119). Cambridge, MA: Massachusetts Institute of Technology Press.

Slatin, J. M. (1990). Reading hypertext: Order and coherence in a new medium. College English, 52, 870-883.

Smith, C. F. (1991). Reconceiving hypertext. In G. E. Hawisher & C. L. Selfe (Eds.), Evolving perspectives on computers and composition studies: Questions for the 1990s (p. 224-252). Houghton, MI/Urbana, IL: Computers and Composition Press/National Council of Teachers of English.

Smith, J. B., & Smith, C. F. (1990). Writing, thinking, computing. Poetics, 19, 121-142.

Smith, J. B., Weiss, S. F., & Ferguson, G. J. (1988). A hypertext writing environment and its cognitive basis. (Report No. TR87-033 Chapel Hill, NC: University of North Carolina at Chapel Hill, Department of Computer Science.)

Sommers, E. A. (1985). Integrating composing and computing. In J. L. Collins & E. A. Sommers (Eds.), Writing on-line: Using computers in the teaching of writing (p. 3-10). Upper Montclair, NJ: Boynton/Cook.

Spitzer, M. (1985). Selecting word processing software. In J. L. Collins & E. A. Sommers (Eds.), Writing on-line: Using computers in the teaching of writing (p. 29-36). Upper Montclair, NJ: Boynton/Cook.

Spivak, G. C. (1976). Preface. In J. Derrida, Of Grammatology (G. C. Spivak, Trans.), (p. ix-lxxxvii). Baltimore, MD: Johns Hopkins University Press.

Sudol, Ronald A. (1991). The accumulative rhetoric of word processing. College English, 53, 920-932.

Sullivan, P. (1988). Desktop publishing: A powerful tool for advanced composition courses. College Composition and Communication, 39, 344-347.

Sullivan, P. (1991). Taking control of the page: Electronic writing and word publishing. In G. E. Hawisher & C. L. Selfe (Eds.), Evolving perspectives on computers and composition studies: Questions for the 1990s (p. 43-64). Houghton, MI/Urbana, IL: Computers and Composition Press/National Council of Teachers of English.

Swearingen, C. J. (1990). Bloomsday: Doomsday book for literacy? In A. A. Lunsford, H. Moglen, & J. Slevin (Eds.), The right to literacy (p. 215-224). New York: Modern Language Association.

Tuman, M. C. (1983). Words, tools, and technology. Review of Walter Ong, Orality and literacy: The technologizing of the word. College English, 45, 769-779.

Turkle, S., & Papert, S. (1990). Epistemological pluralism: Styles and voices within the computer culture. Signs: Journal of Women in Culture and Society, 16(1), 128-157.

van Dijk, T. A., & Kintsch, W. (1983). Strategies of discourse comprehension. Orlando, FL: Academic Press.

Wahlstrom, B. (1989). Desktop publishing: Perspectives, potentials, and politics. In G. E. Hawisher & C. L. Selfe (Eds.), Critical perspectives on computers and composition instruction (p. 162-186). New York: Teachers College Press.

Yankelovich, N., Meyrowitz, N., & van Dam, A. (October, 1985). Reading and writing the electronic book. IEEE Computer, p. 15-30.