Links between emotion word usage, accuracy, and emotional dysregulation: An integrative analysis

doi:10.21203/rs.3.rs-4213874/v1

Download PDF

Research Article

Links between emotion word usage, accuracy, and emotional dysregulation: An integrative analysis

https://doi.org/10.21203/rs.3.rs-4213874/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 14 Nov, 2024

Read the published version in Affective Science →

You are reading this latest preprint version

People vary in the precision with which they report on their emotions, known as emotional granularity, and this precision predicts their ability to regulate their emotions. It is not yet known, however, whether links between emotion regulation and emotional granularity are due to variation knowledge of emotion words -- specifically, individuals’ reported usage, understanding, and ability to accurately define emotion words. In the present report, we combined data from six studies to address this gap in the literature using an integrative data analysis. Participants across the studies reported on their usage and understanding of a list of highly-granular emotion words, and were tested for their accuracy of definition. They also completed questionnaire measures of emotional granularity and emotion dysregulation. Emotion word accuracy and understanding were highly correlated, so individual models tested each separately to predict emotion regulation difficulty. In the model including usage and understanding, we observed a main effect of understanding, such that participants with greater self-assessed understanding of emotion words reported less difficulty regulating their emotions. Similar effects were found for the model including usage and accuracy, such that individuals with higher emotion word accuracy had less difficulty regulating their emotions. Critically, these findings held when accounting for self-reported granularity, suggesting the value of emotion word knowledge measures for predicting dysregulated emotions. Future work should examine how individuals’ emotion word knowledge is also linked to mental health outcomes.

emotional granularity

emotion word knowledge

emotional dysregulation

When people are asked to report on their emotional experiences using researcher-provided terms, they will vary in their ability to make distinctions among related terms. This variation is used to estimate emotional granularity, or individuals’ ability to create experiences of emotion that are nuanced and context-specific (Barrett, 1997). If an individual does not distinguish between neighboring emotion words (e.g., “angry” and “sad”) when evaluating their experiences, it implies that these words are being used less precisely and to refer to a broader affective state (e.g., “bad”), and suggests that this person is unable to flexibly construct emotions that are finely tuned the current set of circumstances. Thus, emotional granularity is lower when two or more emotions are endorsed similarly over time (e.g., “angry” and “sad” strongly covary), but higher when emotion terms are endorsed distinctively (e.g., “angry” and “sad” each reported at separate moments).

Extensive prior research reveals that emotional granularity is related to mental health (for reviews see Dunning et al., 2022; Nook, 2021; & Tan et al., 2022). One explanation of why these links emerge is through efficacious management of one’s emotions (termed, emotion regulation) once those emotions are precisely represented (e.g., Barrett et al., 2001; Kalokerinos et al., 2019; Tugade et al., 2004). Specifically, early evidence suggested that participants with higher emotional granularity used more emotion regulation strategies (identified retrospectively over two weeks) to address their negative emotions in comparison to individuals with lower emotional granularity (Barrett et al., 2001). In turn, higher granularity for positive emotions was found to be a predictor of effective coping and therefore a factor that influences individual differences in resilience (Tugade et al., 2004). More recent evidence suggests that lower emotional granularity, in particular, is associated with ineffective use of emotion regulation strategies in the moment (Kalokerinos et al., 2019), but not the selection of regulation techniques.

Why might granularity be linked to emotion regulation? One reason might be that variation in a person’s emotion word knowledge – specifically, their usage and understanding of emotion words, along with their ability to accurately define emotion words -- supports individual differences in emotional granularity (Kang & Shaver, 2004; Vedernikova et al., 2021). Indirect evidence from the study of alexithymia supports this link. Alexithymia was initially conceptualized as a limited ability to use words (Ruesch, 1948) as symbols for emotional states (Nemiah & Sifneos, 1970), but is now more often assumed to be a disorder of affect regulation (Lane et al., 1996; Taylor et al., 1997). Specifically, one recent model (da Silva et al., 2017) suggests that emotional granularity supports emotion regulation because having a more increased ability to assign meaning to emotional experiences increases subtle distinctions between emotions that enable healthy regulation strategies. In that study, emotional awareness and emotional granularity mediated the relationship between alexithymia and emotion regulation (da Silva et al., 2017). Individuals who are higher in alexithymia also produce fewer types of emotion words and fewer synonyms for target emotion words than do individuals who are lower in alexithymia (Worschack & Klann-Delius, 2013). Critically, a recent meta-analysis of studies between negative emotional granularity and alexithymia showed a significant negative correlation between the two (r = -.10) (Lee et al., 2022), suggesting there is a modest relationship between the ability to represent emotions with words and the tendency to experience emotions in a granular manner.

There is relatively more work directly linking emotion language and emotion regulation, although this research is still in its infancy and there are conflicting findings. Initial work on the topic suggested that emotion word usage—i.e., labeling one’s own emotions -- may benefit emotion regulation (e.g., Kircanski et al., 2012), a phenomenon referred to as affect labeling. There are several potential mechanisms proposed for how affect labeling affects emotion regulation (Torre & Lieberman, 2018). These include either by: 1) by serving as an intrinsic form of emotion regulation, or 2) by bringing online emotion category knowledge that can guide emotion regulation. More recent evidence suggests that labeling one’s emotions may not consistently serve these functions. For example, when participants applied a label to their own negative emotions, this did not lead to decreases negative affect, as would be predicted by an intrinsic regulatory account (Vlasenko et al., 2021). Further, when participants labeled their own emotions prior to being asked to reappraise their emotions, labeling actually hindered reappraisal efficacy (Nook et al., 2021). This suggests that, at least for some regulation strategies that relate to changing one’s representation of the emotional situation, labeling one’s emotions can backfire.

Another account of why emotion language, granularity and regulation may be linked is that emotion labels function to build a robust and flexible set of category representations. For example, it is proposed that labels in the early environment serve to cohere sets of emotion instances that vary widely in their physical, mental, and situational properties (Hoemann et al., 2019). Labels may direct attention to commonalities across these disparate instances, in the process organizing concepts for emotion that can then be generalized to new instances (Shablack & Lindquist, 2019). Indeed, specific labels (e.g., “disgusted”) rather than broad (“bad”) or irrelevant labels (“sitting down”) help young children to link photos of emotional faces to corresponding scenarios (Ogren & Sandhofer, 2022). These findings are broadly consistent with a constructionist approach to emotion, which proposes that emotion words serve as a critical tool for learning emotions as abstract concepts (Hoemann et al., 2019, 2020; see also Hoemann & Barrett, 2019). Finally, there is evidence linking early emotion vocabulary and the development of emotion regulation in early childhood such that emotion regulation ability may be built on a child’s emerging emotional understanding (Cole et al., 2010; Pons et al., 2003; for review, Atzil & Gendron, 2017).

What is not yet known, however, is whether individual differences in emotion word knowledge is related to emotion regulation in adulthood, and account for variation in emotion regulation above and beyond emotional granularity. In the present report, we combined data from six studies to address the relationship between emotion word knowledge and emotion granularity, as they affect emotion regulation? Specifically, we examine how emotion word usage, understanding, and accuracy predict difficulties in emotion regulation using the Difficulties in Emotion Regulation Scale-18 (DERS, Victor & Klonsky, 2016). Further, we also examine whether these measures relate to self-reported emotional granularity and if they predict difficulties in emotion regulation above and beyond self-reported granularity. We tested these relationships using an integrative data analysis (IDA; Curran & Hussong, 2009), a means of pooling individual subject data from independent studies that allowed us to examine whether emotion word knowledge associated with emotion regulation difficulty across studies despite slight variation in study parameters (e.g., number of included emotion words), while accounting for between-study heterogeneity.

Range and Differentiation of Emotional Experience Scale (RDEES)

Emotional granularity was measured using the 14-item Range and Differentiation of Emotional Experience Scale (RDEES) (Kang & Shaver, 2004). This scale is designed to assess the range of a person’s emotional experiences, including how attentive people are to their feelings, the amount they are open to experience, their ability to understand others’ feelings, and adjust socially. Participants respond to items such as “I usually experience a wide range of emotions” or “I am aware of the subtle differences between feelings I have” on a 5-point Likert scale, with 1 indicating “does not describe me very well” and 5 indicating “describes me very well” (see Appendix A). Four of these items are reverse-coded. Scores are averaged across all items. Higher scores indicate higher emotional differentiation (granularity), whereas lower scores indicate lower emotional differentiation. A factor analysis (Kang & Shaver, 2004) of all initial items showed 14 items were sufficient to produce two distinct factors: “range” and “differentiation”, with and alpha coefficient of .85 (.83 for 7-item range subscale and .79 for the remaining 7-item differentiation subscale). In those studies, study 1 used self-report ratings; study 2 used peer rating. The correlation between the two subscales ranged in study 1 between .30 and .47, and between .42 to .57 in study 2.

With respect to construct and convergent validity, the RDEES has been found to be significantly negatively correlated with the Toronto Alexithymia Scale (TAS), and significantly positively correlated with the Emotional Range Test, Levels of Emotional Awareness Test, and the Trait Meta-Mood Scale (TMMS) (Kang & Shaver, 2004). In study 1 (Kang & Shaver, 2004), the differentiation scale, but not the range scale, was significantly positively correlated with only two of the three subscales of the TMMS scales (clarity and repair subscales), but the range subscale was not. In study 2, the differentiation subscale showed greater correlation to an emotion card sorting task (i.e., generated more emotion categories) than the range subscale (r = .33 vs. .20). Thus, the differentiation subscale has good construct validity, such that emotional experience is related more to fine-grain categories than range of emotional experience. Additionally, the differentiation subscale (study 2) also was more strongly correlated with interpersonal relationship than the range subscale (r = 0. 40 vs. 0.14), showing that emotional differentiation is more important than emotional range for good relationship maintenance, even after controlling for social desirability. Emotional differentiation was also more strongly associated than emotional range with both self and peer ratings of interpersonal relationship quality. Although the two aspects of emotional complexity were correlated, the differentiation scale was more strongly related to knowing one's own feelings and understanding others’ feelings (compared with emotional intensity and mood variability). For these reasons, we only used participants’ scores from the differentiation subscale in our analyses.

Difficulties in Emotion Regulation Scale (DERS-18)

Difficulties in emotion regulation was measured using the 18-item version of the Difficulties in Emotion Regulation Scale-18 (Victor & Klonsky, 2016). The DERS is based on a clinically-useful conceptualization of emotion regulation “that was developed to be applicable to a wide variety of psychological difficulties and relevant to clinical applications and treatment development (Gratz, 2007; Gratz & Tull 2010)” (Bjureberg et al., 2016, p. 284). Specifically, the conceptual definition of emotion regulation on which the DERS is based emphasizes the functionality of emotions and focuses on adaptive ways of responding to emotional distress, including the: (a) awareness, understanding, and acceptance of emotions; (b) ability to control behaviors when experiencing negative emotions; (c) flexible use of situationally-appropriate strategies to modulate the intensity and/or duration of emotional responses, rather than to eliminate emotions entirely; and (d) willingness to experience negative emotions as part of pursuing meaningful activities in life. Participants respond to items such as “When I am upset, I have difficulty controlling my behaviors” or “When I am upset, I have difficulty focusing on other things” on a 5-point Likert scale, with 1 indicating “almost never” to 5 indicating “almost always” (see Appendix B). Three of the items are reverse-coded. The minimum overall score is 18, and the maximum score is 90. Higher scores indicate greater difficulties with emotion regulation. Internal consistency and reliability are reported for the 16 item version (Bjureberg et al., 2016). The internal consistency of the DERS-16 has an alpha of .92, and the test-retest reliability was good (ρ =.85, p < .001). In the original validation study (Bjureberg et al., 2016, study 2), the DERS-16 was significantly positively correlated with experiential avoidance, mindfulness, and negative emotionality, psychiatric symptoms, and most clinically- relevant behaviors. These were generally of the same magnitude and direction of the original DERS.

Emotion Words

An initial set of 40 emotion words was selected from the ANEW list (Bradley & Lang, 1999) to maximize coverage of the affective circumplex (Russell, 1980) in terms of valence, arousal, and emotion category, while also balancing for word length and difficulty (i.e. age of acquisition; see Baron-Cohen et al., 2010). Extensive piloting was performed on these 40 words which included 30 online student ratings of arousal and valence (1-5 scale), emotion category sorting (sad, calm, happy, fear, disgust, angry, happy, surprised, or none of the above), and frequency of use (1-5), among other variables not related to the current study. From these 40 words, 20 emotion words were selected based on at least two emotion words per each of the eight emotion categories. The remaining four were distributed across positive valence/low arousal, positive valence/high arousal, negative valence/low arousal, and negative valence/high arousal (see Appendix C). The specific words selected for each emotion category and the four remaining words were selected to have similar use by participants from the pilot ratings.

Participants and Specific Procedures by Study

Dataset 1

As part of a larger study on social media use and emotional health, nine undergraduate students from the University of Massachusetts-Dartmouth and 30 participants from the general population (advertised via social media) between the ages of 18-25 years (n = 32 females, n = 4 males, 2 = prefer not to say, 1= did not answer) completed an online Qualtrics survey which assessed their usage, understanding, (1 = I use this word/ I know what this word means; 4 =I do not use this word often/ I do not understand or am not confident that I know the meaning of this word/), and accuracy of 20 pre-normed highly granular emotion words. For each emotion word, participants were asked their usage, understanding, and then accuracy. The accuracy question was not asked if participants indicated a “4” on understanding for that word, as it did not seem to make sense to ask participants to choose a definition if they did not know what the word meant. For the accuracy questions, the incorrect answers were the correct answers from other words, randomly assigned and equally repeated throughout the list. Accuracy was calculated as the percentage correct over trials. We decided that total accuracy for non-answered/non-attempted questions should be counted as incorrect. Usage and understanding were averaged over the 20 emotion words. Participants also completed the RDEES and DERS surveys. This experiment was part of a larger study which also collected information on participants’ online social media usage and emotional health, and included the Mental Health Continuum (MHC). Participants were entered into a raffle for the chance to win one $50USD gift card at the end of the study for their participation. Data was collected in 2019 under IRB #18.084

Datasets 2 & 3

One hundred fourteen undergraduate students from the University of Massachusetts-Dartmouth completed a Qualtrics survey similar to that described for Dataset 1, in which they also rated each presented emotion word on valence, arousal, and embodiment. Participants were required to be between 18-25 years of age to participate; no further demographic data were collected. The study included the additional measures described above and the same measures and words as above. Participants received a $10USD gift card or one research credit for participation if they were enrolled in Psychology 101 as part of the course requirement. Participants subsequently completed an emotional face perceptual discrimination task in which they viewed individual morphed faces created from combining (using morphing software) sequences of five faces between each of two emotions. In different trials, each emotion was morphed with one of five other emotions. Four individual poser’s faces were used (Barrett face set)1[1]. Morphs were shown one per trial followed by a second face that was either the same exact face or a face that depicted slightly more of one of the emotions (slightly less of the other). Participants were asked to judge whether the second face matched the first face using a computer keyboard. Data was collected in 2019 under IRB #18.084.

Dataset 4

One hundred twenty three undergraduate students from the University of Massachusetts-Dartmouth completed a Qualtrics survey similar to that described for Dataset 1, except the MHC and gender question did not function and did not record data. Participants were entered into a raffle for the chance to win one $50USD gift card at the end of the study for their participation. Data was collected in 2020 under IRB #18.084.

Datasets 5 & 6

Thirty-six (Dataset 5) and 38 (Dataset 6) graduate students from the Kansas City University completed a Qualtrics survey similar to that described for Dataset 1, with the addition of six other surveys in Dataset 5 (n = 36) (Schutte Suite Emotional Intelligence Scale (SSEI), Emotion Regulation Questionnaire (ERQ), Satisfaction with Life Scale (SWL), MHC, TMMS, and the TAS), and these same scales minus the SSEI in Dataset 6 (n = 38). Instead of 20 emotion words, only 14 emotion words were used, and three new words were added to replace words which had a consistently high (>98%) emotion accuracy. As with other data sets, the accuracy question only appeared if participants did not report a “4” for usage (and now also for understanding). In addition, participants completed two emotion perception tasks (see Datasets 2 & 3) in the first PI’s current laboratory at Kansas City University (KCU) in 2022 and 2023. Participants received a $10 USD gift card to Amazon for completion. The study which contributed to Dataset 5 was approved through IRB KCU #1906743 and the study which contributed to Dataset 6 was approved through IRB KCU #1942774. Participants contributing to Dataset 5 included additional training and a long-term intervention of emotion word training and experiential momentary sampling for additional compensation (up to $100 USD over 3 weeks).

We conducted a fixed-effects IDA in which we treated study membership as a property of each participant’s data (Curran & Hussong, 2009) while testing the effect of emotion word knowledge on emotion regulation difficulty using a linear model across all datasets (N = 268). To examine the effect of emotion word knowledge above and beyond self-reported ability to differentiate between emotions, in a first step we fitted a model predicting emotion regulation difficulty (DERS-16) from study (1-6), the differentiation subscale of the RDEES, and their interaction. Then, in a second step, we added the three measures of emotion word knowledge (usage, understanding, and accuracy) and their interactions with study. Emotion word knowledge and self-report variables were standardized by dataset. We used deviation coding to compare each level of the study factor to the grand mean without setting any particular dataset as reference (Brehm & Alday, 2022). As a consequence, the last level of the study factor was not represented in the initial regression results; we accounted for this by manually resetting the contrast and re-running the models to get coefficients for Dataset 6. Descriptive statistics by study are presented in Table 1.

Table 1 Descriptive Statistics by Dataset

Variable	Dataset 1	Dataset 2	Dataset 3	Dataset 4	Dataset 5	Dataset 6
N	39	40	74	123	36	38
Emotion regulation difficulty (DERS-16)	40.31 (11.80)	40.88 (11.87)	38.12 (12.36)	43.09 (11.76)	34.39 (10.09)	36.89 (9.41)
Differentiation (RDEES)	2.29 (0.74)	3.32 (.80)	3.67 (0.60)	2.49 (0.72)	3.74 (0.66)	3.75 (0.72)
Usage	-2.11 (0.42)	-2.14 (0.35)	-2.33 (0.43)	-2.27 (0.41)	-2.02 (0.38)	-2.05 (0.44)
Understanding	-1.18 (0.50)	-1.09 (0.15)	-1.20 (0.40)	-1.17 (0.35)	-1.15 (0.18)	-1.11 (0.14)
Accuracy	.90 (.17)	.92 (.12)	.92 (.15)	.91 (.14)	.89 (.11)	.86 (.12)

Note: Means (Standard Deviations) provided for emotion word knowledge and self-report variables.

Regression results are presented in Table 2. Analyses were run in R using the lme4 and lmerTest packages (Bates et al., 2015; Kuznetsova et al., 2017). All tests of significance are two-tailed at ɑ = .05.

An initial model including study, differentiation, and their interaction revealed no main effect of either predictor on emotion regulation difficulty. There was a significant interaction between study level 1 and differentiation (β = .46, t(338) = 3.25, p = .001). Separate models per study showed that differentiation was a significant positive predictor in Dataset 1, such that participants who reported being better able to distinguish between emotions also reported more difficulty regulating them. This relationship was nonsignificant or negative in all other studies (Table 2), suggesting limited generalizability of this effect.

A subsequent model including study, differentiation, usage, understanding, and accuracy, as well as the respective interactions with study, also revealed no main effects. In addition to the interaction between study level 1 and differentiation (β = .43, t(320) = 2.96, p = .003), there was an interaction between study level 3 and usage (β = .25, t(320) = 2.25, p = .03). Separate models per study showed that usage was a significant positive predictor in Datasets 3 and 4, such that participants who reported more frequent use of the emotion words also reported more difficulty regulating their emotions. This relationship was strongest in Dataset 3, and we did not find statistically significant evidence for the effect in Datasets 1, 2, 5, or 6 (Table 2).

Table 2 Integrative Data Analysis Regression Results

Parameter	β	SE	t	95% CI
*Step 1:* DERS ~ Study + Differentiation + Study: Differentiation
Intercept	-.001	.06	-0.02	-.12, .11
Dataset 1	-.005	.14	-0.04	-.28, .27
Dataset 2	.001	.14	0.01	-.27, .28
Dataset 3	.001	.11	0.01	-.22, .22
Dataset 4	.002	.09	0.02	-.18, .18
Dataset 5	.001	.15	0.01	-.29, .29
Dataset 6	.001	.14	0.004	-.28, .28
Differentiation	-.08	.06	-1.29	-.19, .04
Dataset 1 x Differentiation	.46	.14	3.25**	.18, .75
Dataset 2 x Differentiation	-.12	.14	-0.82	-.39, .16
Dataset 3 x Differentiation	.20	.11	1.83†	-.02, .42
Dataset 4 x Differentiation	-.06	.09	-0.59	-.24, .13
Dataset 5 x Differentiation	-.23	.15	-1.53	-.52, .06
Dataset 6 x Differentiation	-.27	.14	-1.86†	-.55, .02
*Step 2:* DERS~ Study + Differentiation + Usage + Understanding + Accuracy + Study: Differentiation + Study: Usage + Study: Understanding + Study: Accuracy
Intercept	.002	.06	0.03	-.11, .11
Dataset 1	-.01	.14	-0.04	-.28, .26
Dataset 2	-.001	.14	-0.01	-.27, .27
Dataset 3	-.003	.11	-0.02	-.21, .21
Dataset 4	.01	.09	0.15	-.16, .19
Dataset 5	-.002	.14	-0.01	-.28, .28
Dataset 6	-.002	.14	-0.02	-.28, .27
Differentiation	-.07	.06	-1.08	-.18, .05
Usage	.04	.06	0.64	-.09, .17
Understanding	-.13	.09	-1.44	-.31, .05
Accuracy	-.06	.09	-0.68	-.25, .12
Dataset 1 x Differentiation	.43	.15	2.96**	.15, .72
Dataset 2 x Differentiation	-.11	.14	-0.78	-.40, .17
Dataset 3 x Differentiation	.16	.11	1.46	-.06, .38
Dataset 4 x Differentiation	-.04	.09	-0.47	-.23, .14
Dataset 5 x Differentiation	-.21	.15	-1.41	-.51, .08
Dataset 6 x Differentiation	-.23	.15	-1.54	-.52, .06
Dataset 1 x Usage	-.01	.16	-0.09	-.34, .31
Dataset 2 x Usage	-.14	.15	-0.94	-.42, .15
Dataset 3 x Usage	.25	.11	2.25*	.03, .47
Dataset 4 x Usage	.14	.10	1.48	-.05, .33
Dataset 5 x Usage	-.08	.15	-0.54	-.39, .22
Dataset 6 x Usage	-.16	.17	-0.95	-.49, .17
Dataset 1 x Understanding	.44	.31	1.41	-.17, 1.05
Dataset 2 x Understanding	-.28	.16	-1.78†	-.60, .03
Dataset 3 x Understanding	-.22	.20	-1.12	-.61, .17
Dataset 4 x Understanding	-.14	.19	-0.75	-.50, .23
Dataset 5 x Understanding	.14	.16	0.85	-.18, .46
Dataset 6 x Understanding	.07	.16	0.40	-.26, .39
Dataset 1 x Accuracy	-.31	.32	-0.98	-.93, .31
Dataset 2 x Accuracy	.22	.16	1.33	-.11, .54
Dataset 3 x Accuracy	.16	.20	0.81	-.23, .55
Dataset 4 x Accuracy	.02	.18	0.09	-.34, .37
Dataset 5 x Accuracy	-.05	.17	-0.28	-.38, .28
Dataset 6 x Accuracy	-.04	.19	-0.22	-.41, .33
*Step 2A:* DERS ~ Study + Differentiation + Usage + Understanding + Study: Differentiation + Study: Usage + Study: Understanding
Intercept	.001	.06	0.03	-.11, .11
Dataset 1	-.01	.14	-0.05	-.28, .26
Dataset 2	-001	.14	-0.01	-.27, .27
Dataset 3	-.002	.11	-0.02	-.21, .21
Dataset 4	.03	.09	0.15	-.16, .19
Dataset 5	-.001	.14	-.01	-.28, .28
Dataset 6	-.002	.14	-0.02	-.27, .27
Differentiation	-.07	.06	-1.22	-.19, .04
Usage	.03	.06	0.45	-.09, .15
Understanding	-.17	.06	-2.81**	-.29, -.05
Dataset 1 x Differentiation	.46	.15	3.15**	.17, .74
Dataset 2 x Differentiation	-.14	.14	-0.98	-.42, .14
Dataset 3 x Differentiation	.16	.11	1.51	-.05, .38
Dataset 4 x Differentiation	-.04	.09	-0.39	-.22, .15
Dataset 5 x Differentiation	-.22	.15	-1.48	-.51, .07
Dataset 6 x Differentiation	-.23	.15	-1.58	-.52, .06
Dataset 1 x Usage	-.03	.16	-0.17	-.34, .29
Dataset 2 x Usage	-.10	.14	-0.72	-.38, .18
Dataset 3 x Usage	.27	.11	2.50*	.06, .49
Dataset 4 x Usage	.16	.09	1.67†	-.03, .34
Dataset 5 x Usage	-.10	.15	-0.65	-.39, .20
Dataset 6 x Usage	-.20	.14	-1.41	-.48, .08
Dataset 1 x Understanding	.16	.16	1.00	-.15, .47
Dataset 2 x Understanding	-.20	.14	-1.46	-.48, .07
Dataset 3 x Understanding	-.10	.11	-0.90	-.32, .12
Dataset 4 x Understanding	-.14	.10	-1.36	-.35, .06
Dataset 5 x Understanding	.20	.15	1.34	-.09, .48
Dataset 6 x Understanding	.09	.15	0.60	-.20, .38
*Step 2B:* DERS~ Study + Differentiation + Usage + Accuracy + Study: Differentiation + Study: Usage + Study: Accuracy
Intercept	.001	.06	0.02	-.11, .11
Dataset 1	-.01	.14	-0.05	-.28, .27
Dataset 2	-.001	.14	-0.01	-.27, .27
Dataset 3	-.001	.11	-0.01	-.21, .21
Dataset 4	.01	.09	0.12	-.17, .19
Dataset 5	-.001	.14	-0.01	-.28, .28
Dataset 6	-.002	.14	-0.01	-.28, .27
Differentiation	-.07	.06	-1.08	-.18, .05
Usage	.04	.06	0.64	-.09, .17
Accuracy	-.13	.07	-1.97*	-.26, -.0004
Dataset 1 x Differentiation	.44	.15	2.97**	.15, .73
Dataset 2 x Differentiation	-.10	.15	-0.65	-.38, .19
Dataset 3 x Differentiation	.16	.11	1.44	-.06, .38
Dataset 4 x Differentiation	-.05	.09	-0.57	-.24, .13
Dataset 5 x Differentiation	-.21	.15	-1.40	-.51, .09
Dataset 6 x Differentiation	-.24	.15	1.12	-.53, .05
Dataset 1 x Usage	.001	.17	0.01	-.32, .33
Dataset 2 x Usage	-.17	.15	-1.13	-.45, .12
Dataset 3 x Usage	.27	.11	2.39*	.05, .49
Dataset 4 x Usage	.14	.10	1.43	-.05, .33
Dataset 5 x Usage	-.08	.15	-0.53	-.39, .22
Dataset 6 x Usage	-.16	.17	-0.96	-.50, .17
Dataset 1 x Accuracy	.03	.16	0.16	-.30, .35
Dataset 2 x Accuracy	.16	.15	1.12	-.12, .45
Dataset 3 x Accuracy	-.08	.11	-0.66	-.30, .15
Dataset 4 x Accuracy	-.14	.11	-1.34	-.35, .07
Dataset 5 x Accuracy	.02	.15	0.11	-.29, .32
Dataset 6 x Accuracy	.01	.17	0.06	-.33, .35

Note: Significant parameters highlighted in bold. ** p < .01, * p < .05, † p < .10

To follow up on these results, we examined the bivariate correlations between the raw (unstandardized) emotion word knowledge variables across studies. We found that understanding and accuracy were highly correlated with each other (r = .73), but less so with usage (understanding and usage: r = .13; accuracy and usage: r = .14) (Table 3).

Table 3 Correlations among Variables

Variable	Variable	n	Pearson’s Partial Correlation	p value
Usage	Understanding	350	0.115*	0.032
Usage	Accuracy	350	-0.190***	< .001
Usage	DERS-16	350	-0.058	0.277
Usage	RDEES_ differentiation	350	-0.076	0.159
Understanding	Accuracy	350	-0.609***	< .001
Understanding	DERS-16	350	0.220***	< .001
Understanding	RDEES_ differentiation	350	0.012	0.819
Accuracy	DERS-16	350	-0.176***	< .001
Accuracy	RDEES_ differentiation	350	-0.020	0.714
DERS-16	RDEES_ differentiation	350	-0.056	0.295

Note: Controlling for study (Dataset) *p < .05; **p < .01; ***p < .001

With these relationships in mind, we ran two separate versions of our second-step model: one with study, differentiation, usage, and understanding; the other with study, differentiation, usage, and accuracy.

In the model including usage and understanding, we observed a main effect of understanding (β = -.17, t(326) = -2.81, p = .01), such that participants with greater self-assessed understanding of emotion words also reported less difficulty regulating their emotions. We observed the same interactions between study level 1 and differentiation (β = .46, t(326) = 3.15, p = .002) and between study level 3 and usage (β = .27, t(326) = 2.50, p = .01).

In the model including usage and accuracy, we observed a main effect of accuracy (β = -.13, t(326) = -1.97, p = .05), such that participants with higher performance on the emotion word definition task also reported less difficulty regulating their emotions. We once again observed the interactions between study level 1 and differentiation (β = .44, t(326) = 2.97, p = .003) and between study level 3 and usage (β = .27, t(326) = 2.39, p = .02).

The present study examined how emotion word knowledge relates to emotional granulation and emotion regulation (i.e., difficulties in emotion regulation). Specifically, we examined how self-assessed usage and understanding of emotion words, as well as demonstrated definitional accuracy of these words, related to individuals’ self-reported difficulties in managing their emotions. We examined these relationships across six studies using an integrative analysis approach. We found that individuals with greater self-assessed understanding of emotion words reported less difficulty regulating their emotions. We also found that individuals who more accurately defined emotion words reported less difficulty regulating their emotions. These findings only emerged when these effects were examined in separate models, suggesting that these effects are not independent of one another. Further, the effect sizes were similar, suggesting that the simpler measure of asking participants to self-assess their understanding of emotion words may be a sufficient means of capturing emotion knowledge (and predicting meaningful outcomes). It should be noted, however, that accuracy was only assessed when participants indicated they understood the word. This suggests that measuring accuracy above and beyond understanding does not account for emotion regulation difficulties, although it is possible that other assessments of accuracy may provide added value. For example, it may be that the accuracy of self-generated definitions for emotion labels (e.g., Nook et al., 2020) better predicts emotion regulation than the current assessment of picking the correct definition.

We did not observe reliable effects of emotion word usage in predicting emotion regulation difficulties. This is somewhat inconsistent with recent findings suggesting that the breadth of one’s active emotion vocabulary is linked to worse mental health symptoms (Vine et al., 2019) and this breadth is evident in individuals with bipolar disorder (Entwistle et al., 2023). The present usage measure was based on self-report, and so it may diverge from active emotion vocabularies assessed in natural language. Consistent with this idea, when emotion word fluency was measured in a decontextualized manner (i.e., adapting a classic semantic fluency paradigm), individual differences did not predict emotion regulation or depressive symptoms (Hegefeld et al., 2023). Together, these findings suggest that usage measures likely have complex and contextually bounded relationships with outcomes in the emotion regulation and perhaps other mental health domains.

We also did not observe that our measures of emotion word knowledge related to self-reported emotional granularity (i.e., the ability to differentiate between similar emotions), nor did we observe robust relationships between emotional granularity and difficulties in emotion regulation. Importantly, word measures predicted difficulties in emotion regulation, even when accounting for emotional granularity. One possible reason that the granularity measure has limited predictive utility in this context is that individuals have limited insight into their level of granularity. An important future direction will be to compare the present word-knowledge measures against estimates of emotional granularity derived from reports of momentary experience. A second possible reason for these null findings is because emotional granularity may depend on evaluating emotion words in relation to complex, often ambiguous events, and simply assessing knowledge of word meanings may fall short of capturing variation in contextual usage. Future work should examine whether self-reported emotional granularity sufficiently captures such dynamics in emotion.

Future work should examine whether individuals’ emotion word knowledge also links to mental health outcomes. Lower emotional granularity is linked to clinically significant forms of behavioral dysregulation, including eating disorders (Mikhail et al., 2020; Rommel et al., 2013) and substance abuse disorders (Emery et al., 2014; Kashdan et al., 2010), whereas higher emotional granularity appears to be protective against non-suicidal self-injury in individuals with borderline personality disorder (Zaki et al., 2013). Further, lower emotional granularity predicts the severity of depressive symptoms in both adolescence and adulthood (Starr et al., 2020; Tomko et al., 2015; Willroth et al., 2020). Given the bidirectional relationships between mental health outcomes and difficulties with emotion regulation (Dunning et al., 2022; Nook, 2021; & Tan et al., 2022), it will be important to directly examine whether emotion word knowledge predicts mental health outcomes.

Ethical Approval

Studies 1- 4 were approved by the IRB at University of MA – Dartmouth, #18.084. Study 5 was approved through the IRB at Kansas City University, #1906743. Study 6 was approved through the IRB at Kansas City University, #1942774. All participants signed either electronic or paper consent forms stating their voluntary willingness and informed consent to participate. Data were collected anonymously, with the exception of information pertaining to payment/remuneration, collected at the end of the surveys electronically and stored separately from responses.

Funding

K.H. was supported by the Research Foundation – Flanders (12A3923N).

Availability of Data and Materials

Survey measures and word list are available in the Appendices. Data is available by emailing the corresponding author.

Atzil, S., & Gendron, M. (2017). Bio-behavioral synchrony promotes the development of conceptualized emotions. Current opinion in Psychology, 17, 162-169. https://doi.org/10.1016/j.copsyc.2017.07.009
Baron-Cohen, S., Golan, O., Wheelwright, S., Granader, Y., & Hill, J. (2010). Emotion word comprehension from 4 to 16 years old: A developmental survey. Frontiers in Evolutionary Neuroscience, 2, 109.
Barrett, L. F., Gross, J., Christensen, T. C., & Benvenuto, M. (2001). Knowing what you're feeling and knowing what to do about it: Mapping the relation between emotion differentiation and emotion regulation. Cognition and Emotion, 15(6), 713–724. https://doi.org/10.1080/02699930143000239
Barrett, L. F. (1997). The relationships among momentary emotion experiences, personality descriptions, and retrospective ratings of emotion. Personality and Social Psychology Bulletin, 23(10), 1100-1110.
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01
Bjureberg, J., Ljótsson, B., Tull, M. T., Hedman, E., Sahlin, H., Lundh, L. G., ... & Gratz, K. L. (2016). Development and validation of a brief version of the difficulties in emotion regulation scale: the DERS-16. Journal of Psychopathology and Behavioral Assessment, 38, 284-296. ttps://doi.org/10.1007/s10862-015-9514-x
Bradley, M. M., & Lang, P. J. (1999). Affective norms for English words (ANEW): Instruction manual and affective ratings (Vol. 30, No. 1, pp. 25-36). Technical report C-1, the center for research in psychophysiology, University of Florida.
Brehm, L., & Alday, P. M. (2022). Contrast coding choices in a decade of mixed models. Journal of Memory and Language, 125, 104334. https://doi.org/10.1016/j.jml.2022.104334
Cole, P. M., Armstrong, L. M., & Pemberton, C. K. (2010). The role of language in the development of emotion regulation. In S. D. Calkins & M. A. Bell (Eds.), Child development at the intersection of emotion and cognition (pp. 59–77). American Psychological Association. https://doi.org/10.1037/12059-004
Curran, P. J., & Hussong, A. M. (2009). Integrative data analysis: the simultaneous analysis of multiple data sets. Psychological methods, 14(2), 81–100. https://doi.org/10.1037/a0015914
da Silva, A. N., Vasco, A. B., & Watson, J. C. (2017). Alexithymia and emotional processing: A mediation model. Journal of Clinical Psychology, 73(9), 1196-1205.
Dunning, D. L., Wright, G., Bennett, M. P., Knight, R., & Dalgleish, T. (2022). What role does emotional granularity play in adolescent depression and anxiety? A scoping review.
Emery, N. N., Simons, J. S., Clarke, C. J., & Gaher, R. M. (2014). Emotion differentiation and alcohol-related problems: The mediating role of urgency. Addictive Behaviors, 39(10), 1459-1463.
Entwistle, C., Horn, A. B., Meier, T., Hoemann, K., Miano, A., & Boyd, R. L. (2023). Natural emotion vocabularies and borderline personality disorder. Journal of Affective Disorders Reports, 14, 100647. https://doi.org/10.1016/j.jadr.2023.100647
Gratz, K. L. (2007). Targeting emotion dysregulation in the treatment of self‐injury. Journal of Clinical Psychology, 63(11), 1091-1103.
Gratz, K. L., & Tull, M. T. (2010). Emotion regulation as a mechanism of change in acceptance-and mindfulness-based treatments. Assessing mindfulness and acceptance processes in clients: Illuminating the theory and practice of change, 2, 107-33.
Hegefeld, H. M., Satpute, A. B., Ochsner, K. N., Davidow, J. Y., & Nook, E. C. (2023). Fluency generating emotion words correlates with verbal measures but not emotion regulation, alexithymia, or depressive symptoms. Emotion, 23(8), 2259–2269. https://doi.org/10.1037/emo0001229
Hoemann, K., & Feldman Barrett, L. (2019). Concepts dissolve artificial boundaries in the study of emotion and cognition, uniting body, brain, and mind. Cognition and Emotion, 33(1), 67-76. https://doi.org/10.1080/02699931.2018.1535428
Hoemann, K., Khan, Z., Feldman, M. J., Nielson, C., Devlin, M., Dy, J., ... & Quigley, K. S. (2020). Context-aware experience sampling reveals the scale of variation in affective experience. Scientific Reports, 10(1), 12459. https://doi.org/10.1038/s41598-020-69180-y
Hoemann, K., Xu, F., & Barrett, L. F. (2019). Emotion words, emotion concepts, and emotional development in children: A constructionist hypothesis. Developmental Psychology, 55(9), 1830-1849. https://doi.org/10.1037/dev0000686
Kalokerinos, E. K., Erbas, Y., Ceulemans, E., & Kuppens, P. (2019). Differentiate to Regulate: Low Negative Emotion Differentiation Is Associated With Ineffective Use but Not Selection of Emotion-Regulation Strategies. Psychological Science, 30(6), 863–879. https://doi.org/10.1177/0956797619838763
Kang, S.-M., & Shaver, P. R. (2004). Individual Differences in Emotional Complexity: Their Psychological Implications. Journal of Personality, 72(4), 687–726. https://doi.org/10.1111/j.0022-3506.2004.00277.x
Kashdan, T. B., Ferssizidis, P., Collins, R. L., & Muraven, M. (2010). Emotion differentiation as resilience against excessive alcohol use: an ecological momentary assessment in underage social drinkers. Psychological Science, 21(9), 1341–1347. https://doi.org/10.1177/0956797610379863
Kircanski, K., Lieberman, M. D., & Craske, M. G. (2012). Feelings into words: Contributions of language to exposure therapy. Psychological Science, 23(10), 1086-1091.
Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest package: tests in linear mixed effects models. Journal of Statistical Software, 82(13).
Lane, R. D., Lee, S., Reidel, R., Weldon, V., Kaszniak, A., & Schwartz, G. E. (1996). Impaired verbal and nonverbal emotion recognition in alexithymia. Psychosomatic Medicine, 58(3), 203-210.
Lee, K. S., Murphy, J., Catmur, C., Bird, G., & Hobson, H. (2022). Furthering the language hypothesis of alexithymia: An integrated review and meta-analysis. Neuroscience & Biobehavioral Reviews, 104864. https://doi.org/10.1016/j.neubiorev.2022.104864
Mikhail, M. E., Keel, P. K., Burt, S. A., Neale, M., Boker, S., & Klump, K. L. (2020). Low emotion differentiation: An affective correlate of binge eating? International Journal of Eating Disorders, 53(3), 412-421.
Nemiah, J. C., & Sifneos, P. E. (1970). Affect and fantasy in patients with psychosomatic disorders. Modern Trends in Psychosomatic Medicine, 2, 26-34.
Nook, E. C., Satpute, A. B., & Ochsner, K. N. (2021). Emotion naming impedes both cognitive reappraisal and mindful acceptance strategies of emotion regulation. Affective Science, 2(2), 187-198.
Nook, E. C., Vidal Bustamante, C. M., Cho, H. Y., & Somerville, L. H. (2020). Use of linguistic distancing and cognitive reappraisal strategies during emotion regulation in children, adolescents, and young adults. Emotion, 20(4), 525-540. https://doi.org/10.1037/emo0000570
Nook, E. C. (2021). Emotion differentiation and youth mental health: current understanding and open questions. Frontiers in Psychology, 12, 700298.
Ogren, M., & Sandhofer, C. M. (2022). Emotion words link faces to emotional scenarios in early childhood. Emotion, 22(1), 167- 178. https://doi.org/10.1037/emo0001063
Pons, F., Lawson, J., Harris, P. L., & De Rosnay, M. (2003). Individual differences in children's emotion understanding: Effects of age and language. Scandinavian Journal of Psychology, 44(4), 347-353. https://doi.org/10.1111/1467-9450.00354
Rommel, D., Nandrino, J. L., Antoine, P., & Dodin, V. (2013). Emotional differentiation and parental bonding in inpatients suffering from eating disorders. British Journal of Clinical Psychology, 52(2), 215-229. https://doi.org/10.1111/bjc.12010
Ruesch, J. (1948). The infantile personality. Psychosomatic Medicine, 10, 134-144.
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39(6), 1161-1178.
Shablack, H., & Lindquist, K. A. (2019). The role of language in emotional development. In: LoBue, V., Pérez-Edgar, K., Buss, K.A. (Eds.) Handbook of Emotional Development. Springer, Cham. https://doi.org/10.1007/978-3-030-17332-6_18
Starr, L. R., Hershenberg, R., Shaw, Z. A., Li, Y. I., & Santee, A. C. (2020). The perils of murky emotions: Emotion differentiation moderates the prospective relationship between naturalistic stress exposure and adolescent depression. Emotion, 20(6), 927- 938. https://doi.org/10.1037/emo0000630
Tan, T. Y., Wachsmuth, L., & Tugade, M. M. (2022). Emotional nuance: Examining positive emotional granularity and well-being. Frontiers in psychology, 13, 715966.
Taylor, G.J., Bagby, R.M., & Parker, J.D.A. (1997). Disorders of affect regulation: Alexithymia in medical and psychiatric illness. Cambridge University Press.
Tomko, R. L., Lane, S. P., Pronove, L. M., Treloar, H. R., Brown, W. C., Solhan, M. B., Wood, P. K., & Trull, T. J. (2015). Undifferentiated negative affect and impulsivity in borderline personality and depressive disorders: A momentary perspective. Journal of Abnormal Psychology, 124(3), 740–753. https://doi.org/10.1037/abn0000064
Torre, J. B., & Lieberman, M. D. (2018). Putting feelings into words: Affect labeling as implicit emotion regulation. Emotion Review, 10(2), 116-124.
Tugade, M. M., Fredrickson, B. L., & Barrett, L. F. (2004). Psychological resilience and positive emotional granularity: examining the benefits of positive emotions on coping and health. Journal of Personality, 72(6), 1161–1190. https://doi.org/10.1111/j.1467-6494.2004.00294.x
Vedernikova, E., Kuppens, P., & Erbas, Y. (2021). From knowledge to differentiation: Increasing emotion knowledge through an intervention increases negative emotion differentiation. Frontiers in psychology, 12, Article 703757. https://doi.org/10.3389/fpsyg.2021.70375
Victor, S. E., & Klonsky, E. D. (2016). Validation of a brief version of the difficulties in emotion regulation scale (DERS-18) in five samples. Journal of Psychopathology and Behavioral Assessment, 38, 582-589.
Vine, V., Bernstein, E. E., & Nolen-Hoeksema, S. (2019). Less is more? Effects of exhaustive vs. minimal emotion labelling on emotion regulation strategy planning. Cognition and Emotion, 33(4), 855-862. https://doi.org/10.1080/02699931.2018.1486286
Vlasenko, V. V., Rogers, E. G., & Waugh, C. E. (2021). Affect labelling increases the intensity of positive emotions. Cognition and Emotion, 35(7), 1350-1364. https://doi.org/10.1080/02699931.2021.1959302
Willroth, E. C., Flett, J. A., & Mauss, I. B. (2020). Depressive symptoms and deficits in stress‐reactive negative, positive, and within‐emotion‐category differentiation: A daily diary study. Journal of Personality, 88(2), 174-184. https://doi.org/10.1111/jopy.12475
Wotschack, C., & Klann-Delius, G. (2013). Alexithymia and the conceptualization of emotions: A study of language use and semantic knowledge. Journal of Research in Personality, 47, 514-523.
Zaki, L. F., Coifman, K. G., Rafaeli, E., Berenson, K. R., & Downey, G. (2013). Emotion differentiation as a protective factor against nonsuicidal self-injury in borderline personality disorder. Behavior Therapy, 44(3), 529–540. https://doi.org/10.1016/j.beth.2013.04.008

[1] Research Tools - Lisa Feldman Barrett - Interdisciplinary Affective Science Laboratory - Northeastern University (affective-science.org)

No competing interests reported.

Appendix.docx

Download PDF

Journal Publication

published 14 Nov, 2024

Read the published version in Affective Science →

Editorial decision: Revision requested
22 Jul, 2024
Reviews received at journal
19 Jul, 2024
Reviews received at journal
21 Jun, 2024
Reviewers agreed at journal
22 May, 2024
Reviewers agreed at journal
07 May, 2024
Reviewers invited by journal
07 May, 2024
Editor assigned by journal
01 May, 2024
Submission checks completed at journal
12 Apr, 2024
First submitted to journal
03 Apr, 2024

You are reading this latest preprint version

Links between emotion word usage, accuracy, and emotional dysregulation: An integrative analysis

Status:

Journal Publication

Version 1

Abstract

Introduction

Methods

Results

Discussion

Declarations

References

Footnotes

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1