From Learning Technology in the European Communities, Ed. S. Cerri and J. Whiting, pp. 79-90, Kluwer Academic Publishers, 1992.

We will describe our 'integrated environment', which combines computer communications with other learning resources, as developed for the DELTA Pre-Pilot Project. From the point of view of learning and pedagogics, we were trying to establish that the use of computers to support learning means very much more than CBT, in which computers are used principally to replace the trainer. The use of computer communications is central to our learning approach. It enables learners to get personal support from their trainers and also from other students. It allows group discussion and group projects to be used as a central learning tool. At the same time, many of the benefits of CBT are retained: the learners and tutors work in their own time and from their homes or workplaces, avoiding the need to travel or interrupt work. Moreover, the possibility of continuing support from trainers, and self-help from groups of learners after a training course has ended is also available.

With this educational approach in mind, we will describe our ideas for the infrastructure of a European Electronic University, which we see not so much as a single organisation, but rather as a networked support structure linking a range of organisations, which might include distance learning institutions, training material developers, student organisations, or government organisations.

Finally, we will describe the software tools we are currently developing to facilitate the development of advanced interactive environments. Our HyperView system enables graphical environments to be developed by direct manipulation, as with Apple's HyperCard, but with multiple layers, graphical inheritance, and flexible replacement of parts of any layer. There is an underlying language which can be used to attach behaviour to any object created in HyperView. It is an AI based-language, which we are calling Joshu, which is inherently concurrent and has the syntax of the language Scheme, a functional sub-set of Lisp.

The predominant characteristic of our time is rapid change. Our technology, organisational forms and even our political and economic systems are undergoing unprecedented rates of change. To cope with this change we need an educated and informed population, and an adaptable workforce. The skills people learn in schools and colleges can no longer last for a lifetime. Continual updating and training are essential.

Moreover, the need for this continual learning is of such a scale that it cannot be confined to full-time education. People need to learn new skills while carrying on with their normal working lives, with learning taking place from the home and in the workplace. Hence the growing importance of open learning.

It is especially important that the systems set up to satisfy these learning needs are efficient from the point of view of the learner, of the provider, and of employers.

From the point of view of the provider and of employers, the learning materials must be easy to create and must reflect the best available practice. Quality is vital, yet the very great effort required to produce computer- based materials today must be reduced. Nonetheless, learning materials must suit local needs.

From the learner's point of view, learning systems need to be convenient to use and low in cost. They must fit in with the learner's daily schedules, yet must provide access to help when required. Where they require the use of computers, the time required to learn to use the computer must be at an absolute minimum. The user environment provided must be intuitively understandable so that the computer doesn't get in the way of learning, and to reduce the back-up support required. Especially important is that the learning approach must suit the way people learn naturally.

The computer-based learning systems we have been developing aim to find ways to satisfy all these requirements. We describe below the learning approaches and the user environments we have been developing. The strategy we are proposing for a Europe-wide infrastructure in which to deploy them aims to create a synergy between learners, trainers, and developers which allows high quality materials, based on best practice, to be widely used, yet which are tailored to the needs of the local culture and local employers.

We envisage a European Electronic University supporting a variety of learning styles, to suit the needs and expectations of local users throughout Europe. As illustrated in our DELTA prototype below, many different types of educational resource can be included. Nonetheless, we have been considering the issue of learning approaches, and would like to briefly present our conclusions.

Learning approaches can be divided roughly into two camps: a didactic approach and an interactive approach.

The didactic approach is that used by traditional educational systems. Material is presented to the learner by the teacher, who has carefully pre-digested it and ordered it into a logical sequence. The material may be presented in the form of lectures, texts, or more recently, computer-based lessons. Learners are expected to comprehend the material, and check their understanding through questions, quizzes, and exercises. In the didactic approach, the emphasis is on the teacher/trainer/expert and their ability to understand and clearly present the material to be learned.

In contrast to this, in the interactive approach the emphasis is on the learning processes experienced by the learner. It sees the learning process as one in which the learner actively constructs an understanding of new material, and so tends to be based around problems, projects and presentations by the learner.

As shown in Figure 1, the interactive approach is more inclusive than the didactic approach. There is still the need for good materials, clear understanding and structuring by the teacher, but this is cast in a broader pedagogic framework, with the learning process at the forefront.

The interactive approach can either be applied at the micro level, where the learner is presented with a series of structured lessons by the teacher, but these lessons are problem based, or at the macro level, in which the course is structured around larger scale projects, in which the learner is given a set of resources to use. The specific learning outcomes of the latter are usually less closely defined than those of the former.

Our DELTA prototype, with its emphasis on communication and the provision of a variety of resources to assist with a project, is clearly an illustration of an interactive learning approach on a macro scale.

This project produced a prototype of a fragment of a hypothetical course on 'Digital Telecommunication for Managers and Engineers'. The aim of the fragment was to teach students about the use and features of high-speed modems. The prototype was built using HyperCard on an Apple Macintosh as its development system.

The prototype incorporated a computer conference and mail system with hypothetical students from all over Europe joining in discussions and group projects. The conference system used was CoSy, running on the Open University's VAX cluster.

We developed a 'front end' to CoSy which automatically connected to the VAX through a modem, then uploaded any previously prepared messages and downloaded any waiting mail and conference messages. These were then repackaged into hypertext format, with the linking structure of the messages preserved. In this way, the conference system could be used conveniently off-line without losing the conference structure.

Moreover, the facilities for navigating through linked sets of messages are much better than those on the same conference system used on-line. Finally, the automatic facilities keep communication costs minimal without the loss of functionality (through loss of linking structures and access to old messages) which normally occurs in conventional systems with automatic downloading.

Figure 2 shows some of the screen designs used. The lower portion of the screen is what we call the 'sill'. It remains constant while all the rest changes. It functions as a personal desktop, to give the user instant access to all the facilities available and to permit rapid switching between one task and another. It includes icons giving access to personal productivity tools (simple word processor, spreadsheet, drawing tools, filing system, etc.), to the course materials, and to the communication system.

In addition to the communications, there were six teaching packages, illustrating different uses of the computer.These included:

Figure 3 below shows some of the screen designs used. The learning style envisaged with these materials was project based, centred around the computer conference system. The project chosen was the specification of a modem meeting certain requirements. Students would be expected to work in groups to produce their project reports. Each student would have the computer-based resources described above, plus some printed material, plus the perspective and knowledge of the other students in the group.

The completed DELTA prototype illustrates the style of computer support system we envisage.
To summarise its features:

Tailor-made materials are designed and developed to suit the specific educational needs of a given employer or educational provider. If professionally produced, they offer excellent results, well suited to the needs of the clients. However, they are notoriously expensive, often requiring several hundred hours of effort for each hour of learning materials produced.

Generic materials are relatively cheap, off-the-shelf commercial packages which are 'generic' in the sense that they are meant to apply to a wide range of users. The large effort required to produce them is thus spread over a large number of users, so that the cost remains modest. A common difficulty with them is that they may be nearly, but not quite, what the user wants. Because they are general-purpose, they cannot take into account the specific needs of a given user.

A third approach, which has been widely discussed in DELTA circles and which we favour, is to produce customisable, or 'partly-formed' materials. This approach combines the advantages of both off-the-shelf and tailor-made systems. Customisable materials would be designed for wide applicability, but can also be easily modified to suit the precise requirements of the user.

Customisable materials should include entire courses as well as smaller components. Thus a would-be developer could look through existing courses to learn about the way material has been structured, what has been included, and what teaching strategies have been used, thus reducing greatly the early stages of course design and development.

The smaller components should include sets of preformed interface objects: buttons, fields, menus, controls, pop-ups and more sophisticated components tailored to the needs of learning. There should also be subject-specific libraries of components. For example, these might include models of classes of machinery, or databases specific to the needs of particular industries.


5.2 Organisational form

Given the preceding discussions of pedagogical approach and user environment, we can now turn to the question of suitable organisational forms for a European Electronic University. It is clear from the diversity of needs around Europe, that a single, monolithic organisation would not be appropriate. We suggest instead, a fairly loose confederation of local organisations, sharing an evolving set of software and hardware platforms, and linked by a variety of telecommunications networks, both terrestial and satellite based.

Each participating organisation would provide some or all of the services shown in Figure 4:
Libraries of courses and course components, tutorial and discussion groups, feedback and accreditation, and research and development to support the other services.

The function of the libraries is as much to support the needs of developers as of students. The students will be primarily interested in complete courses, but may also use component materials to develop projects and presentations. Developers will be using course materials and component materials both for ideas and, where appropriate, as the basis of the learning materials they are creating.

Similarly, some of the discussion groups will be there for the benefit of students, while others will be places for developers to 'meet' to obtain advice and help with the materials they are working on.

The tutorial and discussion groups in general will be based around computer conferences. In this way, participant can work from their own homes and offices, at their own pace. Moreover, it will be possible for tutors to join from distant locations and even 'guest lecturers' to join on a temporary basis.

Some of Electronic University nodes would be based around existing distance learning institutions, such as the British, Dutch, Danish or Flemish Open Universities, etc. They would be expected to provide the full range of services, and would also act as support organisations for the other nodes.

Other organisations linked as part of the Electronic University, might only provide some of the services. For example, training organisations, based around existing firms producing training materials might provide libraries to suit their own clients, accompanied by tutorial and discussion groups, but would rely on the Institutions for accreditation, where appropriate, and for the research and development to support new courses and tools.

Industry groups, supported by particular employers or trade organisations, might provide their own nodes, with libraries commissioned to support the needs of their own sector. They too, might rely on the Institutions for accreditation when needed, but might also offer their own vocational accreditation.

Student-run organisations might also set up their own nodes. They would provide libraries to suit the needs of their own members, with self-help discussion groups, or perhaps with tutors or trainers hired in some cases.

Finally, government funded nodes could provide useful support services, funding research and commissioning materials to satisfy national needs.

The most important feature of the infrastructure we are proposing is that it should create a synergy, which enables Europeans to access to a pool of the best ideas and materials available, rather than only those generated locally. To do this, two conditions are necessary:

1. The libraries and discussion groups of the various nodes must be interlinked, allowing    each node access to the others.

2. Barriers to the use and modification of the materials in the libraries must be minimal.

To best satisfy the second condition, we propose that there be a large pool of customisable materials and tools which are available to members of the Electronic University free of charge rather than sold when needed. These materials, including source code, would form a public library.

Put more precisely, payment for these public library materials would not be made on the basis of use. Obviously, the producers of the materials would have to be paid, and it is sensible for talented developers to be well paid for their services. However, to create the synergy needed, it would be best for payment to be kept separate from use, to maximise the availability of the materials, and to enable later materials to be based on earlier ones. This would create an evolving pool of learning materials, continually representing the current state of the art.

There are a number of excellent precedents for this approach, with UNIX in its earlier days and Kermit as prime examples. The UNIX community consisted of salaried people who produced material which was then freely distributed to other members of the community.

Similarly, in the Electronic University, we would envisage various ways in which development teams would be paid for their efforts: Some would be employees of the educational institutions, whose income arises from course fees and from grants from various sources. Others would be employed by industry organisations, and would produce material especially suited to that industry, or by government organisations. Even the student organisations could contribute material, either produced by their members or commissioned out of membership fees. Finally, free-lance developers could sell materials to the libraries, which, once paid for would become part of the freely available pool.

One final point: There is no need for the libraries of the Electronic University to consist exclusively of free materials as just described. They can also contain commercial products which are sold for use in the conventional way, and moreover, will provide a new market for such products. However, it is likely that the public materials will be those on the leading edge, simply because a much larger number of good minds can be brought to bear on them, as they evolve over time.

We have now described learning approaches, user environments and organisational forms for an Electronic University. The final, and perhaps the key ingredients to make it all work are suitable software platforms which enable interactive materials to be developed easily, and which are inherently suitable for creating customisable materials.

We will describe one such platform, the HyperView system we are developing. However, we should point out that we do not think a single, universal software platform is desirable, as it is unlikely to be suitable for all needs. Rather, Europe would be better served with a small set of software systems, each finding its appropriate niche. With the public library approach just described, all the systems used would continually be changing, and borrowing ideas from one another. The purpose of this description of HyperView is to contribute to the pool of ideas for appropriate software platforms for the Electronic University.

6.1 HyperView's graphical interface

HyperView will provide tools for creating graphical interfaces either by use of drawing and painting tools, or by selection and modification of graphical objects from built-in libraries. A built-in communi- cations application will provide access to external libraries of subject-specific components.

A user environment in HyperView is built up with multiple layers of graphics, each of which may cover part or all of the screen. Each graphical layer is called an 'acetate'. Acetates may be composed either of bit mapped graphics or drawn graphics. What the user sees is then determined by combining one or more acetates into a 'View', which specifies their order and location. A View may also contain sub-Views, which in turn may contain sub-Views, etc.

The basic mechanism of change in HyperView is to replace one (or a group of) acetate layer(s), a View or a sub-View with another. This flexible, multi-layer approach is much more closely matched to the needs of interactive environments than, for example, the relatively limited structure of HyperCard. It permits the developer to structure an application into contexts and sub-contexts, with only the appropriate parts changing when required.

6.2 Concurrency

In general, interactive user environments consist of some parts which are user controls (buttons, pallettes, sliders, menus, etc.), and other parts which are either purely passive, or which react to operations of the controls or to changes to other objects. The user can choose to operate any of the controls at any time, in any order, and the application must react accordingly. In many cases, such as dynamic models, controls should be active while the model is running, so that parameters can be changed.

In conventional graphical applications, based on sequential languages, this user flexibility is a major cause of difficulty. The programmer must anticipate all the possible combinations of actions the user may choose. Controls which are active while other computations are running are especially difficult.

In HyperView, the user interface is seen as inherently concurrent. Controls are generally seen as concurrent computations which are temporarily delayed: waiting to become active until there is some user input. Other objects too, will normally have delayed computations, waiting to be activated either by the controls or by other objects. HyperView thus provides a natural solution to the problem, with programmer support at its deepest levels.

This concurrent nature of the user interface is reflected in the concurrent nature of Joshu, the language of HyperView. It is what gives HyperView its special power and appropriateness for interactive computer applications.

6.3 Joshu

Joshu is the language of HyperView in much the same way as HyperTalk is the language of HyperCard. Any graphical object created in HyperView can be given behaviour by writing a Joshu script. Joshu is actually a non-deterministic, concurrent dialect of the Scheme programming language, a dialect of Lisp. Joshu is concurrent in the sense that there can be several points of computation progressing at any one time.

Joshu is an elegant and sophisticated language which professional programmers will enjoy, in contrast to the languages built into HyperCard and other authoring systems.

6.4 The object systems

An object-oriented approach, with inheritance, is very important in HyperView. It enables developers to create new objects by customising other, more generic objects (for example, objects in a library). It provides an effective mechanism for structuring an application. Changes can be made to an object higher in the inheritance hierarchy which will then apply to all objects which inherit from it, although the lower objects can specifically override the changes if desired.

Sometimes a developer may wish to create an object which is like another object, but is independent of it, in the sense that changes to the original object will not be passed to the new object. This is provided for by an additional mechanism, clone-of.

HyperView offers developers a rich choice of mechanisms to control inheritance and identity, kind-of and clone-of in Joshu, and connection by graphical context in the user interface.

6.5 The HyperView development support environment

The significance of HyperCard, and of Smalltalk before it, as milestones in development systems for interactive computer applications, was in the kind of support they provided for developers. No longer was the developer limited to a text-based programming language, stored in one or more text files, where they were edited, compiled (or interpreted) and run. Now much of the work was done by direct manipulation. Interface objects were created using paint tools or selected from a menu. The behaviour attached to each object was to be found and edited in a script attached to that object. Rather than searching through a text file, a script came readily to hand by simply clicking on the object. The effect of any changes to a script were immediately apparent. HyperCard graphics, buttons and fields could be copied from one place and pasted to another, taking along any associated scripts.

The result was an enormous increase in developer productivity, and in the case of HyperCard, the ability for non-programmers to develop simple applications quite easily. However, in HyperCard, complex applications still require sophisticated programming, and here the limitations of the language and the data structures provided become very apparent.

HyperView combines some of the strengths of both HyperCard and Smalltalk, with extra features unique to itself. From HyperCard there is the direct manipulation of graphical objects linked directly to scripts. From Smalltalk there is the object-oriented approach, extensibility, and the use of browsers to access relevant sections of code. This is combined with libraries of generic objects as the primary organisational tool.

Thus HyperView is ideally suited to the European Electronic University concept, as described above, and will, we hope be a system which greatly extends the possibilities both for the non-programmer and for the professional.

We have described some of the principal characteristics of the types of learning systems needed if Europe is to thrive in the current period of rapid change:

The development of such learning systems is the challenge for Europe and for the DELTA project for the 1990's.