Grand Valley State University
ScholarWorks@GVSU Student Summer Scholars
Undergraduate Research and Creative Practice
2015
Music Video in a New Light: The Color of Music and the Music of Color Gabriel Z. Ellis Grand Valley State University,
[email protected]
Lee Copenhaver Grand Valley State University,
[email protected]
Follow this and additional works at: http://scholarworks.gvsu.edu/sss Part of the Music Commons Recommended Citation Ellis, Gabriel Z. and Copenhaver, Lee, "Music Video in a New Light: The Color of Music and the Music of Color" (2015). Student Summer Scholars. Paper 146. http://scholarworks.gvsu.edu/sss/146
This Open Access is brought to you for free and open access by the Undergraduate Research and Creative Practice at ScholarWorks@GVSU. It has been accepted for inclusion in Student Summer Scholars by an authorized administrator of ScholarWorks@GVSU. For more information, please contact
[email protected].
1
Music Video in a New Light: The Color of Music and the Music of Color Gabriel Ellis and Lee Copenhaver Grand Valley State University Abstract Scholars have posited relationships between the color palettes employed in music videos and song characteristics such as tempo, musical mode, genre, and lyric emotional expression. However, these relationships are typically examined within the context of individual videos and presented as case studies. In this study, we quantified the color palettes of 160 recent music videos from four popular genres by utilizing “color barcodes,” visual representations of the colors employed within videos, and looked for large-scale correlations between color palette and genre, musical mode, tempo, and performer gender. We also used linguistic analysis software to analyze positive emotion, negative emotion, and anger within the lyrics of our sample songs, and compared these to video color palettes. We found that although musical characteristics such as mode, tempo, and lyrical positive emotion have little measurable impact on video color content, factors such as musical genre may impact both video color content and musical/lyrical characteristics such as anger and negative emotion. Introduction and Literature Review Color and Music Associations between music and color have been commonly made by musicians (as in the “cascade of blueorange chords” in Messiaen’s Quatuor pour la fin du temps) and by visual artists (as in Paul Klee’s painting Fugue in Red). Attempts to scientifically relate the two date back to at least Plato, who attempted to relate color to interval harmony, and numerous scholars have tried to model their relationship by drawing on similarities between their physical properties.1 However, such models are of limited usefulness.2 The innate differences between the physics and structure of visual art and music render a true one-to-one correspondence impossible; although the assignation of segments of the visible spectrum to pitches (C = red, C-sharp = orange, etc.) is possible, any such assignation is arbitrary.3
1
Tawny Tsang and Karen B. Schloss, “Associations between Color and Music Are Mediated by Emotion and Influenced by Tempo,” The Yale Review of Undergraduate Research in Psychology, 2 (2008): 83. 2 Karel Lill and Lee Copenhaver, “Comparative Theories of Visual Art and Music: May I Play You a Picture?,” Mcnair Scholars Journal 16 (2012): 37. 3 Ibid.
When attempting to relate music and color, many scholars also reference the neurological phenomenon of synesthesia. Those with this condition perceive consistent and involuntary associations between different modes of sensory experience. When synesthesia involves associations between music and color, it is called chromesthesia; chromesthetes may associate musical characteristics such as pitch, key, or timbre with individual shades of color. However, chromesthesia is also of limited usefulness when it comes to explaining general associations between music and color. True chromesthesia is both rare and inconsistent in its severity, and the associations made by chromesthetes vary significantly from individual to individual. 4,5 However, recent research suggests that nonchromesthetic individuals also make associations between music and color.6,7 Studies have shown that tempo, timbre, musical mode, interval harmony, and note density all affect the color associations of non-chromesthetes in experimental trials.8,9 In order to understand music-color associations among non-chromesthetes, it is necessary to examine the role of emotion. The idea that emotion can be used to mediate associations between music and color is hardly new; in 1942, V. G. W. Harrison wrote: “In the art of the future, it is quite possible that color and music will be combined. If the combination is to be made successfully, however, the connecting link must be the particular feeling which the artist is attempting to convey. Color and music will form one complete whole, but there need not necessarily be any direct correspondence, color for note, between the color and musical score.”10
4 S Schwettman, S Churchill, and D Eagleman, “A Large-Scale Analysis of Music-Color Synesthesia,” April 13, 2013. 5 Lill and Copenhaver, “Comparative Theories of Visual Art and Music: May I Play You a Picture?” 6 Karen B. Schloss, Patrick Lawler, and Stephen E. Palmer, “The Color of Music” (Vision Sciences Society Annual Meeting, Naples, Florida, 2008). 7 William Griscom and Stephen E. Palmer, “The Color of Musical Sounds: Color Associates of Harmony and Timbre in Non-Synesthetes” (Vision Sciences Society Annual Meeting, Naples, Florida, 2012). 8 Schloss, Lawler, and Palmer, “The Color of Music.” 9 Griscom and Palmer, “The Color of Musical Sounds: Color Associates of Harmony and Timbre in Non-Synesthetes.” 10 V. G. W. Harrison, “Colour and Music,” Journal of the Royal Society of Arts 90 (August 7, 1942): 611.
2 As he predicted, recent research suggests that music-color associations in non-chromesthetes tend to be mediated by emotion. When asked to associate a color with a musical selection, individuals rely on the emotional connotations provoked by both stimuli11,12 (so when listening to music that is “happy,” people will typically select a “happy” color to accompany it. 13 ) Both music and color have emotional valence, and there is consistency in the emotions that both colors and musical characteristics typically evoke (i.e., bright colors are typically associated with more positive emotions than dark colors and major-key melodies are typically associated with more positive emotions than minor-key melodies).14 It is therefore possible to identify large-scale trends among the associations experienced by non-chromesthetes, unlike among those made by chromesthetes.15 However, it remains to be seen whether the associations suggested in these trials are manifested in forms of multimedia that include both musical and color elements, such as music video. Color and Music Video Music video has always been a “hotbed of color experimentation,” 16 and the recent advent of Digital Intermediate (DI) technology has given music video directors and artists an unprecedented level of control over color.17 Using DI, colorists can drastically alter the color content of footage in post-production, applying broad, sweeping strokes of color and light or changing details as subtle as the color of a performer’s eyes.18 This gives contemporary directors and artists a new range of options when it comes to the expressive use of color. One potential application is the use of color to mirror musical elements. Carol Vernallis notes that using DI, colorists can “tweak each shot individually, so that shots
11
William Griscom and Stephen Palmer, “Emotional Mediation of Cross-Modal Associations in Timbre-Color Synesthesia,” Journal of Vision 14, no. 10 (August 22, 2014): 1003–1003, doi:10.1167/14.10.1003. 12 Stephen Palmer et al., “Color, Music, and Emotion,” 2012. 13 Stephen Palmer et al., “Music–color Associations Are Mediated by Emotion,” Proceedings of the National Academy of Sciences 110, no. 22 (May 28, 2013): 8836–41, doi:10.1073/pnas.1212562110. 14 Tsang and Schloss, “Associations between Color and Music Are Mediated by Emotion and Influenced by Tempo,” 82. 15 Griscom and Palmer, “The Color of Musical Sounds: Color Associates of Harmony and Timbre in Non-Synesthetes.” 16 Carol Vernallis, Unruly Media: YouTube, Music Video, and the New Digital Cinema (New York: Oxford University Press, 2013), 215. 17 John Belton, “Painting by the Numbers: The Digital Intermediate,” Film Quarterly 61, no. 3 (March 2008): 58, doi:10.1525/fq.2008.61.3.58. 18 Vernallis, Unruly Media, 216.
can be made to reflect musical changes.”19 Vernallis is one of many scholars who have attempted to describe the functions of color in music video, and although much conceptual progress has been made, conclusions in the literature about the role of color in music video are elusive. Perhaps due to the difficulty of quantifying the color content of videos, scholars tend to favor case studies to large-scale analyses, resulting in an oversaturation of micro-level analyses. However, some generalizations are possible. Color is frequently used within music video to evoke emotion and to create or illuminate form and structure. It may also be used to reflect subtle musical elements such as texture and timbre, and may be related to factors such as performer gender and musical genre. Within music video, color is often used to complement the emotional quality of the music. Andrew Goodwin notes that color can be used in a video to evoke emotion in the same way that lyrics, timbre, and tempo can in a song.20 He describes the use of blue-tinted footage to create a melancholy atmosphere in Suzanne Vega’s Luka, which tells the story of an abused child: “The simplest commutation test (imagine the clip in color, or tinted red, purple, or yellow) reveals the contribution of the blue-filtered tinting to the clip’s affective intent. Color is used here to suggest the blues - a connection that is much more than an iconographic pun.”21
Pamela G. Taylor also focuses on the evocative properties of color, suggesting that it can be used to represent “emotions, feelings, or meanings.”22 Color is also used as a structural device within music video. Martina Elicker suggests that “emphatic visual changes” can be used to draw attention to changes in melody or key, for instance that changing palettes can be used to distinguish chorus from verse. 23 Goodwin describes several instances in which switches between monochrome and full-color footage are used to reflect
19 Carol Vernallis, “Music Video, Songs, Sound: Experience, Technique and Emotion in Eternal Sunshine of the Spotless Mind,” Screen 49, no. 3 (September 1, 2008): 286, doi:10.1093/screen/hjn036. 20 Andrew Goodwin, Dancing in the Distraction Factory: Music Television and Popular Culture (Minneapolis: University of Minnesota Press, 1992), 65. 21 Ibid., 66. 22 Pamela G. Taylor, “Press Pause: Critically Contextualizing Music Video in Visual Culture and Art Education,” Studies in Art Education 48, no. 3 (Spring 2007): 70. 23 Martina Elicker, “Popular Music in the Age of the Video Clip,” AAA: Arbeiten Aus Anglistik Und Amerikanistik 23, no. 1 (January 1, 1998): 71, doi:10.2307/43025556.
3 song structure.24 Vernallis describes the use of color to end.25
visually divide a video into beginning, middle, and Changing color palettes can be used to reflect subtle musical characteristics such as timbre, harmony, and texture. Fred Collopy, discussing the use of color in lumia (a form of musical multimedia in which geometric figures rotate and dance across fields of color and light), states, “It is possible to use color combinations to create a variety of moods, much as chord structure and other elements of harmony in music are used to establish moods. Indeed, such uses in music are sometimes even referred to as ‘coloring’ the music.” 26 Vernallis similarly states that changing colors can “draw attention to texture, surface and materiality.”27 Color may also be tied to factors like performer gender and race. Vernallis, noting the frequency with which brown and gold are employed in videos featuring African American performers, suggests that certain colors may be used in order to complement the skin tone of performers.28 She also notes that color is related to gender; that pink for instance can carry connotations of femininity.29 Finally, color may be related to musical genre. Vernallis states that certain colors appear more frequently than others in particular genres of music video. 30 , 31 Goodwin suggests that certain “visual signifiers” may be associated with different genres of music video, although he does not specifically mention color.32 The purpose of our study was to empirically determine whether these suggestions in the literature are apparent in a representative sample of recent music video. Due to the inherent difficulty of quantifying textures, timbres, and form within songs, we focused on the relationships between video color content and easily quantifiable musical characteristics such as performer gender, mode, tempo, genre, and lyrical content.
24
Goodwin, Dancing in the Distraction Factory, 64. Carol Vernallis, Experiencing Music Video: Aesthetics and Cultural Context (New York: Columbia University Press, 2004), 126. 26 Fred Collopy, “Color, Form, and Motion: Dimensions of a Musical Art of Light,” Leonardo 33, no. 5 (2000): 359. 27 Vernallis, “Music Video, Songs, Sound,” 285. 28 Vernallis, Experiencing Music Video, 125. 29 Ibid., 126. 30 Vernallis, Unruly Media, 216. 31 Vernallis, Experiencing Music Video, 125. 32 Goodwin, Dancing in the Distraction Factory, 66. 25
Methodology Video analysis We began by downloading from YouTube a sample of 160 award-winning music videos, 40 from each of four popular genres: rock, pop, hip-hop, and country. Every year, MTV nominates 5 music videos for their “Best Pop”, “Best Rock”, and “Best Hip-Hop” Video Awards and chooses 1 overall winner in each category. For each year between 2008 and 2015 we downloaded the 1 winner and remaining 4 nominees in each of these categories. As MTV has no award category for country videos, we selected the country videos from the pool of winners and nominees for the Country Music Television “Video of the Year” award. The number of nominees for this award varies from year to year, so we randomly selected 5 videos from the pool of award-winners and nominees for each year between 2008 and 201533. A list of the videos that made up our sample can be found in Appendix A. In order to distill the color content of the selected videos into a more easily quantifiable form, we then used the Color Barcode Generator software programmed by Melvin Laily 34 to convert each video into a “color barcode”—a visual representation of the use of color throughout the video. This software is programmed to take 1000 frames at equal intervals from within each video, compress each frame into a column 1 pixel in width and 500 pixels in height, and arrange these columns into a row 1000 pixels in width. The resultant images illustrate the general color scheme of a video and demonstrate the way that the color palette of a video changes as it progresses. (Figure 1) We next used the Image Color Extract software programmed by Kepler Gelotte 35 to find the ten most representative colors of each barcode. This software searched each image and chose the ten most commonlyoccurring colors based on the number of pixels of each color that appeared in the image. This software allows some customization of settings; the settings we used were chosen to compromise between color accuracy and color diversity. If several different shades of red made up the majority of an image, we did not want them to be compressed by the software into a single shade of red. However, we did not want the software to describe each individual shade of red at the expense of describing the less frequently-occurring colors present in the image. The
33
As there were only 4 nominees for this award in 2008, an extra video was randomly selected from 2009. We have deemed this slight discrepancy not detrimental to the goals of our study. 34 Melvin Laily, Color Barcode Generator, version 0.5, 2013, http://arcanesanctum.net/movie-barcode-generator/. 35 Kepler Gelotte, Image Color Extract, version 1.1, n.d., http://www.coolphptools.com/color_extract.
4 Quantization Delta setting of the software, as it is decreased, increases the accuracy of the detected colors but increases the number of similar colors detected. The quantization delta was set to 30, which we found to be the
can be placed on a spectrum from cool to warm is common in color theory. Although specifics of classification vary depending on source, it is generally acknowledged that “colors that appear closer to the red end of the visible
Figure 1. Color Barcode for Beyoncé’s Video Phone, featuring Lady Gaga. Note the way the barcode reflects, left-to-right, the black-and-white introduction to the video, the subsequent introduction of a rapidly flashing red color, and the remainder of the video’s characteristic use of bright red, blue, green, and yellow hues juxtaposed with faded greys and browns.
best compromise between color diversity and accuracy. The options to “Reduce Brightness Variants” and “Reduce Gradient Variants” were both disabled, as they proved detrimental to color accuracy. The ten representative colors detected by the software were expressed as hexadecimal codes, six-digit numbers that digitally represent individual shades of color by describing their hue, saturation, and value. We then used the ColorHexa Color Encyclopedia36 to translate the hexadecimal codes into color names. This software assigned one of 14 names to each code: black, gray, blue, violet, cyan, green, lime green, yellow, yellow [olive tone] orange, orange [brown tone] red, magenta, and pink. As “yellow [olive tone]” and “orange [brown tone]” occurred frequently, we changed all their occurrences to “olive” and “brown”, respectively. “Lime green” occurred rarely; we combined it with “green.” “Pink” occurred in the results of only two barcodes, and we combined it with “magenta.” The result was that each hexadecimal code was translated into one of 12 color names: black, gray, blue, violet, cyan, green, olive, brown, yellow, orange, red, or magenta. In order to measure the average brightness of each color barcode, we also used the Image Color Summarizer software programmed by Martin Krzywinski. 37 This software represents the brightness of each image with a number between 1 and 100; higher numbers represents brighter images. Finally, for summary analysis, we classified each of our 12 colors as “cool” or “warm.” The concept that colors
36
ColorHexa Color Encyclopedia, colorhexa.com. Martin Krzywinski, Image Color Summarizer, version 0.52, http://mkweb.bcgsc.ca/color_summarizer/.
37
spectrum are said to be warm while the colors that appear closer to the blue end are said to be cool.”38 Xiao et al. suggest that the boundary between warm and cool typically falls somewhere between green and yellow;39 for our analysis we therefore placed the boundary between “olive” and “yellow.” Although black and gray are not typically included on the cool-warm continuum, Ou et al. suggest that they fall closer to cool than warm,40 and we therefore elected to classify them as “cool.” Figure 2 illustrates our 12 colors and their classification (“Cool” = black, gray, violet, blue, cyan, green, olive; “Warm” = yellow, orange, brown, red, magenta).
Figure 2. “Cool” and “Warm” Colors
For analysis of Tempo, tempos were recorded using a “tap tempo” web application, which calculates BPM (beats per minute) as a computer space bar is tapped in time with the beat of the song. A minimum of 30 taps was used for each song in order to achieve relative accuracy. Songs that
38
Reynold J Bailey, Cindy M Grimm, and Chris Davoli, “The Real Effect of Warm-Cool Colors,” All Computer Science and Engineering Research 2006, 1. 39 Youping Xiao et al., “The Biological Basis of a Universal Constraint on Color Naming: Cone Contrasts and the Two-Way Categorization of Colors,” PLoS One 6, no. 9 (2011): 5. 40 Li-Chen Ou et al., “A Study of Colour Emotion and Colour Preference. Part I: Colour Emotions for Single Colours,” Color Research & Application 29, no. 3 (2004): 237–238, doi:10.1002/col.20010.
5 switched tempos for only one section, such as the bridge, were classified according to the tempo in which they primarily existed. For analysis of Mode, songs in the Dorian mode were grouped with songs in Minor; songs in the Mixolydian mode were grouped with Major. Songs that repeatedly alternated between Minor and Major (Lady Gaga - Poker Face) were excluded. Songs that switched modes for only one section, such as the bridge, were classified according to the mode in which they primarily existed. For analysis of Performer Gender, songs were classified based on the gender of the musical performer(s), which typically but not exclusively corresponded to the performer(s) featured in the music video. Songs featuring both male and female performers (Katy Perry featuring Snoop Dogg - California Gurls) were excluded from this analysis, as were those performed by musical groups that included both male and female performers (Lady Antebellum - We Owned the Night). Lyrical analysis In order to evaluate the emotional content of song lyrics, we used the Linguistic Inquiry and Word Count (LIWC) software developed by James W. Pennebaker, Roger J. Booth, and Martha E. Francis. 41 LIWC was created to calculate occurrence rates of positive and
and other mistakes. In order to prepare lyrics for analysis, we corrected any obvious remaining typos and replaced designatory comments such as “chorus” with the full text that they represented. We did not alter non-lexical material (“ooh”, “whoa”). Individual text files were created for each song and were analyzed using LIWC. Results In order to examine the effects of performer gender, tempo, modality, and genre, we began by categorizing the videos according to each of these metrics and counting the total occurrences of each color within each category. The graphs in this section represent the relative frequency of our 12 “primary” colors in the order black, gray, blue, violet, cyan, green, olive, brown, yellow, orange, red, magenta. Graphs were prepared using using Microsoft Excel and statistical analyses performed using R.46 Performer Gender Figure 3 graphs the relationship between Performer Gender and video color content. A one-way ANOVA showed that the levels of Warm Colors did not differ significantly by Gender: Female (M = 4.57, SD = 2.73), Male (M = 3.78, SD = 2.32) (F(1, 140) = 2.73, n.s.).
negative emotion words within given texts,42 and has been used effectively to measure emotional expression in both written and spoken text.43,44 LIWC analyzes given texts by calculating the usage rates of words from 80 different linguistic categories, including linguistic processes, psychological processes, social processes, and many miscellaneous others. As we were primarily concerned with words that indicate emotional states, we focused on the categories Positive Emotion (example words for which include “love,” “nice,” and “sweet”), Negative Emotion (“hurt,” “ugly,” “nasty”), and Anger (“hate,” “kill,” “annoyed”). 45
We obtained song lyrics from LyricWikia, as this database reliably provided lyrics that were absent of typos
42
James W. Pennebaker and Yla R. Tausczik, “The Psychological Meaning of Words: LIWC and Computerized Text Analysis Methods.” Journal of Language and Social Psychology 29, no. 1 (2010): 26–28. 43 Erin O’Carroll Bantum and Jason E. Owen, “Evaluating the Validity of Computerized Content Analysis Programs for Identification of Emotional Expression in Cancer Narratives,” Psychological Assessment 21, no. 1 (2009): 86–87. 44 Jeffrey H. Kahn et al., “Measuring Emotional Expression with the Linguistic Inquiry and Word Count,” The American Journal of Psychology 120, no. 2 (2007): 263–86. 45 “LyricsWikia,” accessed August 30, 2015, www.lyrics.wikia.com.
Figure 3. Color Profiles by Performer Gender
Tempo Figure 4 graphs the relationship between song Tempo and video color content. A breakdown of values is shown in Figure 5. A one-way ANOVA showed that the levels of Warm Colors did not differ significantly by tempo (F(3,156) = 0.91, n.s.).
46 R Core Team, R: A Language and Environment for Statistical Computing. 2013, www.r-project.org.
6
Figure 4. Color Profiles by Tempo
Tempo Category
Mean
SD
Fast (131-180 BPM) Moderately Fast (101-‐130 BPM) Moderately Slow (81-‐100 BPM)
4.29 4.29 3.58
2.00 2.46 2.74
Slow (55-‐80 BPM)
3.70
2.33
Figure 5: Mean (SD) ratings for warm colors by tempo.
Genre
Mean
SD
Country Hip-‐Hop
4.40 2.98
2.64 2.41
Pop Rock
4.53 4.00
2.03 2.36
Figure 8: Mean (SD) ratings for warm colors by genre.
Mode Figure 6 graphs the relationship between song mode and video color content. A one-way ANOVA showed that the levels of Warm Colors did not differ significantly by Mode: Major (M = 4.22, SD = 2.29), Minor (M = 3.71, SD = 2.56) (F(1, 150) = 1.68, n.s.).
Figure 6. Color Profiles by Mode
Figure 7. Color Profiles by Genre
Figure 9 plots the average number of warm colors per video by Genre (as we classified each color as either warm or cool, the plot showing the number of cool colors per video per Genre is the inverse of this plot).
Genre Figure 7 graphs the relationship between song Genre and video color content. A breakdown of values is shown in Figure 8. A one-way ANOVA showed a significant effect of Genre upon the levels of Warm Colors (F(3, 156) = 3.52, p = 0.02). Post-hoc Tukey HSD multiple comparisons showed that Hip-hop and Country differed at p = 0.04 and Hip-hop differed from Pop at p = 0.02. All other pairwise contrasts were non-significant.
Figure 9. Warm Colors by Genre Figure 10 plots the relationship between song Genre and average Video Brightness.
7
Figure 10 Video Brightness (Value) by Genre
Figure 11 Lyric Positive Emotion by Genre
Lyrics LIWC analysis results are shown in Figures 11-13. These plots illustrate the relative occurrence of Lyric Positive Emotion, Negative Emotion, and Anger in our four musical genres. A one-way ANOVA showed no effect of Genre upon the levels of Positive Emotion (F(3, 156) = 0.70, p = ns), but showed a significant effect of Genre upon the levels of Lyric Negative Emotion (F(3, 156) = 3.93, p = 0.01). Post-hoc Tukey HSD multiple comparisons showed that Hip-hop and Country differed at p = 0.006. All other pairwise contrasts were nonsignificant. A one-way ANOVA also showed a significant effect of Genre upon the levels of Lyric Anger (F(3, 156) = 10.57, p = 0.00001). Post-hoc Tukey HSD multiple comparisons showed that Hip-hop showed greater levels of Anger than Country (p = 0.00002), and Pop (p = .00001), and Rock (p = .002). All other pairwise contrasts were non-significant.
Figure 12. Lyric Negative Emotion by Genre
8
Figure 15. Video Brightness by Positive Emotion (H = Hip-hop, C = Country, P = Pop, R = Rock)
Figure 13. Lyric Anger by Genre
Figure 14 shows a side-by-side comparison of video color analyses and LIWC analyses.
Figure 14. Side-by-side comparison of video and LIWC analyses Figure 15 graphs the relationship between Video Brightness and Lyric Positive Emotion on the song level using point labels that illustrate song genre. The slope of this relationship, b = 0.113, is non-significant (t = 0.32, ns) with an Adjusted R2 of 0.000.
The only relationship between emotion and color that approached significance was the relationship between use of Video Warm Colors and amount of Lyric Anger. Figure 16 plots the relevant data using point labels that illustrate song Genre. Anger was negatively skewed (there were many songs without any expression of Anger, and only a few with high levels). Therefore we plotted Warm Colors against log10 Anger. The black horizontal and vertical lines are drawn at the median levels. This graph shows that the lower right quadrant of high Lyric Anger and low use of Video Warm Colors is most densely populated by Hiphop songs, especially at the extreme right bottom. Indeed a linear regression of Warmth against log10 Anger does produce a slope of significantly negative slope, b = -0.53, t = -2.55, p = .01, although overall this explains little of the variance in the data (Adjusted R2 = 0.03).
9 continue to refine our knowledge of these relationships, and help lead us to a better understanding of the fascinating art form that is music video. Limitations and Further Research While the Color Barcode Generator and Image Color Extract software were useful ways to quantify video color content, their use necessarily constituted a brute-force approach. Due to the element of compression involved, a color barcode can never represent the entire color content of a video, and ten representative colors likewise cannot represent the entire color content of a barcode. Further research could take a more sophisticated approach to
Figure 16. Lyric Anger by Warm Colors (H = Hiphop, C = Country, P = Pop, R = Rock)
Discussion Our results indicate that of the variables we considered, Genre played the largest role in determining video color content. ANOVA results showed that Genre significantly impacted the levels of video Warm Colors, whereas Mode, Tempo, and Performer Gender did not. LIWC analyses also demonstrate that Genre significantly impacted the presence of Lyrical Negative Emotion and Anger. Figure 14 suggests that videos from genres that exhibit more Lyric Positive Emotion tended to utilize warmer colors, and to be brighter, than videos from genres that exhibit more Negative emotion. However, when we look on a song-by-song level, a large amount of variability is apparent. Figure 15 demonstrates that there was no significant relationship between Lyric Positive Emotion and Video Brightness. Nevertheless, genre differences are evident on a song-by-song basis in the relationship between Anger and Warm Colors. As Figure 16 illustrates, there is an inverse relationship between Lyric Anger and Warm Colors, although this is against a background of considerable individual variability. Although genre plays a significant role in our results, we cannot dismiss other factors as insignificant. It is apparent in the literature that the functions of color in music video are both diverse and subtle. The fact that we have not established obvious relationships between some of our variables does not necessarily indicate anything other than that our current analytical approach is limited. From this preliminary analysis, however, it is apparent that music/color relationships in music video do exist on some levels. We hope that future research will
ensure that less color detail is lost in translation. 47 Additionally, a high level of human involvement in our analysis process (manual collection of lyrics, manual coding of tempo, mode, performer gender, etc.) forced us to use a relatively small sample size. Automation of most or all of the process would allow for the analysis of a much larger sample. As previously discussed, it is apparent in the literature that color in music video frequently reflects subtle aspects of timbre, harmony, and form, and that the changes in color during a video may be more important than the overall palette utilized. Further research could, for instance, determine whether shifts between minor and major modes within music videos are accompanied by consistent changes in color. Recent advances in the spectral analysis of music could also be employed to automate the analysis of musical characteristics such as harmony and timbre.48 Finally, modification of the Color Barcode Generator software such that it could attach time markers to images or convert individual sections of videos into images would facilitate the comparison of musical characteristics and color palettes as they change during videos. As music video is a dynamic art form, it will always be best studied using methods of analysis that take into account change and evolution. Acknowledgements We would like to thank the Grand Valley State University Summer Scholars Program for funding this research project, and Nick Ellis for his advice and statistical expertise.
47 See: Vijay Pandurangan, “Colours in Movie Posters since 1914,” Blog, Vijay Pandurangan, (June 11, 2012), http://www.vijayp.ca/blog/2012/06/colours-in-movie-posterssince-1914/ and Benedict Koehler, “Color Analysis of Flickr Images,” Blog, Beautiful Data, (May 1, 2013), http://beautifuldata.net/2013/05/color-analysis-of-flickr-images/. 48 See: Matthias Mauch et al., “The Evolution of Popular Music: USA 1960–201,” Royal Society Open Science, May 6, 2015, doi:10.1098/rsos.150081.
10 Bibliography Bailey, Reynold J., Cindy M. Grimm, and Chris Davoli. “The Real Effect of Warm-Cool Colors,” All Computer Science and Engineering Research Report Number: 2006-17 (2006) http://openscholarship.wustl.edu/cse_research/166. Bantum, Erin O’Carroll, and Jason E. Owen. “Evaluating the Validity of Computerized Content Analysis Programs for Identification of Emotional Expression in Cancer Narratives.” Psychological Assessment 21, no. 1 (2009). Belton, John. “Painting by the Numbers: The Digital Intermediate.” Film Quarterly 61, no. 3 (March 2008): 58–65. doi:10.1525/fq.2008.61.3.58. Bernard, Jonathan W. “Messiaen’s Synesthesia: The Correspondence between Color and Sound Structure in His Music.” Music Perception: An Interdisciplinary Journal 4, no. 1 (1986). doi:10.2307/40285351. Collopy, Fred. “Color, Form, and Motion: Dimensions of a Musical Art of Light.” Leonardo 33, no. 5 (2000): 355–60. “ColorHexa Color Encyclopedia.” Accessed August 28, 2015. colorhexa.com. Elicker, Martina. “Popular Music in the Age of the Video Clip.” AAA: Arbeiten Aus Anglistik Und Amerikanistik 23, no. 1 (1998): 69–89. doi:10.2307/43025556. Gelotte, Kepler. Image Color Extract (version 1.1), n.d. http://www.coolphptools.com/color_extract. Goodwin, Andrew. Dancing in the Distraction Factory: Music Television and Popular Culture. Minneapolis: University of Minnesota Press, 1992. Griscom, William, and Stephen Palmer. “Emotional Mediation of Cross-Modal Associations in Timbre-Color Synesthesia.” Journal of Vision 14, no. 10 (2014): 1003. doi:10.1167/14.10.1003. Griscom, William, and Stephen E. Palmer. “The Color of Musical Sounds: Color Associates of Harmony and Timbre in NonSynesthetes.” Presentation at the Vision Sciences Society Annual Meeting, Naples, Florida, 2012. Harrison, V. G. W. “Colour and Music.” Journal of the Royal Society of Arts 90 (August 7, 1942): 609–12. Ione, A. “Kandinsky and Klee: Chromatic Chords, Polyphonic Painting and Synesthesia.” Journal of Consciousness Studies 11, no. 3–4 (2004). Kahn, Jeffrey H., Renée M. Tobin, Audra E. Massey, and Jennifer A. Anderson. “Measuring Emotional Expression with the Linguistic Inquiry and Word Count.” The American Journal of Psychology 120, no. 2 (2007): 263–86. Koehler, Benedict. “Color Analysis of Flickr Images.” Blog. Beautiful Data, May 1, 2013. http://beautifuldata.net/2013/05/coloranalysis-of-flickr-images/. Krzywinski, Martin. Image Color Summarizer (version 0.52), n.d. http://mkweb.bcgsc.ca/color_summarizer/. Laily, Melvin. Color Barcode Generator (version 0.5), 2013. http://arcanesanctum.net/movie-barcode-generator/. Lill, Karel, and Lee Copenhaver. “Comparative Theories of Visual Art and Music: May I Play You a Picture?” Mcnair Scholars Journal 16 (2012): 35–39. “LyricsWikia.” Accessed August 30, 2015. www.lyrics.wikia.com. Mauch, Matthias, Robert M. MacCallum, Mark Levy, and Armand M. Leroi. “The Evolution of Popular Music: USA 1960–2010.” Royal Society Open Science, May 6, 2015. doi:10.1098/rsos.150081. Ou, Li-Chen, M. Ronnier Luo, Andrée Woodcock, and Angela Wright. “A Study of Colour Emotion and Colour Preference. Part I: Colour Emotions for Single Colours.” Color Research & Application 29, no. 3 (2004). doi:10.1002/col.20010. Palmer, Stephen E., Karen B. Schloss, Zoe Xu, and Lilia R. Prado-León. “Music–color Associations Are Mediated by Emotion.” Proceedings of the National Academy of Sciences 110, no. 22 (May 28, 2013): 8836–41. doi:10.1073/pnas.1212562110. Palmer, Stephen, Thomas Langlois, Tawny Tsang, Karen B. Schloss, and Daniel J. Levitin. “Color, Music, and Emotion.” Presentation at the Vision Sciences Society Annual Meeting, Naples, Florida, 2012. Pandurangan, Vijay. “Colours in Movie Posters since 1914.” Blog. Vijay Pandurangan, June 11, 2012. http://www.vijayp.ca/blog/2012/06/colours-in-movie-posters-since-1914/. Pennebaker, James W., and Yla R. Tausczik. “The Psychological Meaning of Words: LIWC and Computerized Text Analysis Methods.” Journal of Language and Social Psychology 29, no. 1 (2010). R Core Team. R: A Language and Environment for Statistical Computing., 2013. www.r-project.org. Schloss, Karen B., Patrick Lawler, and Stephen E. Palmer. “The Color of Music.” Presentation at the Vision Sciences Society Annual Meeting, Naples, Florida, 2008 Schwettman, S, S Churchill, and D Eagleman. “A Large-Scale Analysis of Music-Color Synesthesia.” presented at the National Council for Undergraduate Research, April 13, 2013. Taylor, Pamela G. “Press Pause: Critically Contextualizing Music Video in Visual Culture and Art Education.” Studies in Art Education 48, no. 3 (Spring 2007): 230–46. Tsang, Tawny, and Karen B. Schloss. “Associations between Color and Music Are Mediated by Emotion and Influenced by Tempo.” The Yale Review of Undergraduate Research in Psychology, 2 (2008). Vernallis, Carol. Experiencing Music Video: Aesthetics and Cultural Context. New York: Columbia University Press, 2004. ———. “Music Video, Songs, Sound: Experience, Technique and Emotion in Eternal Sunshine of the Spotless Mind.” Screen 49, no. 3 (2008): 277–97. doi:10.1093/screen/hjn036. ———. Unruly Media: YouTube, Music Video, and the New Digital Cinema. New York: Oxford University Press, 2013. Xiao, Youping, Christopher Kavanau, Lauren Bertin, and Ehud Kaplan. “The Biological Basis of a Universal Constraint on Color Naming: Cone Contrasts and the Two-Way Categorization of Colors.” PLoS One 6, no. 9 (2011).
11 Appendix A: List of Music Videos Pop
Hip-hop
Rock
Country
Adele - Rolling in the Deep Ariana Grande - Problem Avicii - Wake Me Up B.o.B. - Nothin' On You Beyonce - 7/11 Beyonce - Single Ladies Beyonce feat. Lady Gaga - Video Phone Britney Spears - Piece Of Me Britney Spears - Till the World Ends Britney Spears - Womanizer Bruno Mars - Grenade Bruno Mars - Locked Out of Heaven Cobra Starship - Good Girls Go Bad Danity Kane - Damaged Ed Sheeran - Thinking Out Loud Fun - Carry On Fun - We Are Young Iggy Azalea - Fancy Jason Derulo - Talk Dirty Jonas Brothers - Burnin' Up Justin Bieber - Boyfriend Justin Timberlake - Mirrors Katy Perry - California Gurls Katy Perry - Last Friday Night Kesha - Tik Tok Lady Gaga - Bad Romance Lady Gaga - Poker Face Mark Ronson feat. Bruno Mars Uptown Funk Maroon 5 - Payphone Maroon 5 - Sugar Miley Cyrus - We Can't Stop One Direction - What Makes You Beautiful Panic! At The Disco - Nine In The Afternoon Pharrell Williams - Happy Pitbull - Give Me Everything Rihanna feat. Calvin Harris - We Found Love Selena Gomez - Come And Get It Taylor Swift - Blank Space Tokio Hotel - Ready, Set, Go! Wisin & Yandel - Abusadora
A$AP Rocky -‐ Fuckin' Problems Asher Roth -‐ I Love College B.o.B. -‐ Airplanes Big Sean -‐ I Don't Fuck With You Childish Gambino -‐ 3005 Childish Gambino -‐ Heartbeat Chris Brown -‐ Look At M e Now Drake -‐ Forever Drake -‐ Hold On, We're Going Home Drake -‐ HYFR Drake -‐ Started from the Bottom Eminem -‐ Berzerk Eminem -‐ Not Afraid Eminem -‐ We Made You Fetty Wap -‐ Trap Queen Flo Rida -‐ Low Flo Rida -‐ Right Round J. Cole f eat Miguel -‐ Power Trip Jay-‐Z -‐ D.O.A. Jay-‐Z -‐ On to the Next One Jay-‐Z and Kanye West -‐ Paris Kanye West -‐ All of the Lights Kanye West -‐ Black Skinhead Kanye West -‐ Homecoming Kanye West -‐ Love Lockdown Kanye West -‐ Mercy Kendrick Lamar -‐ Alright Kendrick Lamar -‐ Swimming Pools (Drank) Kid Cudi -‐ Pursuit of Happiness Lil Wayne -‐ 6 Foot 7 Foot Lil Wayne -‐ Lollipop Lupe Fiasco -‐ Superstar Lupe Fiasco -‐ The Show Goes On Macklemore and Ryan Lewis -‐ Can't Hold Us Mary J. Blige -‐ Just Fine Nicki Minaj -‐ Anaconda Nicki Minaj -‐ Beez in the Trap Nicki Minaj -‐ Super Bass Wiz Khalifa -‐ See You Again Wiz Khalifa -‐ We Dem Boyz
Arctic Monkeys -‐ Do I Wanna Know? Arctic Monkeys -‐ Why'd You Only Call Me When You're High? Cage the E lephant -‐ Shake Me Down Coldplay -‐ Paradise Coldplay -‐ Viva La Vida Fall Out Boy -‐ Beat It Fall Out Boy -‐ I Don't Care Fall Out Boy -‐ My Songs Know What You D id In The Dark Fall Out Boy -‐ Uma Thurman Florence and the Machine -‐ Dog Days are Over Florence and the Machine -‐ Ship to Wreck Foo Fighters -‐ The Pretender Foo Fighters -‐ Walk Foster the People -‐ Pumped Up Kicks Green Day -‐ 21 Guns Hozier -‐ Take Me to Church Imagine Dragons -‐ Demons Imagine Dragons -‐ It's Time Imagine Dragons -‐ Radioactive Jack White -‐ Sixteen Saltines Kings of Leon -‐ Use Somebody Linkin Park -‐ Burn It Down Linkin Park -‐ Shadow of the Day Linkin Park -‐ Until It's Gone Lorde -‐ Royals MGMT -‐ Flash Delirium Mumford and Sons -‐ I Will Wait Mumford and Sons -‐ The Cave Muse -‐ Uprising Paramore -‐ Crush Crush Crush Paramore -‐ Decode Paramore -‐ Ignorance Slipknot -‐ Psychosocial The Black Keys -‐ Fever The Black Keys -‐ Howlin' for You The Black Keys -‐ Lonely Boy Thirty Seconds to Mars -‐ Kings and Queens Thirty Seconds to Mars -‐ Up in the Air Vampire Weekend -‐ Diane Young Walk the Moon -‐ Shut Up and Dance
Brad Paisley - Online Brad Paisley - Welcome to the Future Carrie Underwood - Blown Away Carrie Underwood - Good Girl Carrie Underwood - See You Again Carrie Underwood - Something In The Water Darius Rucker - Homegrown Honey Eric Church - Springsteen Florida Georgia Line - Cruise Hunter Hayes - I Want Crazy Jason Aldean - 1994 Jason Aldean - Burnin' It Down Jason Aldean - The Truth Kenny Chesney - American Kids Kenny Chesney - Don't Blink Kenny Chesney - Everybody Wants To Go To Heaven Kenny Chesney - The Boys of Fall Kenny Chesney - You And Tequila Lady Antebellum - Bartender Lady Antebellum - Lookin' For A Good Time Lady Antebellum - We Owned the Night Luke Bryan - That's My Kind of Night Miranda Lambert - Automatic Miranda Lambert - The House that Built Me Miranda Lambert - White Liar Rascal Flatts - Every Day Sugarland - All I Want To Do Sugarland - Stay Sugarland - Stuck Like Glue Taylor Swift - Love Story Taylor Swift - Mine Taylor Swift - Our Song Taylor Swift - Safe And Sound The Band Perry - If I Die Young Thomas Rhett - It Goes Like This Tim Mcgraw - One of Those Nights Toby Keith - American Ride Toby Keith - Red Solo Cup Trace Adkins - You're Gonna Miss This Zac Brown Band - Toes