Vous êtes sur la page 1sur 32

Social Forces, University of North Carolina Press

Framing the French Riots: A Comparative Study of Frame Variation Author(s): David A. Snow, Rens Vliegenthart, Catherine Corrigall-Brown Reviewed work(s): Source: Social Forces, Vol. 86, No. 2 (Dec., 2007), pp. 385-415 Published by: Oxford University Press Stable URL: http://www.jstor.org/stable/20430747 . Accessed: 01/02/2012 08:15
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Oxford University Press and Social Forces, University of North Carolina Press are collaborating with JSTOR to digitize, preserve and extend access to Social Forces.

http://www.jstor.org

Framing theFrench Riots: A Comparative Study of Frame Variation


David A. Snow,University California,Irvine of Rens Vliegenthart, University Amsterdam of Columbia CatherineCorrigall-Brown, University British of In an attempttoadvance understanding variationand offrame we thataccountfor it, conducta comparative thefactors study were frameddiagnostically Fall 2005 French "riots" of how the We activities acrossa and prognostically. examine theseframing diversesetofactorsand assess therole ideological, of contextual, attributional and temporalfactors hypothesized accountfor to theobservedvariation.The data comefroma content analysis of articleson theFrench riots that appeared in newspapers from a half dozen countries during theperiod inwhich the riotsoccurred. Our findings, based primarilyon variance and regression for analyses, revealvaried support our hypotheses, the and suggest theoretical analyticalutility examiningframe of variationbeyond the French riots, and raisequestionsthatcall forfurther empiricalinquiry regardingframingprocesses. introduced into the social sciences by The frame concept was Gregory Bateson in 1955 and elaborated nearly 20 years later by ErvingGoffman inFrame Analysis (1974). But the concept stimulated littletheorizing or research until the mid-1980s when it formed the cornerstone fora framingperspective on social movements (Gamson et al. 1982; Snow et al. 1986) and was found to be of conceptual utility in research on political communication (Entman 1993). Within both contexts, research has focused primarilyon the identification ofmovement- or event-relevant frames and theireffects.1 While both lines of research have contributed significantlyto understanding the dynamics of social movements (Benford and Snow 2000; Snow 2004) and political communication, particularly in the context of the print media (Scheufele 1999),we know little about the factors that account for variation in frames, particularly with respect to the same event, object or issue. Additionally, there has been relativelylittle comparative research on framing.
We would like to thank members of theUCI Sociology Department's Social Movement & Social Justice was initiallydiscussed and Cas Mudde Workshop, inwhich the_paper

comments. Direct to David A. Snow, fortheir helpful Correspondence Department of 92697. E-mail:dsnow@uci.edu. Sociology, UniversityCalifornia, of Irvine,
? The University NorthCarolina of Press Social Forces,Volume 86, Number 2,December 2007

386 . Social Forces Volume 86,Number 2 . December 2007

In this paper,we address both lacunae by conducting a comparative study of how differentactors framed the Fall 2005 French riots in 12 newspapers in six countries over a three-week period, Oct. 27 through Nov. 18.2 An overview of the riots is followed by a discussion of the relevant framingissues and theoretical considerations and hypotheses. Data base and sampling considerations are discussed along with coding procedures, the operationalization of variables and analytic procedures. We then review our findingsand their implications for a more general understanding of framing processes. The FrenchRiots of 2005 In the early evening of Oct. 27, 2005, three teenagers climbed the wall of a high-voltage electrical substation ina rundown section of the northern Paris banlieue (suburb) called Clichy-sous-Bois, populated mainly by first-and second-generation immigrants from Northern Africa. The youths reportedly jumped the substation wall in order to take refuge from the police, who they thought were chasing them for fleeing a police identity inspection. Whatever the precise motivation for tryingto hide in the substation, two of the teenagers was injured.They were discovered were electrocuted and the third other shortlyafterwardswhen the police noticed, while interrogating youth theyhad brought back to a station house forsuspected burglary of a nearby construction site, that the station computer screens had blacked out and that therewere nearby power failures. News of thedeaths was quicklydisseminated, promptingsome Clichy sous-Bois youth to take to the streets to burn cars and engage invarious motion what Time described as acts of vandalism.This reportedlyset in "a rolling wave of nightly clashes between young Arabs and French riot police that leapfrogged across the suburbs of Paris," eventually reaching "as fareast as Dijon and south toMarseilles." (Graff 2005:37) The rioting unfolded over the course of 20 nights,ending on Nov. 17when the police "declared a returnto a normal situation throughoutFrance, saying that the 98 vehicles torched the previous night corresponded to the usual resulted in2,888 arrests, average." (Wikipedia2005:1) All told, the rioting 126 injured police and 8,973 burned vehicles (Wikipedia2005:1). Not only did the riots attractworld-wide media attention, but that attentionwent well beyond descriptive reportage to consideration of wave questions about causes and solutions, such as: What prompted this be of civilunrest inFrance? On whom orwhat can it blamed?What can be done to prevent its reoccurrence?And who is responsible forremedying the underlyingproblems? In short,much of the discussion, particularly with particularemphasis in the print media, involved framingthe riots,

Framing theFrenchRiots . 387

on what is referredto as diagnostic and prognostic framing within the framing perspective on social movements. Theoretical Issues and Orientation

from Goffman's Frame Analysis (1974), is rooted in the symbolic interactionist principle that meanings do not automatically attach themselves to the objects, events or experiences we encounter. They arise, instead, through interpretiveprocesses mediated by various contextual factors. Applied to all varieties of social phenomena, the idea of framingproblematizes themeanings associated with relevant events, activities, places and actors, suggesting that those meanings are variably contestable and negotiable and thus open to debate and differential interpretation.The product of framing processes are interpretiveframes,which function, likepicture frames, to focus attention by bracketing and punctuating what in our sensual field is relevant and irrelevant, what is "in-frame" and "out-of-frame." They also function as articulation mechanisms by linking together the highlighted elements of the event or setting such that one set of meanings rather than another is conveyed. And they sometimes which perform a transformativefunction by reconstituting theway in are understood as relating to each other, some objects of attention as in the transformationof everydaymisfortunes into injustices in the context of social movements. By suggesting that our action towards things is partly contingent on how they are framed, these functions also highlight the importance of understanding the determinants of framevariation:what is done in relation to events like riotsdepends in part on the various ways they are framed and the relative salience of one framingvis-a-vis others. Frames that define events as problematic and in need of control or repair, inthe context of social movements and events such as the riots, as tasks of "diagnostic," have been characterized intermsof thecore framing "prognostic" and "motivational" framing(Snow and Benford 1988). The as former entails a diagnosis of some event or aspect of life troublesome in need of change, and the attributionof blame for the problem. and Prognostic framinginvolves the articulationof a solution to the problem, includinga plan of attack and frame-consistenttactics forcarryingitout. Research has shown that both diagnostic and prognostic framingcan generate considerable debate resulting in "frame disputes." (Benford addresses the 1993) The finalcore framingtask,motivational framing, a problem by articulating rationale(s)for engaging incorrective "free-rider" activity(Snow and Benford 1988).

This borrowed occurrences subjectively meaningful. conceptualization,

work that renders events and Framing refers to interpretive, signifying

388 . Social Forces Volume 86,Number 2 . December 2007

of frames (Cadena-Roa 2002; McCaffrey and Keys 2000; Rohlinger 2002), and on framingeffects or outcomes (Cress and Snow 2000; Ferree et al. 2002; McCammon 2001). The scant research on framevariation has generally focused on changes intheway an issue ormovement is framed fromone point in time to another (Berbrier1998; Ellingson 1995; Ferree et al. 2002), with even less attentiondevoted to variation in framingthe same event across different actors. This research void is partly due to the dearth of comparative framingresearch. There are a few exceptions, but generally the comparisons have been across states, cities ormovement organizationswithin thesame country(Cress and Snow 2000; McCammon 2001), or, ifcross-national, across only a couple of countries (Dimitrova and Stromback 2005; Ferree et al. 2002). There is a good deal of research that examines the linkbetween movements and themedia, a relationship thatGamson and Wolfsfield (1993) call "transactional." Much of this research bears directly on framing, dating back toGitlin's (1980) examination of the consequences ofmedia coverage of theNew Left. But like many framing studies, Gitlin's focus was on the framingeffects of media coverage. Other research has examined how ideologically opposed movement organizations have strategicallyconstructed media frames to garner coverage during critical moments in the abortion debate, and how those framingefforts are affected by factors such as organizational structure,organizational media discursive opportunitystructures and the standing of an identity, SMO (Ferree et al. 2002; Koopmans 2004; Rohlinger 2002). But none of the studies working at the intersection of social movements and themedia directlyexamine the factors that account for framevariation actors. among more than two different The same lacunae also hold forresearch inpoliticalcommunication. For example, ina widely cited reviewarticle,Scheufele (1999:109) states that

on factors affect development production that mobilizing the frames, the or

As we suggested earlier, the newspapers examined engaged principally in framingthe French riotsdiagnostically and prognostically. Consequently, we focus on these two core framingactivitieswith the the descriptive objective of identifying range of diagnostic and prognostic frames elaborated by different actors in the newspapers examined, and the analyticobjective of accounting forvariation inthese frames. To date, systematic examination of the factors thataccount forframe variations is rare among framingstudies in the two substantive areas in which they are most prominent: the study of social movements and political communication.Within the former, there has been a plethora of framingresearch (fora summary,see Benford and Snow 2000, and Snow 2004), but the research has generally focused on the identification of

Framing theFrenchRiots . 389

"no evidence has yet been systematicallycollected about how various Various qualities innews intermsof framing." factors impact the structural more (cross-national) comparative research to scholars have called for fillthis gap (Benson 2004; Esser and Pfetsch 2004). Probably themost noteworthysystematic cross-national research has focused on selected European Union related issues (De Vreese et al. 2001; De Vreese and work takes Semetko 2004; Semetko and Valkenburg 2000). However, this a verybroad approach using generic frames at the levelof the newspaper actors within article,but does not look at divergent framingby different there ismore interest inthe effects of news the news item.Furthermore, on framing public attitudes and perceptions thanon the factorsunderlying cross-national or cross-time differences. Our analytic objective in this paper, then, is to begin to fillthe void in the framingliterature regardingthe conditions thataccount forframe variation.Four sets of theoreticalconsiderations guide our investigation. The first is ideological.The importanceof ideology inrelationto political set orientation and action has long been accented by students of culture, Marx and Engels politicalprocesses and social movements, rangingfrom (1970) and Mannhein (1936) toGeertz (1973) and Zald (2000). In relation the importanceof government ideological position for tomedia framing, newspaper framingis suggested by findingsshowing journalists' reliance on institutional, political resources, ofwhich the government is generally a themost important, phenomenon that is labeled "indexing." (Bennett 1990) This selection bias is also found in protest event research that from institutional findsnewspaper description bias favoringinformation resources (Smith et al. 2001). We thus expect the political ideology of both country and newspaper to shape the framingprocesses. At the are who are politicallyright micro level, ithas been shown that individuals and that such deference is a key featureof more deferential to authority ideology (Altemeyer1988). For this reason,we hypothesize that in rightist countries thatare under rightist governments or innewspapers that are therewill be a propensity to use state actors as sources politicallyright, attributions(blaminggroups such and invokenon-structural/group-based or as immigrants youth). Furthermore, Budge et al. (2001) show that a rightist political ideology is associated with a greater tendency to focus we hypothesize that incountries solutions.Accordingly, on law-and-order and newspapers skewed to the righttherewill be more focus on short termsolutions, such as law and order. The second set of theoretical considerations concern relevant outcomes. Contextual trends contextual factors that can affect framing such as unemployment and migration rates have long been posited as correlates of riots (Feagin and Hahn 1973) and have been speculatively

390 . Social Forces Volume 86,Number 2 . December 2007

associated with theFrench riots. We examine these connections alongwith another set of contextual factorsderivingfrom observations regardingthe relationship between event proximity and the relativesalience of the event or issue to different sets of actors. Media research suggests that event proximityis an important news value and is likely increase the chance to that the event will be covered and discussed in themedia (Galtung and Ruge 1965). We thushypothesize thatas distance fromthe riotsincreases, the salience of the issue will decrease. In addition,we hypothesize that countries economically and politically more distant fromthe riotsaremore to likely engage indiagnostic framing(as opposed to prognostic framing) because such distance decreases the likelihoodof being affected by the event and thereforereduces interestinthe search forsolutions. The third orienting theoretical consideration is suggested by the attribution perspective insocial psychology (Jones et al. 1972; Ross et al. 1977).Of particularrelevance is thegeneral proposition thattheattribution of responsibilityfor events varies by, among other things, the actor's relationshipto the event.Although attribution was developed with theory individuals in mind,we examine thisproposition and the extent towhich itapplies to collective actors. Specifically, we hypothesize thatcountries less politically and economically proximate to the riots will be more likely to blame the state for the event. This hypothesis is consistent with the fundamental attributionerror, which posits that individuals confronted with negative events aremore likely blame contextual factors; to whereas those seeking to understand negative events happening to others are more likelyto blame the affected individuals (Ross 1977). Extrapolating to the collective level,we hypothesize that the French should blame contextual factors for the riots and other countries should blame the French government itself.Yet, attribution theorywould suggest that French state officialswould be selective about the contextual factors we hypothesize thatstate actors will be targeted forblame. Accordingly, more likely blame the riotson affected groups, such as immigrants to and youth, and less likely see themselves or theirpolicies as responsible. to other actors, such as opposition leaders, riotparticipants Alternatively, and international will be more likely blame the state. to officials, The finalorienting theoreticalconsideration is temporality. That "time matters," as Abbot (2001) argues, holds for the analysis of any social process or sequence, includingframing processes, particularly when the focal event extends over time, as with the riots. Assuming that frames change over timewith changes in the focal events or the competition of other events, we hypothesize that early in the career of an event the will be limited frame repertoire due to the noveltyof the issue. However, we hypothesize that during the escalation of the event therewill be a

Framing theFrenchRiots . 391

proliferation frames as the field of actors expands, with each actor of attempting tomake sense of the riots.Finally, the event becomes less as novel or begins to dissipate, we posit a crystallizationof frames,with some levelof consensus being reached. Inaddition, we hypothesize that the percentage of framing devoted to diagnosis will shiftover time, with less diagnostic framinginthe later weeks of the riotas the identification of solutions becomes the focal concern. Finally, earlier in the riots,there should be more short-term solutions offeredwhich will be replaced by long termsolutions in later weeks. Data, Procedures andMethods Data Source and Sample Selection The data are derived from a content analysis of 418 articles on the French riots that appeared in 12 newspapers in six countries over a three-week period. We focus our analysis on this period because it covers the duration of the rioting,from its inception on Oct. 27 to officialpolice declaration of itscessation on Nov. 17, 1 days after the 0 had peaked on Nov. 7. Figure 1graphs the evolution and decline rioting of the rioting, measured interms of the numbers of burned vehicles as and arrests, and the total event framings in the 12 newspapers. We chose to examine Canada, France,Germany, theNetherlands, the United Kingdom, and the United States based on three considerations. First, inaddition to Francewe wanted several of the countries to be within theEuropean Union.Additionally, wanted several of thecountries to be we outside of theEuropean Union butwith sufficientlongstandinghistorical and political ties to France to ensure more than passing reportage.And third,since our database contained foreign language newspapers, we were constrained by our language repertoireof Dutch, English, French and German. These three considerations mandated that we select our fromthe six countries listedabove. newspapers The selection of the two newspapers fromeach country was guided we by threeconsiderations. First, wanted one of thepapers to be national in scope and one more proximate and oriented to the national capital. We thought thatcapital-orientednewspapers might be more inclined to frameevents inaccord with state interests. Second, we were interested in selecting papers skewed inopposite directions on the Left/Right political continuum.And third,the selection of newspapers was constrained by their availability for inspection. LexisNexis was our newspaper search engine, and thus limited selection to those newspapers indexed by LexisNexis thatalso satisfied the two preceding criteria.

392 . Social Forces Volume 86,Number 2 . December 2007

Figure 1.Attention toFrenchRiots and Riot Intensity 1600 250

CD c) 1200k E 1000:
800U)~~~~~~~~~~~~~~~~~~~U

1400 -

\i--- 200 200 liii \


EI

150,
100

6 400-:_

t\
/ I N

<

100
~~~~~~~~~50

200~

0 --- VehiclesBurnt-

Arrests Made

Total Framings

Article Selection and Coding The data used for the construction of our dependent variables (framing characteristics) come from a content analysis of relevant articles in each of the 12 newspapers. Articles containing both the words "riot"or "riots" and "France" or "French," or the equivalents for each respective language, during the research period were selected from LexisNexis. Each of these articles was coded when 1.) the articlementioned the riots and 2.) contained at least one diagnostic or prognostic framingelement. Inorder to ensure intercoderreliability among the threeauthors,who we did all the coding, a number of procedures were followed. First, coded a small sample of the same-language newspaper articles inorder to reach consensus on an initial of possible diagnostic and prognostic framing set categories. We then refinedthose categories throughgroup discussion. Afterestablishing the final coding scheme, we each coded a small sample of theEnglish newspaper articles forthepurpose of calculating thedegree Formula (1 of intercoder-reliability Holsti's Intercoder using Reliability 969).3 between each of the coders was more than .75. Overall reliability

Framing theFrenchRiots . 393

Table 1: Sample Newspapers and Number ofArtides Newspaper Ottawa Citizen National Post Liberation Figaro Suddeutsche Zeitung Berliner Morgenpost Parool Telegraaf Guardian London Times New Times York Washington Times Total Country Canada Canada France France Germany Germany Netherlands Netherlands UK UK US US Political position left right left right left right left right left right left right Orientation Capital National National Capital National Capital Capital National National Capital National Capital Number of articles 25 19 37 82 55 24 28 22 33 40 27 26 418

A source was coded foreach framing element. The source could be the journalist who wrote thearticleor any other actorwho directly indirectly or addressed the riots in diagnostic or prognostic terms.We clustered sources intoseven categories: media; members of French government Affairs;4 except Nicolas Sarkozy, the FrenchMinister of Internal Sarkozy; French opposition; international actors; residents/participants; and other miscellaneous actors. The presence of a diagnostic framingelement was indicatedwhen one of the above sources made a statement that addressed the character of the problem. For the presentation of our we cluster the diagnostic elements intosixmajor categories, the results, firstthree being non-structural categories or social groupings, and the second threebeing structural: * Riff-raff: riots are senseless, criminal acts of violence rather than rooted insubstantial structuralproblems or politics.5 * Ethnic and religious minority groups: riots are rooted in the character or culture of the ethnic or religious minorities that inhabit the suburbs. * Over-reaction of authorities: riots are associated with the behavior of the police and government officials, especially Sarkozy, who referred to the rioters as "scum." (New York Times, Nov. 10, 2005) * Failure of are minorityincorporation:riots attributedto a perceived failure the incorporation minorities intoFrench society. of of

394 . Social Forces Volume 86,Number 2 *December 2007

* Economy/education: riotsare attributed toeconomic conditions and associated high levels of unemployment and/or limited educational opportunities among suburban residents. * Housing: riots are an outgrowth of the "miserable condition" of suburban housing. A total of 825 diagnostic framingelements were coded. Additionally, we coded for the attribution responsibilityfor the event to a certain of actor (problem attribution). The followingactors were distinguished: the French state ingeneral, the currentgovernment,youth/criminals, police, immigrants,terrorists,religious groups, parents, and others. A total of were coded. 628 problem attributions For coding of prognostic framing elements, which deal with the question of what needs to be done, we differentiate among six major

prognostic frames:

* Law and order: restore order by suppressing the riots, if necessary with tough action by the police or military. * Action program: address underlying social and economic problems, especially by targetingminority unemployment and discrimination in the labormarket. * Better housing: improve poor housing conditions inthe French * Limit immigration: decrease the number of immigrants and close the border to certain groups of foreigners. * Raising children: raise and educate children in the suburbs to be more responsible citizens. * Dialogue: calls for talk among residents, participants, government officials, and the police to facilitate mutual

suburbs.

understanding.

We coded 504 prognostic framingelements. We also coded, when mentioned, the actor that is, according to the source, responsible for the solution. We distinguished seven different responsible actors: government, police, parents, immigrants in general, participants in the riots, social welfare agencies, and others. A responsible actor was mentioned in446 cases. Dependent Variables We are interested in 1.) differences in the salience of the issue, 2.) the use of various sources across contexts and time, and 3.) differences

Framing theFrenchRiots . 395

in the way the issue is framed across contexts (newspapers and countries), sources and time. For cross-context and cross-time comparisons, the research period is divided intothreeseparateweeks. The unitofanalysis iseverycombination of newspaper and week. Salience of the riots is operationalized in line with the operationalization of visibility social movement organizations of by Vliegenthart et al. (2005). They introduce a formula,where the visibility-scoreis sub-linearly dependent upon the number of times the whether this is in the headline organization ismentioned in the article, or body of the text, the placement of the articlewithin the newspaper (front page or not), and the circulation of the newspaper. We adjusted thisformulaslightlytomake itsuitable forour data and interest in making cross-newspaper comparisons. Instead of lookingat thenumber of times the issue ismentioned, we used the number of framing elements (both diagnostic and prognostic)within a certain article and did not distinguish between headline and body of article, as it was extremely rare that the headline included a separate framing element. Furthermore, when making a comparison between the content of newspapers, circulation is not relevant. However, the size of the newspaper interms of the total article contentwas taken intoaccount inorder tomake a comparison between the relativeattention given to the issues by the newspapers. Thus we included in the formulaa coefficientthatcontrolled forthe totalnumber of articles ineach newspaper during a random day halfway throughthe research period (Tuesday,November 8). These considerations result in the followingoperationalization: = 12 * (Eq. 1) visibility,0ots log(2 f(framings))* npaper fp where f(framings)is the total framingelements inan article,npaper the is coefficientcontrollingfornewspaper size, and fp the place of the article in the newspaper (front page is weighted as 2; other pages as 1). To determine the relative use of thevarious sources, we calculated the percentage of each source used inevery week/newspaper-combination. Fortheframing differencesacross contexts (newspapers and countries), sources, and time, the unit of analysis is each present combination of also encompasses country),timeand source. newspaper (that indirectly * Percentage non-structural/group-related diagnosis of total diagnostic framing:Diagnostic framing elements are divided into two categories: one encompassing non-structural/ that include riff-raff group-related diagnoses frames, the over-reaction of authorities and ethnic or religious minority

396 . Social Forces Volume 86,Number 2 . December 2007

and/or immigrantgroups; the other encompassing structural diagnostic framingpointing to deeper problems within society, including the perceived failure of the society's incorporation policies, economy/education problems, and housing problems attributed to the state or current government. For the "other" category, each separate framingelement is classified intoone of the two categories. * Percentage state-attributionof total attribution: Actor-based attributionsare sorted into two categories: one encompassing the authorities (the state, the current French government and police); the other including all other actors. * Percentage short termprognosis of totalprognosis: Prognostic framingelements fall into two categories: one encompassing short term solutions aimed at ending the rioting,captured by law and order framing; the other including all other framings with a long term prognosis. * Percentage state-responsibility of total responsibility: This

between authorities, such as category differentiates

government, police and social welfare agencies, and all other actors. * Percentage look at the diagnostic crystallization: We convergence of framing towards short-term/group-based diagnostic or structural framing. For each unit of analysis the difference between the relative uses of the two categories is calculated. * Percentage prognostic crystallization: We look at the convergence of framing towards short-term or long-term prognostic framing. For each unit of analysis the difference between the relativeuses of the two categories is calculated. * Percentage of diagnostic framing of total framing: The percentage of the total framing that is diagnostic. IndependentVariables we identify Through analysis of the newspaper articles on the riots, the range of frames elaborated and seek to account forframevariation by examining, inaccordance with the previously discussed theoretical considerations and hypotheses, the relative influence of country/ contextual variables, newspaper characteristics, temporal variation and framingsource. The contextual/country level variables included in this analysis are rates of unemployment and immigration,treated as control variables,6 and political and economic proximityto the event. An index of political

Framing theFrenchRiots . 397

and economic proximity was created by combining three separate measures. First, the voting behavior of the each of the countries during theUnited Nation's 60th General Assembly was examined. A correlation between each country's votes and France's votes was then calculated to assess the level of political congruence. Second, trade relations were examined, looking at the percentage of total imports fromFrance by each of the sample countries. This was a proxy foreconomic proximity. Finally, European Union membership was coded: members of the EU and part of the euro-zone (1), EU members but not part of the euro-zone (0.5) and neither (0). Finally, the political orientation of the actors examined (nations, newspapers and sources) and how it which theyframe affects theway in the French riots is considered. Political ideology is coded according to the self-proclaimed political positioning of the political party or parties currentlyingovernment as left(-1),center (0) or right(1). Analysis The analyses are conducted inseveral steps, primarily using variance and regression analyses.7 For each of the independent variables, a model is conducted forthe various clusters of univariate general linear independent variables (newspapers/countries8 and time for visibility and source selection; newspapers/countries, sources and time for framingvariation). For the latter,two-way interaction terms between the three clusters are included. These analyses show between what clusters and interaction of clusters the values of the dependent variable differsignificantly. Those clusters and interactionof clusters that showed a significant difference are included in the following ordinary least square regression. For the newspaper/country cluster, newspaper characteristics (left-right positioning and capital-national orientation) and countrycharacteristics (political and cultural proximity to France, unemployment and immigration rates)were included. For Week 1 andWeek 2 (Week time, dummy variables are employed for 3 is the reference category). In addition, dummy variables for all which is the reference differentsources were included (except "other," were significantdifferences in the interaction category).When there between clusters, additional GLM analyses established which specific variables interactand these are included in the OLS regression. After conducting the regression analysis, another GLM analysis with the residuals of the regressionas dependent variable established whether the significant differences within categories were indeed captured by the independentvariables.

398 . Social Forces Volume 86,Number 2 . December 2007

Results The findings are presented in two sets of tables. The first set, tables 2 through 5, are descriptive, showing the results by country, framingsource and week fordiagnostic framing,problem attribution, Table proposed solution and thedesignated responsible agent actor. In frame by the French government and 2, greater use of the riff-raff Sarkozy is found, consistent with our hypotheses that state actors are more likelyto associate the riotswith non-structural factors. Equally interesting is the finding that internationalactors see the problem as with the failure of minority incorporation being associated primarily (53.3 percent), while the French opposition actors see the reaction of French authorities, including the police, as being too strident and harsh (45.9 percent). In addition, the suburban residents and riot overreactive police as themost participants are more likelyto identify serious problem associated with the rioting(38.8 percent). Finally, over the three-week period there is a progressive decline inconnecting the underlying problem to specific categories of actors, such as riff-raff (from20.9 percent to 13.9 percent) and authorities (from37.3 percent to 11.1 percent) and a corresponding increase in identifying structural such as the economy and education (from 7.2 percent to factors, 16.1 percent), and the failure of minority incorporation (from 17.6 percent to 24.4 percent). In sum, Table 2 shows noteworthy variation indiagnostic framing,particularlyacross the various framingsources and over the three-week time period. Table 3 displays the results concerning attributionof blame. First, we see that the French attribute less blame to theirgovernment (12.2 percent) than the other countries, except for Canada, but attribute more blame to the youth and criminals. It is also noteworthy that the United States attributes more blame to the police than any of the other countries. We also see thatactors outside of France aremuch more likely to blame the French state (54.5 percent) than the various categories of internalactors, even including the opposition. But when we consider the currentFrench government as a source of blame, we find that the French opposition inparticular(58.3 percent) and local residents and riot to participants (29.4 percent) are most likely blame the government.The youth and "criminals"are also targeted as a major source of blame, but not uniformly the various actors. Instead, consistent with the riff-raff by problem designation, the French government and Sarkozy point to the criminalelements among suburban youth as being a significantsource of blame (37 percent and 55.1 percent respectively).Speaking for the French government, de Villepin pointed to "organized criminalnetworks backing the unrest" and to "gangs of youths, in a mindset of thugs, but

Framing theFrenchRiots . 399

.60 0 c

C\.j 0 D > _ 0D a) C;

w Cl

C -

0U

cs -t C,C)

C0

CD D 'to O cn cocs) Co0 r-- U- CD Id Cl -~ L -

csCN C%O0 C- I`-

W) o6 (6

CD 0 ) CD oi Co

N14U)CD

CO)C\ILC)

u o 0 o - IIC\i CS

,co L-

o O 6--

c0 -C C-4C~CY)

CO C-)

icili

N-L c LO L6L6 (6

C Ct r-~ -eoC6

.0
c m

CD C.,

LO -:

LO n L O f-

cs 1t

C4

"q 00

M co

LO CNr w

CD 0 o

LO c co m

CN I,- co r_ L6

~~~~~I s

C0
i;_

14 LO c.0

(U

4.C

"0q

CM

LCZ C 0 N ?i -2 CLs 11 C% 4-0 LL

.'

co

co coC) LcO Nt C-) C-DL Ci CN C1 co N

0 6

L CNt O- c6 c6 C% -

M CO M t C6 6 t- N vn o,CS) o C)

co r..

o t r_ 4 C1 CN

B 0 c

Q) LLO 0

0,, Q

c)J Q Co

COC1)tCO CCCJC~c

tC\ C

LO O CNIC

COsCO~ C\CN

-L

'

U)~~~~c

0CL-6
0

o ^~~~~~

CD
cs

co
62

co cD co cD co co LOCNa)OI) CNC vCNOI

C% O0)C O

Os-C\ 00 CDl

W CN

cs

- CDO O O m
n

cs-

c c OCO

c)i

~~ ol

~~~C )

II Cv e

C---"t 'r-CJCJCC%N O cg W E C C a

C,r-TC) 2! =

CN )C-4r U)

0~~~~~~~~~~~~~C
c c: >.C -1-.l 0 m 0 o C,, Ci)CU C co .

*1-'~

~~~~~~~0

0~~~~~~~~~~~~~C

~~~~~~~~~~~C

cu

400 . Social Forces Volume 86,Number 2 . December 2007


- 0 C-4 r-

\ o cs~~~ ii

00 .o o t

C 0, CO 0

CDO ro

m 00 O

I,- U"

co C --

> .

C P4 LO CN

) LC 04 CYO

LC CI)C)0

-C

- CC oLO o 00c

4cNi

CN O CN COLc

CN -(6

COCNO (CO~C-i

CY) *3II

c- LC) ~Cy)OCT)

C) (II C')

-1-- I SI s1s

U)

Q 000 Q (UE

ci) Ii

~~CO ,,)

,C)C' ) CYN CO~ N- ,1 ( N ,

LON

O)C) C a

oN

co c E

cs

cs

o6

C 6

(6

V-:

cli C-

cc;

oEug
.O U )_ -. - C .o ~ ~~~L _~~E IUCo |)LO(N Cl U" 8 OD "' O CN 11 t C co tc) to co OCO CO C6 )(

-~~ Co

) U")

CDO Mf

Uo CD)c~c -> LOC)C-L c' ( LO. CD s 11 U" 1 (CD 1 (N-

(D

lq

ClNO,0ClClO d _C co'n (t EE) O

-(NLOOOCO

COl T-L O)CON co 't co t 0. co C CD UO

.
~D

C |? |0cOLLO n~~~~~~m I LO cN C)l CD =~~~~~~~~~C CN CD sn ,I C i CO( ? I~JCNCD l 9 n~~~~~M N CN (6 (N (-L 6U(rL (N cn-C\(N(C(N

s o

CN m r

( > r-3 0

1-Cor-U> C N J-L) 6 66ccco " LO(N(N(N

C)N-N oC6 oiL CN(

C% -

co -I,-, |1 6 (NC4 % - 1 co (N\ a %( co | ?vi C\(N

**~

MC4 -q C')N(

-LO(NC\') ?_: aiaJ a

II

(NCN(NC\

U)

0 -. cu am

~~~~~~~~~~~~~~~~~~

00

U)~~~~c

C)C

>U

Framing theFrenchRiots . 401

also of playing a game and wanting to raise the stakes." (NationalPost, Nov. 8, 2005) Sarkozy and the Frenchopposition, on the other hand, point to the immigrantsthemselves as being the root cause of rioting(32.1 percent and 30.6 percent respectively). Finally, only the local residents and participants target the police as major source of blame, which is consistent with theirproblem designation. Throughout this section of the table, the findingsare consistent with the source/actor diagnostic problem designation in Table 2,which iswhat we would anticipate given are thatthatproblematizationand attribution both component features of diagnostic framing. Inthe last section of the table, shifts inattribution over timeare evident with a decline inblaming the government,youth and criminals,and along the police, coupled with a particularly noteworthy increase in immigrant While it is unclear what is driving these changes, they do attribution. suggest some degree of frame crystallizationin that there now appears to be waning competition among the alternativesources of blame. Table 4 speaks to prognostic framingby specifying the relationship between profferedsolutions and country,source and time. We see that France places less emphasis on a law and order solution (38 percent) than the other countries, but this appears to be compensated forby its emphasis on parental control of children (17 percent). This is consistent with some media calls for more parental intervention. noted inParis's As Le Figaro: "Parents are the firsteducators. In areas where families are involved and supported things are better [there is less violence]." (Le Figaro, Nov. 17, 2005). Consistent with the previous findings,the French government and Sarkozy call fora law-and-orderresponse (48.4 percent and 79.8 percent respectively). Sarkozy exclaimed during the second As he week of rioting, would wage "awar withoutmercy" against the rioting immigrants,including immediate "deportation" for those involved (New YorkTimes, Nov. 10, 2005). Incontrast,media journalists, local residents and riot participantsaccent the importanceof an action program thatspells of out steps to remedy the situationover time.A fifth the residents also call forgreater parental control. It is importantto note the strong French which we suspect can opposition call to curtailimmigration (27.3 percent), be attributed largelyto the voice of Jean-Marie Le Pen, President of the National FrontParty. anti-immigrant According to his youngest daughter Marine Le Pen: and political heir, "The National Front predicted and warned thisviolence would happen 20 years ago.... It has been political madness for 30 years since we allowed immigrants to come here as cheap labor at the behest of French It has been impossible to assimilate these bosses.

402 . Social Forces Volume 86,Number 2 . December 2007

LO 1 c

LOOCN nO)C\JCO O 6 O O CoC) C\lCO

~r~)N U co v- C

CN 4

O)0OCN L6 C6 LO C

a C0I 11

CN

. =O

. 0 CNO C

J CC% s N -CNJ

LO COCNJ

- COCO = s

-C O- II O C<) 0m Co tC DnOC

) t 4 W) dw

CO~ '

6 c>it

.s) v

CD

"t) C-i

c o m CD1 -4

to C\i

o OCo o) Oi

) Oi

Lo 01

LO

00
CY o ~ ~ ~

O Y IIC c0 sCi

C "' ~ I . 0t-- ocia

0n O0

H E
co 0~~

.D_II<cs

C~

(N

csN

cn

0) _
sL u) oX .) C) C:) O C??L6C> O CN t m C:l co c>JH6 O C> O LOCO W) OO) 0 COLOXC')o N2cO LCtNJ t r OCOC (cO6cv

?t rjQ

UIIC E m nsO0 O

0~~ Q 011 CO COO CD OC JcO LOCOO) Uc) OLC) C ) CO

+-k

z~

0C

ON

_ C)

c r-00--

COD 1Cl

CO

CDlN

ONC

Co o .i

Co

E c @ C1

LO

5_

c 11E 1 co ML I

C ' -

E
t -0 0 cn

-o

0-4 C: cu

o~~~ COC ~ ~ ~ ~ ~ t? (I -L6 -T OC0(J-C.0 - (OC5Y(D l R 0 CICo (1 ) 0001t I* O L 0Z

o CD 00 N C ct Docuc\c

-NC C__ (1) 0 cq C,4

L)LC)C iL6

"l

Framing theFrenchRiots . 403

people, simply because there are too many of them." (Ottawa Citizen, Nov. 13, 2005) Table 5 speaks to the second component of prognostic framing by directing attention to the actor designated as responsible for the proposed solution. Here we findagreement across the countries that the government shares much of the burden for solving the problem. There is less agreement, however, about the role of the police, with the Netherlands and United Kingdom having contrasting views. Inaddition, France assigns much importance to parents as responsible agents, twice the level of other countries, which is consistent with previously

and theFrenchopposition are evenmore vocal thanother framing sources about the government's responsibility. Also, residents/participants target parents as responsible for carryingout the proposed solutions, which is consistent with theirframingof the problem diagnostically and their of framing a solution. In the table's last section, thereare slight temporal and increased changes, especially inthe call for less police responsibility parental responsibility. Considering together the results in tables 2 through 5, there is across country, source and time.To get a noteworthy variation inframing better handle on what factorsaccount forthese framing differences, the sources and framing second set of tables analyzes issue salience, framing characteristics by lookingat contextual, newspaper, temporal and actor differencesvia regression analysis. The variables included in the regression analyses, presented in tables 6 and 7, were selected based on the outcomes of the GLM models. In general, considerable amounts of variation inthedependent variables are explained by the clusters of independentvariables. In several respects, the context in which framingtakes place matters. Table 6 shows that a decrease inproximityleads tQ a decrease in issue salience. Newspapers in the most proximate countries (Netherlands and Germany) pay almost as much attention to the issue as the French newspapers. Attention to the issue is slightlylower in theUnited States, differentfromFrance,while inboth theUnited but stillnot significantly less coverage Of the riots. Kingdom and Canada there is significantly Our expectation about the positive relationshipbetween diagnostic framing and geographical distance is also confirmed, though the coefficientonly approaches significance. Incountries further away from France the need for problem-definitionis somewhat higher.However, when geographical distance there is no increase in state attribution decreases (see Table 7).

discussed results. Sarkozy Looking attheframing source section Table5, itisevidentthat of

404 . Social Forces Volume 86,Number 2 . December 2007

(-0 -4- CD -6

11- CO C

wCY)-C c

Cn Lo

Oc C4

0N00

CO o- 0

CN

OCN% ) p

II CD

C14 C1

n t U' C NN

O Z O

0 Cu 0~) 0 C%) C Ont

0 z
O

0n

0 O 0 B t) lX .U ~CN t

to

cn

)L u7 C5)
I co a c c o O C X r P-N O) cz LDO C c ) CC O

H tn sO 00 t 0

_cn 0

C) cL

0)

Y
o

cn

~Cl)

a) wc

CL~~~~~~~~~~~~~~~o
; 0 0 )s Os Ms s a)s s O OG)

.0

0 ~~~~ o o 0

0)(0~~~~~~~ ~ ~ Cu~~~~~~~~~~~

PO~~
~ ~ ~

~ CO U" LO M
~ 00 "a )-0 Ma

0CXm

0)~~~~~~~~0L ,1 ~ c-, 00 ~~~ ~~~~~~~~~(~n o~~~~~~~~e

CV C)

o~~~~~~~~~~~~~~~~~~
0 o ~~~~~~~~~~~~~~

C1 E

00 C5 a)o

0~

-a) " ? ?c6

G) s- -CO CC)

o~

CC DCD

LO C

L6 E6 <,,', p6

C~~~~~~~~~~~~~~~~D C) . ~o0) c) )u coEC ~ 0 -I- 0 ( a 5 E'E co"c ZU : O Z~~~UD O O) N D" a u)L_ 0) c -a a) c N

I-

Framing theFrenchRiots . 405

Table 6:Ordinary Least Squares RegressionofTime, Paper and Country on Issue Salience variable Dependent Salienceof issue variables Independent -.272*** Right newspaper -.243*** National oriented newpaper Canada Germany Netherlands United Kingdom States United Week 1 Week 2 F-value model total Adjusted R-squared N -.232*** -.080 .018 -.273** -.144 -.136 .705*** 10.622 .712 36

< < reported Note: *p< .10 **p .05 ***p .01(one tailed); coefficients (betas). are regression coefficients standardized

The political orientation of newspapers and incumbentgovernments also make a difference,but only to a limitedextent. Findings indicate, newspapers report less about the riots for example, that right-leaning than left-leaningnewspapers (see Table 6) and that left incumbent governments have a positive influenceon the use of state attributions (see Table 7). But there is no evidence forother differences caused by political orientations or ideology, except for a greater reliance on the source inthe case of both a left-wing governmentand media as a framing less proximity (results not presented in tables).While we refrainedfrom formulatingexpectations about concrete differences between capital and national-oriented newspapers and used thisdistinctiononly to select an appropriate sample of newspapers foreach country,there are some and meaningful differences: national newspapers report less interesting about the issue, possibly due to the fact thatcapital-orientednewspapers paymore attention to thequestion ofwhether riotscan also occur intheir cities. The Dutch newspaper Parool (Nov. 10,2005), forexample, asks the will diffuse to theNetherlands." (see question whether "the French riots nationalnewspapers also Berliner Morgenpost, Nov. 6, 2005) Furthermore in more state-centered attributions. engage While our expectations about context-dependency are only partly confirmed, the findings presented in Table 7 support most of our

406 . Social Forces Volume 86,Number 2 . December 2007

O C) oo CM_ co rO > C

Lo

*-It o C'1*' XU)1 o ' o tO * 4' - o 4' cn m *4 CD (. oN 4' 4' 4'

LO ' f-cq o, 4' *CD 4' o' to C) '- I LoO00 co 00I n LO* ' 4' '1 v LO ' o3 04 CO o> o4' 4' 4' 4 4 '

0 O'

c cS Z5 Z5m-r 4' 4'

; 4'

c5

CN ' 4'

Cq 04 ' 4'

DI

'4

>C

0U

0 0

'

4'

4' 4*

Ut o o s O
*v' o to

~~~~~~~~~~~~co
CD

1-

.>

0)

_n CD

U7

o) C.) 0) co) 0 0 C

cCO

0)

C' oI)

0 DO

co C4'

4'

4')

LO'

L~

i
0:

a
^0 co

'M.
. M. . .

0 O4 _

0_0CO C - "q CO O, O,0~~~~~~~~~~~~~~~~

OO 11 O-

st

O OM

C> CMa)

cM~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1F O_ cn 4 Uo C) C> W CO * 1-CtCMI o o C C' t (V) v-C* 0) CO ( M1- T 1* * co

CO 0

O,t O,)<=O

cn vf
00 7C 0) CC) C

t 04 *?.

LO UDCO

C F-.-

Z0

.Ci-

sOo t(9s

= gED | > E
t.< )* c .I

co
o

.?
o

so

Co

co
CY) v0r)

0 ~~~~~~~4' 4'~~~~~~~~c 4' ~~ 0 ~~CoC\JCorC)

4' Co 1 c

E.-

0)

n
0

@ S
0)

e '
>

p 9CD S S p S M
0 E

(~O0 O
) (0 0~~~~~~~~~~~~~~~~~~~~-.0,u(C-0

cO
O 0

E co 0 )
v

0 C

C ~ 0 ~ ~ CDC U) cn c 0 4C/) 4' c c'-.i 4' 4' z

0 0)

(D * >U <:

0C~~~~~~a.0U)nc)o

Framing theFrenchRiots . 407

expectations about differences inframing different actors or sources. by Especially noteworthy is the differencebetween state actors and other actors. State actors are more inclinedto invokenon-structural diagnoses compared to other actors, especially international actors.9Here Sarkozy stands out, speaking about the riots unambiguously and stridently, as when calling the rioters"thugs" and "scum." (New YorkTimes, Nov. 5 and 10, 2005) This also leads state actors, and again Sarkozy, to blame non-state actors forthe existing situation, while international actors and attribute the problems to the currentgovernment residents/participants or the French state ingeneral. For the short-term prognosis, there is a pattern similar to the non-structural diagnosis, with Sarkozy calling for law and order,as reflected inhis threatto "clean out the city with a power hose," (Saddeutsche Zeitung Nov. 7, 2005) while other actors seek more structural solutions. This short-term prognosis bySarkozy (and other state actors) is more pronounced in Europe than in Canada and the United States. Finally, there is an interesting difference inthe range of diagnostic frames used by different actors (diagnostic crystallization): while opposition and actors consistently invoke the same frames, international media sources and residents/participants present a wider varietyof diagnoses. Expectations about frame crystallization are only confirmed for diagnostic framing. Crystallization duringthe second week is lowestwhen frames are most diffuse. However, there is no evidence of prognostic over time. Finally,there is some evidence fora shiftfrom crystallization over thecourse of the event, but only for diagnostic to prognostic framing a limited of actors: theFrenchopposition uses more diagnostic framing set inthe first weeks compared to the last two week, and international actors use more diagnostic framing in the first week compared to the second and third week. Temporality is the finalfactorposited to account for frame variation. Expectations about differences in framingduring differentstages of the riots are largelyconfirmed. First, the issue is more salient during theweek with the highest intensity rioting(see Figure 1 and Table 6). of week of Second, the use of various sources differsover time. In the first the event, newspapers provide accounts ofwhat is going on, drawing on and incorporatingresidents/participants and international actors as with the exception of Sarkozy,state sources enter sources. Furthermore, thedebate late.This is due inpart to the hesitant performance of Chirac, which prompted the Saddeutsche Zeitung (Nov. 10, 2005) to raise the we see that followingquestion: "Where is the President?" Additionally, the use of residents and participantsas a source increases from week 1 to 2, but thendecreases when the intensity the riotssubsides. of

408 . Social Forces Volume 86,Number 2 . December 2007

Discussion In an attempt to furtherunderstand the factors that account for variation in interpretiveframes, particularly with respect to the same event or issue, two sets of findings are presented. These analyses were guided by ideological, contextual, attributional and temporal which a number of hypotheses were theoretical considerations from derived. The first set of findings revealed descriptively significant variation in the framingof the riots diagnostically and prognostically, and suggested which of the hypothesized factors appeared to account for the observed variation. The second set of findings established which of the hypothesized and descriptively apparent associations were significantacross several models. Overall, the findings reveal mixed to weak support for hypotheses rooted in ideological considerations; mixed to strong support for both contextually-basedhypotheses and those derived fromattribution theory; and strong support for hypotheses positing temporal variation in the character of the framing.In the case of ideological considerations,where newspapers and government fall on a right/left spectrum is of some influence in affecting the character of framing,but ideological position is not determinative. This muted effect cautions against the frequent tendency in the social sciences to invoke ideology as a blunt hammer to account for much unexplained variance inexplanations of political action and framing.10 Turning to contextual factors, social and political proximityto France was only of limitedimportance in relationto framing. Apparently, relative difference proximity or distance from,noteworthyevents makes little to, inhow theyare framed.Perhaps that isbecause we live ina global,wired world in which social and political proximityis of less influence than in the past. Of fargreater influence in understanding frame variation are the differentactors in the field responding to and commenting upon were the event. Here there are interesting differences inhow the riots framed by state, oppositional and international actors and by residents and participants.Moreover, framingby the various sets of actors was such as the fundamental partlypredicted by tenets of attributiontheory, as attribution error, reflected instate actors tendency to blame the riots on negativelyevaluated groups ratherthanon structuralfactors. we and perhaps most noteworthy, foundconsiderable temporal Finally, variation inaspects of framing.For example, the news salience of the event, the prominence of differentframingvoices, and the proportion of diagnostic to prognostic frames all varied temporally. Some evidence was evident - the tendency for fewer frames to of frame crystallization

Framing theFrenchRiots . 409

more of the available discursive space over consume proportionately time. What makes these findingsparticularlyinterestingis that the time span was relatively brief- only threeweeks. This underscores, among other things,the dynamic character of framing. not only suggest the conceptual Taken together,the foregoingfindings of and analytic utility pursuing the issue of frame variation beyond the French riots, but theyalso raise a number of questions thatcall forfurther of crystallization or around some diagnostic framesand thecorresponding decline ofothers over time? Is it essentially amatter of resonance (Benford and Snow 2000; Snow and Benford 1988), such thatsome framesbecome more credible empiricallyas more information about the event becomes available? Or is itdue to differences in the relativepower of the various sources within the discursive field(Snow, forthcoming), actors or framing such that the framingsprofferedby those with more institutional power, and thus situated higher in the field's "hierarchyof credibility," (Becker 1970) are more durable in the sense of having greater staying power. In otherwords, are some framingsources or voices being privileged, and why?1"Clearly themedia - newspapers in this case - play a critical role innot only selecting some framingsources over others and privileging own framings. But some frames over others, but also ingenerating their caution needs to be exercised so as not to attributehegemonic power of to themedia interms of the framing events and issues. Not onlydoes occur inother contexts, but themedia also operates within a set framing of organizational and cultural constraints (Bunis et al. 1996; Hilgartner and Bosk 1988). It is also arguable that there is variation in the potency of the various events (e.g., civil disturbances, hurricanes, earthquakes) we know even though streaming throughthemedia (Snow, forthcoming), about the range and variationof events. relativelylittle Pursuing these and related questions should enhance understanding of the factors thataccount forframevariationbothwithin themedia and more generally,especially with respect to specific events or topics. The analysis reportedherein is but a step inthatdirection.
Notes 1. In political communication, there has been considerable debate regarding the use of issue/event-specific news frames versus generic news frames (De Vreese 2005). riots inquotation marks because the term is sometimes used as a political label that focuses attention on one segment of the participants and implies something demeaning about them. Such concerns have been noted place

theorization.example, For whatfactors accountforthe empirical inquiryand

2. We

410 . Social Forces Volume 86,Number 2 . December 2007 in relation to the "ghetto hots" of the 1960s in the United States (see Feagin and Hahn 1973: fn4, Pp. vi-vii). We thus use the term riotadvisedly, but we use itbecause was the predominant descriptor of the disturbances that it

gripped France for nearly threeweeks. 3.

The formal equation is as follows: IR= 22M/2(N1+N2). Where M signifies the total number of agreements between the two coders, N1 the total of the coding decisions made by the firstcoder and N2 the total of the coding decisions made by the second coder.

4. We

elected President, as a separate source because of his dominant and exclusive position during the evolution of the riots (see results section). it focuses on the personal refer to this framing as "riffraff" because characteristics of the rioters and because of itsapplication to rioters inother places and eras. See, for example Feagin and Hahn's summary of earlier explanations of U.S. urban "riots." (1973:6-28) data were

treat Sarkozy, who was

5. We

6.

These

collected by the U.S. Central Intelligence Agency (2002). rate is defined as the percentage of the population that is Unemployment currently unemployed. Net migration rate is operationalized as the difference between the number of persons entering and leaving a country during the year per 1,000 persons (based on midyear population). A positive number indicates that there are more people entering than leaving a country.

7.

At a first glance our research design might appear to call for a multilevel approach, with inclusion of several hierarchical ordered units (e.g. newspapers within countries within time-periods). However, multilevel analyses are only appropriate when the samples at the different levels are of sufficient size (15-20), which is not the case here. Furthermore, we do not have a random sample of countries and newspapers, but we chose them purposively. Nor are we interested ingeneralizing statistically our results to all time periods, we seek to establish the relative impact of countries, or newspapers; rather, independent variables on our dependent and regression analyses are appropriate. which variance variables, for

various

8.

Because

a division innewspapers also encompasses a division incountries, it is statistically not possible to examine them separately in the analyses.

9.

Note

state, forexample Azouz Begag.

that the framing of other state actors is significantly different from we if include Sarkozy with framing by the "other" category. However, the other state actors, this coefficient would be highly significant. This difference suggests the presentation of more nuanced diagnoses by the by the Minister for the Promotion of Equal Opportunities,

Framing theFrenchRiots 0 411 10. For furtherdiscussion of the relationship between ideology and framing, see

Snow and Byrd (2007), Oliver and Johnston(2000),Snow (2004:386-404), Westby (2002). 11. See Carragee and Roefs (2004) fordiscussion of the neglect of power
in media framing research. References Abbott, Andrew. 2001. Chicago Press. Time Matters: On Theory and Method. University of

1988. Enemies Altemeyer, Bob. Authoritarianism. Jossey-Bass. Bateson,

of Freedom:

Understanding

Right-Wing

Gregory. [1955] 1972. "A Theory of Play and Phantasy." Pp. Steps to an Ecology of theMind. Ballantine Books. 1970. Sociological Work: Method and Substance. Aldine.

177-93.

Becker, Howard.

Benford, Robert D., and David A. Snow. 2000. "Framing Processes And Social An Overview And Assessment." Movements: Annual Review of Sociology 26:611-39. Benford, Robert D. 1993. "Frame Disputes Movement." Social Forces 71(3):677-701. within the Nuclear Disarmament

Bennett, W. Lance. 1990. "Towards a Theory of Press-State United States." Journal of Communication 40(2): 103-25. Benson, Rodney. 2004. "Bringing The Sociology Communication 21(3):275-92. Of Media

Relations

in the

Back

In." Political

Berbrier, Mitch. 1998. "'Half the Battle': Cultural Resonance, Framing Processes, and Ethnic Affectations inContemporary White Separatist Rhetoric." Social Problems 45(4):431-50. Berliner Morgenpost. unter Kontrolle." Nov. 6, R1. Ian., Hans-Dieter Klingemann, Andrea Volkens, Judith Bara and Eric Budge, 2001. Mapping Tanenbaum. Estimates For Parties, Policy Preferences: Electors, and Governments, 1945-1998. Oxford: Oxford University Press. Bunis,William K., Angela Yancik and David A. Snow. 1996. "The Cultural Patterning of Sympathy toward the Homeless and Other Victims of Misfortune." Social Problems 43(4): 387-402. 2005. in Paris bekommt Unruhen nicht "Regierung inParis is not capable of controlling unrest.]

[Government

412 . Social Forces Volume 86,Number 2 . December 2007 Cadena-Roa, Jorge. 2002. "Strategie Framing, Emotions, and Superbarrio-Mexico City's Masked Crusader." Mobilization 7(2):201-16. Carragee, Kevin M., and Wim Roefs. 2004. "The Neglect of Power Framing Research." Journal of Communication 54(2):214-33. "The Outcomes in Recent

Cress, Daniel M., and David A. Snow. 2000. Mobilization: The Influence of Organization,

of Homeless

Disruption, Political Mediation, and Framing." American Journal of Sociology 105(4): 1063-1104. Claes H. 2005. "News framing: Theory and Typology." Document

de Vreese,

Des/#/7l3(1):51-62. de Vreese, Claes H., Jochen Peter and Holli A. Semetko. 2001."Framing Politics at the Launch of the Euro: A Cross-National Comparative Study of Frames in the News." Political Communication 18(2): 107-22. de Vreese, Claes H., and Holli A. Semetko. 2004. Political Campaigning Referendums. Framing the Referendum Issue. Routledge. in

Dimitrova, Daniela. V, and Jesper Str?mb?ck 2005. "Mission Accomplished? in Elite Newspapers in Sweden and the United Framing of the IraqWar States." International Communication Gazette 67(5):399-417. the Dialectic of Discourse 1995. "Understanding and Ellingson, Stephen. - Public Debate and inAntebellum Cincinnati." Collective Action Rioting American Journal of Sociology 101 (1 ): 100-44. Entman, Robert M. 1993. "Framing: Toward Clarification of a Fractured Paradigm." Journal of Communication 43(4) : -8. 51 Esser, Frank, and Barbara Pfetsch. 2004. Comparing Political Communication: Theories, Cases, And Challenges. Cambridge University Press. Feagin, Joe R., and Harlan Hahn. InAmerican Cities. Macmillan. 1973. Ghetto Revolts: The Politics Of Violence

Ferr?e, Myra M., William A. Gamson, J?rgen Gerhards and Dieter Rucht. 2002. Shaping Abortion Discourse: Democracy and the Public Sphere in Germany and the United States. Cambridge University Press. Galtung, Johan, and Marie H. Ruge. 1965. "The Structure of Foreign News." Journal of Peace Research 1:64-90. Gamson, William A. 1992. Talking Politics. Cambridge Gamson, William Science University Press.

1993. "Movements and Media as A., and Gadi Wolfsfeld. Annals of theAmerican Academy of Political and Social Interacting Systems." 528(1)^14-25.

Framing theFrenchRiots . 413 Geertz, Clifford. Basic Books. 1973. The Interpretation of Cultures: Selected Essays.

Todd. 1980.The Whole World Is Mass Media and the Gitlin, Watching: Making &
Unmaking of theNew Left. University of California Press. An Essay On Goffman, Erving. 1974. Frame Analysis: Experience. Harvard University Press. The Organization Of

Hilgartner, Stephen, and Charles L. Bosk. Problems: A Public Arenas Model." 94(1):53-78.

1988. "The Rise and Fall of Social American Journal of Sociology

Holsti, Ole R. 1969. Content Analysis for The Social Sciences Addison-Wesley Publishing Company.

And Humanities.

Jones, Edward E., et al. 1972. Attribution: Perceiving General Learning Press.

the Causes

of Behavior.

Ruud. 2004. and Media: Selection Processes "Movements Koopmans, in the Public Sphere." Theory and Society and Evolutionary Dynamics 33(3-4):367-91. Le Figaro. 2005. "ANanterre, l'association Zy'va reconstitue le tissu familial pour apaiser les esprits." [The association Zy'va rebuilds the family bonds to life the spirits inNanterre]. Nov. 17, R 8. Mannheim, Karl. 1936. Ideology and Utopia: An Introduction to the Sociology Harcourt, Brace and Company. 1970. The German Ideology. International. of

Knowledge.

Marx, Karl, and Friedrich Engels.

McCaffrey, Dawn, and Jennifer Keys. 2000. "Competitive Framing Processes in the Abortion Debate: Polarization-Vilification, Frame Saving, and Frame Debunking." The Sociological Quarterly 41(1 ):41-61. McCammon, Holly J. 2001 ."StirringUp Suffrage Sentiment: The Formation of 1866-1914. Social the State Women Forces suffrage Organizations, 80(2):449-80. National Post. 2005. "Paris suburb demands New imposes firstcurfew: Prime Minister to call inarmy." Nov. 8, R A10. 2005. "Angry Immigrants Embroil France in Wider rejects

York Times. Nov. 5, PA1.

Riots."

_.

2005. "Chirac, Lover of Spotlight, Avoids Nov. 10, RA10.

Glare of France's

Fires."

414 . Social Forces Volume 86,Number 2 . December 2007 _. _. 2005. 2005. "AVery French Message from the Disaffected." of Crisis Rules Nov. 13, R A1. to combat Riots."

"Chirac to Ask for Extension

Nov. 15, RA14. Oliver, Pamela E. and Hank Johnston 2000. "What a good Idea! Frames inSocial Movement Research." Mobilization 5(1):37-54 Ottawa Ideology and

Citizen. 2005. "French riots bring recruits to far right: Front National party calls for stronger rules for immigration, end to 'political madness.'" Nov. 13, RA13. "Slaan de Franse onlusten over naar Nederland Nov. 10, R 27 ?" [Will the French

Parool. 2005.

riots diffuse to the Netherlands?]

Rohlinger, Deana. 2002. "Framing theAbortion Debate: Organizational Resources, and Movement-Countermovement The Media Strategies, Dynamics." Sociological Ross, Quarterly 43(4):479-507.

Lee D., Teresa M. Amabile and Julia L. Steinmetz. 1977. "Social Roles, in Social-Perception Journal of Processes." and Biases Social-Control, Personality and Social Psychology 35(7):485-94. Effects." Journal of

Scheufele, Dietram A. 1999. "Framing As a Theory of Media Communication 49(1 ): 103-22.

Semetko, Holli A.f and Patty M. Valkenburg. 2000. "Framing European Politics: A Content Analysis of Press and Television News." Journal of Communication 50(2):93-109. Smith, Jackie, John D. McCarthy, Clark McPhail and Boguslaw Augustyn. 2001. "From Protest to Agenda Building: Description Bias inMedia Coverage of Protest Events in Washington, D.C" Social Forces 79(4): 1397-1423.

Snow, David A. Forthcoming. "Elaborating the Discursive Contexts of Framing: Discursive Fields and Spaces." Studies inSymbolic Interaction. _. 2004. Framing Processes, Ideology, and Discursive Fields. Pp. 380-412. The Blackwell Companion to Social Movements. David A. Snow, Sarah Soule and Hanspeter Kriesi, editors. Blackwell Publishing. Snow, David A., and Robert D. Benford. 1988. "Ideology, Frame Resonance, and Participant Mobilization." International Social Movement Research 1:197-217. Snow, David A. and Scott Byrd. 2007. "Ideology, Framing Processes, Terrorist Movements." Mobilization 12(1 ): 119-36. and Islamic

Framing theFrenchRiots . 415 Snow, David A., Steven K.Worden, E. Burke Rochford and Robert D. Benford. 1986. "Frame Alignment Processes, and Movement Micro-mobilization, American Sociological Review 51(4):464-81. Participation." S?ddeutsche _. Zeitung. 2005. "Wir sind imKrieg." [We are in war]. November 7, P 2. is the President?] November 10, P 8.

2005. "Wo istder Pr?sident?"[Where United States

Central Intelligence Agency. 2002. Factbook on Intelligence D.C.: United States Central Intelligence [electronic resource]. Washington, Agency.

Vliegenthart, Rens, Dirk Oegema, and Bert Klandermans. 2005. "Media Coverage and Organizational in the Dutch Environmental Movement." Support Mobilization 10(3):365-81. Westby, David L. 2002. "Strategic Imperative, Ideology, and Framing."Mobilization 7(3):287-304. Wikipedia: the Free Encyclopedia. 2005. "2005 Civil Unrest inFrance." Accessed Dec. 5, 2005 at: http://en.wikipedia.org/wiki/2005_Paris_suburb_riots. Zald, Mayer N. 2000. "Ideologically Structured Action: An Enlarged Agenda Social Movement Research." Mobilization 5(1): 1-16. for

Vous aimerez peut-être aussi