The effects of news report valence and linguistic labels on prejudice against social minorities

ABSTRACT

Combating prejudice against social minorities is a challenging task in current multicultural societies. Mass media can decisively shape prejudice, because it often represents the main source of information about social minorities. In 3 studies in the Czech Republic (N = 445) and Switzerland (N = 362; N = 220), we investigated how prejudice against negatively and positively perceived minorities (the Roma in Study 1, Kosovo Albanians in Study 2, Italians in Study 3) is influenced by a single exposure to a print news report, by manipulating the valence of reports about minority members (positive vs. negative vs. mixed) and linguistic forms for minorities’ ethnicity (nouns vs. adjectives). Positive and negative reports shaped prejudice in the respective directions; the effect of mixed reports mostly did not differ from positive reports. Labeling ethnicity with nouns (e.g., a male Roma) resulted in more prejudice than adjectives (e.g., a Roma man), independent of report valence. Report valence influenced the affective part of prejudice (i.e., feelings toward a minority), whereas language consistently shaped the behavioral part of prejudice (i.e., preferred social distance from a minority).

The proportion of social minorities in Western Europe is growing faster than ever before (United Nations High Commission for Refugees, 2016). For harmonious coexistence of different groups, citizens need to come to terms with the diversification of society, which implies combating prejudice against social minorities (i.e., groups that differ in observable characteristics or practices, such as ethnicity or religion, from the overall population; Goldmann, 2001). Mass media represent a profound source of information about social minorities that can shape prejudice against them (e.g., Das, Bushman, Bezemer, Kerkhof, & Vermeulen, 2009; Visintin, Voci, Pagotto, & Hewstone, 2017).

Despite the impact of mass media on shaping public opinion, there is little research that captures the causal effect of different media features on attitudes toward different social groups (Atwell Seate & Mastro, 2016; Mutz & Goldman, 2010; cf. Saleem, Prot, Anderson, & Lemieux, 2017; Schmader, Block, & Lickel, 2015). Past studies have usually focused on the impact of the content of media news about social minorities, mostly in terms of their valence (e.g., Joyce & Harwood, 2014; Ramasubramanian, 2011, 2015). However, there are subtler features that can influence prejudice beyond the content of news, such as language employed for description of social minorities (Geschke, Sassenberg, Ruhrmann, & Sommer, 2010; Graf, Bilewicz, Finell, & Geschke, 2013). So far, these two important aspects of media news—their valence and language used for describing group members—have been examined in isolation. This research focuses on how prejudice against social minorities can be shaped by the language used for minorities’ group membership (i.e., nouns vs. adjectives) depending on the valence of print news reports (i.e., positive, negative, vs. mixed).

Social minorities in mass media

The discrimination that social minorities suffer in society is reflected in their underrepresentation or negative presentation of their members in mass media (e.g., Larson, 2006; Tukachinsky, Mastro, & Yarchi, 2015). Particularly negatively stereotyped social minorities (e.g., immigrants) are mostly presented in report-based media programs that introduce predominantly negative information (e.g., Dixon & Linz, 2000; Mastro & Greenberg, 2000; Pagotto & Voci, 2013; Visintin et al., 2017). A large analysis of print newspapers addressing immigration in the United States showed that there was a greater amount of negative information about immigrants than positive information (Dragojevic, Sink, & Mastro, 2017). The prevalence of negativity over positivity was more pronounced in information about outgroups compared to ingroups, which suggests that the effect was not simply due to a general tendency of news to cover more negative than positive events. The prevalence of negative information in the news about social minorities is alarming, given the fact that already a single exposure to a negative report about a minority can deteriorate attitudes toward this minority (Mastro, 2009; Saleem et al., 2017).

The situation is further complicated by the fact that the news often represents the major source of information about social minorities. For instance, Italian citizen’s direct contact with immigrants (i.e., face-to-face encounters) and information from films and TV series were much scarcer than information from TV news and newspapers. When distinguishing between positive and negative information about immigrants, negative information from TV news and newspapers was the most frequent type of information about immigrants in Italy (cf. other types of information in Table 4 on p. 184, Visintin et al., 2017). Consequently, other sources of information (e.g., from entertainment-based programs or face-to-face encounters) that are more balanced with respect to the valence of their content (e.g., featuring more positive information) may fail to improve the negative representation of social minorities spread by the news. The prevailing negativity and the widespread nature of news about social minorities can bear direct implications for the difficult standing of social minorities in society (Atwell Seate & Mastro, 2017; Dixon & Azocar, 2007; Pan & Kosicki, 1996).

Following the prevalence of negativity in media news, past studies have predominantly dealt with the effects of negative information (e.g., Atwell Seate & Mastro, 2016; Das et al., 2009; Dixon & Azocar, 2007; for a review see Mastro & Tukachinsky, 2011). However, to understand the effect of media news, it is necessary to compare the effect of negative news to information of other valences. An experimental study comparing the effect of negative, positive, and neutral media footage about Muslims showed that negative information, as compared to neutral and positive conditions, increased perceptions of Muslims as aggressive, as well as support for harsh actions against Muslims in the home country and abroad (Saleem et al., 2017, Study 3). Despite the merit of Saleem et al.’s study, it is challenging to present participants with a truly neutral condition, especially in the context of negatively perceived minorities. Even seemingly neutral news footage about a delay of football practices during Ramadan in Saleem et al.’s study may constitute an infringement of ingroup (i.e., U.S.) values for many and thus lean toward negativity. Moreover, many news reports are not exclusively negative, containing positive and negative information at the same time. Nevertheless, the effect of such mixed valence in news reports has been largely ignored. The lack of testing negative reports against not only positive but also against mixed valence reports is a substantial gap in the literature.

To our knowledge, only one study has so far simultaneously compared the effect of positive, negative, and mixed information about social minorities on prejudice (Joyce & Harwood, 2014). In the context of the U.S. border to Mexico, U.S. participants viewed a TV documentary depicting interactions between a border-patrolmen and an illegal immigrant that was either positive, negative, or mixed. Only the positive TV documentary improved participants’ attitudes toward illegal immigrants; the documentary with mixed content did not differ from the effect of negative content or control group. Until now, evidence is missing whether this pattern of results would hold for different media content, such as print news, and in different intergroup contexts.

Nouns and adjectives communicate group membership

Although the valence of media news about social minorities is obvious, other subtler linguistic means that convey implicit biases about other groups and their members can escape conscious awareness (Franco & Maass, 1996, 1999; Geschke et al., 2010; Maass, Corvino, & Arcuri, 1994). Two linguistic categories that play a crucial role in communicating group membership are nouns and adjectives. Although both nouns and adjectives can be employed to convey membership in a certain social category, nouns have greater inductive potential when it comes to inferring characteristics from the given label than adjectives (e.g., a Jew vs. Jewish; Carnaghi et al., 2008). Nouns support category-congruent inferences (e.g., attends synagogue regularly) and inhibit category-incongruent inferences (e.g., works on Fridays). The same behavior (e.g., likes to work with plaster) is judged as a more prominent and stable feature of a person described with a noun (e.g., an artist) than of a person described by a corresponding adjective (e.g., artistic; Carnaghi et al., 2008).

Nouns for category labels influence not only the perception of persons, but also of groups. Nouns, as compared to adjectives, used for labeling nationality or ethnicity lead to more pronounced intergroup bias—the tendency to favor one’s ingroup over outgroups (Graf et al., 2013). When asked who should have confiscated property from the Second World War returned—whether the ingroup or the outgroup—a representative Polish sample consistently displayed ingroup favoritism. People always favored returning confiscated property to their ingroup (i.e., ethnic Poles) over the outgroup (i.e., ethnic Jews), irrespective of linguistic labels used for ingroup and outgroup membership. Yet, the difference between favoring the ingroup over the outgroup was larger in the noun (i.e., Poles over Jews) compared to the adjective condition (i.e., Polish citizens over Jewish citizens; Graf et al., 2013, Study 3). This implies that subtle linguistic cues do not change the direction of firmly embedded intergroup preferences; however, they are able to make them more pronounced.

The limitation of past studies on nouns and adjectives is that we do not know whether their effects depend on contextual factors such as valence. So far, no research has contrasted the effect of nouns and adjectives in positive and negative texts at the same time. Nouns may always lead to more pronounced prejudice against social minorities or they may strengthen prejudice only in the context of specific valence (e.g., in the previously tested negative contexts).

The present research

To increase the current understanding of how mass media shape prejudice against social minorities, this research examined the effect of print news reports about distinct social minorities in three experimental studies. Specifically, we focused on the effects of report valence and language used to describe the group membership of minority members.

We expected that a single exposure to a report of distinct valence should lead to different levels of prejudice (Hypothesis 1). Specifically, positive reports should lead to less prejudice than negative or mixed reports; and mixed and negative reports should not significantly differ (based on Joyce & Harwood, 2014). With respect to the effect of linguistic forms in contexts of distinct valence, we expected that nouns should lead to more prejudice than adjectives irrespective of report valence (Hypothesis 2; based on Graf et al., 2013; see Table 1 for the summary of predictions).

We distinguished between two components of prejudice that can bear different consequences for behaviors toward members of social minorities. We focused on the affective component of prejudice (i.e., feelings toward social minorities) because it generally represents a stronger predictor of intergroup behavior than cognitive components (i.e., stereotypical characteristics of social minorities; Pettigrew & Tropp, 2008; Stangor, Sullivan, & Ford, 1991). Furthermore, we used a behavioral proxy embodied by the social distance scale that captures the desired psychological distance from social minorities (Esses & Dovidio, 2002). We assumed that distinguishing between different components of prejudice offers a more nuanced understanding of how distinct aspects of print news reports impact prejudice against social minorities. Because there is no consistent evidence from previous studies, this part of our research is merely exploratory. To ensure the effect of news reports was due to experimental manipulation, we further introduced pre- and postmeasurement of prejudice in the design of Study 3, a methodological strength not often present in past studies. In Study 3, we could, therefore, examine the effect of the distinct aspects of news reports, especially in terms of their valence, on the change in prejudice.

To provide more generalizable evidence, we tested the hypotheses in two different intergroup contexts—the Czech Republic (Study 1) and Switzerland (Studies 2 and 3)—and three minority groups—the Roma (Study 1), Kosovo Albanian immigrants (Study 2), and Italian immigrants (Study 3). Although Italian immigrants in Switzerland are viewed positively, the Roma in the Czech Republic and Kosovo Albanian immigrants in Switzerland belong to the most negatively perceived groups. The Czech Republic and Switzerland represent distinct settings with respect to language and the share of social minorities. Both Czech and German are Indo-European languages, but belong to different families: Czech to Slavic and German to Germanic languages. In both languages, adjectives always precede nouns, with which they are associated. With respect to the specifics of the intergroup settings, the Czech Republic represents a country with a very low percentage of immigrants and social minorities (less than 5%; Czech Statistical Office, 2017). In contrast, Switzerland is at the opposite pole, with one of the highest share of immigrants in Europe (up to 25%; Eurostat, 2017). The differences between the three social minorities, the two countries and languages make the chosen intergroup settings optimal for a comprehensive test of how report valence and linguistic forms for labeling ethnicity in print news shape prejudice against social minorities.

Study 1

Study 1 dealt with a highly stigmatized minority in the Czech Republic, the Roma. The Roma belong to the least liked groups in the Czech Republic—only 15% of Czechs would not have a problem with the Roma as their neighbors (Stem, 2016). The stereotypical views on the Roma include both low warmth and low competence, making their image especially negative.

Method

Participants and design

The online questionnaire was presented in Czech and filled out by 588 Czech participants. Participants read one of six fictitious newspaper reports (3× report valence and 2× linguistic form for ethnicity, between participants), and indicated their attitudes toward the Roma. Because the study dealt with the effect of print news reports, we excluded participants who did not believe that the report was real (n = 143), resulting in the final sample of 445 participants (72% women, M_age = 26.18, SD = 6.35).

Procedure and materials

Participants were addressed via email; they had previously provided this along with their consent to be part of a pool for psychological studies. The research was introduced as dealing with perceptions of media reports. The print news reports were ostensibly taken from a local newspaper and informed participants about an assault featuring a Roma man (adjective condition) or a male Roma (noun condition) who was either the perpetrator (negative version) or the savior of the victim of the attack (positive version). The mixed report featured two Roma men, one being the perpetrator and the other the savior of the victim. The ethnicity of the victim and the perpetrator in the positive version, and the savior in the negative version of the report were not specified, implying majority membership. All report versions featured the same number of labels (eight) and were comparatively long (app. 200 words, see Appendix for details).

Prejudice was operationalized along its affective and behavioral components. The affective component of prejudice was measured with a feeling thermometer ranging from 0 to 100 with the anchors cold and warm (Haddock, Zanna, & Esses, 1993), higher values indicating less prejudice (i.e., warmer feelings to the Roma minority). The behavioral component of prejudice was measured with the social distance scale comprising five items (i.e., to what extent participants would be happy if the Roma were their colleagues, neighbors, friends, family members, or partners; adapted from Esses & Dovidio, 2002; α = .85). The responses ranged from 1 (not at all) to 5 (very much) with greater values indicating more prejudice (i.e., higher preferred distance from the Roma minority).

To control whether participants considered the experimental material real, they indicated whether they believed that the report was a real newspaper report with three response options: (a) I believed, (b) I doubted, (c) I did not believe that the report was real. At the end of the questionnaire, participants answered items about their demographic characteristics (age, sex, origin, and immigration background) and were debriefed.¹

Results

Throughout all three studies, p values of .05 or less were considered significant. To estimate the effect of report valence and linguistic forms for ethnicity labels on affective and behavioral components of prejudice against the Roma, we performed a MANOVA since both measures of prejudice were correlated, r = .66, p < .001. Using Pillai’s trace, the MANOVA revealed a significant effect of report valence, V = 0.02, F(4, 878) = 2.41, p = .048, and linguistic forms, V = 0.02, F(2, 438) = 4.87, p < .01, on prejudice. The interaction between report valence and linguistic forms was not significant, V = 0.01, F(4, 878) = 1.07, p = .373, indicating that the effect of linguistic forms was independent of the report valence.²

We followed up the MANOVA with univariate tests to determine the simple main effects. The ANOVAs showed that report valence predicted prejudice on the feeling thermometer, F(2, 439) = 3.64, p = .027, Eta² = .02, and the social distance scale, F(2, 439) = 4.28, p = .014, Eta² = .02 (see Hypothesis 1, Figures 1 and 2). The planned contrasts indicated that participants who read the positive report expressed less prejudice against the Roma (feeling thermometer: M = 39.15, SD = 24.43; social distance scale: M = 3.46, SD = 0.80) than participants who read the negative report (feeling thermometer: M = 32.89, SD = 23.22; social distance scale: M = 3.69, SD = 0.84), 95% CI [1.02, 11.46], p = .019 for the feeling thermometer, 95% CI [−0.41, −0.04], p = .018, for the social distance scale. However, participants who read the positive report expressed the same amount of prejudice as participants who read the mixed report (feeling thermometer: M = 39.39, SD = 20.48; social distance scale: M = 3.43, SD = 0.78), 95% CI [−4.92, 5.30], p = .943, for the feeling thermometer, 95% CI [−0.22, 0.14], p = .691, for the social distance scale. Participants who read the mixed report expressed less prejudice than participants who read the negative report, 95% CI [1.10, 11.76], p = .02 for the feeling thermometer, 95% CI [−0.45, −0.07], p = .007, for the social distance scale.

Figure 1. A significant effect of report valence (negative, mixed and positive) but not linguistic forms (nouns and adjectives) on prejudice against the Roma measured with the feeling thermometer (range 0–100; higher values indicate less prejudice) in Study 1. Error bars are standard errors of the means.

Figure 2. Significant effects of report valence (negative, mixed and positive) and linguistic forms (nouns and adjectives) on prejudice against the Roma measured with the social distance scale (range 1–5; higher values indicate more prejudice) in Study 1. Error bars are standard errors of the means.

Linguistic forms predicted prejudice on the social distance scale, F(1, 439) = 3.73, p = .050, Eta² = .01 (see Hypothesis 2, and Figure 2), but not on the feeling thermometer, F(1, 439) = 0.34, p = .562. Labeling ethnicity with nouns led to higher preferred social distance from the Roma (M = 3.60, SD = 0.81) than labeling ethnicity with adjectives (M = 3.44, SD = 0.81).

Discussion

In Study 1, report valence and linguistic forms for ethnicity shaped prejudice against the Roma after a single exposure to a print news report about the Roma. Our data supported Hypothesis 1 in that reports of distinct valence led to different levels of prejudice. Accordingly, participants who read the positive report expressed less prejudice than those who read the negative report. However, participants who read the positive report did not significantly differ from those who were exposed to positive information combined with negative information in the mixed condition. Furthermore, at odds with evidence form Joyce and Harwood (2014), participants who read the mixed report significantly differed from those who read the negative report.

With respect to the effect of linguistic forms, nouns led to higher preferred social distance from (but not less warmth felt toward) the Roma than adjectives irrespective of report valence, supporting Hypothesis 2. Specifically, participants who read the print news reports where Roma’s ethnicity was labeled with nouns preferred greater social distance from the Roma than participants who read the reports with adjectives.

Study 2

Study 1 showed that a single exposure to a report about members of a negatively perceived minority can shape prejudice against this minority. Study 2 aimed to test this finding using a different language and in another intergroup context with another minority group, namely Kosovo Albanians in Switzerland.

Method

Participants and design

The online questionnaire in German was filled out by 486 Swiss participants, native German speakers. Because we wanted to focus on Swiss majority’s attitudes, we excluded 94 participants with an immigration background and 30 participants who indicated that they did not at all believe that they read a real news report, resulting in the final sample of 362 participants (65% women, M_age = 23.38, SD = 4.44 years). The design of Study 2 was the same as in Study 1, employing six experimental conditions (3× report valence and 2× linguistic forms) with two measures of prejudice.

Procedure and materials

To select a minority to be used in Study 2, we first conducted a media content analysis of Swiss newspapers. The content analysis indicated that Kosovo Albanians are negatively stereotyped in Switzerland due to their association with crime and, as a result, we chose this minority as the target group. In the fictitious print news reports, a male member of Kosovo Albanian minority gets into a fight between two Swiss men and engages in either helping (positive) or aggressive behaviors (negative). In the mixed condition, one Kosovo Albanian engages in helping and another one in aggressive behaviors in the same situation. In each of the three valence conditions, the nationality of minority and majority members was labeled with either adjectives (a Kosovo Albanian man and a Swiss man) or nouns (a male Kosovo Albanian and a male Swiss; see Appendix for the experimental materials).

Subsequently, participants expressed their prejudice against the Kosovo Albanian minority on the feeling thermometer (ranging from 0 to 100) and on the social distance scale (α = .89) as in Study 1. Next, participants were asked about the report credibility. Different to Study 1, we employed a continuous measure, asking participants to what extent they believed that the report was a real newspaper report, on a 5-point scale ranging from 1 (not at all) to 5 (very much). Finally, participants filled out demographic characteristics and were debriefed.³

Results

To estimate the effect of report valence and linguistic forms on affective and behavioral components of prejudice against Kosovo Albanians, we performed a MANOVA because both measures of prejudice were correlated, r = .51, p < .001. Using Pillai’s trace, the MANOVA revealed a significant effect of report valence, V = 0.04, F(4, 712) = 3.49, p = .008, and of linguistic forms, V = 0.03, F(2, 355) = 5.00, p = .007, on prejudice. The interaction between report valence and linguistic forms was not significant, V = 0.00, F(4, 712) = 0.17, p = .952, again indicating that the effect of linguistic forms was independent of the report valence.⁴

Subsequent ANOVAs showed that report valence significantly predicted prejudice on the feeling thermometer, F(2, 356) = 6.66, p < .001, Eta² = .04 (see Hypothesis 1, and Figure 3), but not on the social distance scale, F(2, 356) = 1.74, p = .178 (see Figure 4). As predicted, the planned contrasts revealed that participants who read the positive report (M = 62.25, SD = 19.71) expressed less prejudice (i.e., warmer feelings) than participants who read the negative report (M = 52.87, SD = 21.73), 95% CI [4.32, 14.54], p < .001. At the same time, participants who read the positive report expressed less prejudice (i.e., warmer feelings) than participants who read the mixed report (M = 57.15, SD = 19.33), 95% CI [−10.14, −5.30], p = .048. There was no difference between the effects of the mixed report and the negative report, 95% CI [−1.02, 9.68], p = .112.

Figure 3. A significant effect of report valence but not linguistic forms on prejudice against Kosovo Albanians measured with the feeling thermometer (range 0–100; higher values indicate less prejudice) in Study 2. Error bars are standard errors of the means.

Figure 4. A significant effect of linguistic forms but not report valence on prejudice against Kosovo Albanians measured with the social distance scale (range 1–5; higher values indicate more prejudice) in Study 2. Error bars are standard errors of the means.

Linguistic forms predicted prejudice on the social distance scale, F(1, 356) = 9.03, p = .003, Eta² = .03 (see Hypothesis 2, and Figure 4), but not on the feeling thermometer, F(1, 356) = 0.44, p = .510. Labeling ethnicity with nouns led to higher preferred social distance from Kosovo Albanians (M = 2.92, SD = 0.65) than labeling their ethnicity with adjectives (M = 2.70, SD = 0.76).

Discussion

Study 2 aimed to validate the findings of Study 1 in a different intergroup context with a different language and minority. We found that prejudice against the negatively stereotyped minority of Kosovo Albanians in Switzerland was shaped by a single exposure to a print news report about Kosovo Albanians, supporting Hypothesis 1. Participants who read the positive report expressed significantly warmer feelings toward Kosovo Albanians than participants who read the negative report, similar to the case of the Roma in Study 1. However, in contrast to Study 1, Study 2 provided supporting evidence for the results of Joyce and Harwood (2014), finding a significant difference between the effect of the positive and the mixed report and the lack of difference between the effect of the mixed and the negative report. Swiss participants who were exposed to the positive report about Kosovo Albanians displayed warmer feelings toward Kosovo Albanians compared to participants exposed to the mixed report. At the same time, the warmth expressed toward Kosovo Albanians did not differ between participants who were exposed to negative and positive information in the mixed report and those exposed to solely negative information in the negative report. Labeling the nationality of Kosovo Albanians with nouns led to higher preferred social distance from (but not less warmth felt towards) Kosovo Albanians as compared to adjectives across the different valence conditions, supporting Hypothesis 2.

Study 3

Studies 1 and 2 dealt with media reports about stigmatized minorities and found the effects of valence and language on prejudice in two markedly different intergroup contexts. However, these effects may only occur in the case of negatively stereotyped groups. To examine whether valence of print news reports and language for minority ethnicity can influence prejudice against minorities more generally, we employed a positively stereotyped group of Italian immigrants in Switzerland in Study 3.

Studies 1 and 2 provided conflicting evidence for the effect of the simultaneous presentation of positive and negative information about social minorities because the mixed report once differed from the effect of a negative (but not a positive) report and once from the effect of a positive (but not a negative) report on prejudice (see Table 1 for hypotheses and summary of results). Study 3 thus employed a more nuanced operationalization and measurement of the effect of mixed news reports. Studies 1 and 2 introduced participants to only one version of the mixed report where the negative information preceded the positive information (first an outgroup member attacked, then another outgroup member helped). In Study 3, we controlled for the potentially confounding primacy and recency effects of the presentation of positive and negative information in the mixed article and prepared two versions of the mixed report (one introducing the negative information first vs. the other introducing the positive information first).

Last, in the two preceding studies we only measured attitudes after participants were exposed to the reports, neglecting their initial attitudes toward the given outgroup. In Study 3, we therefore asked about participants’ attitudes before and after they were exposed to the experimental manipulation, to be sure that the difference in prejudice between groups was due to the exposure to the reports of different valence.

Method

Participants and design

The online questionnaire in German was filled out by 252 Swiss participants, native German speakers. Because we focused on Swiss majority’s attitudes, we excluded 11 participants with an immigration background and 21 participants who indicated that they did not at all believe that the article was a real newspaper report (i.e., 1 on a five-point scale), resulting in a final sample of 220 participants (72% women, M_age = 26.36, SD = 8.45 years).

The design of Study 3 was the same as in both preceding studies and comprised six experimental conditions (3× report valence and 2× linguistic forms) and two measures of prejudice. Unlike the two previous studies, we measured the change in prejudice toward Italian immigrants that represented the difference between pre- and post-measurement of attitudes on two feeling thermometers.

Procedure and materials

To distract participants from the fact that we were interested in their attitudes toward Italian immigrants specifically, they were asked to indicate their attitudes toward various socially relevant issues (e.g., waste separation and nuclear energy) and social groups (e.g., Turkish immigrants, Italian immigrants, German immigrants, Spanish immigrants, Italian-speaking Swiss, French immigrants, German-speaking Swiss, French-speaking Swiss, employed mothers, and homeless persons) on feeling thermometers. All labels for nationality or ethnicity of the employed groups were either used in the noun or adjective form to keep the linguistic forms constant in the pre- and postmeasurement of attitudes. To further prevent participants from remembering the exact value of their attitudes toward Italian immigrants on the feeling thermometer before the manipulation, we employed feeling thermometers where exact values were not marked (only the end points and the middle of the scales). We measured the initial level of prejudice only with feeling thermometers because we were primarily interested in the effect of news valence that has been previously shown to impact especially the affective part of prejudice. Furthermore, the social distance scale consists of five items that would be difficult to complete with so many other groups and impossible to use with social issues.

Next, participants were exposed to a fictitious print news report, in which an immigrated Italian (vs. Italian immigrant), the owner of a restaurant, either invites homeless people who are sitting in front of his restaurant for a free meal (positive) or shoos them away (negative). Because we attempted to recreate credible, real-life news reports, we employed stereotypical characteristics of the given minority. While the stereotypical image of the Roma and Kosovo Albanians contains elements of realistic threat, a danger to personal safety, the stereotypical image of Italians is less negative, containing characteristics relating to emotional instability (e.g., quick tempered reactions). Consequently, we toned down the negativity in the report about Italian immigrants in Study 3, as compared to the reports about the Roma and Kosovo Albanians employed in Study 1 and 2.

All conditions mentioned two Italian immigrants (vs. immigrated Italians) who owned a restaurant together to keep the number of presented outgroup members constant (see Appendix for the exact wording). We created two versions of the mixed report with positive information presented first in one and negative information first in the other. Specifically, in one version of the mixed report, one immigrated Italian invites the homeless people for a free meal and another immigrated Italian, his business partner, shoos them away. In the other version of the report, the homeless people are shooed first by one immigrated Italian and then invited in by the other.

Following the manipulation, participants expressed their attitudes toward Italians living in Switzerland on a feeling thermometer (ranging from 0 to 100) and the social distance scale (α = .90). To account for the change in attitudes after reading the report (M = 1.65, SD = 19.46), we subtracted the value on the feeling thermometer after the manipulation (M = 73.50, SD = 21.78) from the value on the feeling thermometer before the manipulation (M = 74.96, SD = 19.46). Consequently, a negative score represents a decrease in prejudice; a positive score represents an increase in prejudice. At the end of the questionnaire, as in Study 2, we asked participants about the credibility of the report, demographic characteristics and debriefed them.⁵

Results

First, we checked whether the order of presenting positive and negative information within the mixed report influenced participants’ attitudes toward Italian immigrants. The order of presentation did not influence either the change in attitudes on the feeling thermometer, t(90) = .70, p = .492, or the social distance scale, t(90) = 1.25, p = .214. Thus, the presentation order of the positive and negative events did not have an effect on participants’ attitudes.

Using Pillai’s trace, a MANOVA revealed a significant effect of report valence, V = 0.07, F(4, 428) = 3.78, p = .005, but no significant effect of linguistic forms on prejudice against Italians, V = 0.02, F(2, 213) = 2.22, p = .111. The interaction between report valence and linguistic forms was not significant, V = 0.02, F(4, 428) = 1.29, p = .272, indicating that the effect of linguistic forms was independent of the report valence.⁶

Subsequent ANOVAs revealed that report valence significantly predicted the change in attitudes toward Italian immigrants on the feeling thermometer, F(2, 214) = 7.83, p < .001, Eta² = .07 (see Figure 5). Report valence did not predict attitudes measured with the social distance scale, F(2, 214) = 0.12, p = .884 (see Figure 6). The planned contrasts showed the expected difference between the effect of the positive report (M = −3.84, SD = 14.70) and the effect of the negative report (M = 8.67, SD = 23.23) on the change in attitudes toward Italian immigrants, 95% CI [−18.84, −6.25], p < .001. However, the effect of the positive report did not differ from the effect of the mixed report (M = 0.82, SD = 18.24), 95% CI [−1.37, 10.72], p = .129. The effect of the mixed report significantly differed from the effect of the negative report, 95% CI [−14.08, −1.67], p = .013. Thus, the change in attitudes of participants who were exposed to both positive and negative information in the mixed condition was not different from those exposed to solely positive information but different from those exposed to solely negative information.

Figure 5. A significant effect of report valence but not linguistic forms on the change in prejudice against Italian immigrants in Study 3. To determine the change in prejudice, the value on the feeling thermometer after the manipulation was subtracted from the value on the feeling thermometer before the manipulation (range −100 to 100; positive values indicate an increase in prejudice, negative values indicate a decrease in prejudice). Error bars are standard errors of the means.

Figure 6. A significant effect of linguistic forms but not report valence on prejudice against Italian immigrants measured with the social distance scale (range 1 to 5; higher values indicate more prejudice) in Study 3. Error bars are standard errors of the means.

Looking closely at the attitudes toward Italian immigrants before and after the experimental manipulation, we found that the negative report caused the greatest change in attitudes (see Figure 5). The one sample t-test indicated a significant difference from zero in case of the negative report, t(66) = 3.06, p < .001, and in case of the positive report, t(73) = 2.25, p = .031. This means that participants’ attitudes toward Italians significantly worsened after the negative report and significantly improved after the positive report. In case of the mixed report, the change in attitudes after the manipulation was not significantly different from zero, t(78) = 0.40, p = .693. The effect of the negative report on the change in attitudes after the experimental manipulation was significantly larger than the effects of both the positive report and the mixed report.

We found a trend of the effect of linguistic forms on the social distance scale corresponding to Studies 1 and 2, F(1, 214) = 3.58, p = .060, Eta² = .02, in that nouns led to higher preferred social distance (M = 2.68, SD = 0.81) than adjectives (M = 2.49, SD = 0.83; see Figure 6). However, despite the effect size similar to the effects of linguistic forms in Study 1 and 2, the difference between nouns and adjectives on the social distance scale was not statistically significant in Study 3. Linguistic forms did not predict the change in attitudes toward Italian immigrants on the feeling thermometer, F(1, 214) = 1.29, p = .257.

Discussion

Study 3 dealt with Italian immigrants in Switzerland to test whether the effects of valence and linguistic forms found in highly negatively perceived minorities would hold in a positively stereotyped group. Furthermore, Study 3 aimed to validate the results of Study 1 and Study 2 in a more methodologically elaborate design.

Similar to prejudice against the Roma in the Czech Republic and Kosovo Albanians in Switzerland, report valence predicted prejudice against Italian immigrants on the feeling thermometer, supporting Hypothesis 1. Nevertheless, due to a more rigorous design of Study 3, we were able to show that the effect was caused by the experimental manipulation, as indicated by the change in prejudice after participants read the print news reports. The change in prejudice of participants who were exposed to the positive report was significantly different from the change in attitudes of participants who were exposed to the negative report. Similar to results of Study 1, participants simultaneously exposed to positive and negative information in the mixed condition did not differ from participants exposed to the positive report, but significantly differed from participants exposed to the negative report. The difference in the effect of the negative and mixed report was not due to the recency effect of the lastly presented positive information in the mixed report of Study 1. In Study 3, the order of presentation of positive and negative information did not make a difference to participants’ prejudice, eliminating the order of presentation as a possible confound from the effect of the mixed report.

Having read the positive or the negative report about Italian immigrants significantly changed participants’ subsequent attitudes toward Italians as compared to their initial attitudes. The effect of the negative report on attitudes following the experimental manipulation surpassed the effect of both positive and mixed reports. This more determining effect of negativity was found in a positively perceived group and despite the fact that the negative information presented in the report was very mild (i.e., shooing someone away). Furthermore, the target of the described negative behavior were members of a highly stigmatized group (i.e., homeless people) that usually do not evoke much sympathy. The effect of negative information on change in attitudes may be stronger than positive information in negatively perceived groups—particularly in cases of serious instances of negative behavior (e.g., brachial aggression), and when negative behavior targets ingroup members (vs. marginalized outgroups). Indeed, a meta-analysis of experimental studies on individual-to-group generalization showed that negative information has a generalization advantage for stigmatized groups, in comparison to positive information for admired groups (Paolini & McIntyre, 2018). Further evidence is offered by intergroup contact literature, where information about group members that matches group stereotype (e.g., negative for stigmatized groups) is more likely to be generalized than information that does not match group stereotype. This effect is due to high category salience evoked by the matching information (Brown & Hewstone, 2005). Group members who do not fulfil expectations based on group stereotypes can be subtyped and thereby excluded from the formation of prejudice about the given group (Fiske & Neuberg, 1990; Richards & Hewstone, 2001).

On the other hand, social cognitive literature holds that the effects of negative information may be especially strong in positively stereotyped groups. Perceivers direct their attention and give importance to clues that are most informative (Fiske, 1980). In this light, stereotype-inconsistent information may provide novel, and thus more important, information about a group than stereotype-consistent information that fulfils the expectations based on group stereotype (Fiske, 1980). In the context of Study 3, members of positively stereotyped minorities are expected to behave positively, thus, the negative information was the one considered most informative, attracting more attention and processing, resulting in greatest change in attitudes after having read the negative report. Consequently, in negatively stereotyped groups, positive information may be the most informative, potentially resulting in greater change in attitudes after having read the positive report, as compared to reading the negative report. More evidence is needed to resolve which information is the most influential in positively and negatively evaluated groups—a test that is beyond the scope of this article.

With respect to the effect of linguistic forms, we found a trend corresponding to the outcomes of Studies 1 and 2 where nouns used for labeling nationality led to higher social distance, as compared to adjectives, irrespective of report valence. The fact that the difference between nouns and adjectives was not statistically significant in Study 3 could be due to a lower number of participants because the effect sizes of linguistic forms were comparable across the three studies.

General discussion

This series of three studies represents a first attempt to determine how a single exposure to print news reports about minority members, combined with distinct labels for their group membership, influences prejudice against social minorities. The studies tested the effects of the two aspects of news reports in settings that differ with respect to the share of social minorities and language. The results showed that prejudice against social minorities can be influenced by a single exposure to a news report. Both researched aspects of the news—their valence and language used for ethnicity—had an effect on prejudice. However, the two aspects of newspaper reports influenced different components of prejudice. Valence mainly had an effect on the affective component of prejudice, whereas the linguistic forms had an effect on the behavioral component of prejudice (see Table 1 for summary of results across the three studies). Thus, while positivity and negativity of news about different social groups can shape how we feel about them, the labels for group membership influence how close we wish to be to these groups. As different aspects of news influence different components of prejudice, it is advisable to distinguish between them in future studies on media influence and use them when tailoring interventions aimed at combating distinct parts of prejudice.

Table 1. Summary of the Hypotheses and Effects of Valence and Linguistic Forms on two Measures of Prejudice Tested in Three Studies

	Valence			Linguistic forms
	Hypothesis 1	Feeling  thermometer	Social distance	Hypothesis 2	Feeling thermometer	Social distance
Study 1	positive < negative positive < mixed mixed = negative	positive < negative positive = mixed mixed < negative	positive < negative positive = mixed mixed < negative	nouns > adjectives		nouns > adjectives
Study 2	positive < negative positive < mixed mixed = negative	positive < negative positive < mixed mixed = negative		nouns > adjectives		nouns > adjectives
Study 3	positive < negative positive < mixed mixed = negative	positive < negative positive = mixed mixed < negative		nouns > adjectives		nouns > adjectives

Note. The > sign indicates a higher level of prejudice, the < sign indicates a lower level of prejudice, the = sign indicates that the given conditions did not differ with respect to prejudice. The corresponding trend in the effect of nouns and adjectives in Study 3 is given in grey.

With respect to the report valence, a single exposure to solely positive information had the most positive effect on participants’ attitudes. Despite the fact that positive information is rather scarce in the news about social minorities, all three studies showed the advantages of disseminating positive news—especially about negatively perceived minorities—to combat prejudice against them. Furthermore, including positive information even into negative reports about social minorities may combat prejudice. In two out of three studies, positive information in the mixed reports balanced out the detrimental impact of negative information that—when presented alone in the solely negative reports—led to most pronounced prejudice. These findings are at odds with the outcomes of the Joyce and Harwood’s study (2014), which found that the effect of mixed content about social minorities did not differ from the effect of solely negative content on prejudice. Our findings are more promising with respect to improving the stance of social minorities in society through mass media, however, the conflicting outcomes need to be put to the test in future studies. More research is needed to explain whether and when the simultaneous presentation of positive and negative information surpasses the effect of solely negative or even positive information on prejudice.

Study 3 included measurement of attitudes before the experimental manipulation, supporting the assumption that the difference in attitudes after reading a positive, negative, or mixed report is due to report valence. A single exposure to a positive report led to improvement in attitudes, a highly optimistic finding that nevertheless needs to be examined by future studies in different intergroup settings, particularly because Italians in Study 3 represent a positively stereotyped group. Alarmingly, the negative report had the strongest effect on the change in attitudes. The greater impact of negativity over positivity is well documented for a wide range of psychological outcomes (for a review see Baumeister, Bratslavsky, Finkenauer, & Vohs, 2001). The higher effect of negative (vs. positive and mixed) information on prejudice against a positively stereotyped group may be even stronger for stigmatized minorities (Paolini & McIntyre, 2018). If negative information has a stronger effect than positive information on prejudice against stigmatized minorities, the only possible means of combatting this effect would be with a greater amount of positive information. Unfortunately, in reality the news about social minorities is predominantly negative (e.g., Visintin et al., 2017), leaving especially stigmatized minorities with small hope of having their image improved through mass media.

Though this is one theory, there is also contrasting evidence to suggest that negative information is especially prominent in shaping attitudes toward positively perceived groups because negative information diverges from the expectations of positivity (Fiske, 1980). Consequently, negative reports about stigmatized minorities would be unlikely to worsen prejudice, as the effect of the unexpected positive information would be stronger than the effect of stereotype-matching negative information. More research is needed to clarify the effects of positive and negative information in mass media reports about positively and negatively stereotyped minorities, before exact predictions about the impact of news of distinct valence can be made.

Our research is novel in that it examined not only the effect of valence of print news reports on prejudice, but also the effect of distinct linguistic labels for group membership. In two different languages (i.e., Slavic and Germanic) and two cultural contexts (Czech and Swiss), using nouns for minority members’ ethnicity or nationality consistently resulted in greater preferred social distance from the social minorities than using adjectives. This is in line with previous findings that nouns lead to more pronounced intergroup bias. Across our studies, nouns always led to higher preferred social distance than adjectives—not only after reading a negative, but also after reading a mixed or a positive report. Based on evidence from previous studies (Graf et al., 2013), linguistic forms on their own cannot eradicate prejudice. However, their repeated and ever-present nature in mass media can contribute to either solidifying or combating prejudice, which is especially relevant in the case of stigmatized minorities.

Limitations and future directions

In all three studies, the effects of report valence and linguistic forms on prejudice were small, but consistent for all three minority groups. Additionally, corresponding effect sizes have been found in previous research (e.g., Graf et al., 2013; Saleem et al., 2017). The fact that a one-off exposure to a single report about a concrete behavior of outgroup members can alter prejudice against the whole outgroup is intriguing, particularly considering the potentially additive effects of news that people consume on daily basis (Mastro & Tukachinsky, 2011).

We tested our assumptions in two different intergroup and language contexts, including three different minority groups that are perceived differently by the majority. Due to the novelty of our integrative approach, our findings should be tested in other intergroup contexts, languages, and minority groups. Despite the fact that we included different intergroup and language contexts, we did not treat them as moderators in the analyses. Future studies may provide a more valid cross-cultural comparison of the established effects that can differ in size depending on the cultural background.

Our research dealt with the effect of news reports about social minorities on majority’s attitudes. Future studies should look closely into the effects of media news and language used for description of minority membership on attitudes held by minorities (cf. Schmader et al., 2015). Effects within minority groups may differ from those found in the majority (Pettigrew & Tropp, 2006), therefore conducting research with social minorities would be an interesting avenue for future research.

Conclusions

The findings of our research are timely and socially relevant due to the increasing number of social minorities with different ethnicities in European countries and the power of mass media to reach a great number of people. On the one hand, the valence of media reports can be easily identified—people are able to discern whether presented information is positive or negative—and as a result they may—at least to some extent—be aware of the influence that mass media has on their attitudes. On the other hand, differences in labels for ethnicity represent subtle language cues that may escape people’s awareness. The difficulty to control the effects of language in polarizing people’s attitudes poses a challenge for responsible media use. There were attempts to abandon ethnicity labels in certain media contexts (e.g., crime reports), however, to our knowledge, no attempts have been made to draw attention to the effects of distinct linguistic forms used for category membership (i.e., nouns vs. adjectives). Information about the role of different aspects of media in enhancing or reducing prejudice is of relevance to professionals communicating about social minorities (e.g., journalists, politicians, social workers, nongovernmental organizations employees) as well as their audience (e.g., general public, social policy makers) and, as such, should be included into school curricula or training for critical media consumption.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

The research has received funding from the European Commission Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie Individual Fellowship grant agreement No 703495 awarded to Sylvie Graf. Study 1 was supported by a grant 17-14387S from the Czech Science Foundation, and by RVO: 68081740 from the Institute of Psychology, Czech Academy of Sciences. Study 2 was supported through travel grants by the Swiss National Science Foundation (IZKOZ1_166739) and the European Association of Social Psychology.

Notes

1. Study 1 also included a measure of ethnic essentialism (6 items adapted from Roets & Van Hiel, 2011) and direct and mass mediated intergroup contact with the Roma (4 items adapted from Pagotto & Voci, 2013) that were not employed in the design of this study.

2. When controlling for participants’ sex, the pattern of results remained unchanged, with sex having no effect on participants’ attitudes, V = 0.01, F(2, 437) = 1.05, p = .352.

3. Study 2 included a measure of ethnic essentialism (6 items adapted from Roets & Van Hiel, 2011), direct and mass mediated intergroup contact with Kosovo Albanians (4 items adapted from Pagotto & Voci, 2013), and perceived typicality of the described behavior for the minority as a whole (1 item adapted from Brown, Vivian, & Hewstone, 1999) that were not employed in the design of this study.

4. When controlling for participants’ sex, the pattern of results remained unchanged, with sex having no effect on participants’ attitudes, V = 0.01, F(2, 354) = 1.39, p = .250.

5. Study 3 included a measure of ethnic essentialism (6 items adapted from Roets & Van Hiel, 2011), direct and mass mediated intergroup contact with Italian immigrants (4 items adapted from Pagotto & Voci, 2013) and perceived typicality of the described behavior for the minority as a whole (1 item adapted from Brown et al., 1999) that were not employed in the design of this study.

6. When controlling for participants’ sex, the pattern of results remained unchanged, with sex having no effect on participants’ attitudes, V = 0.01, F(2, 212) = 0.66, p = .520.

7. The English translation of the reports from Czech and German mirrors the original reports as closely as possible, although it may compromise the way the reports sound in English.

Appendix

Study 1

Positive report

Turbulent night in Brno: Roma man/male Roma saved life of a brutally attacked man⁷

BRNO – A 26-year-old man suffered serious injuries when returning home around midnight yesterday. Not far away from his apartment, he was assaulted by a man armed with a knife. The aggressor beat the man’s face, kicked his stomach and stabbed his arm while he demanded money.

Fortunately, a passing Roma/Roma passerby spotted the assault and without hesitation set out to help the attacked man. The assailant also threatened the Roma man/male Roma with the knife, but lost his weapon after a short fight and fled. However, he did not manage to run far away because of a police patrol called by the saving Roma/Roma savior. The savior also provided first aid that, according to the police spokesperson Denisa Sulcova, saved the injured man’s life.

The courage of the Roma man/male Roma was highlighted by the superintendent of Brno Metropolitan Police Department Jiri Franc: “The Roma hero/heroic Roma deserves acknowledgement because he was able to stand up for a person in need against an armed man. The incident would have had a tragic ending without the help of the Roma man/male Roma.”

The aggressor now faces charges under arrest. “The police assessed the act as mugging and the defendant can be sentenced up to ten years in prison” added the police spokesperson Sulcova.

Negative report

Turbulent night in Brno: Roma man/male Roma brutally attacked a man

BRNO – A 26-year-old man suffered serious injuries when returning home around midnight yesterday. Not far away from his apartment, he was assaulted by a Roma man/male Roma armed with a knife. The Roma aggressor/aggressive Roma beat the man’s face, kicked his stomach and stabbed his arm while he demanded money.

Fortunately, a passerby spotted the assault and without hesitation set out to help the attacked man. The Roma assailer/assaulting Roma threatened with the knife also the man coming to help, nevertheless he lost his weapon after a short fight and fled. However, the Roma man/male Roma did not manage to run far away because of a police patrol called by the savior. The savior also provided first aid that, according to the police spokesperson Denisa Sulcova, saved the injured man’s life.

The courage of the man was highlighted by the superintendent of Brno Metropolitan Police Department Jiri Franc: “He deserves acknowledgement because he was able to stand up for a person in need against an armed Roma assailer/assaulting Roma. The incident would have had a tragic ending without his help.”

The Roma aggressor/aggressive Roma now faces charges under arrest. “The police assessed the act as mugging and the Roma man/male Roma can be sentenced up to ten years in prison” added the police spokesperson Sulcova.

Mixed report

Turbulent night in Brno: A Roma assailer/an assaulting Roma ended up behind bars, a saving Roma/Roma savior is a hero

BRNO – A 26-year-old man suffered serious injuries when returning home around midnight yesterday. Not far away from his apartment, he was assaulted by a Roma man/male Roma armed with a knife. The aggressor beat the man’s face, kicked his stomach and stabbed his arm while he demanded money.

Fortunately, a passing Roma/Roma passerby spotted the assault and without hesitation set out to help the attacked man. The Roma assailer/assaulting Roma threatened with the knife also the coming savior, nevertheless he lost his weapon after a short fight and fled. However, he did not manage to run far away because of a police patrol called by the saving Roma/Roma savior. The savior also provided first aid that, according to the police spokesperson Denisa Sulcova, saved the injured man’s life.

The courage of the man was highlighted by the superintendent of Brno Metropolitan Police Department Jiri Franc: “The Roma hero/heroic Roma deserves acknowledgement because he was able to stand up for a person in need against an armed assailer. The incident would have had a tragic ending without his help.”

The Roma aggressor/aggressive Roma now faces charges under arrest. “The police assessed the act as mugging and the man can be sentenced up to ten years in prison” added the police spokesperson Sulcova.

Study 2

Positive report

Brawl in a Basel bar, an uninvolved Kosovan/a Kosovan uninvolved averts severe injuries

On Sunday, 9. January, shortly before 11 pm, a conflict escalated between two Swiss man, guests in the bar “Rossi” in Basel. Heavily drunk Andreas Knecht accused Kari Grunder of flirting with his girlfriend and started beating him up. An uninvolved guest, a Kosovan/uninvolved Kosovan guest Dalmat Prifti hurried to help Grunder. The fearless Kosovan/Kosovan dreadnought managed to get Grunder out of the bar and call the ambulance. The emergency physician noted that without the intervention of the Kosovan helper/helping Kosovan, the injuries could have threatened Grunder’s life.

Negative report

Brawl in a Basel bar, an involved Kosovan/a Kosovan involved causes severe injuries

On Sunday, 9. January, shortly before 11 pm, a conflict escalated between two Swiss man, guests in the bar “Rossi” in Basel. Heavily drunk Andreas Knecht accused Kari Grunder of flirting with his girlfriend and started beating him up. An uninvolved guest, a Kosovan/uninvolved Kosovan guest, Azem Demiri barged in and joined Knecht’s side. The interfering Kosovan/Kosovan interferer was kicking Grunder who laid on the ground. The emergency physician noted that the injuries could have threatened Grunder’s life because of the assaulting Kosovan/Kosovan assailer.

Mixed report

Brawl in a Basel bar with two interfering Kosovans: One joins in the beating, the other averts severe injuries

On Sunday, 9. January, shortly before 11 pm, a conflict escalated between two Swiss man, guests in the bar “Rossi” in Basel. Heavily drunk Andreas Knecht accused Kari Grunder of flirting with his girlfriend and started beating him up. An uninvolved guest, a Kosovan/uninvolved Kosovan guest, Azem Demiri barged in and joined Knecht’s side. Demiri was kicking Grunder who laid on the ground. Another Kosovan/Kosovan guest Dalmat Prifti, unknown to both sides of the conflict, hurried to help Grunder. The emergency physician noted that without the intervention of the Kosovan helper/helping Kosovan the injuries could have threatened Grunder’s life.

Study 3

Positive report

An immigrated Italian/Italian immigrant invites homeless to his pizzeria

The innkeeping Italian/Italian innkeeper, Alberto Renzi, caused a sensation in the city centre of Bern last Saturday when he invited a group of homeless people for a free pizza to his pizzeria. The homeless lingered in front of the pizzeria owned by the charming Italian/Italian charmer. While his business partner, Paolo Bianchi, counted the daily revenues, the beneficial Italian/Italian benefactor went to the kitchen and fulfilled everybody’s pizza wish. Having finished their pizzas, the group heartily thanked Alberto Renzi and said good bye. When confronted with journalists’ question why did he do such a thing, the immigrated Italian/Italian immigrant replied in a casual way: Why not?

Negative report

An immigrated Italian/Italian immigrant chases homeless away from his pizzeria

The innkeeping Italian/Italian innkeeper, Alberto Renzi, caused a sensation in the city centre of Bern last Saturday when he loudly shooed off a group of homeless people from his pizzeria. The homeless lingered in front of the pizzeria owned by the choleric Italian/Italian choleric. While his business partner, Paolo Bianchi, counted the daily revenues, the raving Italian/Italian raver went to the front of his restaurant armed with a broom. Loudly gesticulating and swinging the broom, prompted Alberto Renzi the group to disappear immediately. Scared by the fierce reaction, the group absconded. When confronted with journalists’ question why did he do such a thing, the immigrated Italian/Italian immigrant replied in a casual way: Why not?

Mixed report [positive – negative]

An immigrated Italian/Italian immigrant invites homeless to his pizzeria – his countryman chases them away

The charming Italian/Italian charmer, Alberto Renzi, caused a sensation in the city centre of Bern last Saturday when he invited a group of homeless people for a free pizza to his pizzeria. The homeless lingered in front of the pizzeria. The beneficial Italian/Italian benefactor fulfilled everybody’s pizza wish. His business partner, the choleric Italian/Italian choleric, Paolo Bianchi, seemed not to be happy with the idea. He stood in front of the eating homeless armed with a broom and prompted them to disappear immediately. Scared by the fierce reaction, the group absconded. When confronted with journalists’ question why did he do such a thing, the raving Italian/Italian raver replied in a casual way: “Why not?”

Mixed report [negative – positive]

An immigrated Italian/Italian immigrant chases homeless away from his pizzeria – his countryman invites them for free pizza

The choleric Italian/Italian choleric, Alberto Renzi, caused a sensation in the city centre of Bern last Saturday when he loudly shooed off a group of homeless people from his pizzeria. The homeless lingered in front of the pizzeria. The raving Italian/Italian raver went in front of the homeless armed with a broom and prompted the group to disappear immediately. His business partner, the charming Italian/Italian charmer, Paolo Bianchi, seemed not to be happy with the idea. He stood in front of the homeless to protect them and then invited them for free pizza. When confronted with journalists’ question why did he do such a thing, the beneficial Italian/Italian benefactor replied in a casual way: “Why not?”