Emotions and associations evoked by unfamiliar music Richard Parncutt and Manuela M. Marin University of Graz, Austria XIX Congress of the International Association of Empirical Aesthetics University of Avignon, August 31, 2006 Current research on music & emotion • revival of topic in past 10-20 years – important role of music in everyday life – good medium for studying emotion in general – ease of making judgements and affective responses (Gaver & Mandler 1987) • …but few clear findings: – difficult to observe (e.g., measurement, laboratory vs. “real world” situations) – unclear relation to non-musical theories of emotion – inter- und intra-individual variability (Juslin & Sloboda 2001) • research focuses on familiar music (e.g., 9th ICMPC, Bologna, August 22-26, 2006) Familiar versus foreign music Questions • familiarity influences music appreciation (Iwanga et al. 1996, Newman et al. 1995) • listeners tend to like familiar music (North & Hargreaves 1995, Peretz et al. 1998, Ali 2005) Why? – difficult question since sometimes negative emotions are preferred (e.g., Schubert 1996, Levinson 1982) • What specific emotions are associated with familiar versus foreign music? – How are these related to liking? – How do these depend on gender and style? implications for research on Otherness in (ethno-) musicology and cultural studies Emotion, familiarity & liking (1) • Preference feedback hypothesis and optimal complexity - positive relationship between liking and voluntary exposure (e.g., Zajonc 1969, North & Hargreaves 1995, Ali 2005) - inverted-U relationship between liking and arousal potential (subjective complexity, familiarity) (Berlyne 1971, 1974) Emotion, familiarity & liking (2) • Music & Russell’s circumplex model of emotion (1980): 2 dimensions of arousal & pleasantness - non-vocal music: liking and arousal level predict emotion (North & Hargreaves 1997) - songs: arousal, pleasantness & familiarity may predict emotion best, not liking (Ritossa & Rickard 2004) - familiarity may bias ratings of pleasantness, arousal, and emotion in music, in addition to liking (Ritossa & Rickard 2004) Emotion, familiarity & liking (3) • cue redundancy model (Balkwill & Thompson 1999) – judgements of emotions in unfamiliar music: acoustic cues more important than culture-specific (Balkwill et al. 2004) • diversity of emotions – more diverse emotions and stronger experience in familiar music (Iwanaga et al. 1996, Gabrielsson & Lindström Wik 2003) • role of cognition – liking and intensity of emotion may depend more on cognitive factors than familiarity (Ali 2005, Fung 2004) • role of style – liking for a piece may depend more on style than piece (North & Hargreaves 1997) Emotion, gender & expertise • Task: link specific emotion words to music – no significant gender differences in children and adults (Kratus 1993, Robazza et al. 1994, Kallinen 2005) • Task: rate emotional qualities of music – no differences between musicians and non-musicians (Kreutz et al. 2002) • Task: rating complexity and liking – musical expertise may dissolve the relationship between liking and complexity (Orr & Ohlsson 2005) Emotion & language • difficulty of defining – concept of emotion – specific emotions • words used to describe emotions can also be – value judgements – metaphors • …and refer to personal characteristics: – physical appearance – arousal and bodily states – personality • little agreement on which words denote „emotions“ and which do not (Plutchik 2002) • clustering of emotion words into broad categories – little agreement on category names and correspondences Method of self-report (free description) Advantages • most direct form of evidence Disadvantages • limited to participant’s language • sensitive to subtleties and • vague correspondences peak experiences • multiple meanings • large vocabulary • not restricted to basic emotions or valence/arousal • individual differences gradations (Juslin & Sloboda 2001, Plutchik 2002) Shaver et al. (1987): emotion prototypes • hierarchical structure of English emotion lexicon: - lists of emotion names (Averill 1975, Davitz 1969, …) prototypicality ratings of emotionness similarity sorting and hierarchical cluster analysis: basic categories, subcategories 135 emotion words sorted by “basic level concepts”: love, joy, surprise, anger, sadness, fear Our definitions • Emotion word – a word that refers to any aspect of emotional experience • Emotion noun – noun derived from emotion word(s) • Emotion category – cluster of emotion words, e.g. • basic emotions (e.g. fear) • arousal levels (e.g. movement) Context & design • course “Listening to broaden musical horizons” – first year of musicology program, Uni Graz – compulsory unit in program • 43 students (23 males, 20 females), compulsory attendance • 13 weekly sessions @ 90 minutes • student pairs selected/presented styles (vocal & non-vocal) • 17 styles x 3 examples x few minutes Criteria for selection of examples Instructions to students Style should: • • • • • • be homogeneous be acoustically recognizable be accepted as “music” by some group be unfamiliar, interesting, challenging for other students contain no major/minor triads or scales be unsuitable as background music for parties Examples should: • be performed/composed by >2 diff. performers or groups Labels of selected styles 11 modern western styles early computer music Peter Lackner’s canonical principle instrumental music of Olga Neuwirth jew’s harp Japanoise Clicks’n Cuts Beat Furrer György Ligeti – Steve Reich Ambient Sound Free Jazz Stomp 6 non-western styles Tibetan Buddhism Huli Aka pygmies Sioux Indians Capoeira Shakti Procedure Listeners were told nothing about the style. During/after each example (~ 3 min), they provided data on paper: • Qualitative data – emotions experienced – associations evoked – sound sources – pitch-time structures • Quantitative data – (un-) familiarity as music 1 = sounds familiar, 2 = neutral, 3 = sounds foreign – (dis-) liking as music 1 = I like it, 2 = neutral, 3 = I do not like it Analysis of qualitative emotion data (1) 1. Selection of emotion words • list first word in emotion field of each response sheet • 3 examples x 17 styles x ca. 35 students = ca. 1600 words • eliminate repetitions 760 words • convert all words to nouns (in German): emotion nouns • group nouns with same stem and similar meaning • eliminate noun groups that occur less than 3 times • Result: 210 emotion nouns Analysis of qualitative emotion data (2) 2. Assignment of emotion nouns to categories 10 native German speakers (or groups): – group 210 nouns into any number of groups, – find one noun as a group labels – comparison and negotiation 21 emotion categories 10 different native German speakers: – assign 210 emotion nouns to the 21 emotion categories – assignment is valid only if most participants agree – 123 of the 210 were assigned, forget the rest Return to original 1600 emotion words: assign 890 words to 21 emotion categories and forget the rest! Emotion category Original German English translation Number of assigned nouns Total no. of occurrences Equivalent entry in list of Shaver (2001) Abscheu antipathy 3 3 disgust Aggression aggression 8 41 irritation Angst fear 15 152 fear Bewegung movement 15 50 - Chaos chaos 4 61 - Feierlichkeit festivity 3 12 - Fremdheit foreignness 1 3 - Freude joy 11 118 joy Hast haste 5 43 - Intensität intensity 7 72 - Langeweile boredom 10 88 - Emotion category Number of assigned nouns Total no. of occurrences Equivalent entry in list of Shaver (2001) Original German English translation Leid suffering 3 26 suffering Liebe love 4 8 love Neugierde curiosity 3 25 - Ruhe calm 7 88 contentment? Spaß fun 3 18 zest? Spiritualität spirituality 5 25 enthralment? Traurigkeit sadness 8 32 sadness Überraschung surprise 1 2 surprise Wahnsinn madness 4 9 - Wut anger 3 16 anger Difference between (unfamiliar) musical and everyday (Shaver) emotions • good correspondence with Shaver’s basic categories love, joy, surprise, anger, sadness, fear (“basic emotions”) • subcategories: - correspondence: suffering, irritation, disgust - no direct correspondence: lust, longing, zest, contentment, optimism, disappointment, pride, horror, nervousness, torment, neglect BUT…most of these concepts occurred in our list of 210 emotion nouns! • in Shaver’s list but not ours: relief, envy, shame, sympathy • in our list but not Shaver’s: chaos, madness, curiosity, festivity, calm, movement, haste, intensity, boredom, foreignness - some of these tend to describe arousal and potency/dominance (cf .Osgood, Suci & Tannenbaum 1957, Russell 1978, 1980, Schlosberg 1954) Frequency of the 21 emotion categories (all data) • specifically musical emotions? (e.g., Scherer et al. 2001-02, Kreutz 2002) - biased toward foreign music? festivity antipathy surprise fear love foreignness madness fun anger curiosity spirituality suffering joy sadness aggression haste movement calm chaos boredom intensity fe st iv ity s l fo urp ove re ri i s an gne e sp tip ss iri ath m tua y ad lit ne y ss fu a cu ng n su rios er ffe ity s ri m adn ng ov e em ss ag h en gr as t es te si on c in a te lm ns ity ch joy bo a re o s do m fe ar frequency Frequency of 21 emotion categories for 11 modern western styles 140 120 100 80 Major emotion categories: fear, boredom, chaos, joy, intensity, calm, aggression 60 40 20 0 emotion categories fo re ig su nes m rpr s ad ise ne an ss ge s u lo r ffe ve cu rin rio g s c h ity ao s s a fu dn n e ag fes ss gr tivi es ty si ha on s s p f te iri ea tu r in a li m ten ty ov si e t bo m y re en do t ca m lm jo y frequency Frequency of 21 emotion categories for 6 non-western styles 80 70 60 50 40 30 20 10 0 Major emotion categories: joy, calm, boredom, movement, intensity, spirituality, fear emotion categories Familiarity versus liking • correlation 0.36 for modern western styles and 0.27 for non-western styles (both p<0.001) (replication of Peretz et al. 1998, Ali 2005) neutral I don’t like it 11 modern western styles I like it sounds familiar 105 87 58 neutral 95 181 134 sounds foreign 47 79 246 6 non-western styles I like it sounds familiar 131 46 49 neutral 70 91 44 sounds foreign 32 28 58 neutral I don’t like it Familiarity versus liking for 18 emotion categories and both styles (N>10) 3 disliked suffering anger chaos boredom madness spirituality fear sadness 2 aggression curiosity intensity festivity liked movement joy fun haste calm love 1 1 2 familiar 3 foreign Effect of style on familiarity and liking Modern western styles are more foreign and disliked (p < 0.001 ) I don‘t like it 3 2,5 2 I like it 1,5 modern western styles non-western styles 1 1 familiar 1,5 2 2,5 foreign 3 I don't like it Individual emotion categories – effect of style on familiarity & liking boredom N = 55 vs. 33 lik.: p<0.1, non-western 3 No differences for 2,5 haste 2 aggression movement aggression (N=29 vs. 12) joy (N=49 vs. 69) movement (N=22 vs. 27) haste (N=27 vs. 16) intensity (N=48 vs. 24) I like it intensity 1,5 joy 1 1 1,5 familiar fun N = 9 vs. 9 fam.: p=0.05 non-western 2 calm* N = 45 vs. 43 fam.: p<0.1 lik.: p<0.05 non-western 2,5 3 foreign modern western styles non-western styles Note: only emotion categories of comparable size were considered in the analysis 3 p < 0.001 mean ratings foreign/disliked Familiarity, liking & gender p = 0.010 2,5 p = 0.800 p = 0.004 2 1,5 1 familiarity liking modern western familiarity liking vs. non-western styles males females No gender difference for positive correlation „familiarity & liking“ in both styles (p < 0.001). Familiarity vs. liking and gender 11 modern western styles I don't like it No differences for movement (N =12 vs. 10), haste (N = 10 vs. 17), calm (N = 20 vs. 25), boredom (N = 33 vs. 22), joy (N = 23 vs. 26), aggression (N = 14 vs. 15) 3 sadness N = 14 vs. 8 fam.: p<0.1, ♀ anger N = 10 vs. 5 fam.: p<0.1, ♂ suffering N = 13 vs. 6 fam.: p<0.1, ♂ fear* N = 61 vs. 71 fam.: p<0.05, ♂ curiosity N = 9 vs. 9 fam.: p<0.1, ♂ I like it 2 males females intensity* N = 29 vs. 19 fam.: p=0.052, ♂ lik.: p<0.05, ♂ chaos N = 22 vs. 31 fam.: p<0.1, ♀ 1 1 2 familiar 3 foreign Familiarity vs. liking and gender 6 non-western styles No significant differences for fear (N=12 vs. 8), spirituality (12 vs. 8), boredom and calm I don't like it 3 boredom N = 21 vs. 12 fam.: p = 0.13, ♂ males females joy* N = 30 vs. 39 fam.: p<0.05, ♀ 2 intensity* N = 16 vs. 8 I like it lik.: p <0.05, ♂ calm N = 22 vs. 21 lik.: p=0.14, ♂ 1 1 familiar 2 foreign 3 Number of emotion and association words and gender All differences highly significant (p < 0.001) mean number of words 2,5 2 1,5 1 males 0,5 females 0 emotions associations emotions associations modern modern non-western non-western western western styles styles styles styles Conclusions (1) Main results • high correlation between familiarity and liking, independent of style and gender • indications of specific musical emotions (biased towards unfamiliar music): movement, chaos, madness… • foreignness and dislike are associated with fear, madness, anger, chaos, aggression • familiarity and liking are associated with movement, joy, calm, love, fun, curiosity Conclusions (2) Dependency on style • modern western styles are less familiar and liked • most frequent emotions in modern western music: fear, boredom, chaos, joy • most frequent emotions in non-western music: joy, calm, boredom, movement • emotion categories: – majority of categories: similar familiarity & liking ratings for both styles – indications of some exceptions Reasons for differences in ratings? specific stylistic features, liking & cognitive responses, methodological problem… Conclusions (3) Dependency on gender • modern western styles: less familiar and liked by females • non-western styles: no difference in familiarity, but more disliked by females • difference in emotional response: higher number of emotion/association words given by females • emotion categories of both styles: (un-) significant differences in several familiarity and liking ratings (indication of musical preferences?) Conclusions (4) • Are these results culture-specific? – Austria? („Austrian musical emotion lexicon“?) – Europe? – the West? • Are these results music-specific? – foreign art – foreign culture – foreign people & societies