Abstract: This paper studies the role of the news media in the criminalisation of ethnic groups, and the relations between criminalisation processes and the racialisation of difference. The claim of the paper is that criminalisation is part of a broader process of social construction, which involves, on the part of the news media, the organisation of topics and issues, as well as processes of labelling and attribution of (individual or group) traits, meanings, causes, and responsibilities. The paper is structured as follows: first, we present an overview of different theoretical and methodological approaches to the study of news media discourse, in order to identify indicators of the criminalisation of ethnic groups. Second, we discuss the two main techniques for analysis: content analysis and discourse analysis. We conclude with a short description of possibilities and problems in the design of (comparative) research in this field.

Key words: criminalisation, racialisation, critical discourse analysis, content analysis, news media

In social science and criminology, news media coverage of ethnic issues (i.e. issues about so-called "ethnic" or "racial" minorities) has been studied as an indicator for social phenomena such as deviance. However, reporting on deviance and crime cannot be taken at face value as a reliable indicator for actual social phenomena since the media actively construct the reality they are reporting. Thus a study that takes the number of incidents of racist violence covered by the media as an indicator of the number of actual incidents is counter-factual. Moreover, it is dangerous, as the publication of these results could end up being used to create social alarm or, alternatively, to "prove" that racist violence is not such a serious problem (see SACCÀ & MARINELLI, 1997). News reports are less problematic if used to trace episodes that are less affected by media selection procedures, such as political demonstrations (cf. OLZAK & OLIVIER, 1999). As far as crime is concerned, some researchers have compared media reporting on crime with actual crime statistics and shown a divergence between the two (for a review of such studies, c.f. MARSH, 1991). In this case, media content is used to show the under- or over-representation of particular ethnic groups in crime news (for example, DIXON & LINZ, 2000). [1]

In contrast, the five approaches on which this paper is focused take the news media as objects of analysis in their own right. These perspectives provide different answers to the question of how and why, as research consistently shows, news on migrants is so often negative and so often about crime. We will discuss the key concepts, as well as the different methodologies and problematic aspects, of each approach, limiting our considerations to the question of media content (although some remarks on issues of news-production and reception will be made). [2]

It is important to note that the bulk of empirical material from this field dates back to the 1970s and 1980s. This research formed part of the general concern with race relations at the time, in particular in the US, the UK, and to some extent other countries such as the Netherlands. More recently, Southern and Northern European countries that have experienced recent immigration have followed up with studies in which questions of representation have been re-examined in the light of this tradition (c.f. BRUNE, 2002; HUSSAIN, 2002). However, it remains difficult to understand to what extent the observed characteristics of migrants’ representations may have changed over recent decades, as the few newer studies on British, French, and Italian data that are reported here seem to suggest. More extensive empirical analyses will have to be done, following similar standards in different countries, in order to examine this further (cf. TER WAL, 2002; TER WAL, 2004). [3]

The mass media have long been considered a sort of mirror, albeit a distorted one, of society. This metaphor has worked in two directions. On the one hand, assuming that the mirror provides the images through which a large part of the social world is interpreted, there has always been a concern about the possible distortion of media representations, which could reinforce prejudice among the public. On the other hand, if prejudice is widespread in society and, consequently, among journalists (who are mostly White), the media is likely to reflect that prejudice. [4]

In North American studies of the 1960s and 1970s this problem was addressed by investigating prejudice and discrimination within newsrooms. These studies showed that the unequal distribution of power between the White majority and Black minority groups in society was also reflected in the composition of the newsroom (BREED, 1955; WILSON, 2002). The small number of Black and Hispanic journalists and their position in non-directive roles helped to explain the stereotyped portrayals of these ethnic groups in the media (GREENBERG & MAZINGO, 1976). Finally, the predominantly White composition of the audience of dominant media has often been cited as a reason for the orientation of news media towards the norms and interests of these projected audiences. [5]

Similar theories have been used to explain the media bias towards reporting on crime. However, the existing evidence for the over-representation of ethnic minority offenders in news is contradictory (cf. GRABER, 1980; MARSH, 1991; GAROFALO, 1981; SACCO, 1995). When Black crime was found to be over-reported, this seemed linked to fears about the threats allegedly posed by Blacks to the White majority group. By contrast, police harassment of Black families was rarely reported (LEY, 1974) as it was considered uninteresting for both the White readership and the major advertisers, for whom Blacks were a group of little commercial importance. Today this is no longer the case, as ethnic minorities constitute an important part of media audiences not only in the US, but also in many, though not all, Western European countries (STATHAM, 2002; D’HAENENS, 2003). [6]

The predominant methodology used in the ‘Minority and the media’ approach is content analysis. Indicators such as the number of articles and the space, position, and font size of the headlines are taken as measures of the distribution of minority news themes. Using this methodology, media bias is observed in various ways. Firstly, there is a problem of unbalanced selection. Crime, race riots, policing, and violence have always been some of the most-covered subjects, while other subjects have been ignored (SCHARY, 1969; WILSON & GUTIERREZ, 1985; VAN SLIKE TURK, RICHSTAD, BRYSON & JOHNSON, 1989; in a different perspective GANS, 1980; KNOPF, 1975; MERTEN, 1986; VAN DIJK, 1991). A study of British press coverage in the 1980s found that reports on crimes allegedly involving Black people have often been given disproportionate coverage (GORDON & ROSENBERG, 1989). Particular ethnic groups may also receive special attention: for example, in the German press Turks have tended to be covered as if they were representative of the entire range of "foreigners" in that country. This group was also more frequently associated with negative personal characteristics, in particular a tendency to crime and violence, than other groups (MERTEN, 1986). [7]

Secondly, bias is indicated by the amount of space allocated to minority opinions, which is always very little even when minority group members are main actors in the news (FISHER & LOWENSTEIN, 1967; SCHARY, 1969; in a different perspective DOWNING, 1980; VAN DIJK, 1991). This means foreigners do not have the opportunity to challenge the negative definitions given of them in the press (MERTEN, 1986). In other words, minorities are mostly represented in the news as speechless actors involved in negative acts (TER WAL, 2002). Similar forms of "biased" or stereotypical representation have been studied in quantitative content analyses using coding categories that qualified actor roles and characters, and in lexicographic analyses (HUFKER & CAVENDER, 1990; BONNAFOUS, 1991). [8]

Journalists’ associations and unions have played an important role in trying to change attitudes related to ethnic issues within their own profession by promoting the use of guidelines and training facilities. A generally increased awareness has also led to more positive reporting practices, as was revealed in a recent investigation into the coverage of (anti-)racism in the UK in the 1990s. One study, based on the coverage of the six-month period prior to the 1997 British general election, claimed that news on ethnic minorities had become the scene of an "anti-racist show", where increasing amounts of space were being given to the opinion of accredited minority actors (LAW, 1997). Nevertheless, the same study found that reporting still remained situated within the dominant discursive field of crime. Similarly, recent research in the UK suggests that while Black African and Asian minorities are now treated in the media far better than in the 1970s and 1980s, negative portrayals are being given of newly arrived ethnic groups such as asylum seekers (STATHAM, 2002). [9]

Although research should not be restricted to this alone, content analysis can be a very useful part of news analysis. Not only does it make it possible to compare the occurrence of "race" and crime themes with that of other themes or the representation and proportion of negative vs. positive portrayal of migrants as compared to other actors, it also allows for the investigation of other, related, analytical dimensions. For example, to:

Clearly, content analysis can only grasp some of the building blocks of social representations, such as labels/denominations for migrants and their recurrence, and further qualitative analysis is necessary. Apart from its methodological limits which will be discussed below, this approach presents a problem regarding the very concept of media bias and distortion: in some cases it is not clear to which model of "unbiased" representation authors are referring. In the case of thematic coverage, what would "fair" representation actually be? How many news items on crime should there be out of the total of items about minorities? As is shown in the next section, the newsmaking approach addresses this problem from a different perspective. [11]

Newsmaking routines are affected by what has come to be defined as "frames" for talking about immigrants, which may reflect stereotypical thinking, or ways to oppose it (for a review of the use of the concept of frame in media research c.f. SCHEUFELE, 1999). However, the newsmaking approach holds that prejudice does not explain everything. The media’s emphasis on immigrant deviance and crime is also the result of routines and constraints inherent to newsmaking. Studies on newsmaking have shown how techniques of news gathering, selection and editing, time and space limits, lack of freedom within the given news format, as well as processes of socialisation within the newsroom and other organisational constraints, help determine news content and perspective. [12]

The newsmaking approach maintains that the selection and presentation of news topics, actors, and events is determined by several conditions. The most central factors commonly identified in the literature of the 1970s and 1980s are: i) news values; ii) news scripts; iii) news themes; iv) ownership, control, and political affiliation; v) differential access to the media; vi) the relations between media and politics; and vii) editorial policies, including relations between journalists and editors-in-chief. [13]

The first general theoretical assumption is that the selection of news depends on dominant news values, i.e. the implicit or explicit criteria adopted by the news media in the selection and framing of events that make it possible for the latter to be sold as news. Important news values are the negativity and recency of events, authority of actors, and consonance of actions with public stereotypes (GALTUNG & RUGE, 1965; GANS, 1980; VAN DIJK, 1988a; for crime news: CHIBNALL, 1977). In their study on racism and the mass media in Britain, HARTMANN and HUSBAND (1974) found that "conflict, tragedy and deviancy" were the main news values dictating the selection of themes. Other studies have also found that crime reporting constitutes a significant part of reporting on ethnic issues in general (GANS, 1980; GRABER, 1980). When migrant crime becomes a news theme in its own right—a topic in which the interest of public opinion is presumed—it may become a news value in itself, favouring the selection and framing of episodes which seem to fit the theme (for "crime waves" cf. FISHMAN, 1978; for migrant crime cf. TROYNA, 1981; MANERI, 1998a, 2001). News values are also found to account for the fact that images of Blacks are frequently distorted through the reduction of complex situations to simple generalisations, to quick and superficial explanations, in order to fulfil the news value of unambiguousness (GORDON & ROSENBERG, 1989). The fact that the circumstantial causes for the occurrence of crime are hardly ever investigated in crime reporting is not necessarily the result of White prejudice or denial of racism. It does imply, however, that the cause of or "blame" for minority crime is more easily attributed to the individual, and by generalisation to minority groups, a mechanism that reinforces and legitimises ethnic prejudice. [14]

Another cause of distortion by simplification lies in the use of news scripts, which perform the function of organising potentially ambiguous elements into easily interpretable stories. New occurrences are often adapted to a pre-existing script—a narrative structure used to write about recursive events that are generally treated in the same way. HOLLAND (1981) has shown how, in Britain, the popular press failed to see that the New Cross fire, which killed thirteen Black victims in 1981, was a racist attack. Accustomed to writing stories featuring Blacks in the role of perpetrators, not victims, of crime, journalists focused their initial reports on the theme of Black turbulence and, following the organization of a demonstration, adopted the script of the "race riot". Similar examples have been found elsewhere in, for example, the Swedish press coverage of an arson attack on a refugee family, when a racially-motivated explanation for the attack was refused (BRUNE, 2002; and for Italy MANERI, 1995). [15]

News making routines such as the use of sources for information and verbal reactions are also constrained by the power relations inside and outside the newsroom. The predominance of institutional actors in the definition of news agendas and contents is explained by the use of these actors by the media as privileged news sources and as sponsors of the news industry. With regard to news on deviance and crime, this means that official definitions of the situation (by the police and other officials) are likely to be prioritised and to receive prominent coverage as well as high credibility (VAN DIJK, 1988a). If less-official sources that could contradict the stereotypical image given are ignored or covered in a less prominent fashion, it is more likely that a one-sided, and possibly biased, image of ethnic groups will be maintained in the media. The choice of institutional sources relies on their direct access to "facts" of supposedly general interest, on their assumed reliability, and on their ability to provide media with continuous inputs presented in an easily retrievable format (TUCHMAN, 1972; GANS, 1980). Institutions also tend to actively phase events to fit the bureaucratic schedules of news organisations (SIGAL, 1973; COHEN, 1980). Organisations such as civil and police services have the necessary resources and professional staff to organise news events and press conferences, and also have information officers and/or "spin doctors" at their disposal who channel news to those outlets preferred by the organisation before it is generally made public (cf. for the Netherlands, SPRENGER & DE VREE, 2004). [16]

Finally, newsmaking routines and conventional news formats are generally determined by editorial policies (GANS, 1980). For example, in the 1990s most Italian news reports on immigration were—for organisational reasons, among others—part of the local or crime news genre ("cronaca"), and often journalists with a general professional interest in this genre were assigned to cover ethnic issues (TER WAL, 1997). Journalistic routines and lack of resources also favour the reproduction of readily-available official figures and accounts. For the same reasons, investigative journalism, background reports, and explanatory accounts are rare (though a recent cross-European study reveals an increase in investigative reporting [TER WAL, D’HAENENS & KOEMAN, 2004]). [17]

The prevalent methodologies used in the newsmaking approach are ethnographic observation, content analysis, and discourse analysis. Related research questions are:

Many scholars analysing these aspects have stressed that newsmaking constraints do not explain away the racist implications of news-media reporting. They also stress that it is not individual journalists who are to be blamed for this. Both observations highlight the importance of examining not only the newsmaking process but also the dominant cultural outlook on and common sense ideologies about ethnic relations, as well as the ideological constructions that are involved in reporting about ethnic issues and crime. [19]

Classic studies representing this approach are the 1970s studies on the social construction of crime and (youth) deviance in the British press (COHEN & YOUNG, 1973; COHEN, 1980; HALL et al., 1978). All of them are characterised by a wider perspective, which inscribes empirical findings in a historical, socio-economic, and political perspective. The reality represented by the news is conceived of as an essentially ideological construction, based on consensual definitions of marginality and deviance reproduced by the media through their institutional sources. According to these studies, political and social elites, acting as "primary definers", provide that limited set of discourses which are used by journalists to make sense of the world. [20]

In their influential study, Stuart HALL and colleagues (1978) argued that a state response to deviance through public order measures could be justified thanks to the coverage of a so-called "mugging wave" in urban areas, a coverage which ignored or dismissed the specific characteristics of a social problem/crisis underlying the events. HALL et al. revealed the constructed nature of mugging, which they considered "not as a fact but as a relation—the relation between crime and the reaction to crime" (p. viii). The creation of "moral panics" (COHEN, 1980; GOODE & BEN YEHUDA, 1994; HUNT, 1997) such as mugging or asylum scares, has been interpreted as a powerful instrument in building a consensus and setting the agenda for the solution of these social problems with law-and-order policies. The moral panic is sustained by appeals to common sense fears grounded in "objective" analyses of the problem at hand. The creation of a moral panic is defined as follows:

"A condition, episode, person or group of persons emerges to become defined as a threat to societal values and interests; its nature is presented in a stylised and stereotypical fashion by the mass media; the moral barricades are banned by editors, bishops, politician and other right-thinking people; socially accredited experts pronounce their diagnosis and solutions; ways of coping are evolved or … resorted to" (COHEN, 1972, p.9). [21]

In this study, Stanley COHEN reconstructed the stages of a moral panic, in which various social actors—judges, politicians, police, media and "public opinion"—played a predictable role, and showed how these actors contributed to the criminalisation of migrants. [22]

From the perspective of discourse, a primary interest has been the analysis of the forms and origins of the myths that ideological signification has built upon. "Riots" in the early 1980s in Britain were defined—and then dealt with—as a criminal, rather than political, phenomenon through reporting centred around the image of a "thin blue line" of police defending the community against an "unprecedented" wave of violence and lawlessness (MURDOCK, 1984). Another central myth in British discourse on migrant crime has been that of the "inner-city", an isolated place or "no-go area" alien to the norms and values of the White middle-class—something potentially destructive for the "British way of life" (BURGESS, 1985). The banlieues in France and several metropolitan areas in Italy have been subjected to a similar construction of deviance (HARGREAVES, 1996; MANERI, 1998b). [23]

In order to provide answers about the role of different institutional actors in the social construction of deviance, qualitative research is required. Some of the most important elements in this research will help to:

A broader perspective in this approach would ideally include not just analyses of media content, but also field observation, interviews of strategic actors, and analysis of official data and documents. However, such work has rarely been done, and the role of the various actors is generally inferred from the analysis of media materials. In a similar fashion, the strongest methodological critique made of research in this perspective concerns the difficult task of analysing both criminal action and societal reaction: although claims are made about the "overreaction" that lies behind any moral panic, data are not systematically analysed to prove it. This problem is difficult to overcome, since official data on crime are very sensitive to reactions to it, and indeed measure the latter as much as the former. [25]

For the analysis of the news, qualitative methods have been preferred to content analysis, but few, if any, systematic textual studies have been undertaken. Moreover, unless clear indicators are defined, a focus on the reporting of a case, by reconstructing the various stages and headlines, sometimes risks being no more than a summary and paraphrasing of what is found in the material, i.e. a narrative or anecdotal reconstruction of the events covered in the news. Discourse analytical theory is aimed at defining categories for the analysis of textual and linguistic data in a meaningful and systematic way. [26]

In discourse analysis news text is the object of analysis, combining insights from the social constructionist and newsmaking approaches with linguistic and social cognition analyses. This approach claims that negative reporting is only partly a result of the above-mentioned routine conditions of news reporting. Of course, the fact that information has to be presented in a condensed and abbreviated manner requires operations of selection, summarisation, and generalisation. News discourse itself is a social and cognitive practice. Negative reporting is produced by (the reproduction of) dominant ideologies that are encoded in the structures (that is, the content and style) of news text (VAN DIJK, 1991). More specifically, discourse analysis maintains that text and talk play a crucial role in the reproduction of ethnic prejudice and racism. Indeed, beliefs and opinions about ethnic out-groups become socially shared through text and talk. Because most in-group members do not have a daily interaction with immigrants, their beliefs and knowledge about the out-group are shaped to a large extent by the media (VAN DIJK, 1987, see also section on news and public attitudes approach below). [27]

The main direction in this approach is that of critical discourse analysis (CDA), and most of the scholars working in the field have a linguistics background. Critical discourse analysis focuses on the roles of ideology and power and their enactment and reproduction through discourse. It criticises the ways in which existing power inequalities and discrimination are maintained and reproduced through discourse. Discourse has not only pragmatic functions of persuasion and credibility enhancement but also socio-political functions of legitimisation and control. [28]

The analysis of news discourse has often combined quantitative analyses (as described above) with qualitative analyses of smaller sets of news discourse (EBEL & FIALA, 1983; VAN DIJK, 1991). These are used to show how news definitions build on the dominant perspective of the powerful and on a mutual reinforcement of official and popular perspectives on the definition of social problems. More precisely, it is shown how news reports regarding ethnic issues reproduces dominant "situation models" which support and confirm negative beliefs about other ethnic groups (VAN DIJK, 1991, 1993). [29]

In order to analyse the role of news discourse in reproducing stereotypical ethnic beliefs and prejudiced attitudes, this approach builds on analytical categories borrowed from social psychology, social representations theory, and social cognition theory. It uses notions of in- and out-group definitions, and assumes that beliefs about out-groups are organised in schemata (prejudiced attitude schemata). For example, the schema for beliefs about refugees is composed of several propositions having to do with claims that refugees are "bogus", "the victim of criminal organisations/smugglers", that they "come in large numbers and cannot be controlled", that they are "welfare scroungers", and "are likely to end up in illegal, deviant or criminal activities". Through an analysis of argumentation strategies and quotation patterns it is shown how such beliefs or claims are used to justify law and order policies or the restriction of the right to asylum (TER WAL, 1996; VAN DIJK, 1991). [30]

The socio-cognitive discourse theory of Teun VAN DIJK (1991) claims that the representation of ethnic groups in news is influenced by pre-existing beliefs and attitudes about the general threat and particular deviant behaviour of migrants that are stored in the "social memory". Beliefs and opinions about ethnic issues are revealed in the presuppositions, assumptions and meanings that are expressed in news discourse through—among other things—topical organisation, lexical choice and syntactic style, and argumentative structures. Similarly, in critical linguistics, studies of the ideological functions of news discourse have been applied predominantly to the syntactic and local semantic levels of discourse (KRESS & HODGE, 1979; FAIRCLOUGH, 1989, FAIRCLOUGH, 1991; FOWLER, 1991). Local semantics are discussed in the section on techniques for analysis. Discourse analysis is nonetheless directed not only at the local level of word and sentence structure and meaning, but also at the level of larger units of meaning, such as paragraphs, and the global organisation of text structure and meaning (topics) (VAN DIJK, 1988ab, 1991). The latter can also be used as a basis for quantitative analysis. [31]

A study of news discourse production and text features in the criminalisation of ethnic groups may include an analysis of the following (see also research techniques, Section 2 below): [32]

The social and institutional routines of newsmaking are related to structural properties of news reports, such as the summarisation of global meaning or the highlighting of salient information in headlines. The appearance of official actors in more prominent positions than non-official and minority actors in the news is also related to these criteria of newsmaking. That is, the most crucial problem definitions are found in the most prominent and recurrent elements of the structures of news reports, such as headlines and other news schema categories (VAN DIJK, 1988b). Therefore, the representation of in- and out-groups and the use of argumentative strategies in headlines and leads of news reports deserve special attention. The textual function of headlines is to represent the global theme or topic; articles are characterised by a biased schematic or thematic organisation when a piece of information with only minor importance in the main body of the article is fore-grounded, i.e. expressed in the headline. [33]

In addition, news outlets tend to "tag" their services as a means to give a first "framing" for the events and situations described. In newspapers this is the "tag" at the corner of the newspaper page or at the beginning of the headline, citing the rubric/subject area within which the news is covered, such as "foreign affairs". For special issues there can be use of more specific tags such as "emergenza immigrati" ("migrant emergency") or "banlieue" ("ghetto"), which can be telling about the framing of these issues in mainstream media. In North-American television news the words "the war on terror" show a similar use of tags. [34]

Quotation patterns reveal the assignment of speaking and social roles to actors with different levels of power and status within, and access to, newsmaking (ZELIZER, 1989; VAN DIJK, 1988b). This means the roles attributed to different actors involved in ethnic issues can also be revealed by the position and way in which they are quoted. In primary news accounts official and expert discourses are frequently set apart to provide the news reports with authoritative statements which signify the importance of the events or the issue at hand. Editorials frequently contain different forms of reported speech and indirect quotation or other forms of "inter-textuality". Finally, interviews offer yet another form of speech interaction that can be analysed in terms of turn-taking, sequencing, and strategic moves. Qualitative discourse analysis of reported speech may identify forms of indirect and direct quotation, several ways of (de)legitimising or up/downgrading statements made by different actors and their confirmation/ contradiction, or repetition. This indicates not only the different news value and access of actors, but also the "reading of events" which is consequently imposed on the audience. [35]

The text structure of editorials is composed of argumentative categories. Editorials do not just comment on immediate events; they also address wider moral and political issues in what could be interpreted as ideological moves. Because of their focus on argumentation, and their giving voice to elite group representatives, editorials play an important role in the justification and/or challenge of official definitions and policies. The use of argumentation strategies in news reports may also show the way in which power relations are embedded in news discourse. That is, it may reveal that arguments produced in a specific format or by a specific group of news actors are likely to be deemed valid/invalid, credible/questionable or functional/ dysfunctional to the justification of a particular response to immigration by a potentially large group of people. Common argumentation strategies in news discourse about crime are:

In order to unravel the thematic and argumentative links between discourses on issues of, for example, poverty, housing, "race", and crime, analysis could focus on discourse on the broader theme of illegality. Illegality is often used in public discourse as a container concept for these different forms of categorizing "problem groups", even when not of the same ethnic mi (or majority!) background. It can be interesting to look at the links between arguments and themes in order to unravel incongruences, gaps in argumentation, or camps in the debate. [37]

Analysing the narrative organisation of the various elements of a text (topics, episodes, and other narrative functions) is another way to reveal the constitution of meaning. In news, a chronological account of an event is rarely given. Rather, in hard news in the quality press, the order of elements proceeds as their news value decreases. The systematic downgrading of a topic or episode is an indicator of its lesser news value. In news with a dramatic construction, such as many feature stories, some narrative sequences may favour interpretations that would be unlikely if the order of the elements presented was different. [38]

In Italy, for example, local protests against migrants were reported following a script that presented a typical ordering of four elements: disorder brought by migrants—neighbourhood protest—police intervention—resolution of the problem. This ordering was suitable for blaming the victims of the protests and presenting policing of migrants as a natural action in the course of events reported (MANERI, 1995). [39]

The theoretical assumptions underpinning discourse analysis regarding the role of dominant ideologies and power have been questioned and criticised. The reason is that if power is taken as an a-priori given, then the results of research tend to simply confirm this role without further qualification. Explanations in terms of power do not allow for qualified comparative research to be made. They rather favour the collection of further evidence used for the purpose of an anti-racist critique. Some feel that this critique is based on a presumption of "intellectual hegemony", by which the critical scholar imposes her/his "own frame of reference on a world already interpreted and endogenously constructed by participants" (WETHERELL, 1998, p.388). Indeed, in recent research, neither the dominant ideologies theory adopted in discourse analysis, nor the primary-definers thesis of the social constructionist approach, have been taken for granted. [40]

For example, HARGREAVES (1996) shows that constructions of difference found in the reporting of the French press on the "banlieues" did not correspond to the official definitions, and that the social movement of immigrants was what first attracted media attention to these areas. However, the outcome of stereotypical and negative representations of immigrants was the same. Indeed, although "primary definers" do not always set the agenda, nor necessarily provide the first definitions adopted in the media, their preferential access and roles nevertheless tend to contribute to a marginalisation of opposing discourses, in particular the discourse of the immigrants themselves. Still, even within discourse analysis it is now generally agreed that media analysis should pay more attention to variety and diversity, in order to account not just for dominant views and reproduction but also for the role of media in the transformation of culture and society (for example, FAIRCLOUGH, 1995). [41]

As a final approach we will discuss the difficult problem of the study of effects in the communication research tradition. Findings of earlier analysis of press coverage combined with attitude analysis suggested, in agreement with the agenda-setting hypothesis, that "the media do not determine how people think, but mostly what to think about" (HARTMANN & HUSBAND, 1974). This important piece of research on the portrayal of ethnic minorities in Britain analysed the recurrence of themes and linguistic devices with predominantly quantitative methods. Other studies found interesting divergences between the representation of different migrant community crimes in the news, on the one hand, and the cognitions of readers about migrant crime, on the other (GRABER, 1980). More recent examples of divergences between minority audiences’ perceptions of their own group’s representation in the news as compared to actual news contents can be found in DEVROE (2004), ROSS (2000), and POOLE (2001). [42]

Research within the discourse analytical tradition has indicated that media coverage does shape the way that people speak and think about immigrants. On the basis of qualitative analyses of interviews in inner-city neighbourhoods in the Netherlands and the US, VAN DIJK (1978) found that the media, along with personal experience and hearsay, formed one of the main sources for White people’s stories about ethnic minorities and thus for the acquisition and reproduction of socially-shared knowledge about ethnic minorities.

"On many occasions, people refer to the media as a source of information or as a source of ethnic opinions with which they may agree or disagree. […] interpersonal communication about ethnic groups, especially in the low-contact areas, is heavily dependent on media information. People mention the media in general, or the press in particular, for "evidence" about the negative characteristics of ethnic groups. Crime is the major topic in this case, although sometimes also other themes are mentioned, such as cultural differences or favourable treatment". (VAN DIJK 1987, p.153) [43]

Most research in this field has opposed the assumptions of traditional attitude research that measures opinions using fixed questions and categorisations. Instead, discourse analysts have stressed the need to understand the variability of attitudes as they can be found through qualitative analyses of sequences of text and narratives, in particular in applications of conversation analysis (BILLIG, 1991; POTTER & WETHERELL, 1988; WINDISCH, 1990). [44]

As part of the analysis of the construction of negative out-group representations and boundary-drawing practices in the news, and their possible effects on public attitudes, it might be possible to examine the correlation between public perceptions of different ethnic groups (and related degrees of racism/xenophobia) and the amount of news coverage on certain ethnic groups (campaigns), and also the amount of political resonance given to transform this into a political issue (for a quantitative study on asylum issues, see KOOPMANS, 1996). In order to research this qualitatively, one would need to combine media analysis with fieldwork or other techniques such as focus group or in-depth interviews. With these same techniques it would be interesting to analyse individual, interpersonal, and organised group discourse before and after the media coverage of foreseeable important issues, such as the arrival of large numbers of refugees (VERKUYTEN, 1997). [45]

Given the state-of-the-art as roughly sketched out in the preceding paragraphs, we can now evaluate the techniques used for research in this area as follows. Firstly, content analysis is a useful approach because it makes it possible to generalize newsmaking features (frequency of news subjects, actors, and their quotes) for a large amount of articles. Secondly, in order to examine the structures and meanings of news, a detailed qualitative discourse analysis is needed. This approach also allows for questions to be asked about the sorts of social identity, the versions of "self" and "other", that the media project (and about the cultural values that these projections entail). Thirdly, discourse analysis can be complemented by ethnographic field work/observation, interviews with privileged actors, analysis of official data and documents, and possibly by semiotic analysis (cf. VAN LEEUWEN, 2000). [46]

In some studies two or more approaches have been incorporated. The general question about the presentation of news, which is studied by asking who was talking about what, in what way, in which position, can indeed be answered by using a variety of approaches. The most obvious combination is that of content and discourse analysis, used as separate but complementary research methods. For example, one can analyse not only the frequency of quotes by various (minority/majority) actors, but also how they are quoted. In this section the analytical categories of the two main techniques of content and discourse analysis are listed and explained further. [47]

Content analysis is a quantitative technique used to study large corpuses of text. Although some indications for the development of a "qualitative content analysis" have been suggested, we will consider it here in its classic quantitative orientation, since such new qualitative elements either transform this technique in something completely different (but similar to what we will describe in the section on discourse analysis) or do not change its quantitative status in a significant fashion. When doing discourse analysis the researcher can chose between two main procedures:

Words and key symbols (metaphors, slogans, stretches of discourse) are chosen according to the research hypothesis, counted, generally classified into more general categories and then analysed using different techniques. The words most frequently selected to study migrants and crime are those used to name (designations) and qualify (epithets and qualifiers) the actors involved. A comparison between British and Dutch newspaper headlines published between August 1985 and January 1986 shows, for example, that the British press preferred racial terms ("Black", "race") while the Dutch press made reference almost exclusively to national or generic terms ("Turks", "refugees") (VAN DIJK, 1988c; 1991). Studies in which all significant words are analysed, generally from headlines, are also frequent. In VAN DIJK’s (1991) study of the British national press the most common words found in headlines were "police", "riot", "Black" and "race". [49]

Lexical categories may be cross-tabulated with other variables (for example newspapers, years, countries, migrants’ origins) or treated with statistical techniques. The analysis of lexical correspondences is frequently used because it makes it possible to present results graphically (with a scatter plot) in a synthetic and efficient manner. What is shown is the structure of the mutual associations of the words studied and/or of the associations with a categorical variable. In this way it is possible to see indicators of the most common discourses (represented by clusters of words) and to individually associate them with, for example, their origin (different newspapers, years, nationalities, countries). Analyses that use lexical indices work mainly at a thematic level and, since they start from simple linguistic indicators, do not allow for the consideration of complex questions. [50]

With this technique it is possible to obtain quantitative data simply by filling out a (electronic) standardised questionnaire (or codebook) that includes questions about communication units (newspaper articles, TV news items, etc.). The variables may measure frequencies of topics, subjects, actors, and quotes. In this case, a list of possible values (typologies of actors, for example) must be arranged after an initial exploration of the materials to be analysed. A possible alternative is to build such a list following an examination of open-ended questions. The use of a residual "other" at the end of a value list is preferable as it allows certain values to be maintained within existing categories. Variables may also try to measure dimensions of the portrayal of the actors (more or less "racialised", threatening, inclined to crime, and so on) and processes. They may also measure the attitude of the journalist towards the actors involved (more or less favourable, sympathetic, etc.), or their perspective on, or "framing" of, a specific issue. In order to achieve a reliable and valid set of variables, the codebook needs to be accurately pre-tested. Especially for variables considering representations and attitudes, codebooks must include detailed explanations of the dimensions studied as well as instructions for the analysts. [51]

Compared to lexical indices, coding categories allow for a deeper investigation of content and a broader range of research questions, but do so in a way that leaves perhaps too much space to the subjectivity of the researchers and their interpretations, meaning these will need to be controlled for reliability. Data produced by codebooks are limited by not being grounded in linguistic indicators. Besides, the more they aim at enquiring into complex dimensions such as the representation of actors, attitudes of the journalist, and so on, the more likely it is that they will come up with vague and abstract conclusions, and the more difficult it will be to achieve inter-coder reliability. Moreover, from such an analysis it may be possible to conclude, for example, that in 30% of the articles migrants are portrayed as a threat without it being possible to further qualify this finding. [52]

Whereas lexical indexing forces the researcher to address simple questions or to draw complex conclusions from rough data through uncontrolled inferential processes, the use of coding categories leads to the recurrence of the problem. The inferential work is done at the beginning, and the results bear no trace of the (linguistic) material used. This is why content analysis should preferably be restricted to the provision of (quantitative and reliable) data at a (simple) thematic level, or at the level of the presence-absence of actors, quotes, and sources. [53]

As stated earlier, discourse analysis differs from traditional content analysis and lexical indexing. In lexical indexing the evaluation of a word, and the different meanings assigned to it in different contexts, cannot be assessed unless the contexts are checked for every occurrence (a very time-consuming exercise), and only at a local level, of co-occurrences of words in the same sentence parts. The influence of structure, ordering, etc. on the meanings of language is completely lost with lexical indexing. In discourse analysis, text is analysed in context, and so words are analysed together with their specific syntactic and pragmatic functions. In addition, the ordering of and coherence between larger chunks of text is considered, something that is not possible in content analysis. Discourse analysis is thus directed at meanings at the pragmatic level of communication underlying local and global semantic structures. [54]

Discourse analysis usually works with a checklist of analytical categories, which are divided into different levels of analysis. The analyst usually starts by writing up a "summary" (following specific rules of global meaning composition) of the text that represents the semantic macrostructure. He or she then passes to the level of local textual analysis, always keeping in mind the context and function of the analysed fragment in the text as a whole. [55]

Although discourse analysis does not allow for large amounts of text to be analysed, it is possible to use quantification to summarise the recurrence of particular discourse analytical indicators, such as topics, argumentation strategies or "topoi", or syntactic choices for particular actions. [56]

The production and processing of news texts is assumed to require cognitive operations, the activation of previous knowledge, and the updating of existing situation models and group- and event-schemata (VAN DIJK & KINTSCH, 1983). Cognitive relevance is revealed in the physical organisation of news texts. The superstructure of news reports or news schema consists of a number of conventional categories, which exhibit a special linear order as well as hierarchical organisation. Superstructures determine what content typically comes first, second, or last in a text (VAN DIJK, 1988a). Some basic elements in the analysis of the global semantic level could be:

Nominalisations, use of intransitive verbs and of the passive voice, and omission of the agent are used in order to conceal in-group agency in the portrayal of negative acts. For example "Eleven Africans were shot dead [...] when Rhodesian police opened fire on a rioting crowd" is a well-known example used by FOWLER, HODGE, KRESS and TREW (1979), to show how syntactic choices may be used to conceal agency, in this case of the police, and to frame the events as racial conflict (for an example of a quantitative analysis of syntactic choices cf. MANERI, 2001). On the other hand, agency and ethnic identity are used in headlines in which "they" are associated with negative topics (illegal entry, protests, and crime). Minorities may thus be portrayed as active, responsible agents, even when they are actually the victims of repressive policing, community measures or other restrictive policies (GRABER, 1980, FOWLER et al., 1979; VAN DIJK, 1991). [60]

Among the discursive devices through which restrictive measures are justified and populist intolerance represented as a natural expression are metaphors of flood, war, walls, and metaphors representing immigration as a pathology, etc., which work to "emotionalise" the facts (CHILTON, 1994; TER WAL, 1991). Rhetoric also functions at the textual level, for example in the organisation of argumentative structures in editorials (see above). [61]

A common criticism of studies of press discourse is that the impact and audience size of the press are minor when compared to those of television. It is true that much research on the press is done in part because the gathering of newspaper materials is much easier than that of TV news items. However, there is reason to believe that the print media still plays an important role. They are cited as a means of credibility enhancement in interviews (VAN DIJK, 1987); they also generally set the agenda for the other media and can play a crucial role in local mobilisations and the definition of most local issues related to crime and migration, which have been shown to have great influence in the construction of discourses about racialised "others" (TER WAL, 1996; MANERI, 1998a). An interesting focus of analysis might be the analysis of the incidence of TV news coverage and TV documentary and talk shows within the development of larger issues in the national and local press, and vice versa. Another interesting task would be to track the "migration" of news scripts, utterances, and interpretative frames from one medium to another. [62]

In addition to news texts, further analysis of related sources such as press releases and news agency dispatches may provide useful insights into newsmaking and social constructionist perspectives. It makes it possible to compare news texts with actual sources, e.g. to verify to what extent journalists actually rely on these sources. Similarly, the special role of the police and court statements, and their specific news value and formulations, may be further investigated by examining police wires or press releases. [63]

Researchers can gather audio-visual data personally, by scheduling recordings of television news, but this is time-consuming and must be planned in advance. For the analysis of large datasets or for periods further back in time, electronic archives must be considered, although the possibilities of accessing archives varies across countries and broadcasting organizations and can be very expensive. For written news, especially in large sets, electronic data seem attractive but are not available for all newspapers, which can lead to ad-hoc justification for choosing those sources that do have electronic versions available. Not all newspapers are available on CD-ROM or accessible via archives or databases. Another problem concerning electronic data regards the graphic format and sometimes the content: web or CD-ROM version of newspaper articles may be different from those of the printed version, which supposedly reached a larger audience. [64]

In fact, in the case of newspapers, data gathering does not often benefit from access to electronic archives or databases. Article retrieval using keywords or subject fields, or even the combination of both methods, gives poor results, unless the research topic is defined in such a way as to match only a few commonly occurring keywords. Even in this case, articles gathered from electronic archives are just a part of the set of articles that can be selected manually, although the gap consists mostly of short articles or local news reports. Electronic data can be very useful at the analysis stage, although only for some quantitative studies. For the analysis of audio-visual material transcription is needed, and in the case of TV discourse, conversation or content analysis must be integrated with analysis of non-verbal communication. [65]

Analyses of news discourse on ethnic issues have often focused on case-study materials. The problem with case studies is that they may be representative only of the particular (not the typical or paradigmatic case). Attempting to generalise from specific and or particularistic data is unsatisfactory: although similar results and patterns do recur across time and in different countries there may still be selectivity in the examples or cases chosen. The advantage of case studies is that they allow the analyst to focus immediately on the most-debated and interesting aspects of the material, to reduce the amount of data, to go more in-depth, and to study the dynamic relationships between the discourses and activities of the various actors at play. For comparative research, case studies also have the advantage of allowing for a selection of the data that makes comparison meaningful. [66]

Case studies have generally dealt with high-profile events of notable policy or institutional relevance, such as the arrival or expulsion of large numbers of refugees (TER WAL, 1997; WODAK, MATOUSCHEK & JANUSCHEK, 1993), or crime incidents elevated to national emergencies (MANERI, 2001). Specific incidents, such as the Rushdie affair (VAN DIJK, 1991), the "race riots" in inner-city areas (MILES, 1984), "racial violence" in London’s East End (KEITH, 1995) and, more recently, analyses of the Rodney King beating (US) and Stephen Lawrence case (UK) have also been studied. [67]

The news material for such case studies may differ to a considerable extent from the day-to-day routine coverage of ethnic issues in urban areas. The latter are as important for the cumulative effects of the reinforcement of negative stereotypes and opinions about ethnic out-groups as the former. Nonetheless, MANERI (1995; 1998a) has shown how, in the 1990s, Italian press campaigns against migrant crime, which were originally covered as individual crises, increasingly became part of routine daily coverage. In such instances, a case study can give insight into everyday dealings with "race" and deviance, making it possible to study, in press and TV, the transformation of the daily local crime news into political and national issues. [68]

The reasons for undertaking an extensive analysis are clear: doing so makes it possible to summarise large amounts of data and to produce easily readable/convincing quantitative results and monitor everyday coverage. The problem of extensive analysis is that it is hardly possible to do anything other than quantitative analysis, with the limits underlined in the previous and the next section, unless one concentrates on a subset, such as a particular format or theme or, alternatively, uses a large research team. One would have to be very selective in extracting interesting data for qualitative analysis. However, from within a corpus of selected news texts for extensive analysis one can adopt the selection criteria of primacy (news peak days, page number) and prominence (headlines and position in the article, quotes, and portrayal of the main actors involved). It is also possible to focus on questions such as when daily crime reporting starts receiving political attention, when reporting is used for social/protest mobilisation, and who does this. [69]

As suggested by research in Italy, another problem may be that coverage of ethnic affairs, especially initially, may pass through long phases of latency, with issues being picked up as emergencies only occasionally, either as a result of political agenda-setting or of media campaigns (MARLETTI, 1989; MANERI, 1995). In these cases choices in the sampling method (for example the selection of particular weeks or days) or in the period of time covered by the research design may produce results valid only for that particular sample or period. [70]

There seem to be more problems than possibilities for comparative research. Some possibilities for comparison within one set of data were mentioned earlier. It is also possible to compare news coverage and discourse at different points in time; such studies do exist (see above). In order to make such time comparisons less descriptive, it might be interesting to account for changes over time and across countries in terms of the discursive management of anti-racist norms, i.e. in norms about what a person can and cannot say about migrants without being accused of "racism". [71]

However, cross-national comparison, as the most widely accepted form of comparison, is more difficult to operationalise, in part because it is not always clear what exactly is meant by this term. There is little consensus as to which analytical categories and which terminology should be used, even within a single approach such as that of discourse analysis, where the similarities across countries are readily observed. This is mostly because there have so far been few attempts at collective and comparative research efforts. The existing results are part of larger monitoring efforts that only look at the position of migrants or ethnic minorities within the framework of studies about news in general, or sometimes with an additional focus on gender and race issues (for example, the WACC Media Watch content analyses; c.f. SPEARS, SEYDEGART & GALLAGHER, 2000). In addition, existing national research tends to use established nation-specific datasets and selection/analysis methods and indicators, so it cannot be used for post-hoc comparative purposes. [72]

Recent attempts at comparative research in this area have been restricted to quantitative analyses. Recently TER WAL (2004) conduced a cross-national content analysis using the same coding instrument in 15 EU countries, in cooperation with the European Monitoring Centre on Racism and Xenophobia and On-Line More Colour in the Media. This quantitative analysis adopted a new approach, in that the same coding instrument was used in all participating countries. The focus was not only on news involving "coloured" people, but included all domestic news stories. This inclusive approach made it possible to compare the impact of deep-seated news values and other recurrent newsmaking practices on general news contents with that on news stories about ethnic minorities. The analysis showed frequent stereotypical images of minorities being relegated to areas of negative societal coverage linked to crime and deviance, as opposed to the appearance of ethnic minority celebrities on the positive end of the spectrum, or members of minorities being portrayed as ordinary people mostly in relation to asylum and migration issues. In addition, ethnic minorities or migrants became a focus when identity issues such as fundamentalism and religion were discussed. Everyday ethnic relations were often covered without reference to the views and perspectives of the minority protagonists themselves. Even in stories about their own job position, minorities were quoted less than their majority counterparts. In the regional press, integration and identity topics were counter-balanced more than elsewhere by the predominantly negative attention paid to minorities in crime news. The study did little to confirm the hypothesis, implicit in discussions about fair portrayal, that local or regional press would pay (or could more easily be persuaded to pay) more positive attention to ethnic minorities. [73]

To date, few comparative studies exist that make any form of systematic qualitative comparisons. VAN DIJK (1989, 1991) did a study of both the Dutch and British press using the same time span and analytical categories. Here, general patterns of discourse and discursive functions turned out to be very much alike, within the limits of different journalistic traditions and the ideological affiliations of the newspapers examined. Quantitative results varied, although similar patterns in topic use and use of quotes were found. The limits of cross-national comparison in discourse analysis lie in the descriptive character of this approach. The discourse theoretical framework accounts for the structure and organisation of text and underlying beliefs, attitudes, and communicative constraints, but not for the variation in social representations. Moreover, the main aim of the discursive studies done in this area has generally been to denounce the unfair representation of migrants, and evidence from various countries was thus primarily gathered to test the recurrence of negative representation. However, techniques such as topic, lexical, and frame analysis do allow for cross-national comparisons with a more explanatory orientation as well. It is recommended that such research be undertaken in the future to analyse the specific features of news media coverage of migrant crime and its relation to the general coverage of migrant issues. [74]

Whether qualitative or quantitative, cross-national comparative research in news analysis works with a number of given differences: countries with different journalistic traditions; different divisions in (political) orientation, editorial policy of newspapers, and programming policies. Also, each country presents different issues related to ethnicity, as well as differences in terms of the position and treatment of migrants, and related categorisations of migrant or minority population groups. Furthermore, there is a problem with different languages, and different denominations for "migrants" (indeed, the term itself is not unproblematic). The signification and significance of the "race" dimension may also differ widely from one country to the other: this could be an interesting focus of analysis in combination with discourses on crime and deviance. Also, in some European societies—partly due to different histories of immigration and integration—explanations for the reported predominance of crime or deviance among specific sections of minority populations may be connected more often to questions of cultural difference (e.g. generational differences or upbringing according to the religious and cultural orientation of migrant parents), while in others they may be linked more with socio-economic problems (deprivation, segregation, and/or discrimination). Informed questions about differences in the definition and treatment of migrant crime by different actors in society, including the influence that European policy and lobby or protest group agendas may have on the definition of such issues, will constitute an important starting point for future research in this area. [75]

Billig, Michael (1991). Ideology and opinions. Studies in rhetorical psychology. London: Sage.

Bonnafous, Simone (1991). L’Immigration Prise Aux Mots. Les Immigrés Dans La Presse Au Tournant Des Années 80. Paris: Kimé.

Breed, Warren (1955). Social control in the newsroom: a functional analysis. Social Forces, 33(4), 326-35.

Brune, Ylva (2002). Sweden. In Jessika ter Wal (Ed.), Racism and cultural diversity in the mass media (pp.395-420). Vienna: European Monitoring Centre on Racism and Xenophobia.

Burgess, Jacquelin (1985). News from nowhere: the press, the riots and the myth of the inner city. In Jacquelin Burgess & John R. Gold (Eds.), Geography, the media and popular culture (pp.192-227). London: Croom Helm.

Chibnall, Steve (1977). Law-and-order news. An analysis of crime reporting in the British Press. London: Tavistock.

Chilton, Paul (1994). "La Plaie Qu’Il Convient De Fermer ...": Les Métaphores Du Discours Raciste. Journal of Pragmatics, 21, 583-619.

Cohen, Stanley (1980). Folk devils and moral panics. New York: St. Martins Press.

Cohen, Stanley & Young, Jock (1981). The manufacture of news. Deviance, social problems and the mass media (2nd ed.). London: Sage.

Devroe, Ilse (2004). "This is not who we are": Ethnic minority audiences and their perceptions of the media. Paper presented at "Communication and Cultural Diversity" Conference, Barcelona. 24-27 May 2004.

D’Haenens, Leen (2003). ICT in multicultural society. Gazzette, 65(4-5), 401-421.

Dixon, Travis L. & Linz, Daniel (2000). Overrepresentation and underrepresentation of African Americans and Latinos as lawbreakers on television news. Journal of Communication, 50(2), 131-154.

Ebel, Marianne & Fiala, Pierre (1983). Sous Le Consensus, La Xénofobie: Paroles, Arguments, Contextes (1961-1981). Lausanne: Institut de Science Politique.

Fairclough, Norman (1991). Discourse and social change. Cambridge: Polity Press.

Fowler, Roger, Hodge, Bob, Kress, Gunther & Trew, Tony (1979). Language and control. London: Routledge.

Galtung, John & Ruge, Mary (1965). The structure of foreign news: the presentation of the Congo, Cuba and Cyprus Crises in four foreign newspapers. Journal Of International Peace Research, 1, 64-90.

Gans, Herbert (1980). Deciding what’s news. A study of CBS Evening News, NBC Nightly News, Newsweek and Time. New York: Pantheon Books.

Garofalo, James (1981). Crime and the mass media: A selective review of research. Journal of Research in Crime and Delinquency, 18(2), 319-350.

Goode, Erich & Ben-Yehuda, Nachman (1994). Moral panics. The social construction of deviance. Cambridge, MA: Blackwell.

Gordon, Paul & Rosenberg, David (1989). Daily racism. The press and black people in Britain. London: Runnymede Trust.

Greenberg, Bradley S. & Mazingo, Sherrie L. (1976). Racial issues in mass media institutions. In Phyllis A. Katz (Ed.), Towards the elimination of racism (pp.309-340). New York: Pergamon Press.

Hall, Stuart, Critcher, Chas, Jefferson, Tony & Clarke, John (1978). Policing the crisis: Mugging, the state and law and order. London: MacMillan.

Hargreaves, Alec (1996). A deviant construction: the French media and the "Banlieues". New Community, 22(4), 607-18.

Hartmann, Paul & Husband, Charles (1974). Racism and the mass media. London: Davis-Poynter.

Holland, Patricia (1981). The new cross fire and the popular press. Multi-racial Education, 9(3), 61-80.

Hufker, Brian & Cavender, Gray (1990). From freedom flotilla to America’s burden: The social construction of the Mariel immigrants. The Sociological Quarterly, 31(2), 321-335.

Hunt, Arnold (1997). "Moral panic" and moral language in the media. British Journal of Sociology, 48(4), 629-48.

Hussain, Mustafa (2002). Denmark. In Jessika ter Wal (Ed.), Racism and cultural diversity in the mass media (pp. 395-420). Vienna: European Monitoring Centre on Racism and Xenophobia.

Keith, Michael (1995). Making the street visible: Placing racial violence in context. New Community, 21(4), 551-65.

Knopf, Terry A. (1975). Rumors, race and riots. New Brunswick, NJ: Transaction Books.

Koopmans, Ruud (1996). Explaining the rise of racist and extreme right violence in Western Europe: Grievances or opportunities? European Journal of Political Research, 30(2), 185-216.

Law, Ian (1997). Exposing racism in the news. Paper presented at International Sociological Association Conference, University of Bologna, December 1997.

Ley, David (1974). The black inner city as frontier outpost: Images and behaviour of a Philadelphia neighborhood. Washington, DC: Association of American Geographers.

Maneri, Marcello (1995). Stampa quotidiana e senso comune nella costruzione sociale dell’immigrato. Doctoral dissertation. Trento: Facoltà di Sociologia.

Maneri, Marcello (1998a). Lo straniero consensuale. La devianza degli immigrati come circolarità di pratiche e discorsi. In Alessandro Dal Lago (Ed.), Lo straniero e il nemico. Materiali per l’etnografia contemporanea (pp.236-272). Genova: Costa & Nolan.

Maneri, Marcello (1998b). Immigrati e classi pericolose. Lo statuto dell’Extracomunitario nella stampa quotidiana. In Marcella Delle Donne (Ed.), Immigrazione-integrazione: stereotipi e pregiudizi nelle relazioni etniche (pp.479-489). Roma: Edizioni Lavoro.

Maneri, Marcello (2001). Il panico morale come dispositivo di trasformazione dell’insicurezza. Rassegna Italiana di Sociologia, 1, 5-40.

Marletti, Carlo (1989). Mass Media e Razzismo in Italia. Democrazia e Diritto, 19(6), 107-26.

Marsh, Harry L. (1991). A comparative analysis of crime coverage in newspapers in the United States and other countries from 1960-1989: A review of the literature. Journal of Criminal Justice, 19(1), 67-80.

Merten, Klaus (1986). Das Bild der Auslaender in der deutschen Presse. Frankfurt A.M.: Dagyeli.

Miles, Robert (1984). The riots of 1958: Notes on the ideological construction of "race relations" as a political issue in Britain. Immigrants and Minorities, 3(3), 252-75.

Murdock, Graham (1984). Reporting the riots: Images and impacts. In John Benyon (Ed.), Scarman and after (pp.73-95). Oxford: Pergamon.

Olzak, Susan & Olivier, Johan L. (1999). Comparative methods and event analysis: Black civil rights protest in the United States and South Africa. In Dieter Rucht, Ruud Koopmans & Friedhelm Neidhardt (Eds.), Acts of dissent: New developments of the study of protest (pp.253-284). Lanham MD: Rowan & Littlefield.

Poole, Elisabeth (2001). Interpreting Islam: British Muslims and the British press. In Karen Ross & Peter Playdon (Eds.), Black marks: Minority ethnic audiences and media (pp.67-86). Aldershot: Ashgate.

Ross, Karen (2000). In whose image? TV criticism and Black minority viewers. In Simon Cottle (Ed.), Ethnic minorities and the media (pp.133-148). Buckingham: Open University Press.

Saccà, Flaminia & Marinelli, Alberto (1997). Più di uno al giorno. Atti di violenza contro gli stranieri nel corso del 1996: analisi di 20 quotidiani [More than once a day. Acts of violence against foreigners in the year 1996: analysis of 20 newspapers]. Università degli Studi di Roma, Dipartimento di Sociologia (Research report).

Sacco, Vincent F. (1995). Media constructions of crime. The Annals of the American Academy of Political and Social Science, 539, 141-154.

Schary, Dore (1969). The mass media and prejudice. In Charles Y. Glock & Ellen Siegelman (Eds.), Prejudice USA (pp.96-111). New York: Praeger.

Scheufele, Dietram A. (1999). Framing as a theory of media effects. Journal of Communication, Winter, 103-122.

Spears, George, Seydegart, Kasia & Gallagher, Margaret (2001). Who makes the news?—Global media monitoring project 2000. London: World Association for Christian Communication.

Sprenger, Mirjam & de Vree, Frank (2004). Schuivende grenzen, de vrijheid van de journalist in een veranderend medialandschap [Shifting boundaries, the freedom of the journalist in a changing media landscape]. Amsterdam: Nederlandse Vereniging Journalisten.

Statham, Paul (2002). United Kingdom. In Jessika ter Wal (Ed.), Racism and cultural diversity in the mass media (pp.395-420). Vienna: European Monitoring Centre on Racism and Xenophobia.

Ter Wal, Jessika (1991). Il linguaggio del pregiudizio etnico [The language of ethnic prejudice]. Politica Ed Economia, (4), 33-48.

Ter Wal, Jessika (1996). The social representation of immigrants: the Pantanella issue in the pages of La Repubblica. New Community, 22(1), 39-66.

Ter Wal, Jessika (1997). The reproduction of ethnic prejudice and racism through policy and news discourse on immigration. The Italian case (1988-1992). Doctoral Dissertation. Florence: European University Institute.

Ter Wal, Jessika (Ed.) (2002). Racism and cultural diversity in the mass media. An overview of research and examples of good practice in the EU Member States, 1995-2000. Vienna: European Monitoring Centre on Racism and Xenophobia. Available at http://www.eumc.eu.int/ [accessed: May,16, 2005].

Ter Wal, Jessika (2004). European day of media monitoring. Quantitative analysis of daily press and TV contents in the 15 EU member states. Utrecht: On-Line More Colour in the Media and Utrecht University. Available at http://www.multicultural.net/ [accessed: May,16, 2005].

Ter Wal, Jessika, d’Haenens, Leen & Koeman, Joyce (2004). Gekleurde Media? Representatie van etnische minderheden in Europese en Nederlandse dagbladen [Coloured media? The representation of ethnic minorities in European and Dutch dailies]. Paper presented at Etmaal voor de Communicatiewetenschap, Universiteit Twente, 17-18 November 2004.

Troyna, Barry (1981). Public awareness and the media: A study of reporting on race. London: Commission for Racial Equality.

Tuchman, Gaye (1972). Objectivity as strategic ritual: an examination of newsmen’s notion of objectivity. American Journal Of Sociology, LXXVII, 660-679.

Tuchman, Gaye (1978). Making news: A study in the construction of reality. New York: Free Press.

Van Dijk, Teun A. (1988a). News analysis. Case studies of international and national news in the press. Hillsdale, NJ: Erlbaum.

Van Dijk, Teun A. (1988c). How "they" hit the headlines: Ethnic minorities in the press. In Geneva Smitherman-Donaldson & Teun A. van Dijk (Eds.), Discourse and discrimination (pp.221-262). Detroit: Wayne State University Press.

Van Dijk, Teun A. & Kintsch, Walter (1983). Strategies of discourse comprehension. New York: Academic Press.

Van Leeuwen, Theo (1995). Representing social action. Discourse & Society, 6(1), 81-106.

Van Leeuwen, Theo (1996). The representation of social actors. In Carmen Rosa Caldas-Coulthard & Malcolm Coulthard (Eds.), Texts and practices. Readings in critical discourse analysis (pp.32-70). London & New York: Routledge.

Van Leeuwen, Theo (2000). Visual racism. In Martin Reisigl & Ruth Wodak (Eds.), The semiotics of racism. Approaches in critical discourse analysis (pp.333-350). Vienna: Passagen Verlag.

VanSlike Turk, Judy, Richstad, Jim, Bryson Jr., Robert & Johnson, Sammye (1989). Hispanic Americans in the news in two Southwestern cities. Journalism Quarterly, 66, 107-113.

Verkuyten, Maykel (1997). Redelijk racisme. Gesprekken over allochtonen in oude stadswijken [Reasonable racism. Conversations about immigrants in old inner cities]. Amsterdam: UP.

Wetherell, Margaret (1988). Positioning and interpretative repertoires: conversation analysis and post-structuralism in dialogue. Discourse and Society, 9(3), 387-423.

Wetherell, Margaret & Potter, Jonathan (1993). Mapping the language of racism. London: Polity Press.

Wilson, Clint C. (2002). The paradox of African American journalists. In Simon Cottle (Ed.), Ethnic minorities and the media (pp.85-99). Buckingham: Open University Press.

Wilson, Clint C. & Gutierrez, Felix (1985). Minorities and the media. Beverly Hills, CA: Sage.

Wilson, Clint, Gutiérrez, Felix & Chao, Lena M. (2003). Racism, sexism and the media. The rise of class communication in multicultural America (3d ed.). Thousand Oaks: Sage.

Windisch, Ulli (1990). Speech and reasoning in everyday life. Cambridge: Cambridge U.P/ Paris: Maison des Sciences de l’Homme.

Wodak, Ruth, Matouschek, Bernd & Januschek, Franz (1993). Österreichs Einstellungen zu seinen ostmitteleuropäischen Nachbarn. Studien zum fremdenfeindlichen öffentlichen Diskurs Österreichs während und nach der "Wende" Von 1989. Wien: Österreichischen Bundesministerium für Wissenschaft und Forschung.

Dipartimento di Sociologia e Ricerca Sociale
Università di Milano-Bicocca
Via Bicocca degli Arcimboldi 8
20126 Milano, Italy

Jessika TER WAL is Post-doctoral Fellow at the European Research Centre on Migration and Ethnic Relations.

European Research Centre on Migration and Ethnic Relations
Faculty of Social Sciences, Utrecht University
Heidelberglaan 2
3584 CS Utrecht, The Netherlands

Maneri, Marcello & ter Wal, Jessika (2005). The Criminalisation of Ethnic Groups: An Issue for Media Analysis [75 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(3), Art. 9, http://nbn-resolving.de/urn:nbn:de:0114-fqs050391.

Forum Qualitative Sozialforschung / Forum: Qualitative Social Research (FQS)

ISSN 1438-5627

Creative Commons Attribution 4.0 International License