Development of a Japanese version of the Awe Experience Scale (AWE-S): A structural topic modeling approach

Introduction

Awe is an emotional response to perceptually and conceptually vast stimuli that transcend one’s current frames of reference (Keltner & Haidt, 2003). It has been studied in various fields, such as psychology, economics, neuroscience, immunology, and public health (Guan et al., 2018; Ibanez et al., 2017; Monroy & Keltner, 2022; Piff et al., 2015; Sawada & Nomura, 2020; Stellar et al., 2015; Sun et al., 2021; Takano & Nomura, 2022b; van Elk et al., 2019). Previous studies have shown that awe is associated with subjective psychological phenomena, such as a sense of self and consciousness, which highlights the need for a standardized questionnaire to assess the various dimensions of awe experiences (Bai et al., 2017; Rudd et al., 2012; Takano & Nomura, 2021). The Awe Experience Scale (AWE-S), a widely used questionnaire that robustly measures the state of awe (Yaden et al., 2019), has been translated into several languages (Landrum et al., 2022; Ling-xiao et al., 2022; Rodrigues et al., 2022; van Houwelingen-Snippe et al., 2020). However, how the multiple facets of awe measured by the AWE-S relate to the representations of awe experiences in different cultural contexts remains unclear. Simultaneously, there is no reliable and valid standardized Japanese version of the AWE-S. Therefore, it is important to develop a standardized Japanese AWE-S and investigate its narrative dimensions not only for a deeper understanding of the complex emotion of awe and revealing the construct of awe in Japan through cross-cultural comparisons, but also for investigating the psychological mechanisms underlying the various effects of awe on mental health and social behavior (Bai et al., 2021; Liu et al., 2023; Piff et al., 2015; Stellar et al., 2018; Sturm et al., 2022; Takano & Nomura, 2021).

The AWE-S questionnaire comprises six factors: time, self-loss, connectedness, vastness, physiological, and accommodation (Yaden et al., 2019). Time captures the perception of time expansion in awe experiences, which may indicate a shift in mental state towards a further mindful, moment-to-moment awareness, such as here and now (Gregory et al., 2023; Kramer et al., 2013; Rudd et al., 2012; Yaden et al., 2019). Self-loss reflects the phenomenon where awe can diminish one’s sense of self, leading to feelings of insignificance (Bai et al., 2017; Piff et al., 2015; Takano & Nomura, 2021). Connectedness indicates that awe experiences can create a greater sense of connection between individuals and their surroundings, which includes the supernatural, such as God(s) (Liu et al., 2023; Van Cappellen & Saroglou, 2012). Vastness pertains to the perception of vastness vis-à-vis the self, where individuals encounter physically or conceptually enormous stimuli (Keltner & Haidt, 2003; Landrum et al., 2022). Physiological captures awe-related physiological phenomena, such as goosebumps (Quesnel & Riecke, 2018; Williams et al., 2022). Accommodation reflects the process of needing to adjust one’s understanding or cognitive framework to comprehend new information or experiences, such as those often experienced during awe-inspiring situations (Keltner & Haidt, 2003; Taylor & Uchida, 2019; Valdesolo et al., 2016). Yaden et al. (2019) showed that the AWE-S, measured after one recalled and described their experience of awe, was positively associated with trait measures of tendency to experience awe, which was salient for the factors of time, connectedness, vastness, and physiological. This suggested that the state measure of awe had enough convergent and divergent validity.

Due to its robustness and usefulness, the AWE-S has been validated in several countries and cultures via translation into other languages, such as Chinese, Portuguese, and Dutch (Landrum et al., 2022; Li & Qian, 2022; Ling-xiao et al., 2022; Rodrigues et al., 2022; van Houwelingen-Snippe et al., 2020). Universal components and cultural variations of awe-related phenomena have been identified via this scale. However, despite the creation and development of other scales to measure awe in Japan, the AWE-S has not yet been translated into the Japanese language (Nakayama et al., 2020; Sawada & Nomura, 2023). To facilitate further comprehensive cross-cultural comparisons, development of a Japanese version of the AWE-S, which can provide a deeper understanding of awe, is necessary.

Previous studies suggested that cultural similarities and differences in awe experiences were important perspectives to understand its generality and heterogeneity (Bai et al., 2017; Nakayama et al., 2020; Nakayama & Uchida, 2020; Razavi et al., 2016; Sawada & Nomura, 2023). Specifically, it was necessary to consider how awe altered the sense of self and differed across cultures, although the small self-effect was a general and robust phenomenon (Bai et al., 2017; Nomura et al., 2021; Takano & Nomura, 2021). Awe could manifest differently among individuals from East Asian cultures, such as Japan, compared to those from Western cultures according to their culturally, historically, and religiously cultivated views of the self. Particularly, the self is represented as being non-substantial, denial, and embedded as default in Japan, a country with Buddhist and Shinto backgrounds. Meanwhile, it is represented as separated and continuous concepts by others in Western cultures (Nomura et al., 2021; Sugimoto, 2020; Takano & Nomura, 2019, 2022a). Similarly, Markus and Kitayama (1991) advocated independent and interdependent self-construals. Independent self-construals, more widespread in North American and Western European cultures, prioritize an individual’s needs, whereas interdependent self-construals, more common in Asian, Latin American, and African cultures, prioritize one’s social interactions. Thus, a Japanese AWE-S could reveal culture-specific patterns, especially regarding sense of self.

From the perspective of state awe, it is important to address how people narrate their awe experiences in their cultural context for a deeper understanding (Nakayama & Uchida, 2020). Since the AWE-S is measured after participants recall and describe (i.e., write) their awe experiences (Yaden et al., 2019), their contents may have a role manifesting the responses to the AWE-S. Studies revealed that ratings on the AWE-S varied based on the elicitors of awe experiences (Graziosi & Yaden, 2021; Yaden et al., 2019). However, the aspects of narrated experience (i.e., the “potential topics”) associated with each factor of the AWE-S have not been investigated beyond the typology of awe experience. For example, looking up at the night sky or touching a grand theory are different types of experiences. However, both could be accompanied with common potential topic of “universe.” Similarly, different awe experiences narrated through the same elicitor could involve distinct potential topics, which could lead to the AWE-S being rated differently.

This study aimed to develop a standardized Japanese AWE-S as well as investigate its relationships with linguistic generative processes underlying the expression of awe experiences. First, we created a Japanese version of the AWE-S via back-translation and confirmed its factor structure and test-retest reliability compared to the original AWE-S. Furthermore, its convergent and divergent validities were examined regarding its relation to the trait-awe (dispositional awe) questionnaire. We used the Dispositional Positive Emotion Scale (DPES) since previous studies demonstrated that awe-related phenomena observed via the DPES could also be seen through the measurement of state-awe (Bai et al., 2017; Piff et al., 2015; Stellar et al., 2018). We controlled other positive emotions to examine specific relationships between dispositional awe and each factor of the Japanese AWE-S. Second, we used a structural topic model (STM) and exploratorily examined whether and how each factor of AWE-S was associated with topics that underlie the narrative of awe experiences (Roberts et al., 2014). Using this model, we can focus not only on the type of experience, but also on the generative process of potential topics underlying how people feel in that experience and express it in natural language.

Methods

Results

A total of 452 participants (222 males, 226 females, and four others, M_age = 42.52 years, SD = 9.69, Range: 20–79 years) were included in the factor analysis, t-tests, and the STM for the Japanese AWE-S (Table 1). Of these, 91 participants (42 males, 48 females, and one other, M_age = 43.37 years, SD = 9.86, Range: 24–71 years) were included in the intra-class correlation analyses to examine the test-retest reliability. In addition, 315 participants (156 males, 156 females, and three others, M_age = 42.64 years, SD = 10.08, Range: 20–79 years) were included in the factor analysis for the DPES, correlation analyses, and multiple regression analyses. The full dataset can be found under Underlying data (Takano & Matsuo, 2023b).

Confirmatory factor analyses were conducted to examine the factor structure of the Japanese AWE-S (Table 2) and DPES. The results showed that the six-factor model provided an adequate fit to the data for the Japanese AWE-S, chi-squared (χ²) (390) = 1070.31, Comparative Fit Index (CFI) = .905, Goodness Fit Index (GFI) = .854, Root Mean Square Error of Approximation (RMSEA) = .062 [90% lower = .058], and Standardized Root Mean Square Residual (SRMR) = 0.068. In addition, the seven-factor model of the DPES fitted well as in previous studies, χ² (644) = 1728.55, CFI = .860, GFI = .763, RMSEA = .073 [90% lower = .069], and SRMR = .064 (Nomura et al., 2021; Shiota et al., 2006; Sugawara et al., 2020). To examine the internal consistency of the Japanese AWE-S, Cronbach’s alpha coefficients were calculated for each subscale (Table 2). Sufficient internal consistency was confirmed, which was consistent with the results of previous studies (Ling-xiao et al., 2022; Yaden et al., 2019). In addition, the intraclass correlations analyses for each subscale of AWE-S showed sufficient and good test-retest reliability (ρs > .60, p < .001, Table 2).

There were differences in the means of some factors of the AWE-S between the Japanese and original versions, as shown in Table 2. Independent t-tests indicated that the mean scores for the self-loss and vastness factors were higher in the Japanese version than in the original. In contrast, those for the connectedness, physiological and accommodation factors were lower in the Japanese version than in the original (self-loss: t (1086) = 4.90, p < .001, vastness: t (1086) = 4.97, p < .001, connectedness: t (1086) = -11.66, p < .001, and physiological: t (1086) = -2.20, p = .028, accommodation: t (1086) = -5.99, p < .001).

Correlation analyses between the subscales of the Japanese AWE-S and DPES were conducted to examine construct validity. Time, connectedness, vastness, and physiological factors were significantly positively correlated with awe (r = .12–.36, p < .038) and other factors of the DPES; however, self-loss and accommodation were not (Table 3). In addition, while the inter-factor correlations coefficients of the Japanese AWE-S ranged from moderate to large, those of the DPES were large (AWE-S: r = .26–.47, p < .001, DPES: r = .44–.83, p < .001).

Linear regression analyses with a robust estimation method showed that the time and connectedness factors were significantly positively associated with awe factor of the DPES, controlled for other positive emotions (β = .21–.23, p < .013). This indicated that these factors of the AWE-S were specifically associated with dispositional awe. These results did not change when controlled for participants’ age and gender.

Furthermore, seven topics were generated by the STM and labeled: “Spirituality,” “Threat,” “Spatiality,” “Universe,” “Scenery,” “Humanity,” and “Aesthetics” (see Figure 1). The top 20 highest probability words are listed in Table 4. In particular, “Scenery” and “Threat” were the two most common topics that emerged in all the responses. Regarding the Japanese AWE-S, participants with higher scores on the time factor were more likely to use words related to “Spatiality” and “Scenery” (Figure 1A). In contrast, they were less likely to use words of “Humanity.” Similarly, positive associations were found between the connectedness factor and “Spirituality” and “Aesthetics” (Figure 1C), vastness factor and “Universe” and “Scenery” (Figure 1D), physiological factor and “Threat” and “Scenery” (Figure 1E), and accommodation factor and “Spirituality” and “Threat” (Figure 1F). In addition, negative associations were found between the connectedness factor and “Threat” and “Scenery” (Figure 1C), vastness factor and “Spirituality,” “Threat,” and “Humanity” (Figure 1D), physiological factor and “Universe” (Figure 1E), and accommodation factor and “Scenery” and “Aesthetics” (Figure 1F). Self-loss was not significantly associated with any topic (Figure 1B).

Figure 1. Parameter estimate of the effect of each factor of the Japanese Awe Experience Scale on the seven topics.

Note. Effects computed via a structural topic model. Parameter estimates are shown as dots. Expected topic proportions, proportion of the total documents (responses) falling into each topic, are depicted by dot size. Error bars represent 95% confidence interval.

Discussion

This study aimed to develop a Japanese version of the AWE-S, a robust questionnaire that measured the subjective state in awe experiences, and investigate its relations to the representations of awe experiences in natural language. We examined the validity of the Japanese AWE-S and explored how each factor of the AWE-S was related to topics that underlie the narrative of awe experiences via a structural topic model. The results showed that the six-factor model of the original AWE-S sufficiently fit the data for the Japanese AWE-S. Furthermore, the mean scores for the self-loss and vastness factors were higher in the Japanese version than in the original. In contrast, the connectedness, physiological and accommodation factors were lower in the Japanese version. In addition, the time, connectedness, vastness, and physiological factors were positively associated with trait-awe (dispositional awe). The relationships with the time and connectedness factors remained constant, even when controlled for other positive emotions, age, and gender. The STM generated seven potential topics, which were differently associated with each factor of the Japanese AWE-S.

The confirmatory factor analysis revealed that the Japanese AWE-S conformed to the six-factor model of the original (Yaden et al., 2019). In addition, Cronbach’s alpha coefficients and intraclass correlations showed that this scale had sufficient internal consistency and test-retest reliability, respectively. Therefore, our results suggest that the structure and reliability of the Japanese AWE-S are comparable to those of the original version, which rendered it suitable for international comparisons.

The time, connectedness, vastness, and physiological factors of the Japanese AWE-S were positively associated with trait measures of tendency to experience awe (i.e., dispositional awe). These results were consistent with Yaden et al.’s results (2019), which suggested that the Japanese version had similar construct validity to the original. Furthermore, we used linear regression analyses and revealed that the time and connectedness factors were specifically related to trait-awe, controlled for other positive emotions. Previous studies suggested that the dispositional awe questionnaire mainly focused on the positivity, beauty, and mindful aspects of awe (example: “I see beauty all around me” (Shiota et al., 2006). Therefore, our results suggested that the AWE-S was a state scale that could capture the other aspects of awe.

Regarding the culturally specific perspectives, the mean scores for the self-loss and vastness factors were higher, while those for the connectedness, physiological and accommodation factors were lower in the Japanese version than in the original. Previous research suggested that the sense of the self was mainly characterized by being non-substantial, denied, and embedded as default in Japan with its Buddhist and Shinto backgrounds, while it is represented as separated and continuous concepts by others in Western cultures (Nomura et al., 2021; Sugimoto, 2020; Takano & Nomura, 2021, 2022a). Consistent with this perspective, Takano and Nomura (2021) revealed that experimentally induced awe reduced one’s sense of self-size and also blurred the sense of boundary of the self. Hence, Japanese people might be more and less likely to respond to self-reductive (i.e., self-loss and vastness) and self-expansive aspects during awe experiences (i.e., connectedness, physiological, and accommodation), respectively.

Regarding the narrative aspect of awe, the STM revealed that each factor of the AWE-S was related to seven topics of descriptions of awe experiences. There were positive (negative) associations between the time factor and “Scenery” and “Spatiality” (“Humanity”), connectedness factor and “Spirituality” and “Aesthetics” (“Threat” and “Scenery”), vastness factor and “Universe” and “Scenery” (“Spirituality,” “Threat,” and “Humanity”), physiological factor and “Threat” and “Scenery” (“Universe”), and accommodation factor and “Spirituality” and “Threat” (“Scenery” and “Aesthetics”). These results were consistent with the characteristics of each factor of the original AWE-S (Yaden et al., 2019). Previous studies also demonstrated that the elicitors of awe experiences had different roles in manifesting responses to the AWE-S (Graziosi & Yaden, 2021; Yaden et al., 2019). Therefore, beyond the typology, this was the first study to reveal that the various psychological elements of awe emerged differently based on the potential topics of the narratives.

It should be also noted that “Scenery” was the most common topic, which was positively related to the vastness factor of the Japanese AWE-S. This topic mainly consisted of nature-related words, such as “landscape,” “mountain,” and “sea.” Previous studies have demonstrated that many Japanese people are familiar with nature worship, a belief that the self is a part of deified nature that is cultivated under the syncretism of Shinto and Buddhism (Sugimoto, 2020; Takano & Nomura, 2022a). Therefore, these results might reflect the Japanese view of the self in relation to nature worship (i.e., non-substantial, denial, embedded).

Interestingly, “Threat,” the second common topic, might be associated with negative aspects of awe (i.e., threat-awe), which is in line with the fact that natural disasters occur frequently in Japan due to its climate and topography. Previous studies suggested that feelings of awe that are triggered by threatening stimuli such as natural disasters could be tinged with the perception of threat (Takano & Nomura, 2022b). Given that even the recollection of awe experiences based on examples that do not include negative elements (e.g., seeing aurora) evokes the feelings of ifu, negative awe in Japanese (Takano & Nomura, 2023a), Japanese people might be more likely to feel threatened during awe experiences due to Japan’s geographic and historical background.

This study has some limitations and directions for future research. First, along with the original version, the effect sizes of the correlations between some of the factors of the AWE-S and trait-awe were relatively low (Yaden et al., 2019). This may have been caused by ceiling effects, which should be investigated in future studies. Second, when writing regarding previous experiences, it was possible that the participants’ ratings were influenced by their memory of the event and not the actual experience of awe. The recollection method allowed us to investigate the narrative dimensions of the AWE-S. As in a previous study (Bai et al., 2017; Valdesolo & Graham, 2014; Van Cappellen & Saroglou, 2012), other manipulation methods (e.g., video stimuli) should be used to confirm the generalizability of our results. Third, this study investigated the relationship between the Japanese AWE-S and self-reported measures to reveal its construct validity. Given that the advantage of the AWE-S was that it revealed psychological processes that could only be captured subjectively, further studies should investigate whether and how this scale was associated with other behavioral, physiological, and neural measures.

In summary, this study is the first to develop a Japanese version of the AWE-S, confirm its convergent and divergent validities, and investigate its narrative dimensions during awe experiences. Our findings contribute to a deeper understanding of awe and revealing its construct in Japan through cross-cultural comparisons. Furthermore, this study provides conceptual and methodological implications regarding studies on awe.