Introduction

We thank the two anonymous reviewers of French Politics for their constructive comments. Sean Mueller also acknowledges financial support from the Swiss National Science Foundation (grant no. 186897).

In a seminal article in comparative politics, Almond (1956, 396) pointed out that one of the fundamental traits of any political system is “the pattern of orientations to political action” it is embedded in, which he referred to as political culture. He further argued that “the USA, England, and several of the Commonwealth countries have a common political culture, but are separate and different kinds of political systems” (p. 397). The existence of an anglophone political culture, i.e. one defined by language, has been confirmed by the World Values Survey (WVS), the most extensive effort to measure political culture across countries, as well as by several other studies (Inglehart and Baker 2000; Schwartz et al. 2009, 136; Minkov and Hofstede 2012, 150–153). Do other communities speaking the same language but spanning different countries also have such a “common political culture”? While the study of political culture has grown greatly since Almond’s observation and the subsequent pioneering volume by Almond and Verba (1963), there has been surprisingly limited interest in exploring the impact of language on political orientations and resulting behaviour.

Next to language, another distinctive component of culture is religion (Laitin and Wildavsky 1988, 591). For Brubaker (2013, 1), “[l]anguage and religion are arguably the two most socially and politically consequential domains of cultural difference in the modern world.” Unlike language, religion has been widely studied and found to be an important determinant of orientations, values, and attitudes (e.g. Siroky et al. 2017). One of the two main dimensions of cultural variation identified by the WVS is that between traditional and secular-rational values, with religion looming large over traditional values. Not surprisingly, the WVS “cultural maps” of the world are largely based on religious background (WVS 2022; Akaliyski and Welzel 2020, 756).

The scarce attention devoted to language in political culture research contrasts with the growing interest in other disciplines, as we briefly review below. Prior research has found that territorially proximate countries tend to display similar cultural patterns. This is attributed to a process of diffusion brought about by frequent contacts and exchanges (e.g. Naroll 1973; Schwartz et al. 2009, 137; Ember et al. 2015, 589–590). While neither explicitly theorized nor, a fortiori, empirically tested in the literature, it seems plausible to attribute the mentioned cultural similarity among anglophones at least partly to continued diffusion facilitated by their shared English language despite nation-state separation and territorial distance.

To shed light on the role of language in shaping political values and attitudes beyond the English-speaking world, we investigate the possible existence and contours of a francophone political culture. This refers to the notion of French speakers exhibiting a similar “pattern of orientations to political action”, in Almond (1956, 396) words. If such a common political culture shaped by the French language were to exist, it would both (a) bring French speakers together across nation-states and (b) set them apart from their co-nationals speaking another language, at the same time.

We explore such hypothesized francophone political culture through an online survey conducted in France, Belgium, Canada, and Switzerland in autumn 2020, at the height of the COVID-19 pandemic. While all four countries are Western democracies, the latter three contain sizeable French-speaking minorities concentrated in their own sub-state entities. Collecting samples of roughly the same size for all these groups allows us to test for cross-country commonality and within-country dissimilarity at the same time. Is having French as one’s mother tongue systematically associated with similar orientations to political action in a way that possessing another mother tongue is not?

Section “Language and culture: brief literature review” briefly reviews literature on the connection between language and culture in political and other social sciences. Section “French and political culture: theory and hypotheses” theorizes the potential of the French language to shape a francophone political culture and derives two main hypotheses. Section “Data and methods” describes our data and methods. Results reveal an only weakly delineated francophone political culture as just defined—limited to similar views on the role of the state for promoting gender quality and cannabis legalization (“Results” section). Section “Discussion and conclusion” concludes by acknowledging limitations and delineating paths forward.

Language and culture: brief literature review

The relationship between language and culture has been studied in many social sciences. There is a long-standing hypothesis in linguistics, cultural anthropology, and cognitive psychology that connects language structures with patterns of thought. Known as the Sapir-Whorf, or linguistic relativity, hypothesis, it states that “the particular language we speak influences the way we think about reality” (Lucy 1997, 291). Given the diversity among the world’s languages (Evans and Levinson 2009), we should thus expect a significant influence on society. Studies have found that this is often the case, although not without qualifications (e.g. Lucy 1992a, b, 1997; Boroditsky 2001; Pederson et al. 2010). Scholars have shown language to be associated with political culture and preferences of linguistic groups across countries via cultural diffusion, be that through the media or other products (e.g. Inglehart and Baker 2000; Schwartz et al. 2009, 136; Minkov and Hofstede 2012, 150–153; Pye and Verba 2015).

The role of language has also attracted considerable interest in economics. While many have focused on what can be broadly summarized as the impact of linguistic diversity (e.g. Desmet et al. 2012), some have directly addressed the connection between language and behaviour, finding robust empirical evidence for the former’s influence on the latter (Chen 2013; Chen et al. 2017; Kim et al. 2017; Na and Yan 2021); though see Gotti et al. (2021) for a cautionary note). In political science, Laitin (e.g. 1977, 1992, 1998) has been a prominent scholar of the politics of language and subscriber to the linguistic relativity hypothesis, arguing that “the language a person speaks influences the way [s/]he perceives and acts in the world” (Laitin 1977, 222). Although his work has not specifically focused on the impact of language on individual behaviour, it nonetheless offers many examples of their connection. More recently, Pérez and Tavits (2022) have shown that language has a significant effect on public opinion across a range of topics, including gender, ethnic relations, and environmental policies.

The literature on multilingual countries is equally replete with statements that values, attitudes, and political behaviour vary along the language border within states (e.g. Erk 2008; Billiet et al. 2006; Hooghe and Stiers 2022). Several scholars have furthermore uncovered linkages across nations. Hofstede (2001, 63), for instance, located French-speaking Switzerland closer to France, and German-speaking Switzerland closer to Germany, particularly on his “power distance” dimension. Schwartz et al. (2009, 137) also found francophone Switzerland to be nearer to France in terms of hierarchy orientations. Likewise, he positioned “French Canada […] closer to […] France than to English-speaking Canada” (ibid.). In a similar vein, Wiseman (2007) highlighted the distinctiveness of Quebec culture within Canadian political culture (also Blanchard 2023). In a study of values across European regions, Minkov and Hofstede (2014, 156–157) reported similar findings: French-speaking Switzerland clusters with France while German-speaking Switzerland lies closer to Germany. Different patterns have, however, been reported for Belgium. Hofstede (2001, 63) found little difference between French- and Dutch-speaking Belgium but considerable distance between the latter and the Netherlands, with both main Belgian linguistic groups situated closer to France. Billiet et al. (2006) in turn identified noticeable differences in political culture between Flemings and Walloons, albeit less sharp than the major differences in voting behaviour between the two communities would suggest (Hooghe and Stiers 2022; Niessen et al. 2022). Minkov and Hofstede (2014, 156–157) reported French-speaking Belgium as clustering with France but Flanders forming a distinct group, yet otherwise confirmed (Hofstede 2001) earlier findings.

These general patterns have been confirmed by research with a narrower focus. For instance, Bridgman et al. (2022, 110 & 121) reported lower levels of generalized trust in Quebec than across the other, majority Anglophone, provinces of Canada (but see Blanchard 2023, 16–17). They also found a significant correlation for language, with bilingual French and English speakers displaying higher levels of trust than francophones only (Bridgman et al. 2022, 116 & 121–122). In Switzerland, too, many authors have found significant differences in political attitudes and behaviour along the French/German linguistic border (e.g. Kriesi et al. 1996; Büchi 2000; Linder et al. 2008; Seitz 2014). French speakers, for instance, are significantly more pro-EU (e.g. Dardanelli et al. 2021: 180), “technocratic” as opposed to ecological (Hermann and Leuthold 2003), in favour of centralization (e.g. Mueller 2015; Mueller and Dardanelli 2014), and more sceptical of evidence-based policy making (Bundi and Pattyn 2022) than German speakers. Chen et al. (2017: 336) found that companies headquartered in German-speaking Swiss cantons “hold significantly more cash” than those based in French-speaking cantons; Brown et al. (2018) that students in German-speaking Swiss cantons have higher financial literacy than their French-speaking counterparts; and Aepli et al. (2021) that voters in German-speaking municipalities value the private provision of certain goods more and that firms offer more vocational training in the German-speaking part. Other studies have found strikingly similar variation along the language divide in Belgium and Switzerland: Chen (2013) reported that francophones save less for retirement than Dutch- or German speakers in both Belgium and Switzerland, while Na and Yan (2021) observed that firms in the French-speaking parts of both Belgium and Switzerland engage more in tax avoidance that their counterparts in the Dutch and German-speaking areas, respectively.

While not all these aspects are equally closely connected to political culture, there is sufficient evidence to suggest that a common francophone political culture spanning different countries may indeed exist. Yet, we still possess only a limited understanding of its precise contours in terms of specific political attitudes as well as of its relative weight in comparison with other factors such as education, age, gender, and political ideology. Our comparative, multivariate analysis tries to answer these questions.

French and political culture: theory and hypotheses

How could the French language specifically shape individual values and attitudes? While the literature on political culture navigates between the importance of shared historical legacies and contemporary exposure, two main testable potential mechanisms emerge out of the works just reviewed: cognition and diffusion. Following Pérez and Tavits (2022, 19–33), the cognitive channel operates through the grammatical structure of French, predisposing French speakers to see the world differently from speakers of other languages—particularly from those grammatically more distant from French. Such grammatical differences have been shown to affect a range of economic behaviours (e.g. Chen 2013) as well as opinions on political issues (Pérez and Tavits 2022).

The diffusion channel, in turn, posits that the French language, via the circulation of cultural products such as books, magazines, TV, radio, and online content, could act as a conduit for the circulation of underlying values and orientations from one French-speaking population to another and across generations (Brubaker 2013). Linguistic proximity also correlates with frequent contact, cross-migration, and common ancestry (Allassonnière-Tang et al. 2021; Holman et al. 2015). French speakers in particular have their own global organization, La Francophonie, whose goal it is precisely “to promote the French language and political, educational, economic and cultural cooperation” among its members using, among others, their own TV channel, TV5MONDE.Footnote 2

The francophone space we are focusing on is characterized by the presence of one dominant community: France’s 66 million inhabitants versus about 7 million francophones in Canada, 4 million in Belgium, and 2 million in Switzerland. We can thus expect diffusion to operate through France’s political culture radiating outwards by influencing francophones in Canada, Belgium, and Switzerland (cf. also Waterbury 2021). For instance, foreign channels have a 69% market share in French-speaking Switzerland. In Belgium, too, the French channels TF1 and France 2 are among the most widely viewed in Wallonia, with their own La Une having just a 20% market share compared to its Flemish equivalent Één with 30% in Flanders (Media DB 2022).Footnote 3

If either, or both, of these channels works as theorized, we should observe the French-speaking populations in our three multilingual countries to display attitudes similar to those characterising the French. But what exactly are these attitudes? To answer this question in a way that helps us map the contours of a possible francophone political culture, we can compare France to Germany, the Netherlands, and the UK as ideal–typical examples of a German, Dutch, and English-speaking political culture, respectively. The relevant literature summarized below suggests France to be distinctive on four dimensions: trust, hierarchy, uncertainty avoidance, and economic and gender equality. In all these dimensions, France is an outlier compared to Germany, the Netherlands, and the UK.

First, different studies have found lower levels of generalized trust in France compared to the other three countries (Beugelsdijk and Schaik 2005, 310; WVS 2005–9, V23). Generalized trust refers to the degree to which citizens trust each other in everyday life. A second dimension is hierarchy or “power distance”. Hofstede (2001, 83) defines power distance as the difference in status and influence between hierarchically superior and inferior persons in an organization. More specifically, hierarchy is understood as a desire to respect and maintain superior rank and reputation. Hofstede (2001, 87) found significantly higher power distance in France in the late 1960s/early 1970s compared to Germany, the Netherlands, and the UK. Schwartz et al. (2009, 137) equally observed a relatively high level of hierarchy, by West European standards, in France. These findings dovetail with studies of business organizations, which record higher levels of hierarchy and status consciousness in France compared to Denmark, Germany, the Netherlands, and the USA (Brossard and Maurice 1974; Schramm-Nielsen 1989; D’Iribarne 1997; Meyer 2017). Orly (2023, 26) even speaks of “la centralité et la verticalité de la tradition autoritaire” (the centrality and the verticality of the authoritarian tradition), in France.

A third dimension is ‘uncertainty avoidance’, which Hofstede (2001, 145–148) defines as a tendency to look for (formalized) structure in social relations to render the future more predictable. This translates into a preference for regulation and standardization, for instance of economic life, and stability, for instance in terms of employment protection. From this perspective, government intervention and legislation are welcomed to reduce social uncertainty (Hofstede 2001, 150, 174 & 176–177). According to Hofstede (2001, 151–152 & 174), uncertainty avoidance correlates with “power distance” and is again highest in France. These findings are echoed by Borre et al. (1995, 248), who report the French being more in favour of government management of the economy in 1981 than the Germans, Dutch, and British. The contrast with US “antistatism” (e.g. Quadango and Street 2005; also Nettl 1968) is especially pronounced.

The fourth and final dimension relates to attitudes to equality, particularly regarding economic and gender issues, i.e. between different income groups in society and between men and women. This then translates into a preference for redistribution and advancing women’s rights. Based on data from the 1970s–1990s Knutsen et al. (1998, 176), Roller et al. (1995b, 179), and Thomassen et al. (1995, 406) all found higher support for socio-economic equality in France than in the other three countries. Roller et al. (1995a, 68–9) further recorded a higher importance attached to reducing wealth inequalities, and Thomassen et al. (1995, 402) greater preferences for equality over freedom. In a similar vein, Lundmark et al. (1998, 267) found a higher level of “feminist political culture” in France in 1983 while the WVS (2005–2009, V61–V63) registered higher support for political, educational, and economic equality between genders, as well as somewhat stronger support for abortion (WVS 2005–2009, V204).Footnote 4 Attitudes to socio-economic and gender equality also indicate a generally more left-wing orientation, in line with reported self-placements for the 1970s/1980s (Klingemann et al. 1995, 192; cf. also Hooghe and Stiers 2022, 663–4; but see Stiers and Hooghe 2023, 688–9).

We can now formulate more specific hypotheses concerning the extent and content of a common francophone political culture. If language trumps or at least equals other covariates of political values and attitudes, we will encounter a francophone political culture as follows:

H1a

Francophones in Canada, Belgium, and Switzerland exhibit a political culture similar to those of the French and different from those of their co-nationals speaking another language.

Reflecting the distinctive aspects of France’s political culture identified above, we expect such a common political culture to be characterized thus:

H1b

Francophones in Canada, Belgium, and Switzerland as well as the French display comparatively lower levels of trust, higher hierarchical and uncertainty-avoidance orientations, and stronger support for economic and gender equality.

The expectation of pro-government intervention and pro-equality orientations is also buttressed by evidence that states and sub-states with an overwhelming francophone majorityFootnote 5 all display above-average levels of public spending relative to their peers. For instance, in 2019 Quebec spent the equivalent of 34% of its GDP, whereas the Canada-wide average was 24% (Statistics Canada 2020, 2021). Quebec is also the only province with a comprehensive, publicly funded day-care system and drug insurance programme. In Belgium, in 2019 the combined spending of the French-speaking community and the Walloon region amounted to €9′800 per capita, compared to 8′100 in the Flemish region/community (StatBel 2021; NBB 2020).

It is reasonable, however, to hypothesize that language is not all-powerful and its effect tempered or even countered by other aspects. Most importantly, the socio-political situation of the French, as a majority language community and titular nation of their state, differs doubly from that of francophones in Canada, Belgium, and Switzerland—all of which are minority communities in multilingual federations. Pérez and Tavits (2022, 95–110) have shown that speaking a minority versus a majority language has a significant impact on attitudes. This common context could bring the latter three closer to each other while at the same time distancing them from the French:

H2a

Francophones in Canada, Belgium, and Switzerland exhibit a political culture similar to each other but different from those of their co-nationals speaking another language and also different from those of the French.

As linguistic minorities, the francophones in our three multilingual countries are also likely to be more aware of their cultural distinctiveness, particularly of perceived “positive” traits such as support for equality and women’s rights, than either the French or speakers of the majority language. They could thus also be more determined to preserve and promote it. Furthermore, as numerical minorities they could also value regional autonomy more than the French, who do not have to fear cultural majority encroachment.Footnote 6 Hence:

H2B

H2b: Francophones in Canada, Belgium, and Switzerland display comparatively higher support for economic and gender equality as well as regional autonomy than members of linguistic majorities, including the French.

As striving for economic and gender equality are left-wing priorities, this hypothesis dovetails with patterns concerning partisan preferences. Indeed, French speakers in Canada, Belgium, and Switzerland have consistently voted for parties of the left to a greater extent than their compatriots speaking another language, at least since the 1970s. Thus, in the 14 Swiss federal elections between 1971 and 2023, the left (Social-Democrats, Greens, and radical left) scored an average of 35% of the vote in French versus just 28% in German-speaking Switzerland; differences were highest in 2019 and 2023, with 13 and 12 percentage points, respectively (OFS 2023b). Across the 15 Belgian federal elections between 1971 and 2019, left-wing parties scored an average of 45% of the vote in Wallonia versus just 26% in Flanders (Dodeigne and Binard 2018; IBZ 2019; Hooghe and Stiers 2022, 658–659). In Canada, too, the Conservative Party of Canada (CPC), the only major clear and consistent right of centre party in federal politics, has not gained a majority of seats in Quebec since 1972, except for 1984 and 1988.

Data and methods

The empirical backbone of this paper is an online survey conducted in October–November 2020 in Belgium, Canada, France, and Switzerland. Samples were purchased from, and the survey was run via, Qualtrics. We base our comparisons on representative samples among the two main linguistic communities in Belgium (750 French and Dutch native speakers, respectively), Canada (610 French and English native speakers, respectively), and Switzerland (750 French and German native speakers, respectively). In France, we collected the responses of 1200 native French speakers. The country of residence was assessed via geofencing; the survey was translated from English into the different languages by country specialists and trialled before launch. Table 1 shows the resulting sample sizes of each community and country if we drop all those not speaking the mother tongue of their region of residence—hence the lower final sample sizes than initially collected. Table A1 in appendix provides some summary statistics, boosting our trust in the comparability of these groups.

Table 1 Sample and group properties
Full size table

The survey itself consisted of four parts: general and specific views on politics; federalism and decentralization; an omnibus section (with questions on COVID-19, democratic innovations, evidence-based policy making, and personality traits); and socio-demographics. Of interest for us here are all the dimensions relating to the theorized components of a francophone political cultureFootnote 7:

  1. 1.

    Trust was assessed through six questions on how much respondent trusted their federal/central government; their regional/provincial/cantonal government; their local executive; journalists; scientific experts; and their fellow citizens. As the resulting Cronbach’s alpha is a high 0.74 and to deal with missing values, we calculated the mean value across all six questions on a scale from none (1) to a great deal (4).

  2. 2.

    Orientations towards hierarchy and uncertainty avoidance were assessed via two questions on the importance of international cooperation (supranationalism) and the danger and corresponding ban on the sale of cannabis. The scale runs again from fully disagree (1) to fully agree (4).

  3. 3.

    Views on economic equality were assessed through four questions on state interventionism.Footnote 8 As the resulting Cronbach’s alpha is a high 0.7 and to deal with missing values, we calculated the mean value across all four questions on a scale from 1 to 4.

  4. 4.

    Views on gender equality were assessed through two questions.Footnote 9 As the resulting Cronbach’s alpha is a good 0.5 and to deal with missing values, we calculated the mean value across both questions on a scale from 1 to 4.

  5. 5.

    Finally, views on regional autonomy were gauged through two questions on power over education and culture being located at regional level. As the resulting Cronbach’s alpha is a good 0.53 and to deal with missing values, we calculated the mean value across both questions on a scale from the international level (1) to each region/province/canton/community alone (5).

Methodologically, given that political culture is primarily an attribute of groups, we first compare replies across communities. Then, in the interest of trying to understand drivers of cultural differences and similarities, we compare the impact of mother tongue with that of country of residence and other socio-demographic variables using OLS regression analyses at the individual level. It is important to bear in mind that generally cultural values vary only to a limited extent across populations, particularly among high-income Western democracies such as those analysed here (e.g. Fischer and Schwartz 2021; Minkov 2013: 409–416).

Results

Group level analyses

To obtain an overview of differences and similarities within and across countries, Fig. 1 shows the mean values for each cultural orientation by group and item.Footnote 10 A first clear tendency is for francophones in all four countries to be more strongly supportive of state intervention to further gender equality and to support a more cautious drug policy (via opposition to cannabis legalization) than non-francophones. However, only the Québécois—i.e. francophone Canadians residing in Quebec—exhibit a markedly greater desire for regional autonomy than anglophone Canadians residing outside Quebec, and only Romands exhibit slightly greater trust than German Swiss speakers. Regarding supranationalism and economic equality, no distinct francophone pattern can be discerned as all groups show similar levels of agreement.

Fig. 1
figure 1

Cultural orientations by country-language group. Note: Shown are mean values per group; all scales run from 0 to 4 except regional autonomy, which runs from 0 to 5

Full size image

ANOVA tests for all four francophone groups (i.e. the French and the three francophone minorities) and six items confirm the absence of significant differences regarding both supranationalism and gender equality. In other words, among French speakers the mean preferences on these two dimensions are virtually identical—regardless of country of residence and demographic majority/minority position. Results for the other dimensions are mixed: on trust and regional autonomy, only Belgian francophones and the French have identical attitudes, while all other comparisons reveal differences. On drug policy, Walloons and the French on the one hand and Canadian and Swiss francophones on the other hand have the same preferences. Regarding socio-economic equality, it is Québécois and Walloons in one camp versus Swiss Romands and the French in the other.

We next repeat this exercise, but this time for group comparisons within the three multilingual federations. This relativises one key finding reported above in that supranationalism is a shared value also across language groups within states, as already Fig. 1 shows. But the same is not true of the other trans-francophone trait, attitudes to gender equality, where, in line with H1, francophones in all three countries studied here have significantly different views from their non-francophone co-nationals. The only other area to reveal significant inter-group differences within states is drug policy—and, for Canada, regional autonomy.

Such aggregate-level analyses can be informative for they tell us what the actual distribution of preferences in a society is.Footnote 11 In other words, group comparisons take into account that different societies are composed differently: for instance, political views clearly also depend on education or economic situation. So, if a given community has a significantly larger share of well-educated, well-off members then this might in turn explain a more progressive outlook. On this basis, the only area for a common political culture among francophones across our four countries in line with H1 concerns gender equality. Note, furthermore, that while attitudes to cannabis among francophones vary as much as between francophones and non-francophones, they do so at a high level of agreement—some are simply even more against legalization than others.

The next section moves the analyses to the individual level, making use of the full range of reply options alongside several control variables. However, while at the individual level the impact of language on attitudes could disappear, it could remain at the level of groups—especially if, as observed before, French-speaking polities place a greater emphasis on university education than their German-, English-, or Dutch-speaking counterparts. For instance, there are almost as many universities in the French-speaking part of Switzerland as in the German-speaking part, although the latter is twice the size of the former in terms of population (OFS 2022). In Canada, Quebec has the lowest university tuition fees in the country. This is why the next section complements but does not replace group-level analyses.

Individual-level analyses

We first look at within-country differences and then turn to possible similarities among francophones across states. For each of the three multilingual federations, six multivariate OLS models are thus specified, one per dependent variable. The main independent variable is a dummy which takes the value of 1 if a respondent’s mother tongue is French, and 0 otherwise (only Dutch, English, and (Swiss-)German speakers are included). We added several control variables (age, education, personal economic situation, gender, urban/rural residence, and religion; see Table A1 for categories and descriptives) to minimize confounding correlations and improve both the fit and accuracy of our models.

Figure 2 shows the resulting (standardized) correlation coefficients for language within the three multilingual federations. Drug policy stands out as not only showing a significant impact throughout, but also as pointing in the same direction: French speakers in Belgium, Canada, and Switzerland are all more likely to agree that “cannabis is dangerous and should not be legalized” than their respective non-French-speaking compatriots. This result is in line with the theorized greater priority for uniformity, equality, and state intervention as opposed to diversity and personal freedom. A separate report from Switzerland (Sotomo. 2021) confirms not only these differences between French and German speakers (p. 25), but also these different rationales: the former prioritize bans and state-led public health campaigns, the latter favour pragmatism and individual self-determination (p. 51). French speakers indeed desire to reduce uncertainty.

Fig. 2
figure 2

The impact of speaking French in Belgium, Canada, and Switzerland. Note: *p < 0.1; **p < 0.05; ***p < 0.01. Shown are standardized correlation coefficients for speaking French vs. the respective majority language controlling for other factors, with 95% confidence intervals. For full models, see Tables A2–A4

Full size image

In Belgium and Switzerland, French speakers are also more feminist in outlook than their Dutch- or German-speaking co-nationals. In Switzerland, they are more trusting and in Canada much more in favour of regional autonomy. Nowhere is there a significant impact of language on support for either economic equality or supranational integration.

To assess similarities and differences among francophones, we next restrict the sample to those with French as their mother tongue and run the same models as before but instead of mother tongue include country dummies as the main independent variable. Figure 3 shows the resulting coefficients, always compared to France as the reference category. On no dimension are all francophones alike even if, on balance, Walloons come out as the most similar to the French (only two significant differences). The Québécois have different values than the French in five out of six dimensions. Swiss francophones are located between the other two. Swiss and Canadian francophones are also the most like each other: both are more trusting and regionalist in outlook but less in favour of cannabis legalization than the French.

Fig. 3
figure 3

The impact of country of residence among francophones. Note: *p < 0.1; **p < 0.05; ***p < 0.01. Shown are standardized correlation coefficients for speaking French but living in a country other than France and controlling for other factors, with 95% confidence intervals. For full models, see Table A5

Full size image

Of course, these differences are not huge in terms of substance, but noteworthy nevertheless. There is thus at least partial evidence for a francophone minority community of values. The Walloons’ exclusion from this could be because Belgian francophones, although always a numerical minority, were hegemonical for a long time and even today enjoy guaranteed political parity with Dutch speakers at federal level in certain regards (e.g. for cabinet positions). Francophone Belgians are also a much higher proportion of the country’s population than their Canadian and Swiss counterparts and their area includes the country’s capital, which could be further reasons for within-state similarity.

Discussion and conclusion

The question about the existence of a common political culture in nation-states has sparked many heated debates around how to define and measure it. The issue is even more sensitive in multinational and/or multilingual states where differences can be politicized. While elements of a common political culture have been found among other (e.g. anglophone) linguistic groups, our aim here has been to look for a possible francophone political culture that would be shared by French speakers living in states as diverse as France, Belgium, Canada, and Switzerland. As we know of no other study of this kind, ours must be understood as a first attempt at exploring at once similarities across and differences within francophone countries.

Our cases serve the purpose of such a comparison very well because despite partial linguistic commonality, the political structures of Belgium, Canada, and Switzerland are radically different from that of France (federal and decentralized as opposed to unitary and centralized). There are also differences in size, location, and history of state- and nation-building. For instance, while in France French speakers are the titular nation and no other language is permitted when dealing with the state, in Belgium, Canada, and Switzerland francophones are a minority, albeit the largest one. Finally, language is heavily politicized in Belgium and Canada but much less so in Switzerland, or only occasionally and not more than other differences (Linder et al. 2008; Stojanović 2021; Mueller et al. 2022). The existence of a common, transnational, even transcontinental francophone political culture would thus be significantly strengthened if similar patterns were to be found against all those odds.

Yet our analyses yield only partial evidence for such a uniform political culture across the dimensions and countries studied here—especially if we also expect there to be differences vis-à-vis non-French-speaking co-nationals. Two questions regarding how much the state should regulate society in terms of gender equality and cannabis legalization were identified as elements of a political culture that was both shared by francophones and not (or at least not as much) by non-francophones. Hence, both our main hypotheses are only partially confirmed.

Further takeaways relate to the countries and groups under study. Belgians are not as divided culturally as electoral results and especially political dynamics would have us believe, confirming findings by Billiet et al. (2006) and Hooghe and Stiers (2022). The opposite is true for Canadians, where the historical “brokerage parties” plaster over numerous differences in political attitudes (cf. also Stiers and Hooghe 2023). Swiss language divisions are as pronounced as electoral and especially the occasional referendum results would suggest and lie between the Belgian and Canadian cases. Among all francophones, the most similar pairs were found to be Walloons and the French, on the one hand, and Swiss-French/Romands and Québécois, on the other. In fact, Quebec’s political culture is the most distant from France’s, but the fact that the Romands also differ in some cultural orientations from the French points to more than just territorial and phonetic distance as well as embeddedness in an English-speaking sea as drivers.

What is left for further research is to shed light on how and why language shapes only certain political attitudes but not others, why it does so in some (national) contexts more than in others, and how relations between (differently sized) majority and minority groups evolve over time. One obvious factor to investigate is the degree to which language is politicized and construed rhetorically into a marker of underlying values and preferences. In short, people could view things differently because they are told they would. However, given that inter-group differences are lowest precisely in the context with one of the sharpest linguistic divides (Belgium), party competition in that sense does not seem to play a role here. Of course, this then raises the question as to what else explains the sometimes stark differences in public policy we mentioned above, from general spending to academic infrastructure and related differences in left-wing party success.

More generally, our research contributes to several literatures. On the one hand, a growing body of research has studied the impact of language on public policies (e.g. Pérez and Tavits 2017, 2022; Cortina and Rottinghaus 2022; Liu 2022). We build on this research in two ways. First, in the wake of social identity studies underlining the prominent place of language (e.g. Hansen and Liu 1997; Huddy 2001; Huddy and Bankert 2017) and empirical research identifying common traits of an anglophone political culture (Wiseman 2007; also Blaydes and Grimmer 2020), we show that elements of a common political culture also exist among francophones across countries and even continents.

Second, although the impact of political culture(s) has been argued to be crucial for understanding political outcomes across countries (e.g. Keyser 2021), little previous research has focused explicitly on the effect of language at the individual level. Previous studies show that interpersonal conversations can be effective tools for political campaigns to change voter attitudes and behaviours (e.g. Gerber and Green 2000; Kalla and Broockman 2020, 2022). However, this large literature has focused on the effects of conversations in the same language. We contribute to this literature by providing unique data on how different languages within a country relate to different political attitudes—or not, which is even more astonishing given the postulated or real centrality of language for identity, interests, and values.

By focusing on language as one of many covariates of political values and attitudes, our study shows that it is not enough to just include language to control for cultural differences. Instead, we argue based on our findings that group identities within and across states deserve a closer look. Future studies should, if possible, analyse how exactly language shapes attitudes within multilingual societies and across borders. To further refine causal identification strategies for the impact of language is an important task for future research.