Abstract
Alexithymia is a trait characterized by decreased emotional response to visual or verbal stimuli. However, Lyvers et al. suggest that alexithymia is positively correlated with the magnitude of emotional response to music (ERM). This study reexamines their findings in two sets of Asian population (China [n = 344] and Japan [n = 341]) using online surveys to investigate the subfactors of alexithymia. The Toronto Alexithymia Scale and the Geneva Emotional Music Scale (GEMS) were used to measure alexithymia and ERM in the participants, along with trait anxiety and music experiences in their daily lives. The obtained data showed that in both the Chinese and the Japanese populations, individuals with higher alexithymia tend to have higher GEMS scores (i.e., higher ERM), which is consistent with the finding of the previous study. Furthermore, alexithymia and ERM are correlated because of difficulty in identifying feelings, a subfactor of alexithymia. These results were not influenced by trait anxiety. In addition, alexithymia showed no correlation with either the frequency of listening to music or the degree of music absorption in daily life. We argued that the nonverbal characteristic of music might be key to the association between alexithymia and ERM.
Alexithymia is a personality or psychological trait that impairs a person’s ability to demonstrate emotional awareness and form interpersonal relationships and social attachments (Sifneos, 1973). It is mainly characterized by the reduced ability to identify and explain emotions experienced by oneself or others, including poor imagination and externally oriented (or concrete) thinking (Goerlich, 2018; Taylor et al., 1985). People with higher as opposed to lower alexithymia tend to exhibit a lower accuracy in recognizing affective contents (i.e., ability to identify emotions) in stimuli (Lane et al., 1996; Parker et al., 1993). Furthermore, people with higher alexithymia demonstrate a decreased ability for emotional responses (or arousal) to stimuli, as measured by self-reports or physiological indexed (Pollatos & Gramann, 2011; Roedema & Simons, 1999).
Most empirical studies on alexithymia have used either visual or verbal stimuli to examine the emotional characteristics of alexithymia. In our daily lives, music is also a popular stimulus, which is often used for voluntarily evoking or modulating affective experiences. Nevertheless, emotional experiences of music in alexithymia have rarely been examined.
Some researchers examining the psychology of music have proposed to classify musical emotions into two aspects (Evans & Schubert, 2008; Gabrielsson, 2001). One is the cognition of emotion expressed by music, which is referred to as perceived emotions. This is concerning the emotion inherent in the music itself. The other is the emotional experience induced by music, which is called felt or induced emotions. This is regarding the listeners’ emotions. These two aspects differ in terms of the locus of emotions—the former is external, while the latter is internal for individuals (Eerola & Vuoskoski, 2011, 2013; Evans & Schubert, 2008).
Few studies have examined the relationship between alexithymia and musical emotions, in particular, the perceived emotions (i.e., recognition of the emotion expressed in music; Allen et al., 2013; Larwood et al., 2021; Punkanen et al., 2011; Taruffi et al., 2017). These studies reported that people with higher alexithymia exhibit poorer cognitive abilities or reduced arousal judgment for emotions in music. Such results are consistent with those of previous studies that used visual or verbal stimuli as mentioned above.
The present study focuses on the felt emotions (i.e., emotional experiences induced in listeners by music), which have hardly been examined in the alexithymia literature. As a matter of wording, music psychology often refers to felt emotions as “emotional response to music” (ERM; Ladinig & Schellenberg, 2012; Schmidt & Trainor, 2001; Vuoskoski et al., 2012). To ensure consistency with an important previous study mentioned below, this study mainly refers to felt emotions as ERM.
To our knowledge, only a recent study by Lyvers et al., (2020) has examined ERM in relation to alexithymia. They hypothesized that people with high alexithymia scores (measured using the 20-item Toronto Alexithymia Scale [TAS-20]; Bagby et al., 1994) will have lower ERM scores (measured using the Geneva Emotional Music Scale [GEMS]; Zentner & Eerola, 2009; Zentner et al., 2008) as people with alexithymia are considered to have decreased emotional experiences. However, unexpectedly, they found a positive correlation between TAS-20 and GEMS scores, although the effects of personality (the Big 5; McCrae & Costa, 1987) and other affective traits (affect, intensity, and empathy) were statistically controlled using regression analysis. They also reported that the TAS score was positively correlated not only with the total score of GEMS but also with its subfactors. Using these data, Lyvers et al. (2020) discussed that the emotional impact of music on people with alexithymia might be comparable to the effects of alcohol or drugs on patients with substance abuse. They specifically speculated that music could help people with alexithymia amplify their affective experiences that are usually poor or support releasing emotions that are usually inhibited in those with alexithymia.
The primary aim of this study was to reexamine the study by Lyvers et al. (2020) to confirm its generalizability. This study was assumed to be a conceptual, rather than direct, replication of Lyvers et al. (2020), as several aspects of the former were different from the latter. First, this study examined the data from two Asian countries: China and Japan. This would complement the previous report, which primarily examined data from Europe (mostly Caucasians), with different ethnicities and cultures. Furthermore, separately examining two different countries (China and Japan) is more beneficial to investigate the generality of knowledge than examining only one country.
Second, the present study examined the subfactors of TAS-20, as well as its total score. Alexithymia has been treated as a multifactor construct since it was first introduced (Nemiah et al., 1976; Taylor et al., 1991). TAS-20, which is one of the most frequently used measurement, comprises three subfactors: difficulty in identifying emotions (TAS-DIF), difficulty in describing feeling (TAS-DDF), and externally oriented thinking (TAS-EOT). Various studies have reported the different characteristics of these subfactors. For example, it has been indicated that TAS-DIF is associated with neuroticism, whereas TAS-EOT is related to extraversion (De Gucht et al., 2004). Other studies have shown positive correlations between anxiety disorder and alexithymia, but only for subfactors DIF and DDF, not EOT (Dalbudak et al., 2013; Devine et al., 1999). Lyvers et al. (2020) only examined the total score of TAS-20. However, considering these differences between the subfactors, it is necessary to analyze these three factors separately to further clarify whether and how each aspect of alexithymia is related to ERM.
Third, this study additionally measured the participants’ trait anxiety. Many previous studies have indicated a positive correlation between alexithymia and persistent negative trait, such as anxiety and depression (Berthoz et al., 1999; Espina Eizaguirre et al., 2004; Franzoi et al., 2020; Li et al., 2015). Therefore, an association between alexithymia and intensified ERM, if any, can be a spurious correlation mediated by the traits of negative emotion. To examine this possibility, this study explicitly examined the anxiety trait as a control variable to clarify the relationship between ERM and alexithymia.
Finally, this study also examined the participants’ daily experiences of listening to music, specifically, the frequency and subjective absorption (concentration) of music while listening. Despite being measured, these variables were not examined by Lyvers et al. (2020). We, thus, explicitly analyzed these items to examine the interpretation by Lyvers et al. (2020), who suggested that music functions as an intoxicant for people with alexithymia. Based on their substance abuse hypothesis regarding the alexithymia–music relationship, it was predicted that people with higher alexithymia would “use the substance” (i.e., listening to music) more frequently with higher subjective absorption, compared to those with lower alexithymia. The present study examined this hypothesis by including these variables (i.e., the frequency, and subjective absorption while listening to music) into its analyses.
In sum, the primary purpose of this study was to examine the relationship between alexithymia and ERM in two sets of the Asian population. TAS-20 and GEMS were used to measure alexithymia and ERM, respectively, by analyzing the subfactors of each scale. In addition, trait anxiety, frequency of listening to music, and the degree of subjective absorption while listening to music in daily life, as well as the demographic data, were measured and entered in the analyses as covariates to further clarify the nature of the possible link between ERM and alexithymia.
Method
Participants
A total of 355 participants from China were recruited using the online survey site Wenjuanxing. Of these, 11 were excluded, as four were minors, five responded with identical choices to all the items, one confessed a fallacious response, and one (whose scores of TAS-20 and State-Trait Anxiety Inventory [STAI] showed outlier values higher than 2 SD plus mean) was excluded as the participant was much older than the rest (80 years). The remaining 344 participants (154 men and 190 women, average age: 29.17 ± 7.62 years, range: 18–58 years) were analyzed.
A total of 343 participants from Japan were recruited using the crowdsourcing site CrowdWorks and asked to answer our Google Forms survey. Two participants, who chose identical responses for every questionnaire item, were excluded. The remaining 341 participants (100 males and 241 females, average age: 36.28 ± 10.53 years, range: 19–68 years) were analyzed. Demographic information of the analyzed sample is presented in Table 1.
Sample Characteristics (Chinese Sample, n = 344, Japanese Sample, n = 341).
This study was approved by the local ethics committee of Kansai University (Graduate School of Psychology #122).
Procedure
Participants responded to the questionnaire online that included fill in the blanks and multiple-choice questions. This questionnaire consisted of demographics, STAI, TAS-20, and GEMS-45 in that order (detailed descriptions of each scale are provided in the “Materials” section). At the beginning of the survey, all participants were informed that their answers are anonymous, and they have the right to withdraw from the survey at any time. The participants then provided their informed consent on participation. The questionnaire comprised 98 items in total, which had to be completed in one sitting. Participants were limited to single-use access to the survey.
Materials
Demographics
Age, gender, nationality, education, preferred music genre, frequency of listening to music, and the degree of subjective absorption while listening to music in daily life were asked at the beginning of the survey (Table 1). The preferred genre of music was categorized following the study by Lyvers et al. (2020). For “the degree of subjective absorption while listening to music,” participants were asked to answer the approximate percentage of time they listen with concentration and immersion in their entire music-listening time.
Toronto Alexithymia Scale-20
The TAS (Bagby et al., 1994a, 1994b) is a 20-item questionnaire to measure alexithymia characteristics. It consists of three subfactors. The first is difficulty in identifying feelings (DIF), which contains seven items regarding the decreased capacity of identifying emotions and distinguishing feelings between physical sensation and emotional responses, for example, “I am often confused about what emotion I am feeling.” The second is difficulty in describing feelings (DDF), which contains five items concerning the deficient ability of communicating one’s feelings with others, for example, “I am able to describe my feelings easily.” The third is externally oriented thinking (EOT), which contains eight items regarding the tendency of individuals to focus their attention externally, for example, “I prefer to analyze problems rather than just describe them.” Each item was rated using a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The sum of TAS-DIF, TAS-DDF, and TAS-EOT scores was considered as the total score. For this study, Chinese (Yi et al., 2003) and Japanese (Komaki et al., 2003) versions of TAS-20 were used.
Geneva Emotional Music Scale
The GEMS has been specifically devised to measure the emotions induced by music. The present study used this scale to retrospectively measure participants’ emotional responses to their favorite music in their daily lives. The GEMS-45 contains 45 labels (e.g., moved, sad, relaxed, happy) for describing musically evoked emotive states across a relatively wide range of music (Zentner et al., 2008, 2009). This scale rates the extent to which music makes participants feel each of the 45 emotional states. It was scored using a 5-point Likert-type scale ranging from 1 (not at all) to 5 (very much). These emotional labels are divided into nine categories and aggregated into three subfactors by summing the item scores of three categories. The first category is Sublimity, which is the convergence of elating and near-paradisiac factors (wonder, transcendence, tenderness, nostalgia, and peacefulness). The second is Vitality, which is the combination of energetic factors (power and joyful activation). The third is Unease, which is the combination of “negative” factors (tension and sadness).
State-Trait Anxiety Inventory
The STAI (Japanese, Hidano et al., 2000; Spielberger, 1983; Chinese, Zheng et al., 1993) consists of 40 items, half of which assess the state anxiety while the remaining assess trait anxiety. The present study used 20 items for measuring trait anxiety only. These items include statements such as “I make decisions easily” and “I lack self-confidence.” All items were rated on a 4-point scale (ranging from almost never to almost always).
Results
Intercorrelations among TAS and GEMS
Pearson’s correlations between the continuous variables were separately examined for the Chinese and Japanese data. In the Chinese data, several combinations of TAS and GEMS factors exhibited significant positive correlations (Table 2). The TAS-Total and TAS-DIF were positively correlated with GEMS-Total (TAS-Total: r = .178, p < .001; TAS-DIF: r = .175, p = .001) and GEMS-Unease (TAS-Total: r = .355, p < .001; TAS-DIF: r = .341, p < .001). The TAS-DDF was positively correlated with GEMS-Total (r = .166, p = .002), GEMS-Sublimity (r = .129, p = .017), and GEMS-Unease (r = .291, p < .001). The TAS-EOT was correlated with GEMS-Unease (r = .204, p < .001).
Pearson’s Correlation Coefficients Among Variables (Chinese Sample, n = 344).
Note. Frequency = frequency of listening to music; Absorption = the degree of subjective absorption while listening to music; TAS = Toronto Alexithymia Scale-20; TAS-Total / DIF / DDF / EOT = total score / difficulty in identifying feelings / difficulty in describing feelings / externally oriented thinking of TAS-20; STAI-T = Trait anxiety; GEMS = Geneva Emotional Music Scale. Correlation coefficients were marked with a “△” when the p values become nonsignificant (⩾.05) post adjustment by false discovery rate (FDR).
p < .001. **p < .01. *p < .05.
In the Japanese data (Table 3), correlations between TAS and GEMS factors were mainly exhibited by TAS-DIF and GEMS-Unease. Although TAS-Total did not show a significant correlation with GEMS-Total, it was positively correlated with GEMS-Unease (r = .241, p < .001). Furthermore, TAS-DIF showed significant positive correlations with GEMS-Total (r = .191, p < .001), GEMS-Sublimity (r = .186, p < .001), and GEMS-Unease (r = .311, p < .001). Interestingly, TAS-EOT showed negative correlations with GEMS-Total (r = -.165, p = .002), GEMS-Sublimity (r = -.163, p = .003), and GEMS-Vitality (r = -.196, p < .001) in the Japanese data.
Intercorrelations of Variables (Japanese Sample, n = 341).
Note. Frequency = frequency of listening to music; Absorption = the degree of subjective absorption while listening to music; TAS = Toronto Alexithymia Scale-20; TAS-Total / DIF / DDF / EOT = total score / difficulty in identifying feelings / difficulty in describing feelings / externally oriented thinking of TAS-20; STAI-T = Trait anxiety; GEMS = Geneva Emotional Music Scale. Correlation coefficients were marked with a “△” when the p values become nonsignificant (⩾.05) post adjustment by false discovery rate (FDR).
p < .001. **p < .01. *p < .05.
Simultaneous multiple regressions on GEMS
To investigate the contributions of other variables to ERM, GEMS and its three subfactors were set as objective variables separately in simultaneous multiple regression analyses (Tables 4–7). In each analysis, age, gender, frequency of listening to music, degree of subjective absorption while listening to music, STAI-T, and TAS-20 subfactors (TAS-DIF, TAS-DDF, and TAS-EOT) were entered as predictors (all variables were standardized). For both China and Japan, TAS-DIF predicted most of the objective variables, that is, GEMS-Total (Figure 1), GEMS-Sublimity, and GEMS-Unease (ps < .016). In addition, TAS-EOT showed negative predictions for GEMS-Total, GEMS-Sublimity, and GEMS-Vitality only in the Japanese data (Tables 4–6).

Partial Regression Plots Showing That GEMS Total Scores Were Positively Associated with TAS-DIF in China (a) and Japan (b) (ps < .004). TAS = Toronto Alexithymia Scale-20; DIF = difficulty in identifying feelings; GEMS = Geneva Emotional Music Scale.
Simultaneous Multiple Regression Predicting GEMS-Total.
TAS = Toronto Alexithymia Scale-20; TAS-Total / DIF / DDF / EOT = total score / difficulty in identifying feelings / difficulty in describing feelings / externally oriented thinking of TAS-20; STAI-T = Trait anxiety.
R2 = .136; Adjusted R2 = .116.
R2 = .181; Adjusted R2 = .161.
= standardized partial regression coefficient. △ = not significant (⩾.05) after false discovery rate correction.
p < .001. **p < .01. *p < .05.
Simultaneous Multiple Regression Predicting GEMS-Sublimity.
TAS = Toronto Alexithymia Scale-20; TAS-Total / DIF / DDF / EOT = total score / difficulty in identifying feelings / difficulty in describing feelings / externally oriented thinking of TAS-20; STAI-T = Trait anxiety.
R2 = .13; Adjusted R2 = .11.
R2 = .173; Adjusted R2 = .153.
= standardized partial regression coefficient. △ = not significant (⩾.05) after false discovery rate correction.
p < .001. **p < .01. *p < .05.
Simultaneous Multiple Regression Predicting GEMS-Vitality.
TAS = Toronto Alexithymia Scale-20; TAS-Total / DIF / DDF / EOT = total score / difficulty in identifying feelings / difficulty in describing feelings / externally oriented thinking of TAS-20; STAI-T = Trait anxiety.
R2 = .16; Adjusted R2 = .14.
R2 = .131; Adjusted R2 = .109.
= standardized partial regression coefficient. △ = not significant (⩾.05) after false discovery rate correction.
p < .001. ** p < .01. * p < .05.
Simultaneous Multiple Regression Predicting GEMS-Unease.
TAS = Toronto Alexithymia Scale-20; TAS-Total / DIF / DDF / EOT = total score / difficulty in identifying feelings / difficulty in describing feelings / externally oriented thinking of TAS-20; STAI-T = Trait anxiety.
R2 = .221; Adjusted R2 = .203.
R2 = .167; Adjusted R2 = .146.
= standardized partial regression coefficient. △ = not significant (⩾.05) after false discovery rate correction.
p < .001. ** p < .01. * p < .05.
Other variables
Data from both China and Japan showed significant positive correlations between the frequency of listening to music and GEMS-Total as well as its subfactors (Tables 2 and 3, ps < .015). The simultaneous multiple regression analyses also showed that the frequency had a significant positive effect on all GEMS variables in data from both countries (Tables 4–7). Regarding the degree of subjective absorption while listening to music, it was also associated with some GEMS variables. In the Chinese data, subjective absorption was positively correlated with GEMS-Vitality. In the Japanese data, however, it was correlated with GEMS-Total and its subfactors (Table 3), and positively predicted each objective variable in the regression analyses (Tables 4–7, right columns). In addition, an important finding of our study was that the frequency and the absorption did not show any significant correlation with alexithymia or its subfactors for both countries (ps > .053).
The STAI-T showed positive correlations with TAS-Total and TAS-DIF in both countries (Tables 2 and 3). Concerning ERM, STAI-T was a positive predictor of GEMS-Unease in both the Chinese and Japanese data (Table 7). In addition, STAI-T was a significant negative predictor of GEMS-Sublimity and GEMS-Vitality in the Chinese data (Tables 4–6, left columns).
Regarding the gender difference in ERM, the regression analyses (Tables 4–7) showed that males tended to exhibit higher ERM scores than females. This was the case when GEMS-Total and GEMS-Unease were the objective variables in the Chinese data and when GEMS-Sublimity was the objective variable in the Japanese data.
Discussion
The current study examined the relationship between alexithymia and the magnitude of ERM in the Chinese and Japanese populations. A previous study by Lyvers et al. (2020) reported a positive correlation between the scores of alexithymia and the degree of ERM. Their results indicated this association even when several traits (the Big 5 personalities, affect intensity, and empathy) and demographic data (age and gender) were controlled using multiple regression analysis (Lyvers et al., 2020). The data of the current study were consistent with that of the previous study. Simple correlation analyses showed a positive correlation between TAS-Total and GEMS-Total in the Chinese population, which was consistent with the results of the previous study. This correlation between TAS-Total and GEMS-Total was not evident in the Japanese data. However, the subscale of TAS, namely, difficulty in identifying feelings (TAS-DIF), showed significant correlations with the GEMS-Total score in both the Chinese and Japanese data. This association between TAS-DIF and GEMS-Total was also observed for both countries in the multiple regression analyses, which was controlled for other variables (i.e., TAS-DDF, TAS-EOT, trait anxiety [STAI-T], and demographic data [age and gender]). Moreover, the regression analyses revealed that TAS-DIF had a significant positive effect on most of GEMS’s factors (except for GEMS-Vitality). As alexithymia is primarily a deficit in understanding emotions, difficulty in identifying emotions (which is reflected by the DIF score) is considered an essential aspect of alexithymia. In this sense, the present study indicated that the degree of ERM is positively associated with the core aspect of the symptoms of alexithymia.
Concerning the subfactors of GEMS, Lyvers et al. (2020) reported that the TAS total score was positively correlated with the subfactors of GEMS: GEMS-Unease (consisting of negative emotions) and GEMS-Sublimity (elating emotions), with a larger effect for the former, but not with GEMS-Vitality (joyful activation and power). The present study replicated these correlations. It further examined the subfactors of TAS, which were not reported in the aforementioned study. In our data, TAS-DIF was significantly associated with GEMS-Total and GEMS-Unease but also with GEMS-Sublimity. The GEMS-Unease is considered to reflect the negative emotions elicited by music, as it is calculated by summing tension- and sadness-related items. This association between alexithymia and negative emotions (i.e., unease) is understandable as alexithymia is often concurrent with negative affects such as anxiety, depression, and stress disorder (Berthoz et al., 1999; Li et al., 2015; Marchesi et al., 2000). In fact, the present data showed correlations between TAS scores and trait anxiety (STAI-T). However, we consider that the alexithymia–ERM linkage is not a kind of pseudo-correlation mediated by anxiety. This is because the STAI-T did not show a correlation with the GEMS score, and the multiple regression analyses showed a significant association between alexithymia and ERM although STAI-T was included in the predictive variables. GEMS-Unease corresponds to negative emotions, whereas GEMS-Sublimity (consisting of wonder, transcendence, tenderness, nostalgia, and peacefulness) represents a kind of mixed-valence emotion including a positive affect. Therefore, we considered that the contents of larger emotional responses to music in individuals with alexithymia are not restricted to negative valence. Thus, the present data indicate that DIF in alexithymia is positively correlated with the magnitude of ERM. This association may not be limited to the negative valence in ERM. We noted that these results replicate that of Lyvers et al. (2020) to some extent.
As mentioned in this study’s introduction, a few studies have examined the music perception of alexithymia (Allen et al., 2013; Larwood et al., 2021; Punkanen et al., 2011; Taruffi et al., 2017). They showed that higher alexithymia is correlated with a lower ability to recognize and verbalize the emotions expressed by music (Allen et al., 2013; Punkanen et al., 2011; Taruffi et al., 2017) or that judgment of the valence and arousal of musical emotion are more neutral in negative musical conditions (Larwood et al., 2021). All these studies examined the perceived emotions (i.e., emotions recognized in the music). On the other hand, our study as well as Lyvers et al. (2020) examined ERM, which corresponds to felt or induced emotions (i.e., emotions experienced in listeners). Against this background, the current study and Lyvers et al. (2020) demonstrated the opposite association between alexithymia and induced music experiences—the positive correlation between alexithymia and ERM.
Regarding the interpretation of the positive association between alexithymia and ERM, Lyvers et al. (2020) suggested a hypothesis. In advance of their study, it was suggested that alcohol drinking in alcohol-dependent patients with alexithymia is mediated by the expectation of intensified emotion due to drinking (Thorberg et al., 2016). Based on this knowledge, Lyvers and colleagues (2020) argued that the link between alexithymia and ERM could be interpreted using an analogy of substance abuse. They considered that people with alexithymia could release or disinhibit their affective experience, which is usually poorly noticed, with the help of the auditory “substance,” namely, music. However, the current data do not seem to support the substance use model of the ERM–alexithymia association. The substance use view implies that people with alexithymia are likely to be addicted to music. Therefore, it is predicted that such individuals would have a higher frequency of listening to music, and may experience a stronger feeling of absorption while listening to music in their daily lives. The current study explicitly examined these variables. The analyses showed that although the frequency of and the degree of subjective absorption in listening to music were correlated with the magnitude of ERM, these measures were not correlated with any of the TAS factors in either country. Therefore, the results of our study do not support the proposition that music works as a “drug” for individuals with alexithymia.
An alternative interpretation of the association between musical emotions and alexithymia could be the role of nonverbal features of music. A multinational study pointed out that people can understand the musical emotions of various cultures without relying on lyrics (Balkwill et al., 2004). Another study that measured physiological activities reported that participants experience more positive emotions when listening to happy music without as opposed to with lyrics (Brattico et al., 2011). Various studies, including the above-mentioned ones, indicate that music’s influence on people’s emotions does not necessarily depend on verbal information. People with alexithymia find it difficult to understand and express emotions in words (the meaning of the Greek-rooted word “A-lexi-thymia” corresponds to “no words for affects” [Sifneos, 1973]). Thus, unlike other types of emotional stimuli, music may be able to elicit more intense emotional responses in individuals with alexithymia due to its nonverbal character.
Somewhat related to the matter of cultural difference, although significant consistencies between the data of China and Japan (reported in the current study) and those of Europe (reported in Lyvers et al., 2020) have been found, there are minor differences among the datasets: the positive correlation between TAS-Total and GEMS-Total was shown in the European and Chinese data, but not in the Japanese data. (Instead of TAS-Total, TAS-DIF was correlated with GEMS in Japan and China.) We consider that this pattern in Japan is due to the subfactor TAS-EOT, which was negatively associated with GEMS in the Japanese data. This negative association between TAS-EOT and GEMS was opposite to (and canceled out) the positive correlation of TAS-DIF with GEMS; therefore, the TAS-Total score showed no correlation with GEMS in the Japanese data, whereas in the Chinese data, TAS-EOT showed just null correlations with GEMS, remaining the significant correlation between TAS-Total and GEMS. Lyvers et al. (2020) only reported the total score of TAS (i.e., they did not report the scores of the subfactors); however, we speculate that the European data exhibited a pattern that would be akin to the Chinese data of our study. As mentioned in the introduction, when studying the relationship between alexithymia and other variables, EOT often showed characteristics different from DIF/DDF (Dalbudak et al., 2013; De Gucht et al., 2004; Devine et al., 1999), and its reliability was also relatively low (Kooiman et al., 2002; Meganck et al., 2008). Therefore, we considered that DIF represents the core feature of alexithymia, and thus, the data of Lyvers et al. (2020) (Europe) and the present study (China and Japan) are essentially consistent.
The data of the current study suggested a gender difference in ERM as males tended to show higher GEMS scores. Males had significantly higher GEMS scores (GEMS-Total, GEMS-Vitality, and GEMS-Unease) in China and higher GEMS-Sublimity score in Japan. Interestingly, higher GEMS scores for males were also reported by Lyvers et al. (2020). We should note that findings on gender differences in ERM have been mixed: Some previous studies on ERM measured using GEMS did not show significant gender differences (Lundqvist et al., 2009; Robazza et al., 1994; Song et al., 2016), while some showed opposite differences, that is, higher GEMS in females (Aljanaki et al., 2016). In future studies on alexithymia and musical emotions, more attention should be paid to gender differences.
The current study has limitations inherent to cross-sectional questionnaire research. In this study, the causality between alexithymia and ERM seems unclear. Retrospective self-reports provide restricted information about implicit and/or physiological emotional responses to music. A possible approach to further exploring the effects of alexithymia on individuals’ ERM would be to measure neural or physiological responses while listening to music by explicitly manipulating the music stimuli. For example, measuring the activities of the reward system in the brain when listening to music (e.g., Salimpoor et al., 2011) would be beneficial to further examine the “substance use” model of Lyvers et al. (2020). Our study alone cannot conclude the validity of the model.
Furthermore, another limitation of this study is represented by the assessment of alexithymia using TAS-20. For example, some researchers consider a distinction of alexithymia; type I alexithymia corresponds to reduced ability of emotional arousal and cognitively recognizing emotions, and type II alexithymia corresponds to normal or high ability of emotional arousal and reduced ability to cognitively recognize emotions (Bermond, 1997). The TAS-20 is considered to assess the cognitive factor of alexithymia (or type II alexithymia; Larsen et al., 2003), or it cannot distinguish between type I and II alexithymia (Vorst & Bermond, 2001). However, other tools like Bermond–Vorst Alexithymia Questionnaire (Vorst & Bermond, 2001) are designed to assess both aspects of alexithymia. Thus, future studies can benefit by using such tools of assessing alexithymia to examine its association with the music-evoked emotion more comprehensively.
Finally, the effect of participants’ state anxiety on the quality of their responses cannot be ignored, although not measured in either of our study or that of Lyvers et al. (2020). Future studies that additionally measure participants’ affective states will help to make the current findings more rigorous and reliable.
Conclusion
In conclusion, this study supported the notion that people with alexithymia experience more intense ERM, as suggested by Lyvers et al. (2020). Furthermore, the present data indicated that the DIF of alexithymia plays a central role in the association between alexithymia and ERM. We speculated that music could cause intense emotions in people with alexithymia due to the more nonverbal characteristics. Future research could investigate the potential of music in assisting individuals with alexithymia in modulating their emotional capacity (for an example of such an approach, see Allen & Heaton, 2010).
Footnotes
Acknowledgements
We wish to thank Prof. Marcel Zentner for providing us with the original version of Geneva Emotional Music Scale and authorizing us to translate and use it.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by JSPS KAKENHI Grant (#19H01766).
