Educational Technology for Developing Countries
ALAN M. LESGOLD
For developing countries, as for all others, technology, especially information technology, is a tempting way to raise productivity. Just as new technologies have made it possible to produce better products at lower cost with less human labor, so it is often believed that technology can improve the learning process and perhaps even substitute for teachers who are not available or not affordable. Because they often have more educational chores to accomplish and less money to invest, the developing countries especially could benefit from the technological leverage of learning. But because educational systems are extremely stable and resistant to change, it is important to establish clearly whether a given technological contribution will sufficiently enhance educational productivity before undertaking any major effort to use it.
In this information age, the process of learning to use information tools may have value in its own right, beyond the value of those tools for promoting other learning. Technology can be a diversion, though, even a barrier to education, if it takes too much of a teacher's time to set up, or if it distracts teachers from productive tutorial interactions with students. This paper provides several different viewpoints on potential technologies for education and training in developing countries.
MATCHING EDUCATIONAL TECHNOLOGY AND CONTENT TO SPECIFIC ECONOMIC GOALS
Different situations and goals require different instructional and technological approaches. One key instructional requirement is to make optimal use of tools and curricula from outside while simultaneously respecting the need to build new knowledge on the existing local culture and knowledge. Educational efforts will fail unless they are attuned to the indigenous culture and its extant knowledge. For example, Patel¹ has shown that simple health-related measures, such as dietary changes to combat nutritional disorders and use of condoms to prevent AIDS and control fertility, are difficult to teach effectively without adapting to existing culturally-based knowledge.
The willingness of developing country governments to invest financially in educational technologies will necessarily depend on the scale of the educational goals being considered and their potential for economic return. It might be rational to spend thousands of dollars per trainee to produce a leadership corps for a high-yield industry that a country finds strategically central. But a developing country cannot afford high-tech solutions for universal basic literacy education unless that technology substantially improves the efficiency and yield of the educational process.
RECENT TRENDS IN ESTABLISHED TECHNOLOGY
In many respects, the problems surrounding the use of educational technology in developing country schools are similar to the problems surrounding its use in American schools. The U.S. educational system is remarkably stable and resistant to technology-driven change, just as the cultures of developing countries are stable and resistant to change. For example, schools, with their limited budgets and complex procurement systems, find it difficult to handle the materials, labor, and consulting costs related to computer technology maintenance, and teachers have limited time to invest in learning new instructional methods. Furthermore, in each case the extra dedication required of teachers to surmount logistic difficulties will be forthcoming only if the positive effects of the technology are clear and substantial. This leads to three basic principles to follow in adopting education tools.
The first principle is that low-maintenance commodities work best.² The American school scene is replete with horror stories of computers sitting idle in closets because the batteries for their clock chips (and parameter memory) died or because there was no money to repair a hard disk that was dropped. Developing countries face similar problems. Thus the best technologies will be those able to function with little or no maintenance (this is not an absolute rule but rather a guideline). Although a countrywide center for training software designers³ or an urban training center for health paraprofessionals might profit from advanced computer technology, a village removed from urban centers might do best with “disposable” wireless phones or radios combined with desktop-published text materials shipped from a regional production point.
Word processors, spreadsheets, symbolic mathematical manipulation tools, and information network clients are among the software tools that are available as commodities. American school systems often decide to purchase special-purpose versions of tools even though generic products with similar or greater capability are available widely in commodity form-the commodity version may have appeared too complex, or the special version may have offered some feature that seemed useful. Commodities, however, are cheaper by orders of magnitude and much easier to maintain. Moreover, an economically powerful customer base supports the maintenance of commodity software, by both purchasing technical support and exercising purchase choices, whereas schools rarely have enough money to pay for good technical support. Some universities, however, have been quite successful in negotiating extremely low-cost package software deals for their students, such as complete office packages (word processor, spreadsheet, database system, and so forth) for prices below $100. 4
The second principle is that easy-to-use tools work best. When a piece of technology is difficult to use, and especially when it is difficult to start using, it has a much lower chance of penetrating a culture to which it is foreign. The slow-changing cultural lore needs time to recognize that it will be worthwhile to master a new technology.
Even simple technologies can be hard to use when their usage is poorly staged. For example, often in U.S. classrooms a teacher wishing to show a video must first locate the massive, unwieldy cart containing the video display monitor and then roll the cart through crowded hallways to his or her classroom. No wonder video is not used as much as it could be! The specific barriers to easy deployment of technology may differ in developing countries, but the same principle-that difficult-to-use devices will not get used-certainly applies universally. Moreover, the threshold for ease of use is quite high. Devices that are not readily accessible to all who approach them will not work for mass education.
The third principle is that tools that foster activities that promote learning work best. This rule may seem obvious, but it is not. Too often tools that seem “educational” do not lead to the learning of anything important. Indeed, a standard marketing strategy in the United States is to claim that a particular piece of software, such as a computer game, is “educational” - perhaps it teaches higher-order thinking skills. While it is productive to use such commodity tools as spreadsheets, word processors, and information server clients (for example, Mosaic) to support classroom activities, it is generally not as effective to invest student and teacher time in learning to use a program when that program supports learning only incidentally. Many computer games fall into this category, as do other systems that take a long time to learn to use relative to the amount of learning they can support.
TWO EXAMPLES OF EDUCATIONAL TECHNOLOGY
How then do these principles apply to specific technologies? Two interesting examples of technologies are described here. Interactive radio, a low-cost scheme that has been used increasingly since 1974, is included as an example of educational technology deployment because of its substantial success, at least for the limited purposes for which it has been used. The second example, which is at the other extreme of technology, is intelligent computer conversations with students. Using new and powerful intelligent systems technology, this option has considerable potential as an approach that is affordable and distributable in developing countries.
Interactive Radio Instruction
Interactive radio is an idea that most educational technologists would not favor because it is not really interactive. 5 Lessons are spoken over the radio, and students respond to prompts from the speaker, nothing the student says is preserved or transmitted to the speaker. Why would any educational technologist want to develop such a system? First, data show that interactive radio is remarkably cost-effective for schooling. Second, radios are ubiquitous and exist in a highly sustainable form. They also are cheap and are easy to maintain; require only batteries or central electrical power; and are easy to operate. A basic radio has only three control functions, usually lodged in two knobs: system activation (on/off), frequency tuning, and volume level. If standard broadcast bands are used, tuning is extremely straightforward. And, most important, many world cultures support the use of a radio. Thus students, teachers, and other learning mediators (for example, parents) already know how to use them.
In most cultures, a considerable amount of learning activity involves verbal interactions. Such verbal interactions are costly, however, since they depend on a teacher, who can deal only with a small number of students at a time. Because most developing countries have too few qualified teachers, they especially would benefit from technological approaches that replace at least part of what a teacher does. For example, it is simple to replace the set of learning conversations in which a teacher initiates each section of a conversation and in which the conversation can continue when the teacher keeps uttering statements that are not determined by what the student has said. Teachers can provide examples, espouse principles, and even pace the progress of students through linear problem solving without necessarily having to react to the specifics of what students say in reply. For example, a teacher can ask a question, pause for an answer, and then explain why a particular answer is good. In addition, when a group of students listens to radio-delivered “interactive” instruction, each student also learns from the responses he hears his friends calling out alongside him. 6
Tilson has reported on a number of studies of the effectiveness of interactive radio in elementary education. 7 One study conducted in Bolivia around 1989 compared students using interactive radio as part of instruction in mathematics to others using only standard textbook instruction. Some of the former also had used interactive radio the previous year for second-grade mathematics. The differences in post-test performance between the two groups were substantial (Figure 1): effect sizes of 1.29 and 2.03 standard deviations for one and two years of radio-delivered instruction, respectively.
In a range of similar studies (but for the second grade) conducted in Bolivia, Nicaragua, and two areas of Thailand, effect sizes of 0.24 to 0.94 standard deviations were obtained for interactive radio. Other ages and other topics also have been investigated. Health education studies have shown that the radio approach is particularly effective for teaching upper primary students the details of oral rehydration therapy delivery and other aspects of gastrointestinal infection in infants.
The interactive radio approach is relatively inexpensive to start up and to sustain. Costs estimates range from a high of $6.40 (the cost per student in 1990 for serving 25,000 students in Bolivia) to a low of about $1 per student to sustain larger efforts once initial capitalization has been completed. These costs are not much out of line with the most cost-effective approach to basic primary education-textbooks.
Interactive radio is not effective for all instructional needs. For more difficult instructional and training needs, interactive radio is not interactive enough or individualized enough. It works exactly where information and prompting can be provided unambiguously and with minimal regard for moment-by-moment changes in students' understanding. Its social effects, however, should not be dismissed (students are part of a learning group that listens to radio together), and it provides effective delivery of factual knowledge, as well as a social comparison base for students who can watch each other's responses.
Because radio also depends on having little or no need for pictorial or diagrammatic information, it would not be the best way to teach physics students how to draw force diagrams or to teach technicians how to operate an intravenous pump in the critical care unit of a hospital. Thus the advantages of interactive radio are very limited for all the reasons that support the use of multiple media in education. Even so, the idea of distant, audio instruction could be refined considerably in the light of recent progress in speech understanding by computers.
Intelligent Coaching of Oral Reading
A few years ago, Jack Mostow 8 at Carnegie Mellon University proposed to build a computer-based coach for elementary school oral reading. Students learn to read from a combination of being read to, being coached on specific symbol-sound correspondences, and especially reading aloud to a teacher or parent. Mostow reasoned that even though the general problem of how to configure a computer to understand speech is not wholly solved, the problem of understanding a child who is reading a known text is far less complex. Basically, it involves verifying that consecutive words of a fixed script are in fact uttered, which is much easier than deciding which word a person has uttered when there are no constraints on the possibilities.
At present, Mostow has completed demonstration versions of an oral reading coach and is building a production version. The system can recognize mistakes and can prompt students when they get stuck. This tool, which does a lot of what a good teacher would do, is needed both in the United States and in developing countries. Equally important, the basic idea that speech recognition is tractable whenever the set of possible utterances is highly constrained can be taken much further. Students can be offered choices to which they can respond orally, thus allowing a real interaction to take place in the audio realm.
The approach is limited to cases in which there is a describable set of alternatives from which the student can choose. The simplest possibility is multiple-choice responses. While this is not the instructional form of greatest interest to educators, it has its uses. Simple choices that can be offered include self-assessments by the student. For example: “Do you understand why you must boil the water you use for oral rehydration therapy?” “Can you see why it is important to give a child with bad diarrhea liquids right away, even if he has a problem that your friends usually treat with dry foods?” 9 More complex choices might be possible if the student is able to handle occasional follow-up questions from the computer (students can handle partially deaf teachers, so they probably will adapt to computers that are not perfect in what they “hear”).
Multiple computer-generated voices, different sounding voices for different roles, are another possibility. This would allow coaching or other feedback to come from multiple sources. For example, a training program in public health could include coached feedback from a scientist coach, a nurse coach, and a politician coach. Thus in an exercise dealing with an epidemic, a proposed quarantine might produce praise from the nurse coach but expressions of horror from the political coach.
While the value of intelligent conversation as a source of learning is clear, it will not be easy to move powerful computers to rural sites in developing countries, nor is it likely that local villagers could maintain these machines. The solution: the student and the computer, which would be located elsewhere, can talk by telephone (see Note 9). Soon it also will be possible to use relatively low-cost wireless communicators that have low-bandwidth video capability.
What determines the potential suitability of a technology for instructional use in a developing country is not the technology considered in vacuo, but rather the technology properly adapted to this purpose. Many technologies that appear too exotic for the developing world can, in fact, be adapted if the criteria of ease of use, cost-benefit ratio, and sustainability are considered carefully in the design of training or learning systems. And, equally true, many technologies that seem ideal can work out poorly if these considerations are not given careful attention.
WHAT MAKES LEARNING HAPPEN?
In the past, learning was seen primarily as data transmission, with perhaps a second component of practice. Recently, however, instructional researchers have become more aware of the constructive character of some learning. Especially when the material being taught is not part of the student's cultural background, learning must be a more active process-one in which the student constructs new knowledge in terms of prior experience. Effective tools for this kind of knowledge construction process either facilitate the student's efforts after understanding or force the student to confront gaps in his existing knowledge.
Each of the ways to foster learning (an incomplete list is shown in Table 1) has its own strengths and weaknesses. For example, learning by being told works only if the teacher and student agree on the meaning of the language that the teacher is using. Even then, this approach encourages confusion between what was said, the desired learning outcomes, and what the student knew earlier and learned later. No one learning approach is best for all circumstances. Thus it is best to develop a clear set of learning goals before selecting a training approach.
Of the two basic types of learning model described in the rest of this section, one type captures the highly circumscribed approach needed when the goal is to form a highly reliable (but perhaps somewhat redundant) cadre of workers able to handle situations routine enough that modest natural selection among workers is sufficient as an approach to the need for qualified practitioners. The second model captures at least the veneer of the “smart worker” concept.
TABLE 1 Forms of Learning
The Beehive Model
Kevin Kelly, 10 citing the work in robotics by Rodney Brooks, 11 has discussed the factors that make biological systems robust. These include parallel, redundant, and highly accurate subsystem modules. These modules are not completely programmed and are somewhat idiosyncratic, with the most adaptive being selected for a given situation. For example, a beehive feeds itself through this combination of reliability, variability, and natural selection. Bees forage somewhat randomly, but whenever a bee finds food, it returns reliably to the hive and does a highly programmed dance to which the other bees respond in a highly programmed way. In this manner, the hive optimizes its search for food without having to understand how it finds the food.
How would one build work teams that have this property? First, a subset of the skills needed for work would be highly drilled and practiced. Second, when novel situations arise, different team members would try different schemes to deal with the novelty, and a highly disciplined decision process would determine which scheme is to be supported. This approach has been partially adapted by some traditional workplaces to handle, for example, vigilance and safety requirements. If any member of a team notices a safety hazard, the entire team reverts to drilled procedures to preserve life and property. Every organization has some beehive-like functions. As a general approach to work force competence, however, the beehive model has serious shortcomings. Without a deep understanding of work situations, teams may not come anywhere close to having this redundant, reliable, but limited capability.
For those situations in which the beehive approach is feasible, certain forms of training are appropriate. A limited set of skills must be highly practiced, and motivational components must be included in the training regimen because of this overlearning requirement. Traditional computer-based drill may be sufficient in a few cases, if it can be tied closely to the situations in which the skills will be used. Richer, more visual, and more concrete simulation environments may be needed. Such approaches as spoken computer responses to trainee actions may be more motivating, and it is likely to be easier for trainees to work in teams, either at a single learning station or over a network.
Even in the area of traditional drill and practice, then, technology might augment performance by providing more realistic practice situations and by affording more motivating real and simulated interpersonal interactions.
The Smart Worker Model
Training and learning must be broadened for the work environment of the future. While the specific requirements for developing countries will be partly a function of the form of each country's participation in the world economy, it is possible that a smaller country will be best served by aggressively pursuing a few niche markets. Competitive advantage will stem from a combination of standard quality-enhancing practices and deep expertise in the niche domains. Each worker will need to understand the goals and functional characteristics of his or her enterprise and will need to adapt continually to different situations. To prepare a cadre of workers for high-productivity penetration of a niche market, one must offer a combination of skill development, practice with complexity, and adaptive problem-solving capabilities.
Based on my own training development experience, the combination of coached apprenticeship and post-performance reflection is especially productive. Apprenticeships allow basic skills to be exercised in realistically complex situations, while reflection on one's performance is the key to generalization and adaptation. Technology can play a role in both apprenticeship and reflection. For example, a system we built called Sherlock provides a simulation of a work environment (specifically, diagnosis of faults in a complex electronic switching system), an intelligent coach that provides advice when the trainee reaches an impasse, and a collection of reflection tools that allows trainees to review their own performances after solving a problem and compare their performances with that of an expert. Trainees also are given access to descriptions and explanations of system function and diagnostic strategy. Like other aspects of computational technology, the cost of building such training systems is dropping rapidly as the basic techniques are perfected, and especially as object-oriented design tools and object databases become more commonplace. 12
The combination of apprenticeship using real or simulated work places with reflection on performance is essential if workers are to be able to deal with emergent situations. Developing countries, like those further along, need smart workers who can reason beyond what they explicitly have been taught. The “general” capabilities that can be taught, however, are not purely abstract; rather, they are grounded in one's experience and in real job situations that span the likely range of situations encountered in real work. It is difficult, perhaps impossible, to generalize from specific experiences while immersed in them. Yet it is not likely that abstract “book learning” will be available in usable form when concrete work situations demand that a worker stretch his knowledge beyond that which has been fully and explicitly mastered. Reflection, comparison of one's work to that of others, and discussion with colleagues are key means of acquiring stretchable knowledge. Technology for learning that supports these approaches will be particularly useful.
SPECIAL NEEDS FOR WORKER AND INSTRUCTOR TRAINING IN DEVELOPING COUNTRIES
Developing countries, like all countries, need to provide their citizens with a high-quality education. But this is a major economic burden to every leading economic power, and it is especially daunting for newer economies.
In addition to covering the traditional subject matter, schools should coordinate their academic goals for students with the skills and education needed by a modern labor force. Even though some advanced countries are studying the cognitive skill requirements for modern productive work, there remains a cultural legacy, too often copied in developing countries, of detaching schooling from work. While rote training for today's jobs, which will not last a lifetime, is foolish, so is abstraction of schooling to the point where it has no ties to past or future life experience. Alfred North Whitehead noted long ago that knowledge acquired in the sterile academy is often “inert,” not coming to mind when relevant to real world situations. This never was a good thing, but it is particularly problematic in an era of rapid change in the nature of work. Because of the rate of change, the culturally shared knowledge about the workplace that has compensated for the removal of school from work life is decreasing. At the same time, a large cadre of people-teachers-have been successful in life by doing just what school people told them to do. It is essential, then, that teachers better understand the workplace and the ways in which schooling can facilitate learning to do specific work. Technology can help with this by bringing information about and simulations of modern work into the school.
Technical Training before Entering the Work Force
Today, productive work usually requires job-specific training beyond general schooling. In some countries, this takes the form of formal apprenticeship programs, usually combining on-thejob practice with classroom time. In others, apprenticeships are less formal or even nonexistent, but prospective workers still must have specific training before they are likely to find jobs. Too often, programs focusing on specific job performance have provided very specific practice, often on outmoded equipment, combined with totally abstracted schooling. For example, a prospective machinist might receive courses in trigonometry and mechanical drawing combined with exercises using a manual lathe, turning relatively cheap and workable metals.
The problem with such schemes is that in periods of rapid change, intuitions of instructors about real work situations do not change fast enough for them to revise the courses and practice opportunities to support the new jobs. A modern machinist, for example, works by programming a computer-controlled machining cell, or perhaps she receives the program from a programmer and checks to see if it is feasible in its sequencing of cuts and consistent with the target product. Furthermore, depending on the value of the metal and the cost of labor, critical concerns about the reliability and duration of the planned sequence of cuts will vary. Some machinists produce dozens of identical items from cheap metal, while others may spend days planning and executing the milling of a single piece of stock costing $250,000. Do math courses and practice making candlesticks prepare one adequately for such work? While no program is quite that antiquated, few current programs reflect sufficient thinking about the connection between training and subsequent work.
It is costly, however, to completely retrain a worker every time a new technology appears. Conceptual support for broader job families would certainly be worthwhile if it could be achieved. Jobs keep changing, it is argued, and thus students should receive broader preparation, but few programs are based on any systematic analysis of the range of likely jobs or of the key abstractions that can facilitate transfer from one job to the next. Recently, I suggested that the same ideas used to develop abstractions of computer software modules and to organize such modules for reuse also be applied to the systematic specification of knowledge requirements for training. 13 Specifically, just as separable pieces of a computer program, called objects, can be organized into abstraction hierarchies (usually called inheritance hierarchies), so knowledge objects can be specified and similarly organized. In such inheritance hierarchies, a more specific version of an object inherits many of its aspects from a more general “parent” object but also contains the unique knowledge needed to adapt to a special situation for which it has been developed.
Such hierarchies of knowledge specifications provide both computer and human tutors with a formal basis for coaching that is focused both on situation-specific competence and on broader understanding. The logic of this kind of coaching is to (1) use past experiences to establish a basis for generalization, (2) then introduce the generalization, and (3) specify what is different about the current situation that requires a specialization of the general approach.
Training can be made much more efficient and effective by using the object-oriented analysis and design approaches that have enhanced software productivity. After all, building a base of reusable knowledge objects to support a domain of software application is really no different than building a clear specification of the Job-specific and more domain-general knowledge requirements of workers who are to be trained.
Training for New Tasks
A learning-by-doing approach also has many advantages for job-specific training. 14 This approach ensures that the knowledge needed is acquired in the context for which it will be needed. Through the use of computer-based simulations, it becomes possible to provide learning-by-doing training even in many areas where it might be impractical or dangerous to allow workers to do the real job before the proper training. Through the use of intelligent coaching, trainees can engage in rich work with all its complexity since the computer coach ensures that the trainee will not reach an insurmountable impasse; rather it converts impasses into learning opportunities. Through the use of post-task reflection, during which the trainee receives critiques of his performance and learns more about how an expert would have attacked the task, computer-based training systems also lay a groundwork for the trainee's subsequent transfer to new jobs.
Improvements in object-oriented analysis and design technology are key to building practical learning-by-doing systems, as well as the availability of relatively powerful desktop computers. 15 An infrastructure for such schooling can be adapted from prior efforts to introduce learning-by-doing and information management tools for training a vanguard work force for two or three niche markets. India, for example, will soon be able to build on the base of one of its niches-an applications software industry.
A Technologically Literate and Innovation-Receptive Population
Anyone seeking to exercise democratic citizen rights in complex societal decisions, to understand economic policy, and to participate in high-quality manufacturing, service, or information work, will require some new knowledge. Whether considering a public issue, such as to ban or not to ban the use of certain hydrofluorocarbons, or doing one's part in maintaining efficient, quality manufacture of consumer goods, people need increasingly to have tools for understanding, examining, and criticizing complex systems. Although developing countries can begin to improve their economic standing by training a small elite work force for a few target niche industries, further growth also will require a populace that understands the modem world and is prepared for modern productive work. But both the complexity and the interconnectivity of systems will pose learning requirements for schools. Students need extended practice in understanding and manipulating systems, talking about them to one another, and envisioning their function from multiple viewpoints, as well as the tools for managing information complexity. Partly because they may lack a universal analytic verbal tradition, the cultures of many developing countries may be better able to assimilate broad systems-based understanding of the natural and technical world than the cultures of the more developed countries. Still, tools for thinking-the kinds of outline tools, electronic lab notebooks, and simulation packages beginning to penetrate schooling in the richer countries-will eventually be important for developing countries. This will require not sweeping calls for two computers in every hut but rather educational goals aimed at competence with complex systems as the needed tools become maintainable, cheap, and easy to use.
REQUIREMENTS FOR SCHOOLING IN DEVELOPING COUNTRIES
Adapting to Local Language and Conceptual Structure
New systems of knowledge are not acquired independently; rather, they are layered on top of one's prior knowledge. For example, Patel found that even after years of immersion in U.S. cultures, women from South India still maintained a knowledge of nutrition that began from the folk knowledge of the regions where they were born. 16 Without understanding the systems of knowledge those women started with, researchers found it difficult, perhaps impossible, to determine what they understood and knew after exposure to European-American nutritional ideas. Similar situations arise when people from North America go to western Asia. For most Americans, the gastric acidity triggered by spicy food is an illness, to be fought with antacids if it becomes too noticeable. Such an attitude overlooks the role played by food in triggering bodily protections against bacteria.
This then is another requirement for efficient and effective instructional design: the explicit building of connections between what people in developing countries already know and certain bodies of knowledge they need to acquire to be productive. The new knowledge must be anchored in what is already understood. This will be the case especially when issues of health are involved, but it will arise in many other situations as well. In commerce, for example, different cultures view the relationships among workers, owners, customers, and managers differently. In more extreme cases, cultural differences in time perception, for example, become important.
One partial antidote to the cross-cultural communications problem is concreteness. Because verbal telling fails when the terms in the message are not understood, demonstration in a real or simulated world often can overcome such misunderstanding. A major consequence of the exponentially expanding bandwidth of computation and communication systems is the ability to replace symbols with enactments and to replace words with pictures, movies, and manipulable displays.
Developing a Teacher/Instructor Training Infrastructure
The developed countries may have a shortage of teachers who understand the modern workplace, but in developing countries teachers who have such an understanding may be nonexistent. Such a situation calls for development of a cadre of vanguard master teachers with a better understanding of the connections between what they teach and the kinds of jobs their students will have in the future. Here again there is potential for using workplace simulations that are provided and coached by a computer system, this time for use by both teachers and students.
A grass-roots, self-help capability among teachers could be fostered by improved communications capabilities. At the lowest level, telephone interactions can be useful in bootstrapping an improved teacher corps. Modest electronic bulletin board and electronic mail capabilities also can be very powerful, although there will be cultural differences in the extent to which a teacher feels it appropriate to pass on his or her ideas to others or to ask others for assistance. Modest levels of access to information servers containing workplace simulations would be a valuable addition and will become feasible as network technology becomes ubiquitous.
Schools needs benchmark quality standards for the education process and for ways of measuring progress in attaining those standards. For developing countries, the only educational outcomes that really matter are those that move people significantly closer to being more productive. This requires continual assessment against clear standards.
Some aspects of the American approach to testing are not likely to help developing countries; for example, U.S. assessment schemes are grounded in a fundamental distrust of the student. In most total quality systems, measurement and benchmarking are possible largely because employees measure products themselves. If a company with a strong total quality management (TQM) program were to announce that it was hiring a major testing company to do all benchmarking rather than trusting its own employees, the effort would likely be a disaster. Total quality management works in large part because employees are trusted to make good assessments. Indeed, many aspects of quality control require employee input. Surely this must be true for quality control in schools as well.
In most U.S. schools, testing is adversarial, with an implicit assumption that the student's self-assessment is not to be trusted. This is unfortunate because often the student is in the best position to judge her own understanding and competence. To the extent that instruction is delivered by computer systems, those systems can be simpler and more effective if student self-assessment is part of the design. Students will need help in assessing their progress, however, because improvements from day to day in such skills as writing, painting, and other communications competencies are not obvious. (My own experience with trying to learn oriental painting supports this conclusion. I spent several years at it, with no sense at all of any progress. Only when I discovered a collection of old paintings my wife had saved over the years could I see that I had been improving slowly over time.) A major role of assessment in schooling is exactly what it is for industry, to help the student notice changes in the rate and quality of production. But that is only possible if the educational system and the student trust each other.
More broadly, students need to appreciate that their many investments in learning over a period of years, even decades, are headed toward valued goals. Indeed, students need to be able to see the consequences of their learning progress over the long term. 17 How else can a student in fifth grade decide whether she is working hard enough to be on track for an eventual job as an engineer? Social comparison and culturally embedded work norms are fine for highly homogeneous cultures in times when jobs and the training they require are stable. Too often today, however, schools serve a mix of students from multiple backgrounds-black versus white, urban versus rural, Ibo versus Yoruba-who do not trust each other's norms. Consequently, self-assessment tools that help the student to chart progress toward goals he values are needed.
Focusing on Areas of Greatest Need
Developing countries cannot afford to follow the lead of the United States on such issues as the extent of university resources to be put in place Even in the United States, far too many students are receiving a low-quality “liberal” education that does not prepare them for work later, but this is being remedied somewhat by a significantly increased investment in postsecondary technical education.
Developing countries face hard choices in deciding which parts of an educational system to build first. It seems sensible, though, to develop a technical education and worker force preparation capability first, deferring the development of elite professional institutions until later. While the latter are important to the maintenance of a country's identity, they are also the most expensive to build, the hardest to maintain at high quality, and the easiest for which to substitute billets at foreign universities.
EDUCATIONAL TECHNOLOGIES WITH POTENTIAL FOR DEVELOPING COUNTRIES
Some technologies that offer a mixture of new ways to convey information and to support thinking have particular potential for education in developing countries. Furthermore, these technologies soon will be available as affordable commodities.
Combinations of Desktop Publishing and Printing Technologies
The tools for desktop publishing are improving rapidly. Word processing systems are increasingly able to produce hypertexts and hypermedia “shows,” and virtually any enterprise can carry out the entire publication process using a combination of powerful word processing software, desktop publishing and presentation software, and new hardware/software combinations for printing. In a high-technology economy, print would be largely eliminated, replaced by network connections and access to such software as Mosaic. 18 Combinations of printed materials and minimal additional technology will continue, however, to be a primary means of conveying information to rural areas of developing countries. New, low-cost, computer-based printing systems, such as that produced by Riso in Japan, offer a direct interface between standard Microsoft Windows print drivers and the production of printing masters. Thus a user can give the same print command that might lead to laser printer output and instead receive a master for a low-cost, ink-based printing system. The total system is easy to use, maintainable, and as economical as a commodity business desktop computer. On the printing side, for intermediate quantities (a few thousand) costs are much lower than for xerographic reproduction. The information revolution is improving the economics of traditional paper media production.
Commodity distribution vehicles that can convey voice, motion picture, animation, or other information are a significant breakthrough. One can argue over how long the current CD-ROM will remain standard, but the costs for this technology are now low enough that many exciting educational possibilities exist. Moreover, compared to magnetic recordings, CD-ROM disks are more robust and can survive a much wider range of environmental exposure.
One particularly important side effect of the CD-ROM revolution and related technologies is the rise of a low-cost video production capability. Whereas video editing and production equipment was once so expensive that it rented for $100$300 an hour, one can now build systems sufficient to produce decent video and deliver it on CD-ROM technology for $20,000 or less. With CD-ROM drives selling for as low as $250, the range of situations for which video is affordable is growing rapidly. Even if only a few video segments are needed to make an instructional system more effective, a central production facility for a country can afford to produce such products.
While the production of single CD-ROM products is quite inexpensive, mass production is somewhat more expensive. A likely scenario is that a developing country would buy some of its training software from elsewhere-perhaps from a country with a niche market in educational media production-and then adapt the software for local use, fitting it to the local culture and the experience of students in that country. The country's central educational production facility would then produce a single CD master that would be sent to an industrial producer for mass production.
Distributed Information Network Schemes
The worldwide information server network that has developed over an extremely short period allows students almost anywhere in the world to access information from almost anywhere else. Although it is not yet clear how long free access will continue, given a bit of luck straight access for educational purposes may continue to be low in cost or free, with enurepreneurial efforts focused on the problem of searching for and organizing information rather than controlling information pipelines. Even if much of the Internet becomes costly, however, it will remain feasible for developing countries to build local capabilities and to share educational content text, images, video, sound, and so forth via network commodity tools such as those initially distributed by the University of Illinois.
A substantial range of instructional possibilities is available using Mosaic because it supports a variety of still and moving image forms, sound, and a limited interactive capability, using text forms. Further flexibility arises from its ability to invoke a program as the response to a request for information-that is, a form submission or a hypertext link can point to a program on an information server that will be executed before any information is returned to the person making the request. This means, for example, that the World Wide Web capability is readily enhanced to include intelligent interactions between the information server and a student requesting information. In my own work building intelligent coached apprenticeship systems, I have found that most of what my system does could be delivered over the Internet since the student response in my system, while it always seems to be rich enough to mimic real work, is actually limited to selections made by pointing to menus and images. Indeed, we are currently pursuing several projects to convert tutorial systems to a client-server model that is based on the World Wide Web of information servers.
A variety of instructional technologies have been developed to take advantage of the basic network delivery schemes and hypermedia representation schemes that have appeared. At present, researchers in companies and universities are designing “shells” for the authoring of “learning-by-doing” systems, and another cadre of technologists is working on improved distributed information network technology. Simple, locally tailorable technology for distributed use and support of learning-by-doing technology will likely be available within the next few years. Combined with the rapid growth of wireless telephony technology, this will enable the establishment of powerful educational networks in developing countries. The vanguard of educational networks will likely be the laptop computer with a CD-ROM drive and wireless modem, but combinations of satellite and fiber-optic transmission probably will soon follow.
Affordable Basic Artificial Intelligence Tools
Rule-based programming already is playing a role in training design. For example, a system developed by National Aeronautics and Space Administration (NASA) - CLIPS - helps the user to conduct a job analysis. The analysis can then be formalized as a series of “productions,” or conditional actions. This set of IF-THEN rules constitutes a simple form of “cognitive” simulation of the work that was analyzed. A wider range of such tools is likely to be available in the future, including some that do a better job of representing work-for example, explicitly representing both conceptual and procedural knowledge.
In the area of speech recognition, machines soon will be able to routinely analyze utterances and, if necessary, translate them into different languages. This capability takes advantage of the constraints placed on language by the specific situations in which it is used, avoiding the general problem of speech understanding, which remains a basic research issue. Nevertheless, instructional designers now need to consider which training or instruction situations need which modalities of interaction. Listening and speaking, while not likely to become the only form of student interaction with training systems, will certainly play a major role in learning environments.
Object-oriented software engineering technology will play a major role in any cost-effective development effort. While such software is clearly needed to develop systems simulations, the new technology will play a big role as well in the design of systems to promote learning. Not only the systems that one is teaching about, but also the very conversations that lead to learning, can best be represented using systems of computational objects. The disciplined use of object-oriented methodology can improve the quality of computer-based simulations and coaches while simultaneously yielding substantial improvements in software development efficiency.
DEVELOPMENT OF TECHNOLOGY FOR EDUCATION AND TRAINING
Improvements in how software can be developed more efficiently are surely needed when even the largest software house can encounter situations in which a major product is delayed more than a year beyond the expected time of initial release, as often happens today. Thus it is recommended that those undertaking the design of educational software use basic object-oriented techniques for effective tool development and deployment. Furthermore, a modular decomposition strategy will work best-namely, to (1) decompose the task into a number of manageably small pieces; (2) use off-the-shelf capability for as many pieces as possible; and (3) replace any piece-production activity that falls behind with one that is more reliable. Object-oriented design strategies should be used to drive the decomposition process.
The overriding strategy in any such effort should include at least three elements. First, avoid expensive technology except when it is really needed-paper and one-way television can handle many situations. Second, commoditize the tools needed for better learning and make sure that they work on minimal, affordable computer systems. And, third, use coached apprenticeship strategies for educating key workers. Multimedia can be used when direct experience is particularly important.
To end on a practical note, it is always useful to look for ways to combine work and the educational use of computers. Many enterprises use powerful computer systems, but few of these enterprises operate 24 hours a day. Either as part of work time or as an after-hours approach to education, companies could use the same machines that support work during the day for training after hours or even during the course of work. For example, a plant owned by ABB in Helsinki was setting up a new production line to make motor drives. Since virtually each drive was a custom job, computers were placed at each worker's station on the production line to convey parts information and also details on the device assembly. The computers also were equipped to provide additional training. A worker who had questions about how to build a particular motor drive could aim his bar code reader at some special codes on his bench, prompting the computer system to branch to a training package (built in cooperation with the Technical University of Helsinki) that delivered truly “just-in-time” training.
U.S.-based Federal Express uses the computer systems of their office workers to provide training to other employees after hours. While the scheme currently aims only at FedEx workers, a broader possibility, in which the computers at several large businesses in an urban center are used by a virtual junior college at night, seems quite feasible.
The bottom line, then, is that the economic forces that have made information technology cheap and ubiquitous in commerce can and should be put to work for education and training. The actual machines and networks of the commercial sector could be used during the off hours, and the commoditizing effect of industry-driven standardization could be used to cut the costs of learning technology. Finally, the multimedia possibilities of the “edu-tainment” world could be used productively to promote more learning-by-doing throughout developing country educational systems.
1. See L. Percival and V. Patel, “Sexual Beliefs and Practices by Women in Urban Zimbabwe: Implications for Health Education,” McGill Journal of Education 28 (1993); and Sivaramakrishnan and V. Patel, “Relationship between Childhood Diseases and Food Avoidances in Rural South India,” Ecology of Food and Nutrition (1993): 31.
[Украинский] [английский] [русский]