DO MASS MEDIA PORTRAY EUROPE AS A COMMUNITY? GERMAN AND FRENCH DEBATES ON EU ENLARGEMENT AND A COMMON CONSTITUTION SILKE ADAM rN Abstract The goal of this article is to analyse whether, and when, the media depict Europe as a common community and whether, and when, they remain nationally confined. Media portray Europe as a community if they (1) make European and member state actors and topics visible, (2) show the mutual connectedness between these actors and (3) give voice to support and criticism while avoiding that nation states wall themselves off. To measure whether mass media depict Europe as a common community and thus potentially foster citizens' integration a content analysis of interactions is systematically combined with empirical network analysis. A comparison of the debates on EU enlargement and a common constitution in the German and French quality press reveals that whether Europe is portrayed as a community varies between and within countries. Such variations seem to reflect the relation between national elites' attitudes and public opinion towards a specific issue. Silke Adam is senior researcher at the Department of Communication Studies, University of Hohenheim; e-mail: adamsilk@uni-hohenheim.de. si 00 o o rN m £ £ 3 a 0 £ * 1 * (R 0 C K Introduction European integration is characterised by the construction of common political and economic structures. These common structures have created a system of governance in Europe that is unique in the world as no other region shows a comparatively high level of state surpassing regulations and international organisations (Kohler-Koch 2000, 11). From the beginnings of the European integration project, the creation of common structures and institutions was pushed forward by political elites of the nation states whereas citizens have, in general, remained more hesitant. Today, there is a considerable gap between the attitudes of the elites towards Europe and the rest of the population in nearly all member states (Mittag and Wessels 2003, 418). Without its citizens however, any future projects for European Union (EU) integration that go beyond a free trade union, possibly dealing with problems that cause dissent, may fail. Three types of attitudes shape the relation between the political system and its citizens (Easton 1965, 172): these are attitudes towards the authorities who are responsible for daily politics, attitudes towards the regime which equals the constitutional order and attitudes towards the political community. The latter will be the focus of this paper. In order to ensure the development of a community formal steps of integration are not sufficient in themselves. The concept also requires that citizens identify with each other (Easton 1965, 185, 325). Such a "we-feeling" seems demanding in Europe as here the nation states have dominated the concept of political community. Following Fuchs (Fuchs 1999, 156) such a community is usually constituted by two mechanisms: by a border line that defines those inside and outside of the community and by internal connections among its members. In small communities day-to-day interactions create border lines and internal connections. States with millions of inhabitants however, call for a different form of community: an imaginary one that is created in the heads of its citizens (Anderson 2003, Fuchs 1999).1 Consequently, formal citizenship as granted in Europe by the Maastricht treaty in 1992, contributes to the development of a European community only insofar as it is subjectively reproduced. Direct experiences with Europe like exchange programs or the common currency £ may foster the development of a community in citizens' heads. If however direct experiences are rare or completely lacking, mass media are regarded as powerful mechanisms that impact on these subjective images (e.g. Zaller and Feldman 1992, 611). The distance of EU politics from a citizens' everyday life thus makes media influence on the forming of a political community in Europe likely. If mass media have any influence as to whether people think of Europe as a community, then it is the national media who are responsible. The lack of supranational media outlets in Europe, audiences that are held captive in their native languages and yj politicians who need to cater to national audiences to be re-elected point to the fact that national media may or may not contribute to shaping a European community in citizens' minds. Studies point to the fact that national mass media actually influence public perception of Europe (see e.g. de Vreese 2004, de Vreese and Boomgaarden 2006). However, this impact cannot be described on the basis of a simple stimulus-response model. The studies already show that the impact differs according to the type of ^ communication flow (one-sided or two-sided) and according to the sophistication 0 £ C > of the recipients. In addition, the studies point out that mass media affect the evaluation of specific policies, but not the evaluation of the overall integration project. rO None of these studies deals with the question of the development of a European ^ community in the heads of the citizens. Before one can analyse how mass media might impact such "we-feelings," we need to step back and answer the question, which type of media content has the potential to impact citizen's attitudes of a community and how one might measure such media content. Consequently, this paper does not offer a study of media impact on EU attitudes of citizens. What it does is (1) to develop criteria which allow analysing mass media content regarding its integration potential, (2) to show how content and network analysis helps measuring these criteria and (3) to apply these considerations to the German and French debates on EU enlargement and a common constitution by quantitatively analysing reporting and commentating of quality newspapers in a three year period (2000 to 2002). The paper closes with a conclusion and a prospect on future research paths. Theoretical Considerations: Mass Media's Potential for Integration Mass media today are regarded as central mechanisms for a society's integration. The identification with a specific political space is not a precondition for politics, but is itself permanently created in communication processes (Eder 2001, 238, Habermas 2001, 102). So far empirical evidence is lacking as to whether the media fulfil this ascribed function (Vlasic and Brosius 2002, 93). Additionally, researchers still struggle with the question as to how the media might contribute to integration. Integration seems possible in two ways (Pöttker 2002, Weßler 2002): on the one hand mass media may integrate by unification or homogenisation (e.g. Schulz 1997). This is the most common perspective in communication research (e.g. Pöttker 2002, 325). Integration in this perspective requires that the media offer standardized programs and outlets for all groups in society, that these groups use the media similarly and that the media's content is uniform and consensus-oriented. Shared agendas in different media thus would be an indicator for mass media's integration potential. On the other hand mass media may integrate through plurality. From this perspective integration is made possible not by homogenisation, but by connecting heterogeneous parts. The media fulfil this function if they show the mutual interconnectedness of a community (Durkheim 1893/1977, 403). Integration from this perspective takes place through conflict communication (Wessler 2002, 72). It requires that the media allow different groups in society mutual observation. It does not demand however, that all citizens consume the media in a similar way so long as this does not lead to a permanent state of ignorance towards existing conflicts. The media content in this model needs to formulate conflicts without debasement of the opposition. Integration through plurality means that societies are characterised by a permanent struggle within their democratic public spaces (Dubiel 1999, 142). Following the idea that a homogenisation of media outlets, usage and content contributes to integration, one can derive criteria that are necessary for public communication in Europe. Researchers that follow this integration model claim that there must either be a supranational, European-wide media system that reaches out to the masses or - as an alternative - a multitude of national media systems that converge while transporting European issues into their national political spaces. This means that these national media report and comment upon the same issue at the same time under similar points of reference (Eder and Kantner 2000, 306). This model of integration through homogenisation has been criticised for ignoring societies' plurality (Wessler 2002). Their functional and social differentiation makes integration by homogenisation, even in traditional nation states, difficult (Pöttker 2002, 331). In the case of the European Union this aspect becomes even more relevant. Research has shown that the common European input that relies on national media to be transported to the audiences is filtered at the national level creating debates that differ between countries (e.g. Adam 2007 a, b). The plurality of national contexts within the European Union leads to different domestic adaptation processes by national actors and, therefore, to different debates. It is likely that this domestication of European politics with national colours will continue as long as national politicians need to cater to national electorates for winning elections and so long as national media depend on the attention of national audiences in order to stay in business. Additionally, integration through homogenisation faces a second problem in the case of the European Union. Even if common European regulations and policies cause parallel debates in different countries one cannot be sure whether the media will show the European-level origin of these policies. If, however, citizens do not become aware of the European dimension of a topic one must doubt whether such a synchronous debate throughout Europe helps in integrating Europe (Koopmans and Erbe 2004, 100f.). Without direct references to the EU itself or to other European countries any such debates are unlikely to create an imaginary community of Europe in the minds of its citizens. To conclude at this stage that the mass media have no potential for integrating Europe through homogenisation would be misleading. However, the plurality and differentiation in Europe makes it likely that if the mass media contribute to integrating Europe at all, they cannot solely rely on mechanisms of homogenisation, but need also to connect Europe through plurality. Researchers in the tradition of this second integration model claim that traditional national public spaces must open up for European and member state actors and issues in order to allow observation. As a result public arenas get linked to each other forming a network of European communication (Habermas 2001, Koopmans and Erbe 2004). The visibility of European actors and issues in the national mass media indicates a debates' integration potential on a first level. This criteria is supported by empirical research that shows that the higher the salience of European issues in people's daily lives, the more the people tend to identify with Europe (Risse 2004). From an integration perspective ^ however, it seems insufficient if traditional national public arenas simply open up their communicative space for European actors and issues. The pure visibility of European and member states actors and issues is a necessary, but not sufficient condition for integrative debates. Following Durkheim (Durkheim 1893/1977, 403) integration requires the experience of mutual dependencies. Fuchs (Fuchs 1999, 156) calls it the connection among the members of a community and the border lines for those staying outside. The mass media show such mutual connections within 0 £ C > Europe if the national and European member-state actors that have become visible are related to each other. Communicative interactions that are depicted in the mass media thus are not limited to traditional nation states anymore, but transcend them -šT cr> within Europe. This interconnectedness is regarded as a second indicator for the integration potential of debates. Debates hamper the development of a European LO community if national and European member state actors are not related to each ^ other within these debates (lack of connection among the members) or if communicative interactions reach beyond Europe (border line definition). But even if debates show the mutual interdependency within the European political space they do not necessarily foster integration. Integrative debates need to depict a dispute constellation (third criterion) that shows conflict as well as cooperation between the involved actors and prevents, at the same time, nation states from walling themselves off from the rest of Europe. A display of positive references is necessary as otherwise citizens lack possibilities for identification and negative references give existing conflicts within society a voice. If debates can avoid the walling-off of the nation states from each other, then purely national interest-definitions (Gerhards 1993, 106ff.) become an exception. To integrate around 450 million Europeans in 27 countries with 23 different languages into a common community is a difficult task. If the mass media contributes to such task, it seems likely that they do so not only through homogenisation but also by connecting Europe through its plurality. Integration through plurality is demanding. It requires that debates in different European countries (1) make topics and actors in Europe visible, that (2) they show the interdependencies between these actors (interconnectedness) and (3) that they formulate positive and negative aspects of European integration without fencing off traditional entities (dispute constellation). Methodical Considerations: Measuring Debates' Integration Potential Only if we can measure criteria empirically, are they useful for research (Vlasic and Brosius 2002, 99). Such measurement allows not only the description of the integration potential of mass media debates, but also marks the starting point for later studies that need to test the impact of the analysed mass media depictions on citizens' attitudes. To measure whether mass media depict Europe as a community in plurality, it is not sufficient to show the pure visibility of European / member state actors and issues on the media agenda, but it does require us to take a closer look at the actor constellations, to see whether and how these actors are interlinked. One perspective that focuses on the linkages and interactions is the network concept. Networks are characterised by actors, positions or organisations that are connected by specific types of ties (Pappi 1987, 13). Debates in this perspective are constituted by speakers and their communicative interactions with other actors. Such speakers - also called frame-sponsors (Gamson and Modigliani 1989, 6) - are either the media themselves, if they raise their own voices, or political actors that make use of the media as a forum to express their ideas. These speakers attribute responsibility, support or critique to other actors and thus organise discourses by making other actors important or ignoring them and by employing frames of inclusion and exclusion. As those actors to whom responsibility, support or critique is attributed can become speakers themselves, complete networks arise that can be termed "symbolic." They differ from social or policy networks because the analysed interactions do not physically take place, but are presented within the media. These networks construct reality as they reflect the media's depiction of political struggles, not the struggle itself. They therefore represent the "symbolic dimension" of politics (e.g. Edelman 1964). To study media debates from a network perspective a specific, quantitative content analysis needs to be applied that focuses on interactions between speakers and addressees. As a first step all contributions within the media under study are selected that refer to the analysed topic. Interaction units (Früh 2001, 242) are coded within these contributions. These interaction units are defined by Koopmans and Statham (Koopmans and Statham 1999) as "claims." A claim is a strategic action in which a speaker publicly expresses his opinion by attributing responsibility and offering support and critique to other actors. Claims are not confined to verbal actions only but also refer to physical ones. An interaction- and actor-centred concept of debates (symbolic networks) is analysed with the help of an actor- and interaction-centred method. This results in data that reveal not only the actors' visibility, but also their constellations. As it is hardly possible to study such relational data with classical statistical instruments, a specific tool needs to be applied: empirical network analysis (Wasserman and Faust 1999). This method of analysis allows the examination of all three indicators for integrative debates (visibility, interconnect-edness and dispute constellations) as it does not regard texts as a treasure trove of isolated actors, opinions and topics, but as symbolic constructions that reveal how conflict, cooperation and targeting connects the involved actors. The method of empirical network analysis provides us with mathematically founded definitions and indicators that reveal the position of actors within a network and the structure of ties that connects them. Which network indicators can be used to study the visibility, the type of dependencies presented and the dispute constellations within debates will be shown in the following. To determine whether debates allow the observation of different actors and their opinions throughout Europe, we draw on the network analytic form of visibility. Visible actors are prominent in a debate which means that they attract attention. y From a network perspective there are two possibilities to do so: firstly, actors can become visible as speakers who attribute responsibility, support and critique to others and thus serve as agenda-setters directing attention to their positions and interest. For this role actors need to actively access the media. Secondly, actors gain prominence if other actors direct their claims towards them. It can be assumed that this form of visibility is less dependent on the selection criteria of the media but more strongly reflects the importance of an actor in the political process. An actor's total prominence sums up its prominence as agenda-setter and attention receiver. Within a network a prominent speaker is thus the source or the object of many communicative relations. This also means that if a speaker attributes responsibility to two different addressees within one claim its prominence doubles compared to a visibility approach without a relational perspective. The same is true for attention mmixnnrr Pr Am t n nr\ pa muc t~\ nnc r\nf mAan mim AVtcfAnrA ir» nAnatAC mif mAonc C > receiving. Prominence thus does not mean pure existence in debates, but means prominence as agenda-setter and attention-receiver. In network analytic terms the former idea is called "outdegree," the latter "indegree" (Freeman 1979). The sum of both indicators results in the total prominence of an actor which is called "degree." Such "degree" analysis allows us to determine whether the positions and ideas of European and member state actors can be observed in debates. Only il cr> if national actors have to share prominence with those from other parts of Europe is the first criteria for integrative debates fulfilled. However, the high prominence values of European / member state compared to national actors in debates tell only ^ half of the story as far as visibility is concerned. Visibility is a two-dimensional concept: the first dimension refers to the actors' visibility; the second one takes the visibility of a topic on the overall agenda into account. The prerequisite for a debate's potential to integrate is its existence. If a topic is not gaining attention at all it is hidden from public view and thus hinders integration. In general one can distinguish between three types of debates: an issue can be a permanent one, an event-jumper or a low salience-issue. A permanent issue is debated over a longer period of time above a specific threshold of attention; an event-jumper gets on the agenda outside routine politics; a low salience issue remains below a threshold that reaches people's attention. Visibility, however, is a necessary, but not a sufficient condition to attribute integrative potential to a debate. On a second level it is required that debates show the interconnectedness of Europe. For such an analysis we must compare how many interactions solely run between national actors, how many involve non-European actors (globalised interactions) and how many connect European actors. Even if European actors are involved a debate's integration potential can vary. If, for example, debates portray French actors in discussion with other French actors and those from Poland attributing responsibilities to other Polish actors, then communication does not cross borders. Such a construction of Europe that resembles traditional reporting from foreign countries (Koopmans and Erbe 2004) would be assumed to hinder the development of a common European community that transcends traditional border lines. Also a portrayal of Europe that relied solely on discussions within the EU institutions might be a hurdle for integration as this type of debate does not reflect the today's nature of the multi-level system of EU governance. Integration by communication requires a third criterion: a dispute constellation that shows conflict and cooperation while at the same time avoiding the wallingoff of traditional entities. Integration through conflict does not require that critical voices are oppressed. By contrast, it requires that conflict and positive references be formulated. To determine whether conflict and cooperation is present in a debate, we calculate a "PN-Index." This index subtracts all negative references within a debate from all positive ones and divides the result by all evaluated communicative references. For each debate this index varies between -1 and +1. A negative value indicates that conflict prevails; a positive value that supportive references are dominant. Integration through conflict communication becomes likely if the index lies close to 0. This indicates that a debate offers citizens positive points of reference while at the same time taking account of their concerns. Last but not least, Europe is only portrayed as a community if interests - expressed by support and critique attributions - are not bound to traditional entities, but transcend them. Only if speakers from different European countries form common coalitions do the media portray a communicative space which is not dominated any more by specifically German, French or Polish interests. Such debates can be expected to foster the development of an imagined European community as traditional nation states do not wall themselves off. To identify coalitions in public debates we follow the idea of Hajer (Hajer 1995) who claims that actors within a discourse coalition make the same story-line prominent. From a network analytic perspective this means that actors form a coalition if they attribute support and critique to the same actors and thus define their interest similarly. To study such coalition structures one can use the concept of structural equivalence (Wasserman and Faust 1999, 347ff.): two actors are structurally equivalent if they attribute support and critique to exactly the same actors. As it is unlikely to find actors with identical reference patterns, structural equivalence analysis seeks to put actors into coalitions that have similar references of support and critique. This kind of analysis is based on two matrices in each debate that are characterised by the involved actors2 and respectively the support or critique relations between them. Actors are analysed in a detailed perspective: they are not only differentiated according to their origin, but also according to their social position (government, etc.) - and in the country of analysis also regarding their political orientation. Such a detailed level of analysis is necessary to determine whether traditional entities shape interest definitions or not. To evaluate how similar speakers are in their support and critique attributions, we calculate for each pair of actors the degree of correlation between their reference patterns.3 High correlation values indicate that speakers make the same story-line prominent whereas low values point to different patterns of support and critique attribution. The similarity between the reference patterns of each pair of actors is summarised in a similarity matrix which shows all Pearson correlations. This similarity matrix is then subject to a hierarchical cluster analysis based on average linkage procedure and to a multidimensional scaling analysis (MDS). Both methods help to identify coalitions as they group actors together which are similar to each other. They differ as cluster analysis is based on a discrete model out of which strictly separable coalitions arise. MDS in contrast works on a continuous model that seeks to place similar actors close to each other in space.4 Comparisons of these two methods show that their results are similar (Wasserman and Faust 1999, 388) but supplementary (for a detailed description to determine coalitions see Adam 2007 a). The Integration Potential of German and French ^ Debates on EU Enlargement and a Common Constitution 0 Applying the identified criteria and methodology to two central European topics - enlargement and constitution building - in two countries that are often regarded as the engines of EU integration, France and Germany, reveals each debates' potential for integration. The analysis is based on content - more precisely claims - analysis data from a seven-country study in Europe called "Europub."5To yj examine the debates on EU enlargement and a common constitution in France and Germany, all the claims in the political and economy section of a conservative and left-liberal quality newspaper that deal with these two issues were coded in each country for the years 2000, 2001 and 2002. As conservative newspapers the Frankfurter Allgemeine Zeitung (FAZ) and Le Figaro, as left-liberal newspapers the Süddeutsche Zeitung (SZ) and Le Monde were chosen. Within the respective period the enlargement as well as the constitutional issues ^ were at the top of the EU policy agenda. At the end of the year 2002, after hard C > disputes, the European Union decided not only on the largest enlargement round in its history - more exactly on the accession of ten new member states - but also C^ on opening the doors for accession negotiations with Turkey. Probably the second ^ most important question was to define the goals and institutional structure of an enlarged Union. Between 2000 and 2002 the heads of state publicly addressed these questions and decided that a Constitutional Convention should start its work in the beginning of 2002. Consequently, all claims that deal with the relation between Europe and the citizens, with the structures and competences of political order in the EU or with core goals for the integration project per se are coded under the label "constitutional issue" (see for a definition European Council 2001). Enlargement claims refer to the question of who should belong, and under which criteria, to the Union and what would be the consequences thereof. As three years of detailed interaction analysis even goes beyond the capacity of such a large research project, this study is based on a sample. Each newspaper has been analysed once per week avoiding a selection of the left and conservative newspaper at the same day. The sample is thus constituted of 312 newspaper editions in each country.6 Such a data set permits the study of the debates' integration potential according to each outlet, according to specific time frames (e.g. summits versus routine politics), according to specific issue fields and according to country differences. Which focus to choose depends on the researcher's own interest. As the last chapter of this paper seeks to understand differences in the analysed debates in the light of their national and issue-specific contexts, this paper is not concerned with producing a data set that is independent of the possible idiosyncrasies of any single source (see for a similar approach Ferree et al 2002). Consequently, the two different newspapers in each country are not analysed separately regarding their news reporting. If, however they act as speakers themselves, then they are treated as such.7 Also differences in the three-year period have been largely neglected as it has not been analysed whether the integration potential changes in the course of time. Within the selected newspaper editions any articles are selected that fit the two topics under study. Each article then is checked for claims. The number of claims indicates the visibility of a topic.8 Within each claim up to three speakers can attribute responsibility, support or critique to up to three addresses.9 To study the visibility of the actors, their interconnectedness and the resulting dispute constellation each relation between a speaker and an addressee coded within such claims is analysed. The case numbers of the following network analysis indicate the exact number of these relations.10 The media's reporting and commentating were coded by native speakers in country teams carefully trained before coding and supervised throughout the whole coding period. The coder trainers of the country teams were constantly in contact to solve coding problems. Two separate reliability tests have been conducted for the project in general: one for the editorials and one for the media's reporting. For the reliability test on editorial coding, coders in each country team coded a random sample of seven commentaries from the Scotsmen, the Times and the Guardian of the year 2002. The inter-coder reliabilities were measured as the average match between the coders. The overall reliability calculated on the core variables of the analysis turned out to be highly satisfactory with an average match of 75%. The reliability of the coding of media reporting was tested on the basis of one issue of The Guardian. The average match between the coders on the core variables was 87%. As the detailed issue delimitation turned out to be problematic, the author checked on the basis of a string variable that captures the content of each coding unit whether it belonged to the two issue fields under study. The reliability for specific variables is indicated at the bottom of the respective tables or figures. The complete reliability tests can be obtained from the author. The proposed study design facilitates findings that at least partly fulfil the integration requirements as it focuses on issue fields with strong European competences and on countries that are regarded as engines of the integration process. Moreover, a very specific source of data has been used: quality newspapers. Quality newspapers are geared towards a public that shows higher interest in politics as the average citizen. They are better staffed with journalists and correspondents what results in a higher quality of reporting and commentating (e.g. Gerhards, Neidhardt et al 1998, 87ff.). As a consequence thereof, the results are not representative for the overall media system of the countries under study. One cannot conclude that if the quality press portrays Europe as a community, the rest of the media system does the same. For some aspects such a generalisation might be valid as Pfetsch et al (2007) show: press commentaries in seven European countries turned out be more strongly affected by a country's political setting than by media formats. For some aspects, however, a generalisation is not possible. Kevin (2003) shows clearly that quality newspapers give European issues and actors more prominence as television or the regional / tabloid newspapers. What the study design, however, allows is to generalise findings that indicate integration barriers. If the results show that even the quality press - whose recipients are at least partly those who have pushed forward EU integration - has problems in portraying Europe as a common community, it is unlikely that the rest of a country's media succeed in doing so. What the research design also allows is to search for factors external to the media organisations that might impact whether media portray Europe as a community or not. y Visibility. There is only a chance that debates inherit integration potential, if Q the actors and issues from the EU and other member states become visible. In the £ German and French debates on EU enlargement and a common constitution nonnational actors become visible (Table 1). Looking at the degree, which indicates an actor's total prominence as agenda-setter and attention receiver, it becomes clear that national actors are not dominating any of these debates. Their share of the total prominence lies between 38.4% in the German debate on a European constitution and 16.9% in the French enlargement debate. The remaining prominence is shared between European institutions, actors from member states and, in the enlargement debates, also with those from the upcoming member states. Non-European actors are marginal in the constitutional debates, but have some prominence in the enlargement debate as here possible upcoming member states like Turkey are relevant. A comparison between the actors' total prominence and their prominence as agenda-setters (outdegree) shows that national actors gain most of their prominence in the role of public speakers. In the German debate C > on the European constitution for example, the national actors' share of the total prominence amounts to 38.4% whereas their share of the prominence as speakers is above 50%. By contrast, transnational actors gain prominence as attention receivers. il O This comparison shows that national actors still have an advantage when seeking direct access to the media. From an integration perspective one can conclude that non-national actors and their positions become visible - more strongly, however, as objects of national attention than as shapers of the debate themselves. Table 1: Visibility of Actors (%) Constitution Ger Constitution F Enlargement Ger. Enlargement F Out-degree Degree Out-degree Degree Out-degree Degree Out-degree Degree EU 21,6 26,2 27,2 39,1 1 8,7 23,8 27,0 38,0 Member countries 26,6 32,1 24,1 24,3 10,8 16,1 22,7 19,5 Upom. member c. 1,0 1,2 2,6 1,9 19,8 20,4 19,0 16,9 National 50,3 38,4 45,2 33,3 42,0 27,6 23,0 16,9 Other 0,6 1,9 0,9 1,3 8,7 12,0 8,4 8,7 % (total) 100 100 100 100 100 100 100 100 N (total) 696 569 1050 348 Basis: all communicative interactions. Remark: the "other"category contains all actors that are neither part of the EU nor involved into access negotiations at the time of analysis. Reliability match of news analysis: claimant 1/2/3 (scope) = 100%/99%/100%; addressee 1/2/3 (scope): 80%/90%/92%. Reliability match of commentary analysis: claimant is by definition the journalist writing the commentary; addressee 1/2/3 (scope): 83%/64%/65%. An analysis of the actors' visibility tells only part of the story. It does not show, however, how much importance an issue gets on the media agenda. Only if debates gain sufficient attention can they possibly portray a European community. The question therefore is: does reporting and commentating reach a threshold that secures people's awareness? Figure 1 shows the intensity with which EU enlargement and a common constitution have been debated in Germany and France over the course of time. This analysis is based on the number of claims coded. The constitutional issue can be regarded as an "event-jumper" in both countries as reporting and commentating follow the same events. The enlargement debate, however, differs in intensity between the two countries. A relatively high intensity of reporting and commentating in Germany is contrasted by a manifest ignorance of this issue in France until the end of the year 2002. Only shortly before the heads of state decided upon the largest enlargement round in European history ever did the French media pick up on the issue. Whether this already qualifies the French enlargement debate as a "low-salience" issue depends on the evaluation criteria. The analysis proposed here cannot judge an issue's visibility compared to other issues on the media agenda. It does allow, however, an evaluation as to whether the readers have had a chance to assert the importance of an issue. In this sense the French debate resembles a low-salience issue: an examination of at least 8 newspapers editions per month contains less than 10 claims on average. If one takes into account the fact that each article contains on average more than one claim, one can conclude that the enlargement issue was largely ignored in France. The lack of debate in France regarding one of the most crucial decisions of the European Union can be judged as a primary obstacle for integrating Europe through media debates. Figure 1: Visibility of Issues U S i 0 £ * i * IR 0 C « fN Claims - I- Constitution Ger —Constitution F A // W ) 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i i • ! cS> ^ - A ^ ^ „O ^eJ^V" ^^^ Claims 0 0 0 4 0 0 0 0 <£» <£» tf' & T" Enlargement Ger — Enlargement F i i i i i i i i i i i i i ^^ ^r^iV^, ili Basis: al claims coded in Germany (FAZ, SZ) and France (Le Figaro, Le Monde). Reliability match of news analysis: news article selection 83%; claims selection within news articles 84%; broad issue field (issue 1/2/3): 96%; 98%, 97%; detailed issue field coding was done by the author on the basis of a string variable. Reliability match of commentary analysis: Selection of commentaries has not been tested as commentaries are defined in a strict sense as the opinion articles of a journalist or editor and appear every day in a specific layout. claims selection within commentaries 99%. (By definition each commentary contains at maximum one claim); broad issue field (issue1/2/3): 98%/74%/95%; detailed issue field coding was done by the author on the basis of a string variable. Interconnectedness. In the following we analyse whether these national and European actors are portrayed as interconnected and thus mutually dependent of each other or whether communicative interactions are confined to traditional border lines. Table 2 shows that in all four debates there are only few interactions that solely involve national actors. In the German debate on a common constitution the share of nationally confined disputes amounts to around 15%; in the French enlargement debate the share is less than 5%. Interactions that involve actors out- 0 side the European community are irrelevant in the constitutional debates, whereas they have some weight in the enlargement issue. This relatively high number stems CO from the fact that those states hoping to join the Union like Turkey or Croatia are 2 involved in the enlargement debates. Most of the interactions, however, run within the European Union, between national, European, member state or upcoming member state actors. A closer look at this European communicative space reveals that most of these interactions also cross border lines. Between 58 and 68% of all communicative interactions within the debates transcend traditional border lines revealing an interconnected European Union with strong dependencies. European topics are discussed here in a manner different from classical foreign news reporting that only shows disputes within a foreign country. These debates truly reflect the interdependencies of a common Union as they go beyond a portrayal of internal struggles within the core EU institutions and show the multi-level governance system. As far as interconnectedness is concerned the debates fulfil the integration requirements. Table 2: Interconnectedness of Debates Percent of Constitution Germany Constitution France Enlargement Germany Enlargement France Global actors involved 1,7 2,2 19,3 15,8 Solely national actors involved 15,1 11,4 11,1 4,6 European actors involved 83,2 86,4 69,6 79,6 X X X X Interactions bound to borders 17,4 17,8 11,4 14,0 Interactions bt EU institutions 10,9 14,4 3,0 7,8 Interactions within a country 6,5 3,4 8,4 6,2 Interactions transcending borders 65,8 68,6 58,2 65,6 N (Relations) 696 569 1050 348 Basis: all communicative interactions. Reliability match of news analysis: claimant 1/2/3 (scope) = 100%/99%/100%; addressee 1/2/3 (scope): 80%/90%/92%. Reliability match of commentary analysis: claimant is by definition the journalist writing the commentary; addressee 1/2/3 (scope): 83%/64%/65%. Dispute Constellation. On an evaluative dimension, mass media depict a common European community, if they are open for critique and support while at the same time avoiding that traditional entities fence themselves off. The analysed debates only partly fulfil these requirements. In the German debates on EU enlargement and a common constitution negative references towards actors are far more prominent than positive ones. The PN-Index for the former debate equals -0.27, the one in the latter debate -0.24. The lack of support relations could be an obstacle to integrating Europe as a common community is likely to face severe difficulties if conflict is the only driving force within it. In contrast to Germany, positive and negative references have an equal share in the French constitutional debate (PN = 0.00). In this debate conflictual and supportive views are given a voice. An in- teresting case is the French enlargement debate. Here, support dominates conflict (PN Index = +0.14). The media portray the topic in a consensual way. This media depiction is contrasted by the actual conflict constellation within the country: there is no other country in Europe in which the citizens refuse any enlargement of the Union by such a clear majority (European Commission 2000, 2001, 2002). Only about a quarter of the French support the accession of the ten mainly Eastern European countries while less even support full membership of Turkey. The French political elite - which means the complete elite in times when the prime minister and the president come from different parties - has, after unsuccessfully fighting enlargement, contributed to paving its way. The actual conflict within the country dividing the elite and the public is thus not depicted by the media. At least the quality press seems to conceal critical voices. This concealment of conflicts is likely to hinder integration as it leads to mute conflicts which have the potential to increase the gap between the citizens and Europe. The degree of conflict within a debate shows only a part of the dispute constellation. A more detailed analysis of the evaluations within a debate reveals which actors form common coalitions as they attribute support and critique to similar addressees. This last analysis reveals whether Europe is portrayed as a community or not. Only if speakers from different European countries form common coalitions do debates carry the potential to foster an imagined European community. To identify coalition structures in debates, actors with similar support and critique attributions are placed in the same coalition as they are regarded as spokespeople for the same ideas. Figure 2 and 3 show the similarity of actors' reference patterns based on multi-dimensional scaling (MDS) and cluster analysis. MDS places actors with similar reference patterns close to each other in space. Those actors that are identified as similar by cluster analysis (the exact results can be obtained from the author) on a correlation level above .4 have been put into the same circles within this MDS space. In the German debate on a common constitution coalitions are predominantly y transnational. Traditional entities do not wall themselves off. The two German newspapers, the FAZ and the SZ, have similar reference patterns as the supranational EU institutions (EU Commission and the EU Parliament). This indicates that the conservative and left-liberal newspapers are not separated by a left-right divide, but resemble each other in their commentating on the constitutional issue. This coalition of supranational EU institutions and German newspapers is strongly tied to a second coalition that is composed of civil society actors from Germany, Ireland and the British government. The conservative opposition in Germany and other German media quoted in the news coverage form a coalition with French actors and other countries. The German government, together with left-leaning German politicians, constitute the only national coalition within the debate. However, this coalition has similar reference patterns to the alliance of the Belgian and Italian rr at 7Arr» iv» Ar» te i-n/^liir1tr>A- IWa kt-A-n/^l-i mAAii KACin AC Ann /lAilifiAn nArvi t-\ACAri nf A nc_ C > governments including the French media. Besides one coalition composed of Austrian actors only, the other alliances transcend border lines in their composition. In sum, coalitions in this debate do predominantly transcend border lines and thus do not seem to be an obstacle to integration. Turning to the French debate on a European constitution, the analysis reveals a stronger tendency for coalition formation bound to traditional border lines com- -šT O Figure 2: Discourse Coalitions in the Debates on Common Constitution Aust parl/par p Constitution Germany Actors' representation in space: MDS; circles: cluster analysis (cut-level .4); grey coloured circles: similarity between coalitions; Program: UCINET 6; Stress value: Ger = .33; F = .32 pared to the analysed German debate. Besides a coalition composed of Irish actors only, there are two coalitions that are mainly composed of French actors. The French EU-opponents form a coalition together with French civil society organisations, French financial experts and - surprisingly - the left liberal newspaper of Le Monde. As a transnational actor it includes only parliamentarians from Britain. This coalition shows clearly that Europe divides France not by a left-right split but creates its own cleavage (Mittag and Wessels 2003, 419). The conservatives in parliament Figure 3: Discourse Coalitions in the Debates on EU Enlargement Actors' representation in space: MDS; circles: cluster analysis (cut-level .4); grey coloured circles: similarity between coalitions; Program: UCINET 6; Stress value: Ger = .34; F = .32; basis: actors with degree >3 and outdegree >1. and the conservative newspaper Le Figaro form another French coalition in which the Constitutional Convention is integrated at a much lower correlation value. The remaining coalitions are transnational. There is, for example, a French-European alliance which closely resembles an Austrian, Polish, Irish and German coalition. In sum, there is a tendency among French actors to form their own coalitions. Singularly the alliance that is composed of EU-opponents and Le Monde with its 2 multitude of French actors involved is characterised by a clear separation between national and transnational actors and has no correspondence to other coalitions. However, it would be too much to claim that this debate is solely characterised by national coalition building as there are also national actors involved in transnational coalitions that transcend borders. The German debate on EU enlargement is dominated by national coalition formations. The analysis reveals three national coalitions in which transnational actors are poorly integrated. The first one is composed of the German government and left politicians. In contrast to the German debate on a common constitution, this coalition is strictly separated from any transnational alliance. In the second national coalition we find the German experts. Thirdly, there is a big national coalition that surrounds the German opposition. In addition to these three primarily national coalitions there are transnational coalitions functioning mainly without German actors. The only exceptions are the analysed media raising their own voices: the FAZ and the SZ once again form a coalition with supranational EU institutions expanded to include the Spanish government and some upcoming member countries. In the French debate on EU enlargement the picture changes. Here we barely find any national coalition formation. French national actors form coalitions together with actors from the EU or other member states. And also those coalitions in which French actors are not involved show a strong tendency for transnational coalition building. Regarding coalition formation, the French enlargement debate does not show any obstacles to integration. Conclusions and Prospects The analysis has shown that debates differ in their potential to integrate Europe in its plurality. The proposed criteria to analyse debates' integration potential do not call for harmonisation between countries, but require that European actors and topics become visible, that actor constellations reflect the dependencies of a common European political space and that debates show conflict and cooperation while portraying a community in which interests are no longer bound to traditional border lines. None of the analysed debates fully fulfils these requirements although the study design facilitates such integrative findings as quality newspapers have been analysed, debates have been studied in countries that are regarded as the engines of the integration process and in issue fields that are strongly europeanised. The debates on the constitutional issue in Germany and France fit the criteria pretty well. In both debates European and member state actors and the issue itself become visible. However, the German debate is still dominated by national speakers - a finding less pronounced in France. Both debates show the interconnectedness of the European political space. With regard to dispute constellations both debates, however, reveal some integration barriers. In Germany consensus and positive references are rare and thus positive identification with a common community becomes difficult. France, by contrast, shows some tendencies for national interest definitions as national actors - at least partly - separate themselves from the rest of Europe. The enlargement debates show stronger integration barriers. Here the German debate makes EU actors and the topic visible and shows the interconnected-ness of a common political space but refrains from showing positive identification points and remains with clear national coalition formations. The dispute constellation in this debate thus does not truly portray Europe as a common community. The French debate on EU enlargement is even more problematic for integration. Although the debate reflects the interconnectedness of Europe, shows a strong visibility of EU and member state actors compared to national ones and a dispute constellation without national coalition formation, it remains hidden from public view. When the largest enlargement round ever in the history of the European Union was decided on, no substantial discussion took place in France. Even when this topic was debated, critical voices were rare. The French debate thus concealed not only the topic itself, but also the conflicts inherent within it. Three criteria have been used to identify integration potentials and barriers in mass media depictions of Europe: the visibility, the interconnectedness and the dispute constellation. These three criteria seem valuable as they leave behind the idea that mass media might only contribute to integration if their contents are homogenized across all European countries. Whether integration in plurality turns out to be more than a hope and whether mass media actually contribute to it, needs to be shown empirically. Therefore a study of media impact needs to be conducted that systematically connects community depictions of the mass media - and this means going beyond an analysis of the quality press - to public perceptions. Such studies are ambitious as a (political) community is defined by a border line and internal connections (Fuchs 1999, 156). Consequently, mass media depictions of communities need to go beyond analysing mass media material as treasure trove of isolated actors, topics and opinions, but study border line definitions and internal connections among its members. The systematic connection of content and network analysis might guide us in this direction as it allows to study more than the visibility of actors and issues showing the way how actors are interconnected and how dispute divides or unites Europe. y The media portrayal of Europe however may not only be regarded as an inde- pendent variable impacting public perception, but also as a dependent variable. In this latter perspective, one might ask which factors trigger a portrayal of Europe that meets these integration criteria. Such a question goes well beyond existing research that focuses on explaining when Europe gets on the media agenda at all. The results of this study make clear that debates vary in their potential to integrate Europe between and within countries. Thus national, issue-specific contexts may help us understanding such variation. One factor that is crucial to describe these national, issue-specific contexts is the relation between elites' attitudes and public opinion11 as the latter determines the national elites' communication strategies (regarding agenda-setting see Gerhards et al 1998, Kollman 1998) and therefore the domestic adaptation of European issues (Adam 2007 a, b). Applying this idea to our four test cases leads to the following hypothesis: only in those situations, in which the national elite represent national citizens, can we find integration potential within debates. In those situations in which citizens do not agree with C > «J-----------------------------------------„. „„—— „----— their national leaders, however, media debates hardly depict Europe as a common community. The strong gap between the elite and the public in France regarding the enlargement issue - the elite supports it whereas the citizens strongly reject it 00 O (European Commission 2002) - results in a silencing of the topic and the inherent conflict. The weaker misfit in Germany between the elite and the public (European C^ Commission 2002) regarding enlargement leads to a strongly negative debate with 2 a national coalition formation. The constitutional debates where lesser integration barriers have been identified, take place in a context where the elite represent the public: the German elite and public support constitution building (European Commission 2002); in France the party system as well as the citizens are both divided on the future shape of EU integration (e.g. Goulard 2002). Consequently, mass media seem to fail to portray Europe as a community if the public rejects what national elites support, as then national politicians avoid discussing the topic at all or refrain to national interest definitions what is strongly reflected in mass media debates. Consequently, one should not overestimate mass media's potential to integrate Europe as their picture of Europe largely depends on the domestic adaptations of common European policies. Such adaptations are difficult, if there is a gap between national elites and the public. Notes: 1. Weber characterises a community by interpersonal interactions and by the affective feelings of belonging together (Weber 1972). The community definition employed here, is based on imagined not interpersonal interactions and it is open for affective and rational - interest or value-based reasons for belonging together (see also Fuchs 2000, 217f.). 2. This analysis has been limited to actors who substantially shape the debate. Only actors involved in at least three communicative interactions and that appear as a speaker at least once have been included. The latter cut level is necessary as otherwise actors would be placed in the same coalition that are similar in the sense that they do not attribute support or critique to anybody. 3. In general there are two main measures to determine similarity: the Euclidean distance and the Pearson correlation. As actors should be placed in the same discourse coalition if they have similar friends and enemies - even if their voices differ in intensity - one needs to draw on the Pearson correlation. The Euclidean distance by contrast measures the absolute difference between actors' reference patterns and thus does not allow to put speakers into coalitions that have similar reference pattern but differ in their outdegree (Backhaus et al 2003, 496, Wasserman and Faust 1999, 374). 4. MDS is based on an iterative optimisation procedure which seeks to place actors with similar reference patterns close to each other in space. The fit between the presented actor constellation and the original similarity matrix results in a stress value. The lower the stress value the better the spatial representation of actors. A hierarchical cluster analysis is used here (Backhaus et al. 2003, 479ff., Wasserman and Faust 1999, 381) based on an average linkage procedure. This procedure is regarded as conservative as it does not show tendencies for dilatation or contraction (Backhaus et al. 2003, 516f.). Such a cluster analysis is visualised with dendrograms that group actors that have very similar reference pattern on a high level. 5. Detailed information on the project that was sponsored by the EU Commission (HPSE-CT2000-00046) can be found at http://europub.wz-berlin.de and in Koopmans' and Stathams' project description (Koopmans and Statham 2002). 6. As the sample for the commentary analysis in the Europub project has been denser as the one for the news reporting analysis, the commentary data had to be weighted in order to create a common data set that allows for a systematic comparison of media and political actors as speakers. 7. The media act as speakers not only in their commentating, but also when they overtly take a position in news reporting. Each commentary is regarded as one claim by one journalist. 8. This analysis is based on 338 articles (518 claims) in the German and 294 articles (482 claims) in the French constitutional debate; on 485 articles (763 claims) in the German and 192 articles (299 claims) in the French enlargement debate. U S i 0 £ * 1 * (R O C 9. Detailed coding schemes can be found in the following codebooks (Adam et al 2002, Koopmans 2002). 10. For such network analysis only those claims can be included that contain at least one communicative interaction between claimant and addressee. Consequently, the network indicators are based on fewer articles than the analysis of a topic's visibility. The exact numbers are: 277 articles (415 claims, 696 relations) for the German constitutional debate; 250 articles (409 claims, 569 relations) for the constitutional debate in France; 392 articles (618 claims, 1050 relations) for the enlargement debate in Germany; 172 articles (267 claims, 348 relations) for the enlargement debate in France. 11. It would be misleading to claim that the relation between the elite's attitudes and public opinion is the only factor that influences the filtering of EU politics at the national level. For a more detailed explanation see Adam 2007 a, b. References: Adam, Silke. 2007 a. Symbolische Netzwerke in Europa. Der Einfluss der nationalen Ebene auf europäische Öffentlichkeit. Deutschland und Frankreich im Vergleich. Köln: Halem. Adam, Silke. 2007b. Domestic Adaptations of Europe: A Comparative Study of the Debates on EU Enlargement and a Common Constitution in the German and French Quality Press. International Journal of Public Opinion Research 19, 4, 409-433. Adam, Silke, Barbara Berkel, Julie Firmstone, Emily Gray, Ruud Koopmans, Barbara Pfetsch, and Paul Statham. 2002. Codebook for Content Coding of Commentaries / Editorials, Anderson, Benedict. 2003. Imagined Communities: Reflections on the Origin and Spread of Nationalism. London: Verso. Backhaus, Klaus, Bernd Erichson, Wulff Plinke, and Rolf Weiber. 2003. Multivariate Analysemethoden. Eine anwendungsorientierte Einführung. Berlin: Springer. de Vreese, Claes. H. 2004. The Effects of Frames in Political Television News on Issue Interpretation and Frame Salience. Journalism & Mass Communication Quarterly 81, 1, 36-52. de Vreese, Claes H. and Hajo G. Boomgaarden. 2006. Media Message Flows and Interpersonal Communication. The Conditional Nature of Effects on Public Opinion. Communication Research 33, 1, 19-37. Dubiel, Helmut. 1999. Integration durch Konflikt. Kölner Zeitschrift für Soziologie und Sozialpsychologie. Sonderheft 39, 132-43. Durkheim, Emile. 1893/1977. Über die Teilung der sozialen Arbeit. Frankfurt a.M.: Suhrkamp. Easton, David. 1965. A Systems Analysis of Political Life. New York: Wiley. Edelman, Murray. 1964. The Symbolic Uses of Politics. Urbana, IL: University of Illinois Press. Eder, Klaus. 2001. Integration through Culture? The Paradox of the Search for a European Identity. In K. Eder and B. Giesen (eds.), European Citizenship between National Legacies and Postnational Projects, 222-44. Oxford: Oxford University Press, Eder, Klaus and Cathleen Kantner. 2000. Transnationale Resonanzstrukturen in Europa. Eine Kritik der Rede vom Öffentlichkeitsdefizit. Kölner Zeitschrift für Soziologie und Sozialpsycholgie. Sonderheft 40, 306-31. European Commission. 2000-2002. Eurobarometer 54.1; 56.2; 57.1. Public Opinion in the European Union. Brussels: European Union European Council. 2001. Die Zukunft der Europäischen Union - Erklärung von Laeken. Ferree, Myra M., William A.Gamson, Jürgen Gerhards, and Dieter Rucht. 2002. Shaping Abortion Discourse. Democracy and the Public Sphere in Germany and the United States. Cambridge: Cambridge University Press. Freeman, Linton C. 1979. Centrality in Social Networks: Conceptual Clarification. Social Networks 1, 215-39. Früh, Werner. 2001. Inhaltsanalyse. Theorie und Praxis. Konstanz: UVK. Fuchs, Dieter. 1999. Soziale Integration und Politische Institutionen in modernen Gesellschaften. Kölner Zeitschrift für Soziologie und Sozialpsychologie. Sonderheft 39, 147-78. _| Fuchs, Dieter. 2000. Demos und Nation in der Europäischen Union. In H. D. Klingemann and F. 1—' Neidhardt (eds.), Zur Zukunft der Demokratie. Herausforderungen im Zeitalter der Globalisierung, 1 ' 215-36. Berlin: Ed. Sigma. Gamson, William A. and A. Modigliani. 1989. Media Discourse and Public Opinion on Nuclear Power: A Constructionist Approach. American Journal of Sociology 95, 1-37. Gerhards, Jürgen. 1993. Westeuropäische Integration und die Schwierigkeiten der Entstehung einer europäischen Öffentlichkeit. Zeitschrift für Soziologie 22, 2, 96-110. Gerhards, Jürgen, Friedhelm Neidhardt, and Dieter Rucht. 1998. Zwischen Palaver und Diskurs: Strukturen öffentlicher Meinungsbildung am Beispiel der deutschen Diskussion zur Abtreibung. Opladen: Westdeutscher Verlag. Goulard, Sylvie. 2002. Frankreich und Europa: die Kluft zwischen Politik und Gesellschaft. In M. Meimeth and J. Schild (eds.), Die Zukunft von Nationalstaaten in der europäischen Integration. Deutsche und französische Perspektiven, 173-95. Opladen: Leske+Budrich. Habermas, Jürgen. 2001. Zeit der Übergänge. Kleine Politische Schriften IX. Frankfurt am Main: Suhrkamp. Hajer, Maarten A. 1995. The Politics of Environmental Discourse. Ecological Modernization and the Policy Process. Oxford: Oxford University Press. Kevin, Deirdre. 2003. Europe in the Media. A Comparison of Reporting, Representation, and Rhetoric in National Media Systems in Europe. London, Lawrence Erlbaum Associates. Kohler-Koch, Beate. 2000. Europäisierung - Plädoyer für eine Horizonterweiterung. In M. Knodt and B. Kohler-Koch (eds.), Deutschland zwischen Europäisierung und Selbstbehauptung, 11-31. Frankfurt am Main: Campus. Kollman, Ken. 1998. Outside Lobbying: Public Opinion and Interest Group Strategies. Princeton, NJ: Princeton University Press. Koopmans, Ruud. 2002. Content Coding of Claims-making. Codebook for EUROPUB.COM Workpackage 2. Koopmans, Ruurd and Jessica Erbe. 2004. Towards a European public sphere? Innovation 17, 2, 92118. Koopmans, Ruud and Paul Statham. 1999. Political Claims Analysis: Integrating Protest Event and Political Discourse Approaches. Mobilisation: An International Journal 4, 40-51. Koopmans, Ruud and Paul Statham. 2002. The Transformation of Political Mobilisation and Communication in European Public Spheres: A research outline. Mittag, Jürgen and Wolfgang Wessels. 2003. The "One" and the "Fifteen"? The Member States between Procedural Adaptation and Structural Revolution. In W. Wessels et al. (eds.), Fifteen into One? The European Union and Its Member States, 413-57. Manchester: Manchester University Press. Pappi, Franz U. 1987. Methoden der Netzwerkanalyse. München: Oldenbourg. Pfetsch, Barbara; Silke Adam, and Barbara Berkel. 2008, forthcoming. The Contribution of the Press to Europeanization of Public Debates. A Comparative Study of Issue Salience and Conflict Lines of European Integration. Journalism: Theory, Practice & Criticism. Special issue on "European Journalism." Pöttker, Horst. 2002. Integration durch Journalismus contra gesellschaftliche Pluralität? Emile Durkheim revisited. In K. Imhof et al. (eds.), Integration und Medien, 323-35. Wiesbaden: Westdeutscher Verlag. Risse, Thomas. 2004. European Institutions and Identity Change: What Have We Learned? In: R. Herrmann, M. Brewer, T. Risse (eds.), Identities in Europe and the Institutions of the European Union, 247-272. Lanham, MD: Rowman & Littlefield. Schulz, Winfried. 1997. Politische Kommunikation: Theoretische Ansätze und Ergebnisse empirischer Forschung. Opladen: Westdeutscher Verlag. Vlasic, Andreas and Hans-Bernd Brosius. 2002. "Wetten dass..." - Massenmedien integrieren? Die Integrationsfunktion der Massenmedien. Zur empirischen Beschreibbarkeit eines normativen Paradigmas. In K. Imhof et al. (eds.), Integration und Medien, 93-109. Wiesbaden: Westdeutscher Verlag. Wasserman, Stanley and Katherine Faust. 1999. Social Network Analysis. Methods and Applications. Cambridge: Cambridge University Press. Weber, Max. 1972. Wirtschaft und Gesellschaft. Grundriss der Verstehenden Soziologie. Tübingen: Mohr. Wessler, Hartmut. 2002. Multiple Differenzierung und kommunikative Integration. Symbolische Gemeinschaften und Medien. In K. Imhof et al. (eds.), Integration undMdien, 56-76. Opladen: Westdeutscher Verlag. Zaller, John and Stanley Feldman. 1992. A Simple Theory of the Survey Response: Answering Questions versus Revealing Preferences. American Journal of Political Science 36, 3, 579-616. S i o £ * 0 C (6 M