Kellaris, J. J. - An Exploratory Investigation of Responses Elicited by Music Varying in Tempo, Tonality, And Texture

Lawrence Erlbaum Associates (Taylor & Francis Group)

An Exploratory Investigation of Responses Elicited by Music Varying in Tempo, Tonality, and Texture Author(s): James J. Kellaris and Robert J. Kent Source: Journal of Consumer Psychology, Vol. 2, No. 4 (1993), pp. 381-401 Published by: Lawrence Erlbaum Associates (Taylor & Francis Group) Stable URL: http://www.jstor.org/stable/1480509 Accessed: 01/05/2009 09:41 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=lebtaylorfrancis. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

Lawrence Erlbaum Associates (Taylor & Francis Group) is collaborating with JSTOR to digitize, preserve and extend access to Journal of Consumer Psychology.

http://www.jstor.org

JOURNALOF CONSUMERPSYCHOLOGY,2(4), 381-401 Copyright? 1994,LawrenceErlbaumAssociates,Inc.

An ExploratoryInvestigationof Responses Elicited by Music Varying in Tempo, Tonality, and Texture JamesJ. Kellaris Universityof Cincinnati

Robert J. Kent University of Delaware

This study explores listeners' responses to music as a function of objective properties of musical sound. Original classical and pop-style compositions were produced with digital sound technology to provide orthogonal manipulations of musical tempo, tonality, and texture. Three dimensions (pleasure, arousal, surprise) emerged from the responses measured. Variance analyses found main effects of tempo on both pleasure and arousal, and main effects of tonality on pleasure and surprise. Texture moderated the influence of tempo such that a positive contribution to pleasure was observed for classical music, and a positive contribution to arousal was observed for pop-style music. Texture also moderated the influence of tonality on pleasure, with more pronounced reactions to tonal variations observed among listeners exposed to classical music. Implications for the use of music in consumer research and the interpretation of past music-relatedfindings are discussed.

Music touches the lives of consumers in many contexts. It is frequently heard as a background feature in stores and offices (Kotler, 1973-1974), in films (Seidman, 1981), in elevators (Morrow, 1991), and in advertisements (Scott, 1990). Music is also an aesthetic product (Lacher, 1989), the consumption of which constitutes a growing, multibillion dollar industry (Gottlieb, 1991). Given this pervasiveness,it is not surprisingto find that the effects of music have been considered in many areas of consumer research, including hedonic consumption and consumer aesthetics (Holbrook & Anand, 1990, 1992; HolRequestsfor reprintsshouldbe sentto JamesJ. Kellaris,433 CarlH. LindnerHall,Department of Marketing,Collegeof BusinessAdministration,Universityof Cincinnati,Cincinnati, OH 45221-0145.

382

KELLARIS AND KENT

brook & Schindler, 1989), mood (Alpert & Alpert, 1989, 1990; Miniard, Bhatla, & Sirdeshmukh,1992), nostalgia (Holbrook & Schindler, 1991), classical conditioning (Bierley, McSweeney, & Vanieuwkerk, 1985; Gorn, 1982; Kellaris & A. D. Cox, 1989), persuasion (Galizio & Hendrick, 1972), attitude formation (Park & Young, 1986), message processing (Anand & Sternthal, 1990; Kellaris, A. D. Cox, & D. Cox, 1993; Maclnnis & Park, 1991; Wallace, 1991), time perception (Kellaris & Altsech, 1992; Kellaris & Kent, 1992), consumption duration (Holbrook & Gardner, 1993), and retail atmospherics (Milliman, 1982, 1986; Smith & Curnow, 1966; Yalch & Spangenberg, 1990). Undeniably, music occupies a place of importancein the lives of consumers. It is commercially significant to marketers and is of increasing interest to consumer researchers.Nevertheless, a clear picture of how music affects consumers has yet to emerge. Several inconsistent or perplexing music-related findings can be found in the research literature. For example, correlational studies have found the presence of music to enhance (e.g., Hoyer, Srivastava, & Jacoby, 1984), inhibit (e.g., Haley, Richardson, & Baldwin, 1984; Sewall & Sarel, 1986), and have no effect (e.g., Stewart & Furse, 1986) on ad performance. Similar inconsistencies have been observed across studies of sung versus spoken messages (e.g., see Galizio & Hendrick, 1972;Wallace, 1991). The role of music as an unconditioned stimulus in classical conditioning studies remains controversial (e.g., Gorn, 1982; Kellaris & A. D. Cox, 1989). Theoretically, positively valenced music should foster good moods, which, in turn, should encourage positive evaluations and behaviors (Gardner, 1985). Yet, interestingly, Alpert and Alpert (1990) found sad music to produce the most positive behavioral intent in their advertising study. We propose that some of the obfuscation regardingthe influence of music on consumers can be mitigated by a better understandingof the multidimensional nature of both musical stimuli and listeners' responses, and by an appreciation of the role of objective properties of sound as antecedents of emotional outcomes. In this study, we briefly explicate the nature of musical sound, then empirically examine three dimensions of response (pleasure, arousal, surprise) elicited by three basic musical properties (tempo, tonality, texture). The results of our exploratory experiment should be of interest to parties wishing to design or select appropriate music for use in research or commercial applications. OBJECTIVEPROPERTIESOF MUSIC Music is commonly defined as the art of organized sound, the purpose of which is to elicit an aesthetic response in listeners (Apel, 1973). Musical sound is multidimensionalin nature. Although there is presentlyno accepted taxonomy of musical variables, music theoreticians generally agree about the objective

LISTENERS'RESPONSESTO MUSIC

383

properties of music and their underlyingdimensions. The major dimensions of musical sound are time, pitch, and texture (Bruner, 1990). Examples of timerelated variables include tempo, meter, rhythm, and duration. Pitch-related variables include tonality, melody, and harmony. Textural variables include timbre and orchestration. (For definitions of these musical terms, see the Appendix to Bruner, 1990.) These variables constitute the raw materials composers use to organize sound, and may be thought of as the objective design features of musical products (Holbrook & Gardner, 1993). Tempo, the speed or pace of music, is perhaps the most basic component of music's temporal dimension (Duerr, 1981) and has long been held to be an important determinant of listener reactions (Hevner, 1937; Rigg, 1940). However, predictions and univariate effects vary across studies. Moreover, the two most widely cited studies examining tempo effects on consumer behavior (i.e., Milliman, 1982, 1986) used differentmusic under fast and slow conditions and thus potentially confounded tempo with pitch and textural variables. Holbrook and Anand (1990) noted the surprisinglack of attention devoted to this variable in recent decades. Tonality refers to the configuration of intervals between pitches in a scale (Apel, 1973);the most common examples are the conventional, diatonic major and minor keys (Piston, 1941). Like tempo, tonality is characteristicof virtually all music and is widely believed to contribute significantly to musical character(Heinlein, 1928; Hevner, 1935). Stout, Leckenby, and Hecker (1990) found mode (i.e., tonality) to have the greatest impact of any musical variable examined in a study of 40 television ads. In addition to major and minor keys, there are also many atonal or nondiatonic tonalities, each with its own aesthetic character.'A general distinction of atonal music is that it tends to sound out of tune, or more dissonant, to untrained, Western listeners than music pitched in conventional, diatonic tonalities. The third dimension of music, texture, is comprised of timbre and orchestration (Bruner, 1990). Timbre (also referred to as tone color) is defined by Dowling and Harwood (1986) as "the differences of sound quality among various musical instruments" (p. 5). Orchestrationrefers to the configuration of instrumentsused in a composition or performance.Texture has receivedless attention from researchers(for a notable exception, see Gundlach, 1935), but has been widely held to shape responses to music (Cooke, 1959). The overall gestalt of an aesthetic object is partially a function of its objective design features. Thus, the subjective character of a piece of music should 'Examplesincludepolytonal,serialistic,andminimalistmusic(e.g.,New Age, spacemusic), as well as manytypesof nonwesternmusicbasedon microtonalsystemsof pitchorganization (e.g., Indianraga;Arabicmaqam;Chinesegongdiaw;Indonesiangamelan,slendro,pelog).As international"worldbeat"musicincreasesin popularity,moreconsumersarebeingexposedto suchnondiatonictonalities.

384

KELLARIS AND KENT

stem in part from objective propertiesof sound. As Gundlach (1935) observed, "It appears quite certain that many pieces of music can elicit in many auditors a fairly uniform characterization solely through factors resident within the musical structure"(p. 642). For example, fast music tends to be perceived as more exciting and slow music as more relaxing; music pitched in major keys tends to be perceived as happier than music pitched in minor (or atonal) modalities. These basic characterizationsare fairly universal within Western culture and are, to some extent, shared across cultures (Gundlach, 1932;Dowling & Harwood, 1986).

RESPONSES TO MUSIC Music has been poetically touted as the universallanguage of mankind (Longfellow, 1835) and as the shorthand of emotion (widely attributed to Leo Tolstoy). It is often used in commercialapplications to evoke feeling responses in listeners (e.g., mood music). Indeed, although music may be capable of influencing the cognitive and behavioral domains of human response, responses to music are most often conceptualized in terms of emotion or affect (Budd, 1985; Meyer, 1956). As Wallace (1991) observed, most consumer research involving music has focused on affectiveinfluences. Music has been used to induce mood in experimental studies of advertising effects (Alpert & Alpert, 1989, 1990), postconsumption product evaluation (Miniard et al., 1992), and associative learning (Gorn, 1982). This is not surprisinggiven music's reputation as the shorthand of emotion. Several conceptual frameworks exist for characterizing feeling responses (e.g., Izard, 1977; Mehrabian & Russell, 1974;Osgood, Suci, & Tannenbaum, 1957; Plutchik, 1980). As recent consumer research has indicated (e.g., Havlena & Holbrook, 1986;Olney, Holbrook, & Batra, 1991;Westbrook & Oliver, 1991), a common thread underlying major typologies of emotion is a reduced space of three dimensions that can be interpretedas pleasure, arousal, and a third factor alternately labeled surprise, novelty, and dominance. In the present research,we adopt an approach analogous to that used by Holbrook and Batra (1987), that is, we selected items from multiple sources and applied a spatial reduction technique. Consistent with previous research on consumer emotions, we expected this approach to produce three dimensions of response. Theory and recent findings in musical aesthetics allow us to go beyond Bruner's(1990) general proposition that "the components of music are capable of having main as well as interact[ive]effects on [feelingstates].. ." (p. 99). For example, one can anticipate a positive contribution of tempo to arousal. Just as the human body adapts physiologically to variations in light and temperature, variations in the auditory environment may evoke analogous adaptive

LISTENERS'RESPONSESTO MUSIC

385

responses. For example, heart rate, blood pressure, breathing rate and amplitude, and other physiological manifestations of affectlessarousal involuntarily increase upon exposure to fast, stimulating music (Lundin, 1985). One might also anticipate a positive contribution of tempo to pleasure, at least across the restricted range of tempi represented in our study, because dissonances linger at slower speeds and are resolved to consonances more quickly at faster speeds (Dowling & Harwood, 1986). In addition, from an information-theoreticperspective,fastertempi producehigherratesof information and levels of informational density (Crozier, 1981), which should increase the "interestingness"of the music and thereby contribute to listeners'pleasure (Berlyne, 1974). Tempo may also contributeindirectlyto pleasurevia arousal in that fastermusic may seem more exciting. Of course, as tempo increasesto a very high speed, one can expect an eventual decrement in pleasure (Holbrook & Anand, 1990).The presentstudy, however, does not explore an extremerangeof tempi that would produce such nonmonotonic effects. Tonality, operating through familiarity and perceptions of consonance/ dissonance, should also influence pleasure (Dowling & Harwood, 1986). Specifically, the familiar, consonant, major and minor keys should sound more pleasing than unfamiliar, dissonant, atonal modalities due to cultural conditioning (Scott, 1990) and the smoothness/roughness of the physical sensations produced by the tones (Helmholtz, 1885/1954). Furthermore,earlierempiricalfindings (Gundlach, 1935) and long-standing musical intuition suggest that affect and arousal reactions to variations in speed and tone differacrossmusical textures,that is, textureshould act as a moderator. For example, the effect of tempo on pleasuremay depend upon instrumentation because listenershave differentexpectations for various types of musical ensembles. Likewise, the effect of tempo on arousal should be more pronounced for music orchestrated to emphasize the beat or pulse, such as pop music (which tends to feature percussion). This is because music's ability to arouse derives principallyfrom its temporaldimension(Holbrook & Anand, 1990). Theory and previous research provide less guidance about the effects of musical variables on the elusive third dimension of listener response. However, it seems likely that atonal modalities should evoke greater feelings of surprise among listeners.

METHOD Overview We conducted an exploratory experimentusing a 3 x 3 x 2 between-subjects design. Treatments included three levels of musical tempo (fast, moderate, slow) and tonality (major, minor, atonal), and two levels of texture (classical,

386

KELLARIS AND KENT

pop). Dependent variables included pleasure, arousal, and surprise,measured via multi-item semantic differentialscales. The procedure involved randomly assigning subjects to treatments, exposing them individually to a digitally synthesized musical treatment via headphones, and having them fill out a questionnaire. Subjects Subjects were 288 undergraduate students (164 men, 124 women) recruited from a subject pool at a large midwestern university;missing data on 5 cases left a usable sample size of 283. Ages ranged from 18 to 56 years (M = 20). There were no music majors or professional musicians in the sample. The mean self-reported extent of formal musical training was 3.41 on a 7-point scale ranging from very limited (1) to very extensive (7). The mean level of interest in music was 5.76 on a 7-point scale ranging from not at all interested(1) to extremely interested(7). Over 87% of the subjects reported listening to music very frequently. This indicates a reasonably informed, but musically amateur, group of subjects for whom the object of study is personally relevant. Course credit was offered as an incentive. Stimuli The stimulus materials were audio cassette recordingsof original classical and pop-style instrumentalmusic produced in a digital sound studio at a conservatory of music. We used classical and pop-style selections to provide contrasting textures and to enhance the generality of our findings. We composed the original classical piece and commissioned an original pop-style composition. Unfamiliar, original compositions were used to avoid extramusical associations due to prior exposure. The classical composition was written in 18th-centurycontrapuntal style and scored for flute, English horn, bassoon, and cello. The pop-style music can be described as a "Paula Abdul-type dance beat" scored for synthesizer, strings, brass chorus, chorus of synthesized human voices (vocalizing on the syllable "ah"), electric bass, and drums. Thus, the compositions representtwo contrasting textures differing in timbre, instrumentation, and style. Perhaps the most psychologically salient textural differencesbetween the compositions were the use of natural, acoustic sounds in the classical piece versus the use of synthesized, electronic sounds, and the inclusion of drums, in the pop-style piece. Nine versions of each composition were produced: one in each of three tonalities (major, minor, nondiatonic atonal version2),with each played at fast 2The term atonal is used here to mean unconventional, nondiatonic organization of pitch. Given this definition, there are multiple options for operationalizing atonality. The atonal

LISTENERS'RESPONSESTO MUSIC

387

(180 beats per minute [BPM]), moderate (120 BPM), and slow (60 BPM) speeds.3 To avoid confounding tempo with duration, two versions of each treatmentwere recorded. One version consisted of the entire musical selection; thus, playing time varied across tempo conditions (with slower speeds resulting in longer durations). The other version truncated (faded) the piece to the same duration across tempo conditions. Average playing time was about 3 min. Each version (variable or constant duration) was heard by half of the subjects in each treatment group. Recordings were made using digitally sampled sounds. Digital sampling technology involves analyzing the patterns of sounds produced by instruments (and other sound sources), then reproducingthem electronically.This technology, commonly used in industry, is capable of reproducing sounds that are virtually indistinguishable from the original, sampled sound sources. The musical compositions were entered into a Macintosh computer through a Baldwin MCX-1000 electronic keyboard and stored in digital form using ProPerformerMusical Instrument Digital Interface sequencer software. This allowed us to edit the musical scores to produce multiple versions containing orthogonal manipulations of tempo and tonality while holding all other dimensions of the music constant. Thus, melodic contour, rhythms, and other musical variables were invariant between versions within compositions. The edited digital information was sent through a Proteus sampler to produce the sound signal. The signal was routed to a Nakamichi MR-2 recorder through a Lexicon LXP-1 effects unit to produce high-quality sound recordings. Procedure At the time of recruitment, subjects were told that their participation was sought for a music study. The purpose of the study was not disclosed. They were told that the study would involve going to a listening lab, hearing a short tape, and filling out a questionnaire. As individuals arrivedat the lab, they were each given a set of headphones; versions of the compositions used in this research were created as follows. For the classical-style piece, each instrument's part was pitched in a wholetone scale using a different starting pitch. A wholetone scale divides the octave into seven (rather than eight) pitches, spaced at even intervals (rather than combinations of whole and half steps), and thus sounds more dissonant than conventional major and minor scales. For the pop-style piece, each instrument's part was pitched in a different key (pandiatonic), then random pitches in each part were altered to produce additional dissonance. Although classical and pop operations were produced by different techniques, the results were similar in terms of dissonance and novelty. 3Tempo levels were based on musical convention and precedent in music psychology (Lundin, 1985), and consumer aesthetics (Kellaris & Kent, 1991). Although other studies have examined broader ranges of speeds (e.g., Holbrook & Anand, 1990), we restricted our range to what was judged realistic or believeable for our stimulus compositions. In the judgment of a musical expert whom we consulted, versions slower than 60 or faster than 180 BPM would not likely be perceived as normal.

388

KELLARIS AND KENT

a randomly assigned, numbered audio cassette tape; and a (matching) numbered questionnaire. An attendant directed each subject to an audio carrel where simple printed instructions guided them through the procedure. Seals had been placed around the questionnaires; subjects were instructed not to break the seals until after they listened to the tape. Unobtrusive observation of a subset of subjects found no violations of this instruction. After listening to a tape and completing a questionnaire, subjects returnedthe materials to a lab attendant. The procedure took about 12 min. Measures Responses were measured via semantic differentialscales adapted from multiple sources, including previous researchin consumer/empirical esthetics (e.g., Berlyne, 1974; Crozier, 1981; Holbrook & Anand, 1990). Instructions read as follows: "Please rate the music you just heard on each of the following scales by placing an 'X' in the appropriatespace for each scale." The following items appeared after a prompt that stated, "The music I heard was: pleasant(7)/ unpleasant(l), interesting(7)/boring(l), *unappealing(7)/appealing(l), stimulating(7)/ soothing( ), complex(7)/ simple(l), *unenergetic(7)/energetic(l), familiar(7)/unfamiliar(l), arousing(7)/calming(l), *ugly(7)/beautiful(1), refined(7)/crude( 1), likeable(7)/unlikeable(1), *unexciting(7)/exciting, tasteful(7)/tasteless( ), unusual(7)/ordinary(1), *soft(7)/loud(1), surprising(7)/ predictable(l)." (Starred items were reverse coded.) We preferredto use a relativelyshort inventory of items to capture listeners' reactions before they dissipated (see Allen & Madden, 1989). Based on the findingsof Havlena and Holbrook (1986), we expected our reducedset of items to provide a parsimonious representationof the dimensions typically underlying more extensive inventories. Dimensionality of Responses An exploratory factor analysis (principal components, varimax rotation) extractedthree factors with eigenvalues of 5.30, 3.85, and 1.55 that accounted for 66.9%of the total variance. The first factor was composed of items relating to hedonic pleasure. (See Table 1 for factor loadings.) The second factor was composed of items relating to arousal. The third factor seems to echo Mehrabian and Russell's (1974) dominance dimension, Osgood et al.'s (1957) potency dimension, and Plutchik's (1980) surprise category of emotional response. This factor also recalls Berlyne's (1960) novelty construct, recently referredto as the uniqueness of stimulus information (Olney et al., 1991). We labeled this factor surprise. As in previous consumer research (cf. Russell, 1980; Russell, Weiss, & Mendelsohn, 1989), the third dimension of emotional response accounted for a relatively small proportion of the variance (9.7%). Two items (interesting, *unexciting) were dropped due to high cross-loadings across factors.

LISTENERS' RESPONSES TO MUSIC

389

TABLE1 FactorLoadingsFromExploratoryFactorAnalysis Scale Item Pleasant/unpleasant Interesting/boring Unappealing/appealinga Stimulating/soothing Complex/simple Unenergetic/ energetica Familiar/unfamiliar Arousing/calming Ugly/beautifula Refined/crude Likeable/unlikeable Unexciting/excitinga Tasteful/tasteless Unusual/ordinary Soft/louda Surprising/predictable

Factor 1: Pleasure

Factor 2: Arousal

Factor 3: Surprise

.87 .60 .84 - .06 .06 .34 -.12 -.17 .86 .73 .92 .58 .80 - .09 - .35 .14

-.10 .41 .20 .77 .38 .74 -.18 .84 - .08 -.36 .09 .64 -.12 .12 .72 .21

- .03 .30 .03 .04 .63 .12 .52 .08 -.14 -.06 -.02 .16 .00 .77 .12 .76

aReverse coded.

Validity Assessment Following a procedure recommended by Gerbing and Anderson (1988), a confirmatory factor analysis using LISREL VI (Version 6.6) was conducted to assess the factor structure of the retained items. Several more items were dropped due to high (> 2) normalized residuals. Final statistics for the measurement model based on the retained items indicate an acceptable level of fit, 2 = 79.99, df = 24,p < .01;GFI = .942;AGFI = .891; RMS = .058. Three composite scales were formed by summing and averaging related items. Alpha reliabilities were .88 for the pleasure scale (*ugly/beautiful, tasteful/tasteless, refined/crude, pleasant/unpleasant), .81 for the arousal scale (stimulating/ soothing, arousing/calming, *soft/loud), and .70 for the surprise scale (complex/simple, unusual/ordinary, surprising/predictable). RESULTS Multivariate Analysis of Variance (MANOVA) Because our dependent variables-pleasure (P), arousal (A), and surprise but distinct (S)-are conceptually nomologically interrelated, we assessed their = intercorrelations, rp,A -.23, p < .001; rps = -.07, ns; rAs = .31, < .001. Given intercorrelated p dependent measures, we assessed treatment effects via MANOVA to avoid Type I error inflation. Results are summarized in Table 2.

390

KELLARIS AND KENT TABLE2 Overview of Variance Analyses MANOVA Results

Independent Variable

Dependent Variable

Tempo

Pleasure Arousal Surprise Pleasure Arousal Surprise Pleasure Arousal Surprise Pleasure

Tonality

Texture

Tempo by tonality

Tempo by texture

Tonality by texture

Tempo by tonality by texture

Arousal Surprise Pleasure Arousal Surprise Pleasure Arousal Surprise Pleasure

Arousal Surprise

ANO VA Results p <

F

df

p <

ns

4.45 6.47 0.88 39.04 0.63 9.20 6.76 74.55 1.80 2.29

2, 265 2, 265 2, 265 2, 265 2, 265 2, 265 1, 265 1, 265 1, 265 4, 265

.01 .002 ns .001 ns .001 .01 .001 ns ns

.01

0.47 0.40 3.20

4, 265 4, 265 2, 265

ns ns .04

.02

5.62 1.93 5.47

2, 265 2, 265 2, 265

.01 ns .01

2.32 1.06 1.36

2, 265 2, 265 4, 265

ns ns ns

0.77 0.95

4, 265 4, 265

ns ns

Wilks' A

F

df

.9034

4.57

6, 526

.001

.7064

16.64

6, 526

.001

.7754

25.39

3, 263

.001

.9473

1.20

12, 696

.9370

.9463

.9398

2.89

2.45

1.37

6, 526

6, 526

12, 696

ns

The MANOVA reveals a pattern of significantmain and interactiveeffects on the combined dependent measures representingdimensions of response to music. Given the MANOVA findings, we performed a series of univariate analyses of variance (ANOVAs) on each dependent variable. Results of these analyses are also reported in Table 2. Descriptive statistics appear in Table 3. Effects on Pleasure Tempo produced a significant main effect on pleasure, F(2, 265) = 4.45, p < .01, W2 = .02. Texture moderated this effect, F(2, 265) = 3.20, p < .04, 02 = .01, such that the increase in pleasure between slow and fast speeds was significant for the classical music, t(94) = 3.62, p < .001 (one-tailed), but not for the pop-style music, t(90) = .45, ns. This interaction is illustrated in Figure 1.

LISTENERS' RESPONSES TO MUSIC

391

TABLE3 Means and StandardDeviationsfor ExperimentalMainEffects Arousal

Pleasure

Totala Tempo Slow Moderate Fast Tonality Major Minor Atonal Texture Classical Pop style

Surprise

M

SD

M

SD

M

SD

4.71

1.25

3.17

1.49

4.33

1.26

4.48 4.69 4.95

1.26 1.42 1.00

3.42 3.60 4.10

1.39 1.35 1.64

4.30 4.23 4.46

1.31 1.25 1.21

5.19 5.02 3.92

1.10 1.02 1.24

3.65 3.64 3.84

1.60 1.43 1.45

3.93 4.37 4.69

1.29 1.13 1.23

4.87 4.54

1.34 1.14

3.06 4.39

1.24 1.42

4.23 4.43

1.16 1.35

aN = 283.

Tonality also produced a significant main effect on pleasure, F(2, 265) = 39.04, p < .001, 02 = .19, with the consonant (major and minor) tonalities producing more pleasurethan the dissonant, atonal music. Texture moderated this effect, F(2, 265) = 5.47, p < .01, co2 = .02, such that a significant decrement in pleasure is observed between major and minor tonalities for the classical music, t(94) = 2.07, p < .02 (one-tailed), but not for the pop-style music, t(90) = -.51, ns. The decrementin pleasure between minor and atonal music was significantfor both classical, t(94) = 5.78, p < .001, and pop music, t(91) = 3.61, p < .001. This interaction is illustrated in Figure 2. Effects on Arousal Tempo produced a main effect on arousal, F(2, 265) = 6.47, p < .002, c2 = .03. Texture moderated this effect, F(2, 265) = 5.62, p < .01, co2= .025, such that the increase in arousal from slow to fast speeds was significantfor the pop music, t(90) = 4.47, p < .001, but not for the classical music, t(94) = .39, ns. This interaction is illustrated in Figure 3. Effects on Feelings of Surprise Tonality produced a significant main effect on the surprise dimension of listener response, F(2, 265) = 9.20, p < .001, c2 = .055. Less consonant versions

of the stimulus music were rated as more complex, unusual, and surprising, irrespectiveof speed or texture. Statistical differenceswere observed between major and minor modes, t(186) = - 2.46, p < .015; major and atonal modes,

392

KELLARISAND KENT 5.5

5-

ci)~~~~~~~~~~~~~~~~~,

4.5 -I:

I 44 SLOW (60)

1

1

MODERATE(120) TEMPO OF MUSIC (BPM)

FAST (180)

Pop Music Classical Music o

----

FIGURE1 Interactive effect of musical tempo and texture on the pleasure dimension

of listenerresponse. t(187) = -4.16, p < .001; and the minor and atonal modes, t(187) = - 1.89,

p < .06. There were no other main or interactive effects on this dimension of response. Summary Pleasure was influenced by the interactions of both tempo and tonality with texture. Faster speeds and more consonant keys increased the pleasantness of classical (but not pop) music. Arousal was influenced by the interaction of tempo with texture. Faster speeds produced greater arousal among subjects exposed to pop (but not classical) music. Feelings of surprisewere influenced by tonality such that less consonant keys (i.e., atonal relative to minor, minor relative to major) were rated as more surprising. Effect sizes ranged from .01 to .19. Although an effect magnitude of 1%may

LISTENERS'RESPONSESTO MUSIC

393

6 -

5.5 -

5

\

< a 4.5 -

4

3.5

MAJOR

MINOR

ATONAL

TONALITY OF MUSIC

Pop Music ClassicalMusic

2 Interactiveeffectof musicaltonalityand textureon the pleasuredimension FIGURE of listenerresponse.

seem diminutive, this is typical of behavioral research and within the conventional range, especially for interactive effects, which tend to be smaller (see Peterson, Albaum, & Beltramini, 1985).

DISCUSSION This study examined the effects of three important objective stimulus properties of music (tempo, tonality, texture) on consumers' responses to music. Three dimensions of response (pleasure, arousal, surprise) emerged from our inventory of 16 (reduced to 10) items. Consistent with Bruner's(1990) general proposition, a significant portion of music's influence on listeners' responses appears to stem from the objective properties of sound and their interactions. Faster tempi led to more pleasure for subjects exposed to classical music. Previous work in empirical/consumer aesthetics has observed a positive contribution of arousal to pleasure up to a point of inflection (e.g., Holbrook &

394

KELLARISAND KENT 5.5 -

S

4.5 -

i

3.5 -

3 -

2.5

2 SLOW (60)

MODERATE (120)

FAST (180)

TEMPO OF MUSIC (BPM) Pop Music Classical Music

--FIGURE3 Interactive effect of musical tempo and texture on the arousal dimension of listener response.

Anand, 1990). Therefore, it may seem reasonable to anticipate that the influence of tempo on pleasure operates, at least partly, through arousal. However, such a mediation does not appear to exist in our data, because tempo did not significantly increase arousal among subjects exposed to classical music (see Baron & Kenny, 1986). Apparently tempo contributed to pleasure independently of an indirect effect through arousal. Faster tempi produced greater arousal among subjects exposed to our pop-style music. This effect may stem from the orchestrationof our pop music, which emphasized percussion. The drum sounds may have drawn greater attention to temporal aspects of the music, therebyleading to increasedarousal at higher speeds. Tonality affected pleasure for subjects exposed to classical music such that the major key condition was more pleasurable than the minor key condition, and the minor key condition was more pleasurable than the atonal music. These effects may be due to the relative salience of music's pitch dimension (as

LISTENERS'RESPONSESTO MUSIC

395

opposed to temporal dimension) in classical music. Indeed, the orchestration of the classical music may have drawn more attention to pitch-related elements, such as melody and harmony. Not surprisingly,exposure to dissonant, atonal music produced the least pleasure across the textures. Texture moderated the effects of tempo and tonality on pleasure, and the effect of tempo on arousal. However, whereas the manipulations of tempo and tonality were perfectly pure (thanks to digital technology), our manipulation of texture subsumed several variables. Many of these were texture related, but several were not (e.g., melody, style). Thus, we cannot isolate any single structural variable responsible for the observed texture effects. Importantly, listeners' expectations may differ notably for pop versus classical music. The confoundment of such expectations with musical textures in our study represents a limitation that should be addressedin future research.For now, we can only speculate that it may have been the relative saliency of pitch (vs. time) in the classical music that augmented the tonality effect on pleasure (see Figure 2), and the relative saliency of time (vs. pitch) in the pop music that augmented the tempo effect on arousal (see Figure 3). However, it is also possible that normative expectations of speed contributed to the results. Increases in tempo may have had relatively less impact on reactions to pop music because the faster speeds departed less from expectations than they did for classical music. Implications for the Use of Music in Consumer Research Although we consider empiricalresearchin musical aesthetics to be in an early stage of development, our findings would seem to suggest a number of preliminary implications for the use of music in consumer research. First, it is clear that responses to music are at least partly a function of objective musical properties.Thus, studies that examine effects of the mere presence(vs. absence) of music present an interpretationalchallenge, because effects could be quite differentdepending on the traits of the music used. Indeed, inconsistent findings are to be expected when properties of music vary across studies. For this reason, consumer researchers may wish to avoid music versus no music manipulations in experiments and presence versus absence of music comparisons in correlational studies. Studies examining the influence of individual musical properties should use the same (or highly similar)music across conditions, varying only the attribute or attributes of interest. Although this was not always practical in the past, recent advances in digital recording technology facilitate such control. When different pieces of music are used to manipulate a musical variable, musical properties will be confounded, making it difficult to isolate specific causal antecedents. Our findings suggest that researcherswishing to induce pleasant feelings without altering arousal may do so by manipulating musical tonality (rather

396

KELLARIS AND KENT

than speed). This is important because arousal may influence attention and other processes that researchersmay wish to avoid confounding with pleasant affect. In situations calling for the induction of positive feelings, the use of music pitched in a major key is indicated. In situations calling for the induction of negative feelings, the use of atonal music may be indicated. If atonal music is unacceptablefor a given purpose (e.g., because it is unrealistic or incongruent), negative affect can be induced with music pitched in minor keys. However, according to our findings, differencesin affect between minor and major keys may emerge only if the music is orchestratedto emphasize the pitch (vs. time) dimension. Implications for the Interpretation of Past Findings Our findings may help to explain apparent inconsistencies in past musicrelated findings. For example, in associative learning studies, hedonically pleasing (vs. displeasing) music has been paired with conditioned stimuli (e.g., Gorn, 1982; Kellaris & A. D. Cox, 1989). When upbeat pop music is used to operationalize the positive unconditioned stimulus, hedonic valence may be confounded with arousal, which may activate higher levels of attention. Differences in attention to the associative learning task may help to explain why Gorn's (1982) classic finding was not replicated when Kellaris and A. D. Cox (1989) controlled for structural properties of music. A similar implication holds for the interpretation of findings from mood research.For example, Alpert and Alpert (1990) used fast, major key and slow, minor key music to induce happy and sad mood states. Contrary to the prediction that positive moods should generally encourage positive outcomes, the Alperts found sad music to produce the most positive purchase intent. An explanation for this finding, offered in their article, is that the sad music was congruent with the advertised product (a "missing you" greeting card) and thus had a reinforcingeffect. Maclnnis and Park's (1991) study of musical fit (congruity) and ad processing supports this explanation. Our findings may provide additional insight. Independent of tonality, the faster tempo (and louder dynamics) of the happy music used in the Alpert and Alpert study may have induced greater arousal, leading to higher levels of attention. Congruity has been found to moderate the impact of music-inducedattention, such that attention-gettingmusic that is incongruentinhibits message processing (Kellaris et al., 1993). Thus, Alpert and Alpert's happy music may simply have been more distracting than the sad music due to arousal- and attention-inducing features, thereby reducing the persuasive impact of the ad. In sum, our study suggests that, when music is used to induce differentialmood states, researchers should manipulate tonality and hold speed constant to avoid confounding pleasant feelings with arousal and its unintended consequences.

LISTENERS'RESPONSESTO MUSIC

397

Unfortunately, many previous studies in our field do not describe the stimulus music used as fully and conscientiously as Alpert and Alpert. Our findings suggest the importance of characterizing music in terms of its objective stimulus properties, as well as how these properties fit or depart from listeners' expectations. Without such information, it is difficult to identify specific causal influences or assess the boundary conditions within which effects operate. Limitations and Issues for Future Research The present study has several limitations that should be acknowledged. First, the duration of exposure to the musical stimuli was relatively brief-about 3 min. It is possible that different responses to music could develop after more prolonged exposures, such as one would experience in stores, restaurants, or during concerts. In addition, our subjects received only one exposure to a musical stimulus. In certain contexts (e.g., advertising) listeners receive multiple exposures over time. Responses may change as listeners become more familiar with a piece of music. Also, our stimulus music was featured as the focal object in a directed-attention task. The influences of music may differ when listeners receive passive exposures to background music, such as one might experience in retail environments. Indeed, Yalch and Spangenberg (1990) found different effects of background versus foreground music, although different music was used under the two exposure conditions. Future research should examine reactions to the same piece or program of music under different listening conditions. This study examined the influence of three representative properties of music. Given the present findings on tempo, tonality, and texture, future studies should examine the influence of other variables (e.g., rhythm, melodic contour). A further limitation is that our design used only one exemplar per cell, which clearly imposes limits on the generality of our findings. In future research, multiple examples of each treatment combination could be created. Furthermore, the musical examples in our study were purely instrumental. Stimulus propertiesmay be less predictive of responses to vocal music because vocal music may bear richer associational meanings to listeners (Scott, 1990). Responses to music do not stem solely from physical sound, but ratherfrom the interplay of music and characteristics of individual listeners. Therefore, future research should examine the role of listener characteristicsin shaping responses to music. This article examined in some detail the influence of musical properties on responses to music, but did not consider the impact of those responses on other consumptive outcomes such as one might expect when music is a feature of an ad or of a retail environment.We must commend this important next step to future research.

398

KELLARIS AND KENT

ACKNOWLEDGEMENTS This research was funded in part by a College of Business Administration Summer Research Fellowship at the University of Cincinnati. We thank Dr. Frederick Bianchi of the University of Cincinnati College Conservatory of Music and engineering student Jim Grote for their help with stimulus production. Thanks are also due to Chris Allen, Murali Chandrashekaran, Tony Cox, Frank Kardes, Robert Schindler, associate editor Gerry Gorn, and the Journal of ConsumerPsychology reviewers for their helpful comments on earlier drafts of this article.

REFERENCES Allen, C. T., & Madden, T. J. (1989). Gauging and explaining advertising effects: Emergent concerns regardingconstruct and ecological validity. In P. Cafferata & A. M. Tybout (Eds.), Cognitiveand affective responsesto advertising(pp. 327-351). Lexington, MA: Lexington. Alpert, J. I., & Alpert, M. I. (1989). Background music as an influence in consumer mood and advertisingresponses. In T. K. Srull (Ed.), Advancesin consumerresearch(Vol. 16, pp. 485-491). Provo, UT: Association for Consumer Research. Alpert, J. I., & Alpert, M. I. (1990). Music influenceson mood and purchaseintentions. Psychology & Marketing, 7, 109-133. Anand, P., & Sternthal,B. (1990). Ease of message processing as a moderator of repetition effects in advertising. Journal of Marketing Research, 27, 345-353. Apel, W. (1973). Harvarddictionaryof music (2nd ed.). Cambridge, MA: Belknap Press. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research:Conceptual, strategic,and statisticalconsiderations.Journalof Personality and Social Psychology, 51, 1173-1182. Berlyne, D. A. (1960). Conflict,arousal, and curiosity. New York: McGraw-Hill. Berlyne, D. E. (1974). Studies in the new experimentalaesthetics:Steps towardan objectivepsychology of aesthetic appreciation.Washington, DC: Hemisphere. Bierley, C., McSweeney, F. K., & Vanieuwkerk, R. (1985). Classical conditioning of preferences for stimuli. Journal of ConsumerResearch, 12, 316-323. Bruner, G. C., II. (1990). Music, mood, and marketing. Journal of Marketing, 54, 94-104. Budd, M. (1985). Music and the emotions. London: Routledge & Kegan Paul. Cooke, D. (1959). The language of music. London: Oxford University Press. Crozier, J. B. (1981). Information theory and melodic perception: In search of the aesthetic engram. In H. Day (Ed.), Advances in intrinsicmotivationand aesthetics (pp. 433-461). New York: Plenum. Dowling, W. J., & Harwood, D. L. (1986). Music cognition. San Diego, CA: Academic. Duerr, W. (1981). Rhythm in music: A formal scaffoldingof time. In J. T. Fraser (Ed.), The voices of time (pp. 180-201). Amherst: University of Massachusetts Press. Galizio, M., & Hendrick, C. (1972). Effect of musical accompaniment on attitude: The guitar as a prop for persuasion. Journal of AppliedPsychology, 2, 350-359. Gardner,M. P. (1985). Mood states and consumerbehavior:A criticalreview.Journalof Consumer Research, 12, 281-300. Gerbing, D. W., & Anderson, J. C. (1988). An updated paradigm for scale development incorporating unidimensionalityand its assessment. Journal of Marketing Research, 25, 186-192.

LISTENERS' RESPONSES TO MUSIC

399

Gorn, G. J. (1982). The effects of music in advertisingon choice behavior:A classical conditioning approach. Journal of Marketing, 46, 94-101. Gottlieb, A. (1991). The music business: Just like a ringing bell. The Economist, 321, S3-S5. Gundlach, R. A. (1932). A quantitative analysis of Indian music. AmericanJournalof Psychology, 44, 133-145. Gundlach, R. A. (1935). Factors determining the characterizationof musical phrases. American Journal of Psychology, 47, 624-643. Haley, R. I., Richardson, J., & Baldwin, B. M. (1984). The effects of nonverbal communications in television advertising. Journal of AdvertisingResearch, 24(4), 11-18. Havlena, W. J., & Holbrook, M. B. (1986). The varieties of consumption experience:Comparing two typologies of emotion in consumer behavior. Journal of ConsumerResearch, 13, 394-404. Heinlein, C. P. (1928). The affective characterof the major and minor modes in music. Journalof ComparativePsychology, 8, 101-141. Helmholtz, H. (1954). On the sensationsof tone. New York: Dover. (Originalwork published 1885) Hevner, K. (1935). The affective character of the major and minor modes in music. American Journal of Psychology, 47, 103-118. Hevner, K. (1937). The affectivevalue of pitch and tempo in music. AmericanJournalof Psychology, 49, 621-630. Holbrook, M. B., & Anand, P. (1990). Effects of tempo and situational arousal on the listener's perceptual and affective responses to music. Psychology of Music, 18, 150-162. Holbrook, M. B., & Anand, P. (1992). The effects of situation, sequence, and features on perceptual and affective responses to product designs: The case of aesthetic consumption. Empirical Studies of the Arts, 10, 19-31. Holbrook, M. B., & Batra, R. (1987). Assessing the role of emotions as mediators of consumer responses to advertising. Journal of ConsumerResearch, 14, 404-420. Holbrook, M. B., & Gardner, M. P. (1993). An approach to investigating the emotional determinants of consumption durations:Why do people consume what they consume for as long as they consume it? Journal of ConsumerPsychology, 2, 123-142. Holbrook, M. B., & Schindler, R. M. (1989). Some exploratory findings on the development of musical tastes. Journal of ConsumerResearch, 16, 119-124. Holbrook, M. B., & Schindler, R. M. (1991). Echoes of the dear departed past: Some work in progress on nostalgia. In R. H. Holman & M. R. Solomon (Eds.), Advancesin consumerresearch (Vol. 18, pp. 330-333). Provo, UT: Association for Consumer Research. Hoyer, W. D., Srivastava, R. K., & Jacoby, J. (1984). Sources of miscomprehensionin television advertising. Journal of Advertising,13(2), 17-26. Izard, C. E. (1977). Human emotions. New York: Plenum. Kellaris, J. J., & Altsech, M. B. (1992). The experience of time as a function of musical loudness and gender of listener. In J. Sherry & B. Sterthal (Eds.), Advancesin consumerresearch(Vol. 19, pp. 725-729). Provo, UT: Association for Consumer Research. Kellaris,J. J., & Cox, A. D. (1989). The effects of backgroundmusic in advertising:A reassessment. Journal of ConsumerResearch, 16, 113-118. Kellaris, J. J., Cox, A. D., & Cox, D. (1993). The effect of background music on ad processing: A contingency explanation. Journal of Marketing, 57(4), 114-125. Kellaris, J. J., & Kent, R. J. (1991). Exploring tempo and modality effects on consumer responses to music. In R. H. Holman & M. R. Solomon (Eds.), Advancesin consumerresearch(Vol. 18, pp. 243-248). Provo, UT: Association for Consumer Research. Kellaris, J. J., & Kent, R. J. (1992). The influence of music on consumers' temporal perceptions: Does time fly when you're having fun? Journal of ConsumerPsychology, 1, 365-376. Kotler, P. (1973-1974). Atmospherics as a marketing tool. Journal of Retailing, 49, 40-64. Lacher, K. T. (1989). Hedonic consumption: Music as a product. In T. K. Srull (Ed.), Advances in consumerresearch (Vol. 16, pp. 367-373). Provo, UT: Association for Consumer Research.

400

KELLARIS AND KENT

Longfellow, H. W. (1835). Outre-Mer.New York: Harper & Brothers. Lundin, R. W. (1985). An objectivepsychology of music (3rd ed.). Malabar, FL: Krieger Publishing. Maclnnis, D. J., & Park, C. W. (1991). The differentialrole of characteristicsof music on high and low involvement consumers' processing of ads. Journal of ConsumerResearch, 18, 161173. Mehrabian,A., & Russell, J. A. (1974). An approachto environmentalpsychology.Cambridge,MA: MIT Press. Meyer, L. B. (1956). Emotion and meaning in music. Chicago: University of Chicago Press. Milliman, R. E. (1982). Using background music to affect the behavior of supermarketshoppers. Journal of Marketing, 46, 86-91. Milliman, R. E. (1986). The influenceof backgroundmusic on the behavior of restaurantpatrons. Journal of ConsumerResearch, 13, 286-289. Miniard, P. W., Bhatla, S., & Sirdeshmukh,D. (1992). Mood as a determinantof postconsumption product evaluations: Mood effects and their dependency on the affective intensity of the consumption experience. Journal of ConsumerPsychology, 1, 173-195. Morrow, L. (1991, January 7). In one ear, in the other. Time, p. 84. Olney, T. J., Holbrook, M. B., & Batra, R. (1991). Consumerresponses to advertising:The effects of ad content, emotions, and attitude toward the ad on viewing time. Journal of Consumer Research, 17, 440-453. Osgood, C. E., Suci, G. J., & Tannenbaum, P. (1957). The measurementof meaning. Urbana: University of Illinois Press. Park, C. W., & Young, S. M. (1986). Consumer response to television commercials:The impact of involvement and background music on brand attitude formation. Journal of Marketing Research, 23, 11-24. Peterson, R. A., Albaum, G., & Beltramini, R. F. (1985). A meta-analysis of effect sizes in consumer behavior research. Journal of ConsumerResearch, 12, 97-103. Piston, W. (1941). Harmony. New York: Norton. Plutchik, R. (1980). Emotion. A psychoevolutionarysynthesis. New York: Harper & Row. Rigg, M. G. (1940). Speed as a determinerof musical mood. Journalof ExperimentalPsychology, 27, 566-571. Russell, J. A. (1980). A circumplexmodel of affect. Journalof Personalityand Social Psychology, 39, 1161-1178. Russell, J. A., Weiss, A., & Mendelsohn, G. A. (1989). Affect grid: A single-itemscale of pleasure and arousal. Journal of Personality and Social Psychology, 57, 493-502. Scott, L. M. (1990). Understanding jingles and needledrop: A rhetorical approach to music in advertising. Journal of ConsumerResearch, 17, 223-236. Seidman, S. A. (1981). On the contributions of music to media productions. EducationalCommunication and TechnologyJournal, 29, 49-61. Sewall, M. A., & Sarel, D. (1986). Characteristicsof radio commercials and their recall effectiveness. Journal of Marketing, 50, 52-60. Smith, P., & Curnow, R. (1966). "Arousal hypothesis" and the effects of music on purchasing behavior. Journal of Applied Psychology, 50, 255-256. Stewart, D. W., & Furse, D. H. (1986). Effectivetelevisionadvertising-A study of 1000 commercials. Lexington, MA: Heath. Stout, P. A., Leckenby, J. D., & Hecker, S. (1990). Viewer reactions to music in television commercials. JournalismQuarterly,67, 887-898. Wallace, W. T. (1991). Jingles in advertisements:Can they improve recall? In R. H. Holman & M. R. Solomon (Eds.), Advances in consumer research (Vol. 18, pp. 239-242). Provo, UT: Association for Consumer Research.

LISTENERS' RESPONSES TO MUSIC

401

Westbrook, R. A., & Oliver, R. L. (1991). The dimensionality of consumption emotion patterns and consumer satisfaction. Journal of ConsumerResearch, 18, 84-91. Yalch, R. F., & Spangenberg, E. (1990). Effects of store music on shopping behavior. Journal of ConsumerMarketing, 7, 55-63.

Acceptedby GeraldJ. Gorn.