Abstract
Psycholinguistic inquiry can provide insight into the way the words people use reflect psychological states, including emotional states. There is limited research on the use of Linguistic Inquiry and Word Count (LIWC) software to investigate the psycholinguistic properties of emotional memories related to music. The purpose of this study was to test the extent to which LIWC can be used in the analysis of autobiographical memories related to music. Participants were undergraduates (N = 99) at two large universities in the southeastern United States. Each participant was asked to write about times in their lives when music or experiences with music made them feel positive or negative emotions. The researchers conducted a content analysis of participants’ responses and used LIWC software to quantify emotion words (positive and negative), pronouns, causal thinking words, and insight words. Participants used significantly more positive than negative emotion words to describe positive memories of music, but there was no significant difference between the rates of negative and positive emotion words to describe negative memories of music. The content analysis revealed a similar trend: 51% of participants described mixed, conflicting, or changing emotions when describing negative experiences, whereas descriptions of positive experiences tended to be highly positive. Many participants wrote about social experiences and musical achievement. Results of this study offer insights on how humans describe music-related autobiographical memories. LIWC appears to be a useful tool for method triangulation when performing qualitative analysis of emotions in autobiographical memories of musical experiences.
Understanding how human memories—especially those memories related to previous experiences with music—interact with emotions is an important goal for psychologists and musicians. In order to better understand autobiographical memories of music, Gabrielsson conducted a series of studies on strong experiences with music (SEM), which he summarized in a book chapter published in 2010 and a book in Swedish in 2008 and in English in 2011. Participants (N = 953) in his studies were asked to describe the “strongest, most intense experience of music” they ever had (Gabrielsson, 2010, p. 551). Upon conducting a content analysis, Gabrielsson identified seven basic categories of responses, one of which was feelings/emotion. He also described sub-categories of emotional responses, including positive, negative, and different (mixed, conflicting, and changed) feelings/emotions (Gabrielsson, 2008/2011). Negative feelings often depended on associated circumstances (e.g., unhappiness in relationships, illness, loss of loved ones) rather than the music itself. Some participants reported mixed feelings related to musical experiences, including bittersweet feelings or sadness over pleasant events from the past that may not happen again.
Other researchers have observed that people describing SEM sometimes report mixed feelings. In a qualitative analysis, Lamont (2011) drew on a positive psychology framework to explore pleasure, engagement, and meaning in university students’ SEM. She found that most intense experiences students described were positive and occurred when attending live events with others. When participants did describe negative experiences, which happened most frequently while listening to recorded music, the experiences were not overwhelmingly negative (i.e., participants described mixed emotions) and the SEM “. . .seemed to help people to come to terms with their condition” (Lamont, 2011, p. 236). This finding seems to imply that participants were using music for affect regulation, an important function of music listening, particularly for younger adults (Groarke & Hogan, 2016).
In addition to affect regulation, mixed feelings in response to music-related autobiographical memories might well be framed within the context of nostalgia, which has been described as a blend of positive and negative feelings associated with a longing for the past (Wildschut et al., 2006). Investigations of music-evoked nostalgia specifically have indicated that nostalgia was more often identified in songs that were “autobiographically salient, that were familiar to them, that they found arousing, that evoked in them greater numbers of positive and negative emotions, and that gave rise to mixed emotions” (Barrett et al., 2010, p. 394). However, those effects were not independent of individual differences between listeners, particularly nostalgia proneness. Researchers have identified other individual differences, including personality, age, gender, and listener activity, that correlate with emotional reactions to music (Juslin et al., 2011).
Investigating a musical phenomenon similar to SEM, Schäfer et al. (2014) used a grounded theory methodology to examine intense musical experiences, which are defined as unforgettable experiences with music that stand out and involve changed perceptions, deep feelings, and intense physical reactions. Their results indicated that listeners who feel intense musical experiences tend to experience altered states of consciousness, which can lead to feelings of harmony and self-realization—feelings that can motivate them to achieve the same harmony in their daily lives. Schäfer and colleagues reported that listeners’ references to feelings “indicated that there was no rational or analytic thinking during the experience” (2014, p. 537).
Psycholinguistic Inquiry with LIWC
Along with qualitative analyses, psycholinguistic analysis, the study of how language relates to underlying psychological processes (Crystal, 2008), is another way to understand emotional responses to music. Psycholinguistic analysis can be accomplished through the use of text analysis software, such as Linguistic Inquiry and Word Count (LIWC; Pennebaker Conglomerates, Inc., n.d.), which allows researchers to investigate patterns of word usage in people’s written text and transcribed speech (Pennebaker, Booth, et al., 2015). By examining people’s word usage, researchers are able to examine covert psychological processes (Chung & Pennebaker, 2007; Tausczik & Pennebaker, 2010). The software analyzes text documents using a comprehensive internal dictionary of various linguistic dimensions (e.g., affect words, money words, positive emotion words, third-person pronouns, future words, etc.). Researchers can choose the linguistic dimensions of interest for a particular investigation, and results are calculated as percentages of total word count for each dimension, thereby controlling for length/word count (Pennebaker, Boyd, et al., 2015).
Using LIWC, researchers have identified linguistic patterns that people tend to use in speech or in writing when they are experiencing positive emotions that are distinct from patterns that appear when people are experiencing negative emotions. For example, people writing about a positive experience tend to use more first-person plural pronouns (“we”), whereas people tend to use more first-person singular (“I”) pronouns when they are sad and more third-person pronouns (“she,” “he,” or “they”) when angry. People experiencing negative emotions tend to use words indicative of causal thinking and insight (Pennebaker, 2011; Rude et al., 2004).
Uses of LIWC in Investigations of Emotional Responses to Music
Several researchers have used LIWC in investigations of music-evoked autobiographical memories (MEAMs), which often involve strong positive emotions, nostalgia, and some mixed feelings (Jakubowski & Ghosh; Janata et al., 2007). Using a naturalistic approach, Jakubowski and Ghosh (2019) examined participants’ self-reported MEAMs and music listening habits described in a diary over seven days; they found that MEAMs were experienced fairly frequently (once per day, on average) and occurred in response to a variety of music. Analysis of LIWC data indicated a high proportion of positive emotion words, a notably low proportion of negative emotion words, and a relatively high use of social process words (especially words referencing family and friends) in their entries. Similarly, results of a questionnaire study including a large corpus of popular music indicated that approximately 30% of songs cued MEAMs for participants and that the MEAMs usually involved positive emotions (Janata et al., 2007). These two studies (Jakubowski & Ghosh, 2019; Janata et al., 2007) were examinations of MEAMs and listening habits.
Using LIWC in a different way, Yinger and Springer (2019) tested the feasibility of using LIWC software to measure listeners’ emotional responses to music previously identified as happy or sad in character. Participants listened to two recorded excerpts of music—one happy and one sad—and described the feelings and emotions that they perceived and felt in response to the music. Consistent with the researchers’ hypotheses, listeners wrote more positive emotion words than negative emotion words in response to the “happy” excerpt, with the reverse trend being evidenced in their responses to the “sad” excerpt. Although Yinger and Springer’s results provided evidence to support the feasibility of using LIWC software to study emotional responses to music, additional evidence is needed to support the validity of LIWC in investigations of emotional responses related to autobiographical memories of music.
Scholars have noted several limitations to using LIWC for text analysis, particularly the risk of losing contextual meaning, since LIWC does not consider context, irony, sarcasm, or idioms (Tausczik & Pennebaker, 2010). Some researchers have attempted to address this limitation by using a qualitative approach concurrently with LIWC analysis (Czechowski et al., 2016; Firmin et al., 2017). Renz et al. (2018) have suggested using LIWC alongside qualitative content analysis as an intramethod approach to triangulation; however, we only identified one study in which LIWC and qualitative analysis were used with the same narrative data on emotional responses to music (Garrido et al., 2016). Garrido et al. (2016) used LIWC alongside a qualitative analysis as part of an experiment to examine the effects of a listening program on mood. To ensure that their participants complied with the instructions to listen to the stimuli, they asked participants to record their thoughts about the listening selections in a diary. Those diary entries were analyzed using LIWC software, as well as a qualitative thematic analysis. Garrido and colleagues found that listeners’ free responses to a sad music playlist had more sad emotion words and that positive emotional responses were reflected in participants’ uses of words related to social inclusion, motion, achievement, and religion. We did not find any studies in which researchers included pronouns as a variable of interest when using LIWC to better understand emotional responses to music, and we found no studies that used LIWC to investigate word usage that may be indicative of strong experiences with music (Gabrielsson, 2008/2011, 2010).
Need for the Study
Taken together, findings from previous literature have indicated that LIWC can be used as a measure of emotional response to music (Yinger & Springer, 2019) and that the software can identify emotions associated with autobiographical memories (Jakubowski & Ghosh, 2019; Janata et al., 2007). What remains to be known, however, is the degree to which LIWC can be used to investigate participants’ positive and negative autobiographical memories associated with music. Given the software’s ability to quantify emotions based on rates of word usage, it could serve as a useful measure to accompany a qualitative content analysis. By using this “method triangulation” (Juslin, 2016, p. 198) to explore people’s autobiographical memories associated with music, it will not only provide new data to explain how people experience music; it will also allow for a comparison of findings with previous research on strong experiences with music (Gabrielsson, 2008/2011, 2010).
The purpose of this study was to test the extent to which LIWC can be used in the analysis of autobiographical memories related to music alongside a content analysis. The first objective was to use LIWC, a quantitative method, to study the words that college students used to describe experiences with music that made them feel positive or negative emotions. The second objective was to better understand, through qualitative content analysis, the content of students’ writings about experiences with music that made them feel positive or negative emotions. By analyzing the same written responses with LIWC and a qualitative content analysis, we hoped to identify convergence and discrepancy between an objective approach (using LIWC to count words) and a subjective inductive approach (qualitative content analysis) to understanding autobiographical memories of emotional experiences with music (Czechowski et al., 2016). By investigating psycholinguistic data generated by LIWC along with themes identified though a content analysis of college students’ autobiographical musical experiences, this study will help future researchers and educators better understand the links between emotions and memories related to musical experiences. Specifically, we addressed the following research questions:
Will there be any differences in the proportions of words (positive emotion words, negative emotion words, pronouns, causal thinking words, and insight words), as measured by LIWC, used to describe positive versus negative memories related to music?
What emotions and experiences do college students write about when asked to recall positive and negative emotional memories involving music?
Hypotheses
Based on the findings of previous researchers (Kahn et al., 2007; Yinger & Springer, 2019), we hypothesized that participants would use more positive than negative emotion words to describe an experience with music that made them feel positive emotions, and more negative than positive emotion words to describe an experience with music that made them feel negative emotions. We also predicted that, in line with previous research on psycholinguistic profiles related to positive and negative emotions (Pennebaker, 2011; Rude et al., 2004), participants would use more first-person plural pronouns when describing positive experiences with music and more first-person singular pronouns, third-person singular pronouns, causal thinking words, and insight words when describing negative experiences with music. Regarding the second research question, we anticipated that the emotions and experiences described in participants’ free responses would match the SEM framework described by Gabrielsson (2008/2011, 2010) to some degree; however, we maintained an exploratory approach to the analysis to allow additional insights to emerge.
Method
Participants
Results of a power analysis using G*Power software (Faul et al., 2007) indicated a necessary sample size of 80 participants for this study. (Input parameters included an alpha level of .01, a power level of β = .8, and an effect size of d = 0.4.) Undergraduate students (N = 99, 50 female, 47 male and 2 non-binary) were recruited from two large universities in the southeastern United States. Students from courses for both music majors (e.g., introductory music education courses or music theory courses) and non-music majors (e.g., music appreciation or introduction to music courses) participated in the present study. They ranged in age from 18 to 31 years, with an average age of 19.82 years (SD = 1.96). There were 61 music majors, 34 non-music majors, and four students who did not identify their major.
Instrument
Participants responded in writing to two writing prompts, which were adapted from writing prompts used by Kahn et al. (2007) and Gabrielsson (2010). Feedback on the exact wording of the writing prompts was solicited from a panel of three experts who have published research on the measurement of affective responses to music. The finalized writing prompts used in this study were as follows:
Write about a time in your life when music or an experience with music made you feel negative emotions. You may write about any experience with music, but it is important that you write about a time when you felt negative emotions.
Write about a time in your life when music or an experience with music made you feel positive emotions. You may write about any experience with music, but it is important that you write about a time when you felt positive emotions.
Procedure
After receiving IRB approval from the University of Kentucky and Florida State University, we obtained permission from instructors of undergraduate music courses at our respective institutions to spend five minutes during class explaining the study to students and asking for volunteers to participate in the study. Students who were interested in participating in the study signed up for a 15-minute time slot. Following completion of a demographic questionnaire, the researcher or a research assistant played a recording for participants, which consisted of instructions and the two writing prompts. Pre-recorded instructions were used to control for experimenter effects and to standardize protocol across data collection sites. The order of presentation of the two writing prompts was counterbalanced between groups, and participants were given five minutes to respond in writing to each prompt.
LIWC Analysis
After data collection was complete, the research assistant typed all written responses and prepared Word document files for LIWC analysis. Based on previous research on linguistic profiles of people experiencing positive and negative emotions (Pennebaker, 2011), we selected the following LIWC categories for analysis: total word count, positive emotion words (e.g., “love,” “nice,” “sweet”), negative emotion words (e.g., “hurt,” “ugly,” “nasty”), first-person singular pronouns (e.g., “I,” “me,” “mine”), first-person plural pronouns (e.g., “we,” “us,” “our”), third-person singular pronouns (e.g., “she,” “her,” “him”), causal thinking words (e.g., “because,” “effect”) and insight words (“think,” “know”). We used SPSS 25 to conduct all statistical analyses.
Content Analysis
In order to answer the second research question, we used both deductive and inductive approaches to content analysis, using elements of the strong experiences with music (SEM) descriptive system (Gabrielsson, 2010), classifying excerpts based on the type of musical experience in which participants engaged (making music or listening to music), and generating new themes based on the data itself. We also noted specific emotion words that participants used to describe their experiences. We followed the approach to content analysis described by Elo and Kyngäs (2008), which consists of three main phases: preparation, organizing, and reporting. We each read all the written responses (N = 198) independently in an attempt to immerse ourselves in the data and made notes on our observations. After discussing initial observations and reaching agreement on how to categorize elements of the written responses, we again read all of the written responses independently and attempted to classify each written response as one of the following: (a) positive feelings; negative feelings; or mixed, conflicting, or changed feelings (Gabrielsson, 2010) and (b) making music or listening to music. Both of these categories were drawn from Gabrielsson’s (2010) SEM framework. During this deductive analysis, we also noted whether responses included sub-categories described by Gabrielsson (2010), emotional response ascribed to the music itself or due to other circumstances, using music to influence feelings. On subsequent reviews of the data, we conducted an inductive analysis, following an iterative process using Microsoft Excel to derive codes. We then compared classifications and discussed any discrepancies in order to reach agreement on themes and sub-themes.
Results
LIWC Analysis
Because data were non-normally distributed, we used non-parametric statistics to compare linguistic variables written in response to the positive and negative emotion prompts, applying a more stringent alpha level (α = .01) for multiple comparisons. A series of Mann Whitney U tests revealed no significant differences for any of the LIWC variables based on the order of writing prompts (negative emotion prompt first versus positive emotion prompt first), p > .01.
Results of a Wilcoxon signed rank test indicated that participants wrote significantly more positive emotion words (Mdn = 5.48) than negative emotion words (Mdn = 0.63) in response to the positive emotion prompt, Z = 8.37, p < .001, r = .59. However, there was no significant difference for the negative emotion prompt (p = .033, r = .15). Participants wrote approximately the same number of positive emotion words (Mdn = 2.67) as negative emotion words (Mdn = 3.45) in response to the negative emotion prompt. Participants also wrote significantly more positive emotion words in response to the positive emotion prompt compared to the negative emotion prompt, Z = 7.32, p < .001, r = .52, and significantly more negative emotion words in response to the negative emotion prompt compared to the positive emotion prompt, Z = 7.88, p < .001, r = .56. Medians, means, standard deviations, and effect sizes for LIWC variables are shown in Table 1.
Word Counts and Percentages of LIWC Variables in Responses to Positive and Negative Emotion Writing Prompts.
Note. Pos. = positive, Neg. = negative. Word count is measured as the number of words. Medians, means, and standard deviations are percentages of word count. The p values and effect sizes (r) are the results of the Wilcoxon signed rank tests.
Participants wrote significantly more positive than negative emotion words in response to the positive emotion prompt, p < .001 *, r = .59.
There were no significant differences in participants’ rates of positive and negative emotion words written in response to the negative emotion prompt, p = .033, r = .15.
p < .01.
Participants used first-person plural pronouns (e.g., “we”) significantly more often in response to the positive emotion prompt (M = 1.34, SD = 1.85) than they did in response to the negative emotion prompt (M = 0.67, SD = 1.09), Z = 2.89, p = .004, r = .21. (Means and standard deviations are reported here because the median values were zeros.) In contrast, participants used third-person singular pronouns (e.g., “she” and “he”) more often in response to the negative emotion prompt (M = 1.18, SD = 1.75) than in response to the positive emotion prompt (M = 0.46, SD = 1.26), Z = 3.60, p < .001, r = .26. There were no significant differences in the proportions of first-person singular pronouns (e.g., “I”), causal thinking word or insight words participants used in response to the positive and negative emotion prompts, p > .01.
Content Analysis
The most frequently expressed emotion in responses to the positive emotion prompt was happiness (n = 40), followed by joy (n = 19), whereas sadness (n = 26) was the most frequently expressed emotion in responses to the negative emotion prompt, followed by anger (n = 14). Tables 2 and 3 show the frequency with which participants in the present study reported the positive and negative emotions, respectively, identified in Gabrielsson’s research (2010), ranked from lowest to highest arousal.
Positive Emotions from Gabrielsson’s SEM Descriptive System Reported by Participants in the Present Study, Ordered from Lowest to Highest Level of Arousal.
Note. SEM = strong experiences with music. Some participants used more than one emotion when writing about positive experiences with music.
Negative Emotions from Gabrielsson’s SEM Descriptive System Reported by Participants in the Present Study, Ordered from Lowest to Highest Level of Arousal.
Note. SEM = strong experiences with music. Some participants used more than one emotion when writing about negative experiences with music.
Many participants (68%) wrote solely about positive feelings in response to the prompt about positive emotional experiences with music, whereas approximately half (49%) of participants wrote solely about negative feelings in response to the prompt about negative emotional experiences with music. Participants expressed mixed, conflicting, or changed feelings when writing in response to both prompts, although they did so more frequently in response to the negative emotion prompt. Participants were fairly evenly divided in terms of the type of musical experience they described (making music or listening to music) when writing about both positive and negative experiences. Some participants wrote about more than one type of musical experience. Participants’ positive and negative emotions were more often due to other circumstances rather than ascribed to the music itself. Trends in types of feelings and musical experiences expressed in written responses to positive and negative emotion prompts are shown in Table 4.
Types of Feelings, Musical Experiences, and Sources of Emotion Reported in Participants’ Written Responses (N = 99).
Note. Some participants wrote about making music and listening to music or ascribed their feelings to the music itself and to other circumstances.
We identified two themes through inductive content analysis: social experiences and musical achievement. Within the theme social experiences, we noted two sub-themes that connected several of the categories identified through deductive analysis: experiencing positive emotions while sharing musical experiences with others and experiencing negative emotions due to interpersonal stressors. Within the theme musical achievement, we noted two sub-themes that connected several of the categories identified through deductive analysis: positive emotions related to musical achievement and negative emotions related to musical underachievement (see Figure 1 for examples of quotations representative of each sub-theme).
Themes, sub-themes, and examples identified through content analysis.
We noted that participants wrote about feeling positive emotions related to the shared experience of making or listening to music with others. One participant wrote the following about a particularly memorable performance that illustrates positive emotions while making music with others:
We performed in this huge cathedral, and the sound and power of our voices raised together was indescribable. I remember feeling almost weightless I was so full of elation and joy. This was truly one of the strongest, most influential and aesthetic experiences of my life. People in the audience came up to us crying afterward, and that communal spirit I felt with a group of strangers who were the audience is truly one I will never forget.
Participants tended to talk about experiences with people differently in their descriptions of experiences with negative emotions than they did in their descriptions of experiences with music and positive emotions. Rather than describing shared musical experiences, when participants wrote about other people in their descriptions of negative emotions, they tended to write about interpersonal stressors, particularly with music teachers or other members of a musical ensemble but also with family or friends. One particularly intense description of an experience with negative emotions involved a participant’s mother. In this excerpt, it is notable that the participant used first-person singular pronouns (“I,” “me,” “my”) and third-person singular pronouns (“she”) yet no first-person plural pronouns (“we,” “us”). Even though this experience involved another person, the lack of first-person plural pronouns suggests that the participant did not feel a close connection to her mother during this experience:
One time my mother was drunk. She blasted rock n’ roll through the tiny apartment and screamed at me. I could not hear my own thoughts. The music just added to the fear and pain. Every time I hear music similar to what she played, the memory haunts me. I don’t think the music caused the pain, but it definitely added to it. My mother took all of my belongings and threw them outside. As I sit there crying, the music made me anxious and nervous.
In addition to social experiences, many participants wrote about positive emotions related to their own musical achievement (“I will forever be proud of my band program for having a high standard for excellence”) or that of other performers (“Every corps sounded flawless”). In contrast, participants’ musical underachievement (“I played every imaginable wrong note and came out crying”) or that of other performers or teachers (“He didn’t put time into teaching, just ran the pieces a couple times and called it okay”) was often related to negative emotions.
The following excerpt illustrates a challenge that several participants noted when asked to describe an experience with negative emotions and music that may be part of the reason why so many of the responses to the negative emotion prompt included mixed, conflicting, or changed feelings. It is possible to experience or recognize emotions with a negative valence while not viewing the experience as a negative one:
There has never been a time in my life that I hear music and think negatively. As a musician, I listen to music in order to feel positive about the world around me. Does some music make me cry? Yes, it does. But it is not because the music is bad, it is just that the composer’s writing technique conveys an emotion that humans can project. However, music has made me feel tense but immediately relaxed following a proper resolution.
In the excerpt above, feeling positive about the world even when listening to music that makes one cry is an example of mixed or conflicting feelings, and feeling tense and then relaxed is an example of changing feelings. Figure 2 shows additional representative quotations from participants demonstrating the tendency to write about positive emotions and use first-person plural pronouns when describing positive experiences with music and the tendency to describe mixed, conflicting, or changed emotions and use third-person pronouns when describing negative experiences with music.
Representative quotations from participants.
Discussion
Results of this study provided information about the words people use when writing about positive and negative autobiographical memories of musical experiences. When writing about experiences with music when they felt positive emotions, participants used significantly more positive emotion words and first-person plural words (“we”). This finding supports our observation that participants’ experiences with positive emotions and music often involved close friends or groups of people connected by shared musical experiences. In contrast, when writing about experiences with music when they felt negative emotions, participants used significantly more third-person singular pronouns (“she” and “he”).
These findings support previous research suggesting that people who experience positive emotions tend to use more first-person plural pronouns, whereas people experiencing anger tend to use more third-person singular pronouns (Pennebaker, 2011). Although previous researchers have found that people experiencing sadness tend to use more first-person singular pronouns (“I,” “me;” Rude et al., 2004), there was no significant difference in the present study between first-person singular pronoun use when writing about experiences with negative and positive emotions. This could be due in part to the fact that many participants wrote about negative emotions other than sadness, such as anger or anxiety, because sadness, anger and anxiety have different linguistic profiles (Pennebaker, 2011).
Regarding the second research question, results of the content analysis indicated that participants described a variety of positive and negative experiences with music and associated emotions. In some of the situations described by participants, the music itself provided the source of the emotional experience, yet more frequently, the context/situation was the source of the emotional experience, particularly when describing negative emotions. This trend reflects Gabrielsson’s findings (2008/2011) that negative feelings in SEM often depended on associated circumstances rather than the music itself. Participants in the present study frequently described experiences with music that reminded them of other people, a finding that is largely consistent with research by Janata and colleagues (2007), whose participants described autobiographical associations with music that were most often connected to one or more persons. As was the case in Lamont’s (2011) study of university students’ strong experiences with music, many participants wrote about attending live events with others when describing experiences with music that made them feel positive emotions.
Results of the present study are consistent with Gabrielsson’s (2008/2011, 2010) studies of SEM in that participants described many similar feelings associated with musical memories (see Tables 3 and 4). However, there were observed differences between the results of these studies as well. For example, although the majority of Gabrielsson’s participants reported that strong experiences occurred while listening to music, participants in this study reported memories associated with both making music and listening to music. It is possible that this difference between experiences reported by participants in Gabrielsson’s study and the present study could be related to differences in the musical backgrounds of participants themselves. Participants in the present study were college students, many of whom were music majors, whereas participants in Gabrielsson’s studies represented a larger range of ages.
Participants wrote about more solely positive feelings (68%) than mixed, conflicting or changed feelings (32%) in response to the positive emotion prompt, yet their responses to the negative emotion prompt demonstrated a different trend. Over half (51%) of participants described mixed, conflicting or changed feelings related to their experience—more than those who described solely negative feelings (49%). In Gabrielsson’s studies (2008/2011, 2010), fewer participants reported mixed, conflicting or changed feelings (21%) than in the present study. The explanation for these contrasting findings remains unclear, but the results could be due to the different experiences participants chose to write about in their responses. As mentioned previously, a number of participants in the present study described an experience with music that was due to a negative experience with a music teacher. Some described this in mixed terms, while others described the experience as unilaterally negative. For example, one participant described a very strict piano teacher who caused her to feel embarrassed and ashamed of making mistakes, yet she also indicated that those negative experiences had positive outcomes “because it helped me push myself to learn the music and practice it critically in times I would otherwise have given up.” Therefore, those findings that differed from Gabrielsson’s (2008/2011, 2010) may simply be a function of the specific experiences described in the present study.
Results of this study, as evidenced by both the LIWC data and the content analysis, demonstrated differences that reflect how humans process emotions related to autobiographical memories of musical experiences. Participants in the present study wrote more positive emotion words when describing positive experiences with music, and the experiences they mentioned were more uniformly positive. Regarding negative experiences in music, they wrote similar proportions of positive and negative emotion words, and the experiences they described exhibited more mixed, conflicted, and changing feelings. These results may be due to the nature of episodic memory in general, or they could be a result of some listeners’ ability to deactivate displeasure circuits when listening to music. Because autobiographical memories are episodic rather than semantic (Snyder, 2016), they are usually dynamic and may even be changed by the simple act of memory retrieval (Scully et al., 2017). Therefore, these differences may be reflective of the dynamic qualities of episodic memories—especially those memories related to negative musical experiences. An alternative explanation could come from the work of Garrido and Schubert (2013), who proposed a dissociation theory of aesthetic enjoyment to explain the enjoyment some people feel when experiencing negative emotion while listening to music. Garrido and Schubert suggested that some people are able to enter states of absorption and deactivate displeasure circuits when listening to music. This theory could help explain the mixed, conflicting, or changed emotions that about half of participants expressed when writing about experiences with music that made them feel negative emotions; perhaps some of these participants enjoyed experiences with music that made them feel negative emotions.
The finding that participants used similar proportions of positive and negative emotion words (as measured by LIWC) to describe experiences with negative emotions and music is congruent with the finding from the content analysis that participants often described mixed, conflicting or changed feelings when writing about an experience with negative emotions and music. This congruence suggests that LIWC could be used in mixed methods research on emotional responses to music to elucidate or quantify themes identified through qualitative analysis. Future researchers might investigate other LIWC variables (e.g., social processes, cognitive processes, perceptual processes) alongside themes related to experiences with music identified through qualitative content analysis.
Limitations
One limitation to using LIWC to study emotional responses to music is that LIWC only groups emotions into broad categories (i.e., affect; positive or negative emotion words; and anxiety, anger, or sadness). Not every writing sample will include language from every LIWC dictionary, which means that there could be a high proportion of responses with values of zero, skewing data and leading to violations of assumptions of normality. However, since LIWC allows researchers to ask open-ended questions about musical experiences and then quickly process a large volume of data, it a useful tool to complement other measures of emotional responses to music. Although there were statistically significant differences between positive and negative emotion prompts for first-person plural and third-person singular pronouns, these differences may not have much practical significance, since the differences in proportions were so small (see Table 1).
Suggestions for Future Researchers
Future applications of LIWC may be useful to researchers in gaining a better understanding of the musical experience for listeners of all levels of training and experience. In particular, an examination of how non-musicians respond to live music performances would be beneficial and could be accomplished using LIWC software. Another possible avenue for future research would be a comparison of listeners’ written and spoken reactions to musical performances, which would give researchers data describing potential differences based on these two response modes. In light of research suggesting that personality, age, gender, and listener activity correlate with emotional reactions to music (Juslin et al., 2011), future researchers should investigate the correlation between these characteristics of participants and psycholinguistic variables evident in written or spoken responses to questions about emotion in positive and negative experiences with music. Finally, it would also be helpful to examine listeners’ word usage in their responses to musical performances across time as they gain more musical training, thereby offering longitudinal implications about how listeners’ responses to music might change as a function of musical training. Regardless, future investigations of this type should be conducted because many aspects of emotional experiences with music remain to be understood. Nevertheless, based on the congruence in the present study, which used a multi-strategy approach, between the results of the quantitative analysis conducted using LIWC and the qualitative content analysis, LIWC appears to be a useful method of triangulation when attempting to understand emotional responses to autobiographical memories of musical experiences in depth.
Footnotes
Acknowledgements
The authors would like to thank Cecilia Wright for assistance with recruitment, data collection, and data preparation and Amanda L. Schlegel, Jason M. Silveira, and Matthew L. Williams for providing feedback on the writing prompts.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the University of Kentucky College of Fine Arts Research Enhancement Grant.
