Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
To cite this article: Jan Bosman & Louk Hagendoorn (1991) Effects of Literal and Metaphorical
Persuasive Messages, Metaphor and Symbolic Activity, 6:4, 271-292, DOI: 10.1207/s15327868ms0604_3
To link to this article: http://dx.doi.org/10.1207/s15327868ms0604_3
In the first part of this study, the relative effectiveness of metaphorical and literal
persuasive political messages is investigated. Participants read a literal or a
metaphorical description of a racist political party and subsequently filled out a
questionnaire about this party's policy and the ways in which to counteract it.
Results suggest that, although metaphors do influence the participants' ideas,
literal messages are more effective. In the second part of the study, we find that
persuasive messages do not enhance agreement with statements that are congruent with the message but, rather, suppress agreement with incongruent statements. These results are discussed in the context of persuasion theory.
In 1964, in an article on language intensity, J. W. Bowers provided the starting
point for a number of experiments on the persuasiveness of metaphors. Language intensity is the degree to which the attitudes of the speaker are reflected
by the language used, and metaphors, according to Bowers, are the most
powerful device with which to reflect these attitudes. The experiments following Bowers's study indicated that metaphors can facilitate persuasion. However, all of these studies concentrated on "intense" metaphors: metaphors that
are primarily emotional at the expense of cognitive content (e.g., Siltanen,
1981: "Marihuana can screw up brain functions"). Moreover, the experimental
designs were focused singularly on the detection of changes in the degree of
favorableness toward the topic, thereby precluding the discovery of cognitive
metaphor effects unrelated to favorableness. An example of this is a message
claiming that "Peter is a fox," followed by a measure of how positive or
negative the subject feels about Peter and not whether he considers Peter to
be cunning.
Contrary to the aforementioned, Bosman (1987) conducted an experiment
specifically designed to assess the cognitive effects of metaphors: Subjects read
Requests for reprints should be sent to Jan Bosman, Instituut Voor Massacommunicatie,
Katholieke Universiteit Nijmegen, Postbus 9108, 6500 HK Nijmegen, The Netherlands.
272
BOSMANANDHAGENDOORN
273
comparing two metaphor conditions, the results inevitably express the differences between the effects of these two metaphor conditions. Effects that both
metaphor conditions have in common go unnoticed, and possible effects of
Metaphor A on the implications of Metaphor B are erroneously attributed to
Metaphor B. Even though it does not seem likely that the metaphors affect
agreement with each other's implications, it is more appropriate to assess the
effects of the metaphors by comparing the metaphor conditions with a "neutral" condition in which the subjects have not been exposed to some sort of
persuasive message.
The present study is an extension of Bosman (1987). In Part 1 we address
the problem of the effectiveness of metaphorical language as compared to the
effectiveness of suitably designed literal paraphrases. We do this by expanding
Bosman's two-factor design (two different metaphors applied to two political
parties) into a three-factor design in which the third factor is metaphorical
versus literal language. In effect we compare the results of an exact replication
of Bosman with the results of a replication in which the metaphors have been
substituted by literal paraphrases. To ensure maximum comparability, we
employ Bosman's original design, criticized earlier because of the odd chance
that it might conceal any shared effects of the metaphorical messages. The
inadequacies of this design are remedied in the second part of this study, where
the effects of the metaphorical and the literal messages are assessed with the
aid of neutral conditions in which the subjects have not been exposed to a
persuasive message.
The aim of the first part of this study is to compare the persuasiveness of
literal and metaphorical messages. In addition, we explore the persuasibility
of different categories of respondents with respect to both types of messages.
We discuss the differential persuasibility of our respondents at some length as
this discussion leads up to a paradox that is subsequently solved in Part 2. Our
objective in the second part of this study is to obtain a more accurate assessment of the persuasive effects of literal and metaphorical messages.
274
a Dutch pro-Nazi party that existed before and during World War 11. The
written introduction of the questionnaire, explaining our motives for conducting the survey, contained a quotation expressing an unspecified politician's
views regarding the CP or the NSB. This quotation was intended to influence
the subjects' ideas about the CP or the NSB. In line with Bosman (1987), one
of two metaphors was used to describe either the CP or the NSB, thus creating
a 2 X 2 design.
In addition, literal paraphrases of the metaphors were constructed, extending the design into a three-factor design in which the effects of two different
metaphors or their literal paraphrases on two different political parties were
investigated.
Independent Variables
275
makes your car run but not for the conclusion that it will make your car run
faster than other brands. To obtain the last type of conclusion, we need some
brand X to compare our Gasoline A to, and preferably a brand with some
prestige, as the validity of the comparison depends on the quality of brand X.
Hence we devised our brand X with two criteria in mind: (a) The paraphrases
should be plausible paraphrases for our metaphors, and as our research is
focused on the effects of the metaphors on the audiences' cognitions (b) the
paraphrases should have (as much as possible) the same implications as the
metaphors.
Starting out with the core metaphors of the metaphorical quotations, "The
CP (or NSB) should be unmasked" was paraphrased as "The CP (NSB)
presents itself as better than it is," and "i9e CP (NSB) sproutsfrom the fertile
soil of adverse societal conditions" was paraphrased as "The CP (NSB) is
caused by adverse societal conditions." Extended versions of the paraphrases
were used as the literal quotation (see the Appendix).
Dependent Variables
The dependent variables consisted of the cognitive implications of the metaphors. The mask metaphor, characteristic of the comments from the VVD
upon the unexpected electoral gain of the CP, is inspired at least partially by
a moral indignation about the CP and consequently should be appreciated for
its "intense" qualities. On the other hand, "unmasking" the CP involves more
than mere moral indignation. The inferences that can be drawn from the
metaphor may reveal some of the VVD's political ideas. For instance, the mask
metaphor implies a distinction between a real and sinister CP and its presumably inoffensivepublic face, thus enabling the W D to condemn the CP without passing sentence onto the publicly visible elements of the CP's policy. The
mask metaphor also implicitly absolves those who voted for the CP as apparently having been attracted and deceived by the party's mask. Moreover, the
mask metaphor denies the importance of societal problems as the 'ffertilesoil"
of the CP's "growth," the very essence of the metaphor used by the leftoriented parties.
It is not the goal of our research to determine whether these and other
inferential possibilities indeed have guided the VVD or the left-oriented parties
into their choice of metaphor. Although it is not unlikely that these metaphors
provide a concise picture of each party's ideas about the CP, our interest is
not in the metaphors' ability to reflect the speaker's ideas but in their ability to
transmit these ideas to an audience. Therefore, all inferences that could be
derived from the mask and the soil metaphors were collected, and our questionnaire contained these inferences as statements about the CP or about the
NSB. A questionnaire item derived from the mask metaphor is: "The goals
publicly advocated by the CP (NSB) are, as such, not all that bad." An item
276
BOSMANANDHAGENDOORN
derived from the soil metaphor is: "It doesn't make any sense to fight the CP
(NSB); that would be nothing more than fighting a symptom."
The questionnaire contained 13 items derived from the mask metaphor, 10
items derived from the soil metaphor, and a number of filler items. The aptness
of these items as derivations from the metaphors was evaluated by a panel of
10 judges (see Bosman, 1987, for more details).
Subjects had to indicate their agreement with these items on a 5-point scale,
ranging from complete agreement (1) to complete disagreement (5). The sum
of the mask items, the sum of the soil items, and the difference between these
two scores were used as an indication for the effects of the metaphors and their
literal counterparts. The sum of the mask items must be lower (indicating more
agreement with these items) for the subjects in the mask conditions, the sum
of the soil items lower for the soil conditions, and the difference score (Sum
Mask - Sum Soil) should be lower for the mask conditions. Obviously, the
difference score does not add any new information but expresses the joint effect
of the other two variables.
Subjects
Three hundred forty-six university students (142 psychology students and
204 law students) participated. The number of subjects in the eight experimental conditions ranged from 40 to 46.
Subjects were contacted through their courses in the beginning of 1987
and asked to participate in the experiment on a voluntary basis. Although
psychology students at our university are required to earn a number of credits by participating in research, only 60% of the psychology students were
willing to participate. More than 90% of the law students, who did not
receive any type of compensation for their participation, filled out the questionnaire.
The mean age of the subjects was 22.7 years, and 60% of the participants
were women. As in Bosman (1987), political preferences of the psychology
students were predominantly left oriented. The political preferences of the law
students were more equally distributed along the political continuum. Consequently, the total sample contained a considerable number of right-oriented
respondents (37%). Approximately 72% of the respondents indicated the CP
as the political party they disliked most.
Results
Our main interest in the first part of this article is in the relative effectiveness
of metaphorical and literal messages as instruments of persuasion. We specify
the effects of both types of messages in regard to respondents' gender and
political affiliations.
Variable
Condition
CP
Main
Effect
NSB
1;
df
Metaphorical Conditions
Sum mask
items
Mask
Soil
37.2
37.4
36.6
37.4
Party
Message
0.1
0.3
1
1
.74
.28
Sum soil
items
Mask
Soil
27.2
26.5
28.4
27.0
Party
Message
1.7
2.6
1
1
.20
.05
Difference
score
Mask
Soil
10.0
10.9
8.3
10.4
Party
Message
1.3
2.4
1
1
.25
.06
Literal Conditions
Sum mask
items
Mask
Soil
36.7
38.9
36.9
37.6
Party
Message
0.5
3.1
1
1
SO
Sum soil
items
Difference
score
Mask
Soil
Mask
Soil
28.2
27.2
27.7
25.1
3.1
5.8
1
1
.08
.01
8.6
11.6
9.2
12.5
Party
Message
Party
Message
0.6
11.1
1
1
.46
.04
.001
Note. For the mask and the soil items, a lower score indicates more agreement. Significances
are one sided for message effects and two sided for party effects.
278
BOSMANANDHAGENDOORN
and law students indeed reveal a higher effect size for psychology students (2.2)
than for law students (1.1). Nevertheless, the effect size for psychology students is still considerably lower than that found by Bosman. Another difference between the two experiments is that the earlier experiment was scheduled
immediately following another research project in which the participants filled
out a lengthy questionnaire about their conception of ethnic minorities, including repetitive semantic differentials. Although difficult to substantiate, the
larger effect size of Bosrnan's study might be due partly to the effects of fatigue,
numbing the participants' resistance to persuasion, or to a greater sensitivity
to racism evoked by the preceding questionnaire about ethnic minorities.
Finally, it should be noted that public interest in the CP was at a peak during
the first experiment. Since then, the CP has lost most of its electoral support,
and consequently public interest has faded.
Whatever the reasons for the smaller effect size in the present experiment,
these reasons probably apply to the literal conditions as well. Table 1 shows
the results of a two-way ANOVA on the literal conditions. Again, there are
no significant party effects and no interaction effects. All message effects are
significant, and the effect size of the literal conditions (3.2) is twice the effect
size of the metaphorical conditions (1.5). Separate analyses for psychology and
law students confirm the previous conclusion that the smaller effect sizes of
the present experiment are partly attributable to the inclusion of law students.
Again, the effect size for law students is small (1.0), and the effect size for
psychology students is substantially larger (6.0).
The evidence seems to favor literal messages as the most effective means of
persuasion. The effect size of literal messages is twice the effect size of metaphorical messages, and, disregarding the cross-grained law students, the effect
size of literal messages is nearly three times as large. However, an overall
ANOVA with style (metaphorical vs. literal) as the third factor does not show
up significant interaction effects between message and style for any of the
dependent variables. (Interaction between message and style, and not the main
effect of style, is indicative of differences in effect size.) Even a separate analysis
for the psychology students, where differences in effect size were larger, only
produces a marginally significant interaction (p = .07) for the difference score.
Summarizing these results, it seems safe to conclude that, when a suitable
paraphrase is available, metaphorical language has no special persuasive advantage over ordinary literal language. On the contrary, the evidence is in
favor of literal language, for although not statistically significant, effect sizes
for literal language are twice as large.
Who Is Persuaded?
Anyone trying to convince Anita Bryant that homosexuality is beneficial for
the soul is in for a rough time. On the other hand, the present authors are likely
to be a pushover for anyone arguing that this is an important article. Obvi-
Condition
Men
Metaphor
Literal
Women
Metaphor
Literal
Total
Metaphor
Literal
Right
Left
Total
3.1
1.2
0.7
0.1
3.0
2.1
2.2
6.3**
2.1a*
5.0**
0.7
0.8
2.5*
4.6**
1.5
3.2**
-2.9
-2.5
280
BOSMANANDHAGENDOORN
ous section), women, and left-oriented subjects are more affected by the messages than are law students, men, and right-oriented subjects.
The different results for psychology and law students can be attributed to
different compositions regarding sex and political affiliation: 24% of the psychology students are men and only 14% are right oriented, as compared to
54% and 52%, respectively, of the law students. Most of the right-wing men
(88%), for whom we found negative effect sizes, are law students.
The fact that women are more easily persuaded than men appears to be well
documented. Karlins and Abelson (1970, p. 89) contended that "this is one of
the most consistent and reliable findings in the field of persuasion." It is
tempting to conclude that women are more gullible than men. However, some
underlying variable might be responsible for the differences in effect size.
For instance, Scheidel (1963) proposed that it may be easier to persuade
people in areas they are not knowledgeable about, that women are generally less
knowledgeable about politics, and that these facts (combined with the persuasion researchers' predilection for research in the area of politics) have stage
managed the conclusion that women are more easily persuaded than men.
This position was taken by Eagly (1978), for instance. Eagly referred to a
number of studies demonstrating that "individuals are more readily influenced
to the extent that they lack information about the topic or regard it as trivial
and unimportant" (p. 96) and a number of studies showing that "men are more
knowledgeable and interested in political and economic areas than are women"
(p. 96). Repeating Scheidel's (1963) argument, she continued to point out that
the majority of the studies showing a significant sex effect in persuasibility are
concerned with political issues. Unfortunately, Eagly does not comment on the
proportion of no-sex-effect studies that deal with political issues, thus reducing
the value of her argument.
The evidence pertaining to the "less informed, more persuaded" hypothesis,
however, is conflicting. Scheidel's (1963) female subjects were more influenced
in an area they were supposed to be knowledgeable about than in other areas,
and Johnson and Taylor (1981) reported that politically sophisticated subjects
and subjects familiar with the specific topics of the persuasive message were
persuaded more than others. Bosman (1987) found no significant effects of
political sophistication, measured as a combination of the subject's political
interest and consumption of political information. On the other hand, the
present study yielded a significant Political Sophistication X Message interaction for the difference score, F(1, 330) = 5.0, p = .026, indicating larger
persuasive effects for politically unsophisticated subjects. Moreover, men were
significantly more "politically sophisticated" than women, t(180) = 5.3, p =
.000. Nevertheless, including political sophistication in the analysis did not
eliminate the significant Sex X Message interaction for the mask items, although it did reduce the Sex X Message interaction for the difference score
to a nonsignificant level (p = . I 1).
281
282
.01, and the difference score, F(1, 330) = 7.1, p = .008. As in the case of
sex and political affiliation, these interaction effects are not explained away
by political sophistication or the CP as the most disliked party. Hence, the
change of focus toward the negative effects of right-oriented men does not
improve the explanatory powers of political sophistication or the CP as the
most disliked party. It does, however, change the outlook on the problem
we are facing. The close-to-zero effects of men and right-oriented subjects
are the net effect of combining positive and negative persuasive effects and
should not be interpreted as resistance to persuasion. Therefore, theories involving the idea of resistance to persuasion, as applied earlier, are not quite
to the point. We instead should attempt to explain the negative effects for
right-oriented men.
Sherif, Sherif, and Nebergall (1965) argued that the effectiveness of a
persuasive message depends on the distance between the recipients' prior
attitudes and the position advocated by the message. Beyond a certain
point, the message ventures into the "latitudes of rejection," where it either
will be ineffective or will cause the recipient to change his or her position in
the opposite direction. If we are to explain the negative reactions of the
right-oriented men along these lines (suspending the question of why rightwing women do not behave similarly), then these negative effect sizes
should be caused primarily by the soil metaphor and its literal paraphrase,
as the soil metaphor was originated by left-wing political parties and alludes
to the typical left-wing theme that the human being is a product of historical and societal conditions, whereas the mask metaphor was exclusively employed by the right-oriented W D . In unraveling the effect sizes of the
right-oriented men, however, it is revealed that they are minor for the soil
items (1.0 and -0.3 for the metaphorical and the literal conditions, respectively) and negative for the mask items (-3.8 and -2.3, respectively). This
means that right-oriented men, after reading the mask message, agree less
with the mask inferences than they do after reading the soil message. Apparently, the latitude of rejection of the right-wing men is in their own
backyard! We return to this puzzling phenomenon in Part 2.
In summary, Women and left-oriented subjects are more affected by the
metaphors and their literal paraphrases. Unwilling to accept the easy explanation that these groups are downright more gullible than men and rightwing subjects, we set out to find underlying variables responsible for these
results. Although we did succeed in isolating a number of additional variables that interacted with the effect sizes, these variables did not explain the
differences in effect size for sex and political affiliation. Changing our focus
from the zero effects for men and right-oriented subjects to the negative
reactions of the right-wing men unearthed the paradox that these right-wing
men rejected the inferences of a message that should be well within their
latitude of acceptance.
The design includes the three-factor design of Part 1 of this study: Subjects
received a questionnaire about the CP or the NSB (party factor). A citation
in the introduction, allegedly by "one of the party leaders," contained the mask
metaphor or the soil metaphor (message factor) either in its original metaphorical form or in its paraphrased literal form (style factor). In addition, we
constructed a neutral questionnaire for both the CP and the NSB. The obvious
way to do this is simply by deleting the citations. To ensure a thorough reading
of the metaphorical and the literal citations, however, two questions were
inserted in the questionnaire, requiring the subjects to reread the citations.
Rereading the citations and answering questions about them obviously forces
the subjects to collect their thoughts about the CP or the NSB and possibly
influences their ideas about these parties. We therefore chose not to delete the
citations and the questions about them in the neutral questionnaire but to
substitute the citations with "neutral" citations. Neutral here does not imply
neutral toward the CP or the NSB but neutral with respect to the experimental
variables: the mask and the soil metaphors and their implications. It would be
futile to create citations neutral toward the CP or the NSB as these citations
are embedded in a negative introduction about these parties. In other words,
all conditions are equal in that they include a negative presentation about the
party involved. In essence, the neutral citations are composed of the connect-
284
ing sentences in the metaphorical and the literal citations: "We can no longer
neglect the CP (NSB). The CP (NSB) is a danger to society and it is the duty
of all democratic parties to put a stop to this danger." As in Part 1 of this study,
the sum of the mask items and the sum of the soil items were used as dependent
variables. The difference score (Sum Mask Items - Sum Soil Items) was not
employed because it expresses the joint effect of the mask and the soil metaphors (or their literal counterparts) and is not suitable when the message
conditions are compared with the neutral conditions.
Subjects
In the first part of this article, we compared the mask conditions with the soil
conditions, assuming that the mask metaphor or its literal counterpart would
induce more agreement with the mask items and, likewise, that reading the soil
metaphor would change the participants' agreement with the soil items. Table
3 presents the mean scores of the neutral conditions on the mask and the soil
items and the mean scores of the metaphorical and the literal versions of the
mask and the soil conditions, averaged for the CP and the NSB conditions.
In the second half of Table 3, these means are converted into a set of more
easily digestible numbers.
TABLE 3
Mean Scores on Mask and Soil Items for the Metaphorical
and the Literal Versions of the Mask and the Soil Conditions as
Compared to the Neutral Conditions, and the Conversion of These
Mean Scores Into Effect Sources
Mean Scores
Variable
Condition
Metaphor
Literal
Effect Source
Sum
mask
items
a. Mask
c. Soil
b. Neutral
36.9
37.4
36.8
38.2
Target effect
Nontarget effect
Total effect
Sum
soil
items
c. Mask
a. Soil
b. Neutral
27.8
26.7
28.0
26.2
Target effect
Nontarget effect
Total effect
36.8
26.8
Metaphor
Literal
(b - a )
(c - b )
(c - a )
-0.1
0.6
0.5
0.0
1.4
1.4
(b - a )
(c - b )
(C - a )
0.1
1 .O
1.1
0.6
1.2
1.8
285
The difference between the mean scores of the mask and the soil conditions, or the effect of metaphors and literal paraphrases as calculated in Bosman (1987) and in Part 1 of this article, is referred to as the total effect (c
- a). According to our initial assumptions, the total effect reflects the effect
that reading Metaphor A or Literal Paraphrase A has on the subject's position concerning the items associated with A. We refer to this effect as the
target effect of a message, and it is calculated as the difference between the
neutral and the mask conditions for the mask items and as the difference
between the neutral and the soil conditions for the soil items (b - a). Whenever the target effect does not equal the total effect, the difference between
these two can be expressed as the difference between the soil and the neutral
conditions for the mask items and the difference between the mask and the
neutral conditions for the soil items: (c - a ) - (b - a ) = (c - b). We
refer to this as the nontarget effect of a message. Nontarget effects register
the impact of metaphor or Literal Paraphrase A on the items associated with
B. Target and nontarget effects can be either positive (in line with the intentions of the message) or negative (contrary to the message). Table 4 presents
an overview and definition of the effects that are subsequently distinguished.
Contrary to our expectations, the target effects are close to zero, and the
outlines of the total effects are chiseled out of nontarget effects. The significances of the total effects are reported in Table 1. Predictably, none of the
target effects is significant. The nontarget effects, on the other hand, yield
significant effects of the literal paraphrases on the mask and the soil itemsF(l, 160) = 2.75, p = .05, and, F(l, 165) = 2.82, p = .05, respectivelyTABLE 4
An Overview of the Terminology: Negative and Positive Effects, Target
and Nontarget Effects of the Mask and the Soil Conditions
Effect
Condition
Mask
In accordance with the
intentions of the message
Contrary to the intentions
of the message
Soil
In accordance with the
intentions of the message
Contrary to the intentions
of the message
On Message-Related
Inferences
On Non-Message-Related
Inferences
286
BOSMANANDHAGENDOORN
and a close to significant effect of tlie mask metaphor on the soil items, F(1,
171) = 2.51, p = .06.
In Part 1 we found negative message effects for right-oriented men, both for
the metaphorical and the literal conditions. These results might indicate negative nontarget effects, contrary to the positive nontarget effects of the other
respondent groups. The proper way to proceed in testing the significance of
the nontarget effects is to exclude right-oriented men from the analysis. Excluding the right-oriented men yields significant nontarget effects of the literal
paraphrases on the mask and the soil items, F(l, 133) = 6.2,p = .007, and,
F ( l , 132) = 4.1, p = .023, respectively. Again, both for the literal and the
metaphorical conditions, the contributions of the target effects toward the total
effect are close to zero. The total effects are almost completely produced by
nontarget effects.
Thus, our initial assumptions turned out to be incorrect. No support was
found for the hypothesis that target effects are responsible for the persuasive
power of literal and metaphorical messages, and the hypothesis that nontarget
effects are not responsible must be rejected.
Additional support in favor of the importance of the nontarget effects can
be derived from Figure 1 . Figure 1 presents the relative contribution of the
FIGURE 1 Composition of the total effect: the relative importance of target and nontarget effects.
287
target and the nontarget effects to the total effects for all the subgroups that
thus far have been discussed.
In evaluating Figure 1, it should be noted that the subgroups are not
mutually exclusive. The order in which the subgroups appear in Figure 1 is
determined by the size of the total effect. It is obvious that there is little
connection between the target effect and the total effect, whereas the nontarget
effect is more than a fair predictor of the total effect. The target effects are close
to zero, and, except for the right-oriented subjects, the nontarget effects are
always larger than the target effects. The exceptional position of the rightoriented subjects is the logical outcome of merging the negative nontarget
effect of the right-oriented men with the positive nontarget effect of the rightoriented women. The nontarget effects provide a close replica of the total
effects: Psychology students, women, and left-oriented participants are more
easily influenced than are law students, men, and right-oriented participants;
the literal conditions are more effective than the metaphorical conditions, and
the participants in the NSB conditions changed their position more than those
in the CP conditions. And, as conjectured before, even the negative effects of
the right-oriented men are produced by the nontarget effects.
In Part 1, we observed that these negative effects of the right-oriented men
were attributable to the negative effects of the mask metaphor on the mask
items. Proceeding from the assumption that Metaphor A or Literal Paraphrase
A brings about changes in agreement with the items related to A (i.e., a target
effect), we concluded that the right-oriented men reacted negatively to the
mask messages. This conclusion runs counter to the plausible hypothesis that
negative effects are to be expected when the distance between the recipients'
prior attitudes and the position advocated by the message is considerable (i.e.,
when the message is in the recipients' latitude of rejection). However, as the
negative effects of the right-oriented men on the mask items are the result of
nontarget effects, they have not been produced by the mask messages but by
the soil messages! This is shown in Table 5. The right-wing subjects' position
on the soil items is barely affected by any of the messages, and their change
in position on the mask items is produced by nontarget effects (i.e., by the soil
TABLE 5
Effect Sources for Right-Oriented Men
Variable
Effect Source
Metaphor
Literal
Total
Sum
mask
items
Target effect
Nontarget effect
Total effect
- 1.6
1.7
-3.2
- 1.5
-2.4
-2.4
Sum
soil
items
Target effect
Nontarget effect
Total effect
0.4
-0.2
0.2
0.3
0.3
0.5
- 1.7
-3.3
0.1
0.8
0.8
0.0
288
BOSMANANDHAGENDOORN
message). Exposing right-wing men to the leftist soil message backfires on the
intentions of the soil message, not because it results in less agreement with the
soil items but because it fortifies and aggravates their position on the items
related to the mask messages.
290
REFERENCES
Bosman, J. (1987). Persuasive effects of political metaphors. Metaphor and Symbolic Activity, 2,
97-113.
Bowers, J. W. (1964). Some correlatesof language intensity. Quarterly JournalofSpeech, 50, 415420.
Bowers, J. W., & Osborn, M. M. (1966). Attitudinal effects of selected types of concluding
metaphors in persuasive speeches. Speech Monographs, 33, 147- 155.
Eagly, A. H. (1978). Sex differences in influenceability. Psychological Bulletin, 85, 86-116.
Johnson, J. T., & Taylor, S. E. (1981). The effect of metaphor on political attitudes. Basic and
.Applied Social Psychology, 2, 305-316.
Karlins, M., & Abelson, H. (1970). Persuasion. New York: Springer.
Lakoff, G. (1987). The death of dead metaphor. Metaphor and Symbolic Activity, 2, 143-147.
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of Chicago Press.
Petty, R.E., & Cacioppo, J. T. (1981). Attitudes and persuasion: Classic and contemporary approaches. Dubuque, IA: Brown.
291
APPENDIX
Unfortunately, it is impossible to provide an exact literal translation of the
Dutch quotations we used in our experiment. The quotations that follow
correspond as closely as possible to the original Dutch.
Metaphorical Quotations
Mask: "We have to pull off this party's mask. " "Underneath" (this mask)
lurks a terrible atmosphere of intolerance. We can no longer neglect the
Center Party. The Center Party is a danger to society and it is the duty
of all democratic parties to "reveal" the awful mentality of this party.
We can no longer ignore it; we must explain to the people what lies
"behind" this party.
Soil: We can no longer neglect the Center Party. "The Center Party is
the fruit born of an ill society. Adverse societal conditions are the soil for
the revival of racist feelings among the population. " If we do not take the
necessary steps, racism and the Center Party will continue to "grow."
We have to strike at the "root" of the evil. We can no longer ignore the
party. We have to let the people know how "rotten" racism is.
Literal Quotations
Mask: The Center Party presents itself as better than it is. In fact it has
a terrible atmosphere of intolerance. We can no longer neglect the Center
Party. The Center Party is a danger to society and it is the duty of all
democratic parties to clarify the awful mentality of this party. We can
no longer ignore it; we must explain to the people what this party is really
about.
Soil: We can no longer neglect the Center Party. The Center Party was
caused by a malfunctioning society. Adverse societal conditions have led
to an increase of racist feelings among the population. If we do not take
the necessary steps, racism and the Center Party will continue to in-
292
BOSMANANDHAGENDOORN
crease. We have to strike at the cause of the evil. We can no longer ignore
the party. We have to let the people know how harmful racism is.
Neutral Quotations
We can no longer neglect the Center Party. The Center Party is a danger
to society, and it is the duty of all democratic parties to put a stop to this
danger.