Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

Moving Computing and Education Beyond Rhetoric


by Robert P. Taylor & Nancy Cunniff - 1988

Computers enable us to offer students distinctively alternative paths to certain goals, for instance, graphic representation in the place of verbal statement. Where such alternatives can be implemented, it becomes possible to test their comparative effectiveness with some rigor.

Computers enable us to offer students distinctively alternative paths to certain goals, for instance, graphic representation in the place of verbal statement. Where such alternatives can be implemented, it becomes possible to test their comparative effectiveness with some rigor.


This article assumes that the supply of rhetoric about how computing can help or hurt education is more than sufficient for all conceivable purposes’ and that it is high time to move beyond that rhetoric in an effort to determine exactly how computing affects learning. Where computing is proven useful, it should be more thoughtfully and immediately applied; where proven useless, its application should be discouraged. Carefully designed and narrowly focused research is what is now needed. We argue that computing provides an effective and powerful way of providing alternatives for learners. It describes specific research and suggests how that research fits into the larger picture of what is needed to move beyond rhetoric.


Computing can help learning; however, how it does so is not well understood. Critical perspectives such as those voiced in an earlier issue of this journal1 have obscured rather than clarified the issues by making their judgments on the basis of limited understanding. A realistic assessment of the role of computing in education can come only with a deep understanding of how to match the potential educational power of the computer with the needs of learners. It is the job of serious educators to define the questions to be asked and to conduct the research necessary to develop that understanding.


Since we believe appropriate research can identify how and to what degree computing can assist learning, we have undertaken a program of research along those lines. We summarize part of our work here to illustrate how research can further the understanding of the interaction of computing and learning. Our work involves a narrow area of education and a limited research focus, but the broader issue is an identification of how and for whom computing can be uniquely helpful. Our specific focus investigates a comparison of graphic versus textual representation of concepts. We hypothesized and verified that graphic representation of some concepts is superior to textual representation for at least some learners. Since the computer is the principal means for making graphic presentation in the content area involved (computer science), acceptance of the hypothesis implies affirmation of the belief that the computer is a unique learning tool.


The computer can support many alternative representational forms, such as temporal compression, immediate access to information, simulation of movement and interaction, multidimensional graphic representation, and sound. In this article we concentrate on just one of those alternatives: the graphic representation of information as an alternative to traditional textual representation. We discuss the need for alternatives, and specifically address the potential of the computer for graphic representation in one specific educational context. We suggest that the traditional mold of education via text will be displaced only if there is empirical evidence that alternative representations are viable and even superior in at least some contexts, and conclude by giving a specific example of the superiority of graphic representation from the research we are conducting in the content area of teaching and learning programming. This computer science context is not chosen because it is the only one in which computing can be fruitfully applied or its potential contribution be clearly understood but rather because it is the specific area in which we work. Equally powerful cases could be made in other traditional subject areas, such as history, music, or art.

ISSUE: LEARNING, COMPUTING, AND CONCEPT REPRESENTATION


I soon felt that the forms of ordinary language were far too diffuse . . . I was not long in deciding that the most favorable path to pursue was to have recourse to the language of signs. It then became necessary to contrive a notation which ought, if possible, to be at once simple and expressive, easily understood at the commencement, and capable of being readily retained in the memory.2


The need for alternative ways to represent ideas, concepts, and knowledge has long been acknowledged. Because different learners have differing capabilities for processing, understanding, and remembering material, no single mode of presenting concepts works equally well for everybody. Alternative representational modes are essential for ensuring that information will be understood and remembered by a wide range of learners. Formal education, however, seems to function on a radically different presupposition: Textual representation is adequate for nearly everything. There are many forces at work perpetuating the textual tone of formal education, such as:


1. Teachers teach as they were taught, thus the tradition self-perpetuates.


2. Few teachers have fluency in preparing graphic or auditory materials since teacher-preparation courses focus primarily on the development of skill in preparing textual materials and evaluation tools.


3. The preparation of nontextual materials is time-consuming and difficult.


4. School curricula involve the use of almost exclusively text-based materials.


5. The established methods for student evaluation are textual methods; decisions about a student’s success, promotion, and acceptance to higher levels of education are based almost exclusively on the results of textual evaluations.


This article presents an argument that the overwhelmingly textual nature of school activity nourishes a serious imbalance in American education, one that probably prevents a considerable number of students from reaching their potential. The traditional practice of using text to the exclusion of other forms of representation for both the material to be learned and the means of assessing whether the student has learned that material is limiting and unnecessary. We argue further that computers can support alternative representations of learning materials, and that their use for such alternatives is essential to realizing the revolution in education predicted as a result of the arrival of computing.


Computers are one of the most powerful tools available to educators for the design and delivery of instruction based on a nontextual representation of material. However, although computers increasingly provide the means for infusing alternatives into our educational environments, most of their potential for so doing is yet to be tapped. This is so for several important reasons, including:


Well-established tradition dictates that schooling focus on the creation, manipulation, and processing of text-based materials.


Much educational software is merely a computerization of traditional textual materials.


Educators have limited experience with ways in which the computer can be used for nontraditional, nontextual applications.


There is little clear proof that alternative representations of learning materials improve learning.


The relatively high financial and human costs for the production and use of alternatively represented materials is not justified without such proof.

THE NEED FOR ALTERNATIVES: BREAKING TEXTUAL TYRANNY


The art of making pictorial statements in a precise and repeatable form is one that we have long taken for granted in the West. But it is usually forgotten that without prints and blueprints, without maps and geometry, the world of modern sciences and technologies would hardly exist.3


Although graphic representation was the primary mode of formal human communication for a long time before the advent of alphabetic symbols, since the invention of the printing press in the mid-fifteenth century, text has become increasingly central in formal communication. The ease with which print materials could be mass-produced and disseminated was alluring. Since early printing presses were capable of producing exclusively textual material, graphic representation was minimized (or at least let slip away) not because it was less effective than textual representation, but because reproducing text was more convenient, easier, and faster.


In consequence, we have become a culture whose dominant mode of communication in formal settings is print, a culture where education is almost synonymous with mastery of textual material. However, a vast amount of our information about people and things comes from nontextual, visual images; as television and video become more and more popular and widespread, this is becoming increasingly true. What we are seeing is a dichotomy between formal and informal communication, between how we typically learn in schools and how we learn elsewhere. We do not argue that graphic representation should again become the primary mode of communication. That, most certainly, would be unacceptable and foolish. We lobby for balance.


We have all heard it said that one picture is worth a thousand words. Yet, if this statement is true, why does it have to be a saying? Because a picture is worth a thousand words only under special conditions—which commonly include a context of words in which the picture is set.4


There must be a middle ground where simple graphic images enhance and sometimes replace textual descriptions. Graphic representations can convey information rapidly5 and can be remembered and recalled rapidly.6 In a culture so deeply enmeshed in text as ours, graphic communication will not—and should not—replace oral and textual/print communication; rather it can and should enhance our communication by providing multiple representations of some information and alternative, simpler, or more direct representations of other information.


Provision of and teaching about alternative representations is too often ignored in the creation of teaching materials and curricular activities. Certainly, intuition suggests that graphics are often easier to understand than are textual descriptions of concepts or information. Western education, however, often proceeds as if the “written world” is sufficient for teaching about the real world.7 A vast majority of student learning time is spent manipulating and creating text. New approaches all too often merely involve the use of different, but still text-based, materials. Even school textbooks that have many illustrations are often used as if they were exclusively textual because many teachers are not fluent in the interpretation of graphically represented information, spending little time using such illustrations as charts and maps as focal points of instruction. At the same time, the problem self-perpetuates because students often do not know and are not taught how to interpret nontextual visual representations.8


Additionally, we have perfected the manipulation of text to such a degree that we “trust” textual representations of knowledge and rely on a learner’s ability to interpret and produce text as the sole or sufficient measure of learning. In general, we use a learner’s ability to understand, manipulate, and produce text as the measure of that learner’s general ability and achievement. From the earliest years of school, evaluation of student understanding and achievement is textually based, and evaluation materials include measures of textual manipulation. Certainly, once we deem that a child should be able to read, achievement tests consist almost exclusively of text manipulation. This exclusive focus on text for evaluation suggests that the ability to manipulate text is the most important and most highly valued skill in American education.


An issue left unaddressed by the textual nature of most evaluative instruments is whether such evaluation really measures understanding. Does the ability to textually describe information or concepts universally indicate that a learner understands or can apply the information or concept? In many cases, educators assume that the ability to transform information into words implies mastery of the learning. There are cases where this may be so; however, there may be as many cases where this is a false test of a learner’s understanding and ability to really use acquired knowledge. A more subtle point is: To what extent does our predilection for transforming all knowledge into text cause us to emasculate, alter, or degrade a concept into a variant that can be textually represented so that we can present it in our usual print form?


At any rate, successful students are, too often, those who have learned to create and manipulate text easily, rapidly, and purposefully. Those who have trouble with text are quite often unsuccessful in our educational institutions. The system offers little support to those who are not textually apt, ignoring or abandoning even those with highly developed alternative styles and aptitudes. Computing can and should be used to correct this imbalance.

THE ROLE OF THE COMPUTER IN GRAPHIC REPRESENTATION


Among the reasons why we traditionally make such minimal use of static graphics in education are the following:


1. Because of a predominantly textual training, teachers find it difficult to imagine how graphic representation might be useful in a particular presentation.


2. Because it is so time-consuming to prepare sufficiently accurate graphic representations, teachers sharply limit their use as part of prepared materials, in-class demonstrations, and lectures.


3. Because of the time it takes to generate most serious graphic work, students cannot be expected to produce much of it as part of regular homework assignments.


4. Because of the difficulty of accurately rendering any but the most trivial images and the consequent inherent danger that an image will be either misleading or confusing, even the occasional use of graphic representation is avoided as much as possible, by both teachers and students.


5. Because of the relatively high expense of including graphics in books, publishers constantly press authors to minimize their use wherever possible.


These are also the reasons we make no use of animated images.


The implications are obvious. Though an art student learns best by being able to recolor ten versions of the same colored design or by a retrospective analysis of all the versions a particular design traversed on the way to completion, the time required to produce these would be enormous and the cost of photographing or color copying every member of an extensive set of versions would be inconceivable within a school budget. Although the calculus student might understand the meaning of a function and its first and second derivatives best by seeing, for each of a group of related functions, the three curves corresponding f, f', and f” at twenty different value points, the time required for student or teacher to produce the appropriate rendering means it cannot be done, no matter how valuable it might be.


The computer’s graphic capability changes this dramatically. Many computers now available, and more of those beginning to appear on the market, have the capabilities to directly and radically weaken reasons 2, 3, and 4, and to indirectly weaken 1 and 5. With software now available and increasingly with that beginning to appear, teachers and students can render all sorts of relevant images at high enough speed to make the analysis of a class of cases or flurry of versions perfectly reasonable as a basis for either class demonstrations or homework assignments. The accuracy of computer-generated images is well beyond what even the best teacher can do by hand and the capability to store a developmental sequence of versions or a set of cases is far in excess of the best set of notebooks any teacher or student has traditionally been able to maintain. There is no precedent for what is now available. It is a resource exclusively spawned by and supported through the computer. It awaits only more thoughtful application.

A REPRESENTATIVE RESEARCH CONTEXT: THE LEARNING OF COMPUTER PROGRAMMING


Intuitions about the communicative richness of graphics abound. It seems perfectly reasonable to believe that a picture is worth a thousand words, but without solid empirical evidence supporting that intuition, the argument for graphics as an alternative representational mode runs the risk of being a good idea that will never have an effect on educational practice. Although the research question of whether graphic representation is a viable alternative to textual representation could be studied in many contexts, we chose to investigate the questions in the context of teaching and learning computer programming because we have been teaching programming to beginners for several years, requiring them to learn and use two programming languages, one of which is graphically represented, the other textually.


Teaching and learning programming provides a good environment in which to investigate the effect of alternative representation for several related reasons. First, although experts have always sought out and relied on graphic representations of various sorts to clarify their arguments, programming languages have remained largely textual even though they incorporate ideas that are clearly representable graphically. There have been ongoing attempts to represent different aspects of programming graphically, but most are intermediate representational tools, ultimately requiring the programmer to write the program in a traditional, textual language. Thus, students have been forced to study programming in an environment based solely on symbolic, arbitrary, alphabetic notations. This type of textual environment for learning may be fine for experts and even for students who are inclined to see things textually, but for students without highly developed linguistic intelligence,9 this exclusively textual approach is probably detrimental.


Second, a large number of students study programming on many levels of schooling. Additionally, there are many students who might wish to study programming but who are kept from doing so by the overwhelmingly symbolic nature of the subject matter and the monolithically textual form of available programming languages. A graphic representation provides the concreteness needed by some learners, helping them to grasp the abstraction of programming more readily.


Third, the growing interest in visual programming suggests that some developers, at least, believe graphic tools make it easier to understand the complex action of the computer.10 If these tools help experts to understand complex systems, it seems reasonable to assume that a visual representation of programming constructs and logic would help novices understand programming more easily and quickly. However, while intuition and anecdotally recorded observation may have convinced many that graphic systems are viable, empirical research verifying that viability is badly needed.


Finally, and maybe most important of all, because computer science is a relatively “new subject” we are still trying to figure out how to teach programming: what materials and methods to use, and what sequence of conceptual presentation to follow. Because we are at such an early stage in the development of educational processes and products for the teaching of programming, educational researchers can and should address these questions, and then propose sound, empirically verifiable approaches for the development of alternative approaches.

FPL: A GRAPHIC PROGRAMMING LANGUAGE


For several years we have been teaching programming to novices using First Programming Language (FPL), a graphically represented programming language under development at Teachers College, Columbia University. The testimonies that FPL is learnable, by a large number of novice programmers with little mathematics or science background, reinforced our original intuition that a graphic representation is useful, and have also suggested that we should seek their verification empirically. FPL is fully described elsewhere,11 so the following comments are intended only to convey the barest sense of how it compares to a textual programming language.


FPL is a graphic representation of classical programming designed for teaching programming to novices. It uses icons and spatial arrangement to graphically represent the programming actions. There are eleven FPL icons; each represents a specific programming action. Eight include programmer-inserted text, the variables and constants of the program; three include no text. The icons supersede the “reserved words” or instructions of text-based programming languages and thereby embody flow of control and logic.


FPL currently runs on computers in the IBM-PC family. It is not merely a computer-based flowchart to be used only for planning a program that must itself be rendered in a traditional, textual language to actually run. FPL is a fully functional language that provides programmers with a graphics-based, spatial environment for the creation of executable computer programs.


The major difference between FPL and other programming languages is its graphic representation. Programs in classical programming languages must be read in a linear, proselike fashion even when the action of the program is nonlinear. Because of its unique spatial layout of connected symbols, FPL allows a reader to see her or his program as a map, in a format that more directly emphasizes its logical structure. Figure 1 presents a typical beginning program in FPL and, for comparison, its counterpart in Pascal.


[39_539.htm_g/00001.jpg]

INVESTIGATIONS OF A GRAPHIC ALTERNATIVE


Within the context of teaching programming and investigating the relative merits of a graphic representation, the exact subcontext we chose was: to determine whether for some novice learners a graphic representation of the conceptual material of computer programming is more effective than a strictly textual representation. Our work investigates two “informationally equivalent”12 representations of programming in an attempt to provide sound empirical verification that in one specific educational context graphic representation of concepts can be superior to traditional textual representation. Just as there are many contexts in which the viability of a graphic representation could be studied, so there are many facts within those contexts that can be investigated. Our current work focuses on two facets of programming, finding bugs (errors) in programs and comprehension of programs written by others.

Bugs in Programs Written by Novices


Our early studies of the effect of FPL’s graphic representation focused on the presence of bugs (errors that result in unwanted or unexpected behavior by the computer during execution) in programs written by novices. In two different studies13 we catalogued bugs in FPL programs written by our students and compared those bugs with observations of novices’ Pascal programs made by Soloway and his colleagues at Yale University.14 We analyzed the types and frequency of logical (nonsyntactic) bugs in these programs. Logical bugs reflect errors in or misconceptions about problem solving rather than errors in the syntax of the programming language. Since the FPL and Pascal programs studied were solutions to the same programming problems, we could compare problem solving across two representational modes, graphic and textual.


We found that, although there seem to be some types of bugs that are language-independent and are closely allied to instruction, others appear to be clearly affected by the programming language used for solving the problem. For example, students writing programs in FPL did not misplace program statements, or icons (as they are called in FPL), while students writing in Pascal frequently did so. In the same vein, certain types of program statements were often missing in the Pascal programs while this was not the case in the FPL programs.


We speculated that the spatial and graphic representation of programming constructs allowed novices to see the structure and scope of the program more clearly, resulting in fewer errors. It may be that the individual elements of the graphic representation were more memorable, thus resulting in more correct construction of programs. It is also possible that in learning elementary programming plans,15 the spatial layout of FPL makes the placement of plan elements such as variable updates and running totals easier to remember.

Program Comprehension


In another series of studies we investigated whether novice programmers learning both FPL and Pascal comprehend programs in one language more accurately and/or more rapidly than in the other.16 Using an on-line reaction time system, our subjects viewed a series of short program segments and answered comprehension questions about those segments. Each program segment was coded in both FPL and Pascal; the logical structure of the two was identical but the context and details differed to mask the repetition. The same questions were asked about both versions of each segment. To provide a more discriminative context for data interpretation, the subjects’ visual and verbal aptitudes were measured using standard instruments.


We found that comprehension of FPL program segments was significantly more rapid than was comprehension of Pascal versions of the program segments. The results were even more pervasive than we had anticipated; while we hypothesized that FPL would be comprehended more rapidly by subjects with high visual aptitude, we found that for almost all subjects, regardless of visual aptitude as we measured it, this was the case. Thus, the findings seem to indicate that comprehension of short program segments is more rapid when the segments are graphically represented.


Comprehension speed is important only insofar as the programmer is able to comprehend accurately. Therefore, accuracy was also measured in these studies. So that comprehension reaction time could be measured with confidence, the questions were deliberately designed to be answerable by the subjects involved in the study. Thus, we were not surprised that 91 percent of all questions were answered correctly. This relatively high level of accuracy with respect to both languages involved confirms subject facility with both, minimizing the possibility that results were due largely to subjects’ being more knowledgeable in the graphic language.


Although the data on incorrect answers were small, they were interesting. Of those questions that were answered correctly for one version of the segment and incorrectly for the other, 79.7 percent were incorrect responses to the questions about the Pascal segment while only 21.3 percent involved incorrect answers to FPL segments. Thus, the results strongly support our hypothesis that graphically represented programs would be comprehended both more rapidly and more accurately than their textually represented counterparts.


For novices, comprehension is a critical aspect of learning programming, integrally involved in program construction and debugging. Helping novices with the incremental steps involved in the larger task we refer to as programming will certainly help allay frustration and avoid failure. More rapid and accurate comprehension will certainly contribute to learning, affecting both program writing and, most certainly, program debugging. Our findings suggest that the graphic representation helps novices assess certain aspects of a program segment more rapidly and accurately, thus leading us to conclude that a graphic representation is an important alternative, at least for novices in this domain.

CONCLUDING COMMENTS


This article argues that educators can and should use the computer to provide alternative ways of representing knowledge. In particular, our research focused on using the computer to implement graphics as an alternative to the traditional textual representation of concepts. The results suggest that a full understanding of the effect of computing on education can only come from rigorous and extensive research exploring the relationships between learning and the alternative forms of representation the computer offers.


We began with the assumption that without empirical evidence proving its effectiveness, the computer will serve only as a platform for educational rhetoric, and offered evidence that the computer can positively affect learning, illustrating the argument by reference to FPL, a graphically represented programming language used to teach programming to novices. We summarized some narrowly focused research conducted in one subsection of a single academic subject, computer science. Though such a minute investigation does not provide all of the evidence needed, it certainly does begin to prove that when used to provide alternative representations of material, the use of computing can affect learning. In particular, when used to present certain kinds of material graphically rather than textually, the computer can help a majority of learners comprehend some aspects of that material faster and more accurately.


Limited though it is, we are convinced this is the sort of careful research that must be done if we are ever to fully appreciate exactly how computing can and cannot help us learn. Clearly a very large agenda of research lies ahead. However, as Shneiderman17 has pointed out with respect to another aspect of human-computer interaction, each small piece of research fits as “a tile into the mosaic” of a large agenda.



Cite This Article as: Teachers College Record Volume 89 Number 3, 1988, p. 360-372
https://www.tcrecord.org ID Number: 539, Date Accessed: 5/21/2022 9:08:35 AM

Purchase Reprint Rights for this article or review
 
Article Tools
Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Robert Taylor
    Teachers College, Columbia University

  • Nancy Cunniff
    Teachers College, Columbia University

 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS