Abstract
Aesthetic experiences arise through interaction with sensory objects. These experiences are shaped into aesthetic judgments using verbal concepts. We explored this process in music by surveying 804 participants who described their favorite music using adjectives, resulting in 94 semantic fields based on 7,388 responses and 1,786 unique terms. Using the Cognitive Salience Index (CSI), we assessed the prominence of these terms, considering frequency, order, and sample size. The results reflect individual preferences rather than objective criteria, emphasizing personal perspectives on aesthetic appreciation. Prominent terms were mainly related to mood and emotions, highlighting the role of emotional engagement in musical experiences. Descriptions also included sound characteristics and style-specific terminology, illustrating how stylistic features influence aesthetic judgments. In contrast, aesthetic terms—such as “beautiful”—were less salient, likely due to the personal framing of the task. This study helps explain how people derive aesthetic value from music by linking personal experiences with structured verbal descriptions.
Aesthetic experiences are evoked through the engagement with aesthetic stimuli and often contain an explicit judgment of the perceived quality and value of the stimulus. Recipients listen to music, read literature, and observe visual artwork, and form these experiences into aesthetic judgments, using verbal concepts that seem most suitable to describe their impression of the relevant, meaningful and rewarding characteristics of an object (Knoop et al., 2016; Leder et al., 2004). One prominent method to understand aesthetic judgements in different domains is to ask recipients to verbally describe what makes an object aesthetically appealing. This approach “from below” (Fechner, 1876) takes into account the perspective of the recipients instead of being guided by the opinion of experts or experimenter-based selections of aesthetic descriptors. Several studies identified recipient-based verbal concepts of aesthetics for different object categories including visual art, landscapes, faces, and different design classes (Augustin et al., 2012b; Jacobsen et al., 2004; see also Hager et al., 2012), various materials including leather, metal, and wood (Marschallek et al., 2023), tattoos (Weiler & Jacobsen, 2021), literary genres (Knoop et al., 2016), and music (Istók et al., 2009). Often, the art forms were treated without taking into account the different styles and forms within each class (but see Knoop et al., 2016; Marschallek et al., 2023), or inter-individual preferences. Istók et al. (2009) conducted a study aimed at examining the cognitive, knowledge-based concepts underlying aesthetic responses to music. They requested their participants to provide written expressions of their aesthetic evaluations of music using adjectives. Notably, the adjective “beautiful” emerged as the most frequently mentioned word in the study results. Moreover, adjectives that described emotional or mood states were commonly mentioned, including “touching,” “emotional,” “involving feelings,” “pleasant,” “sad,” and “happy.” The analysis also revealed adjectives that were specifically related to musical elements such as “rhythmic,” “melodic,” and “harmonic.” The findings of this study provide valuable insights into the underlying cognitive and knowledge-based concepts that influence aesthetic responses to music (Istók et al., 2009).
Notably, studies on aesthetic concepts “from below” have typically focused on general descriptions of aesthetic objects. In contrast, our study takes a more differentiated approach by examining personal concepts of music through the lens of individual taste, specifically focusing on music that participants have already evaluated positively—namely, their favorite music. This allows us to access aesthetic experiences that are affectively significant, and personally meaningful. Our approach is grounded in a broad understanding of aesthetics – one that goes beyond formal features, such as beauty, or harmony to encompass emotional, cognitive, functional, and personal responses (e.g., Leder et al., 2004).
When asking for a general aesthetic judgment of an object rather than a personally significant one, the term “beautiful” emerged as the most frequently mentioned (Augustin et al., 2012b; Istók et al., 2009; Jacobsen et al., 2004; Knoop et al., 2016; Marschallek et al., 2023; Weiler & Jacobsen, 2021), followed by “ugly” (Augustin et al., 2012b; Jacobsen et al., 2004), and “suspenseful” (Knoop et al., 2016). Further, a substantial number of emotion-related terms such as “touching,” “emotional,” “involving feelings,” “pleasant,” “sad,” and “happy” were identified in studies on music (Augustin et al., 2012a; Istók et al., 2009) and literature (Knoop et al., 2016). In addition, the studies revealed domain-specific descriptive structural features, e.g., music included features such as “rhythmic,” “melodic,” and “harmonic” (Istók et al., 2009), whereas features of patterns were “round” and “symmetric,” visual arts were “colorful,” buildings were “big,” and cars were “fast” (see Tables 1 and 2 in Augustin et al., 2012b). Other terms point to functional reasons of an engagement with art forms, e.g., “soothing” indicates mood regulation, and “inspiring” intellectual stimulation (Augustin et al., 2012a). The latter was named for film, music, and visual arts with a similar likelihood, the first was most dominant for music (Augustin et al., 2012a). Materials were primarily described in a descriptive way and tended to have neutral valence. “Smooth” was the most significant term for describing materials, followed by other key descriptors such as “hard,” “rough,” “soft,” and “glossy.” Additionally, sensorial qualities such as “warm” and “see-through” were central to the aesthetics of materials, with most descriptors relating to haptic qualities such as “cold” and “heavy” (Marschallek et al., 2023).
Models of aesthetic processing include these aspects as well, but differ in their sophistication and specified relations between those aspects (e.g., Bergeron & Lopes, 2012; Graf & Landwehr, 2015; Leder et al., 2004; Pelowski et al., 2017; Silvia & Brown, 2007). For example, Leder et al. (2004) describe aesthetic appreciation as the result of dynamic interactions between multiple modules, including perceptual analysis, implicit memory integration, explicit classification, cognitive mastering, and evaluation, which can lead to aesthetic judgments or affective responses. Silvia and Brown's (2007) appraisal model emphasizes that emotional responses to art are shaped by individual appraisals—cognitive evaluations influenced by traits, values, and expertise—thereby accounting for pronounced inter-individual differences. Similarly, but with a focus on fluency-based processes, Graf and Landwehr (2015) describe aesthetic experiences by stimulus-based properties (e.g., processing affordances) and perceivers’ needs (e.g., cognitive enrichment). Bergeron and Lopes (2012) identify three core dimensions – evaluative, affective, and semantic – but argue that not all need to be involved in every aesthetic experience. Pelowski et al. (2017) extend these perspectives by proposing a multilayered model that encompasses perceptual processing, emotional responses, meaning-making, evaluation, and physiological states. Despite their differences, these models converge in describing aesthetic experiences as complex interactions between stimulus features, emotional reactions, cognitive processes, and individual dispositions. In line with this broader understanding, our use of the term aesthetics moves beyond narrow, formalist conceptions to embrace the full spectrum of ways individuals experience, and emotionally engage with music in everyday life. This is consistent with Herbert's argument that musical affect spans from subtle mood shifts to deeply transformative experiences. Similarly, Behne's concept of Musikerleben underscores that an aesthetic experience includes not only emotional and cognitive dimensions, but also bodily and social components (Behne, 2004). Concepts such as the Geneva Emotional Music Scale (GEMS; Zentner et al., 2008), the Geneva Music-Induced Affect Checklist (GEMIAC; Coutinho & Scherer, 2017), and the Aesthetic Emotions Scale (AESTHEMOS; Schindler et al., 2017) further highlight the nuanced emotional vocabulary used to describe aesthetic experiences, showing how emotions such as awe, nostalgia, or being moved are not merely outcomes, but integral to the aesthetic judgment itself. These perspectives inform our approach by framing the aesthetic experience as a multifaceted process, essential to understanding why music from different genres becomes favorites.
In the context of our study it is important to address the potential concern that aesthetic experience might be conflated with general or functional uses of music. It is argued here that such conflation is avoided through the specific framing of the present study, which asked participants to describe their favorite music. The act of selecting music as a favorite implies that an aesthetic judgment has already been made – one informed by emotional resonance, perceived value, and meaningful personal experience. Even in cases where favorite music is used functionally (e.g., to enhance motivation during exercise), such uses often rely on aesthetic and affective qualities that were previously recognized and appreciated. In this way, functional uses can be seen not as distinct from aesthetic experience, but as shaped by and dependent on it (Behne, 2004; Leder et al., 2004; Pelowski et al., 2017). Accordingly, the free descriptions provided by participants are interpreted as expressions of aesthetic judgment from the recipients’ perspective – reflecting not only formal or stylistic appraisals, but also emotionally and personally meaningful engagements with music.
In research on musical taste, aesthetic concepts have not been in the focus of studies. But studies are related, when investigating the underlying reasons for listening to specific music, such as musical attributes, personality issues, or functions of music listening (Finnäs, 1989; Flannery & Woolhouse, 2021; Greb et al., 2017; North, 2010; Parzer, 2011; Rentfrow & Gosling, 2003, 2006; Rentfrow et al., 2011; Schäfer & Sedlmeier, 2009). These functional aspects of music listening – such as mood regulation, social bonding, or intellectual stimulation – can be understood as integral components of the aesthetic judgment, rather than separate or purely utilitarian motivations. When individuals evaluate music as their favorite, they often do so because it fulfills personal functions that are emotionally, and cognitively rewarding, thereby becoming part of the broader aesthetic experience. The most important functions are mood regulation, arousal, and intellectual stimulation, which are partially matching preference factors. While favorite music overall puts the listener in a good mood, electronic music energizes and motivates dancing, and classical music is intellectually stimulating (Schäfer & Sedlmeier, 2009). These styles also differ with regard to musical attributes. For electronic music the emphasis is on rhythm, while classical music is complex in structure (e.g., Rentfrow & Gosling, 2003, based on preference factors, that pertain specific musical styles). However, relations between preferences, musical features, and music-specific functions are usually shown by correlations (Greb et al., 2017; e.g., North, 2010; Rentfrow & Gosling, 2003; Rentfrow et al., 2011; Schäfer & Sedlmeier, 2009). Interpretation of genre labels are often based on stereotypical descriptions of particular styles of music (e.g., rock as intense and rebellious, Rentfrow & Gosling, 2003), but there are also approaches to related preferences of musical styles with evaluated, style-dependent, characteristic acoustic and musical features (Finnäs, 1989; Flannery & Woolhouse, 2021; Rentfrow & Gosling, 2006; Rentfrow et al., 2011). While there is basic agreement, some interpretations differ when looking into details. For instance, some studies portray electronic music as energetic and rhythmic (Rentfrow & Gosling, 2003), while others describe it as emotionally engaging and useful for mind wandering (Greb et al., 2017; Schäfer & Sedlmeier, 2009; see also Parzer, 2011). In the mentioned studies, musical styles were assigned specific functions based on their characteristics. However, inconsistencies arise from the use of stereotypical labels for these styles, as they lack differentiation. We can achieve this by closely examining the aesthetic concepts listeners attribute to their favorite music.
In the current study on individual aesthetic concepts of music, we collected detailed information on musical preferences on the one hand, and asked participants to generate adjectives describing their favorite music. By bridging preferences and the exploration of aesthetic descriptions, our study contributes to the comprehension of why individuals engage with and derive aesthetic pleasure from their favorite music. This nuanced approach is integral to unraveling the intricate relationships between aesthetic preferences, emotional experiences, and the multifunctionality of music in listeners’ lives. It is decidedly an approach “from below,” to put weight on the individual perspectives of the recipients.
Methodologically, an important difference is that we included all responses in our analyses, whereas earlier studies decided to analyze the most frequent terms only (Augustin et al., 2012b; Istók et al., 2009; Jacobsen et al., 2004; Knoop et al., 2016; Marschallek et al., 2023). Four studies examined terms mentioned by at least 5% of participants (Istók et al., 2009; Jacobsen et al., 2004; Knoop et al., 2016; Marschallek et al., 2023), while the other study analyzed terms mentioned by a minimum of two participants (Augustin et al., 2012b). This procedure results in a dramatic reduction of information. For example, Jacobsen et al. (2004) collected 2,948 responses from 311 participants, resulting in a pool of 590 different terms, of which 38 items were analyzed. Istók et al. (2009) collected 3,565 from 290 participants, resulting in a pool of 1,084 different terms, of which 43 items were analyzed. The reduction of analyzed terms was a methodologically sound decision. Infrequent terms were not analyzed further because of their potential unimportance. In contrast, we assembled all responses into semantic fields to capture the full information in our data set. For example, singular expressions that would not be analyzed by the 5% cutoff, might belong to the same semantic field. With our method, those expressions are fully captured and analyzed in terms of their salience in the data set. However, we included steps of semantic interpretation in the preprocessing of the data, which is time consuming and naturally adds subjectivity. For the measure of musical taste, we asked for liking ratings on 15 musical styles, as well as on a broad selection of substyles that were further defined by representative musicians and artists.
Methods
Participants
Participants were recruited via Facebook postings, websites, newsletters, and Email lists of university students. A total of N = 804 participants took part in the study, consisting of n = 429 females, and n = 356 males (n = 19 no answer), with a mean age of 27 years (SD = 9), spanning an age range of 15 to 72 years. 1 The majority of participants held either a German Abitur (A-levels diploma, n = 354), or a university degree (n = 343) as the highest degree, and a large group identified as students (n = 569). Among the participants, 291 individuals had over eight years of instrumental or vocal training, while 162 had not received any musical instruction. Furthermore, 172 participants self-identified as musicians, and 49 individuals reported studying music or being professionally involved with music. The latter group had actively played and practiced a musical instrument, including singing, for an average of 4.6 years. In terms of music consumption habits, the participants reported spending an average of 1.1 h per day listening to music with focus and concentration, and 2.1 h per day listening casually (for detailed information on the sample, see Supplemental Material Tables S1A-F). Based on browser information, 69% of the participants were logged in in Germany, 2% in The Netherlands, and others in Switzerland, Austria, and spread across the globe. In Germany, most participants came from Berlin (6.1%), followed by Frankfurt a.M. (4%), Mainz (3.5%), Munich (2.7%), and Hamburg (1.9%).
Procedure
All procedures were approved by the Ethics Council of the Max Planck Institute for Empirical Aesthetics. The online data collection was conducted via the software QuestBack Unipark (https://www.unipark.com). The language of the survey was German. The survey was conducted in two phases: the first phase took place from May 10 to July 11, 2016, and the second phase was extended until May 3, 2017. Before starting the survey, participants were informed about the scientific background of the study and the anonymized data collection. In the first part, participants rated 15 different musical styles on a 7-point Likert scale, ranging from “don't like at all” to “like very much.” Additionally, they were given the option to choose “I don't know”, if they were unfamiliar with any particular style. In the second part, participants evaluated likewise a total of 100 substyles, 3 to 11 for each style. For each substyle, three artists or bands were given as examples in order to define the substyle as clearly as possible. To define the styles included in the questionnaire, a list was compiled based on style labels used by 20 record shops, streaming services, music magazines, and websites (e.g., Saturn, last.fm). Styles that appeared on at least half of the lists were selected: blues, classical, country, electronica, folk, hip-hop/rap, jazz, metal, pop, reggae, rock, and soul. Techno, house, and German traditional music were added due to their growing popularity, and “alternative” was excluded as standalone genre label, resulting in 15 final styles. Note that we included “alternative” in substyles for music from more independant labels, or for more experimental developments, as suggested by the experts, e.g., folk included the substyle “alternative folk (folk punk, anti-folk, indie folk).” The experts were music producers, musicians, and musicologists with expertise in popular music, who were asked to assemble hierarchically nested substyles to represent the breadth of each style with minimal redundancy. Up to three representative musicians were identified for each substyle. Feedback was gathered via questionnaires and focus groups. Additional focus groups were used to compare and refine the selection of substyles and musicians. For example, soul included classic RnB (e.g., Aretha Franklin), soul-funk (e.g., James Brown), 60s soul (e.g., The Temptations), 70s soul (e.g., Curtis Mayfield), pop RnB (e.g., Alicia Keys), and neo-soul (e.g., D’Angelo; see Supplemental Material Table S2 for a complete list of styles, substyles, and artists). In the third part, participants were requested to “please list up to 20 adjectives that best describe the qualities and characteristics of your favorite music.”
We further assessed evaluations (7-point scales) on 24 musical functions (22 from Greb et al., 2017 and two items indicating no specific function), and a 15-item big-five personality inventory (Dehne & Schupp, 2007). Both scales are not in the focus of the current study. The last part consisted of questions on musical expertise and behavior (8 items), and demographic information (10 items; see Participant section).
Preprocessing of the Free Responses
From the 804 participants, two did not give sensible responses (i.e., “oh, no idea”, “asdf”), resulting in responses from 802 participants. Those generated a total of 7,388 verbal descriptions regarding their most favorite music, resulting in 1,786 distinct responses after spelling corrections. Our goal was to include as many responses as possible in our analyses. To do so, we harmonized the responses in a multi-step procedure. A first simple transformation was to turn nouns into adjectives or transitive verbs, e.g., “expression” to “expressive.” Another transformation was to reduce phrases to one word, if possible, e.g., “catchy and easy to remember melody” to “catchy,” but we kept the multi-word expressions “fitting the mood,” “enhancing concentration,” and “evoking memories.” Further, we looked into rare responses and subsumed those under terms that were mentioned often. Taking responses with the highest frequency (mentioned by at least 10 participants) as the basis, the semantically closest term was chosen. All remaining low frequent responses were assigned. That is, the final units of analyses were semantic fields. For example, “calming down”/”loosening”/”laid-back”/”reducing stress”/”minimizing tension” were mapped onto “relaxing.” We obtained the essence of each verbal expression, even the creative ones, such as in “getting goosebumps” to “moving.” These preprocessing steps were done by the first author (EG), followed by discussive evaluations and consensus-based decisions by two authors (EG, EL). To ensure accuracy and consistency, the remaining author (JM), not involved in the initial processes, reviewed the decisions, and validated the categorizations. The final set of 94 semantic fields (Table S3–S6 in the Supplemental Material), was translated into English by the first author with the help of DeepL and reviewed by a native English speaker fluent in German.
Additionally, the sematic fields were grouped into aesthetic categories. Firstly, we compiled them through consensus-based discussions within the team of authors. Secondly, to validate this categorization and reduce the subjectivity of the categorization, we decided to use artificial intelligence in a second step. ChatGPT is a powerful Large Language Model-chatbot (Meyer et al., 2024) and particularly well suited for analyzing language (Bang et al., 2023; Wu et al., 2023). ChatGPT was prompted with: “Please categorize the following words into music-aesthetic related categories” (date: 02.11.2023; OpenAI, 2023). All authors agreed that the found solution was meaningful and appropriate. Consequently, this process resulted in the following categories: Mood and Emotion (including 28 terms), Sound Characteristics (15 terms), Style and Genre (15 terms), Social and Relational (10 terms), Aesthetic and Sensory Qualities (26 terms; Table S3 in the Supplemental Material). The titles of the categories correspond very well with dimensions contained in models of aesthetic experiences (e.g., Bergeron & Lopes, 2012; Leder et al., 2004; Pelowski et al., 2017; Silvia & Brown, 2007).
Analyses
All quantitative analysis was done in R (Core Team, 2024). All scripts and data are publicly available on osf (see Data Availability). In our analyses, we related the verbal descriptions of one's favorite music to the individual's musical preferences. We defined the musical taste of the participants by their liking rating for the musical style plus the mean of the liking ratings across substyles related to the style. Participants rated their liking for a list of styles first, and then listed adjectives to describe their favorite music. Consequently, the description of favorite music might have been broader than just specific to one style. However, the aim was exactly that: to collect descriptions of favorite music and not style-specific descriptions, where participants describe the features of a specific style.
This procedure allowed us to identify the most favorite musical style for 761 participants. We excluded data from 41 participants for whom we could not identify one most favorite style but who had two or more favorites (see Table 1 for the frequencies of responses of all 15 styles). For the style-dependent analyses, we excluded the data of the styles country and German traditional music, because the participant groups were small, resulting in a final set of 6895 observations from 747 participants. A small group size will decrease the reliability of the analyses of semantic fields. We defined as cutoff the mean group size minus the standard deviation of the group sizes, which resulted in a minimum of twelve participants as representatives for the style.
Overview of Frequencies of Participants and Responses for the 15 Styles.
Note. Number of participants, with favorite musical style identified by liking ratings on style and related substyles. Number of responses for each style. The preprocessing steps resulted in 94 different response categories (semantic fields) and the occurrences of these responses are counted here.
For all styles together as well as for each style separately, we analyzed the frequency of the semantic fields and the Cognitive Salience Index (CSI). The CSI considers how often one and the same response is given by different participants, and whether it came to mind early on or later in the list of responses. It can be interpreted as a measure of importance or prototypicality of a response. More specifically, the CSI relates the frequency of a term (freq) to the frequency of all different terms within a subset (n), and the mean position of a term within the list of free responses of the participant CSI = (mpos): frequ/(n*mpos) (Sutrop, 2001).
Note that we chose to interpret only the top ten most salient terms for each style. Since ten is an arbitrary cutoff, it is evident that terms appearing further down the list increasingly overlap with other styles. Methodologically, we checked the discussed terms even beyond this cutoff, before making claims about the salience. It is also worth noting that the large variety of terms found for each style also highlights the variability and breadth of descriptions associated with musical styles, pointing to the large individual differences on a perceptual level: Not everybody enjoys the same music for the same reason. For the interested reader, we added an exploratory, non-metric multidimensional scaling approach in the Supplemental Material.
Results
We first report the analyses of the free responses on a style-independent level, that is based on all responses, irrespective of the specific favorite music of the participants. These analyses give a general overview, and invite a comparison to an earlier study on the aesthetics of music (Istók et al., 2009). Then, we report the findings on a style-dependent level, that is, important terms for the description of favorite music based on the evaluated musical styles.
Aesthetic Concepts of Music in General
The top 43 terms are listed in Table 2 (pertaining to the number of terms reported by Istók et al. (2009)), containing the mean position, frequency, and CSI. In line with previous studies, the focus of the report is on the CSI. The term “relaxing” exhibited the highest CSI (0.062), followed by “touching” (0.060), “emotional” (0.059), “melodic” (0.053), “stimulating” (0.050), “soulful” (0.047), “stylistic” (0.046), “rousing” (0.043), “diversified” (0.042), and “interesting” (0.037) (see Figure 1 “General”).
Top 43 Terms Based on the CSI.
Note. Frequency, that is absolute number of occurrences; CSI = cognitive salience index (Sutrop, 2001), taking into account the frequency and mean position in the free response output.
Top Ten CSI Values from the Style-independent Analysis (“General”) and for Each Musical Style. Note. The Colors Denote the Assigned Aesthetic Categories: White Background/black Letters = Mood and Emotion, Light Grey/black = Sound Characteristics, Dark Grey/black = Style and Genre, Dark Grey/white = Social and Relational, Black/white = Aesthetic and Sensorial Qualities (see Supplemental Material Tables S4 and S6 for a complete list of frequencies and CSI split by style).
Aesthetic Concepts of Music from Different Styles
To investigate the style-dependent level, participants were split into groups based on their most favorite style. Importantly, there was quite some overlap between groups, with the terms “touching,” “relaxing,” “emotional,” “melodic”, and “stimulating” being most prominent (see Figure 1: General). “Touching,” “relaxing,” and “stimulating” together occurred for listeners of blues, classical, electronica, folk, rap, jazz, pop, rock, soul, techno. “Emotional” and “melodic” were both notable for blues, rap, metal, pop, rock, and soul. For all styles, at least two of those five terms were mentioned in the top ten. In addition, style-specific terms occurred for one but no other style in the top ten, such as “demanding,” and “clear” for classical, “inspiring” for electronica, “calm” for folk, “lyrics” for rap, “positive” for house, “loud” for metal, “catchy” for pop, “exciting,” and “honest” for techno. Interestingly, the term “beautiful” emerged only once in the top ten, which was for jazz.
To reduce the data, the terms were grouped into the five aesthetic categories: (1) Mood and Emotion, (2) Sound Characteristics, (3) Style and Genre, (4) Social and Relational, (5) Aesthetic and Sensory Qualities. Figure 1 color-codes the five categories, and the mean CSIs across the semantic fields are visualized in Figure 2. From all five categories, Mood and Emotion played a major role across all styles when describing favorite music: most fields were coloured white in Figure 1, and it was the left-most category with the highest overall mean CSI (0.0230) in Figure 2. Interestingly, Aesthetic and Sensory Qualities played the least important role (lowest overall mean CSI = 0.0116). Sound Characteristics (mean CSI = 0.0190), Style and Genre (mean CSI = 0.0169), and Social and Relational (mean CSI = 0.0122) ranked in between. The gradient from styles within each of the five categories was for some categories steep (e.g., Sound Characteristics), for others more shallow (e.g., Style and Genre). A steep gradient indicates stronger differences in importance across styles. For example, for Sound Characteristics, the mean CSI differed strongly between metal and house (Figure 2). The high importance of Sound Characteristics for metal was based on the high salience of terms such as “melodic,” “loud,” “powerful,” and “hard” (Figure 1), whereas there was not a single term from Sound Characteristics in the top ten descriptions of house (Figure 1), which was predominantly described by Mood and Emotion items (Figure 1), such as “relaxing,” “stimulating,” “rousing,” “motivating,” “positive,” and “soulful”. As terms from Mood and Emotion were most prevalent in all styles (Figure 1), the gradient across Mood and Emotion is not very steep (Figure 2), with metal having the lowest mean CSI from all styles in Mood and Emotion (Figure 2), because Sound Characteristics were of higher importance (Figures 1 and 2).
Importance of the Five Aesthetic Categories for each Style. Note. Mean CSI (y-axis) Across the Terms Belonging to the Five Aesthetic Categories (x-axis), Split by 13 Musical Styles (Labeled, Different Filling Color). Data are Ordered from the Style with Highest to Lowest Mean CSI within Each Category.
Discussion
We investigated the verbal concepts of music aesthetics by asking 804 participants to describe their favorite music using adjectives. Through an analysis designed to retain as many participant-used terms as possible, 94 semantic fields were created, for which we calculated the CSI, serving as a measure of the salience, or, in other words, importance and prototypicality of these terms. The task of describing one's favorite music focused on individual tastes rather than objective criteria, which is reflected in the current findings.
The evaluated terms highlight the importance of individual perspectives on aesthetic appreciation and follow the multifarious processes involved in art perception as outlined in models of aesthetic processing (e.g., Pelowski et al., 2017). Indeed, the most salient terms describing mood and emotions underscore the significant role of emotional engagement with music (Bergeron & Lopes, 2012; Graf & Landwehr, 2015; Leder et al., 2004; Pelowski et al., 2017; Silvia & Brown, 2007). Descriptions of sound characteristics and terms related to style and genre indicate that the quality of the aesthetic object is evaluated based on the specific musical style, reflecting diverse aesthetic perspectives. Aesthetic and sensory qualities, particularly the term “beautiful,” showed the lowest salience in the current study. This particular result is remarkable and stands in contrast with other studies (Augustin et al., 2012a, 2012b; Istók et al., 2009; Jacobsen et al., 2004; Knoop et al., 2016), for which “beautiful” was the most important term. Hence, we succeeded with our goal to explore the individual and experience-based aesthetic judgements rather than the more general usage of terms related to “aesthetics.”
Aesthetic Concepts of Favorite Music in General
When examining all terms without taking musical style into account, within the top ten of the CSI, terms describing emotions as well as musical and stylistic aspects were found. The word forms of “relaxing” (top one), “touching,” “stimulating,” and “rousing” are participles and are used as adverbial adjectives. When used as adjectives, participles represent performed action. In our case, music is perceived to perform this action, inducing a change in the listeners’ state. As such, these word forms strongly refer to the functions of music. All these words fall within the spectrum of emotions, suggesting that participants deliberately use their favorite music for mood regulation on the valence and/or arousal level. The terms establish a clear connection to the multifaceted functions that music can fulfill (Campbell et al., 2007; Greb et al., 2017; Hargreaves & North, 1999; Hennion, 2001; Lamont & Webb, 2010; Lonsdale & North, 2011; North et al., 2000; Tarrant et al., 2000). Notably, “relaxing” and “rousing” represent seemingly opposite descriptions, pointing to the diversity of functions music can serve (Greb et al., 2017; Schäfer & Sedlmeier, 2009). The importance of the emotional impact of music is also shown by the high salience of the terms “emotional” and “soulful.” This aligns with the findings of Kuehnast et al. (2014), who demonstrated that the emotional experience of being moved or touched is often associated with the experience of art, such as film and music (Kuehnast et al., 2014).
Terms describing musical features also rank among the top ten, that is “melodic” (rank four), “rhythmic” (rank ten), a result similar to Istók et al. (2009). This suggests that melody and rhythm are key parameters in preference evaluations, a finding also supported by Amazon music reviews (von Appen, 2007).
Additionally, the CSI analysis underscores the importance of stylistic diversity with adjectives such as “stylistic” and “diversified” ranking among the top ten, supporting the idea that participants value stylistic diversity. This aligns with research suggesting substyles differentiate specific musical preferences within broad categories (Parzer, 2011). The importance of 'stylistic' in our study stands in contrast to its absence in other aesthetic studies—a difference likely due to our method of combining infrequent style-specific responses, whereas other studies typically exclude low-frequency terms. Or results highlight music's particular differentiation into substyles compared to objects, art, or literature. While functional descriptions of music are common, stylistic diversity also signifies fans’ emphasis on uniqueness and the importance of substyle differentiation within their preferred music category.
To sum, the style-independent analysis reveals that mood regulation (from “relaxing” to “rousing”), along with specific musical parameters such as melody, rhythm, and stylistic uniqueness, is highly important for describing the aesthetic appeal of favorite music. Models of aesthetic processes highlight cognitive and emotional aspects, stimulus features, and individual differences in appraisal and needs (Bergeron & Lopes, 2012; Graf & Landwehr, 2015; Leder et al., 2004; Pelowski et al., 2017; Silvia & Brown, 2007), which are confirmed by our study “from below.” Musical functions play a significant role, aligning with experimenter-based research (Hargreaves & North, 1999; Schäfer & Sedlmeier, 2009; Schäfer et al., 2012; Sloboda et al., 2001; Van Goethem & Sloboda, 2011).
Aesthetic Concepts of Favorite Musical Styles
Extending previous studies, the current study design allowed for a differentiation of aesthetic judgments of subgroups within an aesthetic domain, that is musical styles. The findings show an overlap between style aesthetics in many aspects, possibly united by the aspect of being favorite music, and not just descriptions of musical attributes. Nonetheless, some styles show pronounced differences to others. To capture these differences in detail, we will discuss the top ten CSI values for each musical style, within each of the aesthetic categories. We will start with the most salient category (Mood and Emotion) and end with the least salient one (Aesthetic and Sensory Qualities).
Mood and Emotion
Terms from the Mood and Emotion category were most salient for the styles soul, techno, pop, and house (Figure 2). The prevalent terms (Figure 1: General) included “relaxing,” “touching,” “emotional,” “stimulating,” “soulful,” and “rousing.” “Touching” and “emotional” are closely related, with “touching” denoting felt quality and “emotional” referring to perceived quality. These descriptors appear in the top ten for all styles except electronica, folk, and techno (missing “emotional”), and house (missing both), underscoring the emotional impact of music (Menninghaus et al., 2015) across styles. The arousal components of emotion were also included, with styles being “relaxing” and “stimulating” (“rousing,” “energetic”) at the same time, with the exception of metal, for which “relaxing” is not among the top ten descriptors. Arousal is important for music preference (Schäfer & Sedlmeier, 2011). As shown, music can relieve tension and stress (North et al., 2000). Further, mentioning “relaxing” together with “touching” suggests that listeners feel touched by the music's calming effect. Music being characterized as “energetic” indicates that listeners value physical stimulation, prevalent in metal and rap. The importance of “rhythmic” for electronica (top one) mirrors studies showing that electronica encourages dancing due to its rhythm (Rentfrow & Gosling, 2003; Schäfer & Sedlmeier, 2009). Interestingly, almost all styles showed high salience values for both “stimulating” and “relaxing.” For example, in rap, these terms ranked first and second. This suggests that music can be both relaxing and stimulating, resonating with listeners through diverse emotional functions. Furthermore, musical styles are diverse, not only in their categorization into substyles but also on a song-specific level. Music of a particular style often includes some pieces or substyles that are more relaxing, and others that are more stimulating. This confirms earlier criticism regarding attributing clear functions to musical styles based on their characteristics. Instead, it highlights that favorite music can serve a variety of functions (Greb et al., 2017). In addition, the semantic field “stimulating” included both physical (“pushing,” “vitalizing”) and intellectual components (“makes me think,” “horizon expanding”), indicating that music can be “relaxing” and “stimulating” at the same time, thus resolving potential conflicts of opposite arousal levels.
In sum, our results align with studies suggesting that emotions are crucial for music listening (Campbell et al., 2007; Greb et al., 2017; Hargreaves & North, 1999; Hennion, 2001; Lamont & Webb, 2010; Lonsdale & North, 2011; North et al., 2000; Tarrant et al., 2000). In contrast to other studies, listeners did not focus on the positive and negative poles of valence, but on the general term “emotional.” “Joyful” and “sad” were shown to be important for music in general (Istók et al., 2009), and are key emotional ingredients of being moved (Kuehnast et al., 2014). High arousal is particularly significant for styles such as rap, soul/funk, and electronica/dance (Rentfrow & Gosling, 2003) or beat/folk/country (Schäfer & Sedlmeier, 2009), partially overlapping with our findings. Our finding of associating favorite musical styles with high, and at the same time low, arousal is novel.
Sound Characteristics
The Sound Characteristics category is most salient for metal, electronica, and blues. For metal and electronica, Sound Characteristics are more important based on the mean CSI than the Mood and Emotion category. Key terms for Sound Characteristics from the top ten CSIs (Figure 1) are “melodic,” “rhythmic,” “harmonic,” “loud,” “powerful,” and “clear.” The prominence of “melodic,” “rhythmic,” and “harmonic” suggests that listeners highly value these features in their favorite music. Istók et al. (2009) report similar musical features, supporting the notion that musical parameters, particularly melody, are relevant in musical value judgments (von Appen, 2007). While “melodic” appears in the top ten CSI for many styles, “rhythmic” is crucial for only a selection of styles such as blues, electronica, rap, jazz, and pop, as well as “harmonic,” which is salient for soul and jazz. The specific appeal of electronic music, for example, lies in its rhythmic aspects (Rentfrow & Gosling, 2003), with its energizing and stimulating rhythmic structures enhancing the relevance of “rhythmic” (Jerrentrup, 2008; Papenburg, 2001; Wicke, 2015). The terms “powerful” and “loud” are indicative of metal's defining characteristics (e.g., Ackermann & Merrill, 2022), distinctly differentiating the aesthetic concepts of metal enthusiasts from other styles. This suggests that metal listeners maintain well-defined perceptions of their preferred style, emphasizing a clear distinction from other musical styles and underscoring their identification with metal.
Style and Genre
The Style and Genre category is represented by terms describing style specifics, such as “diversified,” “stylistic,” and “versatile,” often without going into the specifics (which are arguably partly described in more detail within the Sound Characteristics category). These terms highlight the appreciation of the musical variety and diversity inherent in various styles which are prominent in the top ten CSIs of many styles (Figure 1). The highest salience of this category was found for jazz, and techno (Figure 2). Together with the prominence of the terms “diversified” and “stylistic” for jazz, and “versatile” for techno, this indicates that these styles exhibit a difference in their separation into substyles (Brackett, 2014; Dayal, 2013; Tucker & Jackson, 2020). The low importance of this category for rap and metal (Figure 2) indicates that listeners of these styles prioritize other aesthetic concepts, focusing more on sound characteristics as well as on mood and emotion. Distinction into substyles is significant for listeners of many styles, as evidenced by the prominence of the terms of “diversified” and “stylistic” in styles but rap and pop. This supports literature advocating for a more detailed differentiation of taste at the substyle level (Parzer, 2011; Siebrasse & Wald-Fuhrmann, 2023). Finally, “lyrics” is a prominent style-specific term for rap, emphasizing the style's focus on wordplay, storytelling, and rhyming, which are central and highly valued aspects of rap culture (Toop, 2012). And the style-specific term “catchy” mentioned for pop aligns with the reported catchiness of its melodies, rhythms, and harmonies (Warwick, 2014).
Social and Relational
In the Social and Relational category, the terms “interesting” (blues, classical, electronica, jazz), and “encouraging” (electronica, house, pop, rock) emerged in the top ten most salient terms (Figure 1). However, comparing the mean CSI across terms for each category, many styles showed low values (Figure 2), suggesting that social and relational factors are less relevant in describing one's favorite music. For example, terms such as “identifying,” “connecting,” and “situational” were mentioned but did not show high saliences (Figure 1). This finding contrasts with other studies that suggest social components are highly important in defining favorite music (e.g., included in the communication and self-reflection factor in Schäfer & Sedlmeier, 2010). Hence, this category might be non-specific in its application to favorite styles overall.
We observed low salience for social and relational terms, despite other studies clearly demonstrating the social reasons for musical appreciation. When asked to describe their favorite music, participants apparently did not focus on these aspects, possibly because they are unaware of them at that moment. This discrepancy can be attributed to the method and analysis used in the current study. Although social and relational aspects are present in the current study, they rarely appear in the top ten salient terms.
Aesthetic and Sensory Qualities
The Aesthetic and Sensory Qualities category exhibited rather low mean CSI values across musical styles. Within this category, folk, metal, jazz, and classical had the highest mean CSI (Figure 2). Between the categories, however, it was the least prominent for many styles (e.g., blues, classical, electronica, house, pop, rock), behind the mean CSIs for Social and Relational (Figure 2). Salient adjectives were found in connection with classical, electronica, folk, rap, jazz, metal, reggae, and techno (Figure 1). “Intelligent” turned up in the top ten for folk, rap, and reggae, “demanding” for classical, “complex” for techno, suggesting that these listeners expect cognitive stimulation, likely achieved through musical complexity (Campbell et al., 2007; Hargreaves & North, 1999; Hennion, 2001; Lamont & Webb, 2010; Lonsdale & North, 2011; North et al., 2000; Tarrant et al., 2000), or through meaningful lyrics in rap (as discussed in Style and Genre).
Notably, “beautiful” occurs only on rank ten in jazz, and is on rank 12 for soul and rap, rank 14 for electronica, rank 16 for house, and rank 18 for classical, pop, and reggae, and above rank 24 for rock, blues, folk, techno, and metal. Overall, “beautiful” was at rank 17 with regard to the CSI and frequency (mentioned 172 times, representing 2.14% of all response). Hence, the term “beautiful” is a descriptor for favorite music but not nearly as important as it was seen in previous studies, when asking for qualities of aesthetic objects in general (Augustin et al., 2012b; Istók et al., 2009; Jacobsen et al., 2004; Knoop et al., 2016). This disparity can probably be attributed to methodological differences in the task. Our study specifically focused on describing favorite music while intentionally omitting the term “aesthetics” in the instruction. Therefore, it can be assumed that previous instructions including the term “aesthetics” triggered the mention of “beautiful” much more than our instruction on favorite music. In the same vein, the bipolarity of “beautiful” and “ugly” observed by (Jacobsen et al., 2004) is absent in our data, with negative descriptors being rare. References to “ugly” in the context of favorite music seem improbable, likely being more relevant to visual elements. Participants in our study also used negatively connotated features such as “hateful,” “evil,” “inhuman,” “satanic,” “insulting,” “unpleasant,” and “difficult” to describe favorite music. However, these terms were infrequent and summarized under the term “negative,” which ranked highest for rap (35th rank), followed by metal (41st rank). Much more prevalent in the current study were adjectives with a positive valence related to positive emotions (“touching,” “joyful”), aesthetic attributes (“good,” “funny”), or social qualities (“connecting”), suggesting that individuals associate favorite music with positive feelings and descriptions.
Our study bridges the gap between personal preferences and aesthetic descriptions, differentiating aesthetic concepts across varying listener profiles, and enhancing our understanding of why people derive pleasure from their favorite music. By analyzing the adjectives listeners use to describe their favorite music, we found that distinct musical characteristics are attributed to different styles (e.g., metal is described as “loud” and “powerful,” pop is described as “catchy”). Emotional descriptors such as “touching” and “stimulating” are common across all styles, indicating that these emotions can be evoked by various styles of music. These emotional descriptors highlight the diverse functions that favorite music can serve, such as being “relaxing” or “rousing,” regardless of style. This suggests that listeners of different musical styles can experience similar emotions through different musical characteristics. Future research should therefore explore not only the emotional effects and functions of music on listeners but also the specific musical characteristics and parameters. This will help uncover why different musical characteristics can elicit the same emotions by its listeners, and therefore, help to map out the aesthetic judgements of affectively significant, and personally meaningful aesthetic experiences.
Supplemental Material
sj-docx-1-art-10.1177_02762374251365160 - Supplemental material for What Elevates Music to a ‘Favorite’? On the Aesthetics of Music from the Perspective of Musical Taste
Supplemental material, sj-docx-1-art-10.1177_02762374251365160 for What Elevates Music to a ‘Favorite’? On the Aesthetics of Music from the Perspective of Musical Taste by Emily Gernandt, Elke B. Lange and Julia Merrill in Empirical Studies of the Arts
Footnotes
Ethical Considerations
All procedures were approved by the Ethics Council of the Max Planck Institute for Empirical Aesthetics.
Consent to Participate
Before starting the survey, participants were informed about the scientific background of the study and the anonymized data collection. They gave consent by proceeding with the questionnaire after the information page.
Consent for Publication
Not applicable.
Authors’ Contributions
EL designed the research and collected the data. All authors conceived and designed the analyses. EG and EL analyzed the data. All authors interpreted the results. EG wrote the first draft of the manuscript. EL and JM revised the initial draft. All authors revised, read and approved the final manuscript.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Open Access funding enabled by the Max Planck Society. This research was supported by the Max Planck Society and did not receive any grant from funding agencies in the public, commercial, or not-for-profit sectors.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online and on https://osf.io/hqpdz.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
