Hypertext and reading cognition
Alec McHoul & Phil Roe
School of Humanities, Murdoch University, Murdoch, Western Australia 6150
'Hypertext' -- a term which is sometimes extended to include hypermedia in general -- refers to software capabilities which allow readers supposedly non-linear forms of access to information via personal computers and terminals. A typical hypertext document would open with a top-level menu or home page which might include conventional texts, audio recordings, still pictures and/or video samples: indeed information of any kind which can be stored digitally. On selecting highlighted or coloured words or phrases, or specially boxed graphic frames, a hypertext reader is led to a further screen containing more words and images which explain or expand the initially chosen item: and so on, potentially indefinitely. Each verbal or graphic point can be thought of as a node in a grid of nodes, such that the path traversed in any particular session of reading will be open to the interests discovered by the reader as she or he passes through the grid. Hypertext documents can be distributed on disk or CD, or else posted on mainframes and accessed through file-transfer-protocol routines or network softwares such as MOSAIC.
Some examples of hypertext/media usage are provided in an introductory document on the World Wide Web titled Guide to Cyberspace 6.1: What is hypertext and hypermedia?
"Here are some simple examples of hypermedia:
You are reading a text on the Hawaiian language. You select a Hawaiian phrase, then hear the phrase as spoken in the native tongue.
You are a law student studying the California Revised Statutes. By selecting a passage, you find precedents from a 1920 Supreme Court ruling stored at Cornell. Cross-referenced hyperlinks allow you to view any one of 520 related cases with audio annotations.
Looking at a company's floor plan, you are able to select an office by touching a room. The employee's name and picture appears with a list of their current projects.
You are a scientist doing work on the cooling of steel springs. By selecting text in a research paper, you are able to view a computer-generated movie of a cooling spring. By selecting a button you are able to receive a program which will perform thermodynamic calculations.
A student reading a digital version of an art magazine can select a work to print or display in full. Rotating movies of sculptures can be viewed. By interactively controlling the movie, the student can zoom in to see more detail." (Anon: np)
In principle, this form of information transfer should mean not only that hypertext information can flow freely to any reader whatsoever but also that, once accessed, any given document can be inspected according to the supposedly free choices of the reader. This double openness -- of access to texts and of addressing their contents -- has led some communications theorists to think of hypertext as revolutionary, as redistributing 'power' away from text producers and towards readers. In this paper we want to argue against such claims -- principally (and perhaps ironically) because they are based on a very narrow conception of reading practices. If reading itself were (and always had been) such a narrowly-conceived social and cognitive practice, there might be some substance to the celebratory and optimistic claims of these hypertext analysts. If not, their celebrations and their optimism may be premature.
It is pertinent in the first place to expose the theoretical substructure and the assumptions which underlie these claims for hypertext. Central to the claims of these communications theorists is their understanding of the hypertext object itself, their reading of certain critical theorists (especially Barthes and Derrida), and the assumption of a self-evident difference between hypertext and traditional print text. Claims of convergences between reader and writer, and between hypertext and contemporary critical theory, are based on a praxis of misreading. The critical issues are effaced by this misreading and by a use of (critical) language that a quotation from Derrida seems adequately to describe: it "betrays a loose vocabulary, the temptation of a cheap seduction, the passive yielding to fashion" (Derrida, 1976: 6).
Definitions of hypertext are continually elaborated against a particular and rigid notion of print text. The definitions accorded to the text are also presumed to be the determinants of reading practices. Delaney and Landow, for example, elaborate their definition of hypertext against a notion of the traditional text, which they define according to three attributes: "that the text was linear, bounded and fixed". Their definition of hypertext is then able to become: "the use of the computer to transcend the linear, bounded and fixed qualities of the traditional written text" (Delany and Landow, 1991: 3). Their extended explanation proceeds negatively, contrasting hypertext with the static form of the book: accordingly, hypertext can apparently be composed and read non-sequentially as a variable structure comprising blocks of text connected by electronic links.
Landow has frequent references to the fluidity and instability of hypertext as opposed to the fixity of print-based text. This is premised on hypertext's electronic status, the fact that it is potentially able to be amended and added to by the reader, and so forth (Landow, 1992). What he calls the convergence of reader and writer tends to efface a significant conflation that slips by without critical comment as to what this move constitutes and what is at stake in the conflation. The simple juxtaposition of the physical and fixed structure of the book (bound by its materiality) versus the electronic fluidity of hypertext does violence to the notion of textuality, collapsing distinct categories. Landow frequently relies on Barthes to elaborate his notion of text, but the distinction between work and text is not so simply elided. Barthes states that:
"The work can be held in the hand, the text is held in language, only exists in the movement of a discourse...; or again, the Text is experienced only in an activity of production." (Barthes, 1977: 157)
In the electronic paradigm, it is the notion of the work that makes no sense. In both electronic and print forms, the text remains in language, existing in the movement of discourse and experienced in the activity of production (reading/writing).
The claim by Landow and others of a convergence between contemporary critical theory and technology, specifically hypertext, is a misreading of both hypertext (as a critical object) and contemporary critical theory. The convergence of terms to which Landow points between these areas is simple appropriation -- the theoretical connections have not been established in any systematic way. It is bizarre and superficial to claim that an important theoretical and practical convergence has taken place simply because a number of terms ('link', 'web', 'network', 'interwoven') happen to be used in both hypertext discourse and in Derridean theory. Derrida's work on writing, for example, concerns writing in general -- a general condition of undecidability preceding all particular signs, texts and communications -- and so hypertext as a form of writing must be implicated just as much as other forms of writing. No 'special relationship' between Derridean conceptions of writing and hypertext has been established despite the claims. That could only happen if one -- wrongly -- thought of Derrida not as a philosopher interested in writing's general preconditions (which he is), but as a prophet of sematic anarchy and the reader's liberation movement (which he most certainly is not) (Nealon, 1992; Lucy, 1995). And Derrida notwithstanding, any claimed relationship between hypertext and 'reader-power' must be problematic, especially given the highly conventional and organised structuration of hypertext. The crucial issues of textuality and textual politics that are paramount to this discussion constantly slip away.
Let us begin with a caution. Hypertext enthusiasts exhibit a certain religious fervour linked with the political panacea of democratisation; they imagine a freely available node-web within which the liberation of the reader is the celebration of the mass(es). However, as we noted above, hypertext is very conventionally structured in terms of both access and address. Turning firstly, then, to questions of access: this is bound to be limited. Hypertext readers are a very select group simply by virtue of the equipment required to access hypertext documents. The minimal equipment needed is a reasonably powerful PC, connection software to a local mainframe and means of access to that mainframe: access in terms of both hardware -- such as a modem or ethernet connection -- and institutional rights which usually come with membership of, say, a university community. (And hence it is not surprising that the hypothetical users in the World Wide Web examples in the first section of this paper have such institutional affiliations.) On top of this, potential readers will need to be skilled in file-transfer routines and in hypertext manipulations themselves. This presupposes at least some minimal form of training, institutional or otherwise. Then we have to consider what types of texts can be delivered in hypertext and who controls this. Hypertext authoring programs such as Authorware Professional, Macromedia Director and Toolbook do not come cheaply. They require even more powerful machines than those required merely to read. They require institutional sanctions which allow writers to 'post' their texts on mainframes. Or else they require industrial links to CD-ROM manufacturers for distribution. This effectively limits hypertext genres either to pro-institutional texts (so that many of the first forms available via WWW were in effect advertisements for universities and museums) or to texts which might be perceived as having a market (games, encyclopaedias, movie guides, and so on). In this sense, hypertext technologies appear, in terms of their propensities for free composition and distribution, much more limited than conventional book technologies. In hyper-space, there is no equivalent of the spirit duplicator.
Now moving away from the question of sheer access and towards questions of address: a central claim among pro-hypertext enthusiasts is that hypertext is 'readerly', as opposed to 'writerly', with this distinction based very loosely on that of Roland Barthes. Hence Landow writes:
"From the vantage point of the current changes in information technology, Barthes's distinction between readerly and writerly texts appears to be essentially a distinction between text based on print technology and electronic hypertext, for hypertext fulfils [to quote Barthes (1974: 4)] 'the goal of literary work (or literature as work) [which] is to make the reader no longer a consumer, but a producer of the text. Our literature is characterised by the pitiless divorce which the literary institution maintains between the producer of the text and its user, between its owner and its customer, between its author and its reader. This reader is thereby plunged into a kind of idleness -- he is intransitive; he is, in short, serious: instead of functioning himself, instead of gaining access to the magic of the signifier, to the pleasure of writing, he is left with no more than the poor freedom to either accept or reject the text: reading is nothing more than a referendum. Opposite the writerly text, then, is its countervalue, its negative, reactive value: what can be read but not written: the readerly. We can call any readerly text a classic text'." (Landow, 1992: 5-6)
There is no doubt, in this reading of Barthes, a terrible category mistake. While Landow wants to make a complete separation between types of text such that the 'writerly' type is conflated with print (and hence closure) and the 'readerly' type with hypertext (and hence openness), Barthes himself is more equivocal. For Barthes, the writerly text denies the reader the pleasure of writing, to be sure. But this is precisely what forces the reader into a readerly position, into the space of "what can be read but not written". He consciously tropes on Nietzsche's idea of a slave ethics in introducing the readerly itself: it arises from a denial of entry into writing; it is a "negative, reactive value". It conforms to the writerly, gives itself over to it, plays its game. It too works with the rule of "what can be read but not written". And that is precisely why it is "classic". Where Landow finds an idealist space of liberation, Barthes only marks the side of the slave who is dependent on the master. In Landow's American liberalism, the oppressive simply has to be named and overcome by a word of negation -- 'readerly'. In Barthes, the apparent opposite always depends on what it opposes, plays its game, and finds ways of operating within the same rules. The readerly and the writerly are two prongs of a single forked instrument -- an instrument which may be writing in general and, if so, it will always contain possibilities of violence, one way or the other.
So there is, in this pro-hypertext position, an initial and foundational category mistake which reads a relational binary as an absolute binary, forgetting the dependence that binaries must always bring with them. Then, having made this move, the celebration can begin in earnest, even in Landow's dreams of his imaginary readers' reactions to his own writing:
"Although you cannot change my text, you can write a response and then link it to my document. You thus have read the readerly text in two ways not possible with a book: You have chosen your reading path -- and since you, like all readers, will choose individualised paths, the hypertext version of this book might take a very different form in your reading, perhaps suggesting the values of alternative routes." (Landow, 1992: 7)
Now, in a sense, the silly joke is out in the open: the master-writer's conditions are always the conditions that allow the slave-reader to be free: "you cannot change", "you have chosen", "you, like all readers, will choose" ... usw. This is precisely what Barthes meant by the 'readerly', and the meta-joke is that Landow could not read him slavishly enough. Overcoming the problems of a 'given' text, whether a book or an electronic node-web, is not a simple problem of negation or of imagined negation. And yet the supposed revolution of hypertext and its terrible 'readerliness' are premised, precisely, on such a simple overcoming: the readerly seems non-violent, but "you ... will choose".
The reason for this convenient reading of Barthes becomes clear when we see how this move is then able to link to one of the long-held objectives of hypertext practice. This objective has been articulated from early on in hypertext's history. Yankelovich's influential and often-cited paper "Reading and Writing the Electronic Book" (first published in 1985) makes this clear:
"Ideally, authors and readers should have the same set of integrated tools that allow them to browse through other material during the document preparation process and to add annotations and original links as they progress through an information web. In effect, the boundary between author and reader should largely disappear." (Yankelovich, 1991: 64)
McKnight et al. (1989), in "The Authoring of Hypertext Documents" note that most writings on hypertext have focussed on reading, on what is presented to the reader, and generally on reader-based research strategies. Authoring becomes something which is always oriented to reading (a very narrow and specific notion of reading), so that many hypertext systems in actual use blur the distinction between author and reader, particularly in cases "where the 'reader' will add links to the document, customise and annotate it, thus making the distinction between the author and reader less clear" (1989: 140).
Yet McKnight et al. also attempt to re-establish the place of the author, and do so by pitting the hypertext author against the author of the conventional book. The crucial point they make here is against the grain of hypertext enthusiasm whose short history has always privileged the reader, and the readerly. They say:
"once we have it in our hands, the whole of a book is accessible to us as readers. However, sat in front of an electronic read-only hypertext document we are at the mercy of the author since we will only be able to activate the links which the author has provided." (McKnight et al., 1989: 140) 
This argument is at odds with the celebration of hypertext as constituting the vanguard of the readers' liberation movement; for it conceives of reading practices as essentially determined by the structure of the text, implying a traditional relationship between author and reader, mediated by intentionality. With these assumptions about reading, it becomes possible to construe the provision of links in a document as choices for a hypertext reader which don't otherwise exist. McKnight et al in fact conceive of the links in a document as a constraint on the reader in that such links specify a structured, organised and thus limited number of options.
So while our own critical position towards unduly celebrating hypertext receives some backing from McKnight et al, it's also true that we part company from them when they construe pre-hypertextual readings (indeed any readings) in terms of a very narrow communications model involving authors' intentions set in place specifically to impart limited information to readers who thereby become victims of the text. What this position misses -- along with the celebrationist position -- is that quite 'ordinary' (including pre-hypertextual and hypertextual) forms of reading cognition can be quite fluid, artful, nodal and so on: there is nothing special about this and this is why there is nothing special about hypertext. Along with the celebrationists, McKnight et al seem to think that what is called 'reading' can only be one thing: a single practice with a set of fixed and identifiable criteria. For us 'reading' has always taken a number of highly diverse forms, some of which just happen to be used in electronic formations.
Returning to the celebrationist position, then: from its obviously spurious claims about an apparently new 'readerliness' comes a further claim which shifts it into the broader field of communications history:
"The strangeness, the newness, and the difference of hypertext permits us, however transiently and however ineffectively, to de-center many of our culture's assumptions about reading, writing, authorship, and creativity." (Landow, 1992: 203)
The impetus is no doubt Ongian (see Ong, 1982). Ong, we may remember -- if we have long memories -- claimed that oral communication was the most authentic and human, that writing technologies all but destroyed that complete presence which the exchange of talk permitted and reflected, and that, eventually, a post-'writerly' (in Landow's sense) communications technology would come to restore us to our authenticity. Ong mentions the telephone as an example: a means of exchange which restores the voice, the natural memory, and the presence of one soul to another. But isn't Landow's reading of hypertext an ultimate version of that; a system which "offers the reader and writer the same environment" (Landow, 1992: 7)? And isn't that shared environment precisely one of pure presence? So if "our culture's assumptions about reading, writing, authorship, and creativity" have anything wrong with them, it's that they don't permit an equal exchange, "the same environment", the auditorium (which is the space of the voice). Hypertext is then supposed to redress this balance, to make all persons equal because they become equal participants in a form of writing which (we hear) totally maps on to conversational exchange. Hypertext is 'readerly' and liberating because it restores the truly human voice (marked by the instantaneous exchange of positions, the dialectic) via an electronic medium. 
Elsewhere, Landow (1991) asserts that since hypermedia changes both the way texts exist and the way we read them, then it requires a new rhetorics and stylistics. Beginning from what he calls the defining characteristic of hypermedia (blocks of text connected by electronic links which emphasise multiple connections), he notes that the potential of hypermedia cannot be realised simply by linking, and that there must also be a range of techniques suited to hypermedia -- stylistic devices and rhetorical conventions. What initially seems promising, however, is just as quickly returned to an informational economy serviced by these 'new' rhetorics and stylistics. The necessity for these techniques, he says, is that they "will enable the reader to process the information presented by this new technology" (Landow, 1991: 81). Rather than engaging with a reconceptualisation of reading, writing, texts and meanings, it remains a matter of the more efficient distribution and dissemination of quite traditional information.
John Slatin's (1991) discussion of hypertext also takes up questions of rhetoric. While also beginning from the assumption that hypertext is very different from traditional forms of text, he attributes this difference to a function of the technology that makes hypertext. The characteristics of this technology, he says, are "various, vast and minute simultaneously -- making hypertext a new medium for thought and expression, the first verbal medium ... to emerge from the computer revolution!". A new medium, he says, "involves both a new practice and a new rhetoric, a new body of theory" (Slatin, 1991: 153).
The first requirement he suggests for a rhetoric of hypertext is that it must take the computer actively into account as a medium for composition and thought, and not simply as a presentational device or as an extension of the typewriter. Although he too contrasts hypertext with traditional (print) text, he does not collapse all reading into a single model. Instead he focuses on the assumptions which each kind of text makes about what readers do and the ways in which assumptions about reading affect an author's understanding of composition. His project is concerned with finding ways of talking about documents that have multiple points of entry and exit, and multiple pathways between these points. What this approach begins to open on to is not only an exploration of the possibilities of the medium (through for example, questions of interactive reading, interactive writing, and co-authorship), but also the language of the medium ('ways of talking').
Slatin's central argument is that rhetoric is typically indifferent to the physical processes of textual production. He notes that the maturity and stability of print technologies have been invisible as technology while such transparency is not yet available in terms of computing technologies. Hypertext and hypermedia, in Slatin's argument are still, and likely to remain, immature and unstable as technologies, and so a rhetoric of hypertext cannot afford to disregard its technological substrate. For this reason, theory and practice in hypertext have, potentially at least, an interesting co-existence and mutual interdependence.
What becomes apparent in the way hypertext practice is organised (because of its orientation to this narrow kind of informational reading), despite the claims, is that it is still based on conventional structures of writing and linearity (albeit with a more clearly defined, and also more clearly limited, multilinearity). The metaphorics of hypertext (and hypertextualism) are illustrative here. Shneiderman and Kearsley (1989: 6), for example, have a section on "hierarchies" in which the predominant metaphors are the "tree" (roots, branches and leaves) and of the "parent-child" (defined as superordinate and subordinate concepts). The definitions and descriptions they provide for these terminologies function as instructions for reading which organise reading cognition in terms of a series of metaphors connected to several of what are now fairly conventional discourses. These metaphors -- browsing, indexing, searching, maps, filters, tours, navigation, etc. -- constitute a conventional conceptual reading apparatus. While the implied function of this apparatus can be read as a bridge or transition between 'old' and 'new' modes of reading practice (enabled by the rigid definition of print text and the reader's relation to it), it appears more as the overlaying of conventional reading practice on new technology. The technology may be new, but the approach to it and the relations to it are wholly conventional. Hypertext has already been colonised by conventional reading practices -- how could it not be since, in a sense, it is thoroughly conventional -- and the colonisers don't seem to have noticed.
What exists as 'theory' about hypertext at this time does not acknowledge the roots of hypertext practice and is seduced by the hype around the vastness of the information potential of the medium. This seduction seems to function around the spatial metaphorics of its reading practice and their relations with the discourses on hyperspace and cyberspace, generated through a confluence of science fiction (cyberpunk in particular), the economics of information, and the technologies of computer science. The narratives that mark out these spatial trajectories bear a remarkable resemblance to colonial narratives of discovery and exploration -- where 'virtual' space (following geographic and then 'outer' space) has become 'the final frontier'. These narratives, it must be remembered, have their roots in pre-existing models of writing, textuality and technological practices -- Neuromancer, after all, was written on a typewriter. Tracing the movements of the narratives of these discourses may tell us more about the structures of reading cognition at work in hypertext than simple reduction to notions of the 'readerly' and the 'efficient' processing of information.
"We're in an information economy. They teach you that at school. What they don't tell you is that it is impossible to move, to live, to operate at any level without leaving traces, bits, seemingly meaningless fragments of personal information. Fragments that can be retrieved, amplified." (Gibson, 1988: 30)
And, of course, fragments that can be read in terms of a number of different discourses. Let us not forget here the cautions raised in the notion of the 'electronic panopticon' -- not in the sense of the Orwellian 'Big Brother', but where, as Provenzo (1992: 187) cautions, "[u]sing Foucault's terminology, the literate individual increasingly becomes an 'object of information, never a subject in communication'". Nor should we forget links to military discourses -- not only in terms of technological development, but also for plans and futures. Shneiderman in fact proposes what he calls his 'Star Wars' plan for American education with his vision of the beneficent patriarch. We are also enthusiastic about computing technology in education but we wonder about this educational philosophy:
"I propose a bold national Strategic Education Initiative (SEI) ... patterned on the concept of the Strategic Defence Initiative (SDI) or the Strategic Computing Initiative (SCI).... Mine is also a Star Wars Plan but it is linked to the image of Luke Skywalker's wise and gentle teacher Obi-Wan Kenobi (played by Alec Guinness) rather than to the terrifying Darth Vader. Instead of 1,000 space-based battle stations, I propose at least 10,000,000 school-based edu-stations, enough to have one for every five students, plus appropriate teacher training and software." (Schneiderman, 1992: 14-15)
Returning to our earlier question: the pro-hypertext position claims its object to be revolutionary by virtue of the supposedly non-linear way in which reading cognition takes place in such electronic environments. Hypertext, then, as the ultimate "nonlinear organisation of information" (Schneiderman and Kearsley, 1989: 158), appears to signal an historic shift: the end of the book, the end of linear writing and reading. In our experience, there is no doubt that hypertext documents do have some unique aspects: they speed up the rate of information retrieval and they do allow certain kinds of access to proceed at a pace which would previously have been thought impossible, or to require massive and painstaking archival research.
To take an example from a pre-hypertext database first: using the CD-ROM version of the OED, as opposed to its print version, allows a reader to find, say, all the words that have come into English from Russian since 1855 -- more or less instantly. The same process could, in principle, be carried out on the print version, but this would necessitate a sequential inspection and selection of each entry in the 13 volumes. But, speed apart, nothing has effectively changed in terms of the process of reading cognition. It's merely that the very hard work of meticulous inspection has been taken over by a disc's scanning head linked to a software instruction. The scanning head proceeds in a precisely linear or syntagmatic fashion, allowing the reader access to a specified field of data which, once generated, appears to have a non-linear or paradigmatic character to it. Because of the speed of computer processing, it appears as if the paradigmatic interest of the reader simply 'leaps' into the foreground. But this neglects the machine-reading component which is, in fact, more fully linear and syntagmatic than any human processing capacity.
The same goes for hypertext documents. The reader's paradigmatic interest is displayed in the unique path which she or he takes through a potentially infinite number of such paths in an information web. But each path, as the computer links from node to node, is a purely linear movement. Then, once retrieved, the image, sound or screen-print may or may not be inspected linearly. However it is inspected, the means of its inspection, at this point, will be precisely as it would be under any quite ordinary conditions of reading. Outside the hypertext environment, print can be inspected either sequentially or, say, globally: such as when one looks at a page for its typographical characteristics. Outside the hypertext environment, still images are routinely inspected in non-linear fashion: in fact it's very hard to know what a linear reading of a photograph could be like -- except that we know that computer scanners can divide photographs into pixels and proceed to reproduce them in a left-to-right, top-to-bottom form. Again, it's the computer technology which is more linear than the human and quotidian method of inspection. Outside the hypertext environment: films and videos can be viewed in 'real' time, sequentially from frame 1 to frame n -- but simple VCR equipment also allows them to be looked at in freeze frame, in reverse, shot by shot, scene by scene and so on. Quite simply then, there is a very broad variety of processes, both inside and outside the hypertext environment, which can be called 'readings'. The celebration of the supposedly new 'readerly', 'exchange-based', and 'non-linear' forms of reading which hypertext permits may, then, be premature. Moreover, it may be based on (in order to be opposed to) a far too narrow conception of what 'ordinary' reading is. Let us turn to this problem.
Hypertextualism, in its opposition to the 'writerly', the 'monologic' and the 'linear' appears to think that, prior to the advent of hypertext, reading was a single process, something like the scanning of a printed book from the first to the last word, with information passing into cognition in a sequence dictated by an author, allowing no space of intervention (no 'turn at talk', as it were) to the reader. Having read, on this picture of things, the reader simply 'has in mind' precisely what an author 'put there' and in the order that the author 'put it there'. There are numerous objections to this picture. A fairly simple one is that empirical analyses of reading have shown that readers do not simply add information bits to information bits in linear sequence. Rather, using what Garfinkel calls "the documentary method of interpretation", a very practical and ordinary form of the hermeneutic circle, readers build up a "gestalt contexture", a pattern of overall meaning which can modify -- or else, be modified by -- subsequent text items (words, sentences, paragraphs, and so on) (Garfinkel 1967; McHoul, 1982).
So here we have an objection to the particular picture of 'ordinary' reading held by hypertextualism. A second, and more serious, objection is that we can find no grounds at all for thinking that reading is a singular process of any kind, no matter what that process might be imagined to consist of. Between sections 156 and 171 of the Philosophical Investigations, Wittgenstein (1968: para 167) rejects the idea that reading is a "particular process" -- especially the quite popular idea that this process is a purely mental one. He asks how it could be that "one particular process takes place" when we read. We might read a sentence in print and then read it in Morse code, to give his own (multimedia) example. In such a case, is the cognitive process the same? We expect that most of us will think not. But Wittgenstein is not dogmatic about this. He wants to know why we come to think of the process as a particular one, as singular. And the tentative answer he gives is that we are perhaps fooled by the uniformity involved in "the experience of reading a page of print". He continues:
"the mere look of a printed line is itself extremely characteristic -- it presents, that is, a quite special appearance, the letters all roughly the same size, akin in shape too, and always recurring; most of the words constantly repeated and enormously familiar to us, like well-known faces." (Wittgenstein, 1968: para 167)
But the uniformity of a page of print, and the repetition effect we get in scanning it -- for all that they point to a surface definiteness and specifiability -- do not mean that reading -- even in this highly 'linear' case of scanning a printed page itself -- is a particular process. Instead, a brief inspection throws up a whole range of differences and distinctions regarding what the concept of reading might cover. Staten (1986: 84ff) speculates that one candidate for the essence of reading might be to specify it as being the derivation of repetitions from an original. And this, again, is one of the directions in which computer metaphors of reading have tended to take us -- such that it is how computers work that becomes the model for 'ordinary' (non-electronic) readings and not vice versa. But then we also have to ask: what is to count as deriving? The problem simply shifts on to another terrain. Perhaps, Staten goes on, we should always refer to the 'systematic' derivation of, for example, sounds from marks. But we all know that it is possible to derive the wrong sounds. If someone does that: are they reading?
Again, we could say that the essence of reading was the presence of a certain kind of inner experience, rather than a derivation. But we may, and do, have this experience while we are asleep or affected by drugs. Are we to say that, then, we are reading?
Instead of looking for a definite and singular characteristic of reading, Wittgenstein suggests that we look upon reading as an "assemblage of characteristics". Moreover, according to Staten, these characteristics will:
"in each separate case of reading ... be variously reconstituted, and in these different reassemblings there will always be the infection of characteristics of what does not correspond to what we want to think of as really, essentially, reading.... It is as though these characteristics had dual membership in two mutually exclusive sets." (Staten, 1986: 85)
To summarise: firstly, we cannot prespecify the characteristics which go to make up reading. Secondly, if we could, we would always find them in new and varied combinations, in any actual case of reading regardless of whether the activity takes place inside or outside electronic environments. Thirdly, we will always find, in amongst them, characteristics which we should not want to associate with reading as such but which are crucial to that actual case. Reading is like soup or slime. We should not want to specify its essence according to any neat digital calculus: not that it has no soul as such -- rather it has a multiplicity of souls and "any one of them could at some stage take over and guide the sequence in its own direction" (Staten, 1986: 103). It is because of, not despite, their pleomorphism that we recognise cases of reading.
Reading, then, is a classic instance of what Wittgenstein calls a family resemblance phenomenon. It is not a single or particular cognitive process -- rather it is a family of such processes, and a family whose members do not depend on the particular macro-technologies (books, computers, teacups, night skies, and so on) which happen to deliver texts. Instead, after Wittgenstein, we could think of the manifold forms that reading can take as technologies in their own right -- many of which can be transferred between macro-technological sites. For example, the ways in which Landow and others describe the 'revolutionary' forms of reading involved in hypertext scanning appear to us to be extremely close to the ways in which readers use reference works such as encyclopaedias. Hardly anyone (except perhaps a proofreader) would read such texts from start to finish. Instead a particular set of interests will lead a reader to an index, then to the selection of an item in print, then (perhaps) to a graphic, or to a cross-referenced item, back to the index, to a different source text and so on. Each item can be thought of as a node, if need be; and (again, if need be) the encyclopaedia and the internal and external texts to which it leads can be thought of as a web of such nodes.
There is nothing new in this. It is a perfectly ordinary procedure and one which is but a minor member of the vast family of possible forms of reading cognition. The fact that it has currently cropped up in a particular electronic macro-technology is cause for neither celebration nor despair. Reading remains a complex family of activities, language games, or technologies. It always already had no single defining characteristic such that hypertext could be different from that characteristic. And it remains like this whether or not -- today -- we are referring to printed or electronic means of delivery (macro-technologies). Everyday life continues pretty much as it always has: perhaps a little faster, that's all.
1. See also Whalley (1993) who provides a similar argument regarding the 'nonlinearity' of hypertext, and against the notion of the strict linearity of conventional text.
2. A colleague fond of hypermedia exchanges objected to our idea that he was setting up a virtual conference: "How can it be virtual if I can see him and he can see me in real time, face to face?"
3. This section is based on a chapter called "Reading Practices" from McHoul (in press).
Anon. nd. Guide to Cyberspace 6.1: What is Hypertext and Hypermedia? At World Wide Web address: http://abbott.com.emr.ca/guide/guide.02.html
Barthes, Roland. 1974. S/Z. Trans. R. Miller. New York: Hill and Wang.
Barthes, Roland. 1977. From work to text. In: Image-music-text, 155-164. Trans. S. Heath. New York: Hill and Wang.
Delany, Paul and George P. Landow. 1991. Introduction. In: P. Delany and G.P. Landow, eds., Hypermedia and literary studies, 3-50. Cambridge: MIT Press.
Derrida, Jacques. 1976. Of grammatology. Trans. G.C. Spivak. Baltimore: The Johns Hopkins University Press.
Garfinkel, Harold. 1967. Studies in ethnomethodology. Englewood Cliffs: Prentice-Hall.
Gibson, William. 1988. Johnny mnemonic. In: Burning Chrome, 14-36. London: Grafton Books.
Landow, George P. 1991. The rhetoric of hypermedia: Some rules for authors. In: P. Delany & G.P. Landow, eds., Hypermedia and literary studies, 81-104. Cambridge: MIT Press.
Landow, George P. 1992. Hypertext: The convergence of contemporary critical theory and technology. Baltimore: The Johns Hopkins University Press.
Lucy, Niall. 1995. Debating Derrida. Melbourne: Melbourne University Press.
McHoul, Alec. 1982. Telling how texts talk: Essays on reading and ethnomethodology. London: Routlege & Kegan Paul.
McHoul, Alec. In press. Semiotic investigations: Towards an effective semiotics. Lincoln: University of Nebraska Press.
McKnight, Cliff, John Richardson and Andrew Dillon. 1989. The authoring of hypertext documents. In: R. McAleese, ed., Hypertext: Theory into practice, 138-147. Oxford: Intellect Books.
Nealon, Jeffery T. 1992. The discipline of deconstruction. PMLA 107: 1266-1279.
Ong, Walter. 1982. Orality and literacy: The technologizing of the word. London: Methuen.
Provenzo, E. 1992. The electronic panopticon: Censorship, control, and indoctrination in a post-typographic culture. In: M. Tuman ed., Literacy online: The promise (and peril) of reading and writing with computers, 167-187. Pittsburgh: University of Pittsburgh Press.
Schneiderman, Ben. 1992. Education by engagement and construction: A strategic education initiative for a multimedia renewal of American education. In: E. Barrett, ed., Sociomedia: Multimedia, hypermedia, and the social construction of knowledge, 13-26. Cambridge: MIT Press.
Shneiderman, Ben and Greg Kearsley. 1989. Hypertext hands-on: An introduction to a new way of organising and accessing information. Reading: Addison-Wesley.
Slatin, John. 1991. Reading hypertext: Order and coherence in a new medium. In: P. Delany and G.P. Landow, eds., Hypermedia and literary studies, 153-169. Cambridge: MIT Press.
Staten, Henry. 1986. Wittgenstein and Derrida. Lincoln: University of Nebraska Press.
Whalley, Peter. 1993. An alternative rhetoric for hypertext. In: C. McKnight, A. Dillon and J. Richardson, eds., Hypertext: A psychological perspective, 7-18. Chichester: Ellis Horwood Limited.
Wittgenstein, Ludwig. 1968. Philosophical investigations. Trans. G.E.M. Anscombe. Oxford: Blackwell.
Yankelovich, Nicole. 1991. Reading and writing the electronic book. In: P. Delany and G.P. Landow, eds., Hypermedia and literary studies, 53-80. Cambridge: MIT Press.