Journal of Personality and Social Psychology1976, Vol. 34, No. 4, 615-624

Speed of Speech and Persuasion

Norman Miller, Geoffrey Maruyama, Rex Julian Beaber, and Keith ValoneUniversity of Southern California

The relationship between speaking rate and attitude change was investigatedin two field experiments. Manipulations of speech rate were crossed with (a)credibility of the speaker and (b) complexity of the spoken message. The re-sults suggest that speech rate functions as a general cue that augments credi-bility; rapid speech enhances persuasion. No support was obtained for infor-mation-processing interpretations of the effects of a fast speaking rate. Theincreased persuasion produced by fast speech could not be attributed to dis-ruption of effective counterarguing. These findings emphasize the importanceof perceptual and evaluative factors in the persuasion process at the expenseof a more rationalistic information-processing view of how man responds whenconfronted with an influence attempt.

Although many of the persuasive communi-cations received during daily life are orallypresented, the characteristics of speech thataffect persuasion are rarely studied. More-over, studies that do examine delivery styletypically treat it as a global variable. Forinstance, Dietrich (1946) studied the dyna-mism of a speaker's style, Bettinghaus (1961)assessed the effects of overall speaking effec-tiveness by comparing trained and uncoachedstudents, and Bowers (1965) examined theeffects of extroverted and introverted deliv^ery using dramatic arts students to simulatestyle. The persuasive effects of more discretecharacteristics of oral delivery, such as inten-sity, pitch, speed of presentation, or specificemotional qualities apparently more rarelyelicit attention.

Several considerations suggest that speechrate might be an important variable. Texts onspeech (Allen, Anderson, & Hough, 1968;Monroe & Ehninger, 1974) have reportedconsiderable individual variation in speakingrate with 120-180 words per minute as thelower and upper limits of normal speech.1 Itis a short step to imagine theoretical implica-

Preparation of this article for publication wasaided by support from National Institute of MentalHealth Grant MH26094, awarded to the first author,and a Haynes Foundation Fellowship to the secondauthor.

Requests for reprints should be sent to NormanMiller, Department of Psychology, University ofSouthern California, University Park, Los Angeles,California 90007.

tions of such variation that might contributeto differences in persuasion. Yet, even thoughFoulke and Sticht (1969) have most recentlystudied the effects of speed of presentation oncomprehension, its effects on persuasion re-main unexplored.

McGuire (1969) characterized attitudechange as a stochastic process that flowsfrom attention through comprehension, yield-ing, retention, and action. Although rate ofspeaking might theoretically be expected toinfluence any of these processes, it initiallyseems plausible to focus on how it affects com-prehension. In their review, Foulke andSticht (1969) showed that when significantresults are found, most are in the directionof lowered comprehension when speed ofspeech is rapid.2

If a fast speaking rate does reduce com-prehension, then it should also curtail per-suasive power (Eagly, 1974). However, sincethe results of most studies in the comprehen-sion literature are not significant, it maymake more sense to look for the effects ofspeed in some other stage of the persuasiveprocess, such as yielding. To focus away fromcomprehension seems particularly sensible, in

1 For example, radio announcers are often notedfor their rapid speech rate. Lumley (1933) deter-mined the average speech rate of 25 radio speakersto be 162 words per minute, with a standard devia-tion of 19.40 words per minute.

2 It should be pointed out that a majority of thestudies have found no significant effects.

615

616 MILLER, MARUYAMA, BEABER, AND VALONE

that the high speeds used in the prior re-search equaled or exceeded the upper limitsof the range of normal speech. Consequently,one might reasonably assume that the speedmanipulations in these studies were suffi-ciently powerful to have shown an effect ifone does indeed exist.

How might speed affect the yielding phaseof persuasion? A fast speaking rate might dis-rupt any covert attempt to counterargueagainst the persuasive content of the message,thereby increasing persuasion. Whereas anyeffects of speed on comprehension or counter-argument disruption can be conceptualized asmessage factors—characteristics of the mes-sage that affect persuasion—speaking ratemay alternatively act as a source factor. Thatis, it might provide cues for viewing a fastspeaker as more knowledgeable, competent,and facile. In this case a rapid speaking rateconstitutes a credibility cue and on this basisshould likewise enhance persuasion.

Although the comprehension interpretationof speaking rate does make a distinct predic-tion—less attitude change when speech israpid—both the counterargument disruptionand enhanced credibility interpretation makethe opposite prediction—greater attitudechange with a rapid speaking rate. How canthese two hypotheses be separated? First,with respect to the counterargument disrup-tion hypothesis, if rapid speech does indeeddisrupt covert counterarguing, then, as im-plied by Kiesler and Mathog (1968), the im-pact of speaking rate should vary with thecredibility of the speaker. According to theirargument, when the message is counteratti-tudinal and the speaker possesses substantialcredibility, counterarguing tends to becomethe most viable mode of resisting influence.This should be particularly true if one ishighly involved in the issue (Osterhouse &Brock, 1970). However, counterarguing is or-dinarily an elaborate internal response, onerequiring time and mental energy. A com-munication spoken rapidly should reduce theavailable time and increase the difficulty ofmaking such covert responses. Consequently,increased speed should facilitate attitudechange. On the other hand, a person listeningto a speaker who lacks credibility can resistinfluence by using a less elaborate internal

response; he can simply reject or discountwhat is said. Under these conditions, speakingrate should not substantially attenuate abilityto resist, and consequently, should have littleimpact on persuasion. Although there are fewdirect tests of the role of counterarguing inthe persuasion process (Miller & Baron,1973), the reasoning above does make a clearprediction. It suggests that if fast speech doesdisrupt counterarguing, there should be aninteraction between speaking rate and credi-bility. Alternatively, if speaking rate simplyfunctions as a credibility cue, no interactionwould be predicted. Instead, as previouslysuggested, fast speech should enhance per-suasion regardless of the speaker's credibility.

The results of two studies that begin todelineate the relation between speed of speechand persuasion (Beaber & Miller, Note 1)suggest that speech rate functions as a cuefor increased credibility. The first experimentof Beaber and Miller investigated rate ofspeaking and credibility of the source in a2X2 design. In addition to finding the credi-bility manipulation to be successful (the high-credibility source was seen as more knowl-edgeable and more trustworthy), the fastspeaker was perceived as more knowledgeableand more trustworthy. In addition, the com-munication was perceived as more complexbut also clearer and easier to understandwhen presented by a fast speaker. These find-ings suggest that a rapid speaker is perceivedas more competent, since it takes a skilledspeaker to present complex material clearly.However, since credibility was manipulatedby varying both ability and trustworthinessin this first study, the specific component ofcredibility that speed enhanced cannot clearlybe specified. Certainly, there are situations inwhich high ability and trustworthiness do notnecessarily occur together. The most obviousexample is a used car dealer who, despite hisexpertise, is not always viewed as trust-worthy and may be seen as even less believ-able if he speaks very rapidly.

Therefore, in a second experiment Beaberand Miller manipulated trustworthiness inde-pendently from expertise to investigatewhether or not the persuasiveness of an un-trustworthy communicator would be enhancedby rapid speech. In the first study, any effects

SPEED OF SPEECH AND PERSUASION 617

of trustworthiness could have been obscuredby the difference in expertise of the com-municator. Whereas in the first experimentthe communication advocated "offshore dril-ling for oil," in the second experiment itargued that "The Mazda is the best car valuearound." The communicator was depicted asa used Mazda salesman with something togain from his persuasiveness (low trust) or asa former Mazda salesman who currentlyworked in a bank (high trust). Thus, exper-tise was held constant. This manipulationseemed most likely to produce an interactionbetween speed and credibility and therebytest the generality of speed as a cue for en-hanced credibility. In this instance, an un-trustworthy fast speaker should appear to beless believable than an untrustworthy slowspeaker; contrarily, a trustworthy fast speakershould be more believable than a trustworthyslow speaker, replicating the outcome of thefirst study.

Manipulation checks showed that the trust-worthy speaker was in fact rated as moretrustworthy and, furthermore, that the manip-ulation was independent of expertise, sincethe high- and low-trust speakers were ratedas equally knowledgeable. As in the first ex-periment, rapid speech enhanced ratings ofboth trust and knowledge. Further parallelingthe first experiment, measures of persuasionalso showed the rapid speaker to be morepersuasive.

In summary, then, these two experimentsstrongly argue that speaking rate operates asa very general and powerful credibility cue.The results clearly contradict information-processing hypotheses concerning rate ofspeech. If fast speech had impaired compre-hension, it should have likewise reducedrather than augmented persuasion. Alterna-tively, as argued in the first experiment, iffast speech had interfered with covert counter-arguing, it should have more strongly facili-tated persuasion when credibility or trust-worthiness was high than when it was low.

If, despite the data of these two experi-ments, one is intent on saving the compre-hension or counterargument disruption hy-potheses, one might argue that the propertests have not been performed. Since both ofthese experiments were laboratory studies,

subjects might have construed the situationas a test. Consequently, the predicted inter-action between credibility and speech ratemight have failed to receive support becausethe situation demanded extremely high atten-tion regardless of the level of the speaker'scredibility. Thinking that they might subse-quently be tested on some aspect of their re-tention or comprehension, subjects might havefocused more on the content of the argumentsregardless of the level of credibility assignedto the speaker. According to this view, where-as faster speech might indeed produce someminor enhancement of credibility, its moreimportant, potent, pervasive, and normal con-sequences are those affecting informationprocessing, whether decreasing comprehensionor alternatively disrupting one's ability tocounterargue. Yet, in laboratory experiments,these effects cannot be observed; attentionand vigilance to content is made so asymp-totically high by the implicit demands of thelaboratory that the normally weaker and lessimportant credibility enhancement effects re-main as the only ones to be observed.

In terms of these arguments, a field settingmight provide a more adequate test of theinformation-processing interpretations ofspeaking rate. Since the public accepts on-the-street interviewing with little suspiciousquestioning, test expectancies might not op-erate when a field experiment is concealedbehind this type of facade.

EXERIMENT 1

Experiment 1 manipulated speech rate andspeaker credibility in a field setting to deter-mine their effect on attitude change and per-ceptions of the speaker. Since in the Beaberand Miller (Note 1) experiments the trust-worthiness dimension of credibility did notinteract with speech rate but rather func-tioned as a cue that unidirectionally enhancedcredibility, no attempt was made to manipu-late orthogonally the competence and trust-worthiness dimensions of credibility.

Method

Subjects and experimenters. Subjects were 359 per-sons living in the Los Angeles area. They were indi-vidually approached in a public area by 1 of 14 ex-perimenters, each of whom conducted multiple repli-

618 MILLER, MARUYAMA, BEABER, AND VALONE

cations of the design.3 Experimenters were male andfemale undergraduates enrolled in an upper-divisionpsychology class at the University of Southern Cali-fornia.

Procedure. Each experimenter assigned a subjectto each cell before starting the next replication. Theorder of cells was determined by random assignmentbut was constant across replications and experi-menters. All subjects were individually run. Theywere approached in a variety of public settings, suchas parks, various on-the-street locations, othercampuses, shopping malls, and door-to-door in vari-ous neighborhoods.

The experimenter approached the prospective sub-ject and introduced himself (herself) as an inter-viewer from radio station KSC for the programPeople's Forum, which was described as being con-cerned with topics of interest to the listening audi-ence. Subjects were told that tapes had been madecontaining listeners' opinions on various topics. Theinterviewer asked the subject to listen to a tape onthe current day's topic ("The danger of drinkingcoffee") and then to give his reaction to it. If thesubject agreed, the interview was continued bydescribing the author of the tape and playing it forthe subject.

Persuasive communication. The communicationwas approximately 400 words in length. It arguedthat coffee is a hazardous drug; caffeine in largedoses is a poison and can cause damage to andfluttering of the heart, headaches, migraines, coldhands, and stomach ulcers; by decreasing appetiteand by leading to insufficient sleep, coffee lowersresistance to disease; by providing momentary en-ergy, it hides symptoms of other medical problems;it adversely affects teeth and gums; it is addictingand causes withdrawal symptoms; and in conclusion,like cigarettes, it should be classified by the govern-ment as hazardous.

Credibility. In the low-credibility condition thespeaker was introduced as a locksmith, whereas inthe high-credibility condition he was introduced asa biochemist. To augment the credibility manipula-tion, errors in pronunciation, grammar, and wordusage were inserted into the speech of the low-credi-bility speaker, while retaining the same content andmeaning. After describing the speaker, the experi-menter played either a fast or slow version of therecorded message and then asked the subject toanswer a few questions.

Speaking rate. The speaker used in this experimentspoke at approximately the same rates as did the onein the Beaber and Miller (Note 1) experiments. Thespeaking rate in the low-speed condition averaged102 words per minute, whereas in the high-speedcondition it averaged 195 words per minute.4 Theserates were achieved by simply instructing the speakerto practice delivering the speech as rapidly andslowly as possible while controlling his level of en-thusiasm and involvement. Tapes were made onstudio-quality equipment. Overall volume was bal-anced across tapes by matching average volumes onfive sample segments of each tape. Cassette record-

ings were reproduced from the master tapes anddelivered via cassette recorders.

Dependent measures. After listening to the tape,subjects were asked to indicate orally their agree-ment with a series of statements. Agreement wasindicated on 10-point Likert scales (midpoint = S.S),with higher numbers indicating stronger agreement.The statements concerned global agreement with themain contention of the communication, perceivedspeaker characteristics (knowledgeability), perceivedclarity of the presentation, and the personal coffee-drinking habits of the subject.

Results

Inspection of the credibility manipulationcheck indicated that the high-credibilitysource was perceived as more knowledgeable:Ms — 7.93 and 5.19 in the high- and low-credibility conditions, respectively; / * \ 1 , 13)= 89.1, p < .001. This effect was determinedby testing the main effect against the Experi-menter X Condition interaction, since thisinteraction was significant.5 This most con-servative test (using the significant interac-tion term) also revealed a significant effect ofcredibility on perceived understanding: Ms= 8.71 and 8.07 in the high- and low-credibil-ity conditions, respectively; 7^(1,13) = 7.8,p < .05. The effect of speech rate on knowl-edgeability approached standard levels ofsignificance: Ms = 6.93 and 6.12 in the fast-

3 The authors wish to express their sincere appreci-ation to Susan Anderson, Evie Baker, Laurie Brand,Pat Brotman, Gayle Caviglia, David George, Med-rith Hose, Judi Kammerman, David Mace, SteveMayer, David Quezada, Alan Seims, Francis Troll,and Jan VanSchaik for their able assistance in thecollection of the data.

4 The high-credibility-slow-speech rate was 99.6,the high-credibility-fast-speech rate was 194.4, thelow-credibility-slow-speech rate was 104.8, and thelow-credibility-fast-speech rate was 195.8 (wordsper minute). For the two Beaber and Miller (Note1) studies, a rate of about 110 words per minutewas used in the slow-speech conditions, while a rateof about 190 words per minute was used in the fast-speech conditions. Despite the similarity of speechrates, the different communicator used in the presentstudy should increase the generalizability of theresults.

5 Different experimenters produced different mag-nitudes of effect with the credibility manipulation.Therefore, each effect was tested against its inter-action with experimenters. In effect, then, we areusing a mixed model with experimenters as a randomfactor, testing each fixed effect against its interactionwith experimenters.

SPEED OF SPEECH AND PERSUASION 619

and slow-speech conditions, respectively; F(l,13) = 4.4, p < .0.6, again, by the more con-servative test. Thus the high-speed speakertended to appear more knowledgeable.

An analysis of variance on the postcom-munication agreement measure (using numberof cups of coffee drunk daily as a covariate)yielded main effects that approached signifi-cance when tested with the conservative ex-perimenter interaction error term: Ms = 6.13and 5.44 in the high- and low-speed condi-tions, respectively; F(l, 13) = 2.80, p < .13;Ms = 6.28 and 5.33 in the high- and low-credibility conditions, respectively; F(l, 13)= 4.41, p < .06. Thus the high-speed com-munication was more persuasive than the low-speed communication.6 The fact that the ex-perimenters used in this study lacked muchexperience (which probably accounts in largepart for the Experimenter X Condition inter-actions) might well have reduced the strengthof the effect of speech rate. Further, some re-searchers have argued that effects in thefield are typically weaker than those inthe laboratory (Hoviand, 1959). In addi-tion, examination of the tapes revealedan instance of a glaring nonfluency (re-peated stuttering over one word) in thehigh-credibility-high-speed speech. As shownby Miller and Hewgill (1964), nonfluen-cies typically lower perceived credibility.Consequently, this should have further re-duced the effect of the speed manipulation.Despite these difficulties, it should be notedthat the data did not support either the com-prehension prediction—less persuasion underhigh speed—or the counterargument predic-tion—a Speaking Rate X Credibility interac-tion showing a speed effect under high-credi-bility conditions but no speed effect in low-credibility conditions.7 Instead, they supportthe view that rapid speech serves as a credi-bility cue and thereby enhances persuasion.

EXPERIMENT 2

Although the credibility hypothesis con-tinued to receive general support from Experi-ment 1, there may still be special circum-stances in which speech rate functions to dis-rupt counterarguing. The communicationsused in both Beaber and Miller (Note 1)

studies and in Experiment 1 did not provideespecially novel arguments to the listener, nordid they deal with particularly unfamiliartopics. In persuasion paradigms that offernovel arguments or topics, adequate process-ing of message content should become moresalient. From this standpoint, it can beargued that for the topics used in the re-ported studies, subjects had counterargumentsreadily available and therefore did not experi-ence the high rate of presentation as beingdisruptive. Elaborate processing of the mes-sage was not necessary. A more sensitive testof the counterargument disruption hypothe-sis might require novel communications forwhich simple retrieval of stock counterargu-

6 All effects are highly significant (p < .01) whenthe standard, and less conservative, within-cell vari-ance is used as the error estimate. However, torevert to this more typical error estimate meansthat these results cannot be generalized beyond theparticular set of experimenters used to collect thesedata. This is of course normally true for virtuallyall experiments reported in psychological journals, inthat few researchers test their effects against thevariability of different experimenters' outcomes.Most use only a single experimenter. If the resultswere significant using the more conservative errorestimate (the interaction with experimenters), theirexternal validity would be enhanced in that theypresumably would obtain with any of the experi-menters from the larger population of which these14 constitute a random sample.

7 In attempting to defend the counterargumentexplanation, some might argue that people primarilyuse covert counterarguing to resist persuasion whenthey are involved with or have a personal concernabout the topic. While we have not explicitly ma-nipulated involvement, we can indirectly explorethis notion by attending to an individual differencemeasure of the extent to which our subjects werelikely to be concerned with the topic. It is reason-able to assume that subjects who report being heavycoffee drinkers are likely to be more concerned andinvolved with a speech emphasizing the dangers ofdrinking coffee. The results of the analysis of covari-ance argue against the view that covert counter-arguing is used to resist persuasion primarily whenone is concerned with the issue. The persuasion ef-fects are found even though personal involvement(number of cups of coffee drunk per day) was co-varied out. If speech rate affects persuasion by inter-fering with counterarguing and if counterarguingonly occurs when the topic is involving, removingthe effect of involvement via covariance analysisshould also eliminate any effect of speech rate onpersuasion,

620 MILLER, MARUYAMA, BEABER, AND VALONE

ments would not suffice as a means for re-sisting influence. Therefore, for the presentexperiment, a novel topic, "The dangers ofhydroponically grown vegetables," was chosento ensure that subjects would be forced toprocess novel arguments.

Although a more sensitive test of thecounterargument hypothesis might be createdby using a more novel communication topicand simply replicating one of the earlier stud-ies that examined the Speech Rate X SpeakerCredibility interaction, a more potent criti-cism of the initial experiments would still re-main unresolved. The predicted Speed XCredibility interaction rested solely on thehypothesized differential salience of counter-arguing as a means of belief defense in high-versus low-credibility conditions. AlthoughKiesler and Mathog (1968) provided experi-mental support for the assumption thatcounterarguing is the preferred mode of de-fense when the speaker possesses high credi-bility, alternate interpretations of their dataare possible (Baron, Baron, & Miller, 1973).In addition, since a credibility manipulationonly indirectly affects orientation toward con-tent, it is difficult to specify the range overwhich content orientation is varied. Failuresto produce the predicted interaction may onlyreflect a weak manipulation of orientationtoward and away from content.

A more direct test of the counterargumenthypothesis should use variables more clearlyrelated to information processing. Theoreti-cally, message complexity should have morepronounced effects on the difficulty of pro-cessing information. Indeed, highly complexmessages may force the listener to restrict hisor her covert activity and focus entirely onmessage comprehension. With complex com-munications, processing difficulties should benear their ceiling, leaving little room for speedeffects. However, with simpler messages, anincreased rate of speech should reduce theavailable time for counterargument produc-tion, thereby facilitating persuasion. In sum,there should be an interaction of speed andmessage complexity on attitude change.

To test the interaction predicted by aninformation-processing model, Experiment 2manipulated speech rate and message com-plexity in a 3 X 2 factorial design using a

"novel" topic. Three levels of speaking ratewere employed to ensure that the effects ofspeech rate are not curvilinear and to extendfurther the generality of earlier findings.

Method

Subjects. A total of 90 males and females, chosenfrom the mall of a large open-air shopping center,participated in the study.

Procedure. Subjects were randomly assigned tothe six cells of the design before replication. Theprocedure and facade largely paralleled that of Ex-periment 1. The experimenter approached subjectsindividually and introduced himself as an interviewerfrom radio station KSC who was interviewing peo-ple on the street for the program People's Forum,which was again described as being concerned withtopics of interest to the radio audience. Subjectswere told that tapes had been made containing lis-teners' opinions on various topics. The interviewerasked the subject to listen to a tape on the topic ofthe day ("The dangers of hydroponically grownvegetables"). If the subject agreed to hear the tape,the interview was continued by describing the authorof the tape and playing the tape for the subject.The speaker was ostensibly introduced as CharlesTravis, a produce manager at Ralph's Ranch Marketin Hollywood. The topic of the tape was chosenbecause it had been previously determined that fewpeople know what the term hydroponics means, letalone understand the arguments favoring or opposinghydroponic growing methods. Lacking arguments,subjects are unlikely to have a defensible attituderegarding the value of hydroponics.

Persuasive communication. The communicationwas approximately 300 words in length. After defin-ing hydroponics, it mentioned a number of "prob-lems" with hydroponic methods: the need for fre-quent tests to ensure proper chemical concentrationsin the growing medium, the small size of crops pro-duced, the excessive cost of and trouble with equip-ment, algae on crops, and finally, cellular defects andgenetic abnormalities of crops.

Speaking rate. Fast, moderate, and slow tapeswere all delivered by the same person (but not thespeaker used in Experiment 1). The average speechrates were 191, 140, and 111 words per minute inthe fast-, moderate-, and slow-speech condition,respectively. The speaker practiced the delivery ofthe speech six times immediately before taping. Inaddition, the slow-speech version was recorded first.It was hoped that these measures would reducedifferences in patterns of inflection between condi-tions.

Message complexity. The content of the two com-munications was the same. As described above, itargued against the use of hydroponic methods ofgrowing vegetables. In the simple communication theideas were expressed in simple (vs. compound) sen-tences with few parenthetical or qualifying clauses.In the complex communication the very same ideaswere expressed in compound sentences (formed by

SPEED OF SPEECH AND PERSUASION 621

combining the simple sentences of the simple com-munication). In addition, the complex message usedextended qualifying clauses, set in a sequence thatwould be confusing without careful attention andcomplete comprehension.

Dependent measures. After listening to the tape,subjects were asked to indicate orally their agree-ment with a series of attitude statements. Agreementwas indicated on a 10-step scale (1-10), with highernumbers indicating greater agreement. In addition tomeasuring agreement with the speaker, there werequestions to measure perceptions of the speaker'sintelligence, knowledge, and objectivity.

Results

Since no predictions were made regardingmain effects for message complexity (i.e., itwas primarily included to investigate thepresence of Complexity X Speed interactions),message complexity was analyzed by a multi-variate analysis of variance across all de-pendent variables. This analysis indicated nodifferences between complexity conditions. Inaddition, there was no hint of a MessageComplexity X Speaking Rate interaction, aspredicted by the counterargument disruptionhypothesis for measures of persuasion (i.e.,agreement with speaker). These data arguethat speech rate does not affect counter-arguing. In consonance with this interpreta-tion, although Chaiken and Eagly (1976)found that complexity decreased persuasion,they found no relation between their com-plexity manipulation and counterarguing.

In the absence of a significant interaction,the focus should turn to the main effects forspeed of presentation. A faster rate of speechproduced more persuasion: 77(2,84) =4.00,p < .02; Ms = 4.62, 5.95, and 6.10 for low-,moderate-, and high-speed messages, respec-tively. Furthermore, in consonance with Ex-periment 1, faster speech resulted in percep-tions of greater intelligence: F(2, 84) = 2.31,P<A; Ms =5.70, 6.45, and 6.83; greaterknowledge: F(2, 84) .= 2.59, p < .08; Ms =6.46, 6.80, and 7.53; and more objectivity:JF(2,84) =4 .41 , p<.01; Ms = 3.83, 3.93,and 5.76.

Taken together, these results provide strongsupport for the credibility enhancement inter-pretation of speaking rate. They offer no sup-port for the conuterargument hypothesis andessentially replicate the data from the firstexperiment and the Beaber and Miller (Note

1) studies. In addition, the linear progressionof means for both the agreement measure andthe speaker perception measures adds greatergenerality to the earlier studies and arguesagainst curvilinear effects within the normalrange of speech rates.

GENERAL DISCUSSION

The data of these field studies support thenotion that rapid speech functions as a credi-bility cue. In this regard, it is noteworthy thatacross several messages and speakers, as wellas in both laboratory and field settings, rap-idly spoken communications were more per-suasive than those spoken slowly. Indeed, onemight be inclined to assert confirmation of anew law: "Beware of the fast talker." How-ever, it is important to point out that whilethese studies primarily manipulated speechrate and did attempt to control some of theobvious variables such as volume, enthusiasm,etc., that apparently naturally covary with it,in no sense can we claim that our effects areentirely or purely attributable to speed ofspeech. There may yet be residual factors,naturally co-occurring with rapid speech rates,that enhance persuasion. Only with computer-assisted construction of experimental materialscan we hope to isolate speed per se as thecritical factor. In support of our own interpre-tations, however, Smith, Brown, Strong, andRencher (1975) have used computer-assistedmanipulation of speech rate and have shownthat greater competence is attributed to thosewho speak at a faster rate.

It is also interesting that fast speech withinthe normal range apparently does not inter-fere enough with reception to disrupt com-prehension. While we have tended to empha-size the imperviousness of comprehension tospeech rate, it is worth noting that it was notdirectly assessed in the present studies. Conse-quently, an ancillary view is that whereasspeech rate may well affect comprehension(when adequately measured), comprehensionin and of itself is not very important in thepersuasion process.

While we have generally argued that thedata oppose the counterargument disruptionhypothesis, one other interpretation warrantsconsideration, namely, the effort justificationeffect, which is derived from dissonance

622 MILLER, MARUYAMA, BEABER, AND VALONE

theory. Rapid speech may enhance persuasionby increasing the effort required to processand comprehend the speech content ade-quately. The enhancing effects of effort onpersuasion are well documented (Osterhouse& Brock, 1970; Zimbardo, 1965; Zimbardo &Ebbesen, 1970). Baron et al. (1973) arguedthat such effort expenditure need not involvephysical activity but might simply be apsychological or cognitive process. If so, rapidspeech might produce its effect as a conse-quence of the increased effort required tocomprehend it.

Another consideration is that the presentresearch used paradigms in which the listenercould not interact with the speaker. The im-pact of stylistic variables in social interactionparadigms may be quite different. High speechrate may reduce opportunities for response ina conversation by eliminating pauses. Underthese circumstances, a listener may find ahigh speech rate frustrating and consequentlymight resist influence.

Finally, the effect of the manipulation ofcomplexity in Experiment 2 warrants addi-tional consideration. From the standpoint oftraditional learning theory approaches to per-suasion, the failure to find a main effect ofless persuasion with a more complex messageseems puzzling. Indeed, Chaiken and Eagly(1976) find empirical support for thisexpectation. Yet, further reflection on andcomparison of their procedures with our ownsuggests numerous reasons for the discrepancybetween findings and, further, suggests cir-cumstances under which one might expect ahighly complex message to reduce persuasion.Complexity should interfere with persuasionin circumstances where comprehension of con-tent critically affects the persuasion process.The following are circumstances that mightmaximize the importance of comprehension.

1. The communication topic should disin-cline one to rely on expert opinion. Rather,one's opinion should rest largely on one's ownvalues in combination with an understandingof the relevant facts. The facts should not betoo technical but relatively easy for the lis-tener to understand and should have a mean-ing instrumental to the appropriate implemen-

tation of one's values when forming a finalopinion on the topic.

2. The communication must be lengthy.The number of facts must be sufficiently sub-stantial, detailed, and interrelated to leaveone puzzled about how one's values apply tothe "proper" concluding opinion, unless onedoes possess a fairly complete understandingof the facts regarding the issue.

3. Finally, one's task should not simplyrequire that one offer an opinion. Instead,stemming from one's opinion, there should beaction or behavioral consequences, which ifnot intended for oneself, should apply to oth-ers. If there are no direct and personal deci-sion consequences, the topic of communicationshould at least be one that is similar to thosefor which there were decision consequencesin the past.

When we consider the topic used by Chai-ken and Eagly (1976) and compare it withour own, our failure to find complexity effectsbecomes more understandable. Our topic, thedanger of hydroponically grown vegetables,was novel to most respondents. It argued,largely on the basis of unfamiliar technicalinformation, for a conclusion that the listeneressentially had to accept on faith. Further, itslength was brief (approximately 300 words).On the other hand, their communication, the"Victoria Company Case," was preceded bya handout that provided background infor-mation concerning a dispute between thecompany and its newly organized union em-ployees over the fact that the managementhad failed to provide the Christmas bonusthat the workers had typically received. Forthis preliminary information, presenting bothunion and company views, subjects were givenapproximately 5 minutes to process the infor-mation. Then they received the "case discus-sion" of a law student who was presumablyplaying the role of the lawyer brought in toarbitrate the case. This "case discussion"constituted the persuasive communication andwas at least three times the length of our ownpersuasive appeal. Finally, it is important tonote that the subjects were led to believethat they were assisting the psychology de-partment in evaluating the ability of studentsat a New England law school to argue legal

SPEED OF SPEECH AND PERSUASION 623

cases. In other words, their expressed opinionwould have action consequences for others*

The procedure used by Chaiken and Eagly(1976) to manipulate complexity largelyparalleled our own. Yet, from the standpointof our preceding arguments, it should be ap-parent that adequate information processing(comprehension) should play a much morecentral role in determining subjects' accept-ance of the recommended conclusion in theircase than in our own. In summary, whereasour field experiments may reflect generaltendencies of persons to avoid complex cogni-tive effort and form judgments based uponsimple decision rules (e.g., the speaker seemstrustworthy vs. untrustworthy, etc.), thereare other instances in which judgments aresufficiently important to ensure that neces-sary cognitive effort will be expended (e.g.,buying a car, being on a jury). In these latterinstances, there are apparently some specificsituations (as described above) in whichcomplexity does affect persuasion.

In conclusion, it is important to note thatthe three hypothesized consequences of afast speaking rate explored in this researchreflect two views of the persuasion process,one emphasizing rationality or the importanceof detailed information processing and theother deemphasizing it. The treatment of thepersuasion process by social psychologists hastended to take a rationalistic view of man(e.g., Festinger & Maccoby, 1964; Hovland,Janis, & Kelly, 1953; Jones & Gerard, 1967;McGuire, 1969). More recently, Miller (Mil-ler, 1968; Miller & Baron, 1973) has arguedthat this view may require modification, withgreater weight being accorded instead to emo-tional and evaluative cognitions—at the ex-pense of information, content-oriented cogni-tions.

If the impact of speaking rate on attitudechange were mediated by either comprehen-sion effects or counterargument disruption,then a highly rationalistic view of the per-suasion process would have been supported.However, the findings, which support thecredibility interpretation of speaking rate, areconsistent with a less rationalistic view of thepersuasion process. Indeed, to put it somewhatsardonically, it may be irrational to rationallyscrutinize the plethora of counterattitudinal

messages received daily. To the extent thatone possesses only a limited amount of infor-mation-processing time and capacity, suchscrutiny would disengage the thought pro-cesses from the exigencies of daily life*

REFERENCE NOTE

1. Beaber, R. J., & Miller, N. The effects of speechrate on attitude change. Unpublished manuscript,University of Southern California, 1974.

REFERENCES

Allen, R. R., Anderson, S., & Hough, J. Speech inAmerican society. Columbus, Ohio: Charles E.Merrill, 1968.

Baron, R. S., Baron, P. H., & Miller, N. The rela-tion between distraction and persuasion. Psycho-logical Bulletin, 1973, 80, 310-323.

Bettinghaus, E. P. Operation of congruity in anoral communication setting. Speech Monographs,1961, 28, 131-142.

Bowers, J. W. The influence of delivery on attitudestoward concepts and speakers. Speech Monographs,1965, 32, 154-158.

Chaiken, S., & Eagly, A. H. Communication modal-ity as a determinant of message persuasiveness andmessage comprehensibility. Journal of Personalityand Social Psychology, 1976, 34, 605-614.

Dietrich, J. E. The relative effectiveness of twomodes of radio delivery in influencing attitudes.Speech Monographs, 1946, 13, 58-65.

Eagly, A. H. Comprehensibility of persuasive argu-ments as a determinant of opinion change. Jour-nal of Personality and Social Psychology, 1974, 29,758-773.

Festinger, L., & Maccoby, N. On resistance to per-suasive communications. Journal of Abnormal andSocial Psychology, 1964, 68, 359-366.

Foulke, E., & Sticht, T. G. A review of research onthe intelligibility and comprehension of accel-erated speech. Psychological Bulletin, 1969, 72,50-62.

Hovland, C. I. Reconciling conflicting results de-rived from experimental and survey studies ofattitude change. American Psychologist, 1959, 14,8-17.

Hovland, C. I., Janis, I. L., & Kelly, H. H. Com-munication and persuasion. New Haven, Conn.:Yale University Press, 1953.

Jones, E. E., & Gerard, H. B. Foundations of So-cial Psychology. New York: Wiley, 1967.

Kiesler, S. B., & Mathog, R. B. Distraction hy-pothesis in attitude change: Effects of effective-ness. Psychological Reports, 1968, 23, 1123-1133.

Lumley, F. H. Rates of speech in radio speaking.Quarterly Journal of Speech, 1933, 19, 393-403.

McGuire, W. J. The nature of attitudes and attitudechange. In G. Lindzey & E. Aronson (Eds.), Thehandbook of social psychology (Vol. 3, 2nd ed.)«Reading, Mass.: Addison-Wesley, 1969.

Miller, G. R., & Hewgill, M. A. The effects of vari-ations in nonfluencies on audience perceptions of

624 MILLER, MARUYAMA, BEABER, AND VALONE

source credibility. Quarterly Journal of Speech,1964, SO, 36-44.

Miller, N. As time goes by. In M. R. Abelson, E.Aronson, W. J. McGuire, T. Newcomb, M. J.Rosenberg, & P. H. Tannenbaum (Eds.), Theoriesof cognitive consistency: A sourcebook. Chicago:Rand-McNally, 1968.

Miller, N., & Baron, R. S. On measuring counter-arguing. Journal for the Theory of Social Be-havior, 1973, 3, 101-118.

Monroe, A. H., & Ehninger, D. Principles and typesof speech communication (7 th ed.)« Glen view,111.: Scott Foresman, 1974.

Osterhouse, R. A., & Brock, T. C. Distraction in-creases yielding to propaganda by inhibiting coun-

terarguing. Journal of Personality and Social Psy-chology, 1970, IS, 344-358.

Smith, B. L., Brown, B. L., Strong, W. J., & Ren-cher, A. C. Effects of speech rate on personalityperception. Language and Speech, 1975, 18, 145-152.

Zimbardo, P. G. The effect of effort and improvisa-tion of self-persuasion produced by role playing.Journal of Experimental Social Psychology, 1965,1, 103-120.

Zimbardo, P. G., & Ebbesen, E. B. Experimentalmodification of the relationship between effort,attitude, and behavior. Journal of Personality andSocial Psychology, 1970, 16, 207-213.

(Received December 22, 1975)