Upload
david-dryden
View
218
Download
0
Embed Size (px)
Citation preview
This article was downloaded by: [Northeastern University]On: 03 November 2014, At: 15:52Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Communication Research ReportsPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rcrr20
The Effects of Passive Verb-ConstructedArguments on PersuasionChristopher J. Carpenter a & David Dryden Henningsen ba Department of Communication , Western Illinois Universityb Department of Communication Studies , Northern IllinoisUniversityPublished online: 02 Feb 2011.
To cite this article: Christopher J. Carpenter & David Dryden Henningsen (2011) The Effects ofPassive Verb-Constructed Arguments on Persuasion, Communication Research Reports, 28:1, 52-61,DOI: 10.1080/08824096.2011.541358
To link to this article: http://dx.doi.org/10.1080/08824096.2011.541358
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
The Effects of PassiveVerb-Constructed Argumentson PersuasionChristopher J. Carpenter & David Dryden Henningsen
Different types of verb voice in a sentence were examined as a possible simple cue within
the unimodel of persuasion (Kruglanski & Thompson, 1999). An experiment was con-
structed by inducing two levels of argument strength (strong vs. weak) fully crossed with
three verb voice types in an independent groups design with 242 participants. Evidence
was found consistent with the prediction that verb voice acts as a simple persuasion cue;
the audience’s enjoyment of reading the arguments was found to be a mediator of the
relationship between verb type and message acceptance.
Keywords: Cognition; Language; Persuasion
The Chronicle of Higher Education once published an admonition to researchers that
they should avoid the passive voice in their writing (Pruitt, 1968). Pruitt called the
passive sentence construction ‘‘a form of naı̈ve hypocrisy at best, or intellectual
dishonesty at worst’’ (p. 461).1 Despite admonitions like Pruitt’s, few studies have
been conducted to study systematically the effects of verb voice on the persuasiveness
of messages. However, it is important to explore the variety of ways that structural
choices like verb voice can alter the persuasiveness of messages because structural
elements may be easily confounded with other inductions like argument strength.
Furthermore, the effects of verb voice could provide practical advice to social influ-
ence professionals such as advertisers and those creating public health campaigns.
Christopher J. Carpenter (PhD, Michigan State University, 2010) is an assistant professor in the Department of
Communication at Western Illinois University. David Dryden Henningsen (PhD, University of Wisconsin,
1999) is an associate professor in the Department of Communication Studies at Northern Illinois University.
Correspondence: Christopher J. Carpenter, Department of Communication, Western Illinois University, 221
Sallee Hall, #1 University Circle, Macomb, IL 61455-1390; E-mail: [email protected]
Communication Research Reports
Vol. 28, No. 1, January–March 2011, pp. 52–61
ISSN 0882-4096 (print)/ISSN 1746-4099 (online) # 2011 Eastern Communication Association
DOI: 10.1080/08824096.2011.541358
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
Finding a clear effect for verb voice would provide an easy way to increase the
persuasiveness of any message. Thus, this study explores the possibility that verb
voice affects persuasiveness by increasing how enjoyable reading the message is when
the audience is involved in shallow processing.
Verb Voice as a Simple Cue
Verb voice may alter how enjoyable arguments are to read and operate as a simple
cue to the persuasiveness of the message. The unimodel (Kruglanski & Thompson,
1999) argues that simple cues are only likely to be persuasive when the audience is
not devoting many cognitive resources to processing the message (i.e., shallow
processing). Work in this area has identified the mechanism for this effect—called
the relevance override hypothesis (Pierro, Mannetti, Erb, Spiegel, & Kruglanski,
2005). This hypothesis suggests that simple cues about the persuasiveness of a mess-
age may affect audience members who do not devote a great deal of cognitive
resources to evaluating that message. In other cases, the audience may devote more
cognitive resources to evaluating a message (deeper processing). In those cases, if
there is information in the message that is relevant to the issue and is complex in
nature (e.g., arguments), the audience will not be affected by the simple cues because
the more relevant and complex information is given full consideration and the simple
cues are given substantially less weight.
Research conducted in an advertising context lends support to these contentions.
For instance, Motes, Hilton, and Fielden (1992, Study 2) found that when an ad was
written using the passive voice, individuals rated it as more believable than when the
ad was written in the active voice. On the other hand, these researchers also observed
that sentences written in the active voice had a more positive effect than those written
in the passive voice on the audience’s intention to actually use the proposed service.
These findings produce conflicting recommendations regarding verb voice.
Verb voice does not directly alter the plausibility or relevance of the arguments in
a message. In other words, it does not make the arguments more or less logical, nor
does it alter the amount or quality of the evidence used to support those arguments.
When audience members devote a great deal of cognitive resources to processing a
message, the unimodel predicts that they would be less likely to be influenced by
irrelevant information such as the verb voice used in those arguments. Therefore,
if verb voice does affect the persuasiveness of the message, it should only act as a
simple cue when the audience does not devote a great deal of cognitive resources
to processing the message. The relevance override hypothesis suggests that when
the audience is processing the message deeply, the strength of the arguments should
override the effect of verb voice such that only argument strength will affect message
acceptance. Thus, the following predictions were made:
H1: Verb voice will combine non-additively with the depth of processing of themessage such that the fewer cognitive resources the audience devotes toprocessing the message (the more shallow the processing), the greater the impactverb voice will have on the persuasiveness of the message.
Communication Research Reports 53
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
H2: Argument strength will combine non-additively with the depth of processing ofthe message such that the more cognitive resources the audience devotes to proces-sing the message (the deeper the processing), the greater the impact argumentstrength will have on the persuasiveness of the message. The nature of that impactwill be that strong arguments will be more persuasive than weak arguments.
In light of Motes et al. (1992) contradictory findings regarding the believability of
a message versus its ability to motivate intended action, we posed a research question
concerning which type of verb voice will be more persuasive when the audience is
processing shallowly:
RQ1: Will active voice, passive voice with the agent present, or passive voice with theagent missing be the most persuasive?
The Motes et al. (1992) study was also examined to determine why different types
of verb voice might be more believable than others. They found that ads using passive
voice messages were sometimes rated as more attractive and as more likely to be read
than ads using other voices. In other cases, however, active voice produced more
positive effects on ad appeal. The attractiveness and appeal of the ad are both factors
that might affect how enjoyable an ad is to read. Rather than seeking to test the
variety of variables that Motes et al. examined, this study focuses on enjoyment of
reading the message in an effort to tie the other variables together. Regardless of
the type of verb voice that is more persuasive, overall enjoyment of reading will be
explored as a possible mediator of the effect of verb voice on the persuasiveness of
the message. Because no previous research has explored this possibility directly, a
research question is asked concerning the possibility that enjoyment mediates the
impact of verb voice on persuasiveness:
RQ2: Will the effect of verb voice on message acceptance under shallow processingconditions be mediated by how enjoyable the message is to read?
Motes et al. (1992) did not examine whether or not a message using an agentless
passive voice argument (e.g., the cheese was spread) would affect their outcome
variables differently than an argument that employs the passive voice with the agent
present (e.g., the cheese was spread by Jane). It is possible that the audience may
dislike the ambiguity concerning the actor in the sentence when arguments are
written in the passive voice with the agent missing. On the other hand, audience
members may appreciate how it shortens the sentences. Given that previous research
has not examined this possibility, this study poses the following research question:
RQ3: Will agent presence affect message persuasiveness or how enjoyable the messageis to read?
An experiment was designed to test these hypotheses and explore these research
questions by varying verb voice and argument strength and then measuring the
amount of cognitive resources each participant put into message processing (depth
of processing), how enjoyable the message was to read, and their acceptance of the
proposal.
54 C. J. Carpenter and D. D. Henningsen
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
Method
Participants
Participants were 242 students (129 men and 113 women) from a large Midwestern
University taking introductory and advanced communication studies courses. Their
mean age was 20.25 (SD¼ 2.88).
Procedures
The participants were given a packet with information and questionnaires. The
packet included the message induction that they were assigned. They were each
assigned to one condition of a 2 (Argument Strength: high or low)� 3 (Verb Voice
Type: active, passive with the agent present, or passive with the agent missing) fully
crossed factorial design. After they read their assigned message, participants
responded to the questions measuring the dependent variables.
Materials
All participants received a scenario that explained that the military was requesting
funds for the development a new nuclear weapon called the Robust Nuclear Earth
Penetrator (RNEP). The weapon was proposed by the Pentagon (Feinstein, 2005).
The scenario explained what the weapon was and how it worked. Argument strength
was induced by presenting two arguments in favor of developing the new weapon
that were either both strong or both weak.
Argument strength was established by pretesting a pool of 20 arguments concern-
ing the development of the weapon. Sixty-four undergraduates in a basic public
speaking class were each presented with one half of the arguments and asked to list
all of the thoughts they had after reading each argument. They then rated each
thought as positive if their thought indicated the argument was strong, negative if
their thought indicated the argument was weak, or neutral if their thought indicated
neither. If their thoughts were 65% or more positive, the argument was considered
to be strong; and if their thoughts were 35% positive or less, it was considered weak
(Petty & Cacioppo, 1986). This process uncovered two strong and two weak
arguments. Verb voice was induced by presenting the two arguments using all active
voice, all passive voice with agent present, or all passive voice with the agent missing
(see the Appendix for the arguments used in each condition, the n for each condition,
and the proportion of positive thoughts for each).
Measures
Thought listing. Thought listing was used to measure the amount of cognitive
resources the participants were putting into processing the message. Based on proce-
dures developed by Cacioppo and Petty (1981), participants were asked to list all of
the thoughts they had while reading the arguments; consistent with prior work, this
Communication Research Reports 55
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
served as an index of how deeply the audience was processing a message. Participants
were asked to write only one thought per line; however, some did not follow this
procedure and simply wrote a paragraph. Under these circumstances, a thought
was counted as one sentence or one idea in cases where sentence structure was
unclear. Unitizing reliability was assessed for 20% of the participants (48 parti-
cipants). Ninety-nine percent agreement emerged between coders on the number
of thoughts generated per participant (M¼ 4.17, SD¼ 2.12). A correlation matrix
depicting relationships among all of the dependent variables is presented in Table 1.
Persuasion. The degree to which the participants held argument consistent views
was measured using a scale constructed of five, 7-point semantic differential items.
The participants were asked to use the scales to rate their feelings about the develop-
ment of the RNEP. The items included good–bad, positive–negative, helpful–harmful,
safe–risky, and advantageous–disadvantageous. Higher scores indicated respondents
held more argument-consistent views than lower scores. The scale was reliable
(a¼ .87; M¼ 3.00, SD¼ 1.34).
Enjoyment of reading. The degree to which the participants enjoyed reading
the arguments was measured using five, 7-point semantic differential scales. The
participants were asked to rate their feelings about reading the arguments in the
persuasive message. The enjoyment of reading scale was comprised of the following
items: fun–boring, interesting–uninteresting, exciting–dull, pleasant–unpleasant, and
fascinating–tedious. Higher (vs. lower) scores indicated that participants found the
arguments to be more enjoyable to read. Pleasant–unpleasant had to be dropped
from the scale to establish acceptable reliability. The measure was reliable (a¼ .87;
M¼ 4.35, SD¼ 1.26).
Results
Preliminary Analysis
Examination of the persuasion scores indicated that the sample was against the devel-
opment of this weapon because the mean score (M¼ 3.00, SD¼ 1.34) was signifi-
cantly below the midpoint of the scale (4), t(239)¼ 11.61, p< .05 (d¼ .75). RQ3
asked if one of the two types of passive verbs would be more persuasive than the
other. Initial examination of the means revealed that the agent presence induction
did not affect any of the dependent variables. RQ3 was answered negatively. Because
Table 1 Correlation Matrix of Dependent Variables
Variable 1 2 3
1. Number of thoughts 0.04 �0.04
2. Enjoyment of reading 0.24
3. Attitude
56 C. J. Carpenter and D. D. Henningsen
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
of this finding, the two types of passive verbs are collapsed together in the analyses
that follow.
Passive Voice as Simple Cue
To examine H1, H2, and the RQ1, the persuasiveness of the message was regressed on
argument strength (strong arguments were dummy coded as 0, and weak arguments
were coded as 1), verb type (active voice was coded as 0, and the two passive conditions
were combined and coded as 1), the number of thoughts the participants listed, all of
the two-way interactions (product terms calculated by multiplying two of the inde-
pendent variables together), and the three-way interaction (the product of all three
independent variables). According to the relevance override hypotheses, the two-way
interaction between verb voice and number of thoughts should be substantial, and the
two-way interaction between argument strength and number of thoughts should be
substantial. There was a significant main effect for verb voice (b¼ .55), t¼ 2.71,
p< .01. The positive slope indicates that arguments constructed with either kind of
passive verb were more persuasive than arguments constructed with active verbs.
There was also a significant main effect for thoughts (b¼ .44), t¼ 2.38, p< .05, such
that listing more thoughts was positively related to scores on the persuasion measure;
this suggests a positive relationship between elaboration and message acceptance.
The main effects were moderated by significant two-way interactions. The two-way
interaction between verb voice and thought listing was substantial and statistically sig-
nificant (b¼�.58), t¼�2.17, p< .05. To interpret the interaction, deep and shallow
processing groups were created by placing the participants who listed more thoughts
than the median (4) into the deep processing group (n¼ 95) and the participants who
listed fewer thoughts than the median in the shallow processing group (n¼ 94). In the
shallow processing group, there was a substantial and statistically significant difference
between the participants based on verb voice, t(92)¼�2.34, p< .05 (r¼ .24). Argu-
ments written in the combined passive voice forms (M¼ 3.35, SD¼ 1.31; n¼ 67) were
more persuasive than the active voice arguments (M¼ 2.67, SD¼ 1.61; n¼ 27). In the
deep processing group, there was no substantial or statistically significant difference
between the verb voice conditions on the persuasion scores, t(93)¼�0.50, p> .05
(r¼ .05). Arguments written in either of the passive voice forms (M¼ 3.00, SD¼1.33; n¼ 60) were only slightly more persuasive than the active voice arguments
(M¼ 2.86, SD¼ 1.30; n¼ 35). This finding is consistent with H1, which predicted that
verb voice would act as a simple cue under conditions of shallow processing and be
overridden by argument strength under conditions of deep processing. This finding
also speaks to RQ1 by showing that passive verb voice arguments were more persuas-
ive than active voice arguments.
The two-way interaction between argument strength and thought listing was
substantial and statistically significant (b¼�.68), t¼�1.95, p¼ .05. In the shallow
processing group, there was no substantial or statistically significant difference on
the persuasion measure between those who were exposed to strong versus weak
arguments, t(92)¼�0.48, p> .05 (r¼ .05)—strong: M¼ 3.10, SD¼ 1.15, n¼ 44;
Communication Research Reports 57
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
weak: M¼ 3.23, SD¼ 1.48, n¼ 51). In the deep processing group, however, there was
a substantial and statistically significant difference on the persuasion measure
between the strong and weak argument groups, t(93)¼ 2.93, p< .01 (r¼ .29). The
strong arguments produced higher scores on the persuasion measure (M¼ 3.35,
SD¼ 1.32; n¼ 44) than the weak arguments (M¼ 2.59, SD¼ 1.22; n¼ 51) for the
deep processing group. This finding is consistent with H2, which predicted that
argument strength would only affect the persuasiveness of the message when the
audience was devoting the substantial cognitive resources necessary to evaluate
argument strength. No other main effects or interactions were statistically significant
or substantial for the persuasiveness of the message.
Enjoyment as a Mediator
RQ3 explored the possibility that if verb voice acts as a simple cue, its effect on the
persuasiveness of the message might be mediated by how enjoyable the participants
thought the message was to read when processing was low. This possibility was tested
using path analysis (Hunter & Gerbing, 1982). For the participants in the shallow
processing condition (thoughts <4), there was a substantial and statistically signifi-
cant effect of verb voice on how enjoyable the arguments were to read, t(91)¼ 3.47,
p¼ .001 (r¼ .34), such that those who read active voice arguments perceived them
to be less enjoyable (M¼ 3.54, SD¼ 1.57; n¼ 27) than both types of the passive voice
arguments combined (M¼ 4.58, SD¼ 1.20; n¼ 66).
To test for mediation using path modeling, the product rule was used (Hunter &
Gerbing, 1982). It states that if the correlation between the first and the third variables
in the causal chain (r13) is within sampling error of the product of the correlation
between the first and the second variable and the correlation between the second
and the third (r12 � r23), then the second variable mediates the relationship between
the first and the third. The product of the relationship between verb type and
enjoyment and the relationship between enjoyment and persuasiveness is r¼ .16.
The corrected obtained correlation was within the 95% confidence interval of P
(.05< q< .5)¼ 95% of the corrected obtained correlation (r¼ .28) between verb type
and persuasiveness (see Table 2 for the corrected and uncorrected correlation
matrix under conditions of shallow processing). Correlations corrected for error of
measurement were used to test the model in order to increase the accuracy of the
model estimates. This path analysis found that enjoyment of reading mediated the
Table 2 Correlation Matrix Under Conditions of Low Processing
Variable 1 2 3
1. Verb type 0.34 0.24
2. Enjoyment 0.40 0.34
3. Persuasiveness 0.28 0.39
Note. The upper triangle reports the uncorrected correlations, and the lower triangle
reports the corrected correlations.
58 C. J. Carpenter and D. D. Henningsen
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
relationship between verb type and persuasiveness under conditions of shallow proces-
sing such that passive verbs were rated as more enjoyable to read and enjoyableness of
reading was positively related to persuasiveness. This outcome answered RQ2 affirm-
atively.
Discussion
The purpose of this study was to examine the effects of verb voice inductions on the
persuasiveness of a policy proposal. The unimodel (Kruglanski & Thompson, 1999)
was used to derive the prediction that verb voice would only affect the persuasiveness
of the message when the audience was processing shallowly. Past research on adver-
tising (Motes et al., 1992) suggested that the enjoyableness of reading the message
might mediate the relationship between verb voice and message acceptance.
The evidence was consistent with the prediction that passive verb voice could act
as a simple cue by making the arguments more enjoyable to read, which, in turn, also
made them more persuasive when the audience was processing shallowly. The simple
cue hypothesis was derived from the unimodel (Kruglanski & Thompson, 1999) and
predicted that verb voice would affect the persuasiveness of the message by acting as a
simple persuasion cue. Specifically, the work of Motes et al. (1992) pointed toward
the possibility that passive voice constructed arguments are more enjoyable to read.
Increases in enjoyment ratings of the arguments were associated with greater per-
suasion among the participants who were processing shallowly. In contrast, those
who devoted more cognitive resources to processing the message discounted the
enjoyableness that passive voice arguments produced and were only persuaded by
strong arguments as per the relevance override hypothesis suggested by Pierro et al.
(2005). Future work should examine other linguistic alterations that might affect the
audience’s enjoyment of reading or hearing a message.
Limitations
One limitation of this study is that, despite pretests showing arguments to be strong,
the average scores on the attitude scale were still below the midpoint in the main
study. This was the case even when the audience was processing deeply and reading
strong arguments. Future research should examine situations where the audience is
less recalcitrant. It is also possible that the Petty and Cacioppo (1986) method of
establishing argument strength is faulty (O’Keefe & Jackson, 1995).
Conclusion
Despite these limitations, this study nevertheless establishes the possibility that linguis-
tic inductions can have an effect on the persuasiveness of a message by altering how
enjoyable the message is to read. Future persuasion research using message inductions
should take care to hold verb voice consistent in order to avoid potential confounds.
Before a preference for the passive voice is recommended to applied fields, more
research is required to both demonstrate this effect and understand its mediators.
Communication Research Reports 59
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
Note
[1] The article went on to explain different ways to use a verb in a sentence. When using the
active voice, the writer of a sentence puts the agent before the verb and the object is placed
after the verb. An active voice sentence is, ‘‘The elephant trampled the trees’’ (Pruitt, 1968,
p. 462). The elephant is the agent and the trees are the object in this sentence. The passive
voice with the agent present puts the object of the action first and the agent after the verb,
producing a sentence like, ‘‘The trees were trampled by the elephant’’ (p. 462). The agent is
still present, but it is moved further into the sentence. The third type of verb voice is
the passive voice with the agent missing, where the target of the action is put first and
the agent is removed from the sentence entirely as in, ‘‘The trees were trampled’’ (p. 462).
A sentence can be written in the active or passive voice, and the passive voice can include the
agent or not.
References
Cacioppo, J. T., & Petty, R. E. (1981). Social psychological procedures for cognitive response
assessment: The thought-listing technique. In T. V. Merluzzi, C. R. Glass & M. Genest
(Eds.), Cognitive assessment (pp. 309–342). New York, NY: Guilford.
Feinstein, D. (2005, April 14). U.S. Senator Pete V. Domenici (R–NM) holds hearing on fiscal year
2006 appropriations for the national nuclear security administration. FDCH Political
Transcripts.
Hunter, J. E., & Gerbing, D. W. (1982). Unidimensional measurement, second order factor analysis,
and causal models. Research in Organizational Behavior, 4, 267–320.
Kruglanski, A. W., & Thompson, E. P. (1999). Persuasion by a single route: A view from the
unimodel. Psychological Inquiry, 10, 83–109.
Motes, W. H., Hilton, C. B., & Fielden, J. S. (1992). Language, sentence, and structural variations in
print advertising. Journal of Advertising Research, 32, 63–77.
O’Keefe, D. J., & Jackson, S. (1995). Argument quality and persuasive effects: A review of current
approaches. In S. Jackson (Ed.), Argumentation and values: Proceedings of the ninth SCA=AFAconference on Argumentation (pp. 88–92). Annandale VA: Speech Communication Association.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral routes
to attitude change. New York, NY: Springer-Verlag.
Pierro, A., Mannetti, L., Erb, H. P., Spiegel, S., & Kruglanski, A. W. (2005). Informational length
and order of presentation as determinants of persuasion. Journal of Experimental Social
Psychology, 41, 458–469.
Pruitt, J. D. (1968). Passive voice should be avoided by researchers. Journal of Higher Education, 39,
460–464.
Appendix
Pretest Ratings
Strong
1. If the United States developed the RNEP [Robust Nuclear Earth Penetrator], we
could create flexible nuclear weapons to replace the old overkill weapons of the
Cold War. (66% positive)
2. If the United States developed the RNEP, it would save so much money, there
would be more money for other programs. (71% positive)
60 C. J. Carpenter and D. D. Henningsen
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014
Weak
1. If the United States developed the RNEP, it would not violate any treaties. (24%
positive)
2. If the United States developed the RNEP, we might actually get a return on our
investment unlike the Cold War nuclear weapons we never actually used. (29%
positive)
Test Materials
Strong arguments� active voice
If the United States developed the RNEP, we could create flexible nuclear weaponsto replace the old overkill weapons of the Cold War. Also, if the United Statesdeveloped the RNEP, it would save so much money, there would be more moneyfor other programs. (n¼ 41)
Strong arguments� passive voice with the agent present
If the RNEP was developed by the United States, flexible nuclear weapons could becreated by us to replace the old overkill weapons of the Cold War. Also, if theRNEP was developed by the United States, so much money would be saved byit, there would be more money for other programs. (n¼ 40)
Strong arguments� passive voice with the agent missing
If the RNEP was developed, flexible nuclear weapons could be created to replacethe old overkill nuclear weapons of the Cold War. Also, if the RNEP was developed,so much money would be saved, there would be more money for other programs.(n¼ 40)
Weak arguments� active voice
If the United States developed the RNEP, it would not violate any treaties. Also, ifthe United States developed the RNEP, we might actually get a return on ourinvestment unlike the Cold War nuclear weapons we never actually used. (n¼ 41)
Weak arguments� passive voice with the agent present
If the RNEP were developed by the United States, no treaties would be violated byit. Also, if the RNEP were developed by the United States, a return on our invest-ment might actually come back to us, unlike the Cold War nuclear weapons thatwere never actually used by us. (n¼ 40)
Weak arguments� passive voice with the agent missing
If the RNEP were developed, no treaties would be violated. Also, if the RNEP weredeveloped, a return on our investment might actually come back, unlike the ColdWar nuclear weapons that were never actually used. (n¼ 40)
Communication Research Reports 61
Dow
nloa
ded
by [
Nor
thea
ster
n U
nive
rsity
] at
15:
52 0
3 N
ovem
ber
2014