Skip to main content

Journal of Hispanic Policy

Topic / Media

Toward an Empirical Analysis of Hate Speech on Commercial Talk Radio


This pilot study uses qualitative content analysis to examine hate speech that targets vulnerable groups, including ethnic, racial, religious, and/or sexual minorities, in commercial broadcasting. The study quantifies a recurring rhetorical pattern for targeting specific vulnerable groups through the systematic use of unsubstantiated claims, divisive language, and nativist code words. For example, Latino immigrants were often coded as criminals and then linked to social institutions that were presented as complicit with immigrants. In this way, target groups were characterized as a powerful and direct threat to the nation. While vulnerable groups are targeted, calls for action from talk radio are then directed against those identified as supporters of these vulnerable groups.


The considerable and often heated debate over hate speech has produced numerous reports, articles, and books. These studies have looked at the issue from a number of disciplinary perspectives, including those of journalism, law, linguistics, economics, history, and philosophy (Butler 1997; Cortese 2006; Dharmapala and McAdams 2003; Kellow and Steeves 1998; Lendman 2006; Lewis 2007; Meddaugh and Kay 2009; Neiwert 2009; O’Connor 2008; Slagle 2009; Tolmach Lakoff 2001). These studies offer valuable theoretical, conceptual, interpretive, and descriptive insights into hate speech, but they often rest upon unsubstantiated empirical premises about the phenomenon itself. To date, there is limited research on hate speech using scientific approaches to medium, content, and impact. The main goal of this pilot study is to develop a sound, replicable methodology for qualitative content analysis that can be used to examine hate speech that targets vulnerable groups, including ethnic, racial, religious, and/or sexual minorities, in commercial broadcasting. This pilot study establishes data-driven descriptive categories for such speech and creates a preliminary baseline or reference point for future research.

The backdrop for this study is the 1993 National Telecommunications and Information Administration (NTIA) Report to Congress, which addressed the role of telecommunications in the commission of hate crimes. The NTIA advises the president on telecommunications and information policy and manages the federal government’s use of the radio frequency spectrum. Mindful of First Amendment protections as well as related federal legislation and policy, the 1993 NTIA report established a definition of hate speech drawn from the Hate Crimes Statistics Act of 1990. Now, two decades later, the NTIA report continues to provide a viable definition for hate speech, but it no longer reflects significant recent changes in federal policy, telecommunications platforms, and programming formats and content. Furthermore, the original study relied on data that was, by the NTIA’s own account, “scattered and largely anecdotal,” and it therefore failed to provide a scientific basis for data assessment, let alone a methodology or baseline for future study.

In developing this pilot study, we considered areas in which we expected to see significant results so as to establish and test data-driven descriptive categories. Future full-scale analysis would need to include a comparative dimension.

Commercial talk radio is the focus of this pilot study. Radio has the greatest penetration of any media outlet (print, broadcast, or digital), reaching 90 percent of Americans each week, and the news-talk format is the predominant radio format in terms of dedicated stations nationwide (more than 1,700) and the second most popular format in terms of audience share (12.1 percent; country music is 13.3 percent) (Houston Santhanam 2012). We examined commercial radio talk programs reaching audiences in Los Angeles County because it is the most populous county in the United States and because Latinos made up nearly half—48 percent—of the county’s population in 2011 (U.S. Census Bureau 2011).

As the fastest-growing and largest minority group in the United States, Latinos represented 16.7 percent of the U.S. population, or about 51.9 million people, in 2011 (U.S. Census Bureau 2011). Noncitizens make up 44 percent of the adult Latino population, of which 55 percent is undocumented (Pew Hispanic Center 2007). Nationally, hate crimes against Latinos, when compared with hate crimes against other racial/ethnic groups, have risen by the highest rate, with a 25 percent increase between 2004 and 2008 (Federal Bureau of Investigation 2004; Federal Bureau of Investigation 2008). This increase may be linked to the media-generated negative discourse against immigrants that has been prevalent on the airwaves. In a 2007 national survey, about 64 percent of U.S. Latinos reported that the immigration debate had negatively impacted their lives, while 78 percent reported feeling that discrimination remained a problem affecting their community (Pew Hispanic Center 2007).


The 1993 report was the result of the NTIA’s mandate to examine and report to Congress on the media’s role in “crimes of hate and violent acts against ethnic, religious, and racial minorities.” The report defined hate speech as either: (1) “words that threaten to incite ‘imminent unlawful action,’ which may be criminalized without violating the First Amendment”; or (2) “speech that creates a climate of hate or prejudice, which may in turn foster the commission of hate crimes” (U.S. Department of Commerce 1993). The definition of hate speech used in this pilot study is derived from this definition as well as the definition used in the hate crimes legislation: hate speech is speech that targets a vulnerable group and threatens or fosters the commission of hate crimes against that group, as defined by law.

Inclusion Criteria

The 1993 NTIA report considered all telecommunication at that time: “broadcast television and radio, cable television, public access television, computer bulletin boards, and other electronic media” (U.S. Department of Commerce 1993). For the purposes of this pilot study, we focused our attention on samples from one medium (radio) and one programming format (news-talk) in one market (Los Angeles County). We decided to look specifically at conservative talk radio, which accounts for 91 percent of total weekday talk radio programming (Halpin et al. 2007). This allowed us to consider exemplary instances with respect to media penetration, a predominant format, and a large and diverse market. Furthermore, by examining the news-talk format, program content could also be measured against established professional journalistic standards, specifically, the Code of Ethics developed by the Society of Professional Journalists (1996).

Segments of thirty to forty minutes were selected from each of three programs: The Lou Dobbs Show: Mr. Independent (syndicated by the United Stations Radio Networks), broadcast on 31 July 2008; The Savage Nation (produced at KFMB 760 AM and syndicated by Talk Radio Network), broadcast on 24 July 2008; and The John & Ken Show (KFI AM 640, Los Angeles), broadcast on 30 July 2008. These programs share certain general features of the news-talk format (news commentary, guest interviews) and focus on conservative topics (anti-immigration and free speech). At the same time, each program has a distinct profile: The Lou Dobbs Show is an example of a program featuring a high-profile media personality who has access to multiple traditional media platforms (at the time of the broadcast, Dobbs hosted radio and television shows). The Savage Nation is a prominent example of popular syndicated talk radio. The John & Ken Show represents successful local market news-talk radio (it is also syndicated nationally).

We selected the broadcast segments shortly after the start of a project grant from the Social Science Research Council. This coincided with the controversy surrounding San Francisco’s status as a sanctuary city for undocumented immigrants. Not surprisingly, this issue is reflected in the transcripts; nevertheless, we analyzed speech targeting any vulnerable group (as defined by the 1993 NTIA report).

Qualitative Content Analysis

In this study, we employed conventional qualitative content analysis, also known as inductive category development, whereby we derived coding categories directly from the textual data, in this case, transcripts of the program segments (Mayring 2000). This approach uses delineated and replicable methodologies that allow the generation of inferences from a given text without being bound to inflexible quantification. Conventional qualitative content analysis is ideal for areas of study supported by little theoretical or research literature (Kondracki and Wellman 2002), as is the case for empirical studies of hate speech.

Conventional qualitative content analysis follows an iterative process, beginning with the repeated reading of the data (transcripts) and the formation of the coding process. Category development involves the derivation of criteria from the data, based on background considerations and research objectives (Barrett 2007). Materials are analyzed for patterns, for which descriptive codes are developed; these patterns may indicate the presence of larger themes. As the analysis progresses, categories are either revised or removed based on their frequency and reliability. The inductive process of category development is followed by the deductive process of category application (Mayring 2000). This involves assigning category definitions within a coding agenda, in essence defining how texts should be coded with a category. Data findings and interpretations are related to pertinent research and literature and, as relevant, to common experience (Barrett 2007).

Establishment of Analytic Categories

Trained readers (undergraduate and graduate students), working in conjunction with the investigators, examined the transcripts for each of the three program segments. Emphasis was placed on identifying the relationship between speakers and targets—basically, who said what, about or to whom, and for what purpose. This allowed the research team to identify targets—vulnerable groups and/or their supporters—through implied and named (specific) references to them as well as through a speaker’s call for action against them (that is, suggesting or implying that listeners might do something that could affect the target).

The research team then identified four types of statements that were made relative to these targets: unsubstantiated claims, flawed argumentation, divisive language (that is, “us-them” constructions), and dehumanizing metaphors. Utterances could be categorized, simultaneously, within two or more of these categories. The readers then developed a fifth analytical category for indexicality, wherein a word (or other sign) points to a context-dependent meaning. Given the labor-intensive nature of coding for indexicality, the readers focused on a ten-minute sample from one of the program segments (The John & Ken Show), adapting and using open source software (Transana) for both audio and transcript analysis. The readers focused on indexical terms, or code words, that pointed to a nativist attitude on the part of the speaker, then determined which indexical terms were used most frequently. Readers also identified patterns of rhythm, stress, and intonation (prosody) and discourse alignment among speakers. The four most recurrent indices in the ten-minute segment of The John & Ken Show were then used to analyze the three transcripts.

The preliminary findings provided data for all three programs in six categories: targeted statements, unsubstantiated claims, flawed argumentation (with a focus on fallacies), divisive language (deixis), dehumanizing metaphors, and selected indexical terms (indices for nativism). To ensure a robust methodology, we reexamined these six categories, first in relation to the transcripts and then in relation to one another. The first step yielded some corrections with respect to the coding process. In comparing the findings across categories, we noted a certain degree of overlap. In some instances this revealed how particular rhetorical strategies resonated with one another, but in other instances it merely produced redundant findings.

Category Refinement

Two categories—dehumanizing metaphors and flawed argumentation—raised particular concerns given the redundancy of their findings with other approaches, the expertise required for credible analysis, and the contention among scholars in each area with respect to methodology and interpretation. Our main concern had to do with whether these categories contributed to the project research objectives by generating reliable findings and a replicable methodology. The interdisciplinary nature of the project and the volatile nature of public debate over hate speech, not to mention practical considerations with respect to the limited resources for full-scale research, required a methodology that could be implemented or replicated by nonexperts.

The inherent difficulty of reliably identifying formal arguments (that is, statements that make and present evidence for a claim) in natural-language contexts and the inability of formal logic to adequately evaluate natural-language argument (for example, identifying logical fallacies to invalidate a claim) led us to reconsider an approach in this direction (Hahn et al. 2009).

As Trudy Govier cautions, “In practice it is often difficult to tell whether people are offering arguments or not, and whenever this interpretive issue is contestable, a comment to the effect that a fallacy has been committed will be similarly contestable” (1982, 6). Although the research team could identify and reach a consensus about “traditional fallacies” in the transcripts (particularly, ad hominem attacks), it proved much more difficult to connect them to a corresponding argument. Doing so depended on the charity of the interpreter, who might fill in premises needed to establish a formal argument (Finocchiaro 1981). We found that another analytical category used in our preliminary findings, unsubstantiated claims, provided a more productive approach, one in which speech targeted at vulnerable groups could be assessed through a standard fact-checking methodology.

Initially, metaphor analysis provided a compelling framework by which we could measure the extent to which the radio programs dehumanized vulnerable groups by establishing the sameness between two unrelated things or ideas. Phrases such as “love is a rose,” “the ship of state,” or “immigrants are a virus” are metaphors that facilitate an understanding of one thing (love, nation-states, immigrants) in the terms of another (flowers, ships on an ocean, disease). This type of analysis has already generated considerable insight into the media depiction of Latinos. In Brown Tide Rising: Metaphors of Latinos in Contemporary American Public Discourse, Otto Santa Ana (2002) provides an empirical analysis of the ways in which the mainstream, and ostensibly liberal, press uses metaphors to characterize immigrants in nonhuman terms, such as a dangerous threat, a virulent disease, an invasion, or an animal-like force. As Santa Ana notes, “These metaphors are not merely rhetorical flourishes, but are the key components with which the public’s concept of Latinos is edified, reinforced, and articulated” (2002, xvi). Cognitive linguists emphasize that metaphors can also operate on a conceptual level. If a linguistic metaphor uses one thing or idea to understand another, a conceptual metaphor uses one “coherent organization of experience” to understand another, as with “life is a journey,” wherein the experience of life is understood through the metaphor of travel toward a destination (and, hence, a purpose) (Kövecses et al. 2010, 4). Thus, the study of linguistic and conceptual metaphors has the potential to reveal the rhetorical strategies and underlying conceptual systems by which vulnerable groups are understood and perhaps even acted against.

In the preliminary analysis for our study, the readers used a consensus methodology to identify linguistic and conceptual metaphors directed against vulnerable groups. These were then reviewed, and only those metaphors that dehumanized members of a vulnerable group were selected. The preliminary findings proved suggestive and resonated with Santa Ana’s findings (describing, for example, Latinos as a threat, disease, or animals) while also revealing other metaphoric constructions used to describe not only vulnerable groups but also race (using terms such as “card game,” “nature,” and “criminals”). Given the small sample, however, we identified only one significant and recurring pattern with respect to dehumanizing metaphors: the repeated use of the terms “illegal alien” or “illegal aliens” and “illegal” or “illegals” to describe immigrants. These terms were also identified in our examination of indices for nativism, wherein they code immigrants as antithetical to the nation. We found the latter approach more productive insofar as it provides a comparative framework with two sets of indices for an external enemy (“anarchist” and terms including “illegal”) and home country (terms referring to community and free speech).


Targeted Statements

Drawing from the 1993 NTIA report and the Hate Crimes Statistic Act, this pilot study foregrounds two features that are crucial to the report’s definition of hate speech: a vulnerable group as the target and speech that threatens or fosters the commission of hate crimes against that group, as defined by law. In that context, vulnerable groups are defined as ethnic, racial, religious, and/or sexual minorities. We included undocumented immigrants insofar as they are associated with an ethnic group (Latinos) in the transcripts. We also gathered data on calls for action against those identified as supporters of vulnerable groups.


Readers identified statements in the transcripts that were targeted at vulnerable groups, then distributed the statements into three categories: implied target, which does not explicitly identify a member of a vulnerable group but the intent is clear; named target, which specifically identifies a member of a vulnerable group; and call for action, in which hosts suggest or imply that an action might be taken against the vulnerable group (see Tables 1 and 2).


Readers identified 148 instances that met the study’s criteria for statements targeting a vulnerable group or a group’s supporters. Seventy-nine percent of these instances (117) targeted vulnerable groups, and 21 percent (31) targeted their supporters. Across the three programs, readers identified thirty-three instances of call for action.

Just over two-thirds of targeted statements focused on undocumented immigrants and Latinos (73 of 117, which includes 4 of 28 instances related to people of color in public office). Averaged on a per-program basis, Latinos (both citizen and undocumented) represented 91 percent (43 of 47, including those in public office) of the targeted vulnerable groups on The Lou Dobbs Show; 43 percent (15 of 35) on The Savage Nation; and 43 percent (15 of 35) on The John & Ken Show. The figure for The John & Ken Show is actually higher, since 34 percent (12 of 35) of the targeted statements in this broadcast segment focused on the residents of “South L.A.” (South Central Los Angeles), an area that is roughly 55 percent Latino and 41 percent African American.

Readers identified two calls for action against a vulnerable group: one was a general call related to immigration and the other focused on people of color in public office who supported immigration reform. Insofar as both were oriented toward the political representation system, each might also have been identified as a call for action against supporters. The other thirty-one calls for action clearly focused on supporters: these were specific elected officials, advocacy groups (ANSWER Coalition, Media Matters for America), and employers of undocumented immigrants. This suggests a rhetorical strategy in which vulnerable groups are targeted and identified as a social problem or threat but the call for action is directed against advocacy groups, public figures (and political administrations), or legal enforcement.

Twenty-six of the thirty-three calls for action focused on Latinos and immigration. Seven related to a report by Media Matters for America that criticized the host’s statements about autism, gays, and Democrats as fascists (The Savage Nation).

Unsubstantiated Claims

The assertion of false, unverifiable, and/or distorted claims emerged as a significant feature of all the segments we analyzed. This finding is important insofar as news-talk programming is presented within a general journalistic framework that is associated with fact-based news commentary and expert-driven interviews on topical issues. The relevant professional organizations—Radio and Television News Directors Association, the International Federation of Journalists, and the Society of Professional Journalists (SPJ)—underscore the importance of both freedom of the press and ethical journalism. The SPJ Code of Ethics (1996) includes the following professional standards for journalists:

* Test the accuracy of information from all sources and exercise care to avoid inadvertent error. Deliberate distortion is never permissible.

* Tell the story of the diversity and magnitude of the human experience boldly, even when it is unpopular to do so.

* Examine one’s own cultural values and avoid imposing those values on others.

* Avoid stereotyping by race, gender, age, religion, ethnicity, geography, sexual orientation, disability, physical appearance, or social status.

* Support the open exchange of views, even views one finds repugnant.

* Distinguish between advocacy and news reporting. Analysis and commentary should be labeled and not misrepresent fact or context.

* Admit mistakes and correct them promptly.

The code is a set of guidelines, as the SPJ notes: “The code is intended not as a set of ‘rules’ but as a resource for ethical decision-making. It is not—nor can it be under the First Amendment—legally enforceable.” As with other professional organizations, membership signals adoption of the standards and thereby also provides a basis for identifying noncompliance with widely held professional standards. In contrast to other professional organizations, however, the SPJ itself does not enforce its code. Instead, it “encourage[s] the exposure of unethical journalism” and notes that its code “is widely consulted and applied in newsrooms and classrooms as the definitive statement of our profession’s highest values and a helpful way to think about the specific and unique journalism quandaries we confront daily.”

While it is beyond the scope of this study to resolve the ongoing debate over the relationship of news-talk radio to journalism and professional codes of behavior, we do note that Lou Dobbs has served on the board of the Society of Professional Journalists. In addition, all three shows analyzed here adopt a fact-finding, truth-exposing stance with regard to their programming content. While often imbued with considerable emotion, opinion is nevertheless usually presented as based on fact.


We employed a standard fact-checking methodology for analyzing claims in the transcripts. Each transcript was marked for explicit factual citations (figures, statistics, percentages) and for arguments or assertions with strong factual implications. We included fact-based claims made by guest speakers and interviewees in our analyses, alongside those of the shows’ hosts, since their claims supported positions with which the host agreed. Each statement of fact was cross-checked with credible sources (published articles from academic and national media sources; official, noncommercial institutional Web sites). The claims summarized in Table 3 are those that we were able to disprove or question with reliable evidence. The statements are organized into three degrees of unreliability: false claims, which were proved to be untrue; unverifiable claims, which were based on facts that could not be verified; and distorted claims, which were based on facts that were exaggerated or taken out of context. Each claim was then correlated to a target—a vulnerable group or a supporter—that would be impacted negatively if the unsubstantiated claim were accepted as true.


In the transcripts, readers identified 114 fact-based claims. Of these, 37 percent (42 of 114) were unsubstantiated, with 11 proven false, 18 found to be unverifiable, and 13 found to be distorted. The reliability of fact-based claims varied according to program: claims on The Lou Dobbs Show were 87 percent accurate; claims on The Savage Nation were 53 percent accurate; and claims on The John & Ken Show were 55 percent accurate. In the transcripts, Lou Dobbs made three unsubstantiated claims, with a guest making one in addition. Michael Savage made one false claim but a significant number of unverifiable and distorted claims (eight in each category). John Kobylt and Ken Chiampou made seven false claims, four unverifiable claims, and fiver distorted claims; their guest made a similar number of false and unverifiable claims.

The targets of these unsubstantiated claims were congruent with the vulnerable groups and supporters that are identified in the targeted statements. In The Lou Dobbs Show and The John & Ken Show, the unsubstantiated claims related either entirely or predominantly to undocumented immigrants and governmental agencies or public officials that were characterized as supporting them or facilitating their negative impact on society. The unsubstantiated claims magnified the sense of an immigrant threat (attributed alternately to immigrants as criminals or public officials as accomplices), overstated the effectiveness of the hosts’ preferred immigration policies, and linked immigrant rights advocacy groups to terrorism. In addition to this focus on immigration, The John & Ken Show made eight unsubstantiated claims related to an initiative to ban the construction of new fast-food outlets for one year in South Central Los Angeles, using these claims to discredit a local elected official (Jan Perry), disparage the “Mexican diet,” and portray economically disadvantaged Blacks and Latinos as inherently violent and undeserving of the public’s support. As with the two other programs, The Savage Nation made unsubstantiated claims related to immigration, but it also made unsubstantiated claims with respect to a wider range of targets, including liberals, Democrats, media, and advocacy groups. Other unsubstantiated claims focused on vulnerable groups identified by their race, religion, or sexual orientation.

Divisive Language (Deixis)

In examining media discourse, it is just as important to analyze word choice and how rhetorical effects are used to appeal to listeners as it is to analyze the factual accuracy of statements. For the pilot study, we focused on one particular way that language establishes, maintains, or reinforces in-group status vis-à-vis a targeted out-group: deixis.

In linguistics, deixis refers to words or phrases that require contextual information in order for the reader or listener to grasp the denotational meaning, that is, to understand the referent (who speaks, to whom and of whom, and where and when the speech occurs) for the deictic term or terms (Lyons 1977; Rapaport et al. 1994). For example, the sentence, “And now we don’t like those people over there” includes four terms that require contextual information: “now” (meaning at the present time), “we” (presumably, both the speaker and the addressee), “those people” (a third party previously mentioned), and “over there” (the spatial location of “those people”). In effect, deictic words “point” to specific persons, places, situations, values, ideologies, and/or group ascription in an often unconscious fashion. Deixis frequently occurs between speakers in the same speech community, where members share the same discussion topics, values, and worldviews and can therefore use deictic phrases within their community without being misunderstood.

Because of their capacity to attribute in-group and out-group status, collective pronouns carry a great deictic charge. In fact, they act as influential shapers of perception and social cognition at a very fundamental level, determining group belonging or lack thereof. Research has demonstrated that collective pronouns utilized to indicate in-group and out-group belonging play a powerful role in inter-group bias (Perdue et al. 1990). This is of great significance as perceived members of an in-group are thought to have more positive attributes in comparison to those of an out-group (Brewer 1979). Collective pronouns pointing at in-group and out-group membership may establish evaluative predispositions toward both positive and negative targets, depending on group membership, in a kind of linguistic conditioning (Perdue et al. 1990). Hence, deixis provides an easy and effective tactic at the level of language, rather than at the level of factualness, for talk radio hosts to establish and maintain a cognitive and ideological bond with their audience.


Readers identified pronouns with a deictic function in the transcripts—those that indicated in-group or out-group status—such as “we,” “us,” “ours,” “they,” “them,” and “theirs,” and also “I,” “my,” “me,” “you,” “he,” and “his,” plus pronouns such as “these,” “this,” “those,” and “that” when they had a deictic role. Each transcript was marked for deictic phrases and words. Those that appeared to refer to a sociopolitical, economic, or cultural division were placed in charts for analysis. Most of these listed instances suggested an “us versus them” framework. Thus, each deictic occurrence was linked to an implied or stated in-group, an implied or stated out-group, or the targeted vulnerable group. An explication of the social function of the passages containing deixis provides context for the project as a whole (the results are summarized in Table 4).


Readers identified 37 passages that relied primarily or extensively on deictic phrases. The number of instances were similar across the three programs, with 13 passages identified for The Lou Dobbs Show, 12 for The Savage Nation, and 12 for The John & Ken Show. The deictic phrases used tended to posit an insurmountable sociopolitical, racial, or cultural divide between a show’s audience and targeted vulnerable groups. Fourteen passages focused on vulnerable groups: 7 on immigrants, 1 on sexual minorities, and 6 on the Black and Latino residents of South Central Los Angeles. In another 15 passages, the target was supporters of vulnerable groups (elected officials, advocacy groups, and the media): The Lou Dobbs Show contained 5 of these passages, including 3 that were focused on Latino elected officials; The Savage Nation, 6; and The John & Ken Show, 4.

In addition, three passages in The Lou Dobbs Show and four passages in The Savage Nation fashioned U.S. leaders (President Barack Obama, House Minority Leader Nancy Pelosi, and the Bush administration) as out of touch with the values of the American public. One passage in The John & Ken Show established the wealthy as an out-group given their obsession with physical appearance and health, thereby situating its listeners somewhere between high-income families and low-income communities. The class positioning for appearance (“their slenderness” in wealthy families versus “everybody’s fat” in South Central Los Angeles) and health (“they take care of themselves” versus “they don’t care”) carries strong racial and ethnic overtones for the poor, which the program hosts describe as “that tribe” in “these areas.”

Michael Savage’s statement about a CNN report on gay Iraqis suggests how a speaker can use deictic phrases to aggregate multiple targets around an apparently simple us-versus-them statement. Savage dismissed the report, commenting, “If the first thing they did with their freedom—that’s what American men have died for so they can be gay in Iraq?” In the contrast between “being gay in Iraq” and “American men,” the passage suggests that gay rights are in direct opposition to American values, as measured by military casualties. Savage’s use of “American men” rather than “American troops” emphasizes masculinity as a positive trait, but it also ignores the fact that American women served and died in the Iraq War. Savage’s rhetorical stance is to claim a dismissive lack of interest (“please leave me alone with that already”), but the effect of the passage is to align Iraqis, homosexuals, gay rights, and the liberal media against Savage and his articulation of “freedom” as defended by “American men.”

As a discursive tactic, deictic passages may even be more effective than explicit calls for action against vulnerable groups, as it requires audiences to accept or at least be constantly aware of the underlying context (the speaker’s set of beliefs) in order to understand the speaker’s comments.

Indexical Terms

Indexicality is a concept that emerges out of linguistics as well as the philosophy of language. Like deixis, it describes references whose meaning is dependent on context. An indexical includes any sign—linguistic expression, behavior, or thing—that points to other concepts, objects, or sentiments. The classic example is smoke as an index of fire, insofar as we associate smoke as a sign that points to fire. In this context, smoke means fire, since we understand that fire produces smoke. The relationship between an indexical sign or code and what it signifies is not necessarily causal, however. As we saw in the discussion of The John & Ken Show in the previous section, the hosts established the word “appearance” as an indexical for wealth and “obesity” as an indexical for low-income Black and Latino families in South Central Los Angeles. Analyses of indexicality offer insight into interpretative processes and the role of language in constructing identity and societal attitudes (Inoue 2004; Ochs 1992). As Mary Bucholtz and Kira Hall note, “Identity is the social positioning of self and other [wherein] indexical associations can also be imposed from the top down by cultural authorities such as intellectuals or the media. Such an imposed indexical tie may create ideological expectations among speakers and hence affect linguistic practice” (2005, 596).


Given the large volume of data that would be generated from coding all three transcripts for indexicality, we focused on a sample drawn from one of the transcripts: ten minutes of dialogue from The John & Ken Show between hosts John Kobylt and Ken Chiampou and guest Jim Gilchrist, founder and director of the Minuteman Project. During this segment they discussed San Francisco’s sanctuary policy in the context of murder (and other crimes) and immigration (lines 45-267 of the transcript). This segment provides an example of identity construction as the “social positioning of self and other” by way of political nativism—the attitude or policy of favoring the native inhabitants of a country over its immigrants.

Analysis of Sample

Readers found that the speakers used indexicality in four ways in the sample segment: (1) the use of code words to establish Latinos, immigrants, and immigrant rights advocates as “other” to the nation; (2) the use of rhythm, stress, and intonation (prosody) to emphasize nativist attitudes; (3) the reinforcement of nativist attitudes through word repetition; and (4) alignment between the hosts and guest.

Readers identified twenty passages in which indexical terms (code words) were used to identify certain groups as “other” to the nation. Terms such as “illegal alien,” “gangbanger,” “killers,” “anarchists,” “calamity,” and “domestic terrorism” indexed Latinos, undocumented immigrants, and immigrant rights advocates, thereby associating these groups with crime, terror, and a foreign enemy. Heightening this message was the juxtaposition of these terms with indices for a vulnerable home nation: “community,” “civilized community,” “freedom of speech,” “founding fathers,” “city,” and “country.” In seventeen passages, the speaker’s rhythm, stress, and intonation heightened the indexical associations. Rising pitch and syllabic emphasis on the indices for crime, terror, and the enemy added a sense of urgency. Stutters and pauses when uttering usually positive or neutral words (“advocates,” “endorsing,” “supporting,” “preference programs”) to describe immigrant rights proponents indexed ridicule, disgust, and condemnation. Four terms were repeated between three and six times each over the course of ten minutes: “illegal alien” (6), “anarchist” (3), “community or communities” (5), and “free speech” or “freedom of speech” (4). The first two index a foreign enemy, and the last two index the home nation.

Finally, readers identified lack of alignment among hosts and guests in the sample segment—passages where speakers did not use the same words and tried to advance different ideas. Prosodic and rhetorical elements such as interruption, talking over each other, and changing the subject also indicated a lack of alignment. In the ten-minute sample John Kobylt and Ken Chiampou frequently did not align: Chiampou characterized the protesters as advocating immorality, while Kobylt identified them as free speech suppressors; Chiampou wanted to talk about the specifics related to California Lieutenant Governor Gavin Newsom and a type of gun, while Jim Gilchrist wanted to implicate Newsom using generalities. Gilchrist also wanted to advance the idea that “safe” communities were now in danger. These differences were reinforced through word frequency. In the nine instances where Kobylt, Chiampou, and Gilchrist seemed to engage and concur with one another, indexical analysis revealed that only in one instance did they all fully align. In seven instances, the hosts used differing words to refer to the same discussion topic, and in one instance they used the identical word but in reference to differing interpretations of it. This lack of alignment does not signal lack of agreement (they are all anti-immigration), but rather an appeal to different audience segments through different approaches to the same end.

Analysis of Transcripts

For the larger pilot study, we counted the occurrence of four indexical terms that had the highest number of repetitions in the ten-minute segment, either as a single word or in a phrase: “illegal” (including illegals, illegal alien, illegal aliens, illegal immigrants, illegal immigration),anarchist,” “community” (including communities), and “free speech” (including freedom of speech, free press, freedom of the press). We then counted the occurrences of these terms in the transcript for each of the three programs, determining frequency by speaker and totals for each term (see Table 5).


Program hosts and guests repeated the four indexical terms 101 times in the transcripts. Terms including the word illegal accounted for 68 percent (69 of 101) of these indexical utterances, with most of them occurring in The Lou Dobbs Show (44). Dobbs used terms with illegal in them 31 times, and his two guests used the code words a total of 13 times, in reference to deportation, statistical data about the decline of recent illegal immigration, employers of undocumented immigrants, and immigration reform. Dobbs most frequently used the phrase illegal alien (or illegal aliens), which dehumanizes undocumented immigrants and strips away broader socioeconomic contexts and factors. Dobbs used one other code word, community, which was not tabulated because it was used outside a nativist framework (Dobbs spoke favorably of technological incentives within the “business community”).

Savage used the term illegal alien as a jumping-off point in a criticism of government, civil servants (California Attorney General Kamala Harris), and public policies (the sanctuary policy). But he also established “this illegal alien” as an index not only for the specific gruesome murder of a father and his two sons but also for “murderers, rapists, and pornographers” in general. Savage used the other three terms to establish an us-versus-them contrast between himself and socially liberal groups. He used “community” to attack liberal values and lifestyles with regard to the Middle East and to criticize San Francisco as a liberal community in connection to the sanctuary policy. In using the different iterations of free speech, Savage positioned himself as a victim of the censorship of the political left, specifically in regard to philanthropist George Soros and media watchdog Media Matters for America. In several instances, Savage directly cited the activities of Media Matters as an impediment to his own freedom of speech and portrayed the organization as detrimental to personal freedom, national freedom, and public knowledge. Savage also called for financial investigations of Media Matters. Finally, Savage used “anarchist” in the context of describing left-oriented politics. The term was used in combination with “socialists, communists, and anti-Americanism” to characterize crowds attending Obama’s campaign speeches in the United States and abroad. The term was also used twice to describe Media Matters as an organization that wants to censor Savage’s views.

In The John & Ken Show, “illegal alien” or some variation was used nine times, largely as an index of immorality and criminality. Like Savage, Kobylt and Chiampou associated “illegal alien” with violent crime: “triple-murdering illegal alien.” They also used the phrase four times to qualify the terms advocates, advocacy groups, and protesters, thereby casting these groups in immoral and criminal terms as well. Guest Jim Gilchrist did not use “illegal aliens,” but he did use “these killers” in an indexical sense that generalizes a single murder suspect to imply that all undocumented immigrants are murderous: “They [San Franciscans] don’t want these killers up here.” In the discussion, however, Gilchrist focused more attention on the protests by immigrant rights advocates as suppressing free speech, a point he repeats four times. Gilchrist was the only speaker on the show to use “community” (five times), mostly as a reference point for fears about the dangers posed by undocumented immigrants. Finally, both host and guest used “anarchist” as an index for immigrant rights advocates. For Gilchrist, “anarchists” becomes a code word by which the following associations are attached to immigrant rights advocates: “delusional,” “misinformed,” “let’s suppress everyone else’s freedom of speech,” and “domestic terrorists.” Interestingly, Chiampou used “anarchist” in a different sense, distinguishing between advocacy groups and “just anarchists who signed up for the mayhem that day.” Gilchrist responded by conflating both senses of the word around “domestic terrorism.”

Conclusion and Recommendations

The findings are based on data generated from broadcast segments of thirty to forty minutes that were selected from three commercial talk radio programs. Even using this limited sample, the qualitative content analysis reveals a significant incidence of speech that incorporates targeted statements, unsubstantiated claims, divisive language, and indexical terms related to political nativism. Our analysis yielded no instances of the kind of hate speech that is defined in the 1993 NTIA report as calling for “immediate unlawful action” (U.S. Department of Commerce 1993). Whether such speech exists on talk radio would require a broader study. Based on the evidence we uncovered, the programs reveal a distinct and recurring rhetorical pattern for targeting specific vulnerable groups that relies on the systematic use of a combination of unsubstantiated claims, divisive language, and indexical terms that point to a nativist attitude. Through this rhetorical pattern, vulnerable groups were defined as antithetical to core American values, which were attributed by the hosts to themselves, their audience, and the nation. A significant and recurring indexical construct was that of (Latino) immigrants as criminals and, by extension, as an imminent threat to the American public. Latino immigrants were also linked to social institutions that were presented as complicit with immigrants. In this way, target groups were characterized as both powerful and a direct threat to the listeners’ way of life (in some measure because they were seen as having captured major social institutions such as the media).

What we see as most troubling in our findings is the extent to which this pattern relies on unsubstantiated claims while the talk radio programs situate themselves within the journalistic context of “news” and “opinion” directed at public policy debate. Although our data included no explicit calls to criminal action, the programs systematically placed unsubstantiated claims in the context of divisive language and indexical associations that drew a sharp contrast between their targets (vulnerable groups and those depicted as in collusion with them) and their ideologically aligned listeners, whom they sought to mobilize. In this regard, it is notable that the program hosts often utilized specific situations and people to exemplify larger themes. Thus, while the targets may have been specific (a political figure, a news organization, undocumented immigrants), the discourse itself had bigger political or policy aims.

The primary goal of the pilot study was to establish a rigorous and replicable methodology for a full-scale study or series of ongoing studies. Although the limited size of our sample does not provide a basis for definitive conclusions, our findings nonetheless identify several distinct features of speech among the talk radio programs and raise useful questions for a full-scale study. These include broader studies into the extent and nature of:

* The discursive pattern whereby vulnerable groups are targeted but calls for action are directed against those identified as supporters of vulnerable groups.

* The use of dehumanizing metaphors, divisive language (deixis), and indexical terms (nativist code words) to establish targeted out-groups as outside the realm of legal protection or participation in public discourse.

* The use of unsubstantiated claims as a cornerstone of political opinion presented as an interpretation of fact. These unsubstantiated claims further serve as the basis for targeting vulnerable groups as an immediate and direct threat to the program audience (and nation).

Other studies could attempt to measure the impact of particular discursive patterns, figures of speech, linguistic expressions, and unsubstantiated claims that target vulnerable groups while also calling for action on the part of listeners.

The major challenge for a study of hate speech involves determining whether some speech on news-talk radio conforms to one of the two definitions for actionable hate speech: speech that threatens or fosters the commission of hate crimes. In this regard, it is important to note that there is no inherent statistical or numerical threshold for any of the analytical categories used in this study that could thereby provide unequivocal evidence for the existence of hate speech or a climate of hate or prejudice. Indeed, determining a causal relationship between media discourse and the commission of hate crimes against vulnerable groups would require other approaches that can measure impact. In this regard, the pilot study lays the groundwork for developing scientific studies that would provide evidence related to impact: for example, biomarkers for increased aggression (based on evidence that salivary biomarkers can measure aggression as demonstrated by Gordis et al. 2006), social psychology surveys, and social network analysis (Wasserman and Faust 1994). We argue that qualitative content analysis provides a necessary component of any such analysis, since it provides verifiable, precise delineations of program content. But the question about the relationship between program content and hate crimes requires a multidisciplinary approach that can provide indicators for impact on different levels: physiological, psychological, and social. Together, these approaches would provide a more complete picture of the nature and impact of program content with respect to vulnerable groups.

This project has numerous implications for policy development. In the past, Latino groups have pushed for change on three fronts with respect to media advocacy and policy: improving on-screen and on-air portrayals, increasing employment (for both talent and executives), and facilitating media ownership. While our project does not explore the fine points of media policy or the public and political debate that surrounds them, we do bring renewed attention to content issues as they impact the Latino population. Our pilot study also highlights the issue of codes of professional conduct for journalists. Moreover, our pilot project may generate questions about control over content production and distribution and how that control affects the representation of vulnerable groups. In this regard, the 1993 NTIA report’s recommendation is more salient than ever: “To combat hate speech with more speech.” But such a goal is also easier said than done. Indeed, one can reasonably ask, exactly how and where will this “more speech” be spoken?

In the United States, the issue of hate speech has been framed largely by First Amendment protections, focusing on freedom of speech and of the press. In some ways, the public discourse about free speech has become more about doctrine than process, presuming that free speech is absolute and fixed rather than a freedom from governmental restrictions that must be defended and defined through specific instances and for which there have been notable exceptions (sedition, war protests, obscenity, and, more recently, free speech zones). In the United States, the system of checks and balances inherent in the Constitution’s separation of powers provides a necessary governance context for adjudicating among competing constitutional rights, for example, in the case of hate speech, freedom of speech (First Amendment) versus equal protection (Fourteenth Amendment).

Furthermore, in adjudicating among competing claims and claimants, the government also plays an indirect role in restricting speech, as in the case of libel suits. With respect to broadcast media, the government determines which corporate interests are allocated radio frequency (and, hence, access to speech within expansive media), defines a regulatory framework that has a direct impact on programming formats (and, hence, on content), and plays an explicit role in imposing fines and censorship around specific content. Our goal is not to question freedom of speech but rather to acknowledge that it exists in the context of the entire U.S. Constitution and our corresponding governance system. In addressing the issue of hate speech and its relation to hate crimes, we need the “more speech” of sound research on the phenomenon itself, so that we have some basis other than unsubstantiated claims by which to understand who says what, about whom, and to what effect on the public airwaves.


Student researchers included Hector Peña, Raul Lugo, Alex Lee, Addy Figueroa, Katherine Branson, Kristy Norindr, Melany Bertrand, Antonio Anfiteatro, and Clifford Hilo. We would like to recognize the exceptional contributions of Hector Peña and Clifford Hilo to this project. Our UCLA colleagues, Otto Santa Ana and Darnell Hunt, provided valuable consultation during the early development of the project. This research is supported in part by a grant from the Necessary Knowledge for a Democratic Public Sphere Program of the Social Science Research Council with funds provided by the Ford Foundation. The program supported collaborations between researchers and advocacy groups. This research project is a strategic partnership between the UCLA Chicano Studies Research Center and the National Hispanic Media Coalition. Additional support was provided by the W. K. Kellogg Foundation, as well as by the UCLA Graduate Division.


Barrett, Janet R. 2007. The researcher as instrument: Learning to conduct qualitative research through analyzing and interpreting a choral rehearsal. Music Education Research 9(3): 417-433.

Boone, Daniel N. 1999. The cogent reasoning model of informal fallacies. Informal Logic 19(1): 1-39.

Brewer, Marilynn B. 1979. In-group bias in the minimal intergroup situation: A cognitive- motivational analysis. Psychological Bulletin 86(2): 307-324.

Bucholtz, Mary, and Kira Hall. 2005. Identity and interaction: A sociocultural linguistic approach. Discourse Studies 7(4-5): 558-614.

Bushman, Brad J., and Craig A. Anderson. 2001. Media violence and the American public: Scientific facts versus media misinformation. American Psychologist 56(6-7): 477-489.

Butler, Judith. 1997. Excitable speech: A politics of the performative. New York: Routledge.

Cortese, Anthony. 2006. Opposing hate speech. Westport, CT: Praeger.

Dharmapala, Dhammika, and Richard H. McAdams. 2003. Words that kill: An economic perspective on hate speech and hate crimes. University of Illinois Law and Economics Research Paper No. LE02-004.

Federal Bureau of Investigation. 2004. Hate crime statistics, 2004. Uniform Crime Reports.

———. 2008. Hate crime statistics, 2008. Uniform Crime Reports.

Finocchiaro, Maurice F. 1981. Fallacies and the evaluation of reasoning. American Philosophical Quarterly 18(1): 13-22.

Gordis, E.B. et al. 2006. Asymmetry between salivary cortisol and alpha-amylase reactivity to stress: Relation to aggressive behavior in adolescents. Psychoneuroendocrinology 31(8): 976-987.

Govier, Trudy. 1982. Who says there are no fallacies? Informal Logic Newsletter 5(1): 2-10.

Groarke, Leo. 2012. Informal logic. In The Stanford Encyclopedia of Philosophy (Fall 2012 Edition). Stanford, CA: Metaphysics Research Lab, Center for the Study of Language and Information, Stanford University.

Hahn, Ulrike et al. 2009. Argument content and argument source: An exploration. Informal Logic 29(4): 337-367.

Halpin, John et al. 2007. The structural imbalance of political talk radio. A joint report by the Center for American Progress and Free Press.

Hennessey, Kathleen. 2010. “Tea Party” convention a forum for woes, worries. Los Angeles Times, February 6, A11.

Houston Santhanam, Laura et. al. 2012. Audio: By the numbers. State of the news media 2012: An annual report on American journalism. The Pew Research Center’s Project for Excellence in Journalism.

Inoue, Miyako. 2004. What does language remember? Indexical inversion and the naturalized history of Japanese women. Journal of Linguistic Anthropology 14(1): 39-56.

Kellow, Christine L., and H. Leslie Steeves. 1998. The role of radio in the Rwandan genocide. Journal of Communication 48(3): 107-128.

Kondracki, Nancy L., and Nancy S. Wellman. 2002. Content analysis: Review of methods and their applications in nutrition education. Journal of Nutrition Education and Behavior 34(4): 224-230.

Kövecses, Zoltán, et al. 2010. Metaphor: A practical introduction. London: Oxford University Press.

Lendman, Stephen. 2006. CNN’s vice president of racism and associates. Montreal: Centre for Research and Globalization.

Lewis, Anthony. 2007. Freedom for the thought that we hate: A biography of the first amendment. New York: Basic Books.

Lyons, John. 1977. Semantics. Vol. 2. London: Cambridge University Press.

Mayring, Philipp. 2000. Qualitative content analysis. Forum: Qualitative Social Research 1(2).

Meddaugh, Priscilla Marie, and Jack Kay. 2009. Hate speech or “reasonable racism?” The other in stormfront. Journal of Mass Media Ethics 24(4): 251-268.

Milbank, Dana. 2010. Glenn Beck and the Oakland shooter. Washington Post, August 1.

Neiwert, David. 2009. The eliminationists: How hate talk radicalized the American right. Sausalito, CA: PoliPoint Press.

O’Connor, Rory. 2008. Shock jocks, hate speech & talk radio. San Francisco: AlterNet Books.

Ochs, Elinor. 1992. Indexing gender. In Rethinking context: Language as an interactive phenomenon, 335–358. Edited by A. Duranti and C. Goodwin. Cambridge: Cambridge University Press.

Perdue, Charles W. et al. 1990. Us and them: Social categorization and the process of intergroup bias. Journal of Personality and Social Psychology 59(3): 475-486.

Pew Hispanic Center. 2007. 2007 National survey of Latinos: As illegal immigration issue heats up, Hispanics feel a chill. Washington, DC: Pew Hispanic Center.

Rapaport, William J. et al. 1994. Deictic centers and the cognitive structure of narrative comprehension. Buffalo, NY: Center for Cognitive Science, State University of New York at Buffalo.

Santa Ana, Otto. 2002. Brown tide rising: Metaphors of Latinos in contemporary American public discourse. Austin: University of Texas Press.

Slagle, Mark. 2009. An ethical exploration of free expression and the problem of hate speech. Journal of Mass Media Ethics, 24(4): 238-250.

Society of Professional Journalists. 1996. Code of ethics.

Tolmach Lakoff, Robin. 2001. The language war. Berkeley: University of California Press.

U.S. Census Bureau. 2011. State and county quickfacts for 2011 estimate.

U.S. Department of Commerce. 1993. The role of telecommunications in hate crimes. National Telecommunications and Information Administration (NTIA). Washington, DC: Government Printing Office.

Wasserman, Stanley, and Katherine Faust. 1994. Social network analysis. Cambridge: Cambridge University Press.


Research in economics involves the development of models with empirical support (Dharmapala and McAdams 2003). Media research has established scientific approaches for impact as it relates to advertising as well as to media violence (Bushman and Anderson 2001).

Our study relies on the original target groups for hate speech put forward in the 1993 NTIA report: “‘Hate speech’ would therefore encompass words and images that ‘manifest evidence of prejudice based on race, religion, sexual orientation, or ethnicity.’” That said, our use of “vulnerable group” as a generalized description for hate speech targets allows for the fact that the groups constituted as vulnerable may change over time or across different contexts.

Media Matters for America, a not-for-profit media monitoring organization, provided audio files and transcripts for The Savage Nation for 21-31 July 2008. More information about obtaining copies of the transcripts used in this study is available on request.

The study of informal logic—the attempt to assess and therefore improve reasoning in ordinary (natural) language—seems to be moving away from a focus on fallacies as a way to evaluate the validity of an informal argument (Groarke 2012). For our purposes, the central issues have to do with the limited ability of informal logic (and “traditional fallacies,” in particular) to address natural-language argument (Hahn et al. 2009). This approach also resonates with at least one attempt to redefine fallacy on the basis of the falsity or truth of the premises (Boone 1999).

SPJ elides a crucial distinction here: the First Amendment protects journalists from governmental censorship, not necessarily from “decision making” related to content made by media corporations.

We do not include Dobbs’s claim, “But, by God, I’m an anti-illegal employer as well” (lines 985-986), which has been proven untrue (Nation, 25 October 2010); that was not known at the time of the broadcast.

In a Los Angeles Times article on the National Tea Party Convention in February 2010, an attendee explained what she wanted from the movement: “Our way of life is under attack. I truly believe they are trying to destroy this country. It’s just hard to say who ‘they’ is” (Hennessey 2010). The deictic phrasing reveals both the speaker’s fears for “our way of life” in “this country” and her confusion about the “they” posing the threat. Mass media has the potential to provide contextual information that shapes these fears—and their us-versus-them configuration—by providing a clear referent for the deictic term “they.”

Media Matters for America describes itself on its Web site as a “progressive research and information center dedicated to comprehensively monitoring, analyzing, and correcting conservative misinformation in the U.S. media.”

In an opinion piece on 1 August 2010, a Washington Post writer outlined a suggestive correlation between program content targeting specific groups and violent acts (Milbank 2010).

The First Amendment is actually broader in scope: “Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of speech, or of the press; or the right of the people peaceably to assemble, and to petition the Government for a redress of grievances.”