INTRODUCTION: LANGUAGE TESTING AND CITIZENSHIP IN SWEDEN
The present article sets out to investigate an aspect of a public debate that took place in the context of the parliamentary election in Sweden in 2002. On 3 August 2002, Lars Leijonborg, then leader of the Swedish Liberal Party (Folkpartiet Liberalerna),1
Folkpartiet (lit. ‘the People's Party’) was founded in 1934 and acquired its current name Folkpartiet Liberalerna (lit. ‘the People's Party’ – the Liberals) in 1990. The party has historically profiled itself as a defender of the individual's “right to decide on his/her life and life choices” (www.folkpartiet.se), which has meant, inter alia, the public condemnation of and explicit action against every form of racism. As for electoral results, the Swedish Liberal Party has had varying electoral success, polling from 22.8% of the votes in 1948 to 4.7% in 1998. In 2002, the Swedish political landscape consisted of seven major political parties (here listed in descending order of results in the 2002 election): the Social Democratic Party (Socialdemokraterna), the Moderate Party (Moderaterna), the Liberal Party (Folkpartiet Liberalerna), the Christian Democratic Party (Kristdemokraterna), the Left Party (Vänsterpartiet), the Center Party (Centerpartiet), and the Green Party (Miljöpartiet – de Gröna). It is important to highlight that the Social Democratic Party has had a hegemonic position in Swedish politics, being nearly uninterruptedly in power since 1936, while the Liberal Party was part of the opposition, with the exception of 1976–1982 and 1991–1994, when the government was formed by center-right coalitions that also included the Liberal Party.
All translations from Swedish are my own.
“Integration” is a key word in Swedish political discourse. According to the official definition proposed in the government bill entitled “Sweden, future and diversity: From immigrant politics to integration politics” (Sverige, framtiden och mångfalden – från invandrarpolitik till integrationspolitik) (Prop. 1997/98), integration is “a reciprocal process” which pertains to both Swedes and non-Swedes and “refers to the possibilities to become part of a bigger entity without being required to sacrifice one's cultural and ethnic identity” (Prop. 1997/98:22ff). Nevertheless, the connotations of integration vary widely depending on the context. It will appear manifest in the extracts analyzed in the present article that integration is envisaged as a “one-way” process that only applies to immigrants, on the one hand, and is taken to be synonymous with immigrants' acquisition of alleged Swedish norms as a prerequisite to becoming Swedish citizen, on the other (cf. Blommaert & Verschueren 1998a:111–16).
According to Swedish legislation, when the government presents a bill before parliament, MPs are allowed to submit a counter-proposal in the form of a motion on the same topic of the bill. In connection with the budget bill, MPs can submit motions on any subject under the jurisdiction of parliament. Before being sent to a parliamentary vote, motions or government bills are first read and discussed in one of the 16 standing parliamentary committees, which writes a proposal of decision (betänkande). Each parliamentary committee covers specific areas of responsibility.
Issues related to citizenship and immigrants fall under the remit of the Committee on Social Insurance.
Admittedly, a language test for naturalization is still an issue in the Liberal Party's agenda (www.folkpartiet.se), and was brought up again – albeit differently – during the parliamentary election in 2006.
Of course, a debate on language testing for naturalization need not necessarily be viewed as an extraordinary phenomenon per se. Debates on the same topic – albeit with different degrees of heatedness – emerged in Europe (Denmark, Germany, the Netherlands, and the UK, to name just a few contexts) during approximately the same period (see, e.g., Blackledge 2005, Piller 2001, Stevenson 2006). Nevertheless, the Swedish debate may be regarded as particularly interesting for two reasons. First, the support for a language test transcended the boundaries of political affiliation. On the one hand, a language test for naturalization was not unanimously welcomed by all representatives of the Liberal Party.7
According to interviews published in the Swedish press, the language testing proposal was particularly cherished by the then leader of the Liberal Party, Lars Leijonborg.
Under a common stereotype, Sweden is often described as a country that has been historically homogeneous with regard to the ethnic composition of its inhabitants. Although it has been extensively demonstrated that this homogeneity is a discursive construction that fails to account for the historical presence of a variety of autochthonous ethnic groups on Swedish territory (e.g., the Finns, the Sámi), the ethnic composition of residents in Sweden indeed underwent a major change as a result of considerable immigration during the second half of the 20th century. As statistical data show,8
In Swedish statistical data, the Swedish population is classified according to individuals' place of birth, and divided into the categories ‘born in Sweden’ (inrikesfödda), which also includes those who were born in Sweden but are not Swedish nationals, and ‘foreign born’ (utrikesfödda), which also comprises those who were born outside Sweden but have been naturalized, as well as Swedish citizens born abroad. Furthermore, if one compares the most recently available statistical data on the number of non-nationals and naturalized nationals (31 December 2003), one can note that the percentage of naturalized nationals tends to be high with regard to people coming from Hungary, Poland, the former Yugoslavia, and the “South of the World,” rather than other “Western” countries such as Spain, Italy, the United Kingdom, or the United States (Statistiska Centralbyrån 2005).
I will employ the term “immigrant” rather than “migrant” throughout the article in order to be textually faithful to invandrare ‘immigrant’, which is a key word in the Swedish context.
The turn to multiculturalism was enshrined, inter alia, in three types of policies that are particularly relevant as a background to the issue under investigation in the present study: (i) the constitution, (ii) language-in-education policies, and (iii) citizenship policies. First, a paragraph of the constitution, which was amended in 1974, established that “opportunities should be promoted for ethnic, linguistic and religious minorities to preserve and develop a cultural and social life of their own.” (Kungörelse 1974). Hence, it can be argued that the state was accorded an increasingly more active role in the protection, defense, and support of minorities (cf. Joppke & Morawska 2003:13). Second, as for language, in 1977 the state endorsed immigrants' rights to maintain and develop their home languages (hemspråk),10
The official denomination was changed from ‘home language’ (hemspråk) to ‘mother tongue’ (modersmål) in 1997. At that time, there was a fear that ‘home languages' would naturalize a link between the languages spoken by immigrants and private domains (“homes”), thus undermining the aim of the new legislation to enhance these languages’ status in official/public domains (Hyltenstam, personal communication)
I am not making any claim about the actual implementation of these policies in local contexts, which indeed had diverse outcomes (cf. Hyltenstam 1996, 1999).
Two years for Danish, Finnish, Icelandic, and Norwegian citizens.
Although proficiency in the Swedish language is not mentioned in any of the Citizenship Acts, the government bill (Prop. 1950) preceding the ratification of the 1950 Citizenship Act (Lag 1950) emphasized that “the applicant's knowledge of the Swedish language should be given great importance in the naturalization process” (cited in SOU 1999:307). Moreover, the promulgation (kungörelse) of the 1950 Citizenship Act established that every application for naturalization should be accompanied by a certificate released by “a teacher, a priest or other competent person” (cited in SOU 1999:307) testifying to the applicant's knowledge of the Swedish language.
The relationship between language and citizenship, however, never left the political agenda and was brought up again by the Moderate Party (Moderaterna) during the 1994 electoral campaign (cf. Boreus 2006a), and was successively debated in 1997, when the government appointed a parliamentary committee to investigate whether the citizenship legislation should be revised, dual citizenship allowed, and the status of citizenship revalorized (Dir. 1997, 1998). The committee produced a report (SOU 1999), on the basis of which a government bill was later presented before parliament (Prop. 1999/2000). In the report, the committee suggested that the citizenship law should be amended and dual citizenship allowed. With regard to the revalorization of citizenship, the committee put great emphasis on the importance of Swedish language skills for all residents of Sweden (SOU 1999:313), and stated that “Swedish language skills are … one of the basic prerequisites of a functioning integration process” (SOU 1999:314). Nevertheless, the committee recommended that “requirements concerning language knowledge or knowledge of Swedish society should not be linked to citizenship” (SOU 1999:313) for “reasons of justice” (av rättviseskäl), given that “the conditions to learn Swedish vary to a great extent from immigrant to immigrant, and this depends on factors which are beyond any immigrant's control” (SOU 1999:316). Notably, the only member of the committee publicly disagreeing on the lack of language requirements was a Liberal Party member of parliament, Karin Ahrland. In the end, the Citizenship Act (Lag 2001), which was eventually ratified by parliament and is still in force today, did not establish any level of language proficiency nor any form of language requirement as a prerequisite for naturalization.
In conclusion, during the past 30 years or so in Sweden, the ideologies of multilingualism and multiculturalism/diversity have informed an official regime in which (i) multilingualism and multiculturalism/diversity are officially promoted, (ii) there is minimal difference between the rights accorded to citizens and those to non-naturalized residents, (iii) dual national affiliation is now allowed, (iv) skills in the Swedish language are viewed as irrelevant for citizenship, and (v) any form of language requirement has been rejected as unjust. Against this backdrop, I will argue in the present article that the public claims advanced in 2002 about the necessity of introducing a language test for naturalization are not simply about objective measurement or assessment of immigrants' language skills, but are the tangible manifestation of a competing language ideology – one could call it an ideology of language testing – that attempts to defy multilingualism and multiculturalism by tying proficiency in one language to knowledge of one culture as the compulsory prerequisite for the granting of rights of membership in Swedish society and the Swedish nation as an imagined community (Anderson 1991). It goes without saying that, in using the singular form “ideology,” rather than “ideologies,” I do not mean to convey that we are dealing with a perfectly coherent set of ideas, values, and representations programmatically and intentionally developed by the Swedish Liberal Party and shared by all supporters of a language test. As already mentioned above, claims in favor of a language test were not expressed solely by representatives or sympathizers of the Liberal Party. Nor were they advanced only by “ethnic Swedes.” However, the singular form “ideology” attempts to capture the dissonant coherence emerging from a “Bakhtinian carnivalesque cacophony of voices” (Stroud 1999:344), which, despite their heterogeneity, cluster into sets of discourses resting on interrelated arguments, values, and assumptions.
Specifically, I will approach the Swedish debate focusing on the textual claims that recognize the importance of introducing a language test. This restricted analytical focus is motivated by a desire to concentrate on the explicit and implicit facets of the ideology of language testing. This means that, through the analysis of relevant texts, I will examine what is manifest, but also – and most interestingly – what remains unsaid because it is presupposed or taken for granted. Furthermore, I will investigate whether there are tensions and discrepancies between these two dimensions. In brief, the argument is that, in a democratic society such as Sweden where public discriminatory or xenophobic discourse is barred by law,14
The so-called Lag om Hets mot Folkgrupp (lit. Act on Agitation against Groups of People) is a section of the Swedish Penal Code (Brottsbalk), which was ratified in 1948 and underwent some amendments, the most recent being in 1988 and 2002. The law criminalizes agitation against “groups of people” on grounds of race, color, national and ethnic origin, or religious creed, and since 2002 (the amendment entered in force on 1 January 2003), also on the basis of sexual orientation.
At this juncture, it is crucial to highlight that ideologies are not merely abstract systems of ideas, values, and beliefs existing in people's minds, but materialize in texts and discourses15
It lies outside the scope of this article to offer a comprehensive review of how different theoretical approaches conceptualize the contentious notions of text, discourse, and ideology, and their mutual relationships. Nevertheless, for present purposes I keep these concepts separate because I view them as belonging to different levels of abstractness (cf. Lemke 1995 and Weiss & Wodak 2003:15). Specifically, ideology can be defined as “any constellation of fundamental or commonsensical, and often normative, ideas and attitudes related to some aspect(s) of social ‘reality’” (Blommaert & Verschueren 1998a:25). However, Blommaert & Verschueren also remind us that “ideas can only begin to permeate social life and action when they find forms of expression” (1998a:26). That is, ideologies as powerful systems of signification become salient and tangible when they are encoded in discourse, understood as “language and other types of semiosis as elements of social life” (Fairclough 2003:26). Simply put, discourse and ideology are tied together insofar as discourses are what Sunderland 2004 refers to as the “epistemological sites” where ideology is materialized and thereby can be brought under close scrutiny. Finally, I envisage text as the most basic unit of semiotic “output” (Sunderland 2004:7) produced by individuals in social interaction (see also Wodak 2001). In Talbot's (1995:24) definition, texts are the most concrete and contingent “fabric in” which discourses and ideologies take form.
THEORETICAL FRAMEWORK
During the past 20 years or so, research on the links between language and policy making has undergone a major epistemological shift leading to a reconceptualization of the role played by language in sociopolitical processes (see also Heller 2006). A growing number of scholars in different disciplines began to argue that the existing body of research had been written from a static, structuralist, and positivist perspective, which envisaged language as an objective mirror of a preexisting social world (Blommaert 1996, Fairclough 1989, Watts 2001). As a response, these scholars urged that research should instead carefully attend to two ideologically laden interrelated processes: (i) the ways in which language (or better, discourse) is itself a crucial component in shaping those social categories (e.g., ethnic or national identity) that had previously been taken as givens; and (ii) the ways in which the discursive construction of social reality is deeply embedded in the (re)production or contestation of power asymmetries and domination. To paraphrase Mitchell Dean's (1994:4) characterization of research as a problematizing practice, scholarship on language and politics started asking (often uncomfortable) questions where others had located answers. In particular, three separate strands of research have been engaged in trying to untie the Gordian knot that links language, power, and ideology in policy making: (i) Critical Discourse Analysis, (ii) the field of Language Ideology developed within North American linguistic anthropology, and (iii) Critical Language Testing. In the remainder of this section, I will start by pointing out a few major commonalities and differences between these scholarly traditions, and I will end by foregrounding the reasons why they should be brought into dialogue to inform one another.
To begin with, Critical Discourse Analysis, Language Ideology, and Critical Language Testing can be broadly grouped together under the heading of “critical social research on language,” insofar as they all draw upon social theory (e.g., the work of Bourdieu, Foucault, and Habermas) in order to deconstruct and make transparent relations of power that contribute to the (re)production of social inequalities between individuals in specific contexts. Furthermore, these three fields of inquiry theorize language as a set of resources and practices to which individuals have differential access. This does not only entail that research should understand the reasons why linguistic resources are unevenly distributed (Blommaert 2005; Fairclough 1995b, 2003; Heller 2006). It also means that research should bring to light the overt and covert mechanisms – language testing being one (Shohamy 2006a, 2006b) – deployed by some individuals in order to determine the value of certain linguistic practices, thereby defining what counts as “good,” “correct,” or “appropriate” language (Blommaert 1999c; Heller 2001, 2006; Lippi-Green 1997).
Nevertheless, the focus on language and the reliance on social theory have been fairly distinct in these scholarly traditions, thus possibly accounting for the existence to date of a fairly clear-cut disciplinary boundary between Critical Discourse Analysis and the field of Language Ideology (for an exception see Blackledge 2003, 2004, 2005, 2006). In the different approaches subsumed under the label of Critical Discourse Analysis, the object of investigation is language as the medium whereby power is enacted and discrimination realized. Accordingly, preference has been given to accurate linguistic analyses of texts produced in various institutional contexts. As Fairclough (2003:204) puts it, “Without detailed analysis, one cannot really show that language is doing the work one may theoretically ascribe to it” (emphasis in original). By contrast, the scholars committed to the field of Language Ideology have drawn attention to discursive events in which language is not only the medium but at the same time the object of discourses through which the social order is produced, reproduced, or contested. Simply put, the question is not just to understand how social differentiation is produced through language, but how social hierarchies are subtly enacted on grounds of linguistic practices, and how the latter become inherently associated with certain speakers and other cultural categories such as morality or aesthetics (cf. Blommaert 1999a, Gal & Woolard 2001, Schieffelin, Woolard & Kroskrity 1998). By the same token, the more recent approach of Critical Language Testing (Shohamy 2001) focuses on language as both the medium and object of policy making. In fact, Elana Shohamy (2001, 2006a, 2006b) argues that through language tests policy makers overtly decide what language(s) should be assessed, thereby covertly affecting educational practices and determining what counts as legitimate language knowledge in a given society. In addition, following Bourdieu and Foucault, Shohamy (2001:117–28) points out that the symbolic power of language tests lies distinctively in their ability to shape social categorization not only in terms of the criteria of inclusion and exclusion (who is to take a test or not, who passes or fails a test, etc.) but also in terms of the values attached to such categories.
Besides the different focus on the function of language in social processes, another important element of difference, especially between Critical Discourse Analysis and the field of Language Ideology, is the way in which social theory is “put to work” (Chouliaraki & Fairclough 1999:16) to inform linguistic (or linguistic anthropological) analysis. At the risk of falling into undue oversimplification, because of the richness of theoretical and methodological approaches adopted by different scholars, one can nonetheless point out that much of the groundbreaking work within Critical Discourse Analysis (e.g. Fairclough 1989, 1992, 1995a, 1995b; van Dijk 1993b; Wodak et al. 1999) employs social theory as a mediating link between the linguistic/argumentative dimension of a text, on the one hand, and the social situatedness of that text, on the other. In line with a clearly political emancipatory stance, the aim of Critical Discourse Analysis is to unmask how texts embody and (re)produce what one could call “grand narratives of domination” – that is, broad ideological frameworks (neoliberalism, racism, etc.) that create or uphold social inequality. Accordingly, the focus of investigation is on texts and their potential ideological effects (see Blommaert & Bulcaen 2000 and Pennycook 2001 for critical perspectives on CDA). By contrast, often sidelining the purely linguistic aspect of texts, studies of language ideologies have taken a more materialist approach, which privileges attention to the “real historical actors” (Blommaert 1999b:7), together with their interests and stakes, in order to examine how available semiotic resources, coupled with certain sociocultural constraints, affect discursive production. In other words, in attempting to map the mechanisms whereby symbolic resources are produced and made more or less valuable, research on language ideologies often relies on social theory in order to trace how specific institutional positions, identities, and contexts of production and distribution are intertwined in (re)shaping relations of domination. This, in turn, has also entailed a crucial problematization of the role played by academics (linguists, sociologists, anthropologists, etc.) in defining boundaries between languages, social groups, and so on, thus often legitimizing political proposals (cf. Heller 2006, Heller & Duchêne 2007, Makoni & Pennycook 2006).
I argue that Critical Discourse Analysis, Language Ideology, and Critical Language Testing should inform one another because each provides us with important theoretical and methodological tools with which to make sense of the complexity of what Blommaert (1999b) calls a “language ideological debate” (see also Blackledge 2005 for a groundbreaking attempt to bring together Critical Discourse Analysis and Language Ideology). According to Blommaert, debates can be defined as “more or less historically locatable periods in which a struggle for authoritative entextualization takes place” (1999b:9). This does not only mean that (language) policy making is a process that engages different social actors (politicians, journalists, individual citizens, etc.) who stake claims in the field of language and thereby generate a multiplicity of texts (laws, policy proposals, newspaper articles, etc.). It also reminds us that the often apparently muddled web of interrelated texts is not produced in an historical vacuum, but is itself highly significative of the intersection of different temporalities – that is, “the here and now” and “the durable patterns which lie beyond the control or awareness of individuals” (Heller 2001:212). Or to put it differently, attending to the multilayered historicity of texts (Blommaert 1999c, Blommaert & Bulcaen 2000) produced in language debates can be crucial for understanding the role played by language in local/national responses to the supranational socioeconomic processes of change that go under the heading of “globalization” (Blommaert 2003, Fairclough 2002). This is not only true insofar as language in postmodern or late modern societies seems to have “become more salient, more important than it used to be, and in fact a crucial aspect of the social transformations which are going on – one cannot make sense of them without thinking about language” (Fairclough 2003:203). The reason also lies in the fact that the importance, or even the necessity, of proficiency in dominant, standardized national languages seems to have become a common argument whereby liberal democracies officially respond to the linguistic diversity ensuing from enhanced human mobility in a globalized world. In addition, the commodification of language, which has been singled out as one of the semiotic components of globalization (see Fairclough 2002, 2003; Heller 2003), rather than recognizing the linguistic diversity and hybridity of diasporic contexts, has entailed that language competence is conceptualized and represented as a set of bounded, marketable communicative skills, often devoid of symbolic functions, that can be advertised, bought, and sold, but also accurately measured and therefore tested (cf. Silverstein 1996:290; see also Leung & Lewkowicz 2006 for reflections on testing and assessment in a context of globalization). In sum, the linguistic tools developed by Critical Discourse Analysis (Fairclough 2003), informed by the semiotic model of how language ideologies tie images of languages to other cultural conceptualizations of the speakers of those languages (Irvine & Gal 2000: 37), and the notion of language tests as rites of institution (Bourdieu 1991; cf. Shohamy 2001) will be useful to shed light on the intersections between texts, discourses, and ideologies at work in a specific historical moment in Swedish society.
Before proceeding, however, it is necessary to make a few observations about the choice of the extracts on which the following analysis will be based. The extracts have been selected from a large corpus of policy documents and newspaper articles dealing with the issue of a language test for naturalization16
The corpus includes (i) policy documents related to the parliamentary committee on citizenship appointed in 1997; (ii) opinion polls conducted during the electoral campaign in 2002; (iii) policy documents produced by the Liberal Party on the language test issue; and (iv) 148 newspaper articles retrieved from two electronic databases (PressText and Mediearkivet) through a search for the keywords språktest ‘language test’ and medborgarskap ‘citizenship’, covering the period 1 April 2002 – 31 March 2003. The electronic search was conducted on 1 June 2005. The two databases cover a wide range of newspapers (broadsheets and tabloids, national and regional, and with different political orientations), which can be taken as representative of the Swedish media landscape. As the two key policy documents on language testing were put forward by the Swedish Liberal Party on 3 August 2002 and 18 February 2003 respectively, I opted for a time scale of one year, which would include the period between these two dates and the months immediately before and after them.
In addition, I have incorporated into the analysis some extracts from mainstream print media because they give a sense of how newspapers can help to (re)produce and disseminate ideologies, potentially affecting people's beliefs about the social world (cf. Blackledge 2005, Blommaert 1999b, Fairclough 1995a, Johnson & Ensslin 2007). The recourse to a broad range of empirical data seeks to give a more nuanced understanding of the ideology of language testing and to substantiate some of the conclusions without the researcher's falling into the trap of undue speculation (see Verschueren 1999 and Weiss & Wodak 2003 for valuable insights into the methodology of conducting Critical Discourse Analysis on a large corpus of data).
LANGUAGE, ICONIC BOUNDARIES AND THE MAKING OF ‘IMMIGRANTNESS’
Irvine & Gal 2000 have pointed out that language ideologies produce social group identities in discourse through semiotic processes of recognition and/or misrecognition of linguistic differences between those groups. In particular, they identify three interrelated semiotic processes, which they call iconization, fractal recursivity, and erasure. Iconization can be defined as a process whereby linguistic features or varieties that index social groups are transformed into “iconic representations of them, as if a linguistic feature somehow depicted or displayed a social group's inherent nature or essence” (Irvine & Gal 2000:37). Fractal recursivity “involves the projection of an opposition, salient at some level of a relationship, onto some other level” (38). Erasure is “the process in which ideology, in simplifying the sociolinguistic field, renders some persons or activities (or sociolinguistic phenomena) invisible” (38). Drawing on Irvine & Gal 2000, the first aspect of the ideology of language testing I seek to illustrate in this section is the way in which language knowledge and language practices are central features for constructing opposing social groups, and how a perceived lack of the Swedish language is construed as a problem for Swedish society at large. For clarity of exposition, I will first look at iconization and erasure, and will then move on to fractal recursivity.
“Olle” and “Berit” are used here as typical names of ethnic Swedes.
The common denominator – be it explicitly stated or implicitly presupposed – of the claims proffered by the proponents of a language test is that immigrants lack, totally or partially, the Swedish language. As one can see in the example above, which is taken from an op-ed article published in the popular daily Aftonbladet, the author uses the expressions ‘immigrants speak poor Swedish’ and ‘immigrant children without language’. It is also relevant to point out that these statements were written by a 19-year-old man who migrated from Iran to Sweden at the age of two (according to a biographical note at the end of the article). Analogous to the case described by Blommaert et al. 2005 in Belgium, the language proficiency of immigrants' children seems to be assessed here on the basis of “a language regime valid in a particular national order” (Blommaert et al. 2005:213; see also Silverstein 1996:285). In other words, different languages have different values in a specific national context, and this hierarchy of valued linguistic practices determines what counts as “good” or “bad” language proficiency. It appears clear in the extract that presumable multilingualism (in the languages spoken by immigrants) remains unrecognized, and the children of immigrants are represented as “language-less” (cf. Blommaert et al. 2005:213). Related to this is the fact that the lack of Swedish is not described as the result of the interplay of external social and economic factors, but as if it were an intrinsic characteristic of all ‘immigrant children’. In fact, the author links language knowledge to a widely accepted, albeit contentious factor, which, according to both experts (psychologists, psycholinguists, etc.) and laypersons, underpins human thought and human behavior, namely motivation18
One of the anonymous reviewers correctly points out that lack of “intention” to “integrate” is a common argument in other European debates on immigrants. I employ “motivation” in order to be faithful to the Swedish texts in which it represents a recurrent key word. Albeit contentious, the purported “lack of motivation” evokes the image of immigrants as lazy, and accordingly reluctant to learn the Swedish language. Therefore, in this context, a language test is represented as a form of pressure that would oblige immigrants to learn Swedish.
By tying language attainment to a purportedly intrinsic feature of human nature, the author portrays a lack of Swedish among “immigrant children” as if it were the exterior picture, or icon, that displays a trait of their deepest essence – the absence of motivation. Furthermore, the iconic relation between lack of motivation and lack of language knowledge is a causal one. Simply put, ‘immigrant children’ are deficient in Swedish because they lack motivation. This finding is consistent with Irvine & Gal's (2000:37) observation that “[iconization] entails the attribution of cause and immediate necessity to a connection … that may be only historical, contingent, or conventional” (my emphasis). Finally, by making the lack of language knowledge an icon of ‘immigrant children’, the author erases the social, ethnic, linguistic, and cultural diversity that characterizes this social group and replaces it with a general homogeneity. In other words, an extremely heterogeneous linguistic reality is represented as a linguistic absence, which is productively used to represent youths with different backgrounds as a homogeneous group.
Another aspect of iconization also needs to be addressed. In tying linguistic features and/or varieties to a social group, iconization more or less implicitly creates an opposition along what one could call an iconic boundary between that very social group and other social groups that, factually or purportedly, do not share the same characteristics. In the extract above, the lack of Swedish is not only an icon that portrays children with different ethnic and social backgrounds as an imagined homogenous group. A lack of Swedish also marks implicitly an iconic boundary between these children and all other children who do not lack the Swedish language – that is, ethnic Swedes.
Although there is no doubt that “[a] basic anthropological insight is that ways of talking about the ‘Other’ are ways of talking about ourselves” (Woolard 1989:276) and vice versa, I suggest the notion of iconic boundary as a conceptual tool to make more explicit and to scrutinize more precisely (i) the ways in which iconization operates in creating group oppositions, and (ii) how iconization melds with the other semiotic process identified by Irvine & Gal 2000, fractal recursivity. In fact, by applying the concept of iconic boundary, we are better equipped to understand how iconization creates a dichotomy between social groups, which is simultaneously at work on different coexisting and interrelated semiotic tiers, such as the ethnic, the linguistic, the moral, the political, and the economic. The analysis of the extract below, which is taken from the Liberal Party's report on language requirements (Folkpartiet 2003), can help exemplify this point.
The words infödda and invandrade pose some translation problems. They are both nominalized past participles that literally mean ‘the native-born’ and ‘the immigrated’, respectively. In the text, they are used synonymously with svenskar ‘Swedes’ and invandrare ‘immigrants’.
Drawing on a study published by the Swedish Integration Board (Integrationsverket 2000), the Liberal Party claims that there is a substantial difference between Swedes and immigrants in terms of political participation. Here the low degree of political participation manifested by immigrants is explained as the effect of a list of factors: lack of economic resources, lack of Swedish language skills, lack of civic skills, lack of civic competence, and lower employment rates. As Fairclough (1989:188) proposes, “Where one has lists, one has things placed in connection, but without any indication of the precise nature of the connection.” In this very example, however, the Liberal Party gives the reader an indication of how to interpret this list of causes. In fact, a deficiency in Swedish language skills is singled out as the main cause of immigrants' low level of engagement in political issues. Analogous to what has been observed above, the social and economic factors underlying second language attainment are obscured, and the wide linguistic diversity among immigrants is erased and replaced with a homogeneous deficiency. In brief, a lack of Swedish is treated as if it were an independent social variable in the light of which immigrants' political behavior can be explained.
If we look at the opposition constructed between immigrants and Swedes, the overt saliency put on Swedish language deficiency functions to demarcate an iconic boundary between immigrants, who lack Swedish, and Swedes, who do not. In addition, the opposition between these two groups is salient not only at a linguistic level, but it is also causally projected onto another domain, a political one, in which immigrants are opposed to Swedes in their degree of political commitment, as an inherent result of their lack of Swedish.
Political participation, however, is not the only domain affected by the lack of Swedish, as is illustrated in the example below, taken from an op-ed article written by Nalin Pekgul, chairperson of the National Federation of Social Democratic Women in Sweden.
Despite criticizing the Liberal Party's proposal, the author points out that the whole language testing issue is important because it has emphasized the importance of a common language in achieving a united society. The author goes on to give evidence of the significance of a common language. First, knowledge of ‘good’ Swedish provides access to the job market. Here, one should pay attention to two lexical items: the adjective ‘good’ and the adverb ‘obviously’. While ‘good’, as an evaluative marker of Swedish, adds an unspecified level of proficiency in the common language, the adverb ‘obviously’ makes this level of proficiency a self-evident condition to access the job market in an ‘easier’ way. That is, communicative skills in ‘good’ Swedish are attributed a central role in employability. Second, knowledge of Swedish is said to be a necessary condition to access the public sphere because it enables everyone who speaks it to ‘use his/her democratic rights’ and ‘participate in political life’. That is, a common language enables the speaker to be an active citizen.
Most interestingly, learning Swedish has another facet. The author claims that knowledge of Swedish invests immigrant adults with authority in their role as parents and frees their children from their role as interpreters for their parents. But what may sound like a trivial issue conceals subtle social and moral underpinnings. In fact, in the rhetorical question at the end of the extract, the author demonstrates a belief that adults, who it is assumed do not speak Swedish, acquire knowledge of Swedish society through their children, who have learned the Swedish language. Here, unlike in extract (1), it is presupposed that children do know Swedish, whereas their parents lack it. Furthermore, deficiency in the majority language is subtly tied to negligent parenthood. The arguments of the extract to some extent remind us of a well-known 1995 court case in Amarillo, Texas, in which a judge ordered a mother not to speak Spanish to her child as a condition for keeping the custody of that child, claiming, “If she starts [school] with the other children and cannot even speak the language that the teachers and others speak, and she's a full-blooded American citizen, you are abusing that child” (cited in May 2003:103; my emphasis). In other words, a lack of the majority language, and, by implication, proficiency only in minority languages, is linked to the incapacity to act as “proper” parents. Notably, in the Swedish example, proficiency in Swedish is not just envisaged as a communicative tool that parents need to pass on to their children, thereby granting them a better life and career success. Proficiency in Swedish is represented as a necessary prerequisite of discerning ‘what is right and what is wrong’ – that is, of moral knowledge.
In sum, the author in extract (3) implicitly creates an iconic boundary between those who can speak Swedish and those who cannot, and ties language skills to employment, social cohesion, authority, social/cultural knowledge, and morality. In agreement with Irvine & Gal's (2000:38) observations, a dichotomy created at a linguistic level is projected onto other domains (economic, social, cultural, and ultimately moral), and generates a causal relation along a chain of oppositions: (i) having/lacking the Swedish language, (ii) employment/unemployment, (iii) having/lacking authority, (iv) understanding/not understanding Swedish culture, and (v) having/lacking morality.
To conclude, textual analysis of the examples in this section has illustrated that a perceived or presumed knowledge of the Swedish language and its converse are fruitfully employed as semiotic resources to construct opposing group identities in discourse (cf. Blackledge 2005:176; see also Richardson 2004, who makes a similar point about the construction of Muslims as the “Other” in British media discourse by virtue of an alleged lack of cultural practices tied to Britishness). Moreover, the establishment of a causal relationship between the linguistic and the other levels of opposition entails that a lack of Swedish is construed as a social rather than a merely linguistic problem because it has economic, social, cultural, and ultimately moral implications for Swedish society at large. Or to put it another way, a lack of the majority language “is seen to consign one inevitably to the social and economic margins” (May 2003:103; emphasis in original), while possession of the majority language is turned into “a realizable asset that can be achieved so as to increase overall personal value in one or another recognizable symbolic paradigm” (Silverstein 1996:291). It is also possible to see that the general lack of the Swedish language as an icon of “immigrantness” erases the extremely diverse and varied ethnic, social, and linguistic composition of this heterogeneous group and replaces it with an alleged inherent homogeneity. Finally, although it is impossible to trace the exact ethnic, social, and religious characteristics that are implied or assumed under immigrant as a group category employed in the texts of the corpus, the extracts above provide a glimpse of the ambiguity carried by this category, which constantly slides from being restricted to those who have actually moved into Sweden (e.g., the adults in extract 3) to including all “people who [more or less] deviate from Swedishness” (Hertzberg 2003:56) (e.g., immigrant children in extract 1). As Hertzberg (2003:56) remarks, the immigrant in Swedish discourses “summarises a series of co-existing categorisations which originate from other phenomena than migration alone” and encompasses more generally all those who categorize themselves and are perceived as non-Swedes.
LANGUAGE TESTING AND SOCIAL COHESION
Although the previous section was primarily concerned with conceptualizations of language practices and how they are employed to construct social groups in discourse, it also emerged that different arguments were advanced to buttress the language testing proposal. Working with a definition of discourse as patterns of semiotic resources whereby social actors represent and signify the social world (see, e.g., Fairclough 2003), and keeping the focus on the categories of citizenship and language testing, it was possible to observe in the corpus of data that apparently diverse and multiple representations and meanings of these categories cluster together into a discourse, which can be termed a discourse of social cohesion. Far from being the only discourse at work in the debate on language testing, the discourse of social cohesion most clearly exemplifies the tensions and discrepancies between its explicit formulations and its implicit assumptions and implications.
Previous research (Blackledge 2004, 2005; Blommaert & Verschueren 1998a; May 2001) has pointed out that the notion of a shared language as a commonsense measure to achieve social cohesion is a frequent rhetorical trope in political discourse. Social cohesion is without doubt the pivotal topic in both extracts (3) and (4). In extract (4), the author of a letter to the editor published in the influential daily Dagens Nyheter concentrates on the communicative function of a common language, which makes communication possible between individuals sharing a common territory, whatever linguistic background they may have.
The author clearly states that one common language is a prerequisite for the good functioning of a country, although this seems to be more important in the public sphere than in the private one. Moreover, that one official language is a necessary precondition for the unity of the public sphere and the creation of social cohesion can also be inferred from the last part of the extract, in which an explicitly negative evaluation is made of those countries that have more than one official language.
While the communicative function of a common language is emphasized in extract (4), a more complex scenario emerges in extract (3), which was partly analyzed in the previous section. The author here overtly states that it is important ‘to have a common language if one wants to achieve a united society’. Two presuppositions are triggered by the irrealis statement, by the lexical choice, and by the high density of modality (cf. Fairclough 2003:171–73): (i) social cohesion is a desirable goal, and (ii) a common language is a prerequisite of social cohesion. As I have already sketched, the author goes on to present the benefits of having a common language. First, a common language makes it ‘easier’ to gain employment. Second, a common language enables political participation.
If one draws on an established distinction in sociology between Gesellschaft ‘society’ and Gemeinschaft ‘community’, and views them not as mutually exclusive terms of an opposition but as two coexisting dialectical dimensions of a nation-state (cf. Johnson 2005:122–24), one could say that a common language is presupposed in extract 3 to give access to the economic realm of the labor market and to the civic dimension of the rights and duties that pertain to Sweden as a society. Furthermore, the subtle and implicit links established in the last part of the extract between learning the Swedish language and acquiring knowledge of Swedish culture, which is equated with morality par excellence, are evidence that a common language is also assumed to give access to the symbolic/cultural dimension of Sweden as a community “imagined” (Anderson 1991) around a perceived set of shared cultural and symbolic features.
It is also relevant to point out that the author is not talking about the importance of the Swedish language only for those immigrants who might want to apply for Swedish citizenship, but more generally about the pivotal role language knowledge plays for all non-Swedes and their life in Sweden. In other words, taking the particular issue of language testing and citizenship as a point of departure, the author moves on to deal with a more general topic, integration between Swedes and non-Swedes.
That citizenship and integration are powerfully and problematically enmeshed in each other is further substantiated in the Liberal Party's report on integration (Folkpartiet 2002):
After explicitly maintaining that the issue of citizenship is important ‘in an integration perspective’, the Liberal Party suggestively proposes that citizenship should be ‘the external sign that one has fulfilled some criteria’. In other words, citizenship is viewed as a sign that symbolizes the institutional recognition of the fulfilment of a set of prerequisites that are allegedly indispensable if an individual is to be awarded the identity of Swedish citizen, and thereby fully enter Swedish society. However, it is not immediately clear what kind of criteria or prerequisites the proponents of a language test are referring to. An insight is given in a later section of the report:
Here ‘some’ knowledge of the Swedish language is taken to be a prerequisite of the possibility to ‘benefit from’ citizenship – that is, to be an active citizen in Sweden. This is so because knowledge of the Swedish language is said to be a fundamental precondition for acquiring knowledge of Swedish laws and other unspecified ‘circumstances’. Despite the vagueness of this formulation, ‘laws and other circumstances’ alludes not only to legal norms but also to all “control mechanisms – plans, recipes, rules, instructions … for the governing of behaviour” (Geertz 1973:21), which can be broadly called “culture.” Simply put, the Liberal Party's view of what is necessary to become a Swedish citizen is built on a linear relation between language knowledge and knowledge of a society and its cultural rules, as if understanding the latter necessarily required the former.
Taken together, these extracts demonstrate the idea that social cohesion and integration between immigrants and Swedes are achieved by way of a common language. Furthermore, textual analysis exemplifies the fact that this idea is founded on the presupposition that knowledge of a common language (Swedish) does not merely give access to the civic domain of rights, duties, and political participation and the economic sphere of the labor market, but is actually the only way that immigrants will properly understand a given society, together with its laws, life, and cultural norms. That is, a common language is a precondition for accessing both Swedish society and the symbolic/cultural dimension of Sweden as an imagined community. The fact that we are dealing with a symbolic/cultural dimension rather than a mere civic one of a presumptive citizen's rights and duties is further confirmed by the recurring claim that a language test is a “marker” of the importance of the Swedish language.
To recapitulate, the examples above illustrate that claims in favor of a language test for naturalization are not only about the importance of Swedish as a lingua franca that makes communication possible among individuals sharing a given territory, irrespective of their linguistic backgrounds. Nor are they solely about citizenship in the juridical sense of the term – that is, the statutory rights and duties of a citizen in a given society. These claims are just as much about the nation-state as an economic, political, and cultural system of signification (Bhabha 1990), which enables the process of imagining a sense of national “we-ness” on the basis of an alleged common culture (see also Stroud 1999). Specifically, within the discourse of social cohesion, language testing and citizenship function as a proxy for the rules and conditions of membership in the civic realm of privileges accorded to citizens, on the one hand, and in the symbolic domain of the nation as an imagined community, on the other. In other words, language testing and citizenship are powerful semiotic means through which social actors in a democratic society may discursively engage with the civic and symbolic dilemmas posed by immigration and the management of a de facto ethnically, linguistically, and culturally diverse society.
DISCUSSION: LANGUAGE TESTING AND SOCIAL FRONTIERS
Besides being related to the nation-state, the discourse of social cohesion is explicitly concerned with the creation of a unity or totality that comprises both immigrants and ethnic Swedes. In other words, one could say that we are dealing with a discourse of inclusion through which different social/ethnic groups are brought together to form a whole. However, in this section I will show that there are a number of tensions between the explicit argument that a language test may positively affect social cohesion at a civic and symbolic level, and the exclusionary effects of a language test for naturalization. Bourdieu's (1991) notions of rites of institution and habitus can be instructive in an attempt to explain these discrepancies.
Bourdieu starts from the consideration that rites of institution have the power to “act on reality by acting on its representation” (1991:119), and that “to institute, to give a social definition, an identity, is also to impose boundaries” (1991:120; emphasis in original). Bourdieu goes on to argue that a rite of institution can be defined as
an act of communication, but of a particular kind: it signifies to someone what his identity is, but in a way that both expresses it to him and imposes it on him by expressing it in front of everyone … and thus informing him in an authoritative manner of what he is and what he must be. (1991:121; emphasis in original).
That is, rites of institutions are socially performative acts that, in establishing or reasserting an imaginary boundary, also called a social frontier (Bourdieu 2000), impose normative identities on the persons these rites concern (cf. Stroud 2004:208). Furthermore, Bourdieu emphasizes that two boundaries merit particular attention in relation to rites of institution: (i) the boundary between those who have and have not yet experienced the rite, and, most interestingly, (ii) the boundary between these two groups and those “who will not undergo it in any sense” (Bourdieu 1991:117). As Bourdieu (1991:117) puts it, such rites “institut[e] a lasting difference between those to whom the rite pertains and those to whom it does not pertain.”
Following Bourdieu, language testing for naturalization can be described as a rite of institution insofar as it is a social ritual, which officially defines a compulsory precondition for being ascribed the identity of citizen – namely, knowing a given language. As a rite of institution, language testing for naturalization creates a given social reality for those who undergo it. In successfully passing the test, the non-citizens who prove their knowledge of a given language are deservedly transformed into citizens, and in failing the test, those who do not prove such knowledge are reaffirmed in their social position of non-citizens. In this way, a language test ascribes identities to and institutes a boundary between those who pass and those who fail the test. Furthermore, a language test for naturalization also demarcates another boundary between those to whom the rite pertains – immigrants – and those to whom the rite does not pertain, ethnic Swedes.20
It is true that from a strictly legal perspective, a language test would definitely not pertain to indigenous national minorities (e.g., Finns and Sámi). Yet, one has to bear in mind the ambiguity of the meaning of “immigrant” as a group category in Swedish discourse, which often encompasses all those who are not perceived as Swedes, despite their actual citizenship. As an example, one can take an interview with the leader of the Center Party, Maud Olofsson, who, in answering the question whether immigrants should pass a test to become Swedish citizens, answered: “No. To begin with, we have minorities in Sweden who do not have Swedish as their mother tongue, for example those who speak Sámi and Tornedalian Finnish. Should they lose their citizenship then?” (Till att börja med så har vi minoriteter i Sverige som inte har svenska som modersmål, till exempel de som pratar samiska och tornedalsmål. Ska de då tappa sitt medborgarskap?) (Svenska Dagbladet, 29 August 2002). By referring to national minorities such as the Sámi and the Tornedalians as non-native speakers of Swedish, Olofsson gives proof of the muddle of knowledge surrounding multilingual skills among minorities, citizenship, language testing and, most interestingly, who a test would pertain to. Furthermore, only since 2000 have five indigenous or historical minorities been accorded the status of national minorities (Prop. 1998/1999): Finns, Jews, Roma, Sámi, and Tornedalians. Before then, these minorities were grouped together in political discourse with immigrant minorities under the common label ‘immigrants and minorities’ (invandrare och minoriteter) (SOU 1977).
If we first consider the boundary between those who pass and those who fail the test, a language test for naturalization as a rite of institution is far from being a social practice that can achieve full inclusion of social groups into a civic unity of citizens with the same rights and duties. As an immense body of literature within the field of second language acquisition (SLA) has demonstrated (see the collection edited by Doughty & Long 2003 for a relevant sample of critical insights), a range of individual, linguistic, economic, and social variables may affect SLA in a variety of ways in different contexts. Given all the possible mutual combinations of these factors, uniform SLA is highly improbable, if not impossible. Furthermore, it is far from clear which role motivation plays for SLA and its interplay with the aforementioned variables, on the one hand, and which factors influence motivation, on the other (cf. Dörnyei & Schmidt 2001, Dörnyei & Skehan 2003). Finally, critical research on language tests (see Shohamy 2001 for a comprehensive overview) has underscored that the high stakes associated with institutional language tests, as in the case of tests for citizenship, rather than being an incitement to language learning, may have negative psychological effects on test takers, thereby affecting their language performances. Taking all these considerations together, one can conclude that there will always be somebody who fails a language test for naturalization, whatever level of language proficiency is established for passing it. After all, one might also wonder why a nation-state would introduce a test at all, if not in view of ruling out those who might fail. Consequently, there will always be somebody who will be excluded from rights and duties on a par with ethnic Swedes and other immigrants who actually succeed in the test. Against this backdrop, one can say that there is a paradox, which could be called a civic paradox, between a discourse of inclusion, which explicitly advocates a language test for naturalization as a practical means to achieve social integration, and the inevitable exclusion from the civic domain of rights and duties of the nation-state that befalls those who fail the test. Furthermore, this exclusion would be given legitimacy by the official status of a language test administered by state institutions.21
Some proponents of a test recognized that exceptions should be made in the case of old people, language impairments, and so on. Rather than mitigating the proposal, the recognition of the need for exceptions presupposes that the test is ultimately discriminatory against certain categories of people.
More subtle, perhaps, is the cultural boundary established between those who pass and those who fail the test. One of the assumptions underpinning the Liberal Party's proposal is that knowledge of the Swedish language necessarily implies knowledge and understanding of Swedish life and culture, which in turn is taken as a necessary precondition for being a Swedish citizen. I also showed that a language test is employed in the discourse of social cohesion as a semiotic resource that enables the imagining of the nation as a cohesive community, access to which can be granted or denied depending on parameters of language proficiency, which are also taken as parameters of cultural and moral understanding. This discourse of inclusion is analogous to the one studied by Parekh 2000 in the British context (see also Blackledge 2005:64). Parekh (2000:55) argues that “inclusion is offered on terms already set by the wider society, it involves assimilation, sharing current norms of what it means to be a British or a good citizen.” As this form of inclusion is based on given norms of Britishness (or, in the present case, Swedishness), it intrinsically forecloses the possibility of reshaping these norms, and thereby precludes the recognition and acceptance of each person's individual traits, which is the basis of a truly multicultural society.
However, such a discourse of “inclusion,” in which membership is accorded upon fulfillment of given linguistic and cultural norms, conflicts with the Liberal Party's explicit aim that “Sweden shall be a multicultural, tolerant, and human society” where people's distinctive cultural traits should be respected:
In other words, there is a paradox, which could be termed a cultural paradox, between an explicit attempt to recognize and accommodate cultural diversity and to construct Sweden as a multicultural and tolerant society, on the one hand, and on the other, the more or less implicit presupposition that knowledge of one language (Swedish), and by implication, of one culture (the Swedish one) is the precondition for becoming a Swedish citizen. This can be taken as an example of what Billig (1995:87) calls a “syntax of hegemony,” in which one part of a totality, in this case the Swedish language and culture, is reasserted through a language test as the metonymic representation par excellence of Sweden and Swedishness. In this way, the Swedish language and its supposed necessary counterpart, Swedish culture, are given greater worth, while all other languages and cultures de facto belonging to Sweden as a polity are devalued as irrelevant to citizenship. This entails that the only acceptable identity as a citizen here is that of an individual who speaks the Swedish language and knows the Swedish culture, while all those who cannot prove that they meet these criteria need to be officially barred (cf. Blackledge 2005:213ff), which is evidently in conflict with the acceptance of a multiplicity of individual distinctive cultural traits advocated by the Liberal Party.
Finally, Piller 2001 and Stevenson 2006 convincingly argue that discourses of citizenship are deeply enmeshed in discourses of national identity. However, as Blackledge (2005:52) points out, this does not mean that “affiliation through national identity and affiliation through citizenship are … necessarily the same.” Rather, identification with the nation as an imagined community (Anderson 1991) and membership in the nation-state as a political body may be discursively constructed in opposition to each other, for example by way of a language test, which sets a boundary between those to whom a language test pertains – immigrants – and those to whom it does not, ethnic Swedes. This boundary can be understood if we analyze the way in which Sweden conceives of the nation-state in the light of the aforementioned distinction between society and community.
In the case of Sweden's response to immigration, there is no doubt that since the abandonment of assimilationist policies in the 1970s, there has been an attempt to organize and manage civic society according to the principles of equality and ethnic, linguistic, and cultural diversity, which is constantly reiterated in official documents, and is strongly advocated by the Liberal Party as well (see extract 7). Nevertheless, one cannot but recognize that Sweden as a symbolic community has been historically constructed around an ethnic, cultural, and linguistic commonality, which allegedly characterizes the majority group (ethnic Swedes) and has been made to stand metonymically for the nation as a whole (Hertzberg 2003, Löfgren 1993, Oakes 2001). As far as language is concerned, the previous sections and several other studies (Hertzberg 2003, Stroud 2004) have demonstrated that “good,” “correct,” or “perfect” command of Swedish, as opposed to deficient or hybrid varieties of the Swedish language, are relevant ways of demarcating “Swedes” from “non-Swedes” in a variety of private and public settings. Moreover, the “flagging” (Billig 1995:93), or constant reiteration, of this alleged ethnic, cultural, and linguistic commonality reminds Swedes of their belonging to a national imagined community.
What emerges here is the double edge of national identity as both a construction that is continually produced and reproduced in various discourses, on the one hand, and on the other, as a constructive cognitive framework through which the individual can experience a sense of belonging to an extended community. As previous research (Wodak et al. 1999) has pointed out, the constructive and constructed aspects of national identity can be captured by way of Bourdieu's notion of habitus. Bourdieu (1977:72) defines habitus as a system of durable dispositions resulting from the material conditions that constitute a particular type of environment. In Blackledge's (2005:32) interpretation, habitus is “a way of being which has been inculcated through patterns of behaviour of the group in its history, culture, language and other norms” (my emphasis). In this respect, habitus is shaped by a specific set of objective historical, cultural, and social conditions. Nevertheless, Bourdieu emphasises that habitus is not only constituted by but also constitutive of social reality. In fact, the dispositions that constitute the habitus do not belong only to the mind but are “embodied, turned into a permanent disposition, a durable way of standing, speaking, walking and thereby of feeling and thinking” (Bourdieu 1977:93–94). Through the process of embodiment, “the habitus gains a history and generates its [own] practices [over] time even when the objective conditions which give rise to it have disappeared” (Nash 1990:433–34, cited in May 2001:45). Simply put, although it does not foreclose the possibility of change, the Bourdieuan habitus helps explain the reproductive processes through which members of a social group acquire specific sets of dispositions in early socialization, and how these dispositions are reified as shared tangible habits of thinking, talking, behaving, and so forth, which can only exceptionally be fully acquired and inscribed in the bodies of those who have not undergone the same socialization process. This can be explained as a result of the fact that “habitus, as a product of history, ensures the active presence of past experiences which tend also to normalise particular cultural and linguistic practices, and their constancy over time” (May 2001:46; my emphasis).
If read through the lens of Bourdieu's notion of habitus, Swedish national identity is, on the one hand, the product of reiterated acts through which Swedes represent themselves as a “we-group” who share a purportedly homogenous ethnic, cultural, and linguistic habitus. On the other hand, as “the habitus implies a ‘sense of one's place’ but also ‘a sense of the other's place’” (Bourdieu 1990:131), national identity is a cognitive framework through which Swedes can both identify and position themselves within the imagined community and, at the same time, identify and position themselves in opposition to others who do not fully share the same habitus – inter alia, by means of linguistic practices that deviate from a perceived homogeneous Swedish norm. This means that, as long as Swedish national identity is produced and reproduced around a given linguistic, ethnic, and cultural commonality, people with different or hybrid habituses (e.g., immigrants, but also to a certain extent their children) will always be excluded from the imagined national community.
Against this backdrop, a statutory language test administered by state institutions is a social ritual that would not contribute to achieving a totality encompassing Swedes and immigrants on equal terms. Rather, a language test, based on a static and given relationship between one language, one culture, and one community, would further legitimate a symbolic boundary between a national community of “we,” ethnic Swedes (svenskar), who are assumed to share a common ethnic, cultural and linguistic core, and a “they”-group of Swedish citizens (svenska medborgare), naturalized immigrants who, despite having given proof of their knowledge of the Swedish language, will never be unreservedly perceived as entirely fitting into the given common ethnic, linguistic, and cultural Swedish norm, which is a prerequisite for access to the symbolic/cultural dimension of the nation as an imagined community.
CONCLUSION: WHY LANGUAGE TESTING FOR CITIZENSHIP? WHY NOW?
In the previous section I attempted to single out the paradoxes underlying the language testing proposal and its potential ideological effects. I will conclude by trying to position the emergence of the Swedish debate against the wider panorama of ongoing historical processes of change and their effects on Swedish public discourses of language and immigrants. We know that language and citizenship have been more or less overtly present on the Swedish political agenda during the past 20 years or so. As mentioned in the introductory section, the proposal of introducing language requirements for granting Swedish citizenship was advanced by the Moderate Party during the 1994 political campaign, the terms and conditions of Swedish citizenship were brought under discussion of a parliamentary committee between 1997 and 1999, and the 2001 citizenship law did not mention any level of proficiency in the Swedish language as a prerequisite for naturalization. Interestingly, there was no major public response on any of these occasions. Nonetheless, a heated debate on the issue did emerge in 2002. In the remainder of this section, I will take up Shohamy's (2006b) challenge to ask ourselves: Why language testing for citizenship? Why now?
Following Blommaert 1999b, 2005, answering these questions requires an investigation of three interrelated factors that enable the emergence and development of a language ideological debate: (i) the specific interests of the social actors involved in the debate, (ii) the discursive conditions about what is allowed to be said or not at a specific historical moment, and (iii) the discursive resources available at that very moment.
Given the role of the Liberal Party as the central actor in the debate, the first aspect we need to consider relates to the party's interests in bringing the issue of language testing for citizenship onto the political agenda only a year after the new citizenship law had been passed. In this regard, the language testing proposal can be viewed as a critique of the previous official decisions on immigrants' naturalization process. Furthermore, the fact that the language testing proposal was raised during an electoral campaign might be interpreted as a strategic move to win votes and possibly the election, which in turn entails a presumed consensus of the electorate on the issue (extracts 1 and 4, albeit not necessarily representative, could be taken as examples of such consensus at the grassroots). The assertion that language testing for naturalization was a component of an electoral tactic could be strengthened furthered by the observation, a posteriori, that language requirements were seldom the topic of public discourse after 2003, but were one of the topics brought up again by the Liberal Party during the electoral campaign in 2006.
On the other hand, there are a few elements that might reveal that the debate on language testing for naturalization is the manifestation of an ongoing ideological change within the Liberal Party, coupled with a modification of the conditions regulating Swedish public discourse on immigrants. As Kristina Boreus 2006a, 2006b observes in a detailed analysis of Swedish electoral campaigns since 1988, there has been a clear shift in rhetoric whereby the Liberal Party gradually moved from advocating the rights of immigrants and asylum seekers, and, by implication, underscoring the state's intervention to grant these rights, to highlighting immigrants' duties toward the state. Specifically, Boreus (2006b:73) shows that in 1988 the Liberal Party explicitly focused on the importance of respect for other cultures and religions, and therefore the need for defending immigrants' status in Swedish society. By contrast, the picture was quite different in 2002, when “immigrants” became associated, and thereby nearly synonymous, with “problems.” This rhetorical shift has been interpreted as the manifestation of a move toward a hybrid ideology in which a neoliberal emphasis on individual freedom unconstrained by state intervention is coupled with a neoconservative “enhancement of certain values such as religion, family and nation” (Boreus 2006b:80; my translation).
Nonetheless, according to Boreus, the rhetorical shift is not an exclusive feature of the Liberal Party but is part of a more general trend among most Swedish parties to represent immigrants in contexts of social problems, such as criminality, dependence on subsidies, school failure, unemployment, and female oppression. In particular, as far as the 2002 electoral campaign is concerned, there seems to have been “an accrual of links between immigrants and negatively valued phenomena which immigrants themselves were made responsible. These links constituted a pattern which was not contradicted in other parts of the material” (Boreus 2006a:182). Or to put in another way, negative other-presentation (cf. van Dijk 1993b) seemed to have become “an established cognitive scheme in 2002 while it had occurred only sporadically in previous electoral campaigns” (Boreus 2006a:182). Boreus suggests that this change can be understood if read in the light of the modification in the discursive conditions of Swedish political discourse linked to the emergence of the right-wing populist party New Democracy (Ny Demokrati) at the beginning of the 1990s. After its creation in 1990, the party immediately gained media recognition and political success, polling 6.7% of the vote in the parliamentary election in 1991. However, the fortunes of New Democracy soon declined, and the party did not manage to be represented in the Swedish parliament in the following elections. The relevant aspect is that New Democracy was the first party in Sweden that openly advocated the abolishment of state-financed home language instruction, and publicly associated “immigrants” with “problems” (e.g. criminality, subsidies, and HIV) (Boreus 2006a:82). Despite recognizing that there is no evidence of a “direct influence” of New Democracy's rhetoric on the other parties' later electoral publications, Boreus (2006a:186) argues that New Democracy was a conduit of discriminatory and racist discourses, which were brought into authoritative public arenas such as media and parliamentary debates, thus contributing to making these discourses somewhat acceptable and legitimate.
One should also add that the beginning of the 1990s was marked by a crucial event in Swedish politics: Sweden's accession to the European Union. As I have demonstrated elsewhere (Milani 2006, 2007), EU accession played an important role in rescaling the discursive regime of language in Sweden. To put it succinctly, until the 1990s the status of Swedish as the official national language had been taken for granted, and its function as a symbol of national identity had been played down by Social Democracy since World War II (cf. Oakes 2001), thus leading (among other things) to the absence of a law sanctioning Swedish as the official national language. In this context, the recognition of Swedish as de jure official language in the EU institutions, together with the perception among Swedish academics and politicians of the position of Swedish in the EU as a language “threatened” by English, led to an increased salience of Swedish as an explicit target of political intervention aiming at securing its status vis-à-vis English.
Having established the factors that might have affected the change of conditions underlying public discourses on language and immigrants, what remains to be demonstrated now is why specifically the introduction of a language test for citizenship was proposed in 2002. This leads us to the issue of what discursive resources were available at this particular historical moment. This, in turn, can be understood if we view the Swedish debate not as an isolated phenomenon but as the local outcome of transnational discourses on language testing that were circulating in those years. It lies beyond the scope of the present article to trace the exact relationships between the Swedish context and all the other European debates. Nonetheless, it can be useful to present briefly some of the links between the Liberal Party's proposal in Sweden and the amendment of language policies for naturalization in Denmark.
Danish politics experienced a major change in 2001, not only because there was a shift to the right after nine years of Social Democrat led governments, but also because the newly established government (a coalition of Venstre, the Danish Liberal Party, and Konservative, the Danish Conservative Party) entered an agreement with the right-wing populist Dansk Folkeparti, the Danish People's Party, and took a hard line on immigration. This resulted in an amendment of citizenship legislation in several respects (1 July 2002), which entailed, inter alia, (i) a tightening of restrictions for family reunions, (ii) an extension of the period of residence required for citizenship application, and (iii) a formal documentation of knowledge of Danish language, society, history, and culture as a prerequisite for the granting of Danish citizenship (see Circulæreskrivelse 2002). The interesting aspect here is that these changes were advocated by a so-called liberal party in a context of geographical proximity only two months before the Swedish Liberal Party published its language testing proposal. Although it might appear obvious that there is a spatiotemporal connection between the Danish and the Swedish debates, the Swedish Liberal Party constantly refuted any comparison between their proposal and the restrictive legislative amendments in Denmark. As Lars Leijonborg forcefully put it in a newspaper interview, “There is a diametric difference between us and them. We want more immigrants and they want fewer, so a comparison does not apply” (Svenska Dagbladet, 4 August 2002). Similarly, in a later TV report on the Swedish Liberal Party (Dokument inifrån, 7 May 2006), Leijonborg denied once again any possible influence from Denmark. Nonetheless, in the same interview, the former leader of the Swedish Liberal Party, Maria Leissner, openly revealed that the Danish Liberal Party had been taken as a model for developing new political strategies since the mid-1990s.
To conclude, language testing for citizenship is certainly not a new concern typical of late modernity. However, it seems to have become an increasingly relevant object of discourses in European contexts of migration. Taking Sweden as a case in point, I have shown that the textual claims in favor of a language test are a site not only where opposing social identities of Self vs. Other are constructed, but also where identity politics is shaped (see Blackledge 2005, Schieffelin & Doucet 1998, Schmidt 2006, Stroud 2004, and Woolard 1989 for similar conclusions about the effects of different language ideologies in a variety of settings). That is, these claims are ultimately about who is or should be allowed to be a Swede, and thereby included in the symbolic community, on the one hand, and on the other, who is not and accordingly needs to be legitimately excluded (cf. Blackledge 2005:37). Through the misrecognition of language practices (the lack of Swedish) as icons of immigrants, a language test is explicitly called for as a practical way of diminishing social differentiation, and as a measure of “enablement” (Silverstein 1996:300) of immigrants. However, as a rite of institution, a language test actually contributes to rather than challenges the reproduction of social differentiation, thereby legitimizing the exclusion of certain groups from both the civic and symbolic domains of Sweden as a nation-state (see also Blackledge 2005, 2006). To paraphrase Jacquemet's (2005:263) suggestive description of the conceptualization of language competence in diasporic contexts, the increased prominence of language testing for citizenship is the manifestation of a language ideology that some people, when threatened by human mobility, activate in order to raise the membership bar that regulates access to the in-group at a specific historical and discursive juncture.
