| Empowering the reader? |
| There is much rhetoric about hypertext which stems from literary theory1.
In particular this stresses the importance of the reader's input into a hypertext document as compared with a book. Hypertext readers must actively choose their own path through the hypertext maze. Moreover, in a full hypertext system they can add their own comments and annotations, and hence assume a status equal to that of the author. The distinction between author and reader breaks down. While this may be a meaningful and worthy aim in literary studies and in literary education, these advocates are misleading. This is mainly because they start from a position 1For example, Landow and Bolter. |
| 39 |
| where literature (however it is to be understood) is central rather than peripheral
1. In sum, the problem with literary theory is its concern with literature. This point is illustrated by reference to the following examples of typical and important types of text used in post Second World War Europe and America. It is in the context of these, rather than poetry and literary criticism, that hypertext should be discussed. Important types of texts, some illustrations: |
| Tax forms Advertisements and mailshots Telephone directories - discussion of Minitel may be relevant here. Technical manuals - indexes to aeroplane or car spare parts are ideally suited to hypertext. Cookery Books |
| Note that this is a non-exhaustive list. In these sort of documents it would simply
be inappropriate to encourage a reader to make amendments to a telephone directory or a parts catalogue, although it would be very useful if a reader could easily correct mistakes. But in these texts, and in the many others like them, author and reader have quite different interests. If a hypertext document consists of a web of interconnected passages through which readers may voyage as they please then this affects the methods by which readers can construct arguments. The sense in which an argument is constructed in a book corresponds to the physical completeness of a book referred to in Chapter Two. The progress of a reader, who proceeds from the first chapter to the last reflects the progression through an argument The physical layout of the chapters in a book clearly determines the direction of its argument. Without this fixed structure it becomes hard to develop an extended argument, because the sequence of steps set out by the writer can no longer be assumed to reach the reader's attention in a meaningful order. This fixed structure within the book is of assistance to the reader, who is, however, not obliged to follow the sequence. Within a progressive argument, complications, contradictions and peripheral issues may be considered and accommodated. I have just made a cross-reference back to a previous chapter and could have added another 1For example, Young (1990:119). |
| 40 |
| forward to a section (unwritten when I first drafted this). Such cross-references
are acknowledgements of the existence of complications to my argument. However, while I wish to indicate these problem areas, I do not wish to deal with them at this stage. I confirm that they are considered elsewhere in the text, and I continue. Would the same effect be achieved by inserting a hypertext link? Perhaps, but I would then have to choose where to end it. There may be a reader of this chapter who has not read Chapter Two. If I considered that they should be encouraged to browse through it, I would make a reference to "the previous chapter" without page references. A hypertext equivalent would make a link to the beginning of the chapter, or could place the onus on the reader by prompting them to choose their own word, perhaps from a list of 30 relevant key words, inviting the reader to browse through the contexts in which those words occur. But it is arguable whether this has the same effect as a throwaway comment in a conventional text. |
| Further sceptical comments on hypertext Problems of scale |
| Hypertext is well suited to small (e.g. Nielson Hypertext 87 report) or highly structured datasets. Examples are computer manuals, aeroplane parts catalogues, and many commercial applications. However, datasets are often large or very large; think of the archives of a national institution, a Government Ministry, or a large business. Even some academic materials are of sufficient size for hypertext to seem less convincing. The Hartlib Papers contain twenty million words, and the even smaller Naga video disc exhibits some of the problems to be discussed below. There are two major types of problem. One is recovery, the identification and location of those elements of information which the reader seeks, the problems of which are described by Blair and Maron (see page qqq above). Words are used inconsistently, and the same topic can be discussed using quite different terms. This implies that no simple keyword search will lead to all the relevant places and to none of the irrelevant ones. When trawling through a large database this is not a trivial problem. The other problem afflicting large hypertext systems is that of being |
| lost in hyperspace... |
| 41 |
| When the size of the dataset increases, the links become numerous and complex. Problems
then rise when a diagram of the whole is attempted1. The purpose of such a diagram is as a map to guide readers around the web of information. Ideally it may serve as an equivalent to a table of contents (and index) of a book. If it is not possible to draw a fully comprehensible map, then a simplification must be made. Simplification can be effected in either of two ways. Firstly, a schematic view can be presented, which does not show all the links and pieces of information. This jeopardizes the reader's ability to find important and relevant passages. Secondly, the sorts of links built in to the system can be simplified. The problem is: how to encourage browsing without obliging the reader to make multiple choices at every step. If there are fewer prompted links from one item, the reader may miss connected and relevant information. On the other hand, if many links are prompted, the reader may choose one which is unhelpful. This is likely to occur frequently, since many readers will not have the knowledge needed to choose a link without following a fruitless path; all they learn is that they do not want to be there. From that point, retracing one's footsteps may not be easy. From each node several paths may branch. It may be difficult for the reader to return to their own path. Hypertext can seem not so much amazing as a maze. The problem for the designer of a hypertext system is how to encourage browsing while allowing reader easily to return to their point of departure. It will by now be clear that the writer is sceptical of some of the more extreme claims that have been made for hypertext. This chapter concludes in hypertextual fashion with some longer quotes from other authors on this subject. Jay David Bolter is a hypertext enthusiast. Of the cognitive connection between literacy, orality and computers he says: |
| Hard and Soft Structures |
| Alphabetic literacy differs from literacy in syllabic writings or in Chinese word-writing. The skills needed to write and read a papyrus roll are different from those required to produce and read a written book, and now electronic writing demands yet another set of skills. (By Classical Chinese or Japanese standards, most of us in the West are barely literate, because our handwriting is poor and we have no interest in calligraphy.) In each case the needed skills go beyond the mechanical ones of holding the pen or turning the pages. The writer must learn how to structure and locate text in the visual space provided, just as the reader must learn how to make sense of texts in that space. Such |
| 1Compare Nielson and illustration from Conklin. qqq |
| 42 |
| verbal visual structures are conventions that change in time, as anyone knows who
has tried to find his or her way through a medieval or ancient manuscript. Ancient papyrus and medieval codices present the modern reader with different and alien writing spaces. When ancient, medieval, or even Renaissance texts are prepared for modern readers, it is not only the words that are translated: the text itself is translated into the space of the modern printed book. In translating dialogues by Plato, the editor marks each change of speaker, punctuates sentences, divides long replies into paragraphs, and gives the whole book a table of contents and perhaps an index. Greek readers in the 4th century BC would have had none of these visual cues as they scrolled their way through the papyrus. Plato, Chaucer, and even Shakespeare long after the invention of printing are not merely respelled for modern editions; they are restructured and therefore rewritten1. |
| In the context of a discussion of the provision of manuals for the users of computers
we find some scepticism being expressed about the benefits of hypertext systems. What holds for computer manuals will certainly also hold for several other kinds of text (see my list above). |
| What is it about hypertext that so captures our fancy? I believe that it is the way
in which link traversals model the imaginative process. Hypertext systems give readers (often total) control to move through a complex space of datapoints, exploring those items that tweak their interest, following branches that whim dictates. The question here, however, can be stated bluntly. Is this what readers of computer books really want - or more importantly - really need? Is this the best way to communicate specific, instructional ideas between the writer (whose job, after all, is to help the reader understand on an intellectual rather than emotional level) and the reader (who has turned to the text specifically for intellectual enlightenment rather than entertainment or pleasure)? I maintain that the unstructured roaming encouraged by conventional hypertext systems is antithetical to the reader of both. We have no need of hypertext systems to teach us this lesson. The evidence is abundant in reader reactions to traditional 'reference' documents. Such documents provide a compendium of information; the reader is usually left to his own devices to figure out how these myriad 'chunks' of information interact and pertain to the solution that he is trying to discover. Readers are adamant in their dislike of such forms of documentation, except in those instances when brief, specific memory refreshment is in order. (...) They demand tutorials, and task-oriented handbooks... anything to provide structure, to synthesize the data into meaningful answers to specific problems. Readers have little time and less interest in exploration, they want to be led. When we generalize from the small scale bewilderment that characterises such behaviour to the confusion threatened by unstructured access to the vastly expanded information space offered by the networks, databases, and electronic libraries which are becoming commonplace we cannot help but pity the reader who is told to 'enter and explore' to locate the answer to his problem. Now he has an even greater problem; much more than being 'lost in hyperspace,' he might actually be in danger of the paralysing vertigo that is said to afflict astronauts engaged in extravehicular activities when they confront the endless depth of space that stretch beyond comprehension in all directions. 'Enter and explore' is a condemnation, not a solution. The reader's need for guidance has much more to say about the role of writers. For more than mere compilers of discrete information 'chunks' that are to be assembled at the whim of the reader, writers provide a critical service: they synthesise the available information into forms designed to address the particular needs of particular readers, |
| 1Bolter (1991: 40/41). |
| 43 |
| and thus add important value to the process. They construct meaning out of madness. They take a resume and transform it into a biography1 |
| A little later, Jaynes concludes: |
| Unstructured roaming through the hypertext landscape may be appropriate for a multitude of situations. But when readers need specific answers to specific questions, a more limited freedom is in fact the more efficacious approach2. |
| Finally, let us turn to a recent survey of hypertext (Mcknight et al 1991) in which
empirical studies of people reading from computers and from paper are discussed. In the introduction they state: |
| It is perhaps a fitting irony that the best way to disseminate our views is via the
paper- based medium of a book. In 1982, Jonassen predicted that 'in ten years or so the book as we know it will be as obsolete as is moveable type today' (Jonassen, 1982 p. 379) - a sentiment which apparently echoed the feeling of the pioneer Ted Nelson twenty years earlier, although in a lecture delivered in 1989 Nelson said he had thought books would be a thing of the past within five years, that is, by 1967 (Nelson, 1989). From our point of view, books will be a feature of information dissemination for many years to come. In this book, therefore, the relevant question will not be 'when will hypertext replace books?' but rather 'when is it better to present the information via hypertext rather than via paper?' When we talk about hypertext, we are talking about people using information to perform a task and hence we will be concerned to elucidate the nature of the tasks which are best supported by the medium3. |
| Near the end of the book they discuss: |
| One particular claim, linked to the 'paperless office' notion, is that hypertext
'frees us from the paper chains of linear text'. In our view, throughout the recent literature of hypertext, paper has had a bad press! Printed texts are generally not linear, either in their semantic structure or in the way in which skilled readers use them. Hence, to claim that hypertext frees us from the confines of paper is to tilt at windmills. Electronic documentation, whether structured using hypertext techniques or not, will not replace paper in the short-to-medium term. If paper is replaced in the long term it will be as much for socio-economic reasons (depletion of resources necessary for its manufacture, falling price of alternatives) as for conceptual reasons. Hypertext is not a universal panacea, nor is it so rigidly defined as to allow only a single style of implementation. Hence, some implementations will support some tasks, but none will support all tasks. It is the nature of the interaction between user, task and tool that will determine performance. In the same way that different text types have evolved in the paper medium, we would expect different types to evolve in the electronic medium. The question is the extent to which hypertext has a role to play in the evolutionary process4. |
| In the following chapter we shall draw back back and take the biggest possible perspective
to look at the whole history of writing so completing the circular structure at the centre of of this book. 1Jaynes (1989: 159-160). 2Ibid. (1989: 161). 3Mcknight et al (1991: 2). 4Ibid. (140) |
| 44 |