MULTILINGUAL FAMILIES AND THEIR POSITION IN EUROPEAN SOCIETIES

Background: European Union (EU) statistics show that the percentage of people living in immigrant or mixed households is growing (Eurostat, 2017). Authors concur that fostering multilingualism brings benefits for the family members’ wellbeing (Portes & Hao, 2002) and cohesion (Müller et al., 2020) and for their connection with the society where they live (Vuorenkoski et al., 2000). Research goal: The aim of this study is to analyse the diversity of multilingual families and to describe their social environment in Europe. We study the proportion of people living in multilingual families, their demographic background, and their integration into society. Furthermore, we classify certain European countries according to the characteristics of their multilingual households to better understand the environments in which multilingual families live. Method: We use European Social Survey (ESS) data on immigration background and integration attitudes to analyse multilingual family members from 18 European countries. Demographic background variables include: (1) immigration background and languages spoken at home, (2) multi-ancestry background, and (3) presence of children in the household. The integration variables are: (4) citizenship of the country of residence; (5) feelings of happiness; (6) feelings of discrimination; (7) self-evaluated economic coping. Finally, we conduct a statistical hierarchical cluster analyses to identify countries that share multilinguals with similar characteristics. The detailed description of variables used in this analysis is provided in the article. Results: Our results show that multilingualism has roots in migration in many countries. However, despite the link between immigration history and multilingualism, the background of multilingualism is not linear, and several deviations and outliers are present. The cluster analysis identified five country clusters: (1) Switzerland the most multilingual country; (2) Belgium, Spain, Ireland, and Germany form an immigration-related, highly multilingual group; (3) France and Sweden form a discriminated-multilingualism cluster; (4) The UK with very low rates of multilingualism and multi-ancestry identity and (5) the final cluster formed by the remaining ten countries with low rates of multilingualism overall. Conclusion: This study contributes to the knowledge about multilingual families in Europe and the heterogenous environments in which they live. Results show that multilingualism is closely related to recent immigration and the migration history of a country, but this relationship is not linear. The results of the study provide a basis to study their different dimensions in further detail, particularly, those related to identity formation and the internal and external motivations to develop an affiliation towards host countries.


Introduction
Growing international migration flows and inter-ethnic marriages lead to an increasing diversification of the ethnic structure of families. Interaction between people with different ethnic backgrounds has become more frequent around the world (Sowa-Behtane, 2016). Statistical information shows that binational families are a growing segment of the population in Europe (Sowa-Behtane, 2017). In 2014, the European Union had 55 million immigrants aged 15 -64, making up 17.7 % of the population. These people lived in around 16.7 million "immigrant households" and in 9.4 million "mixed households". These households formed 14.2 % and 8.0 % of all households in the European Union, respectively (Eurostat, 2017). Studying mixed families is complex due to the wide range of characteristics that influence their identity, including ethnicity, race, religion, culture, and language (Collet, 2006;Collet, 2012;Sizaire, 2017;Varro, 2012). The aim of this study is to analyse the diversity of multilingual families and to describe their social environment in Europe. We study the proportion of people living in multilingual families, their demographic background, and their integration into society. Furthermore, we classify certain European countries according to the characteristics of their multilingual households. This classification allows to better understand the environments in which multilingual families live in different countries. To our knowledge, this is the first comparative paper about the life of multilingual families in Europe. We define multilingual families as households in which at least two different languages are spoken daily. Multilingual families develop for a variety of reasons, including immigration, emigration, mixed marriages, mixed cultural heritage, and language learning. In this paper, our special focus is on analysing the country-level links between migration and multilingualism. From the previous analyses (Medvedeva, 2012) it is known that many mixed-heritage families might have a high level of proficiency in the host country language, especially if their host country language is one of the languages spoken at home. The study of multilingual families is important from the point of view of educational programmes, identity formation (Benet-Martínez & Haritatos, 2005;Collet & Varro, 1996) and integration policies.

The importance of multilingualism
The European Union Commission (2021) argues that multilingualism has benefits such as strengthening intercultural understanding, enhancing employability and mobility, and improving the competitiveness of the EU economy. However, multilingual and mixed-heritage families face different challenges compared to monolingual families, and they develop different acculturation strategies (Sam & Berry, 2010). For example, members of mixed families more often develop bicultural and even multicultural identities that they display with different levels of intensity depending on the circumstances (Benet-Martínez & Haritatos, 2005). They can experience different types of cultural power struggles, for example when making decisions about the development of the cultural traits of their children and social values (Finnäs & O'Leary, 2003;Varro, 2012). Multilingualism is highly significant because it influences the cohesion and wellbeing of individuals and their connection with society. Balanced multilingualism is defined as proficiency in at least two languages at a good level. The precise level of those language skills is disputed and is treated differently by different authors (Macias, 2021). Balanced multilingualism is positively associated with minority language maintenance and child wellbeing (Müller et al., 2020), but it also may cause language conflicts (Macias, 2021).
Multilingualism has psychological consequences for individuals, it plays an important role in identity formation, and it has social consequences. It can have an impact on social status, employment opportunities, choices about marriage, and decisions about where to live. In addition, fostering multiculturalism in society results in better affiliation of mixed heritage family members with their host country and positive interactions with majority groups (Igarashi, 2019). Multilingualism is an especially interesting phenomenon in family situations: it involves mixed heritage couples, their children, and the host society. The development of parents' bilingualism is interesting because it involves the challenge of individuals acquiring proficiency in a new language in adulthood. For a minority-culture parent, acquisition of the majority-culture language facilitates integration into the host society. In parallel, mixed heritage identity is better appreciated by all members of the family when the majority culture parent can also communicate in the minority language (De Houwer, 2011). Mixed-heritage family identities and multilingualism can be enabled or oppressed depending on the host society's disposition towards multiculturalism. Research indicates that there are connections between higher levels of creative thinking, intercommunication skills and tolerance, as well as correlations between the language proficiency of children of mixed families and their wellbeing (Hurajova, 2020). Proficiency in both majority and minority languages -balanced multilingualism -results in stronger family cohesion and appreciation of the family's heritage (Müller et al., 2020). At the same time, it leads to better relationships with the majority culture members in the host society. Some authors have described this as 'balanced identities ' (Vuorenkoski et al., 2000). As opposed to the above, lower proficiency in any essential languages, or unbalanced multilingualism, can become an obstacle for developing stronger family relationships and integration into the host society (Portes & Hao, 2002). Furthermore, studies have also refuted the idea that multilingualism might cause linguistic confusion, slower learning and identity conflicts among children and adolescents (Genesee, 2015). Even studies conducted among third generation adolescents with low levels of proficiency in their minority language show that they think that learning and maintaining their minority language is important both for communication and identity formation (Mills, 2001).
The importance of the policy environment Multilingualism can be appreciated by the host society as cosmopolitan capital, i.e., having multicultural awareness, knowledge, and skills to operate internationally and, eventually, gain upward social mobility (Weenink, 2008). There is evidence of a correlation between the national identification of immigrants and minority groups and the implementation of policies that foster multiculturalism (Igarashi, 2019). Generally, more favourable acculturation attitudes in the host society result from higher levels of integration (Berry, 2007). Multicultural identities tend to blossom under the favourable reception of cultural differences by the host society. However, multiculturalism can be conditioned by biased perceptions as well as by political and historical factors. In this respect, the perceived status of the language in the host society plays a significant role and is influenced by social and political connotations that can change at different historical times. The outcome of this process is known as language ideologies (Karpava et al., 2018), and they condition the language ecology -the interaction of languages with their environment (Garner, 2005).

Method
We used data from the European Social Survey (ESS) to analyse the characteristics of multilingual family members from 18 EU countries. ESS data is suitable for the analysis of multilingual households because it includes data about migration background, home language, cultural and socio-economic background, attitudes, beliefs, and personal behaviour. Compared to population censuses, the ESS provides more up-to-date and richer data about the life of multilingual families.
The ESS data are collected every second year through face-to-face interviews. Individuals are selected by using random probability methods with sampling frames of individuals, households or addresses. The ESS results represent the residential population aged 15 and over in each country. We use design weighs for data from 18 countries and from five ESS rounds (2008 -2018) for the analysis ( Table 1). The key selection criteria for the countries was that they had a complete dataset from all these rounds in order to avoid period selection bias. The data from the different rounds were pooled into one database. The total number of respondents per country available for analysis is presented in Table 1. Table 1 shows, that the pooled database includes enough people from multilingual families and enables us draw conclusions about every country separately as well.  Table 1 also presents information about the immigration history of the countries for analyses. We use two variables to describe the immigration situation in each country: the percentage of the population who are foreign-born and recent immigration intensity (average number of immigrants per 1000 population in 2010 -2019). In the sample, we see countries with a high share of total immigrant population -the United Kingdom, Ireland, Sweden, Estonia, Switzerland, France, and Belgium -and countries with recent, continuous, intensive immigration like Ireland, Switzerland, and Sweden. Some other countries, like the Czech Republic, Lithuania, and Poland, have a comparatively limited history of immigration.
The ESS data are collected in the most common languages spoken in each country. All countries provide questionnaires in languages that represent at least 5 % of speakers from the whole population. The share of multilingual families might be underestimated in the surveys, especially in recent immigrant groups who have not yet learned the language of their country of residence and fail to participate in the survey due to the language barrier. The ESS collects data about reasons for non-response, and based on these data, we can argue that the degree of bias due to limited language skills is not large (Table 1). Norway, Switzerland, Belgium, and Sweden had the highest share of non-response due to language, but even in those countries, the percentage of people dropping out from the survey due to limited language skills was less than 4 % of the initial sample.

Variables
Multilingual households are identified through a question about home languages. The ESS asks respondents, "What language or languages do you speak most often at home?" If respondents marked two different languages, we recorded them as respondents from multilingual families and refer to them as multilinguals or multilingualism in this paper. We also analyse the use of home languages among multilingual households. For this purpose, the first and second language spoken in the household were classified in four groups: (1) Official language(s) of a country (including official regional languages), (2) English as a foreign language, (3) other European Union languages, and (4) other non-European Union languages. We used two types of variables to analyse the background of multilingualism: demographic variables and variables reflecting the integration of multilingual families into the society. Multilingualism is often related to migration heritage, and the integration variables are similar to those used in migration studies. The first group of variables describes the demographic background of persons living in multilingual households: (1) immigration background (whether the respondent was born in the country or whether the parents of the respondents were born in the country) and languages spoken at home, (2) whether respondents living in multilingual households also have multiancestry background, and (3) whether the respondent has children in the household. Multilingualism is often related to migration heritage, but it can be also a result of multiculturalism without migration. To investigate each country's conditions and provide better classification of countries, we use multiple indicators about migration. First, we assume that multilingualism is directly related to migration and want to know how many immigrants (individuals not born in the country) are living in multilingual households. Second, multilingualism may be a relic from the immigration of previous generations, mixed marriages or persistent multiculturalism in the country. We capture this segment with the indicator "Born in the country and multilingual (%)". The third indicator reflects the share of second-generation immigrants in the multilingual group. These are people who live in multilingual households and were born in their current country of residence, but at least one of their parents was born in some other country (%). This indicator reflects long-term or persistent multiculturalism traditions in these households. The ESS also allows to investigate people's feelings about their origins. Namely, the ESS questionnaire asks people to disclose their ancestral background. "Ancestry" is understood as "descent" or "family origins". The question asks "How would you describe your ancestry? Please use this card to choose up to three ancestries that best apply to you." Interviewers can code a maximum of three ancestries in total. If more than three are mentioned, interviewers ask the respondent to select three. We recorded respondents with at least two different ancestry backgrounds as a separate group and analysed the link between multilingualism and multiorigin self-identification. To describe the prevalence of this group within a country, we use the variable "Multi-ancestry persons among multilingual persons (%)". We also classify ancestry background information in three groups: (1) local ancestry from the current country or region of residence, (2) other European Union ancestry and (3) non-European Union ancestry and analyse the link between identity and multilingualism.
Finally, we compare the share of multilingual households with children with the rest of the population to describe the potential future of multilingualism. Multilingualism can be passed as a tradition to the next generation via learning processes, and this indicator describes the demographic potential of multilingual households.
The second set of variables analyses the integration of multilingual persons. For this purpose, we used variables reflecting the similarity of multilingual households to the rest of the population. The following variables belong to this category: (4) citizenship of the country of residence -the difference between respondents from multilingual families and the rest of the population; (5) happiness difference between multilingual and monolingual groups; (6) feelings of discrimination -the difference between multilingual and monolingual groups; (7) self-evaluated economic coping -the difference from the rest of the population. Citizenship of the country of residence can be seen as an important indicator of political integration. Citizenship reflects the closeness and commitment of persons to a country, but at the same time, it can be influenced by citizenship laws (waiting period, special requirements). Nevertheless, it serves as a good variable for describing closeness to the country. Acquisition of citizenship is considered as one of the most objective indicators of political integration (Paparusso, 2019). Happiness is often used as an overall indicator of wellbeing in scientific literature. Studies also show that there is a correlation between balanced multilingualism and family wellbeing (Macias, 2021;Müller et al., 2020). Happiness is measured by the ESS with the question: "Taking all things together, how happy would you say you are?". Participants are asked to rate their happiness on a scale from 0 (extremely unhappy) to 10 (extremely happy). In this paper, we use the difference between the average happiness of multilingual people and that of the rest of the population as an indicator of the gap in overall wellbeing. Due to the fact that it is correlated with many other wellbeing variables (health, income, social relations, etc.), we leave those other indicators out of the analysis because we assume that they are already covered with happiness. The only exception is economic coping. We use self-evaluated economic coping and assess the difference of multilingual families from the rest of the population. Economic coping is measured in the ESS questionnaire with the question "Which of the descriptions comes closest to how you feel about your household's income nowadays?" The answer options are 1. "Living comfortably on present income", 2. "Coping on present income", 3. "Finding it difficult on present income", and 4. "Finding it very difficult on present income". Feelings of discrimination mainly captures negative emotions. For this indicator, we use the difference between multilingual and monolingual groups. Perceived discrimination is related to negative feelings and subjective assessment of unfair treatment in society and is considered a good indicator for measuring the perceptions of society. The ESS asks respondents if they would describe themselves as being a member of a group that is discriminated against in the country. If respondents answer positively, a follow-up question asks, "On what grounds is your group discriminated against?" The respondents are not provided with a scale or a list of options to answer this question, but their open-ended answers are recoded into different categories. Because we study multilingual families, we are particularly interested in perceived discrimination on the basis of language. Feelings of discrimination are sometimes dependent on awareness of discrimination and, therefore, may be difficult to compare across countries. In order to tackle the methodological issues, we use a variable showing the gap between multilingual people and the rest of the population.
After the descriptive analyses of variables, we conduct a statistical hierarchical cluster analyses to identify countries that share multilinguals with similar characteristics. We are interested in clusters of countries because they describe the environment and conditions in which multilingual households live. The cluster analyses method helps to group together objects that are the closest from each other according to difference between input variables. Its method uses repeated calculation of distance measures between objects, and between clusters once the objects begin to be grouped into clusters (Institute for Statistics Education, 2021). The outcome is usually represented graphically as a dendrogram, where countries that are close to each other according to input variables also stand close to each other on the figure.
The following is the list of input variables in the cluster analysis. Their detailed description and data (Tables 2 -3) of input variables are available earlier in this paper: -ESS respondents who speak two or more languages at home -Multilinguals among the population not born in the country -Multilinguals among the population born in the country -Second-generation immigrants among multilinguals born in the country -Multilingual households with children (difference from the country average of households with children) -Multi-ancestry persons among multilinguals, -Citizenship among multilinguals (difference from the rest of the population) -Happiness feelings (difference of multilinguals from the rest of population) -Multilinguals who belong to a minority group that is discriminated in the country (difference from the rest of the population) -Discrimination because of language -Difficulty with economic coping (difference from the rest of the population).
All variables except the share of persons from multilingual households were divided by 10 for the analyses. This gives a stronger impact to the first indicator, the level of multilingualism in the clusters, because this is the most important indicator for our analyses.

Results
The share of persons from multilingual households in European populations varies from 2 % to 38 %. Table 2 shows that Switzerland is an extreme exception, followed by Germany, Spain, Belgium, Ireland, and France. These are the countries with the largest percentages of persons from multilingual households, whereas Czechia, the UK, Norway, Poland, and Portugal have a very low share of multilinguals. Estonia, Finland, Lithuania, Netherlands, Slovenia, and Hungary are in the middle with 6 -9 %.

Immigration, demographic background, and languages spoken at home
Not surprisingly, the percentages of multilinguals in the immigrant group are higher compared to percentages in the native-born population ( Table 2), showing that multilingualism has roots in migration in many countries. There is also a strong positive correlation between multilingualism in the immigrant and native-born populations. For example, Germany and Switzerland have the largest percentage of multilinguals in both groups, and the UK and Poland have the lowest (Table 2). Despite the link between immigration history and multilingualism, the background of multilingualism is not only a simple immigration story. The slightly positive trend between multilingualism and the total share of the foreign-born population has several deviations and outliers. For example, in Switzerland, close to 40 % of people live in multilingual households, but the foreign-born population is close to the European average (16 %). Also, Spain, the Netherlands, Lithuania and Czechia have comparatively more multilingualism in proportion to their share of foreign-born population. The United Kingdom stands out as a country with a large foreign-born population but a very low level of multilingualism (Table 2). As expected, a strong correlation is visible between multilingualism and recent immigration intensity (Figure 1). Switzerland is an outlier again as the country with a very high percentage of persons from multilingual households and the high average recent immigration rate. However, multilingualism in this country is comparatively higher than immigration indicators would predict in the case of a linear relationship. Also, Ireland presents high rates of recent immigration and high rates of multilingualism.
Germany, Spain and France stand out with relatively high rates of multilingualism compared to their recent immigration rates. Conversely, in Norway and the UK, multilingualism remains low compared to recent immigration rates (Figure 1).

Figure 1 Multilingualism and average recent immigration rate
To measure the long-term impact of migration on multilingualism, we also analysed multilingualism among in the country-born second-generation immigrants. The UK has the highest percentage of second-generation multilinguals, followed by Estonia and Slovenia, while the lowest numbers are in Poland, Hungary and Spain (Table 2). Next, we describe the language diversity in multilingual households (Figure 2). Official national and regional languages represent a large percentage of all languages spoken in multilingual households across all countries, except Lithuania. English is the most prominent foreign language across all countries. It is most prevalent in multilingual households in Poland, Hungary, Norway, and Portugal. Sweden, Slovenia and Spain have some of the lowest rates. Estonia and Lithuania have some of the highest percentages of non-EU languages among multilinguals, with Russian, Ukrainian and Belarusian featuring as predominant languages.

Figure 3 Ancestry background among multilingual respondents
We also analysed ancestry across multilingual household respondents (Figure 3). In all countries except the UK and Lithuania, ancestry of the country of residence prevails over the other groups. In the UK, only 27 % of multilinguals identify their ancestry with the country where they live, which is the lowest rate among multilinguals. Hungary, Spain, Finland and Poland have the highest rates of local ancestry. Spain's ancestry rates are higher due to the regional identities recognised in the country -e.g., Catalan, Basque, and Galician. Countries have substantial diversity in the non-European identity group. Non-EU ancestry rates are highest in Slovenia, followed by the UK, Estonia, and the Netherlands, but in several Central and Eastern European countries like Poland, Hungary, and Czechia, non-European identity is practically missing.
The other variable helping to capture the multicultural diversity of multilingual households is the indicator about self-identified multi-ancestry, i.e., respondents with at least two ancestries. Countries with rather high multiculturalism -Belgium, France and Spain -have the highest scores of diversity here with 40 -60 % (Figure 4), whereas in Switzerland, only 24 % of individuals from multilingual families have multi-ancestry identity. The UK again has the lowest identity diversity among the countries surveyed ( Figure 4). France and Belgium, the countries with the highest multi-ancestry levels, have close to 60 % local identity mixed with 16 -17 % other EU identity and about 21 -26 % non-European identity (Figure 3).

Figure 4 Multi-ancestry background among multilingual persons
The presence of children in multilingual families gives some indication about the possible future of multilingualism. Multilingual households have more children compared to monolingual households in all countries surveyed. Multilinguals in the UK also have the highest positive deviation from the average of the rest of population in terms of having children, followed by Finland, Sweden and Switzerland (Table 2). In Estonia, Hungary, Ireland, the Netherlands, Poland, Portugal, Slovenia and Spain, multilingual families do not differ from the rest of the population in terms of children.

Integration of persons from multilingual households
Generally, people from multilingual families are less likely to be citizens of their country of residence (Table 3). Comparatively, the highest gap can be seen in Ireland, the UK and Switzerland, followed by Germany and Norway. As previously shown, some of these countries also have high percentages of immigrants. The relative percentage of multilinguals who belong to a group that that they believe is discriminated against is highest in France, the Czech Republic and Belgium, but there is no difference from the rest of the population in Estonia, Poland and the UK (Table 3). We are especially interested in language and discrimination. The results show very small percentages of discrimination against multilingual respondents because of language. Lithuania, Estonia and Finland have the highest percentages, with around 5 percent of respondents feeling discriminated against because of their spoken language, while less than one percent feel discriminated due to the language they speak in Ireland, Norway, Poland, Portugal, Slovenia and Switzerland. Multilingual households also seem to have more difficulties in coping economically in comparison to the rest of the population. In this respect, the Netherlands has the highest percentages, followed by Belgium and Lithuania, while the lowest percentages are found in Portugal and Poland (Table 3). Finally, we also analysed how happy multilingual households are compared to the rest of the population (Table 3). Ireland and Hungary present the highest positive gap, whereas the Czech Republic and Lithuania have the most negative results. No differences are shown in Belgium, Estonia, Finland, Germany, the Netherlands, Poland, Sweden, Switzerland and the UK. Figure 5 shows the distribution of countries in five large clusters. Switzerland significantly stands out from the rest of the countries. A second cluster of countries is formed by Belgium, Ireland, Germany, and Spain and the third cluster by France and Sweden. The rest of countries are similar to each other, with only one outlier -the UK. Table 4 provides the main average characteristics for those clusters.

Figure 5
Groups of countries according to hierarchical cluster analyses. The most similar countries stand also closest to each other Switzerland forms its own group and is an outlier with many variables. It has the highest rates of multilingual persons (Table 4), multilingualism among immigrants, second generation multilingualism, and multilingualism among people born in the country. There are more multilingual household with children in Switzerland compared to other groups of countries. On the other hand, far fewer multilinguals have citizenship in Switzerland compared to other countries, although perceived discrimination is low, and no differences in economic coping or happiness can be found. 24 % of multilinguals have a multi-ancestry background. Altogether, Switzerland is the most multilingual country with long traditions of multilingualism and rather good environment for multilinguals.

Table 4
Average values in the clusters The third cluster formed by France and Sweden has fewer (12 %) multilingual persons compared to the two first groups (Table 4). However, these countries have high rates of secondgeneration multilingualism and multilingualism among immigrants (45 %). The rate at which respondents identified feelings of being discriminated against are the highest among the clusters. We can call this the discriminated multilingualism group.
The UK is distinct from the rest of the countries due to its very low multilingualism and multiancestry identity and the low citizenship rate of multilinguals (Table 4). However, it has an extremely high share of second-generation multilinguals and many multilingual households with children. It has the lowest rates of multilingual households among locally born people and immigrants. It may be that the power of the English language eclipses other multicultural traits. Feelings of discrimination in the UK do not differ between multilingual and monolingual households.
Finally, the cluster formed by the remaining ten countries (10 group) has generally low rates of multilingualism, including in all different sub-groups of immigrants (Table 4). Compared to the UK, multilinguals in this group identify more often as having multi-ancestry. Only secondgeneration multilingualism is below average in this cluster. The rate at which respondents reported discrimination is average, but the multilinguals have politically integrated at a high rate -the citizenship gap is the lowest of all the clusters.

Conclusions and discussion
Multilingualism has significance for the wellbeing, social capital, mobility, competitiveness and cohesion of multilingual family members, as well as for ensuring their affiliation with the society in which they live (Müller et al., 2020;Gaspar, 2010;Weenink, 2008). The aim of this study was to analyse the lives and backgrounds of persons in multilingual households in European countries and to classify the countries according to the environment in which multilinguals live. This study contributes to knowledge about multilingual families in Europe and the heterogeneous environments in which they live. The European Social Survey results show that the percentage of persons living in multilingual households varies from 2 % to 38 % in Europe. In Switzerland in particular, but also in Germany, Spain, Belgium, Ireland, and France, many residents live in multilingual households, whereas this rate is far lower in Czechia, the UK, Norway, Poland, and Portugal. In Estonia, Finland, Lithuania, Netherlands, Slovenia, and Hungary, the rates of multilingualism are moderate. Multilingualism is closely related to recent immigration and the migration history of a country, but this relationship is not linear. For example, Switzerland is at the high end of the multilingualism spectrum, and the UK is at the bottom, despite the fact that large proportions of the populations of both countries are foreign born. Countries present different associations between their foreign-born population and the percentage of multilingual households. These differences are a result of the immigration history and, in some cases, of state policies to create an environment that is favourable towards multilingualism. In this respect, analysing policies regarding the promotion of multilingualism could be the subject of future studies. The analysis of languages spoken at home among multilinguals shows that English features as an important second language in nearly all of the countries in this study. This is the result of the status of the English language as a de facto lingua franca in the European Union (Cogo & Jenkins, 2010), which can also have an impact on its use by international couples as the common language spoken in their households to overcome the barrier of learning the local language. Portugal, Poland and Hungary have the highest rates of English speakers in multilingual households. Estonia and Lithuania are top countries with the highest non-European Union language speakers in multilingual households, with Russian, Ukrainian, and Belarusian as the predominant languages. They are followed by Slovenia, where other Balkan languages coexist, and then Sweden and the UK, with closely equivalent rates between European Union and non-European Union languages. Another important finding is that multilingual households across all countries have lower rate of citizenship when compared with the average population and they have more problems coping economically. This is explained largely with immigration related background of multilingual households but deserves further investigations. Nearly fifteen percent of respondents from multilingual households across all countries believe that they belong to a minority subjected to discrimination, with the highest averages appearing in Czechia, France and Belgium. Further research will be necessary to understand the reasons behind this finding. The cluster analyses helped to identify the most similar country groups. Analysis reveals five types of multilingual countries in Europe. Switzerland forms a separate group, which is underpinned by the fact that the country was established through the principle of "unity in multiplicity". This principle emphasises the multicultural nature of the country and suggests the protection of cultural minorities (Matyja, 2018). The four official languages of the country have benefited from this principle, and our findings show that so do newly arriving immigrants. Conversely, the UK can be described as a multi-ethnic country where the English language prevails and low proficiency in this language can be regarded as a threat to social cohesion (Jaworska & Themistocleous, 2017). Latest UK census (UK Government, 2020) shows that only South Asians and people of Polish descent preserve their languages well. For example, in Ireland, where English is also an official language, the landscape of multilingualism is more diverse. This is because the Irish language shares official status with English (RTE, 2018). Also, Belgium, Spain, Germany, the rest of the countries in the highly multilingual cluster with Ireland, have a long history of multilingualism or immigration impact. Belgium has more than one official language, and four languages have official language status at the regional level in Spain. In Germany, the influx of immigrants arriving in the country in recent years is widening the multilingual landscape (Adler & Beyer, 2018). In the cluster formed by Sweden and France, multilinguals experience more feelings of discrimination. Despite their current multi-ethnic appearance, both countries' societies are facing a variety of challenges to integrate new arrivals (Hübinette & Arbouz, 2019;Leland, 2015). Finally, the fifth cluster formed by the remaining ten countries presents lower rates of multilingualism in comparison to the rest of the clusters. In most of these countries, there is only one official language.
Limitations and future analyses: This study uses a sample of countries taken according to the availability of full dataset, described in the methods section. We worked with data from 18 countries, but in the future, we would like to include more countries in the analysis.
To our knowledge, this is the first comparative paper about the life of multilingual families in Europe. Therefore, the results of the study provide a basis to study their different dimensions in further detail and, in particular, those related to identity formation and the internal and external motivations to develop an affiliation towards host countries.