| Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting |
|
The major sections of this paper are summaries of experimental findings pertaining to (1) the influence of ethos upon the effect of the communication, (2) techniques for generating or changing ethos, and (3) measurements of one or more aspects of ethos and attempts to assess the relative levels of ethos of individuals or groups. Kenneth Andersen and Theodore Clevenger, Jr.: From Kenneth Andersen and Theodore Clevenger, Jr., "A Summary of Experimental Research in Ethos," Speech Monographs, 1963, 30, 59-78. Reproduced with permission of the authors and publisher. INFLUENCE OF ETHOS UPON THE INTENDED EFFECT OF THE COMMUNICATION Experiments concerning ethos have dealt with many and varied topics: with the effects of differences in prestige, credibility, likeableness, and other variables upon attitudes toward political-social issues, upon evaluations of art and literature, and upon learning; with the relative effectiveness of majority and expert opinion and the relative susceptibility of the sexes, different age groups, and per sons of various educational levels to prestige suggestion; and with the temporal effects and the permanency of the attitude change and the learning induced by different levels of ethos. It is important to remember that these studies, which arise from such fields as psychology, speech, sociology, and education, are quite diverse in origin, that many of the experimenters did not use rhetorical terminology, and that many of them also did not perceive a relationship between their studies and ethos. Studies are included, however, if the independent variable is a difference in treatment which is basically related to ethos and if the dependent variable is some measurement which is basically a communication effect index. 1. Such as John Highlander, "Audience Analyzer Measurements and Informational Effects of Speaker Variables in Radio Talks," unpubl. diss. ( Wisconsin, 1953) ; Franklyn Haiman, "An Experimental Study of the Effects of Ethos in Public Speaking," unpubl. diss. (North western, 1948) ; also briefly reported in SU, XVI (Sept., 1949), 190-202. 2. Such as Charles Osgood, George Suci and Percy Tannenbaum, The Measurement of Meaning (Urbana: University of Illinois Press, 1957) ; Erwin Bettinghaus, "The Operation of Congruity in an Oral Communication Situation," unpubl. diss. ( Illinois, 1959). THEORETICAL AND METHODOLOGICAL DIFFERENCES Studies differ so much in the definition of ethos and in certain other theoretical and methodological features that an analysis of these distinctions is a necessary preliminary to reporting the experiments. 1. FIXED ETHOS VS. CONGRUITY HYPOTHESIS. In most studies the ethical element is treated as relatively fixed in value during the communication act, and persuasion is construed as the linking of a proposition with an approved source for a positive effect or a disapproved source for a negative one. However, in some recent studies, especially those using semantic differential measurement, ethos is regarded as flexible, because during the act of communication alterations in the image of the speaker may be caused either by the sender's propositions or by other situational factors. 2. ETHOS ASSUMED VS. ETHOS MEASURED. Early studies of ethical effects commonly followed the pattern of employing two sources (such as Franklin Roosevelt and Herbert Hoover) assumed to differ greatly in credibility, prestige, or some other ethical component and then comparing the attitude change for Group I, which received the message credited to the first source, with that for Group II, which received the same message except that it was ascribed to the second source. This method assumes that for the group of subjects in question, the experimenter can determine intuitively the relative levels of ethos of the given sources. Recent studies, in contrast, have tended to measure ethos. Experimenters have either selected their sources on the basis of pretests of credibility or chosen them arbitrarily and then checked for credibility differences by direct measurement after the completion of the experiment. The last of these techniques, of course, is valid only if one is willing to espouse the fixed ethos model; for if the image of the speaker may change during the speech, a measurement rendered after the address may be quite deceptive concerning ethos at the outset. 3. TOPIC-ORIENTED VS. TOPIC IRRELEVANT ETHOS. The assumption for the majority of the studies apparently is that the prestige, the credibility, or some other ethical characteristic of the speaker varies from one topic to another. Thus, in most of the studies of expert opinion the authorities were selected because they were reputed to be well informed on the topic of the experimental message.6 Some studies, on the other hand, seem to be based on a concept of generalized credibility and to discount or ignore the possibility that the prestige varies from topic to topic. ------------------ 3. Such as Helen Lewis, "Studies in the Principles of Judgments and Attitudes: IV. The Operation of 'Prestige Suggestion,'" Journal of Social Psychology, XIV (1941), 229-256. 4. Such as Muzafer Sherif, "An Experimental Study of Stereotypes," Journal of Abnormal and Social Psychology, XXIX 11935 t, 371-375; Herbert Kelman and Carl Hovland, "'Rein statement' of the Communicator in Delayed Measurement of Opinion Change," Journal ol Abnormal and Social Psychology, XLVIII (1953), 327-335. 5. Such as Malcolm Moos and Bertram Koslin, "Prestige Suggestion and Political Leader ship," Public Opinion Quarterly, XVI (1952), 77-93; Irving Lorge with Carl Curtis, "Prestige, Suggestion and Attitudes," Journal of Social Psychology, VII (1936), 386-402. 6. Such as Clare Marple, "The Comparative Susceptibility of Three Age Levels to the Suggestion of Group Versus Expert Opinion," Journal of Social Psychology, IV (1933), 176 186. 7. Such as Raymond Bernberg, "Prestige Suggestion in Art as Communication," Journal of Social Psychology, XXXVIII (1953), 23-30; William Michael, Bernard Rosenthal, and Michael DeCamp, "An Experimental Investigation of Prestige-Suggestion for Two Types of Literary Material." Journal of Psychology, XXVIII (1949), 303-323. 8. Such as Helen Lewis, loc. cit. Such as Herbert Birch, "The Effect of Socially Disapproved Labeling upon a Well Structured Attitude," Journal of Abnormal and Social Psychology, XL (1945), 301-310; David Cole. "'Rational Argument' and 'Prestige-Suggestion' as Factors Influencing Judgment," Sociometry, XVII (1954), 350-354. -------------------- 4. AVERAGE VS. INDIVIDUAL MEASURE. Although the assumption in most studies is that the experimental group as an entity places the communicator at a certain level of prestige, 7 in some studies ethos is regarded as differing from one subgroup to another, and data are treated separately for such variables as sex, occupation, educational status, and political affiliation. 8 A few studies even consider the prestige of the source in respect to each individual auditor.6 Whereas in the first two types of experiment the usual statistical test is for the significance of difference between means, in studies of the individual auditor the common method is correlation. 5. EXTENT OF AUDIENCE ANALYSIS. Finally, the studies differ in that some examine audience characteristics, whereas others do not. Both approaches have interpretative hazards as well as distinctive advantages. In studies which assess the effect upon attitude change of such audience properties as sex, age, and educational level it is also possible (although infrequently done) to investigate the interaction of ethos with each of the audience variables. Thus, a study in which two levels of prestige are employed with an audience of men and women can include data on the effect of prestige level upon attitude change (ignoring sex), the difference in the persuasibility of the sexes (ignoring prestige), and differences in the relative susceptibility of the two sexes to prestige and non-prestige communication (the interactions). Careful interpretation, however, is necessary: First, the experimenter must distinguish overall persuasibility differences between the sexes (main effect of sex) from prestige-suggestibility (the interaction). Second, where prestige is taken with reference to the entire sample of subjects, he must note the possibility of confounding prestige level with sex-that is, a source may not have the same prestige for the two sexes, and this difference may result in a spurious sex-by-prestige level interaction if prestige level is measured as a group average. Thus, some of the results seeming to show greater prestige-persuasibility for women than for men may have been products of concealed differences in the prestige level of the source for the two sexes. Within the limits of the five methodological distinctions described above, the studies of the effects of ethos present a reasonably harmonious body of findings. In the following pages those studies employing the conception of a fixed ethos model will be presented first, and the limited number employing the congruity model will follow. STUDIES ASSUMING THAT ETHOS IS FIXED A number of studies which employ the relatively common fixed ethos model indicate that certain ethical factors can produce changes in attitude toward political and social issues. Arnett, Davidson, and Lewis found that a group of graduate students shifted significantly toward agreement with graduate educators on Harper's test of liberalism." The study was conducted without a control group, however, and during the lapse of four weeks between the two administrations of the test, factors other than prestige may have operated to produce the observed shifts. Birch studied the effect of political labels of Fascist or Communist and Reactionary or Liberal on college students' judgments of two statements." No significant differences in preference for the two statements were observed, but this conclusion may be misleading. The fact that ninety-nine percent of all subjects favored one statement over the other may have masked any possible prestige effect. While the preceding studies were concerned with the effect of referential group or class prestige upon attitude change, a number of studies have been directed toward an investigation of the prestige of individuals. Saadi and Farns worth found greater acceptance for dogmatic statements which were attributed to well-liked persons than to the same assertions when attributed to disliked individuals. Lorge and Curtis found a significant tendency for subjects to shift opinion toward the supposed position of a prestige source, but they found no significant negative shift when the proposition was linked with a disapproved source. In apparent conflict with these findings are the results obtained by Lewis. She reported that college students remained relatively unchanged in the evaluation of statements and that they tried to explain away the "prestige source" through rationalization." Unhappily, the conclusions to the study show the bias of an author who quite evidently hoped to support an hypothesis: for example, she describes rank-order correlations of a magnitude of .50 as "high." This bias renders suspect the assertion that informal interviews with the subjects and free responses revealed that suggestion, when effective, usually redefined an ambiguous situation. A more satisfactory design for testing a similar hypothesis was that employed by Moos and Koslin, who discovered that vague quotations were those which were the most likely to be influenced by attribution to differing sources." Hastorf and Piper, using a variety of problems, studied the effects of sup posed ratings of businessmen and educators on the attitudes of subjects. They found that all groups, including one which was instructed to duplicate its pretest responses and ignore the supposed ratings, shifted significantly." Smith found that printed propaganda statements when labeled as fact produced greater belief than when labeled as rumor. The success of the "fact" label, however, clearly varied with the prior attitude of the subject and with the relation of the alleged "fact" to truth» The objective of all of the above studies was to assess the effects of prestige upon judgment of political and social issues, and the method in all instances was to link a source with a proposition but to provide no message by which the source supported the proposition. A question of more immediate interest to students of speech is whether differences in the speaker's prestige significantly influence the persuasive outcome of a speech. ------------------------ 10. Claude Arnett, Helen Davidson, and Hallen Lewis, "Prestige as a Factor in Attitude Changes," Sociology and Social Research, XVI (1931), 49-55. 11. Loc. cit. 12. Mitchell Saadi and Paul Farnsworth, "The Degrees of Acceptance of Dogmatic Statements and Preferences for Their Supposed Makers," Journal of Abnormal and Social Psychology, XXIX (1934), 143-150. 13 Loc. cit. 14 Loc. Cit. 15 Loc. cit. 16. A. H. Hastorf and G. W. Piper, "A Note on the Effect of Explicit Instructions on Prestige Suggestion," Journal of Social Psychology, XXXIII (1951), 289-293. 17. George Smith, "Belief in Statements Labeled Fact and Rumor," Journal of Abnormal and Social Psychology, XLII (1947), 80-90. 18. Loc. cit. 19. Edward Strother, "An Experimental Study of Ethos as Related to the Introduction in the Persuasive Speaking Situation," unpubl. diss. (Northwestern, 1951). 20. Stanley Paulson, "Experimental Study of Spoken Communications; The Effects of Prestige of the Speaker and Acknowledgement of Opposing Arguments on Audience Reten tion and Shift of Opinion," unpubl. diss. ( Minnesota, 1952) ; also briefly reported in SM, XXI (1954), 267-271. 21. Carl Hovland and Wallace Mandell, "An Experimental Comparison of Conclusion Drawing by the Communicator and the Audience," Journal of Abnormal and Social Psychology, XLVII (1952), 581-588. -------------------- Haiman presented to three groups a tape-recorded speech variously attributed to Thomas Parran, Surgeon General of the United States; to Eugene Dennis, Secretary of the Communist Party in America; and to a "Northwestern University Sophomore." Not only was Parran rated significantly more competent than the other two, but also, as measured by the Woodward Shift-of-Opinion Ballot, his speech was significantly more effective in changing attitude than was either of the other two. The "Dennis" and the "Sophomore" speeches did not differ significantly.'" Employing essentially the same techniques-a tape-recorded speech, differing introductions, and the Woodward ballot-Strother and Paulson in separate studies obtained results similar to Haiman's. Not only did Strother find significant differences in the persuasiveness of the "Parran" and the "Dennis" speeches, but also he noted that only those who thought they had been listening to Dennis wrote unfavorable comments concerning the speech techniques employed. Paulson attributed a taped speech to a political science professor and to a student. For female auditors there was no significant difference in the effects of the "two" speeches, but among the male auditors the proportion of those shifting opinion was greater for the group which thought it had been ad dressed by the professor. The supposed differences in prestige level in the experiments cited above were assumed to be quite large, and the methods of establishing the prestige levels were straightforward and obvious. On the other hand, Hovland and Mandell, in an effort to assess subtler sources of the speaker's image, manipulated credibility through the suggestion of differing degrees of selfish interest and self-motivation. The non-significant difference in attitude change which the speakers produced was very small, but the audiences, apparently reacting to their presumed prejudices, rated the "unbiased source" as the significantly fairer and more honest of the two. Since these evaluations were rendered after the speech, the initial ethos of the two sources, the point at which the "biases" of one began to emerge, or the ways in which the images of the two speakers changed during the speech are unknown. A study by Kraus likewise suggests the possibility of evaluating indirect, implicative sources of ethos. Using pairs which were racially homogeneous and others which were racially heterogeneous, he compared whites with Negroes in respect to their persuasiveness in filmed discussions of segregation issues. The results indicated that arguments favorable to integration were more persuasive when advanced by the heterogeneous pairs, and Kraus explained the results in terms of differing levels of credibility. All the studies mentioned thus far have dealt with ethos as determined by the position or reputation of the source. Messages, if used, have been standardized so that the only variable was the introduction given the speaker. Other studies, in contrast, have been designed so that some internal message elements have been varied systematically. Gilkinson, Paulson, and Sikkink, who incorporated or excluded authority quotations in two versions of the same speech, found that both versions engendered a significant shift in attitude with only a trend to favor the inclusion of authorities. In another study Sikkink similarly employed quotations, but neither attitude shift nor ratings of convincingness showed significant differences» While the use of authorities certainly has persuasive implications beyond the ethical dimension (and indeed the authors of these experiments apparently did not consider ethos the critical variable), the fact that the speaker was not evaluated as significantly more convincing when he used authorities suggests that citing reputable sources does not necessarily enhance ethos--as some theorists have suggested. The two studies above are included within the fixed ethos model because the prestige of the authorities seemingly served directly as the basis for the shift in opinion, if any. Other experimenters varied the procedure by apparently employing authorities for the purpose of altering the image of the speaker; this altered image, in turn, was to serve as the warrant for the persuasive effect. (Possibly both effects could occur.) Studies of attitude changes dependent upon such at tempts at artistic ethos are reported in a subsequent section of this paper. Historically parallel to the study of the effects of ethos upon political and social attitudes has been the study of its effect upon judgments of literature, art, and matters of personal taste. In three experiments in Turkey and at Harvard Sherif found correlations of .45 to .53 between rankings of authors and subsequent rankings of passages to which authors' names were randomly attached. ------------------ 22. Sidney Kraus, "An Experimental Study of the Relative Effectiveness of Negroes and Whites in Achieving Racial Attitude Change Via Kinescope Recordings," unpubl. diss. ( Iowa, 1959) ; SM, XXVII (1960), 87-88. 23. Howard Gilkinson, Stanley Paulson, and Donald Sikkink, "Effects of Order and Authority in an Argumentative Speech," QJS, XL (1954), 183-192. 24. Donald Sikkink, "An Experimental Study of the Effects on the Listener of Anticlimax Order and Authority in an Argumentative Speech," Southern Speech Journal, XXII (1956), 73-78. 25. See p. 212. Still other implications for a theory of ethos stemming from the authority quotation problem will be discussed in a subsequent paper. ----------------------- Sherif asserts that the name of the author exerts an influence upon ratings of passages. Michael, Rosenthal, and DeCamp matched authors with prose and poetry passages and found little evidence of the effect noted by Sherif. Although they claimed methodological improvements over the Sherif study, their rank-of summed-ranks technique actually produced a measure of dubious statistical reliability." The entire study was conducted in such a manner that results con firming the Sherif finding were highly unlikely. The interpretation of their inconclusive results as evidence contrary to the Sherif hypothesis seems unjustified. More recently, in India, Das, Rath, and Das studied the effect of author prestige upon evaluations of poetry. Working with quite small groups and crude statistical measures, they concluded that prestige influenced judgment greatly but that this effect was weakened when the factors of understanding and merit were stressed. Judgments of art seem to be similar. Data obtained by Farnsworth and Misumi displayed a trend indicating that recognition of the artist's name had some favorable effect on the evaluations of pictures." In another experiment Bernberg found that positive and negative evaluations of alleged art critics significantly affected the judgments by artistically naive students with regard to seven of ten paintings. Cole presented abstract finger paintings for discussion in small groups. In situations in which the art teacher presented judgments in opposition to those of the group, significant shifts occurred only when the teacher was present. A peer leader, to cite a second finding, secured significant shifts only when he also presented pseudo-rational arguments." Again, similar effects have been found in the area of personal taste and perceptions. Duncker presented a story to nursery school children in which a fictional hero endorses a food actually less desirable than an alternative selection. ------------------- 26. Loc. cit. 27. Loc. cit. 28. The problems in the use of a rank-of-summed-ranks technique are discussed by Roger Nebergall, "Some Applications of Measurement Theory to the Judgment of Speech Contests," unpublished paper read at the Central States Speech Association Conference, April 8, 1960. 29. J. P. Das, R. Rath, and Rhea Stagner Das, "Understanding Versus Suggestion in the Judgment of Literary Passages," Journal of Abnormal and Social Psychology, LI (1955), 624-628. 30. Paul Farnsworth and Issei Misumi, "Further Data on Suggestion in Pictures," American Journal of Psychology, XLIII (1931), 632. 31 Loc. cit. 32 Loc. cit. ------------------ The after-effect was decidedly positive-a large percentage of the children selected the endorsed food when given a choice. Over a period of twelve days, how ever, the selection of the less satisfying food declined to the level of a control group. Some of the initial preference for the less desirable food was reinstated by recalling the story, but this effect degenerated very quickly. Donceel, Alimena, and Birch presented adults and high school and college students with personality descriptions of themselves. These supposedly came from tests and expert evaluations, but actually were determined by chance. Under mild suggestion a significant number of students accepted these statements as valid, and under strong suggestion all subjects yielded. They accepted as true the false descriptions of their personalities and reversed previous answers to questions in a personality test. Aveling and Hargreaves found personal suggestion capable of affecting performance in a variety of perceptual and psychomotor tasks, but they also secured evidence of strong negative suggestibility among some of their subjects." Although there is little reason to suppose that those elements of ethos which are designed to obtain attitude change are also capable of producing differences in learning, a small number of studies pertain to this possibility. Weiss taught responses to groups of students, one of which was told that the answers were untrue. No differences in learning occurred, but what was learned correlated with the attitude change which took place during the experiment. Paulson found no significant differences in retention between high and low ethos sources, although certain audience variables did appear to be related to learning. Sikkink's results were substantially the same." An experiment by Harms shows that doze test scores are somewhat higher when the speakers are high in status than when they are low. The inferred reason for this result is that high-status speakers are more "comprehensible." A further result, secured through a differential analysis of listener groups, is that listeners respond with greater comprehension to those from their own class than to speakers from either a higher or a lower class." ------------------------------ 33. Karl Duncker, "Experimental Modification of Children's Food Preferences Through Social Suggestion," Journal of Abnormal and Social Psychology, XXXIII (1938), 489-507. 34. Joseph Donceel, Benjamin Alimena, and Catherine Birch, "Influence of Prestige Suggestion on the Answers of a Personality Inventory," Journal of Applied Psychology, XXXIII (1949), 352-355. 35. F. Aveling and H. L. Hargreaves, "Suggestibility with and Without Prestige in Children," British Journal of Psychology, XII (1921-1922), 53-75. 36. Walter Weiss, "A 'Sleeper' Effect in Opinion Change," Journal of Abnormal and Social Psychology, XLVIII (1953). 173-180. 37. Loc. cit. 38 Loc. cit. 38. Leroy Stanley Harms, "Social Judgments of Status Cues in Language," unpubl. diss. (Ohio State, 1959) ; SM, XXVII (1960), 87. ----------------------- The above studies were concerned with the effects of the ethos of individual communicators. A smaller number of investigations have attempted to compare the effects of expert opinions with those produced by majority opinion. Using as a criterion the frequency with which the subjects reversed their preferences so as to conform to the prestige group, Moore measured the relative influence of majority and expert opinions upon judgments of grammar, ethics, and music. The two sources were about equally effective except with respect to grammar, where the majority opinion prevailed by a ratio of 10 to 7. The primitive design of this experiment may have concealed other differences. An experiment by Marple, who found that both the group and experts influenced opinions about solutions to seventy-five assorted problems, reinforced Moore's results. Majority opinion was roughly one-third more effective than ex pert opinion with students and roughly one-fifth more effective with adults:" With respect to religious beliefs( Burtt and Falkenburg discovered that opinions of both the majority and experts influenced judgments significantly, that expert (clerical) opinions tended to have greater influence than majority views in some matters of religious belief, and that a contrary tendency existed in other areas. Incidental findings of a number of studies bear upon the question of the relative susceptibility of various audience types to prestige as a means of suggestion. Within the narrow range which an undergraduate psychology class affords, Hovland and Mandell found that personality and intelligence were not related to prestige-suggestibility. Kersten reports a similar finding for intelligence; but Wegrocki reports a tendency for intelligence to be negatively associated with prestige-suggestibility. Strother discovered no shifts in opinion which correlated with either sex or the urban-versus-rural dimension, but he did find that members of the audience with initially neutral views on the speech topic were significantly more responsive to variations of ethos than were either the pro or the con groups. Kersten, Paulson, and Pross obtained results confirming those of Strother. ------------------ 40. Henry Moore, "The Comparative Influence of Majority and Expert Opinion," American Journal of Psychology, XXXII (1921), 16-20. 41. Loc. cit. 42. Harold Bunt and Don Falkenberg, Jr., "The Influence of Majority and Expert Opinion on Religious Attitudes," Journal of Social Psychology, XIV (1941), 269-278. 43 Loc. cit. 44 Barbara Kersten, "An Experimental Study to Determine the Introduction upon the Persuasive Speech that Followed," unpubl. State College, 1958). 45 Henry Wegrocki, "The Effect of Prestige Suggestibility on Journal of Social Psychology, V (1934), 384-394. 46 Loc. cit. 47 Loc. cit. 48 Loc. cit. 49 Edward Proas, "A Critical Analysis of Certain Aspects of Ethical Proof," unpubl. diss. ( Iowa, 1942) ; Paulson, loc. cit. Effect of a Speech of thesis (South Dakota Emotional Attitudes," ----------------------------------- Siklcink found that women rated the persuasiveness of all speeches significantly higher than did men, but that women were neither easier nor harder to in fluence than men. Cathcart also concluded that sex was not significantly related to persuasibility. Pross reported some indication that women were the more suggestible, and Wegrocki also concluded that girls, as compared with boys, tended to be more suggestible and to react more strongly to sympathetic propaganda." Paulson found that women reacted more but retained less information. Freshmen, also according to Paulson, tended to shift less in response to the high ethos source than did upper classmen, but there was no guarantee that the freshmen and the upperclassmen perceived the high ethos source in the same light." Cathcart found that education, speech training, and subject matter competence had no effect on persuasibility. The discovery by Aveling and Hargreaves of great differences in suggestibility on a number of perceptual and psychomotor tasks leads to speculation that two sharply divided groups, the suggestible and the contrasuggestible, may exist. They found no tendency, however, for suggestibility to correlate with any of a number of psychometric variables. Marple found that high school and college students shift more than do adults." A single study has illustrated the possibility of investigating the effects of audience size upon the relationship between ethos and attitude change. Knower compared the effect of delivering a speech in an audience situation with giving the speech to one auditor at a time. The speech in the individual situation was somewhat more effective, women were more influenced than men, and women speakers obtained greater attitude shifts than did men. ------------------------ 50. Loc. cit. 51. Robert Cathcart, "An Experimental Study of the Relative Effectiveness of Four Methods of Presenting Evidence," SM, XXII (1955), 227-233. 52. Loc. Cit. 53. Loc. cit. 54. Loc. Cit. 63. Loc. cit. 66 Loc. cit. 67. Franklin Knower, "Experimental Studies of Changes in Attitudes: I. A Study of the Effect of Oral Argument on Changes in Attitude," Journal of Social Psychology, VI (1935), 315-347. ----------------------- In the audience situation, however, male speakers obtained greater shifts than did women. Most of the studies described above deal primarily with the immediate effects of prestige, credibility, and other ethical elements. Hovland and his associates, however, have investigated the temporal effects of the source upon persuasion. In one of these experiments Hovland and Weiss held all of the message elements constant except for factors which produced an impression of high credibility for one source and low credibility for another. The subjects exposed to the former stimulus shifted in significantly greater numbers on immediate post tests of attitude than did those receiving the message with low credibility. Over a period of one month the favorable effect, however, decreased, and the subjects exposed to the "inferior" source moved toward agreement with the attitudes expressed in it. Hovland postulated a "sleeper effect"-that in the absence of further stimuli agreement with high credibility sources decays while agreement with low credibility sources grows. The possible explanation is that the subject forgets the source but retains the information and the essential arguments." In a specific test of the sleeper hypothesis, Kelman and Hovland found that a high ethos source, who was rated significantly fairer, better qualified to speak, and of sounder judgment than a supposedly low ethos source, produced significantly greater attitude shifts. Over a three-week period, however, the extent to which subjects agreed with the positive source decreased significantly, and the extent to which they agreed with the negative source increased non-significantly. Reinforcing the recall of the sources by playing back the introductions of the tape-recorded messages produced greater agreement with the high prestige speaker and less agreement with the one of low ethos in an experimental group than occurred in a control group which received no repetition of the stimuli." In a variation of the above approach Weiss determined that a group exposed to a low credibility source showed less regression toward its original attitude than did a group exposed to a high credibility source." Also supporting the sleeper effect is the finding that over a period of time those who originally disliked a communicator became slightly more positive to ward him while those who had originally liked him became slightly less favorable (non-significant). The results of Duncker's study of the effect of prestige suggestion upon children's food preferences also confirm the Hovland sleeper effect findings in respect to both the decline of the effect over time and the renewal of strength following reinstatement. STUDIES ASSUMING THAT ETHOS IS VARIABLE Diverse as the studies discussed above appear to be, they share a common model of ethos--that is, they are all based on the assumption that the speaker's image is relatively fixed throughout the period of communication. In sharp contrast with this view is the ethical model based on a congruity principle enunciated by Osgood. Intended to explain many psychological functions, the congruity principle holds that an image (or meaning) depends upon the other concepts with which it is associated and thus is subject to perpetual change. Among the factors causing these variations are the successive parts of the message. -------------------------- 58. Carl Hovland and Walter Weiss, "The Influence of Source Credibility on Communication Effectiveness," Public Opinion Quarterly, XVI (1961), 635-650. 59 Loc. cit. 60 Loc. cit. 61 Arthur Cohen, "Need for Cognition and Order of Communication as Determinants of Opinion Change" in Order of Presentation, eds. Carl I. Hovland et al. (New Haven: Yale University Press, 1957), pp. 79-97. 62 Loc. cit. 63 Osgood, Suci, and Tannenbaum, loc. cit.; Charles Osgood and Percy Tannenbaum, "The Principle of Congruity in the Prediction of Attitude Change," Psychological Review, LXII (1955), 42-55. ---------------------- Drawing upon this generalized congruity hypothesis, Tannenbaum formulated predictions of attitude change toward communication sources and then compared these estimates with the results obtained when college students were ex posed to written messages. Since the correlation was .91, the conclusion is that attitude changes of the college students in this experiment conformed to the congruity hypothesis. A study of the same hypothesis applied to public speakers showed that the congruity model predicted changes in attitude somewhat better than chance alone." This study, however, failed to produce the goodness of fit observed in the Tannenbaum experiment." Bettinghaus hypothesized that the difference between these results was caused by the presence of a greater number of elements in the cognitive structure for oral than for written messages. Extending the congruity model to four elements-speaker, central proposition, speech composition, and delivery--he obtained results which fit his extended model significantly better than they do the two-element model (speaker and central proposition) employed in the earlier experiments." GENERATING OR CHANGING ETHOS Unlike the studies discussed in the preceding section, which typically attempted to assess the utility of a presumed or measured ethos, the experiments discussed below are concerned with the means of generating or altering a receiver's image of a communicator. These efforts, in general, fall into two categories: those which tried to establish extrinsic ethos by techniques employed before the message itself began, and those which attempted to create intrinsic ethos by techniques employed by the speaker during the presentation." ----------------------- 64. Percy Tannenbaum, "Initial Attitude Toward Source and Concept as Factors in Attitude Change Through Communication," Public Opinion Quarterly, XX (1956), 413-425. 65. David Berlo and Halbert Gulley, "Some Determinants of the Effect of Oral Communication in Producing Attitude Change and Learning," SM, XXIV (1957), 10-20. 66. Compare the results of Berlo and Gulley with those of Osgood, Suci, and Tannenbaum, P. 212. 67. Loc. cit. 68. Extrinsic ethos is the image of the speaker as it exists prior to a given speech. Intrinsic ethos, comparable to Aristotle's artistic ethos, is the image derived from elements during the presentation of the speech, consciously or unconsciously provided by the speaker. In real life speech situations, the final ethos is a product of the interaction of extrinsic and intrinsic ethos. ------------------- EXTRINSIC ETHOS The following experiments deal with the generation or the modification of a communicator's image by stimuli which are not part of the actual presentation. Since the ethos of the individual depends in part upon the reputation of the group to which he belongs, experiments concerning the alteration of group images are relevant to the concept of ethos. One such experiment showed that very short speeches produced immediate attitude changes in favor of either China or Japan but that over a five-month period significant regression occurred toward the original attitudes. In a similar experiment Roman Catholic school children were found to be quite persuasible to some but not all items in propaganda covering a wide range of topics. Other conclusions were that attitudes toward well known individuals seemed about as subject to change as other attitudes and that reactions toward groups outside the students' immediate experience seemed especially subject to the influence of propaganda.' Closely related to the question of changing attitudes toward individuals is that of building an image. Annis and Meier set out to create an image of an un known source through planted editorials which linked the source with certain opinions and actions. The experimenters assumed that they could predict whether the subjects of the experiment favored or opposed these opinions and actions. As few as seven planted editorials generated the desired image, and most of the effects persisted over a period of four months." Berlo and Kumata studied the effect of a dramatic allegory, "The Investigator," in modifying images. Attitudes toward Joseph McCarthy, the subject of the satire, tended to become more favorable, while attitudes toward the source (the Canadian Broadcasting Company) and toward Congressional committees became significantly less favorable. The experimenters felt that the extreme one-sidedness of the presentation may have caused these "boomerang" effects." Using a single tape-recorded speech, Kersten compared two introductions, one of which employed techniques estimated by experts to focus attention on the speaker and his subject and to build the speaker's prestige and the other of which did not. The persons hearing the speech with the favorable introduction changed opinion significantly more than did those who heard no introduction or the poor one." The confounding involved in the simultaneous manipulation of prestige and attention-focusing elements makes it impossible to conclude that the enhanced prestige of the speaker was the source of the observed difference. In deed, Pross found that an introduction stressing the character, the reputation, and the intelligence of the speaker added little to the persuasiveness of either "ethical" or "non-ethical" forms of a speech.74 Neither Kersten nor Pross actually measured differences in ethos; they assumed that different introductions would affect the variable. The same is true of Highlander's experiment, which seems to show that variable levels of authoritativeness of the speakers do not affect either the likeableness of radio programs or the amount of information gained from them.75 In all such studies it is possible that the experimental treatments failed to take effect in the supposed manner. ------------------------ 62. William Chen, "The Influence of Oral Propaganda Material upon Students' Attitudes," Archives of Psychology, XXIII (1933) ; "Retention of the Effect of Oral Propaganda," Journal of Social Psychology, VII (1936), 479-483. 711 Wegrocki, loc. cit. 71 Albert Annis and Norman Meier, "The Induction of Opinion Through Suggestion by Means of 'Planted Content,'" Journal of Social Psychology, V (1934), 65-81. 72 David Berlo and Hideya Kumata, "The Investigator: The Impact of a Satirical Radio Drama," Journalism Quarterly, XXXIII (1956), 287-298. 73 Loc. Cit. ----------------------- Andersen constructed three introductions designed to establish varying levels of prestige and authoritativeness for speakers dealing with the farm problem. His conclusions were these: (1) Students perceived significant differences be tween a college student and a Professor of Agriculture or a Farm Extension Agent on two scales: (a) the evaluative and the dynamism dimensions of a se mantic differential designed to measure ethos; (b) authoritativeness as estimated by a Likert-type scale. (2) The expected differences between the professor and the extension agent did not result except on the authoritativeness scale. (3) The more rhetorically sophisticated students seemed to perceive differences in ethos that the rhetorically naive students did not. (4) There was no proof that the variations in ethos and authoritativeness affected persuasiveness. A speech of introduction, one should note, creates special theoretical problems; for if the audience image of the introducer is low, this attitude through transfer may affect the ethos of the speaker. For instance, at the time of this writing, a laudatory introduction of a political candidate in the United States per formed by James Hoffa or Fidel Castro might prove a serious detriment to persuasiveness. Since less obvious factors may also affect the experimental situation, it is conceivable that ethos may be more sensitive to such unforeseen and uncontrolled variables than it is to the verbal content of the introductions. --------------------- 74 Loc. Cit. 75 Loc. Cit. 76. Kenneth E. Andersen, "An Experimental Study of the Interaction of Artistic and Non-artistic Ethos in Persuasion," unpubl. diss. ( Wisconsin, 1961). 77. Intrinsic ethos is defined in this study as the image of the speaker which is generated during the presentation of the message. ------------------------------ INTRINSIC ETHOS That changes in ethos result from hearing speeches seems clear from a study of the effect of a campaign speech by Thomas E. Dewey. Comparing ratings obtained before a speech with those recorded immediately afterwards, Thompson found that students raised their estimation of Dewey as a public speaker but did not change their opinions significantly concerning the soundness of his ideas and his acceptability as a candidate." Studies which have altered the presentational elements may be divided into those which have manipulated characteristics of the manuscript and those which have altered such non-manuscript stimuli as the speaker's appearance or his style of delivery. A common type of study is the comparison of the effect of presenting both sides with the effect of giving but one-a distinction which seems to the writers to be ethically significant." In one such investigation Hovland, Lumsdaine, and Sheffield found (1) that the "both sides" presentation was significantly more effective for subjects with a high school education when the weight of evidence clearly supported one side; and (2) that a one-sided presentation was more effective with subjects initially favoring the advocated view and with subjects who had not completed high school." Similarly, Paulson's experiment involved two speeches, one of which omitted opposing arguments and the other of which made the barest mention of them. Opinion changes did not differ significantly, but the "both sides" speech was significantly superior in respect to the amount of information which was obtained. Shanck and Goodman also tested reactions to propaganda which presented equal amounts of argument on both sides or one-sided pro or con arguments. --------------------- 78. Wayne Thompson, "A Study of the. Attitude of College Students Toward Thomas E. Dewey Before and after Hearing Him Speak," SM, XVI (1949), 125-134. 79. The presentation of both sides of an issue is often treated as one aspect of ethical proof. The practice also has logical connotations. It is possible to consider the impact of the treatment of both sides on the image of the speaker and the impact of this image on persuasiveness as distinct from the logical value of the treatment and the resultant persuasiveness. 80. Carl Hovland, Arthur Lumsdaine, and Fred Sheffield, Experiments on Mass Communication: Vol. III of Studies in Social Psychology in World War II (Princeton: Princeton University Press, 1949). 81. Loc. cit. 82. R. C. Shanck and Charles Goodman, "Reactions to Propaganda on Both Sides of a Controversial Issue," Public Opinion Quarterly, III (1939), 107-112. 83. See the studies previously cited by Sikkink, by Cole, and by Gilkinson, Paulson, and Sikkink. ------------------------- That no significant difference was observed might be explained by the extreme subtlety of the propaganda. Another rhetorical element which is sometimes held to carry ethical implications is the use of authority and citations of source. Three studies described earlier in this paper reported that the inclusion of authority did not increase persuasiveness. Cathcart presented four versions of a speech with variations from form to form in respect to the amount of specific evidence and documentation. He found that the forms which supported but did not document contentions and which supported, documented, and specified that the sources cited were experts produced significantly greater shifts at the five percent level than did the form which merely supplied generalizations. A fourth form which supported the assertions and documented fully but did not say that the cited sources were experts was not significantly more effective than the one which merely supplied generalizations. That such differences as were observed were attributable to non-ethical considerations is suggested by the finding that none of the speeches differed in terms of the audience's evaluations of the speaker's competence, enthusiasm, or clarity of ideas. Ludlum constructed a speech in which he incorporated several elements de signed to increase the credibility of the source. His techniques include the acknowledgment of opposing arguments, "leading thoughts rather than forcing," showing alleged facts to be consistent with known facts, showing material to be recent, and manifesting a "high degree of credibility" by means of self-praising statements. Comparing the persuasiveness of this speech with that of a "straight argumentative" address, he found the latter to be more effective." Since he did not measure received ethos, the effect of the variables in the non-argumentative speech is unknown. Moreover, since all of the variables were incorporated in a single speech, it is impossible to isolate the effect of any one of them. If some of the techniques produced positive effects and others acted negatively, the effects may have counterbalanced one another. Thirdly, some of the self-praising statements in the non-argumentative speech may have had an effect quite different from that intended. Finally, argumentative technique may have an ethical dimension for college students, such as those whom Ludlum employed, with the result that the argumentative talk may well have produced a more favorable speaker image than did the speech employing an assortment of "conciliatory" techniques. The experiment by Ludlum points up the importance of specifying carefully any differences in content between speeches intended to produce high credibility and those against which their effects are to be compared. This same consideration applies to an early experiment by Pross, who constructed four forms of a speech on a single topic. Two of these employed techniques of "ethical appeal" (as judged by speech experts) and the other two did not. Length was kept constant. The interpretation of Pross' non-significant findings is difficult, for matching the lengths necessitated the removal of material in order to make room for the ethical elements. As a consequence the two ethical speeches had almost no logical structure. --------------------------- 84 Loc. cit. 85 The problem of separating the logical and the ethical effects of the same complex stimulus is again at issue. The writers believe that a complex stimulus may affect both logical and ethical proof and perhaps pathetic proof as well. 86 Thomas Ludlum, "A Study of Techniques for Influencing the Credibility of a Communication," unpubl. diss. (Ohio State, 1956). 87 Loc. Cit. ------------------------ This investigation and other studies indicate a confusion in the use of the terms ethos and ethical. On the one hand, these terms are used to refer to the audience's image of the speaker, as when it is said that Parran is more credible or higher in ethos than is Dennis; on the other hand, certain types of speech con tent are labeled ethical appeals. For example, a speech which employs many self references and conciliatory elements is described as higher in ethos content than an address which follows a straightforward proposition-and-proof format. Usually, when rhetoricians classify a speech content element as "ethical," they seem to mean that the elements seem to the classifier to be calculated to gain the good will of the audience or to enhance the speaker's ethos. In our present state of knowledge concerning audience response, such a judgment is at best only an educated guess. Therefore, when the results of the Pross and the Ludlum studies are cited in support of the proposition that ethical speeches are no more effective in inducing attitude change than are logical speeches, it should be specified very carefully that the results are based upon analysis of speech content and not upon the image of the speaker which the audience holds. The present writers as rhetorical critics believe that some of the Pross and Ludlum "ethical" speech techniques probably had decidedly negative effects on the ethos of the speaker. The basis of this judgment, of course, is intuitive, not empirical. -------------------- 88 Loc. cit. 89 Loc. cit. ------------------- The message which an audience receives during a speech obviously involves more than verbal (manuscript) stimuli. Several studies indicate that nonverbal factors produce audience judgments concerning the speaker. Haiman found (1) that an audience rated a graduate male speaker higher in competence than it did an undergraduate male and two females; (2) that with content held constant, graduate speakers obtained higher rates of fair-mindedness, sincerity, and likeableness than did undergraduates; (3) that in two experiments shifts of opinion within the audience were correlated positively with the speaker's competence ratings and with nothing else; and (4) that although variations in ratings of likeableness and physical attractiveness could he produced through changes in appearance and demeanor, significant changes in attitude did not result. Many of the variables in the Haiman study are those associated with differences in social status. Harms has shown that, regardless of their own position, listeners in general assign high credibility to speakers of high social status and low credibility to those of low status. Such judgments occur even though the stimulus is nothing more than a short tape-recorded sample of speech. The Harms study further shows that listeners can discriminate class differences with rough accuracy and that they identify the low status speakers somewhat more readily than they do those of superior background. Consistent with these results is the experimental finding that audiences may construct relatively complete assessments of a speaker's personality and physical characteristics on the basis of his voice. Other conclusions to this study were that personality, physical characteristics and occupation were likely to be perceived correctly, that consistency of response (right or wrong!) was a stronger tendency than accuracy of judgment, and that gross psychological characteristics were judged more accurately than physical features. These findings suggest the plausibility of the "truth-will-out" theory regarding the action of subliminal, nonverbal stimuli upon the ethos of the speaker. As the theory goes, an insincere speaker's sophistry will betray itself through unconscious behaviors which act subliminally upon the auditors. An experiment by Hildreth, however, offers no confirmation for this hypothesis. Defining sincerity in terms of the speaker's expressed preference for one side of a controversial issue and using a large number of speakers who filmed speeches on both their preferred and their non-preferred sides, he discovered that audiences were unable to distinguish the sincere from the insincere speeches and that the ratings of the two types of speeches did not differ significantly in effectiveness. Rather, ratings of effectiveness and of estimated sincerity were positively correlated." Unfortunately, methodological considerations render the results of the experiment inconclusive. Since the "sincere" speech was composed, practiced, and delivered first in all instances, the time allowed for composition was very brief, and the making of a film was presumably unfamiliar to a majority of the speakers, a number of factors were operating to enhance performance in the "insincere" presentation as contrasted with the "sincere" one. Indeed, the role which subliminal perception may play in the establishment of ethos has been little clarified by experiments. Drawing upon the "hidden persuader" approach, Steiner found that placing visually superimposed words on a screen at subliminal intensity levels did not alter either the effectiveness of a filmed speech or the judgment of the sincerity of the speaker. Combining prior and intrinsic elements, Strother attempted to study a combination of factors. The addition of ethical techniques either singly or in combination did not significantly increase the persuasiveness of a low ethos source. However, as measured by a hostility scale, the combination of elements apparently surpassed a control speech in allaying hostility toward the low ethos source. In the control presentation neither conciliatory nor special introductory techniques were employed." --------------------- 90. Gordon Allport and Hadley Cantril, "Judging Personality from Voice," Journal of Social Psychology, V (1934), 37-55; also in Hadley Cantril and Gordon Allport, The Psychology of Radio New York: Harper and Row, 1935). 91. Richard Hildreth, "An Experimental Study of Audiences' Ability to Distinguish Between Sincere and Insincere Speakers," unpubl. diss. (Southern California, 1953). 92. George Edward Steiner, "An Experimental Study of the Influence of Subliminal Cue Words on an Audience's Perception of a Filmed Speaker's Sincerity, Effectiveness, and Subject Matter," unpubl. diss. (Southern California, 1959) ; Silf, XXVII (1960), 93-94. 93. Loc. cit. ----------------- In another investigation of combinations of variables Andersen used two tape-recorded speeches, both of which were attributed to three sources described in tape-recorded introductions. The principal results were these: (1) Despite great manuscript variations which speech experts predicted would produce different levels of ethos, the only significant differences between the two speeches were those measured on a dynamism scale. (2) The elements of artistic and inartistic ethos did interact significantly in producing the final image of the speaker. (3) The variations in ethos did not cause a significant difference in persuasiveness." MEASUREMENTS OF ETHOS AND ATTEMPTS TO ASSESS THE RELATIVE DEGREES OF ETHOS In a few instances the development of a measure of ethos has been the main goal of a research project, but more often the measurement of prestige, credibility, or some other ethical component has been ancillary to the study of such presumed results of ethos as preferences, attitude change, and information gain. The methods of measurement in both types of investigation are the same: (1) rankings, (2) socio-grams, (3) "prestige indexes" obtained from attitude change data, (4) linear rating scales, (5) Thurstone-type attitude scales, and (6) devices similar to Likert scaling techniques, including the semantic differential. Perhaps the most elementary method of determining differences among sources in respect to prestige, credibility, likeableness, etc., is to require subjects to arrange the sources in rank order. Sherif, for example, presented a list of six teen authors to a group of undergraduates and asked them to rank the authors according to personal preferences for their writings. A month later the subjects were told to rank sixteen passages in respect to literary merit. Since all of the passages had been written by a single author not included in the list and since literary experts had judged all of them to be of equal merit, the only variable was the false attachment of a different author's name to each excerpt. Correlations between the two sets of ranks were held to represent the effects of "prestige." The replication of the study with similar results in three instances indicates the usefulness of the rank-order technique for simple experiments of this type. The method was to determine the rank order for individuals, to compute rank correlations for individuals, and to draw conclusions from the average correlations. While this technique seems justified, the rank-order method employed by Michael, Rosenthal, and DeCamp was not. In an effort to discredit the "constant stimulus" theory of prestige, these authors worked with mean and median ranks "6-statistics which are generally meaningless. --------- 94 Loc. cit. 95 Loc. cit. 96 Loc. cit. --------------- Cole demonstrated the possibility of using socio-metric data for the determination of certain characteristics of ethos. Using a particular personal characteristic (judgment, personal appeal, etc.) as the basis for socio-metric choices, he selected one or more members of a group as "stars" and then assumed that they were more highly regarded than their colleagues. Under some conditions, these preferred members were as persuasive as authorities from outside the group." Kulp apparently made the first attempt to develop an index of prestige based upon attitude change. In a classic design which was to be repeated with variations many times during the ensuing years, he first administered Harper's test of liberalism to more than three hundred graduate students at Columbia. Later, various subgroups were told that the responses supplied them had been written by social scientists, educators, and other learned persons. The relative amounts of attitude shift toward each of these sources was used as the basis for computing a prestige index for each of the several professional groups. Bowden, Caldwell, and West replicated the essential features of Kulp's study in an experiment using junior high, high school, and college students as subjects and employing a variety of different prestige levels. Sample findings with respect to the economic problems considered were these: "Prestige of the educators seems to increase as progress is made up the educational ladder" and "Ministers received the lowest rank in every case." Underlying these measuring techniques is the assumption that the prestige of a source is directly proportional to the ability to produce attitude shift. In 1938 Lurie formalized this point of view when he defined prestige as "The change in scale value of certain items brought about by attaching the name of the symbol to these items." He obtained scale values for prestige by administering a test of attitude without attaching prestige labels to the items, by administering the same test two weeks later with prestige labels attached, and by then subtracting the scores on the first test from those on the second. The remainder was the index of prestige.'" Naturally, prestige measures obtained in this manner are not pure or independent measures of the variable. Moreover, to use any of these measures to test the hypothesis that prestige induces attitude change is impossible, for the mea sure of prestige is attitude change. In an effort to develop an independent index suitable for testing this hypothesis, Saadi and Farnsworth combined gross ratings of "like," "indifferent," and "dislike" by the formula 100 [(L %I) (L I + D)] to obtain a score for likeableness based on group data.'" --------------------- 97. Loc. cit. 98. Daniel Kulp, II, "Prestige, as Measured by Single-Experience Changes and Their Permanency," Journal of Educational Research, XXVII (1934), 663-672. 99. A. O. Bowden, Floyd Caldwell, and Guy West, "A Study in Prestige," American Journal of Sociology, XL (1934), 193-203. 100. Walter Lurie, "The Measurement of Prestige and Prestige-Suggestibility," Journal of Social Psychology, IX (1938), 219-225. 101. Loc. cit. --------------------------- The multiple-choice aspect of the Saadi-Farnsworth measure was an early precursor of an obvious means of measuring various aspects of ethos-the rating scale. An early experimenter with this type of measurement was Lorge, whose subjects rated seventy sources on a five-interval scale ranging from "those individuals whose opinions you respect most" to "those individuals for whose opinions you have least respect." More recently, Hovland and Weiss employed a five-point linear scale of "trustworthiness" to evaluate the credibility of two sources." The well-known study by Haiman used a variety of scales. In one phase of his experiment two nationally prominent public figures were evaluated on nine point scales of reputation and competence. In other parts of the investigation student speakers were rated on similar scales for the qualities of sincerity, fair-mindedness, physical appearance, conceit, competence, and likeableness. In addition to being one of the first experimental research workers to recognize explicitly the multidimensionality of ethos, Walter made the earliest effort to apply recognized test construction methods to the problem of creating a measuring device. His specific project was the development of an instrument to measure a single factor, the evaluation of character. Beginning with nearly 400 character describing statements and employing both the Thurstone sorting techniques and the Seashore rating methods, he developed two tests of twenty-two items each. ----------------- 102. Loc. cit. 103. Loc. Cit. 104. L oc. cit . 105. Otis Walter, Jr., "The Measurement of Ethos," unpubl. diss. (Northwestern, 1948). 106. Charles Osgood and Ross Stagner, "Analysis of a Prestige Frame of Reference by a Gradient Technique," Journal of Applied Psychology, XXV (1941), 275-290. 107. Loc. Cit. 108. Loc. Cit. 109. Loc. cit. ----------------------- When applied to such individuals as Franklin Roosevelt and "The person with the best character I have known," the two forms of the test were normally distributed, distinguished among intuitively perceived gross character levels, and correlated well (.86) with each other. Applied to two speakers in the classroom, the two forms correlated extremely well (.96). The Osgood and Stagner use of bipolar nouns in a set of scales to rate occupations and occupational groups was a forerunner of the semantic differential technique. They found that the prestige of jobs and workers could be determined through the use of their scales. Although Walter asserted the multidimensionality of ethos and although Haiman's technique actually employed a poly-dimensional approach, until recently no practical way of employing multivariate measures of ethos in research seemed to exist. Now the semantic differential technique makes such research possible. Berlo and Gulley, Berlo and Kumata, and Bettinghaus used the differential to measure attitude toward the communicator, but in each in stance they reported only one dimension of the semantic space, the evaluative aspect of the image. "Although it does not tap much of the content of an attitude in a denotative sense . . . it does seem to provide an index to the location of the attitude object along a general evaluative continuum." Employed in this manner, the semantic differential is similar in many ways to a traditional Likert scale in which a number of judgments concerning the concept are rendered on a linear scale and the sum of the scale values recorded by the subjects is used as a more-or-less uni-dimensional measure of the single property with which the scale is concerned. Andersen developed a semantic differential which was specifically designed to measure ethos. Employing terms garnered from theoretical and experimental literature and securing responses to famous living people from freshmen engineering and physical education students, he obtained two major dimensions (evaluative and dynamism) in the images."' Berlo carried out a similar study, but he used a greater number of concepts and more students than did Andersen. Berlo also employed an oblique solution, whereas Andersen's method was the orthogonal factor solution. Inspection suggests that the two structures were not essentially dissimilar if allowance is made for the difference in the factor rotation methods. SUMMARY Despite the great number of experimental studies relevant to ethos, the scope of this concept is such that the findings are not yet sufficiently numerous and sophisticated to permit definitive conclusions about the operation of ethical proof. The finding is almost universal that the ethos of the source is related in some way to the impact of the message. This generalization applies not only to political, social, religious, and economic issues but also to matters of aesthetic judgment and personal taste. Some evidence even shows that "prestige-suggestion" can affect the appetite for certain foods and can influence performances of perceptual and psychomotor tasks. On the other hand, there is not enough evidence to suggest that the amount of information gained from exposure to a message is related to the ethos of the source-at least this lack of relationship seems to be true of college populations. The effect of ethos, again according to many studies, has a temporal dimension. In other words, when the stimulus is not renewed, material presented by a high ethos source loses in persuasiveness and that given by a poor source gains. Recall of the source reestablishes some of the initial effect, but the improvement which renewal produces decays more rapidly than does the original increment. Some auditors appear to be more susceptible to ethical appeal than others; some may be contrasuggestible. However, there is no evidence to show that suggestibility to prestige correlates well with intelligence, education, speech training, subject-matter competence, age, or sex. The only variable which seems clearly related to differences in suggestibility to prestige is the initial attitude toward the topic or the purpose: consistently, those who are neutral initially shift more often than do those who are at one extreme or the other. ------------------ 110. Osgood, Suci, and Tannenbaum, p. 195. 111. Loc. cit. 112. David K. Berlo, "An Empirical Test of a General Construct of Credibility," unpubl. paper presented at the SAA convention, New York City, December 29, 1961. ---------------------- Research shows that expert opinion may be about as influential as majority opinion in inducing attitude change. While most experimentation has been conducted in a fixed ethos model, re cent research shows that a congruity model can be used to predict attitude change toward both a communicator and his topic. Incorporating elements concerning speech composition and delivery increased the usefulness of the model. Printed and oral propaganda can succeed in creating and altering images of groups or of individuals, but attempts to produce unfavorable reactions to individuals may backfire. When this response occurs, the prestige of the criticized person may increase and that of the attacker may decline. Speeches of introduction probably influence the image of a speaker, but most of the evidence on this point is indirect. Certain characteristics of speech affect the ethos of the speaker. No evidence, however, supports the common beliefs (1) that giving "both sides" is a superior way to present controversial material, (2) that citing the sources of evidence increases persuasiveness, and (3) that including conciliatory remarks, statements of self-praise, and other conscious, obvious attempts at ethical appeal enhances the speaker's status. Such non-content stimuli as dress, voice, and manner apparently affect the attitude of the audience toward the speaker, but these factors may not be related to persuasiveness on a given occasion. There is no evidence that the audience can perceive lack of sincerity; rather, audiences appear to react to their evaluations of the competence of the speaker. Many techniques of measurement have been applied to ethos: among these are ranking, socio-grams, prestige indexes, linear rating scales, Thurstone scales, and the semantic differential. Eaph of these has proved useful in assessing one or more of the aspects of ethos. This preceding body of findings suggests certain possibilities for future re search: 1. The dimensions of ethos should be explored through multivariate analysis in terms of different auditors, different speakers, and different speech situations. New measurement techniques, and especially the semantic differential, make this type of research possible. 2. Ethos or ethical proof should be measured in experiments designed so that this variable is not confounded with persuasiveness. 3. The effect upon ethos of the interaction of prior reputation and the artistic elements in the message should be studied. Findings in this area would be of great importance to rhetorical theory. 4. Some research suggests that differences in ethos are not established as easily with some audiences as previous experimenters often assumed. More research dealing with the methods of establishing and modifying ethos is needed. 5. The effect of variations in auditors, situations, and topics upon the function of ethical proof in persuasion should receive renewed attention. The utilization of improved designs and measuring devices can create experimental conditions that may lead to more meaningful results than those obtained in the past. Also in Part 4:
|
Prev. | Next Top of Page | Index | Home |