rockwell et al. (1997) the voice of deceit. refining and expanding vocal cues to deception

Upload: rost-alius

Post on 06-Jul-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    1/10

    The V oice

     of

     Deceit - Page 451

    The Voice of Deceit Refining and

    Expanding Vocal Cues to Deception

    Patricia Rockwell David  B. Buller

    University of Southwestern Louisiana AMC Cancer Research Center

    Judee

     K.

     Burgoon

    University of Arizona

    TTH'S study examined

      the

     nature

      of

     deceptive vocal behavior

     in

      interactive

    situations.  Data from

      an

     earlier study were used

      to

     conduct

      a

     detailed analysis

    of vocal features

      of

     deceptive speech.  Vocal samples were analyzed perceptually

    and acoustically.

     Of

     three categories examined (time, frequency,

      and

      intensity),

    the time variables best discriminated between truthful  and deceptive speakers,

    with deceivers exhibiting shorter message length, longer response latencies,

    slower tempo,

      and

      less fluency than truthtellers. Deceivers also evidenced

    increased intensity range, increased pitch variance,

      and

      less pleasant vocal

    quality than truthtellers.

    Vocal behavior has been found to be a major factor in discriminating between

    deceptive and truthful speakers (DePaulo, Stone, Lassiter, 1985; deTurck & Miller, 1985;

    Zuckerman, DePaulo, Rosenthal, 1981). However, the findings regarding individual

    components of deceptive voices have not always been consistent across studies (Cody,

    Marston, Foster, 1984; Matarazzo, Wiens, Jackson, Manaugh, 1970; Motley, 1974;

    Patricia Rockwell

      (Ph.D.,University  of  Arizona,  1994) is an  Assistant Professor, Department of

    Communication, University  of  Southwestern Louisiana, Lafayette,  LA 70504.  David  B.  Buller

    (Ph.D.,Michigan State University, 1984)is a Senior Scientist, AMC Cancer Research Center, Denver,

    CO.  Judee

      K.

      Burgoon  (Ed.D., West Virginia University,  1974) is a  Professor, Department  of

    Comm unication, University of Arizona, Tucson, AZ 85721. This project w as su pported by the  U.S.

    Army Research Institute (Contract #MDA903-90-K-0113) and th e

     U.S.

     Army Research Office (Grant

    #30235-RT-A

     AS).

     T he views, opinions, and findings in this report are those of the authors and should

    not be construed as an official Departm ent of the A rmy position, policy, or decision.

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    2/10

    Page 452 -

     Communication Research Reports/Fall 1997

    O'Hair,

     Cody,

     

    McLaughlin, 1981). Some deception studies have used acoustic measures

    to

     de termine vocal features

     (e.g.,

     Ekman, CS ulliva n, Friesen, & Scherer,

     1991;

     M otley, 1974),

    but the overwhelming majority have used perceptual measures (e.g., Buller

     

    Aune , 1987;

    Buller, Comstock, Aune, & Strzyzew ski, 1989; Buller, Strzyzewski, & C omstock, 1991; Cody

    et al, 1984; deTurck & Miller, 1985; Hock ing & Leathers, 1980). Occasionally, decep tion

    researchers have used both percep tual and acoustic measures in one study (e.g., an acoustic

    measure of fundamental frequency and a perceptual measure of tempo in Zuckerman &

    Driver, 1985), but no stud ies could be located that measu red vocal features both acoustically

    and perceptually. The pvupose of the present s tudy is to determ ine which vocal behaviors

    best discriminate between truthful and deceptive vocal behavior using both acoustic and

    perceptual measures.

    Vocal behaviors traditionally examined in deception research tend to fall into three

    broad categories as delineated by Scherer (1985). They consist of time, frequency, and

    intensity.

    Time.

     Vocal features concerned w ith the length or speed of an utterance, either vocalized

    or unvocalized, may be considered a com ponent of time . One consistent finding is that of

    shorter overall message duration for deceivers (deTurck & Miller, 1985; Knapp &

    Comadena, 1979; Mehrabian,

      1971;

     Zuckerman & Driver, 1985). The theoretical basis for

    expecting shorter messages in deception may stem from the deceiver's reticence to offer

    more information than necessary.

    A less frequently reported finding is that of slower overall tempo which is a measure

    typically derived from a count of the total number of words or syllables in the utterance

    com pared to the total time of the utterance . Mehrabian (1971) reported slower tem pos for

    deceivers than for truthtellers. The finding of slower tem pos

     is

     consistent with the argu m ent

    that deception increases cognitive difficulty.

    Deception researchers regularly report decreased fluency for deceivers (deTurck &

    Miller, 1985; Marston, 1920; Matarazzo et al., 1970). One commonly noted deception

    disfiuency is the response latency, a pause immediately preceding a deceptive answer

    (deTurck

     

    Miller,

     1985;

     Kraut,

     1978;

     Matarazzo et al., 1970). These various tim e findings

    pro m pt the following h ypotheses:

    H I:  Com pared to truthtellers, deceivers will exhibit: shorter overall

    message duration,

    H2:

      slower tempos,

    H3:

      less fluency, and

    H4:  longer response latencies.

    Frequency.

      The_most_frequently_ reported finding in deception research regardirig

    frequency_is.that deceptive voices increase in general pitch level, or fundam enta l frequency,

    w ith most researchers arguing that increased pitch level

     is

     prom pted by decep tion-induced

    arousal (Apple, Streeter, & Kraus, 1979; H ocking & Leathers, 1980; Streeter, Kraus, Geller,

    Olson,

     

    Apple,

     1977;

     Zuckerman

     

    Drive r, 1985).

    Despite repeated findings, some researchers believe that other measures of pitch, such

    as pitch range or pitch variety, may

     be

     better indicators of deception than p itch level because

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    3/10

    The

     Voice

     of

     Deceit - Page 453

    following hypothesis and research questions regarding frequency are offered:

    H5:

      Comp ared to truthtellers, deceivers will exhibit high er pitch

    levels.

    RQ l: Comp ared to truthtellers, will deceivers exhibit less pitch variety?

    and

    RQ2:

      A narrow er pitch range?

    Intensity.

      Intensity, or loudness, is not studied as often as time and frequency in

    deception research (often due to inadequate recording equipment). Mehrabian (1971)

    repo rted decreased intensity for deceivers. Buller and A une

     (1987)

     measured loudness

     as

     an

    aspect of vocal activity and found th at this composite variable increased for deceivers in

    certain conditions . Findings from research on emo tion suggest that vocal intensity may

    reveal speakers' sincerity or confidence (Alpert, Kurtzberg, &

     Friedhoff,  1963;

     Frick, 1985;

    Kimble & Seidel,

     1991;

     Scherer, 1978). Even so, it is prem atu re to pred ict the direction or

    deg ree of intensity change, so the following research question is presented :

    RQ3:

      ^W hat changgjun intensity are indicative of deceptive behavior?

    Other vocal features.

      Few other vocal features have been examined as possible cues to

    decep tive voices. Thus, it is difficult to hypothesize w hich behav iors may occur. Find ings

    outside of the deception literature suggest potential vocal cues to deception. For instance,

    research on the vocal features of emo tion have dete rmin ed that articulation (as encoded as

    formant frequency) may be associated with perceptions of different emotions (Ross,

    Edmondson, & Seibert, 1986; Scherer, 1978; 1992, Scherer & Oshinsky, 1977) with the

    emotions of pleasantness, happine ss, and surprise exhibiting more clipped articulation tha n

    those of boredom, disgust, and sadness. Deceptive speakers may experience emotions

    differently from those they might experience if they were truthful and may expose these

    differences in their articulation.

    Likewise, vocal quality (comprised of those vocal behaviors moderated by speakers'

    physiology such as amount of resonance, breathiness, or nasality) may change during

    deception. Scherer

     (1985)

     claims tha t a less pleasant voice quality may be an excellent cue to

    deception because it audibly reflects the vocal tension that may accompany deception-

    induced arousal. When a deceiver experiences arousal, the vocal mechanism may und ergo

    certain physiological changes such as increased tension of the vocal folds and disrupted

    breathing, causing the speake rs' vocal quality to sound less resonant and relaxed, and thus

    less pleasant (Scherer, 1978). Therefore, the following research question is posed:

    RQ4: W hat other features of the voice, if any, distinguish between

    deceivers and truthtellers?

    METHOD

    Data for the present study w ere dra w n from an earlier stud y (Buller, Burgoon, Buslig, &

    Roiger,

     1993).

     The original project utilized a2 X 3 X 2X 2 factorial design. Variables analyzed

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    4/10

    Page 454 - Communication Research Reports/Fall 1997

    For the present study, videotapes of speakers telling the truth and lying in response to

    interview-type questions were used for coding vocal behaviors. C oders evaluated speakers'

    vocal characteristics on seven-point L ikert-type scales, with

      1

    representing a low level of

    a particular feature and

      7

    representing a high level of a particular feature. Vocal

    characteristics analyzed included response latencies (short-long), tempo (slow-fast), fluency

    (not fiuent-fluent), internal pauses (few-many) (short-long), pitch level (low-high), pitch

    range (narrow-wide), pitch variety (unvaried-varied), intensity level (soft-loud), intensity

    range (narrow-wide), intensity variety (unvaried-varied), vocal quality (unpleasant-

    pleasan t), and articulatory precision (imprecise-precise).

    Coders (n=6) were all junior or senior unde rgrad uates majoring in Com mun ication at a

    large southwestern university. Coders received over four hours of training on the vocal

    behaviors they were assigned to code, including actual practice in coding vocal behaviors

    from other experiments. The audio portions of the videotapes were content-filtered to

    prev ent coders from being irvfluenced by the verbal content. Coders wo rked in pairs, w ith

    one pair coding time and voice quality, another pair coding pitch and p auses, and a third

    pair coding intensity and articulation. Interrater reliabilities were generally high (ranging

    from a low of

     .85

     to a high of

     .97),

     indicating that coders w ere successful in agreeing on the

    features of voices.

    The tapes we re then submitted to acoustic analysis using software designed

     to

     evaluate

    characteristics of sound signals.^ Features that were analyzed included total amount of

    sound, number of sound segments, number of silent segments, total amount of silence,

    duration of the initial pause before each question, fundamental frequency mean (in HZ),

    fundamental frequency range (in HZ), fundam ental frequency variance, intensity mean (in

    dB.),

     intensity range (in

     dB.),

     and intensity variance.

    Also included in both the perceptual and the acoustic analyses were total length and

    num ber of syllables in each segment as determ ined by auditory co un t These two measures

    were used to determine composite measures of temp o and fluency.

    RESULTS

    To begin the analysis, intercorrelations w ere ru n on the entire set of acoustic variables

    and on the entire set of perceptual variables. An examination of the correlations revealed

    high correlations among the acoustic measures of length, total sound, number of sound

    segments, total silence, and number of silent segments (r>.48); pitch level, range, and

    variance (r>.88); and intensity level, range, and variance (r>.62). This examination also

    revealed high correlations among the perceptual measures of tempo, pause number, and

    pause length

     (r>.44);

     pitch level, range, and variance

     (r>.84);

     and intensity level, rang e, and

    variance (r>.86). Alpha reliability tests verified these grouping s.

    These group s of variables were then subm itted to multiva riate analysis of variance. In

    instances where the analyses produced singular variance-covariance matrices, variables

    were removed from the cluster according to their alpha reliability scores, wi th the va riable

    exhibiting the lowest alpha reliability score in the cluster being removed first. When the

    analysis no longer produced singular variance-covariance matrices, removal of variables

    ceased. From the use of

     this

     criterion, it was d etermined that

     the

     following grou ps of acoustic

    variables could be successfully analyzed together: (1) length, total sound, an d nu mber of

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    5/10

    The Voice of Deceit - Page 455

    sound), fluency (number of silent segments/total silence), initial pauses (response

    latencies), and fundamental frequency variance were analyzed separately.

    These tests also revealed that the following groups of perceptual variables could be

    analyzed together:

     (1)

     intensity level, range, and variance (a =.96), and

     (2)

     pitch level, range,

    and variance ( a =.96).

     The

     perceptual variables of message length (length minus the p rodu ct

    of num ber of pause s and length of pause s), tempo , fluency, voice quality, and articulatory

    precision were analyzed separately.

    MANOVA was used in instances where vocal behaviors were correlated with one

    another as determined by high intercorrelations, high alpha reliability scores, and the

    Bartlett test of sphericity. This procedu re pro duc ed significant results for three of the five

    MAN OVAs. For those variables tha t did not d emo nstrate high intercorrelations, ANOVAs

    we re performed on each variab le separately. For variables with significant un ivariate scores

    derived from significant MAN OVAs (two out of

     six)

     and for those variables that prod uced

    significant results from the ANOVAs (seven out of 11), planned comparisons with 1

      df

    contrasts tested the differences between the truthful answer condition and the deception

    answ er condition in order to pro vide answers to Hypothese s 1-5 and Research Questions 1-

    4.

      When violations of symmetry conditions occurred, as dete rmin ed by the Box-M test for

    homogeneity of variance and Mauchly's sphericity test, the Huynh-Feldt correction was

    employed to adjust the p value . Results of the planned com parisons for the nine significant

    individual measures are reported in Table 1.

    ,

      DISCUSSION

    A major goal of the current study was to determine which vocal behaviors best

    TABLE 1

    Comparison of Significant Vocal Measures in Discriminating Between Truthtellers and Deceivers

    Message Length

    Number of Sounds

    Tempo

    Fluency

    Response Latency (acoustic)

    Response Latency (perceptual)

    Pitch Variety

    Intensity Range

    Vocal Quality

    Deception

    Mean

    1.90

    1.70

    4.30

    4.49

    11 1

    2.69

    24.02

    8.68

    4.18

    Tmth

    Mean

    3.61

    2.32

    4.56

    4.85

    1.33

    2.00

    21.57

    7.33

    4.32

    F-score

    9.37

    4.86

    7.10

    5.35

    21.13

    13.05

    6.99

    6.58

    5.23

    g-value

    .005

    .015

    .012

    .028

    .000

    .001

    .013

    .015

    .029

    eta

    squared

    .11

    .18

    .19

    .15

    .41

    .30

    .18

    .18

    .14

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    6/10

    Page

     456 -

     Communication Research Reports/Fall 1997

    discriminated between truthtellers and deceivers.  As with p revious deception studies, the

    findings of this study were complex and not wholly expected in some respects. Three

    categories of vocal behavior (time, frequency, and intensity) wer e cons idered .

    The time category revealed the greatest number of significant differences betw een truth

    and deception, including shorter messages, fewer so unds , slower tempo , less fiuency, and

    longer response latencies for deceivers. These cues refiect the im pact of cognitive difficulty

    on deceiver vocal behavior. It seems understandable that deceivers may give shorter

    responses because producing longer messages wo uld require greater cognition and perh aps

    increase leakage cues. Also, longer messages provide more information from which

    receivers m ight jud ge sende r honesty. Thu s, deceivers may op t strategically for reticence

    (DePaulo et

     al., 1985;

     O'Hair et

     al., 1981;

     Zuckerman

     

    Driver,

     1985;

     Zuckerman et

     al.,

     1981)

    Slower tempo may

     be

     an overlooked byp roduct of the increased cogn itive load requ ired

    to genera te false information, signal uncertainty , decide which information to divu lge an d

    to hide, and adapt to conversational events (Mehrabian, 1971). It may also be a strategic

    move by deceivers to manage their image and insure fiuency. From a physiological

    perspective, increased arousal may prevent the normal, efficient functioning of the

    articulators, forcing speakers to slow down in order to be comprehensible (deTurck &

    Miller, 1985).

    Certainly, deceivers' lack of fiuency has been a prominent finding in most deception

    studies (deTurck & Miller, 1985; Matarazzo et al., 1970; Miller et al., 1983; Zuckerm an &

    Driver, 1985). Researchers have argu ed that

     as

     deceivers work to create lies and experience

    unpleasant reactions to deceiving, they may lose control of their normal fiuency  (̂ Motley

    1974; Zuckerman et

     al.,

     1981). The large effect sizes noted for both acoustic and perce ptual

    measures of response latencies (Table 1) provide a dramatic indication that response

    latencies are one of the most telltale indicators of decep tive speech. Researchers (deTurck

     

    Miller, 1985; Kraut, 1978; Matarazzo et al., 1970) argue that longer response latencies

    typically noted for deceivers are due to increased cognitive load, allowing deceivers

    additional time to prepare successful deceptive answers. Deceivers also may be behaving

    strategically when they increase response

     latencies,

     by seeming to pon der

     a

     question before

    answering.

    Unlike many other deception studies, this study did not find increased pitch levels for

    deceivers; however, the acoustic analysis showed that deceivers exhibited significantly

    more pitch variety than

     truthtellers.

     This may have occurred because deceivers strategically

    increased their pitch variety to create an impression of a norm ally expressive, truthful voice

    (Scherer, 1985). Few studie s have considered pitch variety as a discriminator betw een truth

    and deceit, and if the present results portray an accurate picture of what deceivers do

    vocally, pitch variety may prove to be a better indicator of deceit than pitch level. Indeed ,

    reports of increased pitch level in previous deception studies may actually have been

    increased pitch variety.

    Deceivers showed a wider intensity range than truthtellers at least on the acoustic

    measures. Although it

     is

     possible that deceivers increased their range of loudness values in

    order to m ake their voices sound mo re norm ally exp ressive (Scherer, 1978; Scherer &

    Oshinsky, 1977), a more logical explanation m ay be that deceivers were not able to control

    fiuctuations in intensity because increased arousal caused increased vocal tension and

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    7/10

    The Voice

     of Deceit

     - Page 457

    Deceivers' voice quality was less pleasant than truthtellers'. Voice quality is far less

    controllable than other vocal behaviors typically studied in deception research such as

    tempo and response latencies (Scherer et al.,

     1985;

     Zuckerman et al., 1981) because it is mo re

    tied to the physiological make-up of the speaker. Thu s it is more difficult for speakers to

    modify than features rnore under the speaker's conscious control such as strength and

    length of sound produced. Hence, changes in voice quality may be more likely during

    deception than changes in tempo or intensity.

    In summa ry, the deceiver's voice reveals mu ch abo ut the deception in progress. This

    study represents an attemp t

     to

     refine and expan d the audito ry p icture of the deceptive voice.

    By expanding the number and type of vocal cues measured and the methods of

    measurem ent, future studies may m ake this picture even m ore clear. Certainly time cues are

    imp ortant, b ut researchers need to mo re carefully exam ine frequency, intensity, and vo ice

    quality as well.

    NOTES

    ' Results of a comparison of the perceptual and acoustic methods used in this

    study

     are

     reported

     in

      Measurement of

     deceptive voices:

     Comparing

     acoustic

     and

    perceptual data,

    pplied Psycholinguistics

     (in press).

    REFERENCES

    Alpert,

     M.

     S.,

     Kurtzberg,

     M. A.,

      Friedhoff,

     A.

     J.

     (1963). Transient voice change

    associated with emotional stimuli.

      rchives

     of

     General Psychiatry, 8,

    326-365.

    Apple,

     W., Streeter,

     L. A.,

     

    Kraus,

     R.

     M. (1979). Effects of pitch and speech rate

    on personal attributions. Journal of Personality and Social Psychology, 37,715-727.

    Barge, J. K., Schlueter, D . W., & Pritchard , A. (1989). The effects of no nverb al

    communication and gender on impression formation in opening statements.  The

    Southern Communication

     Journal

    54 ,330-349.

    BuUer, D. B.,

     

    Aune, R. K. (1987). N onverbal

     cues to

     deception amo ng intimates,

    friends, and strang ers. Journal of Nonverbal Behavior 11 ,269-290.

    Buller, D. B., Burgoon, J. K., Buslig, A., & Roiger, J. (1993).

      The

     language

      of

    interpersonal  deception.  Paper presented to the annual convention of the Inter-

    national Comm unication Association, Washington, D. C.

    Buller, D.

     B.,

     Com stock,

     J.,

     Aune,

     R.

     K.,  Strzyzewski,

     K.

     D. (1989). The effect of

    probing on deceivers and truthtellers. Journal of Nonverbal Behavior 13 ,155-170.

    Buller,

     D. B.,

     Strzyzewski,

     K. D.,

      Comstock,

     J.

     (1991). Interpe rsona l D eception

    I. Deceivers' reactions to receivers' suspicions and probing.  Communication

    Monographs, 58,1-24.

    Burgoon, J. K., Birk, T.,

     

    Pfau,

     M .

     (1990). No nverbal behaviors, persuasion, and

    credibility. Human Communication Research,

     17 ,

    140-169.

    Cody, M. J., Marston, P. J., & Foster, M. (1984). Deception: Paralinguistic and

    verbal

     leakage.

     In R. Bostrum (Ed.),

     Interpersonal Communication Yearbook 8.

     Beverly

    HiUs,

     CA: Sage.

    DePaulo, B. M., Stone, J. I., & Lassiter, G. D. (1985). Deceiving and detecting

    deceit. In B. R. Schlenker (Ed.),  The self and social  life (pp. 323-370). New York:

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    8/10

    Page 458 - Com mun ication Research R epo rts/Fa ll 1997

    McGraw-Hill.

    deTurck, M. A., & Miller, G. R. (1985). Deception and arousal: Isolating the

    behavioral correlates to deception. Human Communication Research, 12 ,181-201.

    Ekman,

     P., O'Sullivan,

     M., Friesen,

     W.

     V.,  Scherer,

     K.

     R.

     (1991).

     Invited article:

    Face, voice, and body in detecting deceit.

     Journal

     of

     Nonverbal Behavior

    15 ,125-135.

    Frick, R. W. (1985). Commimicating emotion: The role of prosodic features.

    Psychological BulleHn, 97,412-429.

    Hocking, J. E., & Leathers, D . G. (1980). Nonv erbal indicators of deception: A

    new theoretical perspective. CommunicaHon Monographs, 47,119-131.

    Kimble, C. E., & Seidel, S. D. (1991). Vocal signs of confidence.  Journal  of

    Nonverbal Behavior 15 ,99-105.

    Knapp, M. L.,  Com adena, M. E. (1979). Telling it like it isn't: A review of theory

    and research on deceptive communications. Human Communication Research, 5,270-

    285.

    Kraut, R. E. (1978). Verbal and no nverbal cues in the perception of lying. Journal

    of Personality and Social Psychology,

     36,

    380-391.

    Marston, W. N. (1920). Reaction-time symptoms of deception.  Journal of

    Experimental Psychology, 3 72-87.

    Matarazzo, J. D., Wiens, A. W., Jackson, R. N., & Manaugh, J. S. (1970).

    Interviewer speech behavior under condition of endogenously-present and

    exogenously-induced m otivational state. Journal of Clinical Psychology, 2 6,141-148.

    Mehrabian, A. (1971). Nonverbal betrayal of feeling.  Journal  of Experimental

    Research in Personality, 5,64-73.

    Motley, M. (1974). Acoustic correlates of

     lies.

     Western

     Speech,

     38,81-87.

    O'Hair,

     H.

     D.,

     Cody, M.

     J.,

      McLaughlin, M.

     (1981).

     Prepared

     lies,

     spontaneous

    Ues,  Machiavellianism, and nonverbal communication.  Human

      Communication

    Research, 7 325-339.

    Ross, E. D., Edm ondson, J. A., & Seibert, G. B. (1986). The effect of affect on

    various acoustic measures of prosody in tone and non -tone languages: A com par-

    ison based on computer analysis of voice. Journal of Phonetics,  14 , 283-302.

    Scherer, K. R. (1978). Personality inference from

     voice

     quality: Tlie loud voice

     of

    extroversion. European Journal of Social Psychology, 8,467-487.

    Scherer, K. R. (1985). Methods of research on vocal comm unication: Para digm s

    and param eters. In

     K.

     R. Scherer 

    P.

     Ekman (Eds.), H andbook of methods in nonverbal

    behavior research (pp . 136-198). Cam bridge : Cambridge University Press.

    Scherer, K. R. (1992). Vocal affect expression as sym ptom , symbol, and app eal .

    In H. Papousek, U. Jurge ns, & M. Papousek (Eds.), Nonverbal vocal communication:

    Comparative  and  developmental approaches  (pp. 43-60). New York: Cambridge

    University Press.

    Scherer, K. R., & Oshinsky, J. S. (1977). Cue utilization in emo tion attribu tion

    from auditory stimuli. Motivation and Emotion,

      1,

     331-346.

    Streeter, L. A., Kraus , R. M., Geller, V., Olson, C , & Apple, W. (1977). Pitch

    change during attempted deception. Journal of Personality and Social Psychology, 35,

    345-350.

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    9/10

    The Voice of Deceit - Page 459

    psychology, V. 14 .  (pp. 1-59). New York: Academic Press.

    Zuckerman, M., & Driver, R. E. (1985). Telling lies: Verbal and nonverbal

    correlates of decep tion. In

     W.

     Siegman 

    S.

     Feldstein

     (Eds.),

     Multichannel integration

    of nonverbal behavior (pp. 129-148). Hillsdale, NJ: Erlbaum.

  • 8/18/2019 Rockwell Et Al. (1997) the Voice of Deceit. Refining and Expanding Vocal Cues to Deception

    10/10