Visual Metaphors and Authoring

John Hedberg
Barry Harper
University of Wollongong

ABSTRACT: With the introduction of interactive multimedia learning materials and the movement towards constructivism as an underpinning philosophy for the design of democratic learning environments, the organizing metaphor of the authoring system has become critical to the effective design of the final learning environment. There is a need for easy-to-use development tools, which incorporate alternative visual metaphors in the development process. There exist a multitude of authoring tools which vary in the visual metaphors they employ to provide access to information and tools for project construction. In this brief discussion we seek to link the visual metaphors with the constraints and advantages about how an authoring tool supports constructivist instructional strategies and how it enables different users to create new knowledge constructs within these environments. We suggest that new metaphors for authoring need to be developed to match current theory. Through a case study of MediaPlant, we have sought to indicate some directions for authoring tools.

Development of educational software has had a long history of use of authoring environments that have enabled instructional technologists rather than programmers to design and develop applications. The advantage of these tools has been that the designer did not need to be highly skilled in high-level languages, but could use a simpler construction set of pre-programmed modules. The disadvantage was that the developer was limited to the pre-programmed modules available and to the underlying assumptions of highly-structured instructional design models adopted by the tool. Unless the developer had access to sophisticated programming skills, and the tool allowed externally written code, the complexity of the resultant project has been limited. If the visual metaphors employed by each tool are examined, then we contend that the types of projects that are built reflect both the types of metaphor and the constraints that it brings to the learning strategies.

In this paper we will refer to the authoring application as a "tool", the collection of resources which are collected together as a "project". Thus several tools can construct standalone "projects" (Director creates a "projector") which can be distributed to learners individually. Some also focus upon the recording keeping aspects of the system and creates student files which are their responses to the embedded questions. Some tools work directly with the visual display, such as Digital Chisel, HyperCard and Oracle Media Objects and all changes and design decisions are represented in visual changes to the display. Other tools such as HyperStudio seek to provide a simple "Meta-view" where the user can create a display directly on the screen and then later through a "slide-sorter" reorganize the presented sequence of the screens. Yet other tools provide a more comprehensive meta-view, for example, Authorware, shows the flowchart of the algorithm being employed as well as the runtime window. Changes in one display are reflected in the other. Director employs a visual metaphor which is time dependent and displays the resources in several windows, from which the user can construct a project by dragging and manipulating the elements. Additional functionality can be provided such as graphics design, but these elements are derived from the original reason for the tools. For instance, Director, has been derived from an animation package and this is the underlying metaphor which drives the construction.

Visual Metaphors in Practice

Designers have to work within the prevailing metaphor of the authoring environment. For example, if a designer employs Director, the prevailing metaphor is theatrical and time dependent. Events occur on a stage and are managed by a score which dictates the depth and movement of each object over time. By contrast, HyperCard, created a card metaphor where links between discrete representations of objects are embedded in an hypertext relationship. Icon Author and Authorware create a complex flow chart (algorithmic) structure which is used to design the experience but is not displayed to the learner as they work through the pre-determined paths that are reminiscent of the traditional concepts of programmed instruction. (In fairness to the origins of Authorware the frameworks are based in the systematic design of instruction and the focus on "question episodes," from which a series of optional paths are taken and responses evaluated to determine the nature of what appears to have been learned).

Several limitations exist with authoring tools: Some tools do not allow cross-platform development and the speed of the resulting instructional packages may be limited by the interpretive layer which the authoring environment is built upon. Speed and machine requirements of the environment may also limit the interface options as well, for example, instantly updating graphical representations of directly manipulated variables, and allowing direct manipulation of screen objects.

However, today most systems do offer cross-platform solutions such as SuperCard, Director, Authorware, and Oracle Media Objects, but many of these products have not been enthusiastically accepted by the school education market requiring excessive amounts of memory to run, or requiring complex skills to make projects with the tool. One package, HyperStudio, has been accepted in schools as it requires less sophisticated hardware to run and the design structures are comparatively simple.

Given the changes to hardware technology options, what frameworks and visual metaphors need be employed to enable more than highly-skilled designers to use the tools for project creation?

A Constructivist Framework for Multimedia Development

There are several key writers who have called for a reassessment of instructional design models for development of technology supported learning environments that assume a variety of constructivist views of the learner. Hannafin and Land (1997) have suggested that we should be aiming for open-ended learning environments, Jonassen and Tessmer (1996-1997) have suggested that we should be aiming at new learning outcomes, and Cunningham and Duffy (1996) have suggested metaphors we teach by.

Cunningham, Duffy and Knuth (1993) devised seven pedagogical goals for designers of constructivist learning environments:

  1. Provide experience with the knowledge construction process
  2. Provide experience in and appreciation for multiple perspectives
  3. Embed learning in realistic and relevant contexts
  4. Encourage ownership and voice in the learning process
  5. Embed learning in social experience
  6. Encourage the use of multiple modes of representation
  7. Encourage self-awareness of the knowledge construction process

Savery & Duffy (1996) also espoused three principles that characterize this philosophical view in modern technology-based learning environments. Their views encompass much of what the previous authors have proposed but do not explicitly address the issues of multiple perspectives or many world views. These principles are:

  1. Understanding is in our interactions with the environment.
  2. Cognitive conflict or puzzlement is the stimulus for learning and determines the organization and nature of what is learned.
  3. Knowledge evolves through social negotiation and through evaluation of the viability of individual understandings.

Recently, Duffy and Cunningham (1996) have developed a list of seven "metaphors we teach by" as basic assumptions for design.

  1. All knowledge is constructed: All learning is a process of construction
  2. Many worldviews can be constructed: hence there will be multiple perspectives
  3. Knowledge is context dependent, so learning should occur in contexts to which it is relevant.
  4. Learning is mediated by tools and signs
  5. Learning is an inherently social-dialogical activity
  6. Learners are distributed, multidimensional participants in a sociocultural process
  7. Knowing how we know is the ultimate human accomplishment

In sum, since we have seen many writers attempting to develop guidelines for developers of software that will support new frameworks of learning. Jonassen and Tessmer (1996-1997) have also questioned the commonly used taxonomies of learning that are the basis of our instructional design models, proposing that engagement with a greater range of learning outcomes is essential for meaningful learning. They have suggested a new framework for specifying the types of learning outcomes modern learning environments should be developing, viz.:

In response to these design suggestions we have sought to change our design model that we have been using in the Interactive Multimedia Learning Laboratory to one that is based on a constructivist approach to instructional processes (Hedberg et al, 1994). A number of multimedia design models have been developed which illustrate the combination of complex learning environments and which also give students their own real control over their learning environment. This model is based on a more organic and iterative approach than traditional instructional systems design as it attempts to frame the design process in a constructivist framework.

Figure 1: The design process used as the basis for interactive multimedia package development.

Phase one takes the basic information derived from a needs assessment and converts it into a description of the Project space-the information which is to be included in the materials, how it is structured, what the target audience understands about the information and how it might be structured for the audience. A possible structuring device might a concept map of the ideas and links that are to be included in the project.

The second phase reviews the basic description and seeks to link the elements through an appropriate instructional or interaction strategy. It also seeks to identify metaphors which help both the design team and the final presentation of the information structure. The outcome of the second phase would be a formal description such as a design brief. The detail would enable the reader to understand the underlying knowledge structures and the ways it is proposed to link them conceptually and intuitively.

The third phase is a third pass at the same material, this time with the express goal of linking the design ideas into a potential presentation structure. One output of this phase would be an interactive mock-up of the interactive materials using such tools as HyperCard or Toolbook to illustrate not only static display of information but also the graphical and visual metaphors used to create understandable links. The information included in this prototype may include visual, motion, static graphics, sound, and data landscapes as appropriate to the concept under development.

Implications for a Software Environment

The key implications for an authoring environment that support the design model would be that the authoring environment should include:

Multiple World Views

As with alternative conceptions of the learning task, the authoring environment needs to provide for the user a series of views of what they are trying to construct. Most visual structures of authoring tools allow for at least two views of the process. All provide a "runtime" view but many also provide a "meta-view," which enables the user to organize elements into a structure or strategy. Several programs also provide for more than the requisite two views. For example, Apple Media Tool provide a series of windows onto its world, in addition to the runtime view, users could see a representation of the screens and links between screens as well as lists of media resources that could be selected from inclusion on each or any screen. Director provides for the meta-view in terms of a score, but other views are also available, such as the cast and controller. The utility of multiple views does need to be reexamined.

Metalevel views are provided by Director which is a time dependent score, with Authorware it is an algorithm for sorting out steps and complex options. mTropolis (recently killed by Quark) provided two views, an object-oriented design view as well as the platform runtime view. Because it used an object-focus, each object could be constructed and then could be used as a self-sustaining unit within each construction project. Thus great ideas or objects could be re-used in several projects.

What world views are most useful? Do they vary between users, for instance, does it differ if we are asking learners to undertake a multimedia product as a demonstration of their understanding, compared to supporting instructional designers/developers create an interactive multimedia title?

Another aspect of the world view within a package is the development of a useful metaphor in which the user can control elements to achieve an outcome. A useful view will enable the cognitive load of the exercise to be diminished (Sweller, 1988). If the metaphor enables intuitive processing then the instruction required to get a novice user to produce something with the tool is reduced. The inherent metaphor of a score was great for creating an animated time-dependent sequence but it is probably less relevant today, where we are asking students to construct their own view rather than respond to a pre-set suite of options. We suggest that the product has been pushed into a role where time is of little relevance but it adheres to the visual metaphor. As we move to a virtual world and use embedded QuickTime VR, what other metaphor might need to be adopted?

Cognitive Tools and Authoring

Several authors especially David Jonassen (Jonassen and Reeves, 1996) have called for the employment of cognitive tools within interactive software. These tools which are designed to assist with generative learning strategies are not often included in software applications. Within a traditional framework these tools are rarely available to users as they complete a tutorial program, but within the context of recent theory what might be provided. We have tried to develop the visual metaphor of the Personal Digital Assistant (PDA) in our product Exploring the Nardoo. <> (See Figure 2) This type of visual tool would very easily be transferable across interactive learning environments.

The questions here we would ask include:

What cognitive tools should be supplied with any application? Can a uniform set of visual representations be developed so that learning new conventions can be minimized?

Figure 2: The Personal Digital Assistant Notebook

The process of using source material within the package in support of an investigation has been enhanced to allow the student to...

The joint combination of note book and viewer better equips students to view and then critically evaluate or compare different representations of the same information concept. By collecting different media representations of the same topic and "flipping" between these representations at their discretion, the student has the opportunity to establish cognitive links between different media forms which compliment each other and support a central theme or information focus.

The support tool also provides the ability to record thoughts and impressions "on-the-fly" whilst examining media stories. This provides the potential for students to reorganize or revise their thoughts to better "make sense" of what they see and hear. Students are able to document their emerging ideas in support of an investigation or problem solving exercise whilst viewing different media. This provides support in the formulation of new schemata in the process of accommodating the new information.

By way of illustration, whilst viewing a video segment related to the issue of water conservation, students may have their attention drawn to a specific aspect such as better ways to manage water use within the household. The Nardoo PDA provides the means by which they are able to search out other related information (text, graphics, radio/television reports) and store it within their notes "alongside" previously captured information. The different perspectives represented within the source media items can be viewed, edited and brought into context within the student's framework of notes they are making.


To this point, we hope we have established that there are a number of issues for which the choice of an appropriate visual representation and or metaphor can support the range of users who might use an authoring tool. Several other aspects about an authoring tool environment might also be considered as part of the development of easily learned skills to express oneself and in more than traditional forms. Key to any environment is the ways it sets up forms of interaction and supports their simple execution by the tool user. To date, we are still struggling with questions such as:

What visual support structures are required to create a learning environment? Is realism enough to make it authentic or does there have to be some way of interacting with the objects so that the task of manipulation is simple?

Under a constructivist framework is the question not about learner control of the interaction devised for them, but the way in which the learner can create the interaction?

Are we really interested in allowing the learner to go down our prepared path and does it matter if the support for options is provided?

How do you support learners in an environment which has visual representations for navigation?

Creating a Project

Often a tool will require a learning overhead to get a project started. Is there a representation which will support users of different skill levels, or does it really require different products? Consider the issues that relate to the design of new projects.

What useful views are there for a construction project? Does it differ from student, professional designer, programmer?

What general support tools should be available in an authoring tool? What cognitive support structures are necessary? Ferry et al, 1997 have demonstrated the importance of using cognitive mapping tools to generate understanding of complex knowledge domains.

How should an authoring tool generate an authentic context? When would a 3D view be required, how should a navigable space be represented and displayed?

What useful visual representations should an object possess, that show its attributes, current state, and actions?

What are the best ways to visualize strategies for moving through a project?

Is an algorithm more useful than a hierarchical information structure?

What automated tracking needs to be included in a package and how should it be represented? Paul Fritze, 1994, has suggested a number of options, do any of these become useful? Remembering the problems of trying to use audit trail data, automatic collection devices rapidly generate lots of data but much of it is un-interpretable (Williams & Dodge, 1993).

What organization and support structures (annotation devices, media specific views,...) are needed to assist in the development of projects which have potentially thousands of items to track and integrate?

Should a level of customization be available to the project developer so they can 'tailor' the visual organization of their work to a world more in tune with the content and interactions they are developing?

Case Study of MediaPlant as an Authoring Environment

With the increasing demand for cross-platform applications and the need to have high performance, in terms of speed and features, the Interactive Multimedia Learning Laboratory team set out to develop a software development environment that would satisfy these needs and also not constrain the instructional design models used. It was envisaged that the development environment would facilitate the production of complex learning environments, i.e., this would be an authoring tool that offered complete flexibility in the design process and also high level performance on entry-level machines.

MediaPlant consists of a development program and a runtime program. The development program is used to construct and test the project, which is then distributed with the runtime or "player" program. The development program is Macintosh-based while the runtime player has been developed for both the Macintosh and Wintel compatible platforms. The software environment is based on a C++ application framework tailored for large scale multimedia development.

The memory requirements for MediaPlant are 8-10 MB for the development program (the more the better) and approximately 2-3 MB for a runtime application that uses one full screen window at a time. An intelligent coaching mechanism built into the MediaPlant engine makes use of any additional memory given to the application, minimizing file access and speeding up screen changes by retaining recently used media elements in memory. The file structure used by MediaPlant is also designed to minimize the number of accesses required, for better performance when run from slower storage systems, such as CD's.

The initial development program metaphor used is based upon a tree file structure. A graphic design interface which will allow flowchart or hypertext visual representation of the design is also under consideration.

The file structure metaphor (Figure 3) provides a similar interface to the Macintosh Finder, allowing easy access to all of the media, screens, screen objects and scripts within a project. Drag and Drop, as well as copy and paste, make adding, editing and reorganization of elements within a project simple. A designer has considerable freedom in how media is organized, including a choice between using external files, or storing media within MediaPlant's containers. Runtime screens can be displayed whenever desired, allowing easy graphical editing of items on each screen, as well as execution and testing of the product.

MediaPlant, within a multimedia software context, has enabled us to provide the user with innovative, integrated tools allowing them to store audio, video or pictures within the user's notebook. Provided the student's file is open on a machine with the CD-ROM inserted, this will display the media in a viewing window within a personal digital assistant which accompanies them throughout the project. The authoring environment allowed the development team to construct two organizing visual metaphors: a spatial geographic landscape to explore through a simple navigation system and a personal digital assistant to display collect and edit the embedded multimedia data. The general information access and structure is not confined to limited models.

Figure 3. Tree file visual structure used in MediaPlant for the authoring process.


As new theoretical views about technology supported learning environments develop, we need to be able to explore these ideas using tools that allow us to set up these environments quickly and easily. Tools that assist the task without complex dialogues for handling variables and options. We also need to review the visual and conceptual functioning of tools that we offer our children in their endeavors to express their ideas using the full range of media available to them. This is an opportune time to review authoring tools that we currently have, to suggest options that derive from educational issues, and to address some of the questions raised in this paper.


We wish to acknowledge the contribution to these ideas by the team who developed the Exploring the Nardoo products, in particular, the project manager Rob Wright, and our senior programmer Grant Farr.


John Hedberg
Barry Harper
Faculty of Education
University of Wollongong
Wollongong NSW 2522


Interactive Multimedia Learning Laboratory, copyright 1998, March 25, 1998

The buttons that appear below will be found at the top and bottom of each page of the discussion. The first button will take you back to the previous page (in this case, to the beginning of paper #25). The middle button will take you to the ITForum home page. The last button takes you forward into the discussion as it progressed on-line.