Media's Relationship With Stereotyping

Journal of Communication ISSN 0021-9916

RESEARCH ARTICLE

News Media’s Relationship With Stereotyping: The Linguistic Intergroup Bias in Response to Crime News Bradley W. Gorham S.I. Newhouse School of Public Communications, Syracuse University, Syracuse, NY 13244

This paper examines the linguistic intergroup bias (LIB) in the context of people’s interpretations of a race-related television news story. The LIB suggests that people use more abstract language to describe stereotype-congruent behaviors, particularly when that person is a member of an out-group. This study of 208 White adults manipulates the race of a suspect in a TV news crime story and examines how race influences the abstractness of the language viewers use to describe the suspect. The findings offer support for the LIB being induced by crime news and show that news media use is significantly related to the presence of the LIB. This suggests that stereotypical news coverage may subtly influence the interpretations people make about members of other social groups. doi:10.1111/j.1460-2466.2006.00020.x

Mass communication researchers and social psychologists often point to the mass media as an important source of stereotypical images, with the assumption that these images produce or reinforce prejudice and discrimination on the part of audiences. For example, Allport (1954/1979) noted back in 1954 that U.S. newspaper stories that describe criminals as Negroes will likely affect the audience’s perception of African Americans: ‘‘Yet so frequently to associate Negro with crime is bound to leave a lasting effect on readers, particularly if this association is not offset by news items favorable to the colored group’’ (p. 201). Devine and Elliot (1995) note that ‘‘stereotypic images of Blacks persist in the dominant media .. As a result, stereotypes are perpetuated within the culture in subtle, yet highly effectual ways’’ (p. 1149). Such assumptions are not only intuitively appealing but also seem reasonable in light of the findings of cultivation research. Research from this perspective has generally shown a small but consistent relationship between the amount of television viewed and beliefs about the social world (Morgan & Shanahan, 1997; Potter, 1994) such that frequent viewers of television tend to believe that the real world is similar to the world that is portrayed on television. Given that television systematically distorts

Corresponding author: Bradley W. Gorham; e-mail: [email protected]. Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

289

Linguistic Intergroup Bias

B. W. Gorham

these portrayals in ways that favor social elites (see Greenberg, Mastro, & Brand, 2002), cultivation researchers argue that television is an important tool for maintaining dominant ideology (Gerbner, Gross, Morgan, Signorielli, & Shanahan, 2002). Although many content analyses have uncovered the rich and complex ways that media texts might support dominant understandings of race, power, and society, they cannot by themselves tell us whether the messages are really having these effects among viewers and are truly performing the ideological work ascribed to them (Gorham, 1999). Similarly, one of the enduring criticisms of classic cultivation research is that without specifying the mechanisms that lead from television exposure to effect, even this important stream of research is open to questions of spuriousness (Hawkins & Pingree, 1990). Thus, in order to better understand the ways in which media help maintain stereotyping and prejudice in a society that is less tolerant of outright bigotry (Devine & Elliott, 1995), we need to look more closely at how people interact with mass media messages about race. Recent research in our field has used the tools of social psychology to uncover the various ways people interact with media content involving race (e.g., see Oliver, Jackson, Moses, & Dangerfield, 2004). This attention to how audiences process media messages is especially important given that people can react to ambiguous information in stereotype-congruent ways, even when they consciously reject prejudice (Devine, 1989; Monteith, 1993). In addition, although overt bigotry has generally declined in the United States, some researchers suggest that it has been replaced by a more subtle form of modern racism. Gaertner and Dovidio (1986), for example, present evidence for what they call ‘‘aversive racism,’’ characterized by Whites maintaining both sympathetic views for the plight African Americans have had to endure with feelings of discomfort or disdain for Blacks. Thus, in a world where mass media offer contradictory images of African Americans as both successful news anchors and dangerous criminals on the same newscast (Entman, 1990), people may well hold contradictory beliefs about minorities. Media images may contribute to the maintenance of subtle racism without audience members even being aware of it. This paper examines how Americans’ reactions to messages about race can reveal the complex and subtle ways that media use is related to the perception of difference between people of various social groups. More specifically, this paper focuses on linguistic intergroup bias (LIB), defined as the use of different levels of abstractness in language depending on whether a person is talking about someone who belongs to an in-group or an out-group and whether that person is behaving in a way consistent with the group’s stereotype. Only by better understanding the interpretations of race-related news, can we figure out how to help news producers and news audiences avoid reinforcing negative stereotypes. Groups and social knowledge People tend to divide themselves into reasonably coherent groups. What is at issue is not the presence of human groups but the differences people see between themselves 290

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

and members of other social groups. As opposed to those who assert that social groupings reflect naturally existing co-occurrences in the environment (see Rosch, 1978), Wittenbrink, Hilton, and Gist (1998) argue that the perception of differences in the social environment is the result of people applying ‘‘naive theories’’ about the existence of coherent categories onto the people they see around them. People’s lay theories about the world, they argue, suggest not only that people can be divided into groups, but also that there are explanations for these divisions. They therefore ‘‘see’’ coherent categories among others and organize their social knowledge around these perceived differences. ‘‘These theories provide subjective explanations that structure the social environment and define the partitions the perceiver imposes upon it. They explain what a given group of people is like, what attributes the group members share, and, more importantly, why they share these attributes’’ (Wittenbrink et al., 1998, p. 49). Explanations about social groups rarely reflect people’s direct experiences with groups, and so they are more likely found in the social knowledge shared by members of a culture. This view of social categorization allows for the functioning of ideology in the differentiation of social groups and the stereotypes that describe them. Indeed, concern for the relationship of stereotypes, social group categorization, and existing structures of power and dominance were part of both Lippmann’s (1922) and Allport’s (1954/1979) use of the term ‘‘stereotype.’’ In his definition, Allport noted that a stereotype’s purpose ‘‘is to justify (rationalize) one’s conduct in relation to the category’’ (p. 191). Stereotypes, then, represent a link between the categories we use to define the social world and the power structures that govern that world. As van Dijk (1987) argues Ideologies organize large portions of our social life and are based on fundamental goals, interests, and values . . Hence, ideologies are the cognitive reflections of our social, political, economic, and cultural ‘‘position’’ within the social structure . . This means that ideologies, even less than their component attitudes, are not individual, but group based. (p. 194)

Thus, ideology in this cognitive sense is the practice of our individual experiences and raw perceptions (the episodes of our lives) being defined by socially constructed semantic knowledge, with those definitions supporting existing structures of difference. Our perceptions serve our group interests by making attributes in the social environment that might explain differences perceptually salient. For example, Wittenbrink et al. (1998) contend that there is nothing in the environment that naturally makes race the basis of group differentiation. Indeed, this was not the case for a long time: Race was not seen as an important marker of distinction in ancient Egypt or Greece, and race as a biological explanation for difference was not posited until the slave trade came under increasing intellectual and economic attack (Montagu, 1997). Thus, it was the operation of ideology that ‘‘created’’ race as a salient characteristic by which to group people. This perspective is given added weight in light of the findings of social identity theory and research on the attribution error. According to social identity theory Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

291

Linguistic Intergroup Bias

B. W. Gorham

(Tajfel & Turner, 1986), people’s self-concept involves beliefs about their own abilities and their perceptions of the social groups to which they belong. Because people wish to maintain a positive social identity, they will strive to create favorable comparisons between their in-groups and out-groups when their social identity is threatened (Crocker & Luhtanen, 1990; Perdue, Dovidio, Gurtman, & Tyler, 1990). Thus, people denigrate the out-group in an attempt to make the in-group look better. Research also highlights another aspect of this intergroup dynamic: Members of the out-group are seen as being relatively homogeneous, in that the attributes that are said to belong to them are assumed to hold for most members of that group. Ingroups, in contrast, are assumed to feature more variety among members (Linville, Salovey, & Fischer, 1986). The ultimate attribution error (Pettigrew, 1979), in contrast, attempts to explain the behaviors of in-groups and out-groups as functions of either internal or external causes. Pettigrew built on Heider’s (1958) concept of the fundamental attribution error by showing that people who perceive a person performing a negative behavior will be more likely to attribute that behavior to dispositional (internal) explanations if the person is from an out-group, whereas they will attribute the same negative behavior to situational (external) factors if performed by an in-group member. Likewise, positive behaviors will likely be attributed to situational causes when performed by out-group members and dispositional causes when performed by members of the in-group. There appears to be a preference for attributions that help serve the interests of the in-group (Hewstone, 1990). In this context, our beliefs about racial groups have a cognitive component that predisposes us to recognize racial distinctions as salient and more readily attribute negative behaviors performed by members of other racial groups as being reflective of the group as a whole. Because our stereotypes about these groups are stored as simply another type of schema, they are subject to all the processing characteristics that seem to involve all schema: They aid in parsing incoming information, they can be primed, they help structure expectancies, and they can direct our perception of subsequent information. But as has already been mentioned, these schema reflect the operation of ideology in the social environment and thus reflect the distinctions that we apply to the human groups we encounter. Communication research has highlighted the ability of stereotypical media content to prime particular explanations for the behaviors shown in that content. Gilliam, Iyengar, Simon, and Wright (1996) manipulated the race of a suspect in a crime story to examine the effects this manipulation would have on television viewers. Subjects were shown a 15-minute segment of a local newscast, including a crime story in which the mug shot of the suspect was ‘‘painted’’ to alter the suspect’s complexion. Following the newscast, subjects completed a lengthy questionnaire that included questions about the causes and significance of crime, their preferred methods for dealing with the problem, and their stereotypes of various social groups, including African Americans (p. 16). Gilliam et al. (1996) found a significant main effect for the race of the suspect. Subjects expressed more concern for crime and were 292

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

more likely to attribute the causes of crime to group characteristics for the African American suspects compared to the White suspects. Johnson, Adams, Hall, and Ashburn (1997) also included a measure of attribution in their priming experiment; they were interested in how the level of violence in a story might prime readers to evaluate a subsequent African American defendant differently than a White defendant. Given that hostility is a prominent part of the stereotype of Black men (see Devine & Elliot, 1995), Johnson et al. (1997) suggested that if people are primed to think about violence, they may evaluate an African American defendant differently than a White defendant. Violence should trigger the Black stereotype when viewers encounter an African American defendant and lead to more dispositional attributions for African American suspects than for White suspects. In what they thought was a perception and recall task, subjects in the Johnson et al. (1997) study read three stories that either did or did not describe violent acts. In a second study on ‘‘decision-making processes,’’ subjects read an irrelevant article and an article that dealt with a man who was arrested for vandalizing automobiles. The story included background information about the man and a photo of either a Caucasian man or an African American man, or no photo (a race-unspecified condition). Johnson et al. found that ‘‘attributions of defendant behavior did not vary as a function of story violence level for the White and race-unspecified defendant. In contrast, for Black defendants, attributions were more dispositional in the violent condition than the nonviolent condition’’ (p. 85). Furthermore, attributions were more dispositional overall for the Black defendant than either the White or the raceunspecified defendant. Johnson et al. suggest that the primed information (violence) is considered more applicable to Blacks than Whites because it is such a prominent trait of the stereotype of African American men. Thus, the prime had more of an effect when subjects subsequently read about an African American man because it triggered the rest of the stereotype, leading to more dispositional attributions. These results suggest that priming stereotypes of African Americans triggers cognitive processing that follows the ultimate attribution error. This process thereby enhances the ability of the news to produce not just stereotype-congruent interpretations about an individual suspect but also interpretations about the larger group that support dominant racial ideology. Language and the structure of racial thought It is thought by some linguists that the structure of language can reveal important information about the structure of thought (Graesser, Gernsbacher, & Goldman, 1997; Lakoff, 1987; van Dijk, 1997). If that is the case, then the language people use in connection with their intergroup encounters should reveal information about the way groups are perceived. One potentially useful finding for communication researchers comes from what Maass, Salvi, Arcuri, and Semin (1989) call the linguistic intergroup bias, which is the use of different linguistic properties in intergroup contexts. Between social groups with strong negative feelings for each other, Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

293

Linguistic Intergroup Bias

B. W. Gorham

Maass et al. (1989) found that speakers tend to use different types of language depending on the group involved (whether in-group or out-group) and the valence of the behavior involved (positive or negative). Subjects in the Maass et al. study tended to use more abstract language when describing a positive behavior of an in-group member or the negative behavior of an out-group member. However, when describing a negative behavior of an in-group member or a positive behavior of an outgroup member, speakers tended to use more concrete language. Language abstractness, first discussed in Semin and Fiedler (1988), is defined by the use of four types of descriptors: descriptive action verbs (DAVs), considered the most concrete; interpretive action verbs (IAVs); state verbs (SVs); and adjectives (Adjs), which are the most abstract. Abstract language describes highly generalized person dispositions, detached from specific and observable behaviors, and enduring states (e.g., honest, creative, believe, envy). Concrete language, in contrast, describes specific and observable behaviors with clear beginnings and ends, which may or may not have semantic connotations (e.g., kiss, visit, help, threaten; Semin & Fiedler). Furthermore, abstract language is seen as being more informative about the person involved as abstract descriptions are seen as being more stable over time. In an intergroup context, language abstractness can also be thought of as a marker of situational or dispositional attribution for the observed behaviors. Concrete descriptions of a behavior are bounded by the situation in which they occur. They firmly locate the action in a specific place and time but do not tell us much about the situation beyond that. Abstract language, in contrast, implies much more about the disposition of the person involved that is independent of what has been observed. It assumes that the observed behavior is a manifestation of a larger tendency, thus implying that some knowledge exists about the person in question. If all that is different about the person is the group to which he or she belongs, then abstract language implies that knowledge about the social group can be applied to that specific person. Language abstractness, then, can reveal the structure of thought about the group in question as it can reveal the extent to which people think observed behaviors are tied to the circumstances or instead are the result of inherent traits of that group. Subsequent work by Maass, Milesi, Zabbini, and Stahlberg (1995) demonstrates that the LIB is a function of the expectancies that stereotypes generate concerning the likelihood of behaviors. If a behavior is seen as being consistent with what one would expect given the stereotype, then the LIB is much more likely to occur. This suggests that the LIB is the result of cognitive processing rather than the result of motivational processing to favor an in-group, although they acknowledge that such a finding does not mean that in-group favoritism is unimportant for language use. Karpinski and von Hippel (1996) echo this finding by showing that the LIB helps people maintain their expectancies despite the presence of incongruent information. They conclude from this that the LIB should be viewed as part of the attributional bias of intergroup perception. Building on this idea, research has also looked at the LIB in relation to prejudice. von Hippel, Sekaquaptewa, and Vargas (1997), for example, demonstrate that a 294

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

measure based on the LIB can be used as an implicit measure of prejudice for both African American and gender stereotypes. Schnake and Ruscher (1998) show that high-prejudiced White subjects describe the stereotypical behaviors (both positive and negative) of an African American using more abstract language compared to low-prejudiced subjects. Furthermore, von Hippel et al. (1997) also found that their LIB results were unrelated to an explicit measure of prejudice, suggesting that although people may be able to control their explicit reactions to race, their cognitive processing nonetheless reveals stereotype-congruent tendencies. This finding reflects others from the psychological literature that stereotype-congruent cognitive processing can coexist with conscious processing that is genuinely low in prejudice. Thus, the LIB does indeed seem to reveal something important about the structure of White people’s beliefs about African Americans. Gorham (2002) applied the LIB to the study of audience reactions to a television news crime story. In that study, White undergraduate students were first surveyed about their perceptions and use of news media as well as their endorsement of stereotypes and prejudice. A week later, participants were shown the first 10 minutes of a newscast—a newscast in which the race of a suspect in a crime story was manipulated—and then asked a series of questions about what they saw. Included were open-ended questions concerning the suspect seen in the crime story. Gorham compared the language used to describe the suspect by participants who had seen an African American to the responses from subjects who had seen a White suspect. He found evidence of the LIB in the participants’ open-ended responses to questions about the suspect: A greater proportion of the descriptions were Adjs, the most abstract descriptor, when the suspect was Black than when he was White. This result suggests once again that cultural stereotypes of African Americans had been activated in these White undergraduates by the crime story and that these stereotypes had influenced how the students talked about the suspect. That study, however, suffered from some questions about the reliability of the coding as the coding of the more concrete DAVs and IAVs was unreliable. Although coding open-ended responses for the LIB was not unprecedented, as Schnake and Ruscher (1998) had also coded open-ended responses for the LIB, such a method trades a certain level of reliability for the added richness of the data. Scholars often measure the LIB by showing participants some stimulus material and then asking them to choose from one of four descriptions of the event. These descriptions correspond to the four levels of abstraction (e.g., (a) John hit her, (b) John hurts her, (c) John hates her, and (d) John is aggressive) (Semin & Fiedler, 1988). Thus, the present study is an attempt to replicate the findings of the earlier research using closed-ended items. Thus, the chief hypothesis for this study is: H1: The language White participants choose to describe an African American suspect will be more abstract than the language chosen to describe a White suspect.

In addition to looking for evidence of the LIB in people’s reactions to a specific story, this study can also examine the relationship between news media use and the Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

295

Linguistic Intergroup Bias

B. W. Gorham

presence of the LIB. Can television news cultivate the perception of difference, as revealed by the LIB? There is ample evidence that although entertainment television may be more generous in its portrayals of African Americans today than it was in the past (Greenberg et al., 2002), television news tends to portray African Americans and other racial minorities in ways that support dominant stereotypes of race. In a content analysis of Chicago local news programs, Entman (1992) found that where statistical differences existed between portrayal of Whites and Blacks, the difference was always in a direction that made Blacks appear threatening in some way. For example, African American crimes reported in the news were more likely to be violent or drug related, and the suspects were more likely to be shown in physical custody, whereas less likely to be named or heard from than their White counterparts. Similarly, African American political actors were much more likely to be seen arguing that the government violated or should serve Black interests than White political actors were seen promoting racial self-interest. Although acknowledging the problems associated with making assumptions about effects from content analyses, Entman (1990, 1992) argues that patterns such as these may work to support the development and maintenance of McConahay’s (1986) modern racism. Findings from Romer, Jamieson, and de Coteau (1998) and Dixon and Linz (2000b, 2000a) echo these same conclusions. Romer et al. (1998) examined the race of perpetrators, victims, and bystanders presented in television news stories in Philadelphia. By comparing the relative proportion of Whites and Blacks in these stories to real life crime statistics for the same period, the researchers hoped to decipher whether the patterns reported by Entman (1992) and others were the result of realistic group conflict or the operation of dominant racial ideology in the form of ethnic blame discourse. Their results replicated those of Entman in that nonWhites were more often shown as perpetrators of crime and Whites were more often shown as the victims; they also went a step further and ‘‘demonstrated that this disparity could not be explained using typical rates of inter- and intragroup victimization as reported by the police’’ (Romer et al., 1998, p. 298). The same pattern of results was found in Dixon and Linz’s (2000b) study of television news in Los Angeles and Orange Counties. Furthermore, Dixon and Linz (2000a) found that Whites were overrepresented, and Blacks and Latinos underrepresented, as police officers in television news stories about crime. If the presence of the LIB is considered a marker of dominant ideology about social groups by indicating the perception of difference based on group affiliation, then we might expect that those who consume more television news would be more likely to exhibit the LIB, that is, we would expect that, compared to people who watch relatively little television news, those who watch more television news would be exposed to more portrayals of Whites and Blacks in systematically different and stereotypical ways. Such portrayals might cultivate the perception of Whites and Blacks as different and hence strengthen the tendency to use more abstract language following a story about a Black suspect. Although much of the research about portrayals of African Americans in news has been done using local television news, 296

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

there is some evidence that national news has also featured systematic differences (Jamieson, 1992). Thus, the next hypotheses focus on the ability of television news to cultivate the LIB. H2: The LIB will be present for frequent users of local television news use but not for relatively light users. H3: The LIB will be present for frequent users of network television news use but not for relatively light users. H4: The LIB will be present for frequent users of cable television news use but not for relatively light users.

However, it is less clear how other television programming in general might be related to the LIB. Cultivation scholars (e.g., Gerbner et al., 2002) argue that, although the percentage of African Americans on television has increased dramatically since the 1980s (see Greenberg et al., 2002), the overall symbolic messages of television programming have not changed very much. If Whites are still systematically portrayed in ways that support their dominance, then exposure to television in general should support the development and maintenance of stereotypes and prejudice. Thus, cultivation would predict that people who watch more television should be more likely to exhibit the LIB than less frequent viewers. In contrast, one criticism of cultivation is its insistence on medium-level measures rather than on more content-specific ones (Potter, 1994). This criticism seems especially relevant in light of the dynamic television environment of today as close to 80% of the households in America receive cable television and audiences are increasingly fragmented across dozens of channels. There are also more, and more varied, portrayals of people of color on television than ever before, so the relationship between television viewing and the LIB may not be as straightforward as cultivation researchers would posit. Thus, one research question is: RQ1: What is the relationship between overall television viewing and the LIB?

Finally, this study also investigates the relationship between the LIB and other news sources. Although crime news has been a staple of the newspaper since the era of the Penny Press, we should not automatically assume that newspapers reflect television news’ racial disparities and lead to more stereotypical processing. Because newspaper use has often been associated with increased knowledge about news (Bennett, 1989), one might think that it could be associated with less stereotypical processing and hence be less likely to relate to the LIB. Furthermore, web news sources have been little studied in this context, so it is left as a research question as to how these two news sources will be related to the LIB: RQ2: What is the relationship between newspaper and web news use and the LIB?

This study attempts to improve on the earlier study of the LIB in response to TV news not only by expanding the links between LIB and media use but also by Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

297

Linguistic Intergroup Bias

B. W. Gorham

examining a broader sample of adults rather than using a convenient sample of college undergraduates. Students are not typical in their media-use habits, and although the validity of results using student samples may not be an issue for processing-oriented research questions (see Pingree et al., 2001), the generalizability of findings will nonetheless be enhanced by broadening the base of our research participants. Thus, this study attempts to look for more evidence of the LIB in people’s reactions to television news stories about race in an adult sample using more standard measures. Methods

The results described in this paper were part of a larger study on people’s reactions to race-related television news. The sample used in this study consisted of 208 White adult members of the staff of a midsized private university in the Northeastern United States. Data from 23 people of color who participated in the study are not reported here. The participants had a mean age of 40.44 years, but despite efforts to recruit men, the sample was severely skewed toward women, with 77.9% of the participants being female. Similar to Gilliam et al.’s (1996) study, the researcher and an assistant recruited university staff, excluding students and faculty, to participate in an ‘‘impressions of TV news’’ study by offering to pay them $10 for their time. Sessions were scheduled for lunch hours and late afternoons over the course of 2 weeks, and participants selfselected which session to attend. Sessions were held in a room with a large video screen and varied in size from 5 to 33 people. Upon arrival, participants were told by the White research assistant that they would be surveyed about the media-use habits and their opinions concerning TV news, that they would watch a short video, and that they would then be asked to respond to the video. To encourage open and honest responses, participants were also told about procedures the researchers would use to safeguard the anonymity of their responses. Once they consented to participate, the respondents were given a previewing survey. The previewing survey asked respondents to estimate how many a minutes a day they spent watching local, national network, and cable television news; reading a daily newspaper; and reading a news Web site (like CNN.com or NYTimes.com). It also asked respondents to estimate how many hours of total television they watched in a day. Other items on the survey, which are not reported here, asked about participant’s perception of news credibility, their motivations for viewing TV news, and other areas. After completing the survey, participants watched a video of the first 8 minutes of a television news broadcast from Baton Rouge, Louisiana, which contained the experimental manipulation. The particular version of the video for each session was chosen by the researcher prior to the arrival of the participants, rather than randomly selected, to try to ensure a similar number of participants in the four conditions. Participants saw one of four versions of the news broadcast, after which they were 298

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

given postviewing surveys to complete. The videotape consisted of an opening ‘‘Got Milk’’ commercial and then the newscast to the first commercial break. The newscast contained four news stories, a weather segment, and two teases before the first commercial break. The two news anchors, a male and a female, were both White, as were the reporters in all the stories. Unless otherwise noted, all the people presented in the news stories were White. The first story in the broadcast, lasting 30 seconds, centered on the rebuilding of a chemical plant in Baton Rouge that had recently been damaged by an explosion. The second story in the newscast, nearly 2 minutes and 15 seconds in length, focused on the families of several victims of murder, for which a suspect had not yet been caught or even identified. Near the end of the story, the stand-up reporter announces that police, who have been tight lipped in their investigation, will only say that they want to talk to ‘‘this man.’’ At this point in the story (with only about 15 seconds left in the package), the manipulation occurred, and subjects saw one of four different versions of the story. In all versions of the story, the audio track, in which the reporter describes the man’s height and weight but not his race, remains the same. In two versions of the story, a close-up photograph of an African American man’s face is shown surrounded by a red border. These two versions differed only in the actual photograph. Similarly, two versions featured close-up photographs of two different White men. Two different African American men and two different White men were used in the photographs to ensure that the results would be because of race and not because of some unusual feature of a particular man’s face. This seems especially prudent given that research on social perception suggests that facial characteristics can influence social judgments (Berry & Wero, 1993). To test for the similarity of the faces within racial conditions, photos of four different Black men and four different White men were judged using Rhodes’ (1988) facial ratings scales before the tapes were selected for use. Ten White graduate students rated the faces, and the two that were the most similar for each race in terms of how closely their scale scores corresponded were used for the stimulus tape. Two brief stories and a weather segment followed the crime story. Two teases, one of which featured an African American football coach and one that featured an African American state senator, were the final segments before a brief lotto screen and a commercial. It is at this point that the assistant turned off the videotape and handed out the postviewing surveys. The first 10 items of the postviewing survey asked respondents about their mental activities during viewing, such as reflecting on the content and seeing connections to their own lives, and their overall impressions of the quality and credibility of the newscast. The next series of items contained the LIB measure. For each of the stories in the newscast, participants were asked to choose which of four answers best summarized that story. Each of the choices corresponded to a level of linguistic abstraction, such as ‘‘the female anchor waved to the crowd’’ (concrete DAV) to ‘‘the female anchor is friendly’’ (abstract Adj). For the crime story, participants Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

299

Linguistic Intergroup Bias

B. W. Gorham

were asked which of the following they would choose to summarize the man’s role in the story: The man police want to talk to probably hit the victims (DAV); the man police want to talk to probably hurt the victims (IAV); the man police want to talk to probably hated the victims (SV); and the man police want to talk to is probably violent (Adj)1. The LIB, then, is operationalized as a greater proportion of participants using the abstract descriptors like Adjs following exposure to the African American photo compared to those exposed to a White man’s photo. Thus, it is a between-subjects comparison of the abstractness of the language used to describe people of the two groups. The remainder of the survey included a mix of close-ended and open-ended questions used in the larger study and not reported here. The final items on the postviewing survey measured demographic data but also included Brigham’s (1993) Attitude Toward Blacks (ATB) scale, a 20-item, multifactor measure of prejudice. Results

As a group, the 208 participants were relatively moderate news users, watching an average of one local (M = 32.27 minutes, SD = 42.23) and one network newscast (M = 24.46 minutes, SD = 31.55) a day as well as some cable news (M = 15.90 minutes, SD = 24.78) and reading a modest amount of news in the paper (M = 17.66 minutes, SD = 26.34) and on the Web (M = 18.02 minutes, SD = 26.38). The high standard deviations indicate that some people were much more frequent users of news media than others. The group also watched an average of 3.12 hours of television a day (SD = 3.51). The results from the two White and Black conditions were first compared to check for any intrarace differences. Chi-square tests of independence between the conditions on the LIB revealed no significant differences in how participants reacted within the two racial conditions (White 1 and White 2: x2 = 1.18, df = 3, ns, Cramer’s V = .12; Black 1 and Black 2: x2 = 2.80, df = 3, ns, Cramer’s V = .16), so the data from the two White suspect conditions and the two Black suspect conditions were collapsed into single White and Black conditions. These two conditions were then compared, and there were significant differences in the responses between participants who saw a White suspect and a Black suspect for both concrete and abstract descriptors (x2 = 17.36, df = 3, p = .001, Cramer’s V = .29). A greater proportion of respondents (59.6%) who saw the White suspect chose the concrete (IAV) description (‘‘the man who police want to talk to probably hurt the victims’’) compared to those who saw a Black suspect (33.6%). A significantly larger percentage of those who saw the Black suspect (58.4%) chose the abstract (Adj) description (‘‘the man who police want to talk to is probably violent’’) compared to those who saw a White suspect (31.5%). Thus, the language participants chose was more abstract following exposure to an African American suspect, and H1 is supported. To test for a relationship between news media use and the LIB, participants were categorized as light (or infrequent) or heavy (or frequent) users of each news medium 300

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

Table 1 Cross-Tabulation of Linguistic Level of Response by Race of Suspect, Controlling for TV News Use Medium

Media Race of Percent Choosing Use Suspect Linguistic Level DAV IAV

Local TV news

Light

White Black Heavy White Black Network Light White TV news Black Heavy White Black Cable TV news Light White Black Heavy White Black

2.2 5.6 0.0 1.7 2.2 3.9 0.0 3.2 2.1 1.8 0.0 5.4

SV

N

x2

p

Cramers V

Adj

60.0 2.2 35.6 45 4.79 ns 38.9 5.6 50.0 54 59.1 13.6 27.3 44 17.36 .001 28.8 3.4 66.1 59 64.4 4.4 28.9 45 4.41 ns 43.1 5.9 47.1 51 54.5 11.4 34.1 44 15.05 .002 25.8 3.2 67.7 62 59.6 4.3 34.0 47 6.43 ns 35.1 7.0 56.1 57 59.5 11.9 28.6 42 15.65 .001 32.1 1.8 60.7 56

.22 .41 .21 .38 .25 .40

Note: df = 3 for all comparisons. Linguistic level refers to answers to the question ‘‘The man police want to talk to probably .’’ DAV = hit the victims; IAV = hurt the victims; SV = hated the victims; Adj = is violent.

based on median splits of their responses to the appropriate items. Chi-squares were then run on each of the groups using cross-tabulations. As Table 1 highlights, news media use is indeed related to the presence of the LIB: For each of the television news variables, the chi-square of the LIB responses was not statistically significant for the light users but was for heavy users, supporting the notion that frequent users of TV news are more likely to exhibit the LIB than infrequent viewers of TV news. Thus, H2, H3, and H4 are supported. As for the research questions, chi-square tests also revealed that heavy users of television are more likely to engage in the LIB than light users of television, suggesting that cultivation may indeed be at work (see Table 2). Newspaper use also showed a significant relationship with LIB: Increased use of the local newspaper was associated with more abstract descriptions following the Black suspect compared to the White suspect. A different pattern emerged for use of news Web sites, however. Infrequent users of Web news exhibited the LIB a bit more strongly than those who used news Web sites more frequently. However, because both chisquares are significant, Web site use does not impact the LIB, suggesting that the Web is the only news medium of those tested here that has no relationship to the development of the LIB. One last result is worth noting, although it was not the subject of a hypothesis or research question. When participants’ scores on the Attitudes Toward Blacks scale were dichotomized into high- and low-prejudiced groups and compared to the LIB Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

301

Linguistic Intergroup Bias

B. W. Gorham

Table 2 Cross-Tabulation of Linguistic Level of Response by Race of Suspect, Controlling for Media Use Medium

Overall TV

Media Use Light Heavy

Newspaper

Light Heavy

Web news

Light Heavy

Race of Suspect White Black White Black White Black White Black White Black White Black

Percent Choosing Linguistic Level DAV

IAV

SV

Adj

2.2 1.7 0.0 5.7 1.9 3.7 0.0 3.4 0.0 6.0 2.3 0.0

53.3 38.3 65.9 28.3 63.5 42.6 54.1 25.4 60.9 34.3 58.1 32.6

2.2 5.0 13.6 3.8 5.8 5.6 10.8 3.4 8.7 1.5 7.0 8.7

42.2 55.0 20.5 62.3 28.8 48.1 35.1 67.8 30.4 58.2 32.6 58.7

N

x2

p

Cramers V

45 60 44 53 52 54 37 59 46 67 43 46

2.70

ns

.16

22.53

.001

.48

5.03

ns

.22

12.76

.005

.37

14.69

.002

.36

7.67

.05

.29

Note: df = 3 for all comparisons. Linguistic level refers to answers to the question ‘‘The man police want to talk to probably .’’ DAV = hit the victims; IAV = hurt the victims; SV = hated the victims; Adj = is violent.

responses, there was no difference between the two groups (low prejudice: x2 = 11.71, df = 3, p , .01, Cramer’s V = 34; high prejudice: x2 = 10.08, df = 3, p = .02, Cramer’s V = .31). Both groups exhibited the LIB, meaning that how one scored on the prejudice scale did not have an impact on how one answered the LIB. This is in line with previous findings (von Hippel et al., 1997) that explicit prejudice measures will show no relationship with implicit measures such as the LIB and hence reflect the cognitive biases that occur outside of conscious awareness or control. Discussion

This study examined the LIB, which is the use of either more concrete or more abstract language based on the behavior and the group membership of the subject being described. The LIB is considered an implicit marker of group differentiation and stereotyping, especially in the context of behaviors that conform to the stereotype for that particular group. Using an experimental design where the race of a suspect in a crime story was manipulated, this study found that White adults were more likely to endorse more abstract descriptions of an African American suspect and use more concrete descriptors for a White suspect. Furthermore, this tendency was related to news media use, such that people who more frequently watch television news, overall television, and read the newspaper are more likely to exhibit the LIB. The results support the idea that race-related news stories will prime dominant stereotypes of race and that these stereotypes, once triggered, guide interpretation in 302

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

ways that may help cultivate dominant views of race. A greater proportion of the participants who viewed an African American suspect in a television news story about unsolved violent crimes chose more abstract language to describe the man’s role in the story compared to participants who viewed a White suspect. The language used to describe the White suspect, in contrast, suggests a more limited bounded action rather than an inherent state of being. The interpretations viewers gave of the suspect were more abstract—and thus more dispositional, less situational, and reflecting more of an understanding that the individual represents his group—when that suspect was an African American than when he was White. Given research about the attribution error and how people systematically tend to favor the in-group over the out-group in their perceptions and thought processes, perhaps these White participants thought the more abstract descriptions were more ‘‘natural’’ and fit the story better than the more concrete ones when that suspect was Black, thus revealing the operation of dominant stereotypes at work. It is important to keep in mind that these results do not necessarily reflect conscious thought processes. Instead, the results suggest that such biased processing is likely going on without the awareness or control of the participants, given that there was no relationship between the LIB and an explicit measure of prejudice. The experiment was designed to afford participants a degree of anonymity so that they would feel comfortable expressing their true feelings: Respondents were never asked for their names, were paid in cash, and were instructed to place their surveys in an envelope and place the envelope in a box away from the research assistant. Although the group of university staff was largely low in prejudice according to their ATB scores—on the ATB scale, where the range of possible scores is from 20 (very negative attitudes toward Blacks, thus very high prejudice) to 140 (very positive attitudes toward Blacks, thus low prejudice), this group had a mean of 109 and a median of 111—there were people nonetheless who scored in the high-prejudice area of the scale. Finding the LIB statistically separate from the explicit prejudice measure further supports the idea that biased processing is occurring without the conscious awareness or control of the participants. Instead, it would appear that these respondents were cognitively inclined to see these social groups as being different and to see being violent as an expected behavior for African Americans. Thus, the automatic activation of stereotypical schema for African Americans influences the language processes of the respondents, even if they would not consciously endorse the stereotypes. Although this study cannot speak to the causal nature of the relationship between long-term media use and the perception of social groups, it does find a necessary condition for cultivation: a correlation between media-use variables and responses that support dominant ideology. Frequent use of local, network, and cable television news, as well as frequent use of the local newspaper, were all associated with the presence of the LIB in ways that support negative stereotypes of African Americans. Furthermore, overall television use was also associated with the LIB, perhaps suggesting that despite the plethora of channels and choices in today’s media environment, the symbolic message that African Americans are violent is nonetheless hard to escape. Social Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

303

Linguistic Intergroup Bias

B. W. Gorham

psychologists often suggest that the social world is awash in stereotypical portrayals of African Americans, and this observation is supported by content analyses of media content, particularly news. The argument advanced by Devine (1989) and others is that such images of the cultural stereotypes help develop, maintain, and reinforce schema that reflect the cultural stereotypes and that these stereotypes then influence information processing. This study demonstrated that implicit, stereotype-congruent responses can be primed by media images and that these responses are more likely for people who consume more television and television news. Although they do not offer any evidence on the direction of causality, these results do offer more pieces of the puzzle as to how media might cultivate dominant views of race. At the same time, use of the web for news did not have an impact on the LIB. Although it is difficult to draw any sound conclusions from this without knowing more details about which Web sites the participants visit, perhaps this finding denotes a difference between national and local print-oriented media. ‘‘If it bleeds, it leads’’ is arguably a mantra for all news outlets, but it seems to be much more the purview of local media. Perhaps the Web sites of large news organizations with national reputations do not feature as much crime news about African Americans as more local media, or perhaps people are not exposed to those stories to the same extent as they are with other media. It cannot be determined from this data what is different about Web news, but it does seem noteworthy that this news medium did not show the same relationship with the LIB. Beyond the lack of detail in the media-use measures (it would be interesting, e.g., to compare which cable news networks people primarily watched), there are other limitations in the study that future research should note. Perhaps the patterns would be even stronger if the potential in-group member were not also being constructed as a criminal, that is, perhaps the relevant out-group for some of the of these participants was ‘‘criminals’’ and not necessarily ‘‘African Americans’’ or even ‘‘African American criminals.’’ For some of these largely low-prejudiced working- and middleclass Whites, perhaps the suspect’s race was trumped by the fact that he was a suspected criminal. He would therefore be considered outside of one’s in-group, thus depressing the overall influence of the stereotype on the LIB. Although the fact that the sample was made up of university staff is perhaps a step up from the use of undergraduates, one should nonetheless exercise caution in generalizing beyond this sample. It was a self-selected sample, of course, but it was also an overwhelmingly female one. Although there is no reason to necessarily think that females somehow process information from television news stories differently than males, it is known that women and men are not equal in their exposure to news media. Thus, perhaps the patterns noted are unduly influenced by the female majority of this sample. As relevant as news stories about crime are for studying White’s cognitive reactions to African Americans, future research should think about applying the theories to research designs beyond the criminal paradigm and beyond the strict boundaries of Black and White. Although I am just as guilty of this as the next researcher, studies 304

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

such as this one that use ‘‘Black crime’’ as a research stimulus for Whites may well be playing into the maintenance of the very schema we hope to challenge. Athleticism is also a prominent trait in the stereotype of African Americans (Devine & Elliot, 1995), for example, so perhaps we would find the LIB in a comparison of stories featuring Black and White athletes. Future research should seek to broaden the potential pool of participants used in the research as well as the topics used in the stimulus. Despite these limitations, this study suggests that the LIB should be the focus of more attention by researchers interested in the effects of media content, particularly those interested in how media may implicitly work to cultivate certain views of the world. In addition to research about race, the LIB could highlight important differences in how audiences react to increasingly polarized political media content. Perhaps the LIB could tease out the social constructions audiences have of ‘‘liberals’’ and ‘‘conservatives’’ and highlight the types of content that prime political intergroup processing. In any case, the LIB is an implicit and unobtrusive way that language can reflect the dominant ideology of race. Given the complex ways in which people can maintain both sympathetic, low-prejudiced views of minorities while simultaneously harboring feelings of discomfort (Dovidio & Gaertner, 1986) or make interpretations that are congruent with stereotypes, despite their conscious egalitarian beliefs (Devine, 1989; Monteith, 1993), uncovering more of the subtle ways that media influence our thoughts about race is a positive step toward being able to intervene in the process. Only by understanding the many ways in which viewers apply meanings to media content about race, can we hope to use mass media to further social justice and equality. Acknowledgments

A research grant from the Newhouse School was used to fund this research. The author thanks Dave Kurpius for his help in securing the stimulus tapes and Jamie Butler for her work in securing participants. Note 1

A series of items based on von Hippel et al.’s (1997) study was also used, in which participants were asked to individually rate how well the phrase from each linguistic level described the man’s role in the story. The results from these items were not statistically significant and are not reported here. It may be that participants were thrown off given that every other story was asked about only using a forced-choice measure.

References Allport, G.W. (1954/1979). The nature of prejudice (25th anniversary ed.). Reading, MA: Addison-Wesley. Bennett, S. E. (1989). Trends in Americans’ political information. American Politics Quarterly, 17, 422–435. Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

305

Linguistic Intergroup Bias

B. W. Gorham

Berry, D. S., & Wero, J. L. F. (1993). Accuracy in face perception: A view from ecological psychology. Journal of Personality, 61, 497–520. Brigham, J. C. (1993). College students’ racial attitudes. Journal of Applied Social Psychology, 23, 1933–1967. Crocker, J., & Luhtanen, R. (1990). Collective self-esteem and ingroup bias. Journal of Personality and Social Psychology, 58(1), 60–67. Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal of Personality and Social Psychology, 56(1), 5–18. Devine, P. G., & Elliot, A. J. (1995). Are racial stereotypes really fading? The Princeton trilogy revisited. Personality and Social Psychology Bulletin, 21, 1139–1150. Dixon, T. L., & Linz, D. (2000a). Overrepresentation and underrepresentation of African Americans and Latinos as lawbreakers on television news. Journal of Communication, 50(2), 131–154. Dixon, T. L., & Linz, D. (2000b). Race and misrepresentation of victimization on local television news. Communication Research, 27, 547–573. Dovidio, J. F., & Gaertner, S. L. (1986). Prejudice, discrimination, and racism: Historical trends and contemporary approaches. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 1–34). New York: Academic Press. Entman, R. M. (1990). Modern racism and the images of blacks in local television news. Critical Studies in Mass Communication, 7, 332–345. Entman, R. M. (1992). Blacks in the news: Television, modern racism, and cultural change. Journalism Quarterly, 69, 341–361. Gaertner, S. L., & Dovidio, J. F. (1986). The aversive form of racism. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 61–90). New York: Academic Press. Gerbner, G., Gross, L., Morgan, M., Signorielli, N., & Shanahan, J. (2002). Growing up with television: Cultivation processes. In J. Bryant & D. Zillmann, (Eds.), Media effects: Advances in theory and research (2nd ed., pp. 43–68). Mahwah, NJ: Erlbaum. Gilliam, F. D., Jr., Iyengar, S., Simon, A., & Wright, O. (1996). Crime in black and white: The violent, scary world of local news. Press/Politics, 1(3), 6–23. Gorham, B. W. (1999). Stereotypes in the media: So what? The Howard Journal of Communications, 10, 229–247. Gorham, B. W. (2002, August). The linguistic intergroup bias in interpretations of a race-related crime story. Paper presented at the Association for Education in Journalism and Mass Communication, Miami, FL. Graesser, A. C., Gernsbacher, M. A., & Goldman, S. R. (1997). Cognition. In T. A. van Dijk (Ed.), Discourse as structure and process (pp. 292–319). London: Sage. Greenberg, B. S., Mastro, D., & Brand, J. E. (2002). Minorities and the mass media: Television into the 21st Century. In J. Bryant & D. Zillmann (Eds.), Media effects: Advances in theory and research (2nd ed., pp. 333–352). Mahwah, NJ: Erlbaum. Hawkins, R. P., & Pingree, S. (1990). Divergent psychological processes in constructing social reality from mass media content. In N. Signorielli & M. Morgan (Eds.), Cultivation analysis (pp. 35–50). Beverly Hills, CA: Sage. Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley. 306

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

B. W. Gorham

Linguistic Intergroup Bias

Hewstone, M. (1990). ‘‘The ultimate attribution error’’? A review of the literature on intergroup causal attribution. European Journal of Social Psychology, 20, 311–335. Jamieson, K. (1992). Dirty politics. New York: Oxford University Press. Johnson, J. D., Adams, M. S., Hall, W., & Ashburn, L. (1997). Race, media and violence: Differential racial effects of exposure to violent news stories. Basic and Applied Social Psychology, 19(1), 81–90. Karpinski, A., & von Hippel, W. (1996). The role of the linguistic intergroup bias in expectancy maintenance. Social Cognition, 14(2), 141–163. Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. Chicago: The University of Chicago Press. Linville, P. W., Salovey, P., & Fischer, G. W. (1986). Stereotyping and perceived distributions of social characteristics: An application to ingroup-outgroup perception. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 165–208). New York: Academic Press. Lippmann, W. (1922). Public opinion. New York: MacMillan. Maass, A., Milesi, A., Zabbini, S., & Stahlberg, D. (1995). Linguistic intergroup bias: Differential expectancies or in-group protection? Journal of Personality and Social Psychology, 68(1), 116–126. Maass, A., Salvi, D., Arcuri, L., & Semin, G. (1989). Language use in intergroup contexts: The linguistic intergroup bias. Journal of Personality and Social Psychology, 57, 981–993. McConahay, J. B. (1986). Modern racism, ambivalence, and the modern racism scale. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 91–126). New York: Academic Press. Montagu, A. (1997). Man’s most dangerous myth: The fallacy of race (6th ed.). Walnut Creek, CA: Altamira Press. Monteith, M. (1993). Self-regulation of prejudiced responses: Implications for progress in prejudice-reduction efforts. Journal of Personality and Social Psychology, 65, 469–485. Morgan, M., & Shanahan, J. (1997). Two decades of cultivation research: An appraisal and meta-analysis. Communication Yearbook, 20, 1–45. Oliver, M. B., Jackson, R. L., II, Moses, N. N., & Dangerfield, C. L. (2004). The face of crime: Viewers’ memory of race-related facial features of individuals pictured in the news. Journal of Communication, 54(1), 88–104. Perdue, C. W., Dovidio, J. F., Gurtman, M. B., & Tyler, R. B. (1990). Us and them: Social categorization and the process of intergroup bias. Journal of Personality and Social Psychology, 59, 475–486. Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport’s cognitive analysis of prejudice. Personality and Social Psychology Bulletin, 5, 461–476. Pingree, S., Hawkins, R. P., Hitchon, J., Gilligan, E., Radler, B., Kahlor, L, et al. (2001). If college students are appointment TV viewers. . Journal of Broadcasting and Electronic Media, 45, 446–463. Potter, W. J. (1994). Cultivation theory and research: A methodological critique. Journalism Monographs, 147, 1–35. Rhodes, G. (1988). Looking at faces: First-order and second-order features as determinants of facial appearance. Perception, 17, 43–63. Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association

307

Linguistic Intergroup Bias

B. W. Gorham

Romer, D., Jamieson, K. H., & de Coteau, N. J. (1998). The treatment of persons of color in local television news. Communication Research, 25, 286–305. Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categorization (pp. 27–48). Hillsdale, NJ: Erlbaum. Schnake, S. B., & Ruscher, J. B. (1998). Modern racism as a predictor of the linguistic intergroup bias. Journal of Language and Social Psychology, 17, 484–491. Semin, G. R., & Fiedler, K. (1988). The cognitive functions of linguistic categories in describing persons: Social cognition and language. Journal of Personality and Social Psychology, 54, 558–568. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), The Psychology of intergroup relations (2nd ed., pp. 7–24). Chicago: Nelson-Hall. van Dijk, T. A. (1987). Communicating racism: Ethnic prejudice in thought and talk. Newbury Park, CA: Sage. van Dijk, T. A. (1997). The study of discourse. In T. A. van Dijk (Ed.), Discourse as structure and process (pp. 1–34). London: Sage. von Hippel, W., Sekaquaptewa, D., & Vargas, P. (1997). The linguistic intergroup bias as an implicit indicator of prejudice. Journal of Experimental Social Psychology, 33, 490–509. Wittenbrink, B., Hilton, J. L., & Gist, P. L. (1998). In search of similarity: Stereotypes as naive theories in social categorization. Social Cognition, 16(1), 31–55.

308

Journal of Communication 56 (2006) 289–308 ª 2006 International Communication Association