As a graduate student I studied Chinese with a native speaker who was not very confident about his English. Some short grammatical explanations had been prepared by the department and recorded by an English native speaker. The Chinese teacher, instead of reading these grammar lessons to the class or passing them out in printed form, instead took us to the language lab and played the recorded lessons for us. I was not the only student who found these lessons extremely difficult to process and remember. The lessons were written in a fairly dense academic style and plainly not intended to be read aloud. Although I did not learn much Chinese grammar from these experiences, I did realize that the auditory channel was not optimal for this type of learning. These data stayed with me until I was introduced to Richard Clark's (1983) "mere vehicles" hypothesis.
Arguably the most famous theoretical dictum in the field of instructional technology is Richard Clark's assertion that media do not influence learning. After an extensive review of the literature Clark (1983) concluded that, "The best current evidence is that media are mere vehicles that deliver instruction but do not influence student achievement any more than the truck that delivers our groceries causes changes in our nutrition" (p. 445). Although there has been research which has called Clark's hypothesis into question (e.g., Kulik, Kulik, & Bangert-Drowns, 1984), he has continued to defend it, essentially unchanged (Clark, 1992). He has defended his position by arguing that even where apparent differences are detected they are rendered dubious by various forms of confounding. Some of the forms of confounding that Clark has cited are the novelty effect, the "John Henry Effect," editorial bias, unequal instructional strategies, unequal opportunity to learn, and unequal quality of instructional design. He has since abandoned the editorial bias argument, but continues to defend the basic notion that media do not influence learning (Clark, 1994).
Because of my above mentioned experiences, I have long felt that Clark's hypothesis is wrong and could be shown to be such by a simple experiment.
A first step in answering questions about the effects of media is to define the word medium. Surprisingly, Clark (1983) does not define what he means by media. Nonetheless, it is apparent that he intends some kind of "common sense" interpretation. Thus, television, film, radio, books, and teachers are considered to be different media. This approach leads to some problems, since some computer systems are now described as "multi-media" although they cannot carry more than two channels of information (visual and auditory) and in this sense they are no different from film. Also, pictures and text can be presented by both books and computers, but we would not want to say that a story delivered on paper is in the same medium as a story delivered by computer screen. Clark addresses this "symbol system" problem and points out that symbol systems are not uniquely associated with any particular medium. Thus, a particular encoding of information is not a medium. Pictures are not a medium, although pictures can only be delivered though a visual medium.
For the purposes of this paper, a medium will be a particular physical information delivery technology, some of which include storage capacity as well. For example, there are several types of auditory media: magnetic tape, radio, CD-ROM players, computers. Audio media form a class of media? Assuming all are above a certain quality threshold, I assume an auditory message delivered by one will be instructionally equal to a message delivered by another. Needless to say, some of these media afford interactivity, which complicates the matter. However, as Clark pointed out, comparing interactivity to non-interactivity is not a test of media, because interactivity is not a necessary characteristic of any medium, it is an instructional design option.
In the present experiment I will be comparing two media, but I will assume that each is representative of a class of media, all of which share necessary characteristics.
Meanwhile, while reviewing the literature for a different purpose, I unexpectedly came across a line of published research which seriously undermines Clark's position. The research has largely been done by a pair of scholars, Adrian Furnham and Barrie Gunter, and their colleagues, while studying the effects of mass communication on the recall of factual information. These researchers are apparently unaware of Clark's hypothesis and thus have not addressed it.
Although Furnham and Gunter have found a variety of results under differing conditions, they have consistently found that subjects remember material presented in a print medium better than identical material presented in an audio medium or a combined audiovisual medium. Gunter (1987) concluded that this was due to the INHERENT CAPACITIES of these different media to convey knowledge. These differences were found for samples drawn from populations of schoolchildren, university students, military personnel, and non-students. Across these categories the most consistent result was that subjects REMEMBER BETTER from print materials than audio or audiovisual materials.
The following table (adapted from Furnham, Gunter, & Green, 1988) illustrates their findings.
|News||128 school children||P > A = AV||Gunter, Furnham, & Gietson, 1984a|
|68 students||P > A > AV||Furnham & Gunter, 1985|
|117 military students||P > A = AV||Gunter & Furnham, 1986|
|101 adults||P > A > AV||Furnham & Gunter, 1987|
|Political broadcast||65 students||P > AV > A||Gunter, Furnham, & Leese, 1986|
|Advertising||69 students||P > AV > A||Furnham, Benson, & Gunter, 1987|
|Magazine program||63 students||P > AV > A||Furnham, Proctor, & Gunter, 1988|
|Science program||60 students||P > AV A|
|Furnham, Gunter, & Green, 1988|
|P > AV = A|
(easy and hard)
|60 students||P > AV = A|
|Furnham, Gunter, & Green, 1988|
|P > A > AV|
|P > AV > A|
|P = AV > A|
P = paper
A = audio
AV = audiovisual
Furnham and Gunter's research methodology is best exemplified by their article, "Memory for information from a party political broadcast as a function of the channel of communication" (Gunter, Furnham, & Leese, 1986). An approximately five minute television broadcast was delivered to university students in three groups. Group one received the TV broadcast in its normal form. Group two received the broadcast with audio only. Group three received a transcribed script of the broadcast. Time was held constant in all treatments. Memory was tested with free recall, cued recall, and multiple choice questions, in that order. Results, as reported above, indicated that retention was superior with print materials. Note that the content was designed for TV and that the print condition was the least "natural." These methods and results are typical.
Although Furnham and Gunter's studies have repeatedly found significant differences in achievement with different media, their work is not free from minor, but possibly confounding, factors. First, different media may not be equally "novel" or familiar. This objection cuts both ways. If TV were superior you could argue that this was due to the novelty effect. If the printed text were superior you could argue that performance was better because it was familiar. Thus the argument is equivocal and cannot be resolved without an experiment which holds novelty approximately equal.
Second, reading may be superior to listening because of time differences. Although Furnham and Gunter held total time constant, normally people can read faster that an announcer speaks. Thus the subjects in the print condition may have been able to read the text more than once. Clark mentions this "reviewability" problem in his original article, although in a slightly different context. He noted that the reviewability of computer-based presentations had been touted as a feature of the "new" media. Clark dismissed this by arguing that a teacher can review material also, so reviewability is not a unique characteristic of any medium. However Clark's argument begs the question of whether there are some media which inherently lack reviewability (such as TV) and therefore may be at a disadvantage instructionally. Although reviewability is a nearly inherent characteristic of non-volatile media under normal conditions, it can be controlled for under experimental conditions and if controlled would eliminate this objection.
Another possible confound is the students' attitudes towards the different media. Salomon (1984) offered the AIME (Amount of Invested Mental Effort) hypothesis, which states that students may view TV as "easy" and books as "hard" and may not attend equally to both. This is not an inherent characteristic of the medium, but rather a cultural attitude which may differ in different communities. If this were true it would save Clark's hypothesis in the cosmic sense, but it would not save it for limited areas such as countries, because the differential effects of media would be entirely predictable and "necessary" within a given cultural context. However, if similar results are obtained across widely different populations, it becomes implausible to argue that relatively fluid factors like "attitudes" are the underlying cause.
Before I discovered the Furnham and Gunter research, I was designing a research study to test Clark's hypothesis. It happened that this experimental design resembled Furnham and Gunter's studies with one exception: It seemed to eliminate the possible confounds.
The experimental design was very simple. It involved presenting the same verbal material sentence-by-sentence by computer in two media, sound and text. I selected a short text describing the Wright Brothers' experiments before their famous flight. This text was chosen because the subject was familiar, allowing the students to orient themselves to the material quickly, while keeping the actual content of the passage unfamiliar. This allowed a good measure of comprehension and memory for declarative information. Achievement was measured as in the Furnham and Gunter studies, by having students free-recall propositions from the passage immediately after the presentation. Scores were the total number of true propositions recalled.
The passage was recorded digitally, sentence by sentence, and the reading time for each sentence was recorded. The passage was presented by computer in two different media. The software interface was identical for both treatments, except that for the text version the students saw the text on screen, sentence by sentence, while for the audio version, the passage was presented by headphones, sentence by sentence. The text sentences were on screen for exactly the same elapsed time as it took to read the sentences aloud. Students could not control the pace of presentation once they had started. By presenting an identical passage by computer with the time of presentation equalized, the experiment is virtually as pure a medium comparison as is conceivable. By equalizing the treatments so carefully , this experiment eliminates the novelty, AIME, and time confounding problems discussed above. It also seems to eliminate Clark's general confounding objections.
Subjects (n = 36) were drawn from undergraduates in teacher training and assigned to one of two treatments in the order in which they appeared in the computer lab. The only exception to this was that males and females were assigned separately, so that an approximately equal number of males and females were assigned to each treatment. The students had some experience with computers, but could not be called highly experienced.
Results, which are presented in Table Two, indicate that this study confirms Furnham and Gunter's results. The text group, even with reviewability controlled, remembered significantly (t = 4.119, p < 0.0002) more propositions than the audio group. This contradicts Clark's hypothesis that media do not influence learning.
|Unpaired t value = 4.119||Probability (2-tail) = .0002||df = 34|
|Group 1, n = 17||mean = 6.353||sd = 2.805|
|Group 2, n = 19||mean = 3.105||sd = 1.883|
I communicated my results to Richard Clark and he responded by citing the "Beliefs and Attributions" arguments which he used in his original argument. This argument states that because students attribute different "difficulty" levels to different media and different levels of self-efficacy to themselves, they will invest different levels of effort to learning from those media. However, it is not clear what kinds of conclusions should be drawn from these premises. Indeed it is hard to imagine what evidence Clark would cite supporting such premises. What evidence is there that students predominantly try harder when reading from a computer screen than when listening to a computer through headphones? Why would this correlate with Gunter and Furnham's subjects who were not using computers at all?
Salomon's (1984) study found that efficacy correlated negatively with AIME for students viewing a TV film, but positively with students reading a story. In other words, because students judged TV to be easy, they had high self-efficacy, so they invested less effort (by self-report). Conversely, the students judged print to be difficult, rated their self-efficacy lower, and then tried harder. As a result the print group was able to make more correct inferences. The groups did not differ with respect to factual recognition (Note that the results reported here and by Furnham and Gunter involve recall of facts!). Either way, if such attitudes are evoked by media and then have causal relationships with achievement, it is difficult to hold that "media don't influence achievement" although that influence might be unpredictable at the individual level.
In order to test whether my students might have differing attitudes towards different media, I surveyed a group of 16 students selected from the original pool. I asked them two questions (embedded in a list of 13 questions) about the relative difficulty of learning from tape or books. They rated ease of learning on a five-point Likert scale. Books were rated significantly easier to learn from than tape (t = 2.6, p < 0.02). Students also rated their level of effort with books or tapes. They rated their effort significantly higher (t = 7.3, p < .0001) with books. These questions were administered two months after the experiment and without reference to the experiment, and the wording ("tape" and "books" rather than computer audio and text) was chosen to avoid obvious association with the experiment.
These results indicate that the attitudinal differences towards media are present (as Clark asserted to me). However, they are still problematical. Note that my results are the opposite of Salomon's. My students reported more, not less, effort with the "easier" medium (print). Also rating books easier than tapes does not conform to Salomon's findings. His subjects rated TV easier than print. Thus my students should have tried harder with audio because they rated it more difficult.
There may be several reasons for these differences. Salomon's subjects were sixth-graders while mine were college students. Also, tape is not the same as TV. Self-reports of effort do not necessarily reflect actual effort, especially over short periods of time. Additionally, we do not know that the students in the audio condition (or in Salomon's studies) actually made less effort. Simple observation of the students gave no indication that the audio group was behaving differently from the text group.
It appears that there are two possible conclusions that can be drawn from these data. The first is that there is a direct causal relationship between the medium of instruction and the amount of declarative knowledge retained by students. Students receiving instruction through an auditory medium will retain less declarative knowledge than students learning visually from a text. The second possible interpretation is that older students faced with learning from an auditory medium perceive that medium as more "difficult" than print and therefore invest less effort in their learning (contrary to Salomon), resulting in less retention of declarative knowledge. If my experiment were all the evidence we had, we might judge the second explanation plausible, but given the repeated results of Gunter and Furnham with a variety of subjects and media presentations and given that my results replicate theirs, it is unlikely that all these different subjects, from different countries, dealing with different content (some of it "easy") and presented with differing specific media presentations could have all been under the control of the same "attitudes and beliefs." I don't see how we can conclude other than that MEDIA INFLUENCE LEARNING.
Clark, R.E. (1983). Reconsidering research on learning from media. Review of Educational Research, 53, 445-459.
Clark, R.E. (1984). Research on student thought processes during computer-based instruction. Journal of Instructional Development, 25, 2-5.
Clark, R.E. (1984). "Where's the beef?": A reply to Heinich. Educational Communication and Technology Journal, 32, 229-232.
Clark, R.E. (1985). Confounding in educational computing research. Journal of Educational Computing Research, 1, 137-147.
Clark, R.E. (1985). Evidence for confounding in computer-based instruction studies: Analyzing the meta-analyses. Educational Communication and Technology Journal, 33, 249-262.
Clark, R.E. (1991, February). When researchers swim upstream: Reflections on an unpopular argument about learning from media. Educational Technology, p. 34-40.
Clark, R.E. (1994). Media will never influence learning. Educational Technology Research and Development, 42, 21-29.
Furnham, A., & Gunter, B. (1985). Sex, presentation mode and memory for violent and non-violent news. Journal of Educational Television, 11, 99-105.
Furnham, A., & Gunter, B. (1987). Effects of time of day and medium of presentation on recall of violent and non-violent news. Applied Cognitive Psychology, 1, 255-262.
Furnham, A., Benson, I., & Gunter, B. (1987). Memory for television commercials as a function of the channel of communication. Social Behavior, 2, 105-112.
Furnham, A., Gunter, B., & Green, A. (1988). Remembering science: The recall of factual information as a function of presentation mode. Applied Cognitive Psychology, 4, 203-212.
Furnham, A., Proctor, T., & Gunter, B. (1988). Memory for material presented in the media: The superiority of written communication. Psychological Reports, 63, 935-938.
Gunter, B. (1987). Poor reception: Misunderstanding and forgetting broadcast news. Hillsdale, NJ: Lawrence Erlbaum.
Gunter, B., & Furnham, A. (1986). Sex and personality differences in recall of violent and non-violent news from three presentational modalities. Personality and Individual Differences, 7, 829-837.
Gunter, B., Furnham, A., & Gietson, G,. (1984). Memory for the news as function of the channel of communication. Human Learning, 3, 265-271.
Gunter, B., Furnham, A., & Leese, J. (1986). Memory for information from a party political broadcast as a function of channel of communication. Social Behavior, 1, 135-142.
Kulik, J., Kulik, C., & Bangert-Drowns, R. (1984). Effects of computer based education on secondary students. Paper presented at the annual meeting of the American Educational Research Association, New Orleans, LA.
Salomon, G. (1984). Television is "easy" and print is "tough": The differential investment of mental effort in learning as a function of perceptions and attributions. Journal of Educational Psychology, 76, 647-658.