Instruction Design and E-learning: Issues and Challenges

M Sasikumar,
Centre for Development of Advanced Computing, Mumbai,
ni.iabmumcadc|isas#ni.iabmumcadc|isas
(paper presented at Indo-ASEAN conference on e-learning, Hyderabad)

Abstract

Instruction design is perhaps the most critical component of e-learning today, since the effectiveness of learning is controlled primarily by the nature of instruction. Instruction involves the actual content as well as the way it is delivered. Much of e-learning, as is practised today, do not exploit the capabilities enabled by ICT, and rely on fairly well understood models of instruction. In this paper, we review the important aspects of instruction design and the current status of the field, and look at the challenges ahead. We restrict our attention to the design issues of instruction, in this paper.

1. Instruction Design and E-learning

The term instruction, in this paper, refers to the content that is delivered to a learner in an e-learning scenario. The actual type of e-learning deployed – synchronous or asynchronous, lecture based or activity based, supplementary or blended, etc – are not critical to our discussion. By the definition we use, we include not just the passive content, but the process of delivery of the content. The process of delivery is important in instruction design, since that is what encompasses much of the pedagogical issues.
As is now commonly believed, e-learning as a discipline is yet to mature across the world. Beyond the walls of corporate environment, where e-learning is used for employee training, e-learning has had very limited successes. While, there are factors such as technology, bandwidth, etc which account for part of this lack-lustre performance, there are serious concerns on the quality of content and the way in which it is delivered to the learner.
The overall e-learning environment need to be looked at carefully in determining the culprits for the poor performance of e-learning. A closer look at most of the e-learning setups, private and public, gives us the proof we need. Mostly the e-learning resources consist of e-books – electronic versions of hardcopy textbooks, with almost the same look and feel – and perhaps some slides and web pages. Most of them still fail to involve the learner in the learning process, to any significant extent. Assessment component is often limited to simple multiple-choice recall type of questions, and is sometimes used as a gate to restrict access to subsequent material. Application of concepts learned to practical scenarios is rarely done as part of the framework [1].
Thus for e-learning to achieve the effectiveness that it is capable of, attention is required on a number of issues ranging from a well organised LMS to effective assessment and feedback mechanisms. For example, over the last decade or so, learning management systems have not progressed much in terms of its ability to manage the learning process. There is little new in today's LMS that was not known or available in some form, a decade ago! However, above all these concerns is the necessity for effective content. No matter how good a school/college you have, no matter how well organised your educational system is, if you do not have well thought out learning resources and good teachers, the effectiveness of learning will be low. In this paper, we focus on the issues in designing effective content.
Designing effective content requires attention to three components:

  1. The knowledge/information that the learner is expected to learn. For example, 'Newton's laws of motion'.
  2. The approach (pedagogy) to introduce this to the learner.
  3. The resources used for this purpose - the picture, text, video, animation, example, etc.

For an e-learning student, there is an additional component - which is the medium for delivery and the interface used for this. This includes web, CD, etc.
In the next section, we build up on this categorisation and identify the broad aspects of instruction design. Each aspect is discussed briefly, identifying the relevant issues from the perspective of that aspect. We then focus on a few of these aspects which are the relatively more challenging ones. Section 3 looks at learning theory with focus on constructivism as the emerging focus area. Subsequent sections look at learning environments building on this model and the challenges in design of such environments. Towards the end, we look at issues of reusability and sharing of content, and associated challenges, and also concerns of how much of the instruction design can be automated or machine supported.

Aspects of ID

Considering the scenario of e-learning, one can divide the instructional component into four relatively separate parts.

The learner interface

This concerns the interface provided by the system to the learner. Generally, this is the computer screen, and associated multimedia devices as applicable. One is beginning to see devices such as mobile phones (m-learning) and iPods being targetted as potential delivery devices. These are likely to bring up additional constraints on the interface design and content delivery, given the different nature of these devices.
The concerns in design of learner interface, include the amount of information that should be shown at a time, the placement of the various types of information, synchronisation between various elements on the screen (e.g. Audio, pictures and text), choice of colour, font, etc, and so on. There are a fair amount of guidelines on these concerns. For example [2] provides a summary of studies concerning a number of issues in this realm. One such observation suggests that visuals, if used, on a page must relate to the content on that page; otherwise, the learning effectiveness reduces. This is contrary to the view by some group, that visuals can also serve to reduce the monotony of content and hence help retention of attention. The field of user interface design, and user centered computing also offers a lot of useful resources in addressing these concerns. These provide guidelines on choice of colour combinations, placement of elements, layout, naming, etc.
It is also important to look at physical comfort and cognitive load in the design of interfaces. Ideally, the learner's cognitive load must relate only to the subject that he is learning, and not involve having to remember numerous commands and conventions used on the interface. For example, annotating the parts of a picture directly on the picture, as opposed to providing a description below the picture (often requiring scrolling) reportedly reduces cognitive load and hence improves learning effectiveness.
The degree of attention received by different areas of screen have also been studied. This is important in placing various types of information on the screen.
Some of the aspects that affect the learner comfort are the usage of scroll bar (recommended to minimise), number of mouse clicks required to perform various functions (inversely proportional to the frequency of the task), use of bold face, uppercase, underline, etc in the text, and so on.

Navigation

Navigation refers to the provisions to move around within the resources provided. This has three distinct aspects. One concerns the navigation viewing the e-learning interface as another computer system interface. There is a famous guideline for user interfaces, which is also important and useful here: any given screen should provide information to the user indicating how he got to that screen, what he can do there, and how to get out of that screen. The provision of buttons for various commonly expected tasks (e.g. Take a test, make notes, make a book mark, log out, suspend/resume session) need to keep this in mind. Headers and titles indicating the nature of the current page are also important.
The second aspect of navigation is navigation within the course content. One expects e-content to provide freedom in moving around among the lessons, and not so infreqeuently learners tend to get lost! It is important to provide visual indicators identifying where he is in the content space. Some of the systems incorporate bread crumbs indicating the nesting of content (e.g. Introduction => need => example 1). Often a click on any part of the bread crumb takes you to the appropriate level, providing an intuitive and convenient interface. The section numbering schemes used in books also serve a similar purpose.
Navigation in content also has a pedagogical component. Studies such as those reported in [2] show that a complete learner centered or complete teacher centered approach is, in general, not appropriate. A teacher's guidance on what is appropriate to study (and not to study) at any point is important. Within these guidelines, the learner may be allowed freedom to choose his path. This way, the freedom is balanced with the danger of getting lost or disillusioned with the content. This has implications with respect to navigation. The system must clearly indicate where the user is in the overall content space (how much completed, and how much remaining), what content is now recommended for study, and what content he is allowed to follow. Systems like Vasishta [3] follow a traffic light model for this purpose. A similar approach is followed in the Marathi Tutor system [4] (see figure 1). In a tree-structured content overview at the left of the screen, a green flag indicates lessons that a learner is allowed to pursue, red for the lessons not allowed, and blue for those which are completed. An arrow indicates the current lesson. Learner can expand the lesson overview selectively to know the depth and breadth of the content at various parts.
Similarly, linking content with specific learning objectives, inclusion of advanced organisers, etc are also pedagogic devices for navigation aimed at minimising cognitive load on the learner.

Type of Content

Type of content, here, is primarily a syntactic notion. Content can be produced in different types varying in the media used and the senses for which it is meant. We can categorise the content as visual and auditory, from the point of view of how the system communicates to the learner. You can also categorise the content as text, static images/pictures, videos, simulations, etc. These may need to be read or listened or a combination. These also vary in the amount of interactivity and learner involvement that is incorporated. One can always build in these capabilities, at a higher level than the content per se – for example, the text could be presented in short segments, requiring some interaction from the learner at the end of each fragment.
One concern at this level is the resources the learner will need to have to use the content. This includes relevant softwares (PDF reader, Flash player, movie player with specific format support, Word processor, etc). It is useful to minimise this requirement. For example, providing much of the textual content in HTML or XML format, makes it accessible for most users on the Web, compared to making them into a PDF file or a proprietary document format.
Support for open standards is another important concern at this level. Wherever possible, it is important to use open standards for storing and distributing your content. Proprietary formats for documents, images, etc has multiple disadvantages. On the one hand, you need to insist that every user buy the relevant software for accessing the content. On the other hand, you are dependent on the particular software being supported for longer than the lifetime of your content, failing which your content may become no longer usable. For documents, one could use HTML, or Open Document Format (ODF) [5] which is fast becoming popular. ODF supports documents, spreadsheets and presentations using an XML based public specification. Similar concerns apply for audio and video files as well as images. Here the choices are not that clear – but open formats are fast emerging in these as well.
In the set of resource constraints, bandwidth certainly comes in, when you consider high density content such as high quality pictures and videos. While the available bandwidth is increasing with time, live/synchronous video where every learner listens to the same material at the same time, is mostly a bad idea, except when the class is spread across a small geographic area, well interconnected through a fast LAN/WAN. When the content is predominantly video based, bandwidth should be viewed as a significant constraint. Concerns of privacy prevent the learner to store and watch the video, thanks to use of streaming techniques. This means that whenever he wants to view the content, good quality internet link need to be available.
Type of content is also determined by another concern – the cost. Cost of development, both in terms of time and effort, vary widely among the various options here. Videos are easiest to produce, if a good faculty and good quality recording devices are available. Interactive simulations and animations are the most expensive to produce, since they involve substantial software development on a case by case basis. While text based content is easy on the resources required, producing good quality text is like producing a good quality textbook, requiring substantial effort in editing, layout, organisation, etc.
Video recorded lectures depend on being able to get faculty members who are good in their subject and who have good communication skills including presentation, white/black board usage, etc. The ability to make eye contact with the learner, by looking into the camera while lecturing, is extremely difficult for most faculty members. Therefore, this significant attention gathering device is often lost. For a distant learner listening to the video, either synchronously or asynchronously, a well organised set of slides being used as the base in the lecture is a good pace marker. Again, one finds few videos that make effective use of slides.
Long video sessions with no interaction built into the delivery system is a poor choice for content, given the low attention span of an e-learning student. Videos are best viewed as good supplements on select parts of the content – a video clip to illustrate the functioning of a plant, a dissection of an animal, etc. The strength of e-learning is the personalisation of learning (one person on one computer) and the good computing power at one's disposal. Interactive simulation programs where the learner can explore situations and learn is the best medium which exploits this effectively. However, very little of the content today is available in this form. We come back to this in a later section.

Organisation and Pedagogy

Organisation refers to the order in which the various concepts are introduced and pedagogy the way in which each concept is introduced. These are, actually, the most critical aspects in teaching a subject, and often underplayed in most cases of e-learning.
When there are a set of related concepts to be introduced or when a difficult concept is to be introduced through a series of other simpler concepts, organisation becomes important. Different teachers adopt different strategies for this problem, depending on the nature of the subject, the nature of the learner and the experience of the teacher. The part of instruction design dealing with this issue is called Elaboration theory, introduced by Reigeluth [6]. Approaches here include easiest first, most difficult first, spiral model in increasing complexity, etc. The notion of prerequisites for the concepts are also relevant here, since they provide additional constraints on possible valid orders in which the concepts can be introduced.
While organisation works at the set of concepts level, pedagogy looks at a single concept at a time (mostly). What is the best way to introduce a concept, how do we test the understanding of this concept, what activities are suitable to support understanding and encourage long-term retention of the concept, etc are important concerns here. For example, one can choose to approach the concept through formulation of a mathematical model, a visualisation, a real-life example/story, etc. This is also the most appropriate place for personalisation. For example, one can choose the type of examples or illustrations that best relates to the profile of the specific learner.
The pedagogy model to be used is also dependent on the nature of the concept and the depth of understanding. [7] defines a matrix structure illustrating this. Each cell defines a particular combination of the nature of concept and the depth – recall is the shallowest and find, requiring to find where the concept is applicable and then apply it, is the deepest.

Remember Use Find
Fact
Procedure
Principle

It is useful to identify which of the 9 cells, the intended concept belongs to. This enables a judicious choice of pedagogy. Merril's subsequent work on Instructional Transactional Theory [8], takes this idea further to automatically select a pedagogy given the matrix element.
In [1], Merril identifies a set of first principles of instruction, and gives the following prescription for effective instruction. Learning is facilitated when [1]

1.learners are engaged in solving real world problems
2.when existing knowledge is activated as a foundation for new knowledge.
3.when new knowledge is demonstrated to the learner
4.when new knowledge is applied by the learner
5.when new knowledge is integrated into the learners' world.

These provide a succinct set of guidelines for pedagogy in the design of instruction.

Learning Theory and ID

In the previous section, we briefly outlined the various aspects of instruction design for e-learning, and remarked that pedagogical issues are perhaps the most critical in this. Much of the concerns in the pedagogical aspects come from the theory of learning, a field with rich contents and even richer history. The early models of learning were of behaviourist nature, considering the learner as a partially filled slate with the task being to insert the missing pieces. The teacher would impart the missing pieces to the learner, almost surgically attaching them to the learner.
With increased understanding of how humans learn, this model gave way to what came to be known as cognitivist learning theory. Ideas such as learner model and strategies of instruction were introduced at this stage. The most popular model today is a refinement of cognitivism, named as constructivism [10]. This model introduces a number of powerful notions of how people learn, and also provides powerful directions to exploit the opportunities provided by computer and communication technologies. Not surprisingly, constructivism has caught the attention of many researchers and practitioners in the field of e-learning.
The basic premise of constructivism is that learning is individual, in that learning takes place when a learner revises his current mental model of the domain. Neither the existing model nor the nature of revision is directly under the control of the teacher. Therefore, in general, there is no way to ensure that identical learning takes place among a set of learners, against a given instruction. Each learner absorbs the instruction depending on his existing mental model and other characteristics including his learning and behavioural style, and 'constructs' a revised mental model. The task of the teacher becomes, primarily to facilitate this process as well as possible.
Given the focus of constructivism on individual variations in learning, the model encourages use of exploratory and collaborative learning scenarios, unlike the traditional 'lecture' mode delivery of instruction used in behaviourist models. And the strength of e-learning is precisely in supporting these areas of exploratory and collaborative learning. However, much of the e-learning practised today relies on behviourist model of learning.
Constructive learning environments (CLEs) and learning through games are two major approaches built on constructivist model of learning. Learning through games creates a game-like environment where the learner plays an interactive game, and learning is integrated into the process. While there are debates on the desirability and effectiveness of using non-serious devices such as games for learning, in the face of lack of adequate evidence, the field has strong supporters too.

Constructive Learning Environments

Constructive Learning Environments (CLEs) are learning environments built on a constructivist learning model. These systems provide effective playing grounds for learners to try out what they learn and get constructive feedback. Often the playing ground drives the learning as well, since the focus is on learning whatever is required to handle the assigned task well. The playing ground can take a variety of forms from the simple descriptive problem solving to simulated building of a device. Examples of CLEs are the chemistry lab simulation available on the web (http://www.chemcollective.org/vlab/vlab.php), and the practice environments provided in the www.w3schools.com for learning languages like HTML, DHTML, Javascript, etc.
Along with most lessons, w3schools provide the user two different windows, one where he can modify a code fragment that is pre-loaded, and the other which shows how that modified page would look on a normal browser. This kind of setups enable students to explore the usage of various constructs, in a simulated reality, giving them a feel for how these actually work. It also enables them to experiment with complexities, such as interaction of different constructs, which are difficult to do in a text book or classroom.
The chemistry lab framework, similarly, allows a student to load various chemical constituents used in a lab (acids, bases, salts, indicators, etc) into typical lab equipments like beakers, pippette, etc and mix them as they like. Indicators can be added to check status and transitions of ph values. The system shows elaborate details for each product selected by the learner, and also tracks behaviour of the products loaded in the containers as the products are mixed. These are powerful ways for students to internalise concepts of chemical reaction.
Developing a CLE is a challenging task in general. Unlike simple e-book or video lectures, CLEs require the system to possess a sophisticated domain model. The system need to model the various components of the task (e.g. the various compounds in the chemistry lab scenario), the way these interact with each other so that it can predict the result of such interactions, and the various misconceptions that a learner can have in the domain. Each of these are, in general, open problems. In order to realise practical systems, developers device various engineering approximations and solutions, to make the problem tractable.
The Marathi Tutor [4] mentioned earlier is also a CLE for language learning. Rather than follow a grammar based approach introducing the language through increasingly complex grammatical constructs, the system here adopts a explore-and-learn approach. Various sentences and expressions, and their variants are provided for the learner to analyse and study. This process, can provide adequate feel for the sentence structure and grammatical rules in the target language.
Use of pedagogical agents is an often used technique in high end content like CLEs. These are animated characters, which watches over the learner activities, and provides guidelines/instructions, either voluntarily or on being prompted. Compared to annotations embedded in the text, this is reported to be more effective in communicating such information to the learner. Such an agent can provide hints, ask probing questions to help the learner identify mistakes he has done, and also advise on the course of action to be followed.

Other Aspects

In the previous sections, we saw a number of issues relating to instruction design. Mostly we focussed on the various aspects that make the instruction effective for an e-learning student. There are some aspects that we have not addressed in this paper. Prominent among them is the formulation of a systematic process for design of instruction. Models like ADDIE, ISD, etc [9] have been proposed for this purpose. The overall approach is adapted from traditional design models, and particularly software design process, starting from need analysis to validation/verification. Another important issue is reusability and sharing of content.
Given the high cost of good quality content, reuse of content has been a concern among the e-learning community for long. Reuse is an issue that need to be addressed at multiple levels. At the lowest level is purely syntactic reuse, as in borrowing a picture, a definition, a document, etc as it is and using it elsewhere. In e-learning parlance, this amounts to identifying the relevant software required for the specific content and ensuring it is available for the required content segment. The person borrowing need to figure out any dependencies – direct and indirect – that the borrowed content may have and address them suitably.
The more interesting part of reuse is semantic reuse. When a full concept or topic can be reused as part of a different course or a different audience. This requires more detailed and structured information, such as the intended objectives of the borrowed content, pre-requisite knowledge on the learner, etc. Efforts at gathering such data through meta-data is also a familiar concern. Standards effort like SCORM, etc address these issues. Today, most LMSs and content packages choose to make themselves SCORM compliant, to enable the content to be fitted into other SCORM compliant LMS environments.
Despite the effort in this direction, there are open problems in semantic reusability. SCORM, in its current form, does not address issues in enough depth from this perspective.
Another concern related to reusability is that a lot of the content is proprietary in nature. While the developers may be willing to allow you to use them in your course (for a fee or otherwise), they do not want to hand over the content to you to adapt and plug it in your learning management framework. The solution to this is a service oriented architecture, like web services. In this scheme, all such content, through a suitable wrapper offer a fixed set of services. All licensed clients can send a request for any of these services. This is a powerful way to share resource on the web (not restricted to content) and an emerging paradigm.

Conclusion

Good quality content is, perhaps, the most pressing problem facing successful adaptation of e-learning in non-commercial environments. Ensuring good quality requires attention to a number of different aspects. Aspects such as visual interface and media have received a fair amount of attention. However, pedagogy and related aspects are confined to academic research progrmmes, and yet to reach the practical world. In this paper, we have outlined the major aspects of instruction design, briefly looking at some of the major issues for each of them. The coverage is, by no means, exhaustive given the time and space constraints. The points mentioned here would provide a feel for the kind of issues, and encourage readers to produce better content. There is a fair degree of empirical and intuitive guidelines available, making the field commercially viable today. However, areas such as constructive learning environments, content quality control, adaptive instruction, etc are fertile areas of research. Developments in this area will spurt e-learning into a new realm of effectiveness, and closer to realising its full potential.

References

[1] M David Merril. First principles of instruction. Educational technology, research and development, 50(3), 2002. Available from http://cq-pan.cqu.edu.au/david-jones/Reading/papers/3/first_principles.pdf
[2] RC Clark and RE Mayer. E-learning and the science of instruction. Pfeiffer Press, 2003.
[3] Philip S Tellis, Prem Sreenivasan Narayan, Suresh Dhamapurkar, Sasikumar M, SP Mudur. "Vasistha" - an instruction delivery framework for online learning.
proceedings of the national seminar on e-learning and E-Learning Technologies - "ELELTECH INDIA 2001" - Hyderabad, India.
[4] Archana Rane and M Sasikumar. A constructive learning environment for language learning. Technical Report, CDAC Mumbai, 2006.
[5] Open document format reference. http://www.oasis-open.org/committees/download.php/12572/OpenDocument-v1.0-os.pdf
[6] C. Reigeluth (ed.). Instructional Design Theories and Models. Hillsdale, NJ: Erlbaum Associates, 1983.
[7] MD Merril. Instructional Design Theory. Englewood Cliffs, 1994.
[8] MD Merril. Instructional Transaction Theory (ITT): instructional design based on knowledge objects. in [11].
[9] ADDIE model. http://ed.isu.edu/addie/
[10] The dynamics of theory and practice in instructional systems design. De Villiers, M.R. Unpublished Ph.D. Thesis, 2002. University of Pretoria. http://upetd.up.ac.za/thesis/available/etd-02212003-180121/
[11] CM Reigeluth (Ed). Instructional-design theories and models: a new paradigm of instructional theory. Lawrence Erlbaum Associates. 1999.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License