This is an open access article distributed under the terms of the Creative Commons
Attribution License (Climate change is a fundamental threat to humanity and life on Earth (IPCC, 2023). To mitigate climate change, people need to adopt low-carbon lifestyles, and policies need to facilitate such lifestyles (IPCC, 2018; Newell et al., 2021). Therefore, it is important to understand what motivates people not to deny climate change but to support and demand climate-friendly political decisions and societal changes.
Social identity theory (SIT, Tajfel & Turner, 1979) assumes that identification with a group motivates behaviour on behalf of that group. Researchers have thus argued that identifying with all human beings as an inclusive group—a global identity (McFarland et al., 2019)—should promote climate protection for the sake of all humanity (e.g., Reese, 2016). Indeed, research shows positive relations between the strength of people’s global identity and their pro-environmental intentions and behaviour. However, most of these studies focused on private behaviour (e.g., Joanes, 2019; Loy, Reese et al., 2022; Pong, 2021), or policy support (e.g., Loy, Clemens et al., 2022; Loy et al., 2021; Loy & Reese, 2019). Activism has rarely been investigated (see Renger & Reese, 2017; Rosenmann et al., 2016, for exceptions).
Self-categorisation theory (SCT, Turner et al., 1987) assumes that different aspects of people’s identity can become salient in specific contexts and guide behaviour, depending on situational cues that make the identity aspect accessible. However, only few studies were able to raise global identity salience (e.g., Loy et al., 2021; Reese et al., 2015; Römpke et al., 2019). Thus, respective research has been called for (McFarland et al., 2019). One way to make identity aspects salient is through communication (Seyranian, 2014; Trepte & Loy, 2017)—particularly language usage. For example, words referring to global identity increased global identity salience (Tu et al., 2012). Identity salience, in turn, can affect behavioural intentions. For example, inclusive language referring to people’s social identity fostered intentions to promote renewable energies (compared to individualising language; Seyranian, 2014). Drawing on SIT, we further argue that accusatory, exclusive language could demotivate collective behaviour. Yet, this type of language has, to our knowledge, not been examined in the context of pro-environmental behaviour.
Our study extends prior research by investigating whether inclusive language has the potential to raise global (not only social group) identity salience and thereby promotes the intention to engage in pro-environmental activism and policy support. Moreover, we address the research gap on the relation between global identity and denial of climate change implications. As further novelties, we compare inclusive to exclusive language in the specific context of demonstrations for climate protection. Thereby, we contribute to the practical question how climate activists can garner support effectively. In the following, we outline our theoretical rationale in detail.
Global Identity and Pro-Environmental Behaviour
McFarland et al. (2019) summarised definitions and theories regarding global identity. We refer to the social identity approach (SIT and SCT). SIT describes group memberships as essential parts of identity. People identify with certain social groups—their ingroups—and distinguish themselves from outgroups (Tajfel & Turner, 1979). SCT differentiates three identity levels: personal, social group, and global identity as a human. Hence, all of humanity is understood as a superordinate ingroup people can feel more or less part of (Turner et al., 1987).
By global identity we mean identification with and concern for people all over the world (McFarland et al., 2012), including the dimensions of global self-definition (i.e., identification and connectedness) and global self-investment (i.e., solidarity and care; Reese et al., 2015). According to SIT, individuals tend to benefit their ingroup (Tajfel & Turner, 1979). As globally identified people regard all humanity as their ingroup, they should be motivated to act pro-environmentally to benefit all humanity (Batalha & Reynolds, 2012; Reese, 2016).
Pro-environmental behaviour means committing acts that protect the environment (e.g., conserving energy) and omitting harmful acts (e.g., flying; Lange & Dewitte, 2019). Research often differentiates private (e.g., consumption) from public behaviour encompassing policy support and activism (Stern, 2000). By pro-environmental policy support we mean approval of policies aiming at a better life in a healthy environment (e.g., carbon pricing, pro-environmental subsidies; Drews & van den Bergh, 2016). Pro-environmental activism comprises “civic behaviors that are focused on systemic causes of environmental problems and the promotion of environmental sustainability through collective efforts” (Alisat & Riemer, 2015, p. 14), such as participating in demonstrations or signing petitions.
Past research found that the stronger people’s self-reported global identity, the more pro-environmental behaviour/intentions they indicated. For example, global identity correlated positively with a pro-environmental lifestyle covering a range of behaviours (Lee et al., 2015; Loy, Clemens et al., 2022; Loy & Reese, 2019; Loy, Reese et al., 2022), refraining from flying (Loy et al., 2021), intending to reduce clothing consumption (Joanes, 2019), food waste (Pong, 2021), and animal products (Römpke et al., 2019). Moreover, a stronger global identity was associated with support for pro-environmental policies (Loy, Clemens et al., 2022; Loy et al., 2021; Loy & Reese, 2019). Less research exists on global identity’s relation with pro-environmental activism. However, Renger and Reese (2017) found that the more study participants identified with people all over the world, the stronger their intentions for pro-environmental activism. The degree to which people regard themselves as “world citizens”—a concept similar but not equivalent to global identity (see Carmona et al., 2020, for details)—was positively related to pro-environmental activism (Rosenmann et al., 2016).
The Role of Inclusive Language for Global Identity Salience
SCT states that situational cues can make aspects of people’s identity salient (i.e., top of mind, accessible), which then more likely guide behaviour (Turner et al., 1987). Therefore, some experiments investigated how to situationally increase global identity salience. After exposure to experimental cues, self-reported global identity was measured—sometimes using instruments equivalent to measures in the correlational research mentioned above, sometimes slightly adapted by asking how participants think and feel “now, in this moment” (but it has also been questioned whether situational accessibility can be measured this way; Loy & Spence, 2020). For example, participants who looked at themselves in a mirror and saw a poster with hands from different ethnical backgrounds holding a globe or several national flags in the background reported stronger global self-investment and donated more money to a global charity, compared to a control group. Seeing the posters indirectly predicted donations through stronger global self-investment (Reese et al., 2015). Further ways to increase global identity salience were asking people about past international travel experiences (Loy et al., 2021), or letting them interact with a person from another continent in a simulated chat (Römpke et al., 2019).
Moreover, language usage has been investigated. Participants formed meaningful sentences from sets of scrambled words. These sets included words linked to either local or global identity. Global identity was relatively more accessible than local identity in the global condition (Tu et al., 2012). Seyranian (2014) tested the effect of inclusive language (in this case referring to social identity as students at a certain college) on pro-environmental behavioural intentions. Inclusive language alludes to one’s ingroup and builds a feeling of “who we are”, due to the use of collective pronouns (e.g., “we”) and words like “community” (Seyranian, 2014). Seyranian (2014) argued that the use of collective pronouns is essential because it increases a feeling of familiarity (Housley et al., 2010) and overpowers the separation between communicator and listeners (Cheney, 1983). Moreover, inclusive language is free from prejudices and derogatory speech but focuses on respectful and empathic messaging (Barcena et al., 2020). Using inclusive language referring to students’ social identity (compared to individual language referring to personal identity, e.g., “I”) in a speech advocating renewable energies fostered intentions to collectively promote renewable energies. People in the inclusive (vs. non-inclusive) language condition were more confident about social change (Seyranian, 2014). Thus, inclusive language may have the potential to promote pro-environmental activism and policy support.
An essential part of people’s motivation for activism may be their ingroup identification with other environmentalists or the climate movement in general (Bamberg et al., 2015; see Fritsche et al., 2018, for an overview). SIT argues that social groups tend to distinguish themselves from outgroups, often through language (Tajfel & Turner, 1979). Such outgroup discrimination can be observed at climate demonstrations, where activists use accusatory messages (e.g., “how dare you”, “you are destroying our future”, Spaiser et al., 2022). We refer to language that devalues the outgroup as exclusive language. Past research indicates that using an accusatory “you” can lead to antagonistic responses in the recipient (Kubany et al., 1992). Therefore, inclusive language may be more effective than exclusive language to convince people who do not identify with the climate movement of the urgency to collectively act against climate change. However, due to people’s tendency to devalue outgroups, people who identify strongly with the climate movement might not be demotivated by exclusive language that devalues their outgroup.
Global Identity and Climate Denial
Climate denial has been described as a self-protective strategy when psychological resources are lacking to endure the threat of climate change (Wullenkord, 2022; Wullenkord & Reese, 2021). Climate denial ranges from literal forms (i.e., denial of hard facts), over interpretive forms (i.e., distortion of severity), to implicatory forms (i.e., recognition of facts but denial of implications, but see e.g., Opotow & Weiss, 2000, for another categorisation). Currently, most Germans do not literally deny that anthropogenic climate change exists or distort its severity (Wullenkord, 2022). Still, many people express a subtler implicatory climate denial (Norgaard, 2011): they deny the moral, psychological, and political implications of climate change and their responsibility to behave in a climate protective manner (e.g., through rationalisation of own involvement—“although climate change is a big problem, it is not my responsibility to do something about it”—and avoidance of information or thoughts about climate change, Wullenkord & Reese, 2021). Although implicatory denial may appear less severe, it may be a barrier to climate mitigation if it results in inaction. In fact, the more people engage in all types of denial, the less pro-environmental behaviour they report (Wullenkord & Reese, 2021).
Past research also indicates that climate denial is related to identity. For example, environmental identity (Nartova-Bochaver et al., 2022) and environmental movement identity (Krange et al., 2019) were negatively related to literal and interpretive climate denial; a white conservative male identity was positively related (Krange et al., 2019; McCright & Dunlap, 2011). Moreover, right-wing authoritarianism (i.e., preferring strict authorities, laws, and punishments) and social dominance orientation (i.e., preferring social hierarchies and devaluing low-status groups) were particularly strong predictors of all forms of literal and interpretive climate denial (Jylhä & Akrami, 2015; Jylhä & Hellmer, 2020), and to a lesser extent implicatory denial (Wullenkord, 2022).
Global identity is negatively related to social dominance orientation and right-wing authoritarianism (McFarland et al., 2019), which makes a negative relation to climate denial plausible. However, the relation between global identity and (implicatory) climate denial has, to our knowledge, not been examined. The stronger people’s identification with all humanity, the more they might consider threats to all humanity as relevant and feel responsible to help solve them. In line with this thought, global identity was positively related to how relevant people evaluated climate change for them personally and societally (Loy, Reese et al., 2022), to how strongly they behaved in a climate protective manner (Loy, Reese et al., 2022), and supported climate policies (Loy, Clemens et al., 2022). Moreover, global identity was positively related to a sense of responsibility to be actively involved in global issues (Reysen & Hackett, 2016; Reysen & Katzarska-Miller, 2013), and solidarity with victims of climate change injustice and willingness to collectively engage on behalf of these victims (Barth et al., 2015). Hence, global identity might be associated with acknowledging psychological, political, or moral implications of climate change and thus less implicatory climate denial.
Current Research
In the current research, we experimentally examined impacts of using inclusive (vs. exclusive) language in the context of advocating climate protection. We hypothesised:
-
Inclusive (vs. exclusive) language use increases the salience of global identity (H1), intentions for pro-environmental activism (H2a), and pro-environmental policy support (H2b).
-
Inclusive (vs. exclusive) language use indirectly increases intentions for pro-environmental activism (H3a) and policy support (H3b) through an increase in global identity salience.
-
The language effect is moderated by people’s identification with the climate movement: if participants identify with the climate movement, inclusive language does not increase intentions for pro-environmental activism (H4a) and pro-environmental policy support (H4b), while it does if participants do not identify with the climate movement.
-
Inclusive (vs. exclusive) language use reduces implicatory climate denial indirectly through an increase in global identity salience (H5).
-
The language effect is moderated by people’s identification with the climate movement; if participants identify with the climate movement, inclusive language does not reduce implicatory climate denial, while it does if participants do not identify with the climate movement (H6).
Method
Design and Procedure
This pre-registered online study (see Bauer & Loy, 2024) adheres to the APA’s ethical principles and was approved by the university’s ethics committee within a research program (#2020-297). We acknowledge that the pre-registration did not determine all analytical decisions. We used the software SoSci Survey (Leiner, 2024). After giving informed consent, participants rated their identification with the climate movement, and were randomly assigned to one of three groups: 1) inclusive language, 2) exclusive language, 3) control group. In the experimental conditions, participants evaluated four demonstration placards with either inclusive or exclusive language. Afterwards, they reported their global identity, pro-environmental activism intentions, pro-environmental policy support, and climate denial. Participants in the control group evaluated the placards after the dependent variables. This order allowed for an assessment of the dependent variables without prior language cues, while keeping study duration constant and avoiding suspicion regarding study contents (the study was advertised as a study about placards). All participants received control questions about the stimulus material and socio-demographic characteristics. Finally, they had the possibility to sign a fictitious petition as a situational indicator of pro-environmental behaviour. They were debriefed, could participate in a lottery of four 25 Euro vouchers, and could ask for data deletion. Psychology students could receive class credit (though the study was not restricted to students).
Participants
We recruited a convenience sample through student mailing lists, social media, and personal contacts. A priori power analyses suggested a sample of at least 300 participants (see Loy et al., 2024, Supplement A). After data exclusion (see Figure 1 and Loy et al., 2024, Supplement B), the sample consisted of N = 307 participants (213 female, 90 male, 4 diverse, Mage = 29.91 years, SD = 12.29, Range = 18–75). The majority was highly educated: 47% with (applied) university entrance certificate, 46% with (applied) university degree. More than half were students (60%), followed by employees (30%). Almost all participants lived in Germany (98%; see Loy et al., 2024, Supplement B for a detailed sample description).
Figure 1
Participant Flow Depicting the Allocation of Participants to the Experimental Conditions and the Number of Excluded Participants per Criterion.
Stimulus Material
We created placards in the style found at climate demonstration (see Figure 2). Four contained slogans with inclusive language, matched to four with exclusive language. Participants rated each placard regarding nine characteristics (e.g., creativity), and stated which of the four placards, if any, they would use themselves. These questions were supposed to make people examine the placards closely. To be able to exclude participants with technical problems, we asked whether pictures were viewable without pixelation and with sufficient detail. We also asked participants to recall the slogans in writing.
Figure 2
Placards With Inclusive vs. Exclusive Language in German (English).
Measures
Scale descriptives are provided in Table 1, detailed analyses in Loy et al., 2024, Supplement C, variable wording and codings alongside data and script in Loy and Wullenkord (2024).
Table 1
Psychometric Properties of the Measures
| Variable | n | M | SD | Range | Item number | α | ω | AVE | RP |
|---|---|---|---|---|---|---|---|---|---|
| Identification with the climate movement | 307 | 4.76 | 1.52 | 1.00–7.00 | 4 | .91 | .91 | .73 | |
| Global identity (1-dim) | 307 | 5.19 | 1.12 | 1.10–7.00 | 10 | .92 | .93 | .56 | |
| Global identity (2-dim) | |||||||||
| Self-definition | 307 | 4.84 | 1.29 | 1.20–7.00 | 5 | .88 | .88 | .61 | |
| Self-investment | 307 | 5.53 | 1.09 | 1.00–7.00 | 5 | .88 | .89 | .61 | |
| Pro-environmental activism intention | 306 | 2.63 | 0.70 | 1.00–4.61 | 18 | .92 | .91 | .41 | |
| Pro-environmental activism intention–Rasch score | 307 | 0.03 | 1.88 | -5.56–5.52 | 18 | .84 | |||
| Pro-environmental policy support | 306 | 5.58 | 0.98 | 2.16–7.00 | 19 | .92 | .90 | .39 | |
| Pro-environmental policy support–Rasch score | 307 | 0.77 | 1.79 | -4.23–4.16 | 19 | .83 | |||
| Implicatory climate denial (2-dim) | |||||||||
| Rationalisation | 307 | 2.94 | 1.21 | 1.00–7.00 | 7 | .87 | .84 | .48 | |
| Avoidance | 307 | 2.82 | 1.16 | 1.00–6.88 | 8 | .91 | .91 | .56 |
Note. α = Cronbach’s alpha, ω = McDonald’s omega, AVE = average variance extracted, RP = person separation reliability based on Rasch model.
Identification With the Climate Movement
We included three items used by Wallis and Loy (2021), extended by one further item (e.g., “I feel a strong connection to others who engage in environmental and climate protection”). Participants rated the items on a Likert scale from 1 (does not apply at all) to 7 (fully applies).
Global Identity
We used an adapted version (Loy & Reese, 2019) of the Identification with all Humanity Scale (McFarland et al., 2012; Reese et al., 2015). Participants rated five items on global self-definition (e.g., “I feel connected to people all over the world”) and five items on global self-investment (e.g., “I want to help people all over the world”) on a Likert scale from 1 (does not apply at all) to 7 (fully applies).
Pro-Environmental Activism Intentions
We selected items from the Environmental Action Scale (Alisat & Riemer, 2015) and the General Ecological Behaviour Scale (Kaiser & Wilson, 2000), added own items, and rephrased behaviour to behavioural intentions. Participants rated 18 items, e.g., “I will participate in demonstrations for climate protection, (e.g., Fridays for Future)”, on a Likert scale from 1 (never) to 5 (very often).1
Pro-Environmental Behaviour
As a situational indicator, participants stated their willingness to sign a petition requesting the government to prioritize the 1.5°C limit of the Paris Agreement in the upcoming legislation period (yes, n = 171; no, n = 125; missing, n = 11). Later, they were debriefed that the petition was fictitious.
Pro-Environmental Policy Support
We extended scales used by Tobler et al. (2012) and Loy, Clemens et al. (2022). Participants rated 19 items (e.g., “Introduction of a general speed limit of 130 km/h on highways”) on a Likert scale from 1 (fully against) to 7 (fully in favour).
Climate Denial
We used the Climate Self-Protection Scale (Wullenkord & Reese, 2021), including implicatory and interpretive denial. Participants rated 26 items2 on a Likert scale from 1 (fully disagree) to 7 (fully agree). As pre-registered, we analysed implicatory denial: 1) rationalisation (e.g., “My personal influence on climate change is negligible”), 2) avoidance (e.g., “I try not to think about climate change”), 3) denial of guilt (e.g., “I don’t need to make climate change a matter of conscience”). We excluded denial of guilt due to poor model fit (see Loy et al., 2024, Supplement C).
Results
Correlations Between Study Variables
Table 2 shows bivariate correlations.
Table 2
Correlations With Confidence Intervals
| Variable | 1 | 2 | 3 | 4 | 5 | 6 |
|---|---|---|---|---|---|---|
| 1. Identification climate movement | ||||||
| 2. Global self-definition | .46** [.37, .55] | |||||
| 3. Global self-investment | .56** [.47, .63] | .75** [.70, .80] | ||||
| 4. Pro-environmental activism intention | .69** [.62, .74] | .35** [.25, .45] | .48** [.39, .56] | |||
| 5. Pro-environmental policy support | .61** [.53, .68] | .37** [.27, .46] | .50** [.41, .58] | .64** [.57, .70] | ||
| 6. Climate denial: rationalisation | -.40** [-.49, -.30] | -.29** [-.39, -.19] | -.31** [-.41, -.20] | -.37** [-.46, -.27] | -.39** [-.48, -.29] | |
| 7. Climate denial: avoidance | -.13* [-.24, -.02] | -.05 [-.16, .06] | -.02 [-.13, .09] | -.19** [-.30, -.08] | -.08 [-.19, .03] | .29** [.19, .39] |
*p < .05. **p < .01.
Group Comparison
Participants assigned to the three groups did not differ in the distribution of gender, age, and identification with the climate movement (ps ≥ .388), indicating successful randomization. Table 3 displays the dependent variables per group. As we did not specify scoring in the pre-registration, we report both mean scores as well as factor and Rasch scores resulting from scale analyses (see Loy et al., 2024, Supplement C)3. One-way ANOVAs did not reveal any significant group differences (see Table 3). Specifically testing H1, H2a, and H2b using one-tailed t-tests (not pre-registered), we did not find significant effects of inclusive (vs. exclusive) language, despite tendencies in the expected direction particularly for global self-definition, pro-environmental policy support, and rationalisation (ps < .10; see Loy et al., 2024, Supplement D). Additionally, we conducted a Chi-squared test comparing groups’ willingness to sign the petition. We found no group difference (χ2(4) = 0.39, p = .983).
Table 3
One-way ANOVAs Comparing Dependent Variables Between the Groups
| Inclusive language
|
Exclusive language
|
Control group
|
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Variable | M | SD | M | SD | M | SD | F | p | η2 |
| Global self-definition | 5.01 | 1.18 | 4.72 | 1.30 | 4.81 | 1.37 | 1.23 | .269 | .004 |
| Global self-definition–factor scores | 0.17 | 1.16 | -0.11 | 1.30 | -0.05 | 1.41 | 1.48 | .225 | .005 |
| Global self-investment | 5.60 | 0.93 | 5.49 | 1.03 | 5.51 | 1.28 | 0.39 | .532 | .001 |
| Global self-investment–factor scores | 0.10 | 0.96 | -0.06 | 1.07 | -0.04 | 1.29 | 0.79 | .375 | .003 |
| Pro-environmental activism intentions | 2.67 | 0.71 | 2.62 | 0.65 | 2.59 | 0.73 | 0.66 | .418 | .002 |
| Pro-environmental activism intentions–Rasch scores | 0.14 | 1.84 | -0.01 | 1.76 | -0.04 | 2.03 | 0.46 | .496 | .002 |
| Pro-environmental policy supporta | 5.75 | 0.76 | 5.51 | 1.04 | 5.50 | 1.07 | 2.56 | .080 | .010 |
| Pro-environmental policy support–Rasch scores | 0.96 | 1.59 | 0.64 | 1.93 | 0.71 | 1.82 | 0.98 | .323 | .003 |
| Climate denial–rationalisation | 2.86 | 1.16 | 3.10 | 1.29 | 2.86 | 1.18 | <0.01 | .979 | <.001 |
| Rationalisation–factor scores | -0.05 | 0.76 | 0.10 | 0.88 | -0.05 | 0.79 | <0.01 | .956 | <.001 |
| Climate denial–avoidance | 2.94 | 1.17 | 2.71 | 1.10 | 2.81 | 1.20 | 0.60 | .439 | .002 |
| Avoidance–factor scores | 0.13 | 1.21 | -0.10 | 1.15 | -0.02 | 1.23 | 0.67 | .414 | .002 |
aWelch ANOVA due to unequal variances.
Mediation Analyses
We found no indirect effects of inclusive (vs. exclusive) language through higher global self-definition or self-investment on any of the dependent variables (see Table 4 and Loy et al., 2024, Supplement E for detailed results). Hence, we found no support for H3a, H3b, or H5.
Table 4
Indirect Effects of Inclusive (vs. Exclusive) Language
| Hypotheses | Indirect relations | B | SE | p | 95% CI | β |
|---|---|---|---|---|---|---|
| H3a | IL – GSD – PEAI | 0.06 | 0.04 | .108 | [-0.13, 0.13] | .04 |
| IL – GSD – PEAIa | 0.16 | 0.10 | .131 | [-0.05, 0.36] | .04 | |
| IL – GSI – PEAI | 0.04 | 0.05 | .362 | [-0.05, 0.14] | .03 | |
| IL – GSI – PEAIa | 0.13 | 0.11 | .270 | [-0.10, 0.35] | .04 | |
| H3b | IL – GSD – PEPS | 0.07 | 0.04 | .115 | [-0.02, 0.16] | .04 |
| IL – GSD – PEPSa | 0.13 | 0.09 | .123 | [-0.04, 0.30] | .04 | |
| IL – GSI – PEPS | 0.05 | 0.06 | .347 | [-0.06, 0.17] | .03 | |
| IL – GSI – PEPSa | 0.11 | 0.10 | .269 | [-0.08, 0.30] | .03 | |
| H5 | IL – GSD – R | -0.07 | 0.05 | .127 | [-0.16, 0.02] | -.03 |
| IL – GSD – Ra | -0.05 | 0.03 | .131 | [-0.12, 0.02] | -.03 | |
| IL – GSI – R | -0.04 | 0.05 | .428 | [-0.14, 0.06] | -.02 | |
| IL – GSI – Ra | -0.04 | 0.03 | .283 | [-0.10, 0.03] | -.02 | |
| IL – GSD – A | -0.03 | 0.03 | .238 | [-0.08, 0.02] | -.01 | |
| IL – GSD – Aa | -0.03 | 0.03 | .265 | [-0.08, 0.02] | -.01 | |
| IL – GSI – A | -0.01 | 0.01 | .508 | [-0.04, 0.02] | -.00 | |
| IL – GSI – Aa | -0.01 | 0.02 | .422 | [-0.05, 0.02] | -.01 |
aFactor/Rasch scores. IL = inclusive (vs. exclusive) language, GSD = global self-definition, GSI = global self-investment, PEAI = pro-environmental activism intentions, PEPS = pro-environmental policy support, R = climate denial: rationalisation, A = climate denial: avoidance. Confidence intervals were bootstrapped through 10,000 samples.
Moderation Analyses
Next, we examined identification with the climate movement as a moderator, regressing the dependent variables on language (inclusive vs. exclusive), identification, and the interaction term.4 Details are reported in Loy et al. (2024), Supplement F.
There were no significant conditional effects of language on pro-environmental activism intentions, policy support, or climate denial in the forms of rationalisation or avoidance (ps ≥ .065). Moreover, people’s identification with the climate movement did not moderate language effects on activism intentions (H4a), policy support (H4b), and rationalisation or avoidance (H5; ps ≥ .290). We found conditional effects of identification on activism intentions, mean scores: B = 0.31, SE = 0.03, 95% CI [0.24, 0.37], p < .001; factor scores: B = 0.90, SE = 0.10, 95% CI [0.68, 1.13], p < .001; policy support, mean scores: B = 0.37, SE = 0.05, 95% CI [0.22, 0.48], p < .001; factor scores: B = 0.81, SE = 0.11, 95% CI [0.55, 1.02], p < .001; and rationalisation, mean scores: B = -0.33, SE = 0.08, 95% CI [-0.52, -0.14], p < .001; factor scores: B = -0.24, SE = 0.06, 95% CI [-0.39, -0.11], p < .001; but not avoidance, mean scores: B = -0.12, SE = 0.08, 95% CI [-0.27, 0.04], p = .121; factor scores: B = -0.14, SE = 0.09, 95% CI [-0.32, 0.04], p = .119.
Discussion
We investigated whether using inclusive (vs. exclusive) language in the context of demonstrations for climate protection impacts how much people identify with people all over the world, intend to engage in pro-environmental activism, support pro-environmental policies, and deny climate change implications.
Contradicting H1, we found no significant impacts of language use on global identity salience. This contrasts prior research showing effects of language use on people’s momentary global identification (Tu et al., 2012). However, in this prior study, the examined wording directly referred to global identity, while our wording was generally inclusive without using terms like “global”. Moreover, a sentence-scrambling task might be a more active way to process language than evaluating placards. Further differences could consist in language per se, cultural background, the sample, or the measure. While Tu et al. (2012) conducted their study in English with US students, our study was conducted in German with a more heterogeneous convenience sample, using a different measure. Recent research showed that using different measures for global identity might indeed imply different meanings (Carmona et al., 2020). Finally, using a generic “we” without naming the group (i.e., all of humanity) might have resulted in other identities than global identity becoming salient (e.g., we as activists, we as Germans).
We did not find evidence that inclusive (vs. exclusive) language fosters pro-environmental activism intentions and policy support (H2). This contrasts findings that inclusive language promoted collective action intentions favouring renewable energies (Seyranian, 2014). However, this prior study referred to the social group of college students, while our study aimed to address a more inclusive identity—global identity. Furthermore, Seyranian (2014) compared inclusive to individual language, while we investigated differences to exclusive language and a control group. Moreover, our study differs regarding language and context, as Seyranian conducted her study in English with US students.
In the absence of effects of inclusive language on global identity and pro-environmental engagement, we found no support for H3 assuming an indirect positive effect on engagement through stronger identity salience. However, our results show that the more people identify globally, the more they intend to become pro-environmentally active and support pro-environmental policies. Thereby, our study extends prior research on global identity and pro-environmental behaviour (e.g., Joanes, 2019; Loy, Reese et al., 2022; Pong, 2021; Römpke et al., 2019) and adds to the few results on activism (Renger & Reese, 2017; Rosenmann et al., 2016). Still, we cannot infer a causal direction of this relationship. Global identity might promote pro-environmentalism and/or vice versa.
Contradicting H4, language effects did also not depend on people’s prior identification with the climate movement. Still, identification with people who engage in environmental and climate protection was strongly related to participants’ own intentions to engage in activism in the future. This is in line with studies showing that social identification is one of the strongest predictors of participating in the Swiss and German Fridays For Future demonstrations (Brügger et al., 2020; Wallis & Loy, 2021).
Contradicting H5, we found no evidence for language use to decrease denial of climate change implications through global identity. Nevertheless, rationalisation (i.e., the claim that one’s own actions do not meaningfully contribute to climate change) was negatively related to global identity; avoidance of information and thoughts about climate change was not. While findings for rationalisation were in line with our reasoning, findings for avoidance were not. One may speculate that some people scoring high on avoidance (but not rationalisation) might use avoidance to maintain or increase their engagement in the face of climate change (maintaining functionality, see also Pihkala, 2022; Wullenkord & Ojala, 2023). The correlations of variables in this sample support this assumption.
Limitations and Future Directions
Our study is limited by its highly educated, mostly female convenience sample comprising many students. These characteristics are also often found amongst climate activists (Wahlström et al., 2019). Moreover, the average level of identification with pro-environmental activists was rather high in our sample. Thus, future studies should recruit more diverse samples with less prior contact to the climate movement through quota sampling or, ideally, random sampling. This research could examine generational differences in the assessed variables, and whether age effects differ for activism compared to private pro-environmental behaviour (e.g., Otto & Kaiser, 2014). Additionally, prior experience with demonstrations should be assessed—as active participant or passive observer. Thereby, it will be possible to examine whether inclusive language might be a means to address people without or with only little prior experiences. Moreover, it would be useful to ask in how far people felt personally addressed by the slogans. It is possible that most participants did not feel derogated by the exclusive wording because they saw themselves as part of the ingroup of climate activists.
Relatedly, future studies should assess other social identities that may be evoked by the wording (e.g., we as activists, we as the young generation). Experimental research could vary whether and which social categories are addressed in slogans (e.g., none vs. humanity vs. young people etc.), in order to examine whether activists engage their audiences more if they address specific groups. Moreover, climate justice considerations should be taken into account. For example, messages such as “we as humanity should mitigate climate change” might suggest equal responsibilities of Global North and South, in light of unequal carbon footprints and (historical) contributions to climate change. Phrases such as “you should mitigate climate change” might exaggerate individual responsibilities, in light of structural and systemic barriers.
Average levels of global identity and pro-environmental policy support were high in our sample, which limits the possibility to observe effects of the experimental stimuli. This is another argument for more diverse samples. Moreover, it will be valuable to examine country differences regarding the meaning of global identity and how it is related to pro-environmentalism. For example, the meaning of the human category could differ between individuals (Bain, 2013) but also between countries (IWAH, 2024). It would be worthwhile to compare countries that are currently more or less impacted by adverse effects of climate change and more or less active climate movements.
Future studies could also use larger samples. We had based our power considerations on roughly related prior research, which had examined different variables. Even though we did not find significant differences between our experimental groups, descriptive examination of the mean scores partly shows tendencies towards our hypotheses, but with effect sizes undetectable given our sample size. Replications of our study could use these effect sizes for power analyses. We argue that even small effects of language use might be practically relevant and thus worth detecting, as they likely accumulate over time.
Importantly, one-time exposure to experimental stimuli might not evoke a meaningful response. Global identity in particular might have a strong trait component and develop over longer time periods with manifold influences. Thus, longitudinal research is needed, including repeated exposure to stimuli that might impact global identity and long-term assessment of experimental effects, alongside further possible influences such as personality or education (McFarland et al., 2019). A fruitful path to investigate momentary changes in the salience of global identity compared to other identity aspects might be social identity switching (Zinn et al., 2022), for example, by sequentially addressing different identity levels in slogans.
Finally, the setting and design of our study needs critical reflection. External validity could be strengthened by examining the impact of placard wording directly at demonstrations or through showing videos of demonstrations. These contexts might elicit stronger feelings in viewers of demonstration placards. Internal validity could be increased in laboratory experiments in which the setting of study participation is controlled and less disturbances arise compared to online settings. Moreover, language use could be investigated not only in placards but also, for example, in speeches or flyers.
Practical Implications
How can climate activists motivate others to engage in climate protection? Should they use “you” or “we” in their communications? On the one hand, we cannot infer that inclusive language matters from our study. We did not find any significant effects of using inclusive “we” slogans on demonstration placards on participants’ intentions to engage in climate activism or their support for political change—compared to derogatory “you” slogans or no messages. On the other hand, we cannot infer that language use does not matter, either. Rather, further research with varying methodology on this topic is needed (see suggestions above) to clarify, for example, the following practically relevant questions: Does an accusatory “you” (e.g., “you destroy the planet”) demotivate people without prior experience with climate demonstrations, who feel personally devalued? Should climate activists address specific social groups with their messages?
Conclusion
Using inclusive “we” slogans (vs. accusatory “you” slogans) on climate demonstration placards did not impact how much participants identified with people all over the word, intended to engage in climate activism, supported climate policies, or denied climate change implications. However, the stronger people’s global identity, the stronger their activism intentions and policy support, and the weaker their denial of their own impact on climate change. Based on our findings, we suggest finding ways to increase global identity because such an increase might contribute to climate protection (even though our results are not conclusive about the causal direction). Using inclusive language in the context of climate demonstrations does not seem to make a significant difference. However, we argue that it is too early to abandon the idea that people can be better motivated by inclusive words welcoming them, than by exclusive words derogating them. Rather, we hope that we could encourage other scholars to explore this research field, using diverse samples, designs, and measures.