Hypertext

Hypertext, while having been available on mainframe computers for some time, is a comparatively new comer to the micro computer and computer based instruction. Because of this, its definition is fairly fluid with different writers emphasising particular aspects of it in their definition. For example, Marchionini (1988, 8) describes Hypertext as being:

the electronic representation of text that takes advantage of the random access capabilities of computers to overcome the strictly sequential medium of print on paper. "Hypermedia" extends the nonlinear representation and access to graphics, sound, animation, and other forms of information transfer.

whereas Carr (1988, 7) states that:

Hypertext is a specific form of data retrieval - one that's significantly different from other data retrieval methods.

Due to their very nature, the media that are currently used for the delivery of instruction are designed to be used in a linear fashion. For example, while readers can read a book in reverse order, they will undoubtedly encounter problems as the writer will have designed the book to be read from the beginning to the end. Film and video are totally linear and even a presentation by a teacher is basically linear although it has more flexibility than the other media.

Instruction is presented in a linear fashion through necessity rather than for sound educational principles and it can be questioned whether this is the best method for effective learning. When individuals undertake informal learning, they will not normally approach it in a structured, linear fashion, rather they will explore the situation in a way that interests them and to the depth that they require. If this is the case for informal learning should it also be the method employed in formal learning?

Raker argues that currently available information systems (i.e. paper based ones) have taught us to move through the information in a sequential fashion, and these materials are, in fact, designed to be used in this way. She states, however, that "Careful examination of reading habits reveals that most readers do not begin at the beginning of a book and read straight through to the conclusion", rather the reader skips from section to section, consults a reference, picture or figure and might even go to another text. If this is the way that people learn from books then Raker argues that they need a more efficient and natural way of accessing information (1989, 18).

Hypertext allows the storage of textual material in a non-linear fashion. Links can be created in a Hypertext document which allow for branching. These links are normally established by placing 'buttons' on a particular part of a Hypertext document (the button may simply cover a word, or it might be placed over several paragraphs). When this button is activated, normally by a mouse click, some action takes place. This action could be to simply show a field which contains a definition of the word on which the button was placed, or it might link to another part of the document, or it could link to a completely different document. Using this concept 'three dimensional' documents can be created which can contain, or give access to very large bodies of knowledge. This has very exciting implications for education. Not only can Hypertext provide an efficient way of storing and giving access to the every increasing corpus of knowledge, but it also allows learners to explore the material in a way that suits them.

Raker points out that teachers are going to have to develop new strategies for information management as the amount of available information will increase dramatically in the near future. As she says "In the next few years more data will be generated and disseminated than is currently contained in all of world history." (1989, 125). She also argues that it is necessary to develop ways of linking this new material into the existing knowledge base, and, as a result of this, new information organisational and retrieval system are being developed with Hypertext and Hypermedia being examples of these new technologies (Raker 1989, 18).

Jonassen (1988, 14) suggests that when using a Hypertext system,not only can learners create their own individual path for navigating through the knowledge base, but they can also add to it by annotating "the text by creating notes, explanations, and analogies, and by adding bibliographies or glossaries". He also claims that "The goal of hypertext and hypermedia is to provide an electronic environment to facilitate knowledge exploration by the learner".

Jonassen (1988, 14) states that the learner can gain several advantages from using Hypertext. Not only is the learner provided with access to large bodies of information, but:

That access is often quite unconstrained. Access to the corpus of information can be from many different locations rather than the more sequential access structures provided by traditional text. Hypertext also provides the opportunity for guided tours by the author which sequence access in ways that presumably will make the information more meaningful.

While it is still not known how an individual actually learns, it seems logical to suppose that learning can be facilitated by presenting information to the learner in a suitable structure. Hypertext can allow this. As Jonassen (1988, 14) argues:

What is most distinct about hypertext is its ability to represent in its own structure and presentation of the structure of knowledge that it is attempting to convey. Because hypertext is a node-link system based upon semantic structures, it can map fairly directly the structure of knowledge it is presenting.

This is taken a step further by Bush who states that learners can use Hypertext to:

access and organise information in accordance with their cognitive needs, which results in the best learning effect. Thus, flexible information organization and management are considered essential (Bush 1986 quoted in Park 1991, 24)

Jonassen (1988, 14) makes the point that learners normally have to adapt to the structure of the instructional material imposed by the author whereas, when using Hypertext, they can adapt the structure of the material to suit their own cognitive needs:

In hypertext, readers are not constrained by the subject matter structure or by the author's organisation of the text. Since an individual's knowledge structure is unique, based upon his or her own set of experiences and abilities, the ways that individuals prefer to access, interact with, and interrelate information is also distinct. In order to accommodate the text to the learner, rather than the learner to the text or its implicit structure, the text structure should be under the control of the learner. In order to get the most from accessing information, learners should be encouraged to explore information and even alter it in ways that make more sense to the learner. Hypertext permits these activities.

While as Jonassen (1988, 14) claims:

Hypertext (hypermedia) is a rapidly expanding technology that offers phenomenal instructional potential. Its flexibility in structure and style make it perhaps the most effective technology system to date for individualising instruction.

However, this power does not come without a cost. The various problems encountered by both the authors and users of Hypertext basically come from the fact that it is such a new technology that there is very little guidance available for either the designers or users of such systems (Jonassen 1988, 14).

Jonassen (1988, 14) describes some of these problems as being:

  1. Navigation, especially in unstructured hypertexts.
  2. Where should a user enter a hypertext?
  3. How do users integrate what has been learned from the hypertext into their own knowledge structure? The less structured the hypertext is, the less likely that users will integrate this knowledge into their own knowledge structure.
  4. Cognitive overload (See Dede 1988). The greater the number of learning options available, the greater the cognitive demands placed on the user.

Navigation is a problem identified by many researchers. For example, Roselli (1991, 42) states that:

A major problem with hypertext is that, as the structure of the material becomes more complex, so does the chance of the users becoming lost and disoriented (lost in HyperSpace). Current systems use navigational aids such as browsers, overviews and search and find functions. However, these do not always prove to be effective.

Marchionini expands upon this problem by stating that learners who use Hypertext systems face four problems, "(1) disorientation, (2) distraction, (3) technological progress, and (4) human psychology and sociology." (1988, 9).

Disorientation, according to Marchionini, is especially prevalent amongst novice users of Hypertext systems. One reason for this is the extent of the organisational system necessitated by the very large corpus of information held in the system which add to the cognitive load of the user. Another cause is the facility with which users can jump from one part of the system to another; "users explore and then forget where they are or how they arrived". Yet another cause, according to Marchionini, is that users cannot judge the size of, and their location within the Hypertext. He compares this with reading a book; the reader can determine how many pages there are in it and what page they are currently on. This situation should be improved as learners gain more knowledge about using such systems and designers start to produce more appropriate interfaces. Until then, he concludes that:

For the present, instructors who are concerned with the system getting in the way of student learning must be prepared to intervene to help students and to assist designers in minimising such effects by documenting in detail all instances of disorientation (Marchionini 1988, 9).

The second problem identified by Marchionini, distraction, comes from the fact that learners are given such a high degree of control over the Hypertext environment. While self-directed learning is a "worthy general goal of education" it cannot be assumed that this can be attained by simply giving freedom to the learner. A high level of freedom can become confusing as it brings with it increased decision-making responsibilities and "Cognitive resources may be diverted from content and relationships as learners attend to navigational decision-making". He continues:

Distraction is compounded by the vast quantities of information, easily accessible (a mouse click), much of which may be only peripherally relevant. The rich learning environment can easily become an environment of "hyperchaos". There is danger that students can either miss relevant instructional points entirely, or worse yet, form wrong interpretations of the information. Placing an inquisitive, undisciplined young learner in a rich hypermedia environment may be similar to giving a remote control unit to a student in front of a television with 80 channels of action programming. ... The issue becomes whether students should learn by "browsing the world" or staying "on task" according to the wisdom of their teachers. ... Introducing hypermedia may affect fundamental learning patterns in schools and certainly will affect instructional decison-making on the part of teachers (Marchionini 1988, 10).

Marchionini (1988, 10) states that techniques for adding structure to Hypertext documents are evolving (for example, "guided tours, layers of paths or webs" and these should reduce the amount of decision-making that the learner has to undertake and thus reduce the level of distraction.

Roselli (1991, 42-43) identifies the same problem but refers to it as 'cognitive overload':

The wide range of choices within a hypertext can lead to cognitive overload and users have to continually assess whether or not it is worth continuing along the current path. This cognitive overload can give rise to conceptual disorientation when the users lose sight of the aims of using the hypertext while exploring this mass of information. This often occurs because users take an inappropriate path through the material.

However, while Marchionini sees removing some learner-control by structuring the Hypertext as a solution (or partial solution) to this problem, Roselli (1991, 42-43) does not agree with this and warns that:

While this [cognitive overload] is seen as a major problem with hypertext, the solution to it must not remove the main value of such a system, the users' ability to navigate freely through it

Roselli's (1991, 42-43) solution to this problem is to build a monitoring system into the Hypertext. This system would plot the learners' progress through the Hypertext and warn if it was considered that they were getting too far from the main point and suggest a suitable course of action (presumably somewhat like the 'advisement' strategy proposed by Tennyson et al. (cited in Relan 1991, 12)). Roselli states, however, that the learner should be able to turn off this system whenever it was desired to do so.

This whole problem again raises the issue of user control and Bowers and Tsai (1990, 22) suggest that as Hypertext presents such a different situation to other computer-based instructional media, the whole issue of learner control should be re-examined. They argue that:

While much of the research on learner control has encompassed the allowance of a somewhat restricted freedom of movement between screens or discrete pieces of information, hypertext can facilitate the creation of connections or links within a large body of diverse pictures, facts, or activities; in this way, the learner is actively involved in building the learning environment.

Carr (1988, 9) makes the following comment about learner control:

Hypertext, to a large extent, leaves control of the process in the hands of the user. Your first reaction to this may be that it's as much a drawback as an advantage. After all, one characteristic of teaching/training is that the student must be led. While an expert system may be too rigid and controlling to be an effective trainer, going to the opposite extreme doesn't appear to help a lot.

He makes the suggestion the designer can maintain control of the instructional environment by creating paths along which the learner can proceed, but the learner can decide which of these paths to take. This might be termed 'pseudo-user-control'.

The third problem that Marchionini (1988, 9) suggests that users of Hypertext systems encounter is the technology used. He states that the technology currently used in presenting Hypertexts gives rise to a number of ergonomic problems which affect learning. Marchionini includes such things as poor screen resolution, awkward input devices and poor workstation and furniture design as contributing to this. He states that educators must plan for the "total electronic environment" and not just for the purchase of hardware and software. (This is supported by a survey of students that I undertook in which the majority said that the Newling Computer Laboratory was not a good learning environment and cited such things as badly designed furniture, insufficient space, noise, distractions, and poor lighting to support this claim.) He also notes that when there are vast amounts of information to be handled in a hypertext system, time is of essence and high speed processors must be used (1988, 10).

The fourth problem identified by Marchionini is that of Human psychology and sociology. As education comes to rely more heavily on technology, there will be the normal responses to change and that "we can be assured of confusion, false starts, grandiose expectations, gross underestimations, and hopefully some serious improvements". He notes that there is already resistance to change and burn out by people who have attempted to introduce technology caused by the rapid rate of change in software, hardware and interface. "How learning is affected by these natural effects of change is uncertain, but hypermedia will certainly exacerbate the classic tension between empowering learners with greater freedom to direct their own learning and assuring society that all students learn a common body of skills, concepts and principles" (1988, 11).

He argues that:

One of the most exciting potentials of hypermedia is the quality of learner control it allows. A fundamental problem for teachers and designers is how to shape this potential into quality learner control experiences. We want our students to learn to explore information freely and easily, but with purpose and discipline. The privilege of freedom demands responsibility; the responsibilities of using hyperdocuments include knowing when and how to stay oriented and attentive to goals (1988, 11).

As was stated earlier, Jonassen (1988, 14) claims that it is not only users who encounter problems with Hypertext systems. As Hypertext is a new technology, there are few guidelines and little research to assist the authors of such systems in their work. Jonassen says that some of the problems that are encountered by hypertext authors include:

  1. When creating a hypertext where do you begin?
  2. What structure, if any, should be given to the hypertext? What options should be given to users?
  3. "How can the knowledge structure of the expert or the content structure of the subject matter be manifest and then mapped onto the hypertext?" (1988, 14).

Kearsley (1988, 21-22) presents some authoring guidelines for Hypertext documents which go a little way to answering Jonassens second question regarding structure. Kearsley suggests that producing a Hypertext data base introduces some additional considerations to those involved in normal good writing and that these include:

  • Chunking. Information should be presented in small "chunks" that deal with one idea, theme or topic.
  • Inter-relationships. Each document should contain links to other documents to produce a richer hypertext environment. However, such links should serve a clear purpose.
  • Consistency of document names. The author must keep track of documents as they are created if links are to be established with ease.
  • Master reference list. It is essential to keep such a list as authoring progresses to ensure accurate referencing.
  • Simplicity in Traversal. "Navigation should be simple, intuitive, and consistent throughout the system. Movement through the system should be effortless and not require thought".
  • Screen design. "Screens should be designed so that they can be grasped perceptually. Visual layout is very important in screen design".
  • Low cognitive load. "Minimise the burden on the user's short-term memory. Do not require the user to remember things from one screen to an other".
  • Early reviews. "Subject the data base to technical, legal, and management reviews as early as possible. As the data base becomes larger, changes become more difficult to make".
  • Maintain multiple perspectives. "When authoring, try to balance the technical requirements of the system with the user's perspective and the organisational use of the data base" (1988, 21-22).

In a Hypertext system that consists of more than one document, the first one that the user gains access to is called the root document. It is from this that the user can gain access to other documents in the system. This, therefore, is the most important document in the system as, if users cannot interact satisfactorily with it, they are unlikely to achieve success with the rest of the system. Kearsley (1988, 22), therefore, presents several possible strategies for use in this root document:

  1. Make the root document an overview that contains links to all major concepts in the data base [glossary approach].
  2. Adopt a hierarchical approach in which the links in the root document are major categories. [top-down approach]
  3. Organise the root article as a list or table of contents of the major concepts in the data base. [menu approach]
  4. Make the root document into a tour of the data base [tutorial approach] (Kearsley 1988, 22).

The specific approach to be taken depends on the nature and use of the data base.

Another factor which determines the effectiveness of a Hypertext system is the design of the screen display. Information must be displayed effectively if the user is to gain access to it efficiently. Kearsley states that there are a number of formatting and organisation issues that have to be taken into account when producing Hypertext documents:

Spacing of lines and the amount of information to be included on a single display are a major concern because of the poor readability of many computer screens (30 percent slower to read than printed text). A major principle of screen design is that displays should not be "cluttered" and contain too much information (1988, 22).

Hypertexts can be structured in several different ways. As Boone and Higgins (1991, 21-2) state:

The idea of hypertext as described by Nelson (1984, 1978, 1981) include different types of hypertext forms, from simple links between chunks of related text, to a more loosely structured text navigation system connected to a vast knowledge library including all pertinent information about a subject. Jonassen (1986) described three forms for implementing hypertext: (a) node-link, with chunks of text linked together providing direct access from any piece of text to another; (b) structured, in which the hypertext form serves as a meta-database, controlling access to each of several databases of related information; and (c) hierarchical, similar to structured hypertext, but with content arranged with general concepts broken down into more detailed concepts.

What ever structure is adopted for a Hypertext, Kearsley states that "Knowledge must be structured in a way that supports the mental models that readers may create when they use the hypertext system" (1988, 23).

As was discussed in the introduction to this paper, much of the work that was done in the area of CBI in the early days of microcomputers was technologically rather than educationally driven. Lanza (1991, 18) suggests that the production of Hypertexts could come to suffer from the same technology driven situation. As he suggests:

At present, the main potential risk for the computer-based instruction field is the indiscriminate and simplistic adoption of this new and fascinating technology by instructional designers. The indiscriminate use of hypertext as instructional technology would give rise to a proliferation of programs whose projects would be technology-driven, rather than prompted by real needs and educational aims, and thus certainly of scanty effectiveness (Lanza 1991, 18).

He also suggests that the easy with which a Hypertext can be prepared could be a problem:

The apparent ease of development and production of hypertext-based instructional software could lead developers to disregard methodological and theoretical design issues. Instructional software must be developed by applying a design methodology which promotes, as a first phase, the cognitive analysis of instructional needs and objectives. The tutoring principles applied in order to manipulate the environment to exert a positive influence on the learning process should be derived from current theories on learning and teaching. Only then can the selection of the most suitable instructional technology be made, observing strict criteria for the choice (Lanza 1991, 18).

Jonassen (1988, 15) states that "Hypertext design is theory-rich and research poor". He identifies a number of areas as being ones worthy of research.

The first of these is how to structure hypertexts to ensure the maximum amount of learning takes place. Jonassen suggests that this can be done by finding out how users use unstructured hypertexts. From observations, a method of path analysis should be developed and using this, relate the paths taken by individuals to such things as cognitive style of the learner. An alternative approach would be to structure hypertexts according to various models and to assess the effect on the learners. The knowledge structure of the learners could be identified and this could be compared when unstructurred and structured hypertexts were used.

The next area worthy of research identified by Jonassen (1988, 15) is that of how knowledge structure can be assessed. While Jonassen notes that this can be done using pencil and paper methods, he states that there is now software such as Learning Tool and SemNet which will perform this task. These techniqes can be used to compare the cognitive structure of people with individual differences using different types of hypertexts. He states that "These tools can readily illustrate the reorganisation of knowledge structures, which [can be] defined ... as a conceptual measure of learning" (1988, 15).

The information gained from the path analysis of hypertext documents can be used to generate designs for effective hypertexts. This, according to Jonassen is a 'bottom-up' approach. He continues:

Preassess learner styles and assign users to hypertexts that have structures that are consonant with or dissonant to their preferred learning styles or provide the learners control of the choice. Preferential matching of learners to instructional treatments based upon their individual knowledge structures has been the theoretical goal of designers of intelligent systems. Hypertext provides that possibility (1988, 15).

Jonassen also proposes a 'top-down' approach to the design of hypertexts. The starting point of this is the the content structure, or an expert's knowledge structure, in which case it is assumed that the knowledge structure of the expert will be mapped onto that of the learner. This, however, raises a further consideration, how is an ideal knowledge structure defined? There are several ways of doing this which are identified by Jonassen (1988, 15). The first is to develop a cognitive or semantic map of the expert using quantitative methods. To do this, the expert is required to complete word associations of all the related concepts in the content domain. "The intercorrelations are multi-dimensionally scaled to generate a structural map". This process can also be undertaken using computer software as described above. Jonassen also states that it can be performed by using an intelligent front end to the hypertext using expert systems:

Expert tools are now available that can define and depict the knowledge of an expert. These are examples of frame-based systems, which define frames or schema and instantiate those with information from other frames. These systems are practical replications of schema theory. They enable one to define the knowledge of the expert as well as convert that knowledge into a series of if-then rules. This structure could then be applied to the hypertext (1988, 15-16).

The final area in need of research is identified by Jonassen as being how the information gained from a hypertext can be integrated into the learner's cognitive structure.

Conclusions

Assuming that the ability to access information in a non-linear fashion is of benefit in the learning process, hypertext would appear to have great potential in the production of efficient computer based instruction materials. There are, however, numerous areas that must be researched. Firstly, it must be shown that material from a hypertext can be mapped efficiently on to the learners' knowledge structures. Secondly, the issue of learner control must be resolved. The majority of the work performed so far in this area has been on 'pre-hypertext' materials which were designed to be used in a fairly unstructured fashion. As hypertext represents a totally different learning environment, it cannot be assumed that the current research findings about user control can be extraploated directly to hypertext. If, in fact, it is shown that complete user control is not effective for hypertext, then it will not be desirable to use unstructured hypertext systems for education.

The third area for research is that of navigation. Many researchers claim that users have problems navigating around large, unstructured hypertexts. There would appear to be two main factors which contribute to this. The first is that the navigational tools supplied with the hypertext are not efficient. More guidance should be given to users as to where a link will take them, where they are at the present time and how to return to a previous location. The second contributing factor is that users are not familiar with working in this type of environment. One of the advantages seen for hypertext is that very large amounts of information can be incorporated into one system. However, it can be asked if the use of such a complex system is a suitable way to introduce new users to hypertext or if it would be better to allow them to use smaller and more structured systems until they were familar with the system and how to navigate through it.

While hypertext is seen as having many educational advantages, it is restriced by the fact that it uses only text. This situation is overcome when other media are added to produce hypermedia. This will be dealt with in the next section.