Académique Documents
Professionnel Documents
Culture Documents
Bail Reviewed work(s): Source: American Sociological Review, Vol. 73, No. 1 (Feb., 2008), pp. 37-59 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/25472513 . Accessed: 20/10/2012 06:09
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review.
http://www.jstor.org
Recent or
studies
report boundaries
"symbolic
variation groups
to construct
of "us
of countries, salient
more
than others
processes by first panorama of these poorly understood. This article provides the " the relative salience or "configuration ofmultiple symbolic boundaries in comparing 21 European countries. I usefuzzy-set analyses ofdata from the2003 European Social Survey to create a typologyof symbolic boundary configurations.The results indicate that thesymbolic boundaries deployed by thegeneral public do not correspond to the " Moreover, thedata official "philosophies of integration emphasized in the literature.
suggest previous comparisons in other to explain have focused regions of Europe too heavily where variation data from of how on Western immigration using Europe, began overlooking more recently. I important generate variation hypotheses
quantitative
with
these hypotheses
studies
toward
theory of "boundary-work."
INTRODUCTION
the boundaries of countries neat Although divide people into social groups, the ly conceptual distinctionsused to constructnotions
of "us" and "them" are an equally important
component of social
Direct William
HarvardUniversity, Kirkland Street, 33 Cambridge, MA 02138 (bail@fas.harvard.edu). This research was made possible by fellowshipsfromthe German Marshall Fund and the National Science Foundation (IGERT #98070661). I thankJasonBeckfield, Eric
Rogers Neil Kastoryano, Bleich, Brubaker, Gross,
Ragin, Graziella Silva, William JuliusWilson, ASR editors and review Christopher Winship, and the
ers for their insightful comments and suggestions. All errors are uniquely my own. Previous drafts of this presented the Council at theASA
Douglas 1966; Jenkins 1996). Explaining cross national variation in the relative salience of these "symbolic boundaries" is the central goal of the boundary-work literature (e.g., Kastoryano 2002; Lamont and Molnar 2002; Wimmer 2005). This literature shows thatsocial identities are not only multidimensional but also highly mutable. While religion is an salient symbolic boundary in certain extremely countries, it is largely irrelevant in others and displaced by race, language, or culture?in dif ferent configurations?still elsewhere (e.g., Lentin 2004; Triandafyllidou 2001; Wieviorka 1994). Although a number of idiosyncratic explanations have been provided for such vari ation, they have yet to be synthesized into a theoryof boundary-work. Building on previous small-scale comparative studies, this article advances the study of boundary-work by pro viding the firstpanorama of symbolic bound aries toward immigrants in 21 European
countries.
Annual meeting in of European Studies Annual and the International Conference Science in Tokyo, Sociological Japan. Review,
Immigration is of natural interestto scholars of boundary-work because it reveals the sym bolic boundaries deployed when social bound
2008, Vol. 73 (February:37-59)
American
38
AMERICAN SOCIOLOGICAL REVIEW immigrants and immigration regimes across theContinent. I begin by developing a framework for the macro level and study of boundary-work at the a brief overview of immigration to providing Europe from 1945 to 2003.1 then review the lit erature on symbolic boundaries in threeregions of Europe, highlighting macro-level factors within and between regions. Next, I develop a typology of symbolic boundary configurations by applying a combination of "fuzzy-set" tech niques to data derived from questions about a hypothetical immigrant in the 2003 European
countries, but also due to variation in their understanding of diversity itself. Britain, for example, practices multicultural race relations (Favell 2001), whereas the use of racial cate gories is prohibited under the tenets of repub licanisme inFrance (Weil 2002). Until recently, nationhood inGermany was cast in terms of ancestry (Kastoryano 2002), while thousands of expatriates are denied cultural and legalmem Greece each year (Kiprianos, Balias, bership in and Passas 2003). So-called "Dutch tolerance" rests on religious accommodation (Rath et al. 2001), but Swedish multiculturalism has a dis tinctlysecular heritage (Runbolm 1994). These "philosophies of integration" (Favell 2001) are central to the "xenophobophelia" (Stolcke 1995) of European policymakers wary of being compared with the "race-obsessed" United States. It is yet to be determined, how ever,whether these distinctions also shape the configuration of symbolic boundaries deployed by the general public. This question is central
to numerous
aries are crossed.While Europeans once looked askance at theU.S. "color line," a recent influx of non-Western immigrants has brought ten sions to the fore.The murder ofTheo Van Gogh in the Netherlands in 2004, the fatal beating of a Chinese student in Ireland in 2002, and the perennial debate over laicite (secularism) in France exemplify the severity and variety of such tensions. Today, Europe is perhaps best described as a set of "diverse diversities," not only because of variation in the ethnic and cul turalbackground ofminority populations across
Social Survey.The results indicate thatthe sym bolic boundaries deployed by the general pub lic do not correspond to theofficial philosophies of integration emphasized in the literature. Moreover, the data suggest previous compar isons have focused too heavily on Western
Europe, overlooking variation across other
regions where immigrationbegan more recent ly. I develop hypotheses to explain this new found variation using demographic, and historical data socioeconomic, institutional, from a variety of quantitative and qualitative sources. The discussion and conclusion offer examples of how these hypotheses can be com bined by future studies toward a theory of
boundary-work.
SYMBOLIC BOUNDARIES
The recent "boom in boundary studies" (Wimmer 2005) highlights the significance of social classification across a wide variety of
These studies show considerable vari
Western Europe (e.g., France, tion countries in Germany, and Britain) where immigration began in the immediate postwar period (e.g., Brubaker 1992; Favell 2001; Kastoryano 2002). In con trast, the emerging literature on the "new" immigration countries of Southern and Eastern Europe emphasizes theabsence of philosophies of integrationamong these regimes (e.g., Lentin 2004; Triandafyllidou 2001; Wieviorka 1994).
Because these small-scale comparisons are com
comparisons
of "old"
immigra
contexts.
ation in the "boundaries" developed by groups to separate themselves from others (Abbott 1995; Barth 1969; Baubock and Rundell 1998; Douglas 1966; Jenkins 1996). Boundaries have both social and symbolic dimensions; this arti cle examines the latter. Symbolic boundaries are distinctions made by social actors "conceptual ... [that] separate people into groups and gen erate feelings of similarity and group member
ship." Conversely, "social boundaries are
partmentalized by region, only idiosyncratic explanations for the configuration of symbolic boundaries have been produced. Macro-level comparisons within and between regions of Europe are needed not only to contextualize previous research,but also to explore significant variation in thehistorical, demographic, socio economic, and institutional characteristics of
objectified forms of social differences mani fested in unequal access to an unequal distri bution of resources... and social opportunities" (Lamont and Molnar 2002:168). To be sure, symbolic and social boundaries are closely relat ed.While social boundaries are institutionalized, however, symbolic boundaries shift through
SYMBOLIC BOUNDARIES IN 21 EUROPEAN COUNTRIES classification struggleswhere majority groups attempt tomaintain the privileges attached to their status (Eliasoph and Lichterman 2003). "Only when symbolic boundaries are widely agreed upon can they take on a constraining character ... [and] become social boundaries" (Lamont and Molnar 2002:168). Citizenship laws, for example, are rigid social boundaries, but they are predicated on the flexible distinc
39
tions of symbolic boundaries, which are need ed to define such exclusion (Bryson 2006; Sackmann, Peters, and Faist 2003). In this way, symbolic boundaries are a "necessary but insuf ficient" condition for the creation ormodifica
and timingofmigration, (2) the size and origin of immigrant groups and theirposition in the labor market, (3) citizenship and civic inclusion policies, and (4) philosophies of integration. of Migration
tion of social boundaries and should therefore be viewed as "equally real" (Lamont 1992). To unravel the complex relationship between symbolic and social boundaries, studies of boundary-work emphasize the multidimen Whereas sionality andmutability of the former. social psychological theories of social identity require that groups be categorized as "in
groups" or "out-groups" (e.g., Tajfel 1981), the boundary-work approach I propose requires attention to the relative salience or configura tionofmultiple symbolic boundaries (e.g., race,
language, culture, or human capital). religion,
As Table 1 shows, postwar immigration gener Western Europe ally occurred much earlier in than in Southern and Eastern Europe. Castles andMiller (2003) identifythreegeneral trends of immigration to Western countries between 1945 and 1970: (1) refugee movements after World War II, (2) guest-workermigration from the European periphery, and (3) postcolonial migration. Germany accepted the bulk of refugees in the immediate postwar period,most ofwhom were fleeingEastern Europe. Postwar labor recruitment Southern Europeans, Turks, of
and Moroccans was most prominent in France,
This not only adds much-needed precision to the concept of social identitybut also enables one to ask whether the configuration of symbolic
boundaries competition reveals for social the interests resources. of groups For example, in
Britain, Germany, Switzerland, Belgium, Luxembourg, theNetherlands, Sweden, and Austria. Substantial postcolonial migration from Africa, Asia, the Caribbean, and theMiddle East occurred inFrance, Britain, Belgium, and the Netherlands. After 1970, European integration and eco nomic restructuring altered the sources and tim of immigration flows (Brochmann 1996). ing Although postwar labor migration bolstered the
economies
how the social boundaries aries, one can identify previously protected by race are renegotiated. In this way, theboundary-work literatureattempts to explain why majority groups choose certain
(Goldberg 2006; Lamont 2000). By examining the entire configuration of symbolic bound
tries,global economic decline in the 1970s led borders. Large-scale immi many to tightentheir gration continued, however, through family reunification policies. By the mid-1980s, countries such as France, Germany, and the
of many
Western
European
coun
symbolic boundaries, incorporating some groups while excluding others. The manner in which symbolic boundaries are policed ormade permeable reveals the strategic?although often
subconscious?interests of majority groups.
Netherlands had significant second-generation immigrant populations. Meanwhile, Southern European countries began to experience sub stantial immigration from Latin America, North Africa, and Eastern Europe for the first time. The collapse of the Soviet Bloc and civil unrest in Africa and the Middle East ushered in a new
40
Table 1. Timing ofMigration, 1960 to 2000_ Average Net Migration (Weighted by Total Population) 1960 to 1970
Austria .08 .12 Belgium Britain .01 Czech Republic -.19 Denmark .06 Finland -.37 France .42 .20 Germany Greece -.46 .01 Hungary Ireland -.58 Luxembourg Netherlands -.18Italy .48 .07 .00 Norway -.12 Poland -1.39 Portugal .05 Slovenia -.22 Spain Sweden .03 Switzerland_.58_-^14_.39_.36 Source: Note: Eurostat. occurred between 1945 and 1960 in several of the above countries but data are not Significant migration available for this period. .42 .95 .14 .24
1970 to 1980
.11 .18.30 .11 .03.14 -.04 .11 .00 -.02 -.04 .04 .01 .09.12 .12 .19 .26 .41 .28 -.02 -.02 .76 .24 .10 -.09 .42 .34 .04 .10 .20 -.21 .13 -.06 .14 -.08 .22 -.16 -.02 .68 .09 .08 .09
1980 to 1990
1990 to 2000
.25 .04
.33 -.58
wave of refugeemigration in the mid-1990s. At the same time, illegal migration of non Union European (non-EU) immigrants increasedwhile legal obstacles preventing intra European Union migration began to dissolve. Characteristics Today Variation in the timing and sources of postwar migration toEurope caused vast discrepancies in the size and regional origin of foreign-born populations. Table 2 describes foreign-born populations as a percentage of the total popu of the Foreign-Rorn
pies of such variation, European immigrants from non-EU countriesmake up 6.83 percent of Austrian population but only .47 percent of the the French population. As Table 3 shows, the varied timing and sources of immigration have also led to variation in the role of immigrants within European labormarkets. Citizenship and Civic Inclusion
lation by region of origin.While the foreign born constitute 40.10 percent of the total pop ulation in Luxembourg and 20.45 percent in make up only 2.88 and 1.98 Switzerland, they in Hungary and percent of the population Poland, respectively. Although those born in the Middle East and North Africa are roughly
4.40 percent of the population in France, they constitute less than .03percent of the totalCzech population. Among many other possible exam
made thedevelopment Each of theabove factors of citizenship and civic inclusion policies more pressing in certain countries than in others. While citizenship policy iswell established in "old" immigrationcountries such as France and Germany, many "new" immigration countries such as Spain and Italy did not develop mid-1980s. Among the citizenship laws until the old countries, there is variation in the criteria used for citizenship decisions and even thevery A definition of an immigrant. now classic com parison is drawn between France, where all second-generation immigrants are entitled to citizenship (jus soli), and Germany, where until recently those without blood-based ancestry
- 5*
Totala (Non-EU-15) EU-15 South North Africa Asia East America Caribbean America Oceania
.02
.57 .24 .14 .18 .39 .13 .04 .03 g ? 2.88 Italy .22 3.91 .04 2.48 1.23 .01 .94 .07 .03 .00 Hungary
Spain 5.33 1.55 .40 .89 .18 .05 .13 1.83 .06 .24 j* .01 %1.05.01 7.301.98 2.43.41 .96 Poland .03 .00 .71 1.51 .86 .00 .53 .03 Norway .37 .33 w 3.38
Note: Foreign-born populations by World the to Bank aggregated Cross-national according region P data coding were origin scheme. regional immigrant second-generation of on .08 .08
Belgium Finland3.308.82.201.19 Ireland 10.68 5.80 .70 2.21 1.00 .16 .39 .19 .18 .04 .01 Germanyb 2.32 rf 2.54France.5210.24.08 10.19 .81 .22.47.90Greece.05 7.23 1.02.14.01^2.77 .16 .62 .05 .10 .43 .32 .60 .172.42.10 .01 .20 .03.04 6.44 8 .08.01 .15 .01 .02 10.01 .14 1.58.22 .16 4.40 .39 .19 .42 .12 .01 .07 .65 .02 .21 O Denmark 6.72.42 2.00.56 1.07.65 1.58 .17 .21 .02 .04 Austria 12.12.14 2.60 .33 6.83.07 1.94 .06 .11 .00 .02 10.07 g Netherlands 34.281.51 40.10.36 2.752.67 .511.98 .13.69 .69 .20 .99 .63 .31 1.76 .35 .19 Luxembourg .07 .08 .03 ?
Sweden 12.03 4.29 2.84 2.36 .55 .37 .68 .67 .20 .03 .04 * Switzerland 20.45.50 11.34.41 4.66.67 1.43 .57 .12 .07 C
apercentages Theis sumnot of stateless regional does not? because equal the total who immigrants are or whose origin unknown are shown.
Table Regional 2. Origin the of Foreign the Percentage Born Total Population, of 2002 a as Czech .01 Republic 4.38.01 .27 .17 3.85.01 .03.03 .01 .00
for
Germany
(data
are Statistiches for Slovenia). from available not populations not are currently available. 2
Bundesamt. O
42
Table
Country_Force
Czech
Republic
1.6 Denmark 3.5 Finland 1.6 France 5.2 9.0 Germany Greece 9.5 1.0 Hungary Ireland 6.5 ? +1.1 Italy 3.8 45.0 3.8 3.6 Norway 2.7 Portugal 3.7Spain Sweden 4.6
Luxembourg Netherlands
Switzerland_2L9_2^0_+3_ Source: OECD (data not available a the total population. by Weighted for Slovenia and Poland).
(jus sanguinis) were treated as foreigners regardless of their birthplace (Alba 2005; Brubaker 1992). The distinction between these "civic" and "ethnic" citizenship regimes has been applied throughout Europe (Weldon 2006), although it isheavily criticized for failing to cap
Philosophies
of Integration
Citizenship policies are closely related to the philosophies of integration (Favell 2001), or public ideologies about exclusion and inclu
sion of immigrants, that are created by many
within the civic side of thisdichotomy (e.g., ly Brubaker 2001; Joppke 2005). Table 4 shows each country's score on the Civic Citizenship and Inclusion Index (Geddes et al. 2005), a
ture the nuances within each category (Kuzio 2Q02; Kymlicka 1999). A growing number of most states now fall firm studies suggest that
European governments (see Table 4). Most philosophies of integrationdraw upon the lega cy of nation building or colonial strategy.For example, French republicanisme stresses total while British mul assimilation of immigrants, ticulturalrace relations follows a pluralistmodel loosely based upon a similar colonial policy.
is considerable variation along each dimension. For example, naturalization isgenerously award
Until 2000, Germany's Ausldnderpolitik (for eigner's policy) treated immigrants and their children as "permanent guests" entitled to very few benefits from the state. Many Southern and Eastern European countries either do not have philosophies of integrationor are in theprocess of developing them (Carrera 2006). Those coun
Finally, the index suggests immigrants have more easily integrated into the labormarkets of Belgium
Austria,
Luxembourg, and Denmark. Similarly, Finland and Sweden have flexible family reunification policies, whereas Greece and Austria do not.
ed to immigrants in France regardless of their origin but more tightly restricted inAustria,
tries currentlydeveloping philosophies of inte gration are responding to increasing calls within the EU to design a "Common Agenda for
and Spain
Integration" among all member stateswhereby immigrantsobtain "basic knowledge of thehost society's language, history, and institutions" (European Parliament 2005).
and Greece.
Policy) eg (Foreigner's
and
g Participation
(Multicultural Society) O
France 100.82 113.59 106.28 111.41 96.94 Republicanisme (Republicanism) Multicultural 99.47 102.04
Race
99.03
Sweden 119.89 104.85 113.53 107.14 Mangkulturellt 96.12 81.63 102.04 103.86 117.35 (In development) (In Gedogen development) g(Dutch g Tolerance) 86.73 Inter-Culturalism 86.73 (Varies region)107.63 by Spain 91.84 101.94 Auslanderpolitik 127.55 103.86 (In E development) Belgium Denmark 133.51 107.77 122.45 111.11 (Varies 69.48development) region) 78.64 by 77.29 81.63 103.45 (In 99.03 development) Finland 100.82 115.94 102.04 107.43 (In 93.5111.41 107.43 99.47 81.56 75.6 95.49 113.4 109.42 Residence Labor Market Inclusion Philosophy of Integration15 domain. for Czech in inclusiveness indicate Republic, Data scores Higher available policy each Slovenia, Switzerland. Poland, Norway, not and on details a measFor the more are O bcountries integration Philosophies of modified been have in Q 2003. created immigration generation 2000; several policy Germany's in or extensively since revised second was Long-Term
64.03
107.77 91.79
gj None Poland ? ?
91.51
immigrants
can
now obtain
citizenship
regardless
For of an ancestry. (2005), Geddes OECD, Source: qualitative colleagues and ^sources listed Table in 5.
overvie
None Slovenia ? ? *o
44
AMERICAN SOCIOLOGICAL REVIEW (2001) suggest the legacy of "pillarisation," or religious accommodation, has encouraged reli gious tolerance in the former countries,where as the highly political Muslim community in Britain has provoked public backlash. Not unlike racism, however, there is evidence that overt has become stigmatized in "Islamophobia" many Western countries as well (Cesari and McLoughlin 2005; Kastoryano 2002). Previous studies also suggest that language most important sym and culture are two of the Western European coun bolic boundaries in tries. Again, the literature focuses on the centrality of language and culture in philoso phies of integration. In France, for example, government demands that immigrants learn the language and culture throughle creusetfrancais (the French Melting Pot) provoked a public backlash thatstressed la droit a la difference (the right tobe different) (Todd 1994). More recent ly,Brubaker (2001) identified a strong resur gence of assimilationist rhetoric spurred by the success of the far right in France. Similar, although perhaps less virulent, debates about assimilation have emerged in Britain (Bleich 2003; Favell 2001) and Germany (Kastoryano 2002). The question of assimilation is fiercely con tested inFrance, Britain, and Germany, but the debate is less prominent among noncolonial precedent (Garner 2003; Wimmer 2002; Zolner Danes use the 2000). Zolner (2000) argues that of Grundtvigianism or "bounded principle equality" to distinguish themselves from the colonial atrocities perpetrated by theirneigh bors. Similar observations have been made of "transethnic" patriotism in Switzerland, which stresses linguistic and cultural pluralism, albeit within strictEuropean limits (Wimmer 2002). A final question in the literatureon symbol Western Europe concerns the ic boundaries in possible convergence of attitudes toward immi grants at the regional level.A number of stud in the ies point to growing similarities Western European coun of policies immigration most face similar challenges of tries now that immigrant integration (Brubaker 2001; Joppke 2005). Joppke (2005), for example, argues that most Western countries have shifted the crite ria of citizenship decisions from the character and istics of groups to the credentials voluntarism of individuals. Such arguments are
powers where assimilation has no historical
While race is central to discussions of symbol ic boundaries in the United States, its relevance is highly contested in Western Europe, where race is not defined through hypodescent but ratherthroughnation building, colonialism, and theHolocaust. Analyses of racism in Western often focus on the relationship between Europe philosophies of integrationand public attitudes. For example, Favell (2001:226) emphasizes the "colorblind" ethic offepublicanisme inFrance vis-a-vis multicultural race relations inBritain:
"In France, racism
it has become privatized and Still others suggest that "private" unspoken." racism is equally prominent inFrance, albeit dis guised as republicanisme (Lapeyronnie 1993; Wieviorka et al. 1992). Comparative Todd 1994; in Britain
...
is public
and
spectacular;...
Western Europe. throughout Given the widespread stigmatization of Western Europe, religion has become racism in a primary focus of theboundary-work literature in this region (Zolberg and Long 1999). For argues that example, Goldberg (2006:349) World War II created a "shift inEurope's dom inant fixation of concern and resentment from the figure of 'the black' ... to that of 'the Muslim.'" Again, however, the literatureshows cross-national variation in the relative important salience of religion in the configuration of sym bolic boundaries. Kastoryano (2004) argues that anti-Muslim attitudes are more public in France than inGermany because secularism is an integral aspect of Civic Republicanism, whereas German secularism allows religious pluralism.While theFrench openly demand the
"nationalization"
historical studies of France and Germany sug gest race became part of national identity through conflict between the two nations, long before the arrival of non-Western immigrants (Brubaker 1992;Weil 2002). Another strandof the literature compares the emergence of antiracist attitudes (Lentin 2004; Taguieff 1991) and shows amore general denunciation of race
aremore likely toviewMuslims as Gastarbeiter (guest workers) whose cultural differences are to be toleratedbut not incorporated.Comparing anti-Muslim attitudes in the Netherlands, Belgium, and Britain, Rath and colleagues
of Islam,
she argues,
Germans
SYMBOLIC BOUNDARIES IN 21 EUROPEAN COUNTRIES provoked in part by the growing harmoniza tion of immigration policy at theEU level and an international human rights discourse that stresses civic criteria in citizenship decisions. Although supranational pressures have been shown to produce convergence of government policies, it is not yet clear whether these forces have produced similar effectson public opinion. A number of studies describe an emergent tide of xenophobia based on the notion of "fortress and Favell 1999; Europe" (e.g., Geddes Goldberg 2006; Kastoryano 2002), but it isnot yet clear which symbolic boundaries are most salient in these developing attitudes.These find ingshave scarcely been tested empirically, par ticularly outsideWestern Europe. Southern Europe
45
in nationalism in the place pivotal Mediterranean (Muro and Quiroga 2005). Only recentlyhas the arrival of non-Christian immi grants provoked scholars to analyze the role of symbolic boundaries based on religion (Zapata
via popular culture (Lentin 2004). The role of religion in the configuration of is also unclear, even symbolic boundaries has historically enjoyed a though Christianity
selves as racist (Kiprianos et al. 2003). There is tentativeevidence, however, that Western-style racism isbeing "imported" to Southern Europe
The literatureon boundary-work in Southern Europe highlights the region's abrupt transition from emigration to immigration in recentyears. Comparing Greece, Italy,and Spain with theold immigration countries ofWestern Europe, Triandafyllidou (2001) concludes that symbol ic boundaries in Southern European countries are much more unstable. Indeed, previous research suggests ethnicity and culture in Southern Europe have historical antecedents based on the unique "mixed" character of the Mediterranean, marked by North African and Middle Eastern influences long beforeWestern
states came into existence.1 For these reasons,
Barrero 2003). There is limited evidence that religion is a more important symbolic bound ary in Greece than in Italy (Triandafyllidou 2001). The relative salience of symbolic bound aries based on language and culture has yet to be studied in detail, although both were strong components of colonial strategies in Spain and Portugal (Medrano 2005; Mendoza 2001). on Instead, the literature Southern Europe focus es primarily on perceptions of economic threat induced by the abrupt increase of immigrants in the labormarket and high levels of unemploy ment and illegal migration (Apap 1997; 1997). Sniderman and colleagues Baganha (2000) reportthatItalians have negative attitudes toward immigrants with low human capital. Likewise, Kiprianos and colleagues' (2003) analysis ofmultiple public opinion surveys sug Greeks are among the most likely of gests that all Europeans to blame immigrants for high unemployment. Previous research also suggests that human capital is an important symbolic boundary against the largeAlbanian popula tion in Italy (Vasta 1993) and Greece (Lazaridis and Psimmenos 2000). Europe
previous studies have concluded that race is less salient inSouthern Europe than in Western Europe (Medrano 2005; Triandafyllidou 2001; Wieviorka 1994). For example, Sniderman and (2000) report no difference in colleagues Italians' attitudes toward immigrants from Africa and Eastern Europe. Similarly, previous survey analysis suggests thatGreeks are the least likely of all Europeans to describe them
Eastern Because
1 For were
instance,
race
inGreece
is derived
from the
(Triandafyllidou 2001). Similarly,la razza italiana (theItalianrace) and italianitd (Italianess)do nothave
ancestry
explicit racial connotations, although there is evidence of racist attitudes among Northern Italians toward their southern counterparts, who often have darker
skin(Vasta 1993).
immigrationtoEastern countriesbegan on very recently,the literature symbolic bound aries inEastern Europe primarily focuses on the ethnic "unmixing" (Brubaker 1996) of people communism. brought together under these studies is the Complicating migration of "national" minorities such as the Roma (Brubaker et al. 2006). A small but growing literature has begun to compare how these fac tors shape attitudes toward the rapidly growing fromAsia and population of "new" immigrants the Middle East inEastern Europe (e.g., Nyiri 2003; Phalet and Orkeny 2001; Wallace 2002).
46
AMERICAN SOCIOLOGICAL REVIEW symbolic boundaries based on race, religion, lyzed in the literatureat large, butmost studies focus only on two or threeof these dimensions. Insofar as the theoryof boundary-work empha sizes themutability of symbolic boundaries, to inattention theentireconfigurationrisks over
language, culture, and human capital are ana
Several studies conclude thatof all Europeans, most hostile toward for those in theEast are the eigners (Nyiri 2003; Wallace 2002). The com mon assumption is thatculturalmembership is based on race or ancestry in these countries, although this has been challenged in recent
years (Janmaat 2006; Nyiri 2003; Szoke 1992). Nevertheless, Wallace (2002) reports racism is higher inEastern Europe than inother regions, even thoughnon-European immigrants make up an extremely small proportion of their total population. There is some evidence of growing racism towardAsian immigrants in theCzech Republic and Hungary (Nyiri 2005), although negative attitudes aremost frequently directed African andMiddle Eastern immigrants toward (Nyiri 2003). Other studies suggest thatreligion is a more important symbolic boundary in the Czech Republic than inHungary and Poland (Wallace 2002). There is also evidence that lan guage is a strong symbolic boundary inPoland (Nowicka 2006), as is culture inHungary and theCzech Republic (Nyiri 2003). Finally, recent studies show that symbolic boundaries based on human capital are particularly strong through out Eastern Europe (Nyiri 2003; Phalet and Orkeny 2001; Wallace 2002).
looking symbolic boundaries thatdisplace oth ers. Therefore, the primary goal of this article is toproduce a typology of symbolic boundary
configurations using data on multiple symbol ic boundaries from countries indifferentregions of Europe. The secondary goal of this article is to iden new macro-level variables thatcan be used tify
ly cite country-level phenomena, the compart mentalization of the boundary-work literature has prohibited systematic analysis of historical, demographic, socioeconomic, and institutional variation among countries within and between Western European countries focus on citizenship laws and philosophies of integration,but they neglect the demographic and socioeconomic on factors emphasized in the literature Southern
regions. For example, most comparisons of
to explain theconfiguration of symbolic bound aries.Although explanations of variation in the configuration of symbolic boundaries routine
and Eastern Europe. Conversely, studies of Southern and Eastern countries often overlook the institutionalfactors central to the literature onWestern Europe. Below I explore the rela tionshipbetween my typology and the four axes of variation across countries identified above: (1) the sources and timing of immigration, (2) the size and origin of immigrant groups and
isons ofWestern European countries outnum ber comparisons of Southern and Eastern Europe. While comparisons of France, Britain, and Germany are commonplace, the amount of variation between these countries has yet to be assessed in a wider European context. This is importantnot only because of variation in the development of social boundaries between natives and immigrants within and between regions, but also because it limitsassessment of the possible convergence of symbolic bound Western countries themselves. Finally, aries in
comparison.
Second,
compar
of integration.2 In thisway, I provide the first systematic analysis of these variables in broad cross-national perspective that can be used by futurestudies to develop amore comprehensive theory of boundary-work.
their market, (3) citizenship position in the labor and civic inclusionpolicies, and (4) philosophies
http://www2.asanet.org/journals/asr/2007/
47
Author(s) Bovenkerk, Miles, and Verbunt (1992) (1993) (1994) Todd (1994) et al. (1994) (1998) (2000) (2001) (2001) (2001)
Countries
Britain, France, Netherlands France, Germany Britain, France France, Germany, Netherlands France, Switzerland Race, Race, Italy, Sweden, Human (United Italy Human Race, (United States) Culture, Language, Human Religion,
(1990)
Brubaker (1993) Lapeyronnie and Solomos Wrench Ireland
France, Britain, Germany, States) France, Britain, Belgium, (1997) ApapItaly, Spain Britain, France
(2000) Denmark,
Culture, Language Culture, Human Capital, Race, Capital, Race, Religion Religion, Race, Culture, Language Culture, Race,
Britain, Netherlands Italy, Spain, Britain, France, (United States) Czech Republic, (Slovakia) Race,
(2001)
Germany (2002) Kastoryano Wallace (2002) France, Germany, Poland, Hungary, Religion, Culture Human Capital, Race, Language, Race Race, Human Capital, Culture, Religion
Germany, Austria, Bleich (2003) Britain, France Garner Nyiri (2003) (2003) Ireland, Britain, Poland, Hungary, Slovenia, Rydgren Sackman (2003) et al. (2003) Fetzer and Soper (2004) (2004) (2004) and McLoughlin (2005) France,
Culture
(Romania),
Germany, Netherlands,
Britain, France, Germany Britain, France, Germany Britain, France, Italy, Ireland Britain, France, Germany, Netherlands, Spain
Italy,
a in parentheses are not included inmy study. Countries b The vast comparative literature on social boundaries is not described Soysal 1994).
2005;
Joppke 2005;
members of an ethnicminority in theircountry.3 My total sample comprises 33,258 individuals in 21 countries, averaging 1,584 people per
country.
I derive measures of symbolic boundaries from a unique set of questions in theESS that
two questions were used to deter following status: (1) "Are you a member of a minority group in your country?" and (2) "Have you ever been discriminated against on the basis of your mine minority race/religion/ethnicity?" 3 The
48
AMERICAN SOCIOLOGICAL REVIEW Conversely, race is least important in all coun In in varying degrees. tries, although Luxembourg, for example, themean score for racial symbolic boundaries is .93,but it is 4.12 in Hungary. Religion is relativelymore salient, averaging 3.52 across all countries and ranging as high as 5.87 inGreece. Symbolic bound aries based on human capital (education and occupation) generally fall between themost important (language and culture) and the least important (race and religion) symbolic bound
aries.
ask respondents to evaluate a hypothetical immigrant. Respondents were first asked: "Please tellme how importantyou think each of these things should be in deciding whether someone born, brought up and living outside [country] should be able to come and livehere." They were then shown a card with the follow ing statements: (1) be white, (2) come from a Christian background, (3) speak [one of] the official languages of [country], (4) be commit ted to the way of life in [country], (5) have good educational qualifications, and (6) have work skills that [country] needs. Responses were coded on a 10-point Likert scale where '0' is "extremely unimportant" and' 10' is "extreme ly important." I refer to the six symbolic bound in these questions as race, aries measured religion, language, culture, education, and occu
FUZZY-SET METHODOLOGY
Typologies are analytical tools used to com pare cases?in this case, countries?with ideal types not observed empirically. In practice, however,many of thequantitativemethods used to construct typologies create mutually exclu sive groups, ignoring the likelihood that many countries are in fact combinations ofmultiple types (Ragin 2000). Traditional or "crisp" clus between" types by forcing them into the clus ters they most closely resemble. Although this is inconsequential for studieswith large sample sizes, failure to identify such cases among the 21 countries in this study risks misidentification and misinterpretation of typical symbolic boundary configurations. I address these issues below through a combination of "fuzzy-set" techniques. I use "fuzzy cluster analysis" (FCA) (Dimitriadou et al. 2006) to produce a typolo gy of symbolic boundary configurations and explore itsrelationship to the country-level fac
tors above using measures of "fuzzy consisten ter analysis, for example, ignores countries "in
The "hypothetical immigrant" module of the ESS marks a considerable improvement over previous cross-national surveys of attitudes toward immigrants in Europe. Whereas most previous surveys ask respondents to describe theirfeelings toward"immigrants,""immigrants from outside Europe," or "racial and ethnic minorities," theESS questions are designed to capture important variations within each of these broad categories. By disaggregating atti tudes toward immigrants intomultiple sym bolic boundaries that are compared in the boundary-work literature(see Table 5), theESS measures both the intensity and the form of attitudes toward immigrants. It is thusparticu well suited to the conf igurationalapproach larly to symbolic boundaries adopted here. As with all survey-based studies, however, it ispossible thatESS respondents produced socially desir able responses to the questions, rather than the intimate convictions thatmight be revealed through ethnography.This is particularly rele vant for thequestion on race because the liter ature demonstrates the widespread influence of antiracist discourse inWestern Europe. It remains to be determined, however, precisely how much antiracism has permeated each coun how try (Lentin 2004) and, more importantly, such developments have shaped the entire con figuration of symbolic boundaries in response. Table 6 presents descriptive characteristicsfor the six symbolic boundaries in each country. Language and culture are consistently among the most important symbolic boundaries.
pation.
cy" (Ragin 2006).4 Crisp cluster analysis requires thatcountries trast, FCA assigns countries "membership scores" thatdescribe how much they resemble multiple fuzzy clusters or "sets." This is accom plished by applying a "fuzzy modifier" to the traditional c-means clustering algorithm (Dimitriadou et al. 2006). Consider thematrix mean of 21 countries (i) and their Xy comprised scores for the six symbolic boundaries above
belong to one?and only one?group. In con
For a detailed
traditional Supplement.
49
Country
Race
Austria 2.04 2.26 Belgium Britain 2.39
Religion
3.27 2.71 3.26 3.69 3.57 3.89 3.20 2.49 5.87 4.69 3.47 4.44 2.01 2.67 3.39 4.79 3.83 3.45 3.91 2.32 2.92 3.52 2.01 5.87
Language
7.57 6.99 7.40 6.29 6.41 6.23 7.33 7.77 7.78 7.68 6.38 5.77 8.45 7.42 6.25 6.82 6.00 7.50 5.92 4.35 6.15 6.78 4.35 8.45
Culture
7.14 8.25 7.51 8.24 6.33 6.88 6.28 8.18 7.47 6.30 8.00 8.18 8.95 6.68 7.17 7.95 7.90 5.58 6.57 6.45 7.08 6.08 7.98 7.35 7.73 7.25 6.13 7.57 6.19 6.45 8.95
Education
6.92 6.67 6.25 6.09 6.87 6.29 7.42 6.39 6.91 6.34 6.38 6.77 7.07 8.22 7.79 8.13 6.83 6.12 6.82 6.52 5.73 6.19 6.67 6.05 5.10 5.89 6.92 6.38 7.47 7.21 6.32 6.67 6.09 4.84 4.48 5.98 6.74 4.48 4.84 7.79 8.22
Occupation
Czech
Republic
3.64 Denmark 1.84 2.81 Finland 2.34 France 1.52 Germany Greece 3.64 4.12 Hungary Ireland 2.40
Luxembourg Netherlands
2.55 Italy .93 1.90 2.27 Norway Poland 2.95 2.94 Portugal 2.95 Slovenia 2.94 Spain Sweden 1.31 Switzerland 1.55
Mean
(all countries)
Note:
(j).5 When applied tomatrix Xtj, the fuzzy c means algorithm produces a matrix of mem M for21 countries and k sets, and bership scores akx 6matrix of "set cenrroids"C that describes each fuzzy set. This is accomplished by mini mizing the objective function:
the configuration of symbolic boundaries in
is constrained such that each country'smem mem bership cannot be negative, and the total across all sets is normalized. The bership
0 to 1, and the sum of each country's member
= (M,C)X^fjdtj i J
is membership coefficient of coun where uij the try / in setj, and dtj is the Euclidean distance between observation / and center j. The researchermust specify a value greater than 1 fork and m, the fuzziness index.6The function
ship scores equals 1. Unlike factor scores, FCA membership scores describe relationships between cases, not vari
ables.
boundaries resembles the configuration of clus ter centroids in each set. By plottingmember butiontoassess bias from but did kurtosis, it notyield different results (available from the substantially
author). 6 Here a value of m = 51 performed fuzzy clustering of the entire distri I used the Xie-Beni index to determine applications. the appropriate number of sets (k). Different values of m and k produced very similar results (available from the author). analyses. I used the R software to conduct
bership in a given set axeprototypical, whereas thosewith weak membership in all sets are sim ply atypical. The characteristics of each set are identified through inspection of the cluster cen A troids. country'smembership scores describe how closely its configuration of symbolic
Therefore,
countries
with
strong mem
2 is chosen,
following Bezdek
all
50
AMERICAN SOCIOLOGICAL REVIEW As Ragin (2006) shows, the consistencywith which one set is a necessary condition foranoth er can be calculated as follows: Consistency (Y< X) =
multidimensional space, one can ship scores in assess how exclusive each set is. Moreover, one can easily identifycountries inbetween sets as well. The "crisp" clustering solution can be
data than crisp cluster analysis. Fuzzy sets cannot be analyzed in tandem with non-fuzzy variables unless the latterare transformed into fuzzy sets as well. I calibrate the country-level variables presented in the background sections into fuzzy sets using the method proposed by Ragin (forthcoming) and detailed in theOnline Supplement. Even after all variables are transformed into fuzzy sets, standard correlational techniques cannot be used to explore the relationship between fuzzy sets (Ragin 2006). This is because correlations describe the covariation of variables, whereas the fuzzy-set approach asks whether cases are which subsets of one another.Consider Figure 1, describes the relationship between two fuzzy member sets: X and Y. The figure shows that inX is almost always greater thanmem ship
deduced for reference simplyby grouping coun tries according to their strongest membership. The range of membership scores within each crisp set, as well as the amount of overlap between them, reveals the integrity the crisp of clustering solution. Inmost applications, there is considerable variation within and between crisp clusters. In most cases, FCA therefore provides more precise measures of structure in
^(miniY.Xd)/^)
This formulameasures not only thefrequency of X being greater than but also the Y, magni tude of thisdifference.Large inconsistencies are edged as such. Scores .80 and higher indicate increasingly consistent relationships. Below, I calculate the consistency of the country-level variables (from thebackground sections) with the sets produced by FCA to develop hypothe ses to explain the configuration of symbolic macro level. boundaries at the
penalized, but "near misses" are also acknowl
sidered error. In the fuzzy approach, however, a consistent subset relationship exists between X and Y; meaning thatmembership inX is Y7 almost always a necessary condition for The in the lower right-hand corner are sim points Y which membership in ply considered cases in
in Y. Traditional correlational bership would not reveal a signif techniques, however, icant correlation between the two because the points in the lower right-hand corner are con
For a discussion
of the distinction
between
nec
on cluster validity,
see Part I
of the Online
Supplement.
51
1.0-1
0.9
0.8
0.7
0.6 .g
0.5 -
g 0.4
0.3 #
0.2-
0.1-
0.0
-\-\-1-1-1-1-1-1-\-1-1
0.0
1.
0.1
0.2
0.3
0.4
0.5
0.6
in Set X
0.7
0.8
0.9
1.0
Membership
Figure Example of a Subset Relationship
Table
7.
Characteristics
of Three
Sets
of Symbolic
Boundary
Configurations
Race
Mean (All Countries) 2.44
Religion
3.52 Fuzzy
Language
6.78 Sets Cluster
Culture
7.57 6.19 6.74 Centroids
Education
Occupation
SetA SetB
Note: Cluster
+.43 -.41
SetC_-.75
+.42 -.57
themean
-.68 +.67
for all countries.
-.37 +.15
-.16 +.16
+.16 -.10
-.46_-M_Il57_-M_-.91
and C.9 The three ellipses depict the crisp clus tering solution for reference. Spain, Portugal, Italy,Finland, Poland, theCzech Republic, and Irelandmost closely resemble SetA (in order of the strength of theirmembership). Britain,
the France, Austria, Germany, Belgium, Netherlands, Luxembourg, and Slovenia most closely resemble Set B. Finally, Switzerland, Norway, Denmark, and Sweden most closely resemble Set C. While Spain, Britain, and Switzerland are nearly prototypical of Sets A,
respectively, countries closer to the ori
B, and C,
gin of the axes (e.g., Greece and Hungary) are Most countries scarcely related to the three sets.
52
iBB_-__9____-___r:"'
^ " ""
-jjJJJ^^IfBPP^ M__^_^__i_itt___.
__B_H______^__^_HP^
'
f^BBP^^
Figure Notes: 2. Fuzzy
jJl____________P^^* Scores
*
in Three
--.
Sets
Membership
scores describe how closely its configuration of symbolic boundaries resembles the A country's membership those with low member in a given set are prototypical; sets described inTable 7. Countries with high membership or "length" describes ship are simply atypical. Ellipses depict the "crisp" clustering solution: themajor diameter scores within each crisp cluster whereas theminor diameter or "width" describes overlap
(.51) and C (.29) and Slovenia has significant membership in Sets B (.50) andA (.25). In sum, FCA shows significant variation within and between sets thatwould not be recognized by crisp cluster analysis.
fall somewhere inbetween, and several are best described as in-between sets.The Netherlands, for example, has significant membership in both Set B (.56) and Set C (.28). Likewise, Ireland has significant membership in Sets A
B countries (Britain, France, Austria, Germany, Belgium, theNetherlands, Luxembourg, and Slovenia) are geographically continuous in the "core" ofWestern Europe?if one ignores the all but one of the Set C English Channel?and countries (Switzerland, Norway, Denmark, and The in Sweden) are part of Scandinavia. between countries roughly follow thispattern as well. Slovenia sits between the "core" coun triesof Set B and the "peripheral" countries of Set A. Ireland sits between Sets A and C, and theNetherlands sits between Sets B and C. Because a theoryof boundary-work does not yet exist, a variety of different hypotheses could be developed to further explain theFCA results. Below I develop hypotheses by exploring the consistency of Sets A, B, and C with the four country-level factors discussed in the back ground sections: (1) the sources and timing of
SYMBOLIC BOUNDARIES IN 21 EUROPEAN COUNTRIES immigration, (2) the size and origin of immi mar grantgroups and their position in the labor and civic inclusion policies, ket, (3) citizenship and (4) philosophies of integration. Consistency Calculations
53
tentwith countries that experienced high net migration between 1960 and 1970, 1980 and 1990, and 1990 to 2000. Set B is highly con sistentwith high net migration between 1960 and 1970 and consistentwith high netmigration between 1980 and 1990. Although consistency increases between Set A and net migration by decade, it never reaches the .80 benchmark.
Table 8 describes the consistency between Sets A, B, and C and thefour fuzzy sets thatdescribe increases inmigration by decade proportional to the total population. Set C is highly consis
Returning toTable 1, one finds that the Set A countries (Spain, Portugal, Italy, Finland, Poland, Czech Republic, Ireland,Greece, and Hungary) were all once countries of emigration and only recentlyexperienced large-scale immigration.In contrast, the countries inSets B and C all expe rienced sizeable postcolonial or guest worker World War II. migration in the decades after Given the timing ofmigration, it is not sur prising that Sets B and C are consistent with countries thathave large immigrant populations (proportional to the totalpopulation), while Set
A is not (see Table 8). Set B is consistent with countries thathave large immigrant populations from theMiddle East and North Africa, sub Saharan Africa, South and East Asia, and Oceania. Set C is consistentwith countries that have large immigrant populations from all regions except theCaribbean and Oceania. Set
Table 8. Consistency ofCountry-Level Sets with Sets A, B, and C Symbolic Boundary Configurations (N Timingof Immigration 1970 to 1980
= of Immigrant Population (N 20) .541 .891 .904** Immigrant Population .646 (non-EU-15) East & North Africa Africa .680 .684 .812* .946** .566 .906** South East Latin America Caribbean Immigrant Percent of Labor Tertiary Education15 Change Force3 .557 .872* Force (1998 to 2003)b .895* .641 .714 .704 North .500 Oceania .466 .592 .833* .863* America .500 Asia .563 Asia .526 .818* .825* .622 .678 .637 .558 .926** .927** .883* .929** .673 .756 .900** .880* 1990
_SetA_SetB_Set
21)
1960
to 1970 .590
.932**
.938**
.630
1980tol990 .587 to 2000.743
.719
.851* .766 .966**
.684
.942**
Characteristics Total
.754 .896*
Sub-Saharan
.906**
.756 .840*
.827* .742
Inclusion_.873*_^44_.754
54 A
AMERICAN SOCIOLOGICAL REVIEW Carrera 2006). Moreover, there isno discernable pattern across Sets A and C apart from the lack of philosophies of integration amongmost coun tries that resemble Set A.
abrupt increase in thepercentage of immigrants in the labor force between 1998 and 2003. Table 8 presents the consistency of setsA, B, and C and five fuzzy sets generated from and colleagues' Civic Geddes (2005) Citizenship Index. Because the index is not available for the Czech Republic, Hungary, Norway, Poland, Slovenia, or Switzerland (see Table 4), these results should be considered highly tentative. Set A is consistentwith coun tries thathave generous antidiscrimination,nat market inclusion policies. uralization, and labor Set B is only consistentwith countries thathave generous family reunion policies. Set C is con sistentwith countries thathave generous natu ralization policies and highly consistent with
that have generous long-term resi
is not consistent with countries that have large immigrant populations regardless of region. Sets B and C are consistentwith coun tries inwhich immigrants make up a consider able part of the labor market, while SetA is not. Sets A and C are consistent with countries in which a substantial part of the immigrantpop ulation holds tertiaryeducation. Only Set A is consistent with countries that experienced an
DISCUSSION
Set A: New the European Immigration Periphery Countries on
Countries most closely affiliated with Set A (Spain, Portugal, Italy,Finland, Poland, Czech Republic, Ireland,Greece, and Hungary) share the following characteristics: (1) They are locat ed on theperiphery of theEuropean Union. (2) They were all once sources of emigration and only recently began receiving considerable (3) immigrants immigration. Nevertheless, remain a small proportion of the overall popu lation. Therefore, (4) discourses about immi grant integrationare relatively unsophisticated
countries
in
compared with those in the old immigration Western and Northern Europe. countries of It remains to be determined why racial and religious symbolic boundaries are stronger than average in Set A (see Table 7). One hypothesis is that phenotype and religious dress provide visual cues about group membership that are particularly conspicuous in new immigration countries, precisely because
These cues may
of their homo
contact
Surprisingly, the results suggest that official philosophies of integration do not correspond to the configuration of symbolic boundaries deployed by the general public. France, Britain, and Germany's philosophies of integrationare more often than those compared in the literature of any other combination of countries (see Table 5). Figure 2, however, suggests the configura tion of symbolic boundaries used by thegener al public in these three countries is nearly identical.All three countries are closely affili atedwith Set B, which is characterized by strong linguistic and cultural symbolic boundaries and weak racial and religious boundaries. While this configuration mirrors the emphasis on assimilation in French republicanism^ it runs counter to the pluralist tenets of British multi cultural race relations. The results are even less compatible with Germany's historical emphasis on ancestry, despite a modest growth of assim ilationist rhetoric in government discourse described in recent studies (Brubaker 2001;
between groups, allowing racial and religious stereotypes to go unchallenged (Allport 1958). An abrupt increase in thevisibility of a minor itypopulation may also provoke perceptions of "group threat"(Blumer 1958), as majority group members come to realize thatcertain privileges and status are attached to their race or religion. These perceptions may be reinforced by eco nomic insecurity as well, and Set A is consis tent with countries thatexperienced an influxof immigrants into their labor markets in recent
geneity.
limit positive
years.10 The results might also suggest that antiracist discourse has not yet permeated the periphery of Europe as thoroughly as it has Europe's core.Note thatthesehypotheses are not
10 This may also explain why occupational sym are stronger than average in Set A. bolic boundaries See the Online sive application for a more comprehen Supplement the of "group threat" and "contact"
SYMBOLIC BOUNDARIES IN 21 EUROPEAN COUNTRIES mutually exclusive. For example, positive con tactmay limit perceptions of group threat or facilitate the diffusion of antiracist discourse. Nevertheless, additional research is needed to explore these hypotheses indifferentcombina tions and to provide alternative explanations for the strengthof racial and religious symbol ic boundaries in Set A countries. in
55
Regardless of the extentof second-generation disenchantment, the "integration question" is perceived as a common "social problem" across most Set B countries (e.g., Tissot 2007). With the exception of Slovenia, Set B countries are
Countries most closely affiliated with Set B (Britain,France,Austria, Germany, Belgium, the Netherlands, Luxembourg, and Slovenia) share thefollowing characteristics: (1) They are locat ed in the core ofWestern Europe (except Slovenia). (2) They received substantial immi World War I, either gration in the decades after from guestworker agreementswith countries on theEuropean periphery (including Turkey and North Africa) or from formercolonies. Because now constitutea sizeable of this, (3) immigrants of the overall population, and (4) pub portion lic discourse about immigration has evolved over decades and is therefore more sophisticat ed than thatof thenew immigration countries. The emphasis on language and culture in Set B countriesmay resultfrom natives' acceptance of the permanency of immigration.As second
long-standing participants indiscussions about theharmonization of immigrationpolicy at the EU level, and they are highly aware of each other's integration strategies (Carrera 2006).n This may have produced convergence in the configuration of symbolic boundaries deployed by the general public, because these discus sions are constrained within a universal human rights discourse that stigmatizes group-based exclusion but sanctions individual-level exclu sion based on language, culture, and human capital (Joppke 2005). Nevertheless, addition al research isneeded to explain why such supra national discourse appears to have permeated Set B countriesmore deeply than countries in Set A. Set C: Accommodating Isolationists
generations of nonwhite and non-Christian immigrants come of age, racial and religious distinctions may not only become less con While spicuous but also less politically tenable. public discourse necessarily shifts from the accommodation to the integrationof immigrant
populations, natives may become more con
Countries most closely affiliated with Set C (Switzerland, Norway, Denmark, and Sweden) share the following characteristics: (1) They are located in Scandinavia (except Switzerland). (2) They each received considerable migration afterWorld War II, although they had no colonies fromwhich to recruit. Because these countries initially had relatively small popula tions, (3) immigrants from a variety of differ ent regions constitute a relatively large (4) theyare politically isolated from the core of Western Europe, and discourse about immi grant integration has evolved independently (Runbolm 1994;Wimmer 2002). As above, the weak racial and religious sym bolic boundaries characteristic of Set C coun triescould be explained as resultingfrom strong antiracist discourse, positive contact among natives and non-European immigrants, or the absence of competition between them. None of these hypotheses, however, explain why racial linguistic and cultural symbolic boundaries are farweaker in Set C countries than in all others. Here, deeper historical analysis may be war
proportion of the overall population. Finally,
well. ple,
forced by recent reports of "segmented" or "downward" assimilation of second-generation immigrants (e.g., Alba 2005; Crul and Vermeulen 2003; Silberman,Alba, and Fournier 2007), whose difficulty crossing social bound ariesmay inhibittheir"symbolic" integrationas
There about is growing for exam controversy, "reactive second among ethnicity"
cerned about the longevityof their linguisticand cultural identity. Or, natives may realize that language and culture guarantee theprivileges of were previously "protected" by group status that race or religion. These attitudesmay be rein
generationTurks in Germany (Diehl and Schnell North African counterparts in 2006) and their France (Beaud and Pialoux 2003).
1x
I am indebted
posing thishypothesis.
to an anonymous
56
AMERICAN SOCIOLOGICAL REVIEW contact between natives and immigrants or reduced perceptions of threat between them. Or the coincidence of these national self-under standingswith internationalantiracistdiscourse may have rendered ethnocentric attitudes less politically tenable than elsewhere. Again, much additional research is needed to explore these hypotheses more rigorously. Directions for Future Research
ranted. Tagil (1995), for instance, shows that intergroupdifferences were accommodated in the early history of Scandinavian countries to provide stability against the threatof cultural ly homogenous rivals. This was also true of Switzerland,where geistige Landesverteidigung (spiritual defense of the country) united dis parate linguistic and religious groups intoa sin more from its gle republic in the face of threats powerful neighbors.12 In contrast, linguistic and cultural differenti ation was central to the nation-building strate gies of many Set B countries. Consider, for example, theHerderian tradition inGermany or France's virulent (Calhoun 1993) which Britain repaid in kind "Anglophobia," (Greenfeld 1992). Likewise, race and religion were particularly importantdistinctions in the nation-building strategiesof several countries on theEuropean periphery thatdefined themselves against the threat of non-European and non
Gellner's theorypredicts that linguistic and cul tural boundaries become the primary mecha
nism of intergroup exclusion. While
Christian empires. This line of reasoning builds on Gellner's (1983) theory of nationalism, which suggests that different components of group identity become salient based on threatsfrom external groups.When "human chasms" such as race or religion do not separate insidersfrom outsiders,
it is not
colonies, Scandinavian countries, for example, condemned themfordoing so (Z0lner 2000) and a disproportionate number of welcomed from those colonies.13 In this way, refugees Gellner's theory is not incompatible with the otherhypotheses developed above. For instance, historical emphasis on accommodating dis parate groups may have encouraged positive
12 Note most
clear whether European publics today are aware that such processes might affect theirattitudes toward immigrants, the national self-under standings produced during nation building may be path dependent. As many Western European countries forced their language and culture on
lyze the content of these symbolic boundaries across different situations. For example, it is possible that linguistic and cultural symbolic boundaries are used publicly tomask private racism or Islamophobia. This study provides new theoretical and methodological tools for the studyof boundary work. I argue that treating immigrants as a sin gle out-group neglects importantcross-national variation in the conceptual distinctions used by natives to create notions of "us" and "them."The typology of symbolic boundaries presented above provides much needed context forprevi ous comparisons of two or threeWestern new variations European countries and identifies inother regions of Europe. This discussion also identifies new historical, demographic, and socioeconomic variables thatappear tobe more
To be sure, the configuration of symbolic boundaries isbut one ofmany factors thatshape the integration of immigrants into host soci eties.While the socioeconomic and legal seg regation of immigrants cannot be ignored, neither can the role of symbolic boundaries in creating and maintaining social boundaries. Due to the cross-sectional nature of the data, it is not possible to establish whether the config urations of symbolic boundaries revealed above are theproduct or the source of social inequal While it is likely that causality works in ity. both directions, longitudinal data and historical case studies are needed to explore my hypothe ses in greater detail and in different combina tions. In addition, ethnography and in-depth qualitative research are needed to furtherana
est affiliationiswith Set C. 13 Part II of theOnline Supplement shows that with largerefugeepopu Set C is highly consistent
lations.
promising in explaining the logic of boundary of integration work than the philosophies in the literature.Together, these emphasized contributions constitute a preliminary step must be toward a theoryof boundary-work that more rigorously by future studies in explored Europe and beyond.
SYMBOLIC BOUNDARIES IN 21 EUROPEAN COUNTRIES Christopher A. Bail is a PhD Candidate in the and Department ofSociology atHarvard University a Doctoral Fellow in the MultidisciplinaryProgram-. on Inequality and Social Policy at the KennedySchool
of Government. His other research explores the con and collective nection between symbolic boundaries the global diffusion of culture, and thepol violence, itics ofantiracism International in cross-national and theMinda perspective.
57
New Europe. and the National Question in the UK: CambridgeUniversityPress. Cambridge,
2001. "The Return of Assimilation?
He
in a Transylvanian Town. Ethnicity Press. NJ: Princeton University Princeton, 2006. Making Multiculturalism: Bryson, Bethany. Boundaries Departments. Press. and Meaning Stanford, CA: in U.S. English Stanford University and Ethnicity."
REFERENCES
Abbott, Social Andrew. Research 1995. "Things 62:857-82. Assimilation of Boundaries."
Calhoun, Carrera,
Craig.
1993. "Nationalism
Challenge
Programme.
Papers,
Sixth EU Framework
Reading, Joanna.
in in Italy and Spain." 138-57 Pp. edit and theNew Immigrations, Europe
Crul,
in Maria Ioannis, ed. 1997. Immigration Baganha, Southern Europe. Oeiras, Celta Editora Portugal: Lda. Barth, Fredrik. 1969. Ethnic Groups and Boundaries: The Social Organization of Cultural Difference. London, UK: George Allen and Unwin. Baubock, Blurred Rainer and John Rundell, eds. 1998. and Boundaries: Ethnicity, Migration, CT: Ashgate. Brookfield,
Review Migration Claudia and Rainer Ethnicity' or and Arguments, Labor Migrants
Citizenship. Pialoux. 2003. Violences Beaud, Stephane and Michel Violence Sociale. Paris: Fayard. Urbaines, J. C. and N. R. Pal. 1995. "On Cluster Bezdek, for the Fuzzy C-means IEEE Model." Validity on Fuzzy Transactions Systems 3:370-79. Erik. 2003. Race in Britain Politics and Bleich, France. Press. 1958. "Race Blumer, Herbert. of Group Position." Pacific 1:3-7. and Gilles Verbunt. Bovenkerk, Frank, Robert Miles, 1990. "Racism, Migration and the State in Western A Case for Comparative Europe: Analysis." International 5:475-90. Sociology Grete. 1996. European Brochmann, Integration and Third Countries. Oslo: Immigration from Scandinavian Press. University 1992. Citizenship and Nationhood Brubaker, Rogers. inFrance and Germany. Cambridge, MA: Harvard -. Press. University 1996. Nationalism Reframed: Nationhood Prejudice Sociological as a Sense Review Cambridge, UK: Cambridge University
Review Migration Dimitriadou, Evgenia, Kurt Hornik, Friedrich Leisch, David Meyer, and Andreas 2006. Weingessel. "Miscellaneous Functions of the Department
of
1071 .html). r-project.org/src/contrib/Descriptions/e 1966. Purity and Danger: An Analysis Douglas, Mary. and Taboo. London, of Pollution of the Concepts UK: Routledge. 2003. "Culture Eliasoph, Nina and Paul Lichterman. in Interaction." American Journal of Sociology 108:735-94. Parliament. 2005. European for Integration: Framework Agenda for the Integration of in the European Union." Communication and the "A Common
Statisticsfor the Software (el071), TU Wien." R Retrieved September 24, 2007 (http://cran.
Committee
of the Regions. Adrian. 2001. Philosophies and S. the Idea 2nd ed. New 2000.
of Integration: inFrance of Citizenship York: Palgrave. Attitudes States, toward and France,
Public
Immigration
in the United
58
Cambridge,
Cambridge
University
1992. Money, Morals, & Manners: the French and American Upper of
Steve.
Politics
of Belonging: Andrew,
Favell, eds. 1999. The in and Minorities Migrants Balch, CT: Ashgate. Claire European Brussels,
Brookfield, Europe. Jan Niessen, Alex Jose Peiro. and Inclusion 1983. Nations Blackwell.
and the Boundaries and Class, of Race, Sage Foundation. Immigration. New York: Russell "The 2002. and Virag Molnar. Lamont, Michele in the Social Sciences." Study of Boundaries
2005. Index.
et
les
and Nationalism.
face 2000.
to Greece: Pp. in
Goldberg,
David
Theo.
Ethnic
2006.
and Racial Five
"Racial
Studies Roads
1992. Nationalism:
edited by R. King, G. Lazaridis, Europe, and C. Tsardanidis. London, UK: MacMillan. in 2004. Racism and Anti-Racism Lentin, Alana. Europe. Michigan Medrano, Ann Arbor, 2005. in MI: The University Citizenship of and Press. Juan Diez. "Nation,
Ireland,PatrickR. 1994. The Policy Challenge of Politics inFrance Ethnic Diversity: Immigrant
and Switzerland. Cambridge, 2006. MA: Harvard Press. University Janmaat, Jan Germen.
Spain." Societies of
Old andNew EuropeanMember ofNationhood in States: Partial Support for the Ethnic-Civic
Ethnic and Racial Studies 29:50-78. 1996. Social 2005. Identity. London, UK:
"Popular
Conceptions
Cristobal.
Dimensions 41-65
"Exclusion
in the Liberal
2005. "Spanish and Alejandro Quiroga. Diego 5:9-29. Ethnic or Civic?" Ethnicities Nationalism: Ewa. 2006. "Identity and of Transnational Studies Socio-cultural People 29:1072-86. in
Identities: States Riva. 2002. Negotiating Kastoryano, and Immigrants inFrance and Germany. Princeton, NJ: Princeton -. France 2004. Press. University and Incorporation: "Religion International Islam in
Migration -.
A Regional in Hungary: Center for Policy Studies, Comparison. Budapest: Central European University. 2005. Subalterns? Transnationals "Global Parallel Studies Modernisers or Local Perceptions inHungary." Journal 31:659-74. of Chinese of Ethnic and
and Vaggelis Stathis Balias, Pandelis, 2003. "Greek Policy towards Immigration Immigrants." Social Policy & Administration
37:148-64.
Phalet,
Migration Karen
Minorities
and Antal Orkeny, eds. 2001. Ethnic in Context. and Inter-Ethnic Relations
Utrecht:ERCOMER.
to as a Response 1995. "Prejudice Lincoln. Quillian, Perceived Group Threat: Population Composition in and Racial and Anti-Immigrant Prejudice Europe." 60:586-611. Ragin, Charles American Sociological Review
Kymlicka, Will.
Nationalism."
1999. "Misunderstanding
Pp. 131-40 in Theorizing
-.
Chicago,
Science. Social C. 2000. Fuzzy-Set Press. of Chicago IL: University in Social Research: "Set Relations 2006.
59
-.
"Calibration Versus Forthcoming. Measurement." Oxford Handbooks Methodology: edited by D. Collier, H. Brady, of Political Science, and J. M. University Rath, Box-Steffensmeier. Press. Penninx, Kees Groenendijk, and Oxford, UK: Oxford
Jan, Rinus
Anna. and 2001. Triandafyllidou, Immigrants National Identity inEurope. New York: Routledge. 1993. "Rights and Racism in a New Vasta, Ellie. of Immigration: The Country 83-98 inRacism and Migration Berg. Italian Case." Pp. inWestern Europe,
and Xenophobia:
Weil,
2002.
Qu'est-ce
Frangais?
Weldon,
Faist. 2003. Identity and Integration: Migrants in Western Europe. Aldershot, UK: Ashgate. Richard and Irene Silberman, Roxanne, Alba, Fournier. 2007. the Second 30:1-27. M., and Thomas and Politics Paul de Peri, Rui Pierangelo Piazza. 2000. The Outsider: in Italy. Princeton, NJ: -.
Multilevel
"Segmented
Assimilation
in
Paris: Andreas.
Chicago, IL: UniversityofChicago Press. Stolcke, Verena. 1995. "Talking Culture: New
New Rhetorics Boundaries, Current Anthropology Europe." Laszlo. 1992. of Exclusion 36:1-24.
Migrants
in
Szoke,
and Immigration Emigration Architecture." International 26:305-23. Tagil, Sven. theNordic University Taguieff, PA. Decouverte. Tajfel, Henri.
1995. Ethnicity and Nation Building in World. Carbondale, IL: Southern Illinois Press. 1991. Face 1981. Human Studies in Social au Racisme. Paris: and La
the Integration of Immigrants." on Perspectives Politics and Society 4:243-64. European Aristide R. and Woon Litt Long. 1999. Zolberg, in Europe and the United States." Incorporation Politics and Society 27:5-38. theNation: Debates Z0lner, Mette. 2000. Re-Imaging on Brussels: Immigrants, Identities and Memories. Peter Lang.
"Why
Islam
is Like
Spanish:
Cultural
Groups
Social New
Categories:
Psychology.