2010:172
B A C H E L O R T H E S I S
Perceived Emotions in Music
If musical experience effects perception of emotion in music
Linn Fijal
Luleå University of Technology Bachelor thesis
Audio Technology Department of Music and media Division of Media and adventure management:
2010:172 - ISSN: 1402-1773 - ISRN: LTU-CUPP--10/172--SE
1
Perceived emotions in music
If musical experience effects perception of emotion in music.
C Essay
Linn Fijal 13/5/2010
2 Abstract
We know little about the mechanics behind emotions perceived in music because of the subjective nature of emotional experience. This makes it difficult to measure responses and make comparisons across individuals, since we have a factor that we can’t fully control; the individual’s perception and earlier experiences. But how big is the effect of earlier experiences in perception of music? In this essay the focus is on musical experience and whether musical experience makes a difference in how we perceive emotions in music. Listening tests were done by three groups of subjects with differing experience of music to see if there was any detectable difference. The groups where audio
engineers, musicians and people with very little experience in music. The results from the listening test showed indications of that the musicians perceive more emotions in the music. Any reasons for and against these signs are discussed.
3 Table of contents
1. Introduction 4
2. Aim__ 6
3. Background 6
4. Method 7
4.1. The listening test 7
4.2. Words for emotions 8
4.3. The music 9
4.4. The test subjects 10
4.4.1. The Audio Engineer group 10
4.4.2. The Musician group 10
4.4.3.The non musician group 10
4.5. The Data 11
5. Results 11
5.1. Emotions 18
6. Discussion 19
7. References 20
8. Appendix 23
4
Perceived emotion in music
1. Introduction
How people perceive emotions represented in music and emotion induction caused by music are two different concepts that needs to be separated in this essay. Even though there are some research pointing at the fact that one thing leads to another (Zentner, Grandjean, Scherer, 2008). Maybe it doesn’t even matter if the emotion is felt or perceived because some theories in music emotion induction suggest that the induction of emotion is depending on the perceived emotion and the listening environment. According to those theories music emotion induction is a consequence of the mind mimicking the perceived emotion (Zentner et al, 2008). In that case perceiving emotions in music is a good starting point for further research within the area of emotion induction. This essay will be focused on perceived emotion in music.
The concepts of felt and perceived emotion are in this case separated by the means that perceived emotion refers to intellectual processing; the perception of an intended or expressed emotional character. Felt emotion refers to the introspective perception of psychophysiological changes, which is often associated with emotional self-regulation. For example you can hear a piece of music and perceive it as happy, but you might not get affected by it or feel happier. Emotion or emotional are used in the controlled sense of feeling, meaning the sense of how an emotion feels to us. (Kreutz, Ott, Teichmann, Osawa, Vaitl, 2007)
A lot of research has been done earlier (Zentner et al, 2008, Kreutz et al, 2007, Ekman, 1992, Izard, 2007, Russell, 2003, Watson, Wiese, Vaidya, Tellegen, 1999, Frijda, Sundararajan, 2007, Lundqvist, Carlsson, Hilmersson, Juslin, 2009) maybe because music causing induction of emotion is an interesting subject for many areas of profession within the audio business today. Many audio professionals find they need to create music that suits a specific mood or purpose, or they need to increase or change the emotional content of a piece of music. This is frequently the case for music that will be or has been recorded. At present, there are not many guidelines that tell us how to apply musical techniques such as arranging, performance technique and sound engineering/mixing to achieve specific affects, and this would be very helpful. Take movie sound for example; imagine sound as a tool to induce exactly the right emotion in the audience in every scene of the movie. It could be useful. Imagine how you could use emotion induction with music in commercial causes to lift the credibility of the lifestyle you try to sell. Imagine you knew exactly what kind of sound, rhythms and other features in the music, trigged what emotion and the million ways you could use this knowledge in your work as an audio engineer. With a set of guidelines that told us how musical features mapped to the perception of emotions, we could produce music that was more likely to have the affects we hope to achieve. Emotion induction by music is an area that needs to be
explored, what does trig emotion induction and how does it work? Why does it work and how can it be used in todays audio business? The first step in understanding induction is perception. What emotions are perceptible in music?
First off, when we listen to musical instruments, we perceive many different aspects of the sound.
We hear the notes of the composition. We hear the instrument and how the performer plays it. We also hear the acoustics of the room/hall, or if the music is recorded, we hear the signal processing, microphones, effects and mix. All these things together give the music a certain emotional affect.
5 We know little about the mechanics behind music emotions because of the subjective nature of emotional experience. This makes it difficult to measure responses and make comparisons across individuals, since we have a factor that we can’t fully control; the individual’s perception and earlier experiences. In combination with mentioned earlier, music is a word describing a lot of different sounding material and can’t be rounded up to one easy measured excerpt. So what we got are two factors that are hard to define in a correct way, emotions and music, and we try to find out some links between them. Ways to find the link between music and emotions include self-reporting where subjects choose their own language or the experimenter imposing the words. However, if the method for measuring emotion induction are self-reports for example, there can be hard to find words that describe this felt emotion, and then it is hard to know if all the test subjects use the same words to describe the same emotion. If the problems of describing emotions in words are solved there’s still a problem for the test subject to know if the feeling they have in mind is felt or perceived.
The concepts of felt and perceived emotion are in this case separated by the means that perceived emotion refers to intellectual processing; the perception of an intended or expressed emotional character. Felt emotion refers to the introspective perception of psychophysiological changes, which are often associated with emotional self-regulation. For example you can hear a piece of music and perceive it as happy, but you might not get affected by it or feel happier. Emotion or emotional is used in the controlled sense of feeling, meaning the sense of how an emotion feels to us. (Kreutz, Ott, Teichmann, Osawa, Vaitl, 2007)
I would like to further look in to how people with different musical experience emotionally perceive different music, in the level of affect. To be able to do research on people with different musical experience, three categories have been made to label different groups in musical experience. The groups of people are musicians, audio-engineers and also people with little or no musical experience.
The reason for choosing these groups of people are because they have different expertise in music, these groups of people are often involved in the production of for example music and movies.
Musicians have the performance aspects and a lot of knowledge in musical structure in combination with a stylistic and historical perspective. Non musicians don’t have any expertise in music and audio engineers because they are trained listeners; they make the recordings and create the sound of it, but lack the performance aspect. Some possibilities are that musicians focus more on the technical performance and the playing and therefore are less sensitive to perceiving emotions in the music.
Non-musicians should then be the opposite, since they don’t have any expertise in music, and either perceives fewer emotions in the musical content, because they have a harder time understanding and analyzing what they hear. Or either perceive more emotions than the musicians because they only focus on the music itself. Audio engineers are trained listeners, like the musicians, but probably not as trained when it comes to playing the music. Audio engineers are trained listeners in another way and more focused on timbre, attack, reverberation, sustain, decay, etc. thus another type of specialized knowledge. Another possibility is that the musicians are trained in expressing emotions in music through their performances and therefore understands the different interpret cues that the composer puts in the music to convey emotions.
6 2. Aim
The aim of this study is to see if there are any differences in how these groups perceive emotion and what those cues are, if there are any.
This knowledge would help gaining deeper understanding in music emotion, which might come in handy when working with for example music in movies. It would also prove if musical background makes any difference in perceiving emotions in music. Potential findings would facilitate
understanding in music production and communication between musicians and audio engineers and for example movie directors, in case of making a production together where musical emotion is in focus. Audio engineers rarely work with and for only audio engineers, there are probably musicians involved and in the end, the final product probably reaches out to people without musical
experience. Understanding between these groups of people with different musical experience is important, especially for the audio engineer that often works as an interface between these groups.
This knowledge could help composers to write music that all sorts of audiences could understand and interpret the same. It could help audio engineers in creating sounds for everyone to perceive the same. It could also tell if this matter is worth putting more resources on for further research or if musical experience even matter in how we perceive music.
3. Background
The theories that I’m going to build this work upon say that music can induce emotions but there are some different theories about how and why. What the different researches agree on is that music induces emotions and mostly positive emotions. ( Kreutz et al, 2007, Lundqvist et al, 2009, Zentner et al, 2008)
Some of them believe in the basic emotion theory which says that all emotions induced by music can be divided into a few basic emotions. These emotions (like happiness, sadness, fear and anger) are often communicated in music through different acoustic cues like rhythm, tempo, loudness etc.
these acoustic cues are not unique for music, but can also be found in speech. So the main theory is that these cues, working together, have evaluated from speech and are then being recognized in the music and emotions are being induced in that way. (Ekman, 1992; Izard, 2007)
Another way to look upon emotion induction is the circumplex model, where you think in two different psychological dimensions; valence (negative to positive) and arousal (low to high). (Russell, 2003, Watson et al, 1999)
A third way to look upon music induction is referred to as the theory of emotion refinement that suggests that there is a difference between coarse and refined emotions and that refined emotions are more felt than acted upon. That musically induced emotions should be refined emotions. (Frijda
& Sundararajan, 2007)
There are also theories that humans, through complex psychological processes, mimic the emotions they perceive in the music and thereby feel the emotion that the music communicates. This theory also explain why music mostly induce positive emotions, the concise version is that the listening environment often is to safe and far away from any harm to induce the negative emotions. (Zentner et al, 2008)
7 Then there are a lot of ways to measure music emotion induction, but all of them need further research to cover the holes in them and to tell anything for sure.
The most popularway to measure emotions are through “self-reports” where the test subjects themselves get to describe what they feel. Problems with self-reports are that the test subject sometimes can have a hard time figuring out if the feeling they have are perceived or felt. Another problem is that there are no standard words to describe emotions, which leads to the possibility that several test subjects use different words to describe the same emotion or vice versa, which can be confusing in a research situation. Studies of perceived emotions face similar methodological problems. (Zentner et al, 2008)
Other ways to measure induced emotions are through measuring the physical responses, either in the facial muscles or the autonomic nervous system. In these cases the problems are that there is not really any research done on the coherence of different physical responses. If there are no given links between the physical responses, the physical responses might be totally individual and the
measurements don’t really tell anything useful. When it comes to differences depending on musical background, experience (amateur, professional) I have not found any credible researches or theories yet (Zentner et al, 2008). For these reasons, this study on perceived emotions will use a methodology that is not based on measuring physical responses.
4. Method
A listening test was set up, where people were listening to three different pieces of music and then rated the amount of perceived emotions.
4.1. The listening test
Listening tests were conducted to determine how strongly various emotions were perceived in musical excerpts. The test is based on the idea of filling in scales depending on how strong you perceive a certain emotion. Another motive is that the test should be easy to take and finish correctly, therefore the test needed to be short so the subjects wouldn’t lose focus. This is to minimize the impact of internal distractions like daydreaming. The listening test was designed to be possible to take at home when ever the subjects had any time and energy over. This was to minimize the impacts that a laboratory environment could cause during the listening test. The aim was to have as little internal and external distractions as possible and as little stress as possible, because total attention on the test is important for accurate results (Tellegen and Atkinson, 1974). The autor made the judgment that a neutral test environment where the subject felt safe was more relevant for the test results than the equally good sound quality that a laboratory designed for the cause would guarantee.
The test used a questionnaire that started out with instructions on how to take the test and contact information if they had any further questions. Then there followed questions asking about age, gender, occupation, musical experience e.g. In the listening test, for each of the three musical examples words for similar emotions where grouped together under a scale with a headline per scale, describing the group of words. The test consisted of one test form per each of the three sound excerpts. The aim of the design was to make it easy to overview and find the right emotion
describing word. The test subjects’ task was to mark the exact emotion word and fill in the amount of
8 that emotion perceived in the musical example on the scale. The test form will be found in the appendix.
4.2. Words for emotions
The first problem was to find words describing emotions in a good way. There has been research done on this matter before (Zentner et al, 2008) and they came up with the Geneva Emotional Music Scale, a list of words in French suitable for emotions inducted by music. This list was translated to English by the authors. Since this test was made in Sweden with Swedish speaking test subjects a translation and some rework was made on the scale to represent a Swedish version for perceived emotions and to avoid any linguistic confusion. As a result of lacking knowledge in the French
language I started with the English version. A first translation into Swedish was made. The translation was adjusted a bit to fit the modern Swedish language more correct, based on my sense of the Swedish language. Two copies of the English version were also sent out to two native Swedes with high skills in English. The first one was a 22 years old woman, whom had gone to an English speaking school for 15 years. The second one was a 50 years old man working internationally by lecturing in English. They translated and gave each their version of the list of words which was put together with my version into the final list.
List no. 1, The PPMEW. The words used in the listening test, based on the GEMS.
9 4.3. The music
When selecting the music there was some criterions it had to meet up with. First it needed obvious emotional content, but not too much. The aim was to find a piece of music that expressed one or a few emotions. Preferably, the music should not be commercially available to avoid that the listening test subjects already have a relationship to the music that would color their opinion about it. Other things that could color the test subjects opinion about the music are the instrumentation. Therefore the instrumentation of the music needed to be as transparent as possible, if an ensemble, no soloists that could be favored e.g. In the end the decision was to use plain solo piano, and make sure no test subjects had chosen piano as their main instrument.
The final selection of the music was written by Amanda Björnerås in 2007/2008. The three pieces consists of classical solo piano, recorded in one take each to capture the true performance right there and then at the recording session and presented to the subjects unedited. The original files from the composer was .mp3 for the listening test they were burnt down on a CD for the
distribution, without any further changes in sampling rate or coding. The audio quality of the files was comparable to commercially available MP3s, and there were no expectations that the audio quality would impact the perceived emotions. The pieces were presented in the listening test in their whole length, 57 s, 3.46 minutes and 2.44 minutes. The aim with the music used in the listening test
List no.2, the GEMS byZentner, Grandjean, Scherer, 2008
10 was originally to investigate what you as an instrumentalist could convey to your listeners, are the composers intended emotions the same emotions that the listener perceive or do the listener create their own perceived emotions from their own experiences and preferences. The results of Björnerås investigation indicated that all people have a way to perceive similar emotions in the music but also that the peoples earlier experiences in life effected how they perceived the music. Björnerås investigation is related to this one since it looks upon people’s perception of musical emotion but different since it only looks upon the perception in general and does not involve musical experience.
The first piece of music was written to describe stress. The second one love and happiness. It was inspired by Swedish folk music. The third was inspired by snow and a little bit more modern harmonically and without any obvious rhythm.
4.4. The test subjects
Summed together there was twenty people taking the listening test. Seven in the audio engineer group, seven in the musician group and six in the group of people without any musically related profession, the group called non musicians. All of them volunteered for the test through advertising on facebook. The subjects were mostly from Piteå and Stockholm but there was some exceptions. No one got any compensation for their participation. There was no controlling of the test subjects hearing. The test subjects did not need to test their hearing before taking the test or show any kind of certificate from a recent hearing test. The reason for this was that I wanted to examine how they perceived musical emotions from their everyday conditions in life. The only group that was
guaranteed to have no known hearing problems was the audio engineer group since all of them where students at the School of music and medias in Piteå and needed to have good hearing to access the audio engineer program.
4.4.1 The audio engineer group
The Audio engineer group had an average age of 24.6 years, the subjects was between 20 and 29 years old and of mixed gender. All of them had more than 10 years of experience in playing a musical instrument and listened to music around or more than 15 hours a week. The instruments and
preferred genre of music varied, the most popular ones was guitar, piano and drums on the instrument and rock/pop as the genre.
4.4.2 The musician group
The musician group had an average age of 23 years the youngest was 21 and the oldest 27. This group was also mixed in gender. The most common instrument was guitar and bass, all of them had played for more than 10 years and at least 15 playing hours a week. Their habits in music listening was in average 7-15 hours a week and they preferred very mixed genres but pop/rock was most common.
4.4.3 The non musician group
The non musician group was between 22 and 49 years old, with an average of 27.2 years. This group was also mixed in gender. Two of them had played guitar in their life, but none of them played more than 4 hours a week and both had little playing experience. Their music listening hours was mixed between 0-3 and more than 15 hours a week and the preferred genres varied but the most common was pop/rock.
11 4.5 The Data
The data received in the listening test was first put in to a table to get a good overview and to detect patterns; these tables will be found in the appendix. Then another three figures (Figure 1-3) was made with the same aim as the previous, but also with the focus on the spreading of grading within and between groups. These three figures showed the most frequently graded emotions for each group in each song. This clarifies which emotions most of the subjects perceived in each song. If the figures only would contain the highest rated emotions, the emotions that was rated low and by many of the subjects would be lost, and the emotions that one or two subjects rated high would show. This would be misleading for the results, because what matters for this essay is what the majority of the subjects rated, to be able to draw conclusions representative for the group and not only for one or two individuals. Since the group with the least participants had six people taking the test, the limit for when the results could be representative for the group was set to three test subjects. The limit for the other two groups, with seven test subjects was set the same. (Figure 1-3) is completed with a bar showing the variance for each group. This variance is counted on all emotions, not only the ones in the figure, to illustrate the variance of ratings in the group to. To find patterns within the groups, if some subjects favoured some emotions or if one group favoured some emotions etc, another three figures (Figure 4-6) was made showing how each group rated all three songs.
5. Results
First three figures (figure 1-3)will be presented, one for each song, comparing the groups most graded emotions. After that another three figures will follow (figure 4-6), showing how each group rated all songs, to be able to compare between the songs. This section ends with a deeper look at the emotions and the music. the complete raw results from the listening test will be presented as three tables (table 1-3), one for each group in the appendix.
Looking at the results you can clearly see differences between the pieces when it comes to rated emotions and as well similarities between the groups in the ratings of the different pieces. The main amount of rated emotions are similar, or closely related for all groups in each piece.
Looking closer, to start out with the stress-piece and all the emotions that has been rated by three subjects or more, which are half of the subjects for one group (50%) and a little less than half (approximately 43%) of the subjects for the other groups. The emotion highest rated by the most subjects are for all groups worry, after that the results are a little more mixed with Stress, fiery, thrills, energetic, lively, nervous, joyful and sorrowful. A major part of these emotions can be related to stress in one way or another. This is the piece where all the groups are most consistent; many emotions are rated by all the groups. The variance is also about even between the groups. The figure shows the emotions that more than three subjects, in each group graded in the listening test for the piece referred to as “Stress”.
12 The love piece in the same criterions as above shows a little more spread results. The rated emotions are peacefulness, relaxed, affectionate, inspired, happy, dreamy, sad, calm, spirituality, stimulating and nostalgic. The love piece is interesting to look at, because the ratings are clearly divided between the groups except for peacefulness that all groups rated high. Some ratings could be different words for similar emotion like calm and relaxed, but these are in the same group. So the question you can ask here is if the divided ratings are a coincidence or not.
0 5 10 15 20 25 30 35 40 45
worry fiery thrills stressed joyful energetic nervous lively sorrowful Variance
Stress
Emotions with more than 3 ratings and variance
Non musicians Musicians Audio engineers
Figure no. 1
The figure shows the emotions most commonly rated, group by group for the stress piece. All the ratings from each group are summed together to show which emotions was most frequently rated and how high it was rated by the groups. The emotions in the chart got at least three ratings in the listening test by at least one group. The bars named Variance shows the variance of the whole groups ratings, not only the emotions shown in the chart, to se how spread the groups’ ratings was.
Figure no. 1
13 The snow piece is not written about an emotion, but more a natural phenomenon and possibly could be more spread in its results, and it is spread in the results, even though relatively few different emotions are rated. In the previous experiment that the stress piece was written for, the point was to see if music could convey not only emotions but also a picture. To see if you could convey the whole picture or if the listener creates its own picture based on its own experience and relationship to the emotions conveyed. Björnerås found that it is hard to convey pictures, and that it is most likely the listener that creates its own picture based on associations to the piece. This time it is expected to
Figure no. 2
The figure shows the emotions most commonly rated, group by group for the love piece. All the ratings from each group are summed together to show which emotions was most frequently rated and how high it was rated by the groups. The emotions in the chart got at least three ratings in the listening test by at least one group. The bars named Variance shows the variance of the whole groups ratings, not only the emotions shown in the chart, to se how spread the groups’ ratings was.
Figure no. 2
14 see that even though the composer is focused on conveying a picture or several emotions, the listening test subject still perceive those emotions. The most rated emotions are tender, sadness, excitement, nervous, meditative, dreamy and wonder which can be related to. Worry, nervous and excitement can be describing similar emotions, just like sadness and sorrowful and like meditative and dreamy.
As well as there is likeness between the groups in all pieces, there are no consisting differences between the groups that last throughout all pieces. The average grading/emotion are 12 for musicians and non musicians, 11.5 for the audio engineer. This means no significant difference between the groups when it comes to the sense of amount of emotions. Counting on how many emotions each group rated the audio engineers and musicians both graded 69 emotions and the non musicians graded 73. The following graphs show the groups ratings across all three pieces.
0 5 10 15 20 25 30 35 40 45 50
tender sadness excitement nervous meditative dreamy wonder sorrowful joyful worry Variance
Snow
Emotions with more than 3 ratings and variance
Non musicians Musicians Audio engineers
Figure no. 3
The figure shows the emotions most commonly rated, group by group for the snow piece. All the ratings from each group are summed together to show which emotions was most frequently rated and how high it was rated by the groups. The emotions in the chart got at least three ratings in the listening test by at least one group. The bars named Variance shows the variance of the whole groups ratings, not only the emotions shown in the chart, to se how spread the groups’ ratings was.
Figure no. 3
15
Figure no. 4
Figure no. 4
Shows how the audio engineers rated through the whole listening test.
All the ratings are summed together, and divided in different bars for the three pieces.
This is to show possible patterns in how the audio engineers voted in the listening test.
16
Figure no. 5
Figure no. 5
Shows how the musicians rated through the whole listening test. All the ratings are summed together, and divided in different bars for the three pieces.
This is to show possible patterns in how the musicians voted in the listening test.
17
Figure no. 6
Figure no. 6
Shows how the non
musicians rated through the whole listening test. All the ratings are summed together, and divided in different bars for the three pieces. This is to show possible patterns in how the non musicians voted in the listening test.
18 5.1 Emotions
Looking at the most rated emotions from all three groups, some patterns can be seen. This is only a theory, built upon the fact mentioned in the introduction that there are no universal way to describe emotion. The emotion perceived by the subject might just be as individual as the word he/she uses to label it with. A way to deal with this is to divide the emotions in to four “fields” based on valence (negative – positive) and arousal (high – low) (Figure no.7). This way of grouping the emotions are in many ways ambiguous, since there is more to emotions than valence and arousal, but it can illustrate one way to divide emotions into groups.
The non musicians for example have three words out of eight (13,2% of all ratings summed together) that might be words for the same emotions; peaceful (4,3% of all ratings) , relaxed (4,5% of all ratings) and meditative (4,7% of all ratings). Meditative and dreamy (together 9,6% of all ratings) could also be related in the same way. They also have stressed energetic and nervous (together 14,2% of all ratings for the non musicians), that also possibly could be related to the same emotion.
This is interesting because it is two groups of emotional opposites. The group of musician also have some words that can be related to the same emotion; stressed, worry, nervous and thrills are all
Figure no. 7
This illustration is a version of R. E Thayer’s valence-arousal model, with the words from this case. The underlined words are from Thayer’s original model, and the other ones are
examples of how the emotions from this case could fit in it. _______________
19 words for some kind of tension and they represent 50% of the highest rated. Musicians also rates stimulating high, which they are the only group who does. Almost all of the rates for stimulating among the musicians were in the love piece. The audio engineers excel with two words, which can be translated to the same Swedish word or put to the same emotion; affectionate and tender. Then they have the group of some words for tension, like the other groups. Both non musicians and audio engineers have some word for sorrow, sorrowful and sad, the musicians have none. Beneath you find a mapping picture of the words that was rated higher than 25 for all the groups. Common words are marked with red and some related with green.
6. Discussion
When looking at the results and the figures with the emotions rated more than three times per group, there is a clear difference between the groups in one song, Love. Here you can clearly see how the three groups have rated differently, except for peaceful that all groups rated high. The non musicians have rated that they perceive dreamy, spirituality and sadness, emotions with low valence, whereas the musicians had really high ratings on high valence emotions like stimulating and the audio engineers in agitated and inspired and happy, emotions with high arousal and positive valence.
This is the only song where this distinct difference between the groups has appeared. Since I can’t find the same patterns in the rest of the songs I have to believe that the divided result depends on the different individuals in the groups in combination with the characters of the stimuli. Even though some tendencies are found, there is a bigger possibility that the tendencies depend on the human factors, the ones not possible to control and coincidence.
A pattern that goes through all songs is that the musicians are alone on rating several emotions in every piece of music. The only pattern between the musicians emotions only rated once are that there is no connection between them and none of the emotions are rated only once by the musicians in more than one piece. This might be a sign of that the musicians perceive more emotions in the music. A possible reason for this might be the performance aspect, because of their profession in expressing emotions in music while playing, and that this have made them recognize the musical codes for conveying emotions. Another possibility is that they do not know exactly what they
Figure no. 8
shows the emotion words rated higher than 25, for each group. Words that can be related are marked with green and words that can be seen as common with red.
20 perceive emotionally, so they grade several emotions out of insecurity of what they really hear in the music.
Looking at the grading in whole the non musicians has graded most different featured emotions, which can be viewed upon as a sign of insecurity when grading the emotions. This might be because they have not been trained or conditioned to think of music according to any emotional descriptions.
It might also be a sign of that they are not as used to put emotional words on what they perceive in music. This can be because they are not trained to think in the terms of musical or emotional conformation like musicians. That the audio engineers are trained in how to convey emotions in music are not something you always can take for granted, but in this listening test all of the audio engineers had a musical background. They probably also had some knowledge in sound design, which means that they are familiar in thinking in terms of emotions and audio and how they connect.
The result from the listening tests shows no distinct differences between the groups throughout the whole test, the patterns found are vague. Maybe there are a little too few music excerpts to cover the whole spectrum of perceived emotions. This possibility takes away the chance to draw
conclusions about which emotions are easiest to detect in music and how it affects the results. Since there is no music excerpt written to describe for example happiness, there is no wonder that words for happiness is missing in the list of highest rated. There is also a possibility that the instructions for the listening test were not clear enough for all the subjects to fill in the forms on equal basis. Another source of error is the few test subjects and the fact that most of the test subjects where between twenty and thirty years old and from the institution of music and media in Piteå, and can’t be held representative for all of these professions in Sweden.
To summarize the vague patterns found in the listening test, we first have the distinct differences in the love piece that does not appear in any of the other two pieces. This difference is now seen as a result of coincidence. But maybe some clarity could be found if more research would be done on similar excerpts containing positive emotions or Swedish folk music and more subjects. In the results musicians were also solo grading most emotions, which are seen as either a sign of that the
musicians perceive more in the music or that they are insecure of how to put words on what they hear. To further research this matter some other test form would be more suitable. Maybe
interweaving the subjects could help. The third and last pattern is that the non musicians grade most different emotions of the groups which are interpreted as lack of knowledge in how to put words on musical emotions or insecurity of what they hear. To find out more extended listening tests with interweaves could be a solution. Until more extended research has been made I interpret the results as that no adjustment needs to be done to highlight emotional content in a piece of music to adapt it to a certain group of listeners with a certain musical or non musical background. I draw this
conclusion since from the point where we stand now the only thing we know for sure is that people perceive the music individually. The indications that people with different musical background perceive music different are still too vague to know exactly how to apply them on real audio work.
21 7. References
Björnerås A (2008). Säg det i toner: en undersökning av möjligheten att i musik förmedla känslor och bilder, ISSN 1402-1773/ISRN LTU-CUPP--08/319--SE / NR 2008:319
Ekman, P. (1992). An argument for basic emotions. Cognition & Emotion, 6, 169–200.
Frijda, N.H. & Sundararajan, L. (2007). Emotion refinement. Perspectives on Psychological Science, 2, 227-241
Izard, C. E. (2007). Basic emotions, natural kinds, emotion schemas, and a new paradigm Perspectives on Psychological Science, 2, 260–280.
Kreutz G, Ott U, Teichmann D, Osawa P, Vaitl D (2008). Using music to induce emotions: Influences of musical preference and absorption.
Psychology of Music 2008; 36; 101 originally published online Nov 12, 2007; DOI:
10.1177/0305735607082623
Lundqvist L.O, Carlsson F, Hilmersson P, Juslin P (2009). Emotional responses to music; Experience, expression and physiology. Society for Education, Music and Psychology Research vol 37(1): 61–90 [0305-7356 (200901) 37:1; 61–90]10.1177/0305735607086048
Russell, J. (2003). Core affect and the psychological construction of emotion. Psychological Review, 110, 145–172.
Tellegen, A. and Atkinson, G. (1974) Openness to Absorbing and Self-altering Experiences
(‘Absorption’), a Trait Related to Hypnotic Susceptibility, Journal of Abnormal Psychology83:
268–77.
Thayer R.E (1989). The Biopsychology of mood and arousal, Oxford University Press, ISBN 0-19- 506827-0 (PBK)
Watson, D., Wiese, D., Vaidya, J., & Tellegen, A. (1999). The two general activation systems of affect:
Structural findings, evolutionary considerations, and psychobiological evidence, Journal of Personality and Social Psychology, 76, 820 –838.
Zentner M, Grandjean D, Scherer KR (2008).Emotions Evoked by the Sound of Music:
Characterization, Classification, and Measurement. No. 4, 494–521
22 8. Appendix
First the raw data from the listening test, then follows the listening test in its original form.
23 Table no. 1
The raw data from the listening test. The audio engineer group.
Table no. 1
24 Table no. 2
The raw data from the listening test.
The musician group.
Table no. 2
25 Table no. 3
The raw data from the listening test. The non musician group.
Table no. 3
26 Here follows the listening test, in its complete form as handed out to the listening test subjects.
Instruktioner till lyssningstest Introduktion
Detta är ett lyssningstest för att undersöka uppfattningen av känslor i musik. Den grundläggande uppgiften är att gradera känslorna du uppfattar i tre olika musikstycken samt några frågor om din relation till musik.
Lyssningsmiljö
Lyssningstestet ska genomföras i en lugn miljö som är så störningsfri som möjligt (tyst, lugn, ingen facebook).
Frågeformuläret
Börja med att fylla i den första sidan med uppgifter. När detta är klart bläddra till nästa sida, där du kommer att hitta en rad skalor för olika kategorier. Under kategorierna finns ett antal beskrivande ord. Läs först igenom alla ord för att bilda en uppfattning om dem. Sätt igång musiken i den ordning den kommer på skivan. Lyssningsnivån är valfri så länge den är konstant under hela lyssningstestet.
Lyssna igenom första stycket en gång och fyll i formuläret. Gradera så många skalor du vill genom att kryssa i rutorna, men var noga med att markera den känsla du syftar på för varje skala. Försök att gradera alla känslor du uppfattar i musikstycket. Utifall du inte anser att det finns ord för att beskriva någon exakt känsla så kan du använda rutan nämnd annat under varje kategori. Rutan annat är namngiven med ett nummer, fylls rutan i skrivs föredraget ord på raden efter rätt nummer under rutan. När du lyssnar på nästa musikstycke, vänd blad, så du för varje musikstycke har ett tomt blad att fylla i. Avsluta med att fylla i frågorna på sista sidan och lämna tillbaka formuläret.
Estimerad tid
Testet kommer maximalt att ta 30 minuter att genomföra.
Frågor?
Ring mig på 073 707 5678
Eller kontakta via mail på linfij-7@student.ltu.se När testet är klart
Kuvertet med CD-skiva och samtliga ifyllda papper lämnas alternativt;
- I min brevlåda på ankars 85b, brevlådan heter L Fijal.
- Till mig personligen
- I mitt fack i skolan, finns utanför radioredaktionen/PC-redaktionen, längst in i G-korridoren.
- Postas till min hemadress.
27
28
29
