McKNIGHT, Cliff: Hypertext in Context

Chapter 5 - Creating Hypertext

 
Chapter 5 Contents  > From Chaos to Order, From Order to Understanding? SITEMAP

Producing Conventional 'Linear' Text ]

Figure 13 ]

Exploration ]

Organisation ]

Writing  ]

Tools Used ]


    Producing Conventional 'Linear' Text

    Traditionally, scholarly writing has been concerned with the "transformation of inchoate ideas into coherent structures" (Smith, Weiss and Ferguson, 1987). Although there is no universally accepted model of the writing process, there is general agreement over the components of the main stages Figure 13 represents these stages in terms of the author's activities, the external form of the text and the underlying logical structure. We will briefly sketch out the author’s tasks and then see how hypertext may support them.

      [Top]



    Figure 13: The Stages in the Production of Conventional 'Linear' Text. [After Rada et al., 1989]

     

    Initially, there must be some spur to undertake the work, be it intellectual (an alternate explanation for a body of facts or a resolution to an apparent theoretical inconsistency) or simply financial (a better mousetrap!) At this stage the 'text' will comprise at the very least a vaguely thought out theme and some supporting ideas which may be realised as an outline for approval by a publisher. Likely areas for initial investigation, such as relevant journals or individual authors, may also be known. From this beginning, three successive phases in the production of a text can be identified:

  1. exploration
  2. organisation
  3. writing
      [Top]




    Exploration

    This stage is concerned with general knowledge acquisition or 'fact finding', and involves searching citation indices and abstract journals and all the other activities which are relevant to a literature search. The process may seem disorderly since multiple avenues may be followed in parallel, and relevance thresholds may be initially set quite low. In addition, semantic associations within the information may be missing or erroneous. In order to manage the rapidly expanding collection of references, notes, quotations and photocopies, many authors resort to systems based on the card index file which provide an external memory without imposing any particular semantic structure (linear or otherwise) on the data.

      [Top]




    Organisation

    There is a gradual shift in emphasis away from fact finding and towards orienting, structuring and then re-structuring the growing body of material. This will involve examining the data from different theoretical perspectives and making inferences and deductions. These processes indicate gaps in the knowledge base which prompt further information seeking in an iterative fashion. Hierarchical structures will begin to take shape as superordinate/subordinate relations are perceived (or imposed) and sequences of cause and effect are established.

      [Top]




    Writing

    Smith, Weiss and Ferguson (1987) see this final stage as the culmination of a transition from what is essentially an information network to a hierarchy of concepts and relations. This phase initially involves the selection and arrangement of the knowledge generated during the first two stages to meet the author's rhetorical, or discourse, goals. The author will often have to choose between a number of pieces of supporting evidence for the sake of clarifying the argument. This is followed by a cycle of drafting, editing and revising at a much finer level of detail.

      [Top]




    Tools Used

    Word processing has significantly assisted this process through the ease with which revisions can be made to a text at all stages in the life cycle, but has barely improved on the initial data collection and organisation. Many authors still prefer to use file cards because of the ease of searching and, more importantly, sorting the notes. These requirements can be easily and effectively catered for by using a network-based hypertext system. Individual notes and references can be created as nodes in the network and then repeatedly grouped and regrouped as often as required. Mini-hierarchies, groupings, classifications and so forth can be assembled and discarded without abandoning serious investments of time and effort.

    Trigg and Irish (1987) describe the experiences of 20 authors when using NoteCards, a sophisticated network-based hypertext system, to help in the preparation of a variety of texts. NoteCards was found to be particularly appropriate when the task was considered complex and likely to require major reorganisation rather than simply editing or revision. They identified an overhead concerning the effort required to input the information, which tends to reduce the attraction of using NoteCards for shorter or less complex texts.

    If NoteCards-like systems can be useful aids in the preparation of conventional hierarchical texts, it is less clear that they can help in the presentation of those texts to readers. Experimental evidence from the field of document design suggests that



    "readers understand and learn more easily from texts when the information is set out in well defined structures and when the text provides clear signals of shifts from one part to another"
    (Charney, 1987).



    Thus, guidelines for technical prose (e.g., Kieras, 1985; Hartley, 1978) recommend that texts should be organized hierarchically and that this structure should be supported by

  1. typographic layout
  2. general comments should precede specific information
  3. presuppositions and transitions between sections should be made explicit.

    Probably the strongest support for the importance of hierarchical structuring in the comprehension of expository or technical text comes from the experimental and theoretical work on reading undertaken in the fields of text processing and discourse analysis (see, Kintsch, 1974; Meyer, 1975; Kintsch and van Dijk, 1978). In summary, the widely accepted theory of text comprehension proposes that readers maintain two representations of the text:

  1. The first is a verbatim representation of successive text fragments which are held temporarily (i.e., in short term memory)
  2. The second is called the textbase and is a hierarchical semantic structure extracted from the contents of the initial, or surface, representation.

    Comprehension is assumed to be an essentially 'bottom-up' process in which successive fragments are formed into the basic building blocks of the textbase. These semantic units are called micropropositions and, in total, they comprise the text's microstructure. Through the action of various rules of combination and reduction, this microstructure is successively reduced in an iterative fashion to produce higher order macropropositions and the text's macrostructure. The higher order macropropositions define the 'gist' of the argument contained in the text and are typically reproduced in readers’ summaries. Following criticism of earlier versions of this theory van Dijk and Kintsch (1983) extended it to include a situational model. This refinement allowed for the continuous application of domain specific knowledge from long term memory for the construction of the macrostructure (i.e., the process is recognised as also containing 'top-down' components).


  [Top]

Introduction ]  [ Conversion of Text to Hypertext ] [ Creation of Original Hypertext ] [ Characteristics of Extended Prose Arguments ] [ Hypertext Network or 'Web of Facts' ] [ Fallacy of Simple Networks as 'Ideal' Representations of Knowledge ]  FROM CHAOS TO ORDER, FROM ORDER TO UNDERSTANDING  [ Conclusion ] [References ] [Glossary ]