Covert racism in law and social theory

The U.S. civil rights movement delegitimized blatant, biologically based, segregationist racism and ushered in new legislation and stronger application of constitutional equal protection to address racism. However, these victories were partial because the legal definition of racism fails to match the social reality of racism. Antidiscrimination law largely treats race as biologically derived, racism as rooted in individual bigotry and racial discrimination as based on immutable traits possessed by all group members (Carbado and Gulati Reference Carbado and Gulati2013; Freeman Reference Freeman1990; Haney López Reference Haney López2006; Onwuachi-Willig and Barnes Reference Onwuachi-Willig and Barnes2005; Rich Reference Rich2004). Yet, race is socially constructed through ever-evolving meaning systems, and racism results from diffuse, multiscalar and intersecting forces that organize American life in racially stratifying ways (Bonilla-Silva Reference Bonilla-Silva2022; Haney López Reference Haney López2006; Lawrence Reference Lawrence1987; Omi and Winant Reference Omi and Winant2014). As a result, a chasm exists between race and racism operating on the ground and race and racism recognized by antidiscrimination law.

Racial/Ethnic Minorities (particularly Black people) experience the criminal justice system as an expansive, accumulating and mighty apparatus of oppression where law emboldens racism, legal professionals “do racism” as they “do justice,” and a “street law” is created for “street people” (Murakawa and Beckett Reference Murakawa and Beckett2010; Gonzalez Van Cleve Reference Gonzalez Van Cleve2016: 13, 53).Footnote ³ Conversely, they experience antidiscrimination law as a narrow, disaggregating and weakened shield of protection that is more promissory than real (Murakawa and Beckett Reference Murakawa and Beckett2010). The “social regime of strict scrutiny” omnipresently surveils, disciplines and traps Black people in a liminal space of life and death and safety and punishment, while the legal regime of strict scrutiny narrowly constricts racially conscious remediation efforts and naturalizes and reinforces their racial subordination (Carbado Reference Carbado2022). Hence, although antidiscrimination law vows to eliminate racism, it takes a formalist and skin-deep approach to race, “aggressively disaggregates” systemic racism and racial causation and is “a jurisprudence of racial nonrecognition” rather than the supposed opposite (Murakawa and Beckett Reference Murakawa and Beckett2010; Obasogie Reference Obasogie2010: 612; Edelman et al. Reference Edelman, Smyth and Rahim2016).

Law partakes and is a central site of the social construction of race (Haney López Reference Haney López2006). This jurisprudence of racial nonrecognition is in essence a special form of legal construction of race. Similar to how courts in past prerequisite cases constructed racial categories by determining who was and was not White (ibid), courts in racial discrimination cases construct racial boundaries by determining whether a justification is racial or not. In prerequisite cases, courts constructed race by drawing boundaries between races, with classification as White entailing legal protections and advantages. In antidiscrimination cases, courts construct race by drawing boundaries between racial and nonracial explanations, with classification as racial entailing (equal treatment) legal protections and advantages. In both settings, the law demarcates racial boundaries and creates a prized classification (Whiteness and raciality, respectively), courts narrowly guard eligibility for the prize, and plaintiffs are forced to simultaneously abide by and perpetuate the classification to receive legal protection.

Racial codes arise from the chasm between race and racism on the ground and in antidiscrimination law. They give the user “plausible deniability” of race-based decision making (i.e., the ability to deny that actions were racially motivated, Bilotta et al. Reference Bilotta, Corrington, Mendoza, Watson and King2019; Edelman et al. Reference Edelman, Smyth and Rahim2016; Liu and Mills Reference Liu and Mills2006). Hence, redress for victims of coded racial discrimination depends on whether they succeed or fail to uncloak plausible deniability. If they succeed, courts recognize the accused act as illegal racial discrimination (e.g., Dews v. Town of Sunnyvale, Texas, 2000). If they fail, courts classify the accused act as nonracial and the code provides a legal hack for discrimination with impunity (e.g., Hallmark Development Inc. v. Fulton County, Georgia, 2006). Antidiscrimination law’s inability to crack racial codes is a major cause of their ascent.

Various theories illuminate how covert racism works. Covert racism spreads a narrative that American society is fair and justifies racial inequality on ostensibly nonracial grounds, such as White cultural/ideological values of individualism, meritocracy and free-market competition (e.g., “symbolic racism” in Kinder and Sears (Reference Kinder and Sears1981); “modern racism” in McConahay (Reference McConahay, Dovidio and Gaertner1986); “laissez-faire racism” in Bobo (Reference Bobo2017); “color-blind racism” in Bonilla-Silva (Reference Bonilla-Silva2022)). These values are so pervasive and hegemonic that they form part of commonsense knowledge about race and racial inequality, to the point of making racist beliefs and actions “automatic,” “unconsidered” and even “unconscious” (e.g., “unconscious racism” in Lawrence (Reference Lawrence1987); “commonsense racism” in Haney López (Reference Haney López2003); “implicit bias” in Levinson and Smith (Reference Levinson and Smith2012)). A smaller, though sizable, literature investigates mechanisms, processes and effects of covert racial discourses. Beckett (Reference Beckett1997) exposes how political and media elites socially constructed the war on crime and the war on drugs in the latter 20th century to stoke racial anxieties and advance a conservative agenda. Mendelberg (Reference Mendelberg2001), Haney López (Reference Haney López2014) and Bennett and Walker (Reference Bennett and Walker2018) show how strategic actors imbue race-neutral words like fundamental rights, gun ownership, welfare, crime and criminal justice, and states’ rights with racial meaning to seek political, economic and ideological gains. López-Espino (Reference López-Espino2023)exposes how legal professionals create covert discourses to racialize and coerce obedience from parents in child welfare cases through techniques of silencing, voicing figures of disbelief and labeling their speech as lies and excuses.

Despite these vibrant and growing studies of covert racism, comprehensive knowledge on tropes, mechanisms, evolving processes, and individual and systemic effects of covert racial discourses is lacking. Our research seeks to narrow this gap. We analyze one type of discourse – racial code words. We ask: What forms do contemporary racial code words take? What tropes do they convey? What functions do they perform and with what mechanisms? What knowledge, if any, can we gain about racial code words that could help researchers, legal professionals and social justice advocates detect coded race talk?

We also seek to narrow a disciplinary gap between sociolegal studies and CRT. On the one hand, we respond to sociolegal race scholars’ calls to study race capaciously, operationally and systemically (Gómez Reference Gómez2012; Murakawa and Beckett Reference Murakawa and Beckett2010). We investigate micro-instances of racial code word use and examine their tropes, individual and systemic effects, and linguistic mechanisms. On the other hand, we bring CRT to bear on sociolegal studies of race. The legal realist movement in the early 20th century and the law and society movement in the latter 20th century have made it a common and honored practice to apply social science insights to legal issues (Currie Reference Currie1951; Macaulay Reference Macaulay1963). In recent decades, in the context of race, the neutrality and necessity of applying social science insights to adjudication have been challenged from the left (CRT) and the right (legal formalists), respectively (Moran Reference Moran2010). We deploy CRT insights to inform, energize and stimulate debate in sociolegal studies of race and racism, a practice explicitly advocated by former Law and Society Association President Laura Gómez (Reference Gómez2012).

More concretely, we integrate CRT scholars’ insights as we search for ways to deconstruct the hack of racial code words. CRT scholars have long lamented how antidiscrimination law programmatically denies redress to victims of discrimination based on racially salient features such as housing patterns, hairstyles, names or accents and proposed innovative doctrinal solutions (Carbado and Gulati Reference Carbado and Gulati2013; Lawrence Reference Lawrence1987; Onwuachi-Willig and Barnes Reference Onwuachi-Willig and Barnes2005; Rich Reference Rich2004). Lawrence (Reference Lawrence1987) introduces a cultural meaning test for equal protection adjudication. The test asks courts to act “much like a cultural anthropologist” and investigate whether the government defendant’s conduct conveys “a symbolic message to which the culture attaches racial significance” (ibid: 356). If the court determines that more likely than not “a significant portion of the population thinks of the governmental action in racial terms,” the court will assume that unconscious racism has influenced the decision-maker and thereby “apply heightened scrutiny” (ibid). Rich (Reference Rich2004) shows how race/ethnicity is embodied in various socially coded racial/ethnic markers and offers a racial/ethnic performativity test to better address modern forms of employment discrimination. Under this test, the plaintiff must establish that the performative act on which discrimination is based carries racial/ethnic significance and that the association between the act and the racial/ethnic identity is too strong to require the individual to change. The defendant then must offer a race-neutral value or justification and specifically explain how the plaintiff’s performance thwarts that value. Lastly, Onwuachi-Willig and Barnes (Reference Onwuachi-Willig and Barnes2005) propose a creative transplant of the “regarded as” verbiage in the Americans with Disabilities Act to employment discrimination law to address racial discrimination by proxy. If an employer uses a racial proxy (name, hairstyle, voice, etc.) to determine that a job applicant belongs to a certain race and denies the latter a job based on that determination, the employer has conducted illegal discrimination. Our study integrates these CRT insights into a “racial meaning decoding tool” to support researchers, legal professionals and social justice advocates in detecting racial code words in varied settings.

Canvassing racial code words from scholarly literature

Studying racial code words in empirical settings, particularly those relevant to antidiscrimination law, is essential to understanding and combating covert racism. Yet, such studies face immediate challenges. Researchers lack a framework informed by knowledge of the common features and functions of racial code words to determine whether words might be racially coded in a particular setting. Incomplete and inconsistent if not arbitrary findings may result in the absence of this foundation. Racial code words’ plausible deniability and political sensitivity may expose researchers to accusations of carelessness or subjectivity at best and stoking racial hatred or racism at worst.

A general framework of what racial code words are and how they function is needed to help researchers avoid these pitfalls. Ideally, such a framework would come from repeated empirical investigations into racial code words within and across diverse settings, like local housing or workplace decision making, court cases or police reports. However, this approach would be time- and resource-intensive. Such commitments would be imprudent absent some general knowledge on the forms and functions of racial code words, especially their harmfulness and intractability.

Our research provides this starting foundation by mining a corpus of racial code words from a rich but under-synthesized data source: existing cross-disciplinary scholarly literature on real-life race talk in post-civil rights America. We use a rigorous method of sampling, data extraction and analysis – a systematic literature review (Xiao and Watson Reference Xiao and Watson2019). This method draws findings from scholarly texts by identifying relevant texts through keyword searches, applying inclusion criteria to sample texts most able to answer the research questions, recording and validating data extracted from these texts through iterative review processes involving multiple researchers and drawing findings through qualitative methods of analysis, like thematic coding (ibid). We used these methods to synthesize the rich, contemporary empirical scholarly literature on how people talk about race in diverse settings, canvass the code words used and develop an interpretive framework and later a decoding tool broadly applicable across social settings.

A key strength of studying racial code words through this data is that they have been vetted twice: first as race talk by the authors of the texts reviewed and second as racial code words by our research team. This quality helps to weaken their plausible deniability and counteract claims of researchers’ subjectivity. The interdisciplinarity of this body of literature also helps to uncover common patterns of racial code words across diverse settings and build global insights on their forms and functions.

Studying racial code words from scholarly literature also has drawbacks. The data are filtered at least once and sometimes twice or more through others’ interpretations. This limits the universe of race talk available for analysis within a particular setting. For example, researchers selectively extract instances of race talk from their participant interviews and integrate them into their findings. Instances from news media are filtered first through the perceptions of journalists and then through researchers who analyze the media. Selective filtering also reduces available contextual information on race talk, such as its setting or participant characteristics. This not only creates possibilities for interpretive errors but also limits insight into whether discourse is motivated by racial antipathy. Additionally, the settings studied in this data are shaped by authors’ disciplinary orientations (e.g., political speeches for political scientists and schools for education scholars). Settings that are not strongly connected to a specific discipline are underrepresented (e.g., covert race talk at children’s summer camps). These qualities of our data increase the risk that our findings fail to capture the broad use of racial code words in post-civil rights America and may misrepresent or misinterpret those that are captured.

Our sample of 734 racial code words comes from 97 texts published between 2000 and 2020 that address race and verbal language in real-life environments in post-civil rights America (see Appendix 1 for their references). These texts were drawn from a larger pool of 1,356 texts identified through a systematic keyword search using Google Scholar and our university library databases, which integrate large repositories, including HeinOnline, JSTOR, Web of Science, EBSCO, ProQuest and others. A text was included in the pool if (1) it had at least one keyword anywhere for each of the four dimensions of our research: race, communication, code and bias (see Appendix 2 for the keyword list) and (2) appeared in the first 200 results of our university library search engines (combining all of the keywords below in a Boolean string) or first 10 results in 190 iterations of searching in Google Scholar (due to limitations in combining keywords in a Boolean string), both sorted by relevance. This approach resulted in an initial sample that was broadly related to these dimensions but also had some blind spots (e.g., Gonzalez Van Cleve Reference Gonzalez Van Cleve2016; Nguyen et al. Reference Nguyen, Basolo and Tiwari2013; Tighe Reference Tighe2010). It captured texts that addressed race talk, even if their main topic was not explicitly about it.

We next identified texts most able to answer our research questions by retaining those that met the following criteria, which only 97 of the 1,356 texts met:

1. (1) Published on or after January 1, 2000 in English in an academically recognized press/journal,

2. (2) Used empirical data to answer questions about race or ethnicity and verbal language in real-world environments that were:

   • • In the United States,

   • • After 1967 and

   • • Not from fictional, simulated or other controlled experimental settings and

3. (3) Contained racial code words identified:

   • • By the research team as:

     • Indirect signifiers of racial or ethnic groups that

     • Contain at least one positive or negative value judgment and

     • Contextually implied or salient meanings or

   • • By the author(s) of the text.

Words were indirect signifiers of racial groups when they were embedded in discourse that was identified as race talk by the author(s) of the text or our research team but did not explicitly name the referred to racial or ethnic group in the sentence in which they appeared. They contained at least one positive or negative value judgment when the words, sentence or discourse in which they were embedded expressed positive or negative valence or affect (see Mohammad Reference Mohammad2018). The meanings were contextually implied or salient when the speaker and the audience within a particular setting shared an understanding of the words during the exchange. An example is a White woman’s recounting of conversations with neighbors in a mixed-race Chicago neighborhood who told her to ‘‘Be careful when you go down that street’ and, where there was, you know, sort of a seemingly rougher clientele living on certain streets …’ (Burke Reference Burke2017: 289–290). We identified rougher clientele as a racial code word, because it (1) was an indirect signifier of Black people (as suggested by the author and indicated by the lack of explicit reference to Black people in the discourse), (2) contained a negative value judgment (rougher has negative valence) and (3) was understood as such by the speaker and the listener in a conversation about safety in a particular Chicago neighborhood.

Evaluating whether texts contained racial code words was challenging. Meanings of words (especially coded words) are context-dependent and fluid (Hall Reference Hall and Hall2013). They are constructed through not only convention but also dynamic, ongoing interactions between people with different identities and societal positionalities. The socially constructed nature of race additionally dictates that racial meanings are contested and even embattled (Omi and Winant Reference Omi and Winant2014). For example, although we interpreted a bouncer’s instruction to turn away anyone in big chains as coded language for Black people, someone with a different positionality (e.g., a White teenager who wears big chains to “look cool”) may reach a different conclusion. These challenges manifested in our research, as our two-person research team carries internal differences in racial, disciplinary and linguistic positioning.

We took several measures to address interpretive challenges. First, we independently extracted code words from a pool of 30 texts selected from our initial sample, discussed and resolved discrepancies and enhanced our inclusion criteria. Then, we sorted the initial sample by author last names, divided it between us alphabetically and independently extracted potential code words fitting the inclusion criteria. We recorded them in a database along with the following contextual elements: the (1) lines of text encompassing the code word, (2) temporal and geographic setting, (3) motivating event and (4) characteristics of the referred-to subject, user and audience, including racial identity if available. We then independently evaluated each other’s data, discussed and resolved discrepancies and iteratively revised our criteria and reevaluated the code words until no further changes were made. This resulted in a final sample of 734 code words from an initially extracted pool of 1,197 code words, testimony once again to difficulties in discerning racial meaning.

We explicitly embrace the malleability and contestability of racial meaning in our decoding tool, which we present in a later section. We follow a practice in CRT of treating words as racially coded when community norms or a plurality of parties – in our case, our research team and often the authors of the examined scholarship – believe that race is being signaled (Lawrence Reference Lawrence1987; Rich Reference Rich2004). Our approach also is informed by the logic of dealing with contested or otherwise uncertain facts in the law, such as hearing from opposing sides and making findings based on a preponderance of evidence.

Henceforth, the code words extracted from scholarly texts are italicized and referred to as the “sample”; their citations are denoted by “S” (i.e., sample) and a number representing the order in which their references appear in Appendix 1 (e.g., S2). Virtually all of the racial code words came from data referenced in the texts; they most often appeared in quotes from the data (e.g., from interview participants or news media, as indicated by the use of singular quotation marks) but in some instances they appeared in an author’s synthesis of observations from their data (e.g., a list of racial code words commonly used in a political campaign, as indicated by the use of double quotation marks). We found only a few instances of an author using racially coded language to interpret their data or build an argument, mostly to convey personal experiences or illustrate common examples of coded racism (e.g., S76; S89; S66).

Table 1 shows the sampled code words’ temporal, racial and contextual dimensions. Most appeared between the 40-year span of 1968–2008 and the 8-year span of the Obama Presidencies (36% and 35%, respectively). Code words for Black people were most prevalent (45%), followed by White (25%) and Latinx (11%) people. A sizable proportion referred to people from other racial and ethnic groups (henceforth called “Other people”). A small number of these referred to Middle Eastern, Asian, Native American or Jewish people (n = 34, 18, 3, 3, respectively); most referred to Mixed Race people or Racial/Ethnic Minorities in general. The user and audience were most often White people when race was known (56% and 31% of code words, respectively), but often the race of the user was unknown (e.g., a police officer) or the race of the audience was irrelevant, because it was the public in general (e.g., American voters). The sampled code words occurred across a range of settings but mostly in the media, politics or research; only a small portion occurred in settings with some potential for antidiscrimination protection (e.g., schools, workplaces, marketplaces and law enforcement).

Table 1. Characteristics of sampled code words (n = 734)

Source:  Authors’ own work.

Notes: Code words’ subjects, users and audiences may be of multiple racial groups. Other race refers to Asian, Native American, Middle Eastern, Mixed Race, Jewish or Racial/Ethnic Minorities in general. Code words in a few cases have multiple settings.

We used thematic analysis to discover and record the sampled code words’ explicit and implicit qualities in a codebook (Gaber Reference Gaber2020). Our approach was abductive, meaning it was informed by categories and theories from existing scholarship and new divergent ones found through systematic and iterative data engagement (Timmermans and Tavory Reference Timmermans and Tavory2012). Tropes from covert racism and sociolegal studies were foundational, including perceptions of Racial/Ethnic Minorities’ criminality, deviance from American cultural values like individualism and White possessiveness of American property, economy, governance and citizenship (e.g., Beckett Reference Beckett1997; Bonilla-Silva Reference Bonilla-Silva2022; Carlson Reference Carlson2020; Moreton-Robinson Reference Moreton-Robinson2015). We discovered their nuances and identified additional categories, like White people’s tactics of responding to perceived threats, by iteratively reviewing the data through the lens of theories of covert racism and sociolegal studies (e.g., Beckett Reference Beckett1997; Carlson Reference Carlson2020; Haney López Reference Haney López2014). Finally, we stratified the code words based on their racial subjects and examined trends in their forms and functions, drawing from linguistic, cultural and race theory.

We tell a laboriously researched story about contemporary racial code words. The data are drawn from empirical studies in sociology, communication, education, law and political science and settings as diverse as politics, media, law enforcement, schools and workplaces (see Appendix 1). Our research approach was careful and conservative; the racial code words in our corpus have been twice vetted as race talk, as previously noted. The disciplinary and situational range of our data and the rigor of our methods enables us to identify common ways that racial code words are functioning across settings and offer a new interpretive framework and decoding tool to assist investigations in particular settings.

Nonetheless, the story we tell is imperfect. In addition to the limitations outlined above, we do not bolster the originally sampled texts with texts from their references and citations or conduct parallel data collection by multiple researchers due to the initial sample’s large size and our team’s limited capacity. Taken together, these issues leave some bias unresolved. The racial code words in our sample do not represent the full range of coded racial speech in post-1968 America. There almost certainly are gaps in the forms, tropes and functions of code words revealed by those sampled. The lack of contextual information for some code words, the fluid and contested nature of racial meaning, and our positionality may have led us to misinterpret racial signaling and meaning in some cases. Deconstructing racial code words will take time. Our research offers starting resources, ones that are rigorously grounded in prior literature, existing norms and careful coding but intended to be discussed, verified and refined.

Reverse engineering racial code words in contemporary America

This section disassembles the machinery of the sampled racial code words. We reveal how they integrate tropes that create narratives about racial/ethnic groups with linguistic mechanisms that offer efficient, adaptable and plausibly deniable communication, often to racially stratifying ends.

Forms and tropes

Figure 1 shows the range and relative frequencies of sampled code words for White, Black, Latinx and Other people. The format of the words conveys frequency, with bolded black words appearing ten times or more in our sample, unbolded black words appearing two to nine times and gray words appearing only once. Font size conveys frequency but also code word length, with shorter ones also appearing larger. Parentheticals show the variation of similar code words, including singular and plural versions and qualifications. For example, America, American, Americans and most Americans are shown as (most) America(n(s)). Brackets capture contextually understood but unstated elements, like ‘no people walking around with their pants [down]’ (S56: 562).

Source:  Authors’ own work.Notes: Code words’ formatting reveals their frequencies. Those in bolded black appeared at least ten times; those in unbolded black appeared between two and nine times and those in gray appeared once. Font size is mostly a function of frequency, with more common code words appearing larger, but also of length, with shorter code words also appearing larger. People of Other Races & Ethnicities refers to Asian, Native American, Middle Eastern, Mixed Race, Jewish or Racial/Ethnic Minorities in general.

Figure 1. Forms of code words by the race of their subjects (n = 734).

America and American, suburb and suburbanite, and law and order Footnote ⁴ were the most frequent code words for White people; diversity was the most common for Black, Latinx and Other people. Welfare and thug also commonly referred to Black people; illegal commonly referred to Latinx people and fobby and terrorist commonly referred to Asian and Middle Eastern people, respectively. Most code words appeared only once, indicating a rich mosaic of contemporary coded race talk. Notably, none of the code words for White people appeared 10 times or more (bolded black), while multiple code words for Black and Latinx people had this quality. This may indicate that stereotypes for White people are more nuanced and individualized than those for Black or Latinx people.Footnote ⁵ A few of the code words are racial and ethnic slurs (“towel head,” ‘banana,’ ‘twinkie,’ ‘Coconut,’ ‘Uncle Tom,’ ‘nappy-headed hos’ and ‘Dindu Nuffins’) (S79: 1818; S81: 118; S40: 234; S25: 8; S16: 160); others have explicitly racist elements (‘camel-riding,’ ‘jungle,’ ‘eat dog,’ ‘strapping young buck’ and ‘young buck’) (S52: 33; S23: 288; S82: 195; S65: 59; S14: 707) (Wikipedia Contributors n.d; Wiktionary n.d). These have been fully or partly decoded over their lifecycle, a theme we revisit later.

Code words for White people and Whiteness

Sampled code words for White people were mostly expressed by White people to other White people or the public at large. They convey beliefs about White people’s respectability and privilege and their possession of American lifeways see Table 2). The “White Respectability and Privilege” trope conveys that White people are deserving of their high position in the American social structure, because they create community affluence and fulfill their economic and civic duties (Mills Reference Mills1951; Urciuoli Reference Urciuoli1994). One example is a White man referring to the segregation of White people and Racial/Ethnic Minorities in local schools as ‘they didn’t want the sort of lower areas to assimilate with the upper areas’ (S18: 109). Here, ‘upper areas’ and ‘lower areas’ refer to where White people and Racial/Ethnic Minorities lived, respectively. Code words expressing the affluence of White communities often ground it in suburban single-family homeownership, such as when newspaper managers justified redirecting their investment away from their traditional central city circulation areas as merely trying to serve “affluent readers [in] the suburbs” (S97: 27). Those about White people’s economic citizenship reference their knack for generating and consuming goods and services, as well as their perceived higher human capital. This is evident in a Chicago computer software employer’s view that ‘you do not talk street talk to the buying public,’ where the employees who’talk street talk’ are Black and Latinx people and the’buying public’ is White people (S46: 298–299). Code words about White people’s stronger perceived civic citizenship address their tendency to follow laws and pay taxes, which create safer and better resourced communities. This language is present in a Ku Klux Klan recruitment flier distributed to ‘[l]aw abiding citizens’ in the Houston area (S88: 53).

Table 2. Tropes of sampled code words for White people and Whiteness

Sources:  Authors’ own work. Tropes adapted from Mills (Reference Mills1951); Urciuoli (Reference Urciuoli1994); Moreton-Robinson (Reference Moreton-Robinson2015); Dudas (Reference Dudas2005); Haney López (Reference Haney López2014);  Anderson (Reference Anderson2016) and Drakulich et al. (Reference Drakulich, Wozniak, Hagan and Johnson2021).

Note: See Appendix 1 for the example code word sources.

Another category is “White Possessiveness,” which expresses that America belongs to White people, as opposed to America belonging also to Racial/Ethnic Minorities or White people belonging to America (Moreton-Robinson Reference Moreton-Robinson2015). Being American requires being from or adopting what is considered White male European Protestant culture – values like individualism; property acquisition, control and extraction; advancement through merit and being hardworking and calm. The subcategory Citizenship equates this culture with American citizenship. Journalist Bill O’Reilly expressed this in comparing the ‘non-tradition … constituency’ of ‘single-women, Hispanic Americans, African Americans, whatever’ that reelected Obama in 2012 and the ‘[t]raditional American voters’ that didn’t (S57: 153–154). The subcategory Governance identifies this culture as the basis for American governance, such as when Senate Majority Leader Trent Lott told the Council of Conservative Citizens (a right-wing, pro-White people political group): ‘The people in this room stand for the right principles and the right philosophy’ (S21: 54). Nostalgia conveys longing for a time when this culture dominated American society, to the detriment of Racial/Ethnic Minorities and other marginalized groups, a sentiment often expressed by President Trump and his supporters in calls to ‘Make America Great Again’ (S49: 211; S39: 36).

Many code words link to Whiteness but do not exclusively refer to White people. Some reflect a trope that we call “White Defense,” borrowing from historical and sociolegal studies of White people’s responses to civil rights actions (e.g., Anderson Reference Anderson2016; Drakulich et al. Reference Drakulich, Wozniak, Hagan and Johnson2021; Dudas Reference Dudas2005; Haney López Reference Haney López2014). These mostly convey White people’s reaction to supposed threats posed by Racial/Ethnic Minorities to their respectability, privilege or possession of American lifeways. They refer to subjects of all races, though they most commonly refer to White people, followed by Latinx and Black people. Government policies that sought to foster racial equality or that were racialized to defend racial power asymmetries, like busing during the Nixon election and the war on terror during the George W. Bush and Obama administrations, often were subjects. Some convey White people’s resentment of policies perceived to benefit Racial/Ethnic Minorities, implying a need for action but not directly calling for it. For example, one author expresses her resentment of law schools’ use of affirmative action to admit Black people by arguing that it “flatters their own egos, so that [law schools] can gaze upon their ‘diverse’ realm and bask in their noblesse oblige” (S66: 59). Others call for a response to modify or eliminate these policies to benefit White people, such as some politicians’ and journalists’ advocacy for ‘competency based education’ as a strategy to advantage White people by ending affirmative action (S30: 223–224). These code words reflect desires to maintain White respectability, privilege and possession of America through “colorblind” governance that weakens Racial/Ethnic Minorities’ societal power and success.

Other code words link to Whiteness but refer to Racial/Ethnic Minorities. These reflect the trope of “Passing,” namely perceptions of Black, Latinx, Asian or Middle Eastern people succeeding or not to various degrees in adopting what is recognized as White male European Protestant cultural traits (Urciuoli Reference Urciuoli1994). They are relatively rare in our sample and were variously spoken by White people and Racial/Ethnic Minorities. One subcategory is “Acting White,” such as when an Iranian American man debated whether to ‘wear a baseball cap’ representing an American sports team to the airport to emphasize the Whiteness of his racially ambiguous appearance and avoid being perceived as a ‘terrorist’ (i.e., a Middle Eastern person) (S67: 120). Another is “Becoming White,” such as a Filipino Twitter user’s deployment of the hashtag ‘whitewashed’ to make a self-deprecating joke about being assimilated into White culture, because she is ‘having Starbucks’ when her family ‘eats jollibee’ (a fast-food restaurant serving Filipino cuisine) (S75: 515).

Code words for racial/ethnic minorities

Sampled code words referring to Racial/Ethnic Minorities, like those for White people, were often expressed by White people to other White people or to the public at large. They mostly convey that Racial/Ethnic Minorities cause indirect, direct or possible future harm to White people and to society more broadly (see Table 3). This is especially evident in the “Parasite,” “Contagion” and “Villain” categories. These well-studied tropes evolved from stereotypes for immigrants (particularly Chinese) in the latter 19th century and Black people in the slavery and post-Reconstruction eras; subsequently, they fueled cultural and economic opposition to civil rights, welfare programs and (particularly Latinx) immigration and justified the expansion of the carceral state and oppression of Racial/Ethnic Minorities and immigrants (e.g., Anderson Reference Anderson2016; Beckett Reference Beckett1997; Beckett and Ming Francis Reference Beckett and Ming Francis2020; Carlson Reference Carlson2020; Dvorak Reference Dvorak2000; Gold Reference Gold2012; Haney López Reference Haney López2014; Lens Reference Lens2009; Russell-Brown Reference Russell-Brown1998; Santa Ana Reference Santa Ana2002; Santa Ana et al. Reference Santa Ana, Juan and Sanchez1998; Shabazz Reference Shabazz2015).

Table 3. Tropes of sampled code words for racial/ethnic minorities

Sources:  Authors’ own work. Tropes adapted from Bonilla-Silva (Reference Bonilla-Silva2022); Gold (Reference Gold2012); Beckett (Reference Beckett1997); Russell-Brown (Reference Russell-Brown1998); Dvorak (Reference Dvorak2000); Lens (Reference Lens2009); Haney López (Reference Haney López2014); Shabazz (Reference Shabazz2015); Beckett and Francis (Reference Beckett and Ming Francis2020);  Anderson (Reference Anderson2016); Carlson (Reference Carlson2020); Santa Ana (Reference Santa Ana, Juan and Sanchez1998; Reference Santa Ana2002);  Valles (Reference Valles2021); Gonzalez Van Cleve (Reference Gonzalez Van Cleve2016); Flores et al. (Reference Flores, Escudero and Burciaga2019); Fleury-Steiner et al. (Reference Fleury-Steiner and Fleury-Steiner2009); Longazel (Reference Longazel2012) and Gonzalez Van Cleve and Mayes (Reference Gonzalez Van Cleve and Mayes2015).

Note: See Appendix 1 for the example code word sources.

Sampled code words expressing Parasite convey that Racial/Ethnic Minorities exploit communities, institutions, services, amenities or other aspects of the U.S. social system. They close to equally refer to Black, Latinx and Other people. Most present Racial/Ethnic Minorities as “Takers not Makers,” as exploitative recipients of and not contributors to social welfare programs who have deep personal and cultural deficiencies within economic spheres. The most pervasive example of Taking is ‘welfare queen,’ which Presidential Candidate Ronald Reagan famously used to refer to Black women living extravagantly on public benefits (S65: 58). Another is a news commentator defending the War on Drugs (which disproportionately criminalized Black people) by describing those targeted as ‘living on welfare, food stamps, Medicaid, and Section 8’; the commentator goes on to use Not Making code words by arguing that their communities ‘don’t promote a strong work ethic because most … don’t work themselves’ (S84: 258). A small portion presents Racial/Ethnic Minorities as “Cheaters,” conveying their perceived exploitation of job markets and higher education through biased hiring, promotion and admissions processes (e.g., calling a Person of Color in the workplace an ‘affirmative action hire’ (S76: 113)). Closely related to Parasite is the category “Deficiency,” which expresses anxieties that Racial/Ethnic Minorities (particularly low-income and working-class children and their parents) lack personal traits necessary to contribute to the economy, which may eventually threaten White people and society when they become a Parasite (Valles Reference Valles2021; López-Espino 2023). High school teachers referring to Racial/Ethnic Minority students as ‘at-risk kids’ is an example (S80: 124–125). Parasite and Deficiency code words echo Gonzalez Van Cleve (Reference Gonzalez Van Cleve2016)’s observation of “mope” – a lazy, incompetent degenerate who drains societal resources – as a racially coded term that prosecutors, judges and defense attorneys construct to describe poor Black and Latinx defendants charged with nonviolent crimes and frame them as undeserving of legal protection (ibid: 57–61, 69).

Contagion expresses that Racial/Ethnic Minorities spread something harmful within communities, institutions or other societal domains. Agents of harm include attitudes, values, behaviors and conditions. These most refer to Latinx people in our sample – echoing Santa Ana’s studies of public discourse about Latinx people in 1990s California (Santa Ana Reference Santa Ana2002; Santa Ana et al. Reference Santa Ana, Juan and Sanchez1998), followed by Black and Other people. Like Parasite, Contagion code words exist within socioeconomic spheres of life, but the object of harm is geographic: (1) the physical “Invasion” of White and potentially other racial and ethnic communities by Racial/Ethnic Minorities and (2) the “Corruption” of these communities by conditions associated with Racial/Ethnic Minorities. Invasion code words often use deixis like “these” or “those” to differentiate outsiders from insiders, such as White neighbors referring to Black people on their block as ‘these people’ in expressing safety concerns (S23: 290). Illegal and variants were common Invasion code words for Latinx people. An example is the Tea Party rally sign ‘Freeloading Illegals are Raping U.S. Taxpayers,’ which derives additional potency from its integration of Parasite (‘Freeloading’) and White Respectability and Privilege (‘Taxpayers’) code words (and seeds the trope of Villain (described below) in ‘Raping’) (S65: 152). An example of a Corruption code word is referring to Black and Latinx students attending a White neighborhood high school as the ‘bad element’ (S23: 289). Contagion code words often reflect users’ “legal-spatial consciousness” of how race shapes rights to occupy and behave in space, their attention to transgressions of these norms and their role in reinforcing them (Flores et al. Reference Flores, Escudero and Burciaga2019; Shabazz Reference Shabazz2015).

Villain expresses that Racial/Ethnic Minorities cause direct harm to White people and potentially others by threatening personal safety or private property. These sampled code words most refer to Black people, followed by Middle Eastern and Latinx people. They commonly convey fear that Racial/Ethnic Minorities will commit crimes against White people. The subcategory “Suspect” expresses concern about Racial/Ethnic Minorities engaging in crime based on their comportment, place of residence or behaviors, such as when a juror in a bank robbery case observed about the Latinx defendant, ‘I guess we’re profiling but they cause all the trouble’ (S3: 1135). The subcategory “Offender” addresses perceptions that Racial/Ethnic Minorities commit violent or property crimes against White people. One example is politicians and journalists calling Black people protesting the death of a young Black man in police custody ‘thugs’ (S26: 350). Another is a White high school student telling a Middle Eastern peer, ‘Shut up, terrorist’ (S67: 85). Like Contagion, Villain implies a moral-legal cartography that creates “new forms of knowledge of space, self, and other” and calls for White defense and harsher criminalization of minorities, particularly Black and Latinx people (Fleury-Steiner and Fleury-Steiner Reference Fleury-Steiner and Fleury-Steiner2009: 6; Longazel Reference Longazel2012; Gonzalez Van Cleve and Mayes Reference Gonzalez Van Cleve and Mayes2015; Shabazz Reference Shabazz2015).

Functions

Most of the sampled racial code words have racist effects. Some perpetuate commonsense racism and contribute to a “racial episteme” (a total system of knowledge and knowing about race) that portrays White people and communities as respectable and duly privileged and Racial/Ethnic Minorities and their communities as pathological and duly inferior.Footnote ⁶ Others contribute to systemic racism by supplying the substantive justifications for policies and practices that perpetuate racial inequality. The linguistic mechanisms of euphemism, metonymy and othering enable code words to imbue racial meaning and perpetuate these forms of racism by giving users varying degrees of efficient, adaptable and plausibly deniable race talk (see Table 4). Some evolve, assisted by a technique we call compounding.

Table 4. Mechanisms of sampled code words

Sources:  Authors’ own work. Mechanisms adapted from Cameron (Reference Cameron2012); Crespo-Fernández (Reference Crespo-Fernández2018); Rich (Reference Rich2004); Onwuachi-Willig and Barnes (Reference Onwuachi-Willig and Barnes2005); Carbado and Gulati (Reference Carbado and Gulati2013); Shabazz (Reference Shabazz2015); Littlemore (Reference Littlemore2015); Plumwood (Reference Plumwood1993); Pandey (Reference Pandey2004); Hall (Reference Hall and Hall2013) and Lens (Reference Lens2009).

Note: See Appendix 1 for the example code word sources.

From covert race talk to racist effects

Santa Ana observes that discourse “antecedes society” and “speaks society into existence” by creating categories, conceptual mappings, worldviews and institutions (Reference Santa Ana2002: 317). Our findings affirm his insight. Most of the sampled code words intentionally or unintentionally contributed to racist effects. Many perpetuated commonsense racism by promoting White people’s respectability or Racial/Ethnic Minorities’ pathology (Haney López Reference Haney López2003; Obasogie Reference Obasogie2010). Examples include high school teachers and administrators referring to Black students as ‘Special Ed kids,’ a journalist referring to a White murder victim as an ‘upstanding citizen,’ and Newt Gingrich referring to Black people as those who have ‘no habit of showing up on Monday’ (S80: 182–183; S44: 11; S57: 102). These examples, and the eight common tropes that our sample of racial code words conveys, show how racial code words contribute to epistemic racism – routine and automatic discursive practices that link Racial/Ethnic Minorities (and other minoritized groups like women and people from the Global South) with inferior characteristics like unintelligence, animality and irrationality and link White people (mostly “Western” White men) with superior qualities like higher intelligence, civility and rationality (Beagan et al. Reference Beagan, Bizzeth, Sibbald and Etowa2024; Bhimull et al. Reference Bhimull, Hecht, Jones-Imhotep, Mavhunga, Nakamura and Siddiqi2022; Grosfoguel Reference Grosfoguel2010; Reference Grosfoguel2013; Kubota Reference Kubota2020). Although racial code words in our sample rarely rely on metaphors for creation, the narratives they tell about Racial/Ethnic Minorities are mostly metaphors: Parasite, Contagion and Villain.Footnote ⁷ These vivid, rich and familiar metaphors perform powerful “conceptualizing and signaling functions” (Santa Ana Reference Santa Ana2002: 59), enabling the production of a racial episteme that portrays White people and their spaces as superior and Racial/Ethnic Minorities and their spaces as inferior. These findings indicate that racial code words have become building blocks of a “colorblind” epistemic racism in post-civil rights America.

Other racial code words were used to endorse a policy or practice that perpetuated racial inequality and White privilege. These speakers were frequently a White person advocating for a supposedly pro-White policy or practice or opposing a supposedly pro-Racial/Ethnic Minorities policy or practice. The most common was harsher policing and penalization of Racial/Ethnic Minorities, particularly Black people. Examples include a neighbor advocating for increased policing to stop ‘hooligans’ and a gun rights group calling for ‘responsible gun ownership’ to intimidate Black people – kinds of discourse that Carlson (Reference Carlson2020) labels “gun militarism” and “gun populism,” respectively (S23: 289; S14: 701). Use of racial code words to support harsher penalization of Racial/Ethnic Minorities affirms existing studies on how mainstream discourse on crime perpetuates racial power and control by creating state apparatuses that, instead of governing crime, govern through crime (Fleury-Steiner and Fleury-Steiner Reference Fleury-Steiner and Fleury-Steiner2009; Gonzalez Van Cleve and Mayes Reference Gonzalez Van Cleve and Mayes2015; Murakawa and Beckett Reference Murakawa and Beckett2010; Simon Reference Simon2007). Other common policy-positions included:

1. (1) Delegitimizing or sabotaging policies, initiatives or individuals that sought to empower minorities. For example, conservative critic Craig Smith referred to Barack Obama as a ‘hip-hop president’ (S4: 138).

2. (2) Opposing welfare programs. For example, White politicians referred to welfare recipients as ‘welfare queen[s],’ ‘people on food stamps’ and ‘people who are on crack’ (S35: 2125; S72: 1164; S57: 60).

3. (3) Opposing affirmative action programs. For example, White research participants highlighted that affirmative action beneficiaries ‘don’t keep the averages up’ and are ‘destined to fail’ (S13: 19; S52: 30).

4. (4) Advocating stricter immigration policy. For example, Republican politicians argued that ‘bad hombres’ and ‘illegal aliens’ stem from “open borders” (S35: 2125; S9: 131).

Most private practices involved race-based exclusion, such as from a bar, shopping mall, neighborhood, newspaper, job, intimate relationship or educational or medical resource, such as school children avoiding a ‘terrorist’ peer (S67: 85).

Notably, many code words both perpetuated commonsense racism and supported a racist policy or practice. This shows how covert racism and institutional racism are mutually constituted and reinforced (e.g., Gonzalez Van Cleve and Mayes Reference Gonzalez Van Cleve and Mayes2015; Murakawa and Beckett Reference Murakawa and Beckett2010). Code words that perpetuate commonsense racism sustain a racist cultural/epistemic ecology, supplying the substantive justifications for policies and practices that perpetuate racial inequality. In turn, these policies and practices produce appearances of White people’s respectability and privilege and Racial/Ethnic Minorities’ pathology and inferiority, thus justifying commonsense and epistemic racism (see also Haney López Reference Haney López2006; Lawrence Reference Lawrence1987). Like overt race talk, racial code words direct attention away from racist systems and structures and toward the racial difference of the disadvantaged subject (Gonzalez et al. Reference Gonzalez, Simon and Rogers2022, citing Fields and Fields). However, more convenient than overt race talk, racial code words cloak the co-dependent cultural and systemic racism with plausible deniability and racial innocence and open a new space for cultural/epistemic and systemic racism to co-evolve in a “colorblind” society (Bonilla-Silva Reference Bonilla-Silva2022; Omi and Winant Reference Omi and Winant2014).

Euphemism, metonymy and othering: mechanisms for efficient, adaptable and plausibly deniable race talk

Linguistic mechanisms are instrumental to overt and covert racial discourses (e.g., metaphor in Santa Ana Reference Santa Ana2002; “metapragmatic dismissals” in López-Espino 2023). Santa Ana (Reference Santa Ana2002, Reference Santa Ana, Juan and Sanchez1998) analyzes how use of metaphors such as “immigration as dangerous waters,” “immigrant as animal,” “affirmative action as disease” and “nonEnglish language as barrier” shaped public discourses on immigration, affirmative action and bilingual education and contributed to the passing of anti-Latinx referenda on these issues in 1990s California. We found that metaphors were used mostly to explicitly racialize a subject in our research and hence were not codes, with a few exceptions, such as referring to Latinx immigrants as ‘anchor babies’ (e.g., S22: 35; S57: 42; S14: 712).

Speakers in our sample of racial code words commonly used euphemism, metonymy and othering to covertly racialize a subject and, in turn, perpetuate epistemic and/or systemic racism. Euphemism is the use of agreeable, milder or inoffensive language to refer to something disagreeable, harsher or potentially offensive about a racial group or its members (Cameron Reference Cameron2012). An example is a master’s student using ‘not diverse’ to refer to the harsher fact that Racial/Ethnic Minorities are absent from her classes (S8: 165). Euphemist code words in our sample exhibit three, often overlapping forms: understatement, under-specification and overstatement (Crespo-Fernández Reference Crespo-Fernández2018). Understatements use fuzzy language to lighten a serious topic, such as an online news commentator using ‘old boys network’ to refer to practices of hiring only White men (S53: 197). Under-specifications use general language to refer to something more specific, like an online news commentator using ‘certain group of people’ to refer to Black male athletes perceived to have “foul mouths” at a local mall (S64: 245). Overstatements use upbeat exaggerations to highlight something desirable about the referent, such as an Apple supervisor reminding his employee that ‘Apple products are for the premium market’ (i.e., not Black people) when asked whether “African American English” should be incorporated into Siri speech recognition software (S85: 28).

Euphemism makes conversation polite by downplaying racial dimensions and protects from charges of racism. For example, a Southern politician criticized the growing struggle for racial justice in America with a remark that had the rest of the country voted for Strom Thurmond (a presidential candidate who campaigned on racial segregation) like his state did, ‘we wouldn’t have had all these problems over all these years’ (S48: 203). This latter scenario also exhibits “dog whistle racism,” where politicians knowingly convey racial meaning while denying race talk (Bennett and Walker Reference Bennett and Walker2018; Haney López Reference Haney López2014). Other examples include calls for ‘law and order’ and ‘states’ rights’ and to ‘Make America Great Again’ in the Nixon, Reagan and Trump Presidential Campaigns and Presidencies, respectively (S14: 690; S59: 72; S39: 36).

Metonymic racial code words refer to a racial group or its members using a contextually understood closely associated feature (Littlemore Reference Littlemore2015; Radden and Kövecses Reference Radden, Kövecses, Panther and Radden1999). They have two parts: the descriptor (a racially salient feature) and the described (a member, the entirety or another feature of a racial group). The descriptor-described may be part-whole, whole-part or part-part. Timberland, a shoe brand, is a part that represents the whole of a Black man in a bar’s rule of ‘no Timberlands at the door’ (S47: 135). People, a word for society as a whole, represents the part of White people in a White woman’s observation that ‘people’ get mad at ‘people of other races’ for competing for jobs (S47: 158). Bussing, a part of school racial desegregation, represents another part – Black students going to White schools – in Nixon’s rhetoric against the practice during his presidential campaign (S87: 124). The relationship linking the descriptor and the described may be physical (e.g., police officers referring to a Muslim peer as “towel head”) or conceptual (e.g., high school teachers referring to White students as ‘Gifted and Talented’) (S79: 1818; S80: 124–125). CRT scholars have analyzed various types of coded racism through metonymy, such as discrimination by proxy (Onwuachi-Willig and Barnes Reference Onwuachi-Willig and Barnes2005; Rich Reference Rich2004) and intra-racial discrimination directed against a racial subgroup sharing a common feature, e.g., unassimilated Black people (Carbado and Gulati Reference Carbado and Gulati2013).

The variants Racializing the Location and Inflating the Social Member or Subgroup metonymy were common in our sample. Racializing the Location metonymy refers to a racial group through a closely associated geography. Sampled code words for Black people and White people most frequently convey this. Examples include generalized geographies like ‘inner city’ and ‘urban’ (racialized as Black) and ‘suburban’ (racialized as White) and specific geographies like “Chicago,” “Compton,” ‘Detroit’ and ‘the Bronx’ (racialized as Black) and ‘the West Coast’ (racialized as White) (e.g., S65: 214; S78: 480; S8: 165; S76: 96; S97: 27; S80: 128; S61: 182). Others include a healthcare case worker’s reference to their Black and other Racial/Ethnic Minority clientele as ‘the inner city kids that we deal with’ and a White fraternity mocking Black History Month by hosting a gathering called ‘Compton Cookout’ (S37: 477; S90: 63). The utility of Racializing the Location metonymy stems from enduring imagined and real spatial racial segregation in America, through housing types (e.g., the ‘projects’ racialized as Black), city sections (e.g., ‘South Side’ racialized as Black), entire cities (e.g., ‘West Sierra Linda’ racialized as Racial/Ethnic Minorities in general) and regions (e.g., ‘rural’ racialized as White) (S38: 115; S41: 97-98; S15: 21; S85: 22; Shabazz Reference Shabazz2015).

Inflating the Social Member or Subgroup metonymy refers to a racial group using a notable member or a subgroup. Notable members in our sample include real persons (e.g., ‘Osama’ racialized as Middle Eastern and ‘Jefferson Davis’ racialized as White), stereotypes of real persons (e.g., ‘welfare queen’ racialized as Black), archetypical fictional individuals (e.g., ‘Uncle Tom’ as a Black person who betrays other Black people to curry favor with White people) and generic fictional individuals (e.g., ‘Alice Tang’ racialized as Asian, ‘Tyrone Jackson’ racialized as Black and ‘Joe’ racialized as White) (e.g., S67: 81; S57: 41; S35: 2125; S40: 234; S85: 10; S28: 62). Subgroups often convey racial stereotypes. For instance, a White woman moving into a racially “mixed neighborhood” received advice to stick to streets with ‘university or professional looking people,’ a code word conveying respectable White people in the neighborhood (S23: 290). Other examples are ‘people on food stamps’ (racialized as Black and conveying Parasitism) and “terrorists” (racialized as Middle Eastern and conveying Villainy) (S72: 1164; S65: 36).

Metonymies facilitate race talk not only by achieving plausible deniability through racial proxies but also by reducing the informational, emotional and reputational costs of communication. They pare a “large amount of information” down into a simplified “manageable form,” enabling speakers to use this skeletal form to access the complex concept (Littlemore Reference Littlemore2015: 4–5). For example, wearing ‘big chains’ congeals the perceived pathologies of a young Black man: criminality, irresponsibility, poverty and hypersexuality (S47: 135). Like euphemism, they also relieve speakers from experiencing personal stress and adverse social consequences, like cancellation, that might stem from naming race.

Othering, like metonymy, also contributes to efficient race talk (Santa Ana Reference Santa Ana2002; Santa Ana et al. Reference Santa Ana, Juan and Sanchez1998). It creates a qualitative difference and ideological distance between two racial groups, placing them on the opposing sides of a contextually salient binary of us vs. them (Hall Reference Hall and Hall2013; Lens Reference Lens2009; Pandey Reference Pandey2004; Plumwood Reference Plumwood1993). Othering contrasts sharply with euphemism by explicitly using disagreeable, harsh or potentially offensive language to forge racial difference. Stereotyping is an important tool of othering. It ignores the internal differences among the otherized group, makes them “appear suitably homogeneous” and creates categorical differences that place them in a position of inferiority, which “ground[s] hierarchy” (Plumwood Reference Plumwood1993: 53–55). Otherizing differences in our sample include:

1. (1) Total and insurmountable. Black and Latinx people were dehumanized through words such as ‘animal’ and ‘demon’ (S89: 237; S40: 56).

2. (2) Moral. Some politicians portrayed Latinx people as foreigners who took advantage of America’s social services through producing ‘anchor babies’ (S22: 35).

3. (3) Legal. Latinx and Asian people’s citizenship and belonging were questioned through words such as ‘illegals’ and ‘fobby,’ respectively (S97: 118; S75: 519).

4. (4) Cultural. ‘eat dog’ was racialized as an Asian practice, and “towel head” was a racial proxy for Middle Eastern men (S82: 195; S79: 1818).

5. (5) Interpersonal. ‘rude,’ ‘angry,’ ‘loud’ and ‘opinionated’ were microaggressions against Black people (S47: 81; S76: 205, 109).

These examples reveal that othering is frequently deployed to target Latinx immigrants, a phenomenon also noted by Santa Ana (Reference Santa Ana2002, Reference Santa Ana, Juan and Sanchez1998). While euphemism cloaks race and value judgments in nicer-sounding substitutes, othering amplifies value judgment but avoids direct connections to race. Othering weakens plausible deniability, which makes decoding easier. Yet, it is commonly used despite this risk because, like metaphor and metonymy, it is fundamental to how we think, communicate and form a cultural community. Cultural formation and order depend on sensemaking through differentiation and categorization and marking boundaries against intruding or impure symbolisms (Hall Reference Hall and Hall2013).

Othering often was present among sampled code words that contributed to racist effects, which reveals its potency as a discursive craft of racism (Gonzalez et al. Reference Gonzalez, Simon and Rogers2022; Hall Reference Hall and Hall2013; Pandey Reference Pandey2004). An example is Tea Party activists’ conception of society as made up of ‘workers’ and ‘nonworkers,’ ‘productive citizens’ and ‘the freeloaders,’ which creates a binary of socioeconomic virtue and vice to oppose poverty assistance programs and perpetuate commonsense racism (S1: 14). When stereotype infuses with law, othering is amplified. For example, a Tea Party-backed candidate claimed that ‘waves of illegal aliens [stream] across our border, joining violent gangs, forcing families to live in fear’ (S9: 131). Here, ‘illegal aliens’ conveys legal and moral difference; the law’s making and marking of human beings as citizens, legal immigrants and illegal aliens and the trope of Villain cross fertilize to produce powerful racist effects (Gonzalez Van Cleve Reference Gonzalez Van Cleve2016; Longazel Reference Longazel2012).

Compounding: a catalyst in the lifecycle of racial code words

Racial code words are birthed through the marriage of tropes and linguistic mechanisms like those described above, which enable users to realize complex meaning in an efficient, adaptable and plausibly deniable way. Yet their codedness may wax and wane, with some becoming decoded and transitioning to explicit race talk and others becoming coded and decoded again.

Heavily scrutinized racial code words sit at a crossroads. One path is to become recoded through the discursive practice we call compounding. This technique capitalizes on the syntactic versatility of language by enabling users to stack multiple code words together, graft one code word to a racially neutral expression or remix parts of code words into a new expression to create a new code word. Stacking enables users to efficiently communicate richer or more precise racial meaning, such as a business leader referring to White areas as ‘suburbs in the more affluent communities [that have] college-going culture and much more academically rich environment’ in explaining reasons for enduring social inequality in education (S15: 18–19). Stacking also allows users to recode language decoded in public and academic discourse, like terrorist and street gangs becoming ‘domestic terrorist street gangs’ in a news article commentator advocating for the continuation of the war on drugs (S84: 257). Grafting and remixing enable users to create new code words to racialize new phenomena, like ‘political thug,’ ‘food stamp president’ and ‘gangster government,’ which evolved after the election of President Obama, the first Black President (S26: 351; S14: 707; S57: 48–49). Similarly, a White House staff member’s remixing of ‘kung flu’ cleverly implicates Chinese people in the spread of the 2020 coronavirus pandemic, which first broke out in China (S50: 654). Compounding responds to the inherently unstable qualities of coded race talk, a dynamic that sociolegal scholars also attribute to “rights talk” (Dudas Reference Dudas2005).

Another path is for the racial code word gradually to become more exposed as explicitly racist. Haney López (Reference Haney López2014) traces the political life of code words like bussing, law and order and welfare queen by showing how their racial meanings become uncloaked by scholars, media and others through their overuse by politicians and pundits during election cycles and political administrations. It is at this stage that the race talk might become fully decoded and enter the lexicon of explicitly racial words (e.g., towel head).

The birth and death of individual racial code words do not affect the continuous construction of the racial episteme, because these individual building blocks are perfectly dispensable and replaceable. Like in metaphors, what makes the building block effective is not the particular string of words used, but the conceptual linkages that the words invoke (Santa Ana Reference Santa Ana2002: 30). Other strings of words can invoke the same linkages. Hence, while individual racial code words are dispensable, racial code words as building material are inherently regenerative and indestructible. They are “[s]mall acts of cunning,” endowed with diffused power, subtle arrangements and apparent innocence, and must be subject to profound suspicion (Foucault Reference Foucault1995: 139).

Detecting the codes: a racial meaning decoding tool

We now present a “racial meaning decoding tool” to assist researchers, legal professionals and social justice advocates in detecting coded race talk. The tool integrates our empirical insights – that racial code words deploy tropes and linguistic mechanisms to convey racially stratifying narratives in efficient, adaptable and plausibly deniable ways, often to racially stratifying effects – with a synthesis of CRT insights discussed earlier from Lawrence (Reference Lawrence1987), Rich (Reference Rich2004) and Onwuachi-Willig and Barnes (Reference Onwuachi-Willig and Barnes2005). The tool can assist in evaluating whether words suspected to fit our definition of racial code words – indirect signifiers of racial or ethnic groups that contain at least one positive or negative value judgment and contextually implied or salient meanings (see our explanation in “Canvassing Racial Code Words from Scholarly Literature”) – are potentially racially coded and used for disequalizing purposes. Application involves conducting three analyses and scoring the words in question based on their outcomes. Higher scores imply a higher probability of racial coding and use for disequalizing purposes (see Table 5).

Table 5. The racial meaning decoding tool

Sources: Authors’ own work. Steps adapted from Lawrence (Reference Lawrence1987); Rich (Reference Rich2004) and Onwuachi-Willig and Barnes (Reference Onwuachi-Willig and Barnes2005).

Note: The tool should be applied to words suspected to be indirect signifiers of racial or ethnic groups that contain at least one positive or negative value judgment and contextually implied or salient meanings.

The first step is thematic analysis. This involves assessing whether the words might have racial meaning by determining whether at least one of the eight common tropes of racial code words shown in Tables 2 and 3 is present. The second step is effect analysis. This requires evaluating whether the use of the words could lead to a disequalizing effect on racial power relations. Examples of disequalizing effects include (1) commonsense racism by promoting White people’s respectability or Racial/Ethnic Minorities’ pathology, (2) a policy or practice that perpetuates racial inequality and White privilege and (3) other disequalizing effects. The third step is linguistic analysis. This involves assessing whether the words might have qualities that help to obscure or efficiently convey racial meaning by discerning whether at least one of the three common mechanisms of racial code words shown in Table 4 is present. Words receive a score of 1 for each analysis passed and 0 for each analysis not passed or with unclear outcomes. The final step is to tally the scores of the three analyses. Words scoring between two to three are potentially racially coded and used for disequalizing purposes, while those scoring zero or one are likely not.

Taking two additional steps can help to strengthen the rigor of the tool’s application. First, the user should try to apply shared societal understandings of the words in question, particularly those from the speaker’s and audience’s communities (Rich Reference Rich2004). Integrating local contextual data can help to clarify shared understandings. For instance, demographic data showing disproportionate tendencies of particular racial and ethnic groups to have particular conditions can help illuminate the use of metonymy (e.g., data on differences in geographic location for Racializing the Location metonymy or housing tenure for Inflating the Social Member or Subgroup metonymy). Second, ideally at least two people should conduct the analyses and cross-validate their outcomes. Scores of 1 should be given only when their outcomes agree; scores of 0 should be given when their outcomes are unclear or disagree.

We illustrate the utility of the tool through two examples of words reflecting grounds for opposition to allowing for denser housing development in Arizona (Eland Reference Eland2022; Rosequist Reference Rosequist2022). In the first example, a resident of a predominately White and single-family home suburb opposes updating a plan to allow for denser mixed-use development, because they ‘didn’t move here to be in an urban city.’

Thematic analysis: The region’s racially segregated housing patterns (higher concentration of Racial/Ethnic Minorities and denser housing in the central city and higher concentration of White people and single-family housing in the suburbs) establish a plausible link between race and geography, with the city imagined as Minority and the suburb imagined as White. In this context, the resident’s invocation of urban city plausibly reflects a Contagion theme that the planning change would bring Racial/Ethnic Minorities (and presumed negative consequences) to the suburb. Scoring = 1.

Effect analysis: The ground for opposition is expected to hinder Racial/Ethnic Minorities from living in the suburb, given that they are disproportionately likely to live in denser, multifamily housing in the region. Scoring = 1.

Linguistic analysis: The geographic and racial correlations shown in the thematic analysis reveal that the speaker may be using the linguistic mechanism of metonymy, referring to Racial/Ethnic Minorities by a closely associated geography (urban city). Scoring = 1.

Decision: Total score = 3. The ground for opposition is potentially racially coded and used for disequalizing purposes.

In the second example, residents of a predominately White small town of large lot homes oppose a denser and mixed-use modular housing project targeted to mining workers because it will increase the ‘transient’ population.

Thematic analysis: Most workers employed in occupations related to mining in the region identify as White. The ground for opposition reflects none of the common tropes about White people. However, a sizable minority of these workers identify as Racial/Ethnic Minorities. The ground for opposition may reflect the trope of Contagion if it references miners who are Racial/Ethnic Minorities, since the residents perceive that transient workers may bring harm to the town. Overall, the presence of tropes is uncertain. Scoring = 0.

Effect analysis: It is uncertain whether excluding mining workers from a predominately White town would have a racially disequalizing effect, since it is likely that the mining workers who would live in the proposed housing would be mostly White people. Scoring = 0.

Linguistic analysis: Transiency is a characteristic of miners; thus, metonymy is present. Scoring = 1.

Decision: Total score = 1. The ground for opposition is likely not racially coded and used for disequalizing purposes.

A strength of the tool is that it is designed to deal with the uncertainty and contestability of coded language. It evaluates the possibility of racial coding by investigating multiple dimensions of the words in question and placing them on a spectrum. The deniability of racial meaning is less plausible for words placed higher on the spectrum (i.e., with scores closer to 3) and vice versa for those placed lower (i.e., with scores closer to 0).

The tool has academic as well as practical utility. Researchers can use it to study coded race talk in particular social settings and verify, challenge and refine it for these settings. The tool also encourages legal professionals and social justice advocates to continue inquiry into race-based discrimination claims where antidiscrimination law has so far left off and to update the right to nondiscrimination for a “colorblind” society. In antidiscrimination lawsuits/investigations, legal and other advocates can apply the tool to scrutinize clients’ encounters with suspicious words or proffered justifications and analyze whether they encoded plausible racial meaning in their particular context and were used to produce racially disequalizing effects.

Applying the tool will be challenging and risky. Discerning racial coding is a delicate and fraught process. People’s positionalities influence how they perceive meaning, as previously noted. Racial coding can take myriad forms and no words are indispensable to conveying a racial message. If one word is decoded, others can take its place. Racial code words can adapt to changed circumstances and be invented spontaneously by drawing on potential cues that appear in a discourse. If evidence to support a particular decoding falls short of being incontrovertible, researchers are vulnerable to accusations of being racist (Haney López Reference Haney López2014). Sharing Lawrence’s assessment about his cultural meaning test, the tool “will be beset by the complexities and inadequacies of social interpretation and buffeted by the head winds of political resistance” (Reference Lawrence1987: 388). While deconstructing the edifice of coded racism one building block at a time is slow and laborious, each of the numerous, diffuse, local and quotidian settings of racial discourse is a potential site of discursive and power remaking. Similar to Santa Ana’s call to invent “insurgent metaphors” (Reference Santa Ana2002), to the extent an ever-evolving and regenerative reservoir of racial codes is perpetuating racial domination, society should build a counter-reservoir and systemically, programmatically and ceaselessly uncloak the codes to equalize racial power. Our tool is a first effort in this direction.

Building knowledge to further deconstruct racial code words

Despite claims by the Supreme Court and others that colorblindness is equal treatment and that race-based remediation efforts are a form of racial discrimination (Students for Fair Admissions v. Harvard 2023; Dershowitz Reference Dershowitz2021), racism and racial stratification continue in post-civil rights America, partly under the disguise of colorblindness (Bonilla-Silva Reference Bonilla-Silva2022; Powell Reference Powell2023). Coded language is an instrument of this “colorblind racism.” Through a thematic analysis of 734 racial code words from 97 scholarly texts, this research introduced a new interpretive framework and decoding tool to inform and stimulate future investigations of racially coded language.

We identified eight common tropes of racial code words, four for White people and Whiteness and four for Racial/Ethnic Minorities. Tropes for White people and Whiteness (White Respectability and Privilege, White Possessiveness, White Defense and Passing) convey that White people (1) are respectable and duly privileged and have exclusive possession of America and (2) must conduct self-defense against threats from Racial/Ethnic Minorities, who in turn must strive to assimilate to White norms and values. Tropes for Racial/Ethnic Minorities (Parasite, Deficiency, Contagion and Villain) convey that they are socially, economically, morally and developmentally deficient people who cause indirect, direct or possible future harm to White people and communities and society more broadly. Through these tropes, racial code words fuel commonsense racism and a racial episteme that portrays White people and their spaces as superior and Racial/Ethnic Minorities and their spaces as inferior. They also contribute to systemic racism by supplying the substantive justifications for policies and practices that perpetuate racial inequality. These policies and practices, in turn, produce appearances of White people’s superiority and Racial/Ethnic Minorities’ inferiority, thus justifying commonsense racism (Haney López Reference Haney López2006; Lawrence Reference Lawrence1987).

Certain linguistic mechanisms enable racial code words to perpetuate both epistemic and systemic racism. Euphemism and metonymy obscure the racial nature of the discourse, creating a “colorblind discursivity.” The ensuing plausible deniability helps to neutralize, normalize and legitimize an otherwise troubled discourse of knowledge production (Foucault Reference Foucault1990). This enables speakers to engage in disavowal, “a strategy by means of which a powerful fascination or desire is both indulged and at the same time denied” (Hall Reference Hall and Hall2013: 257). Mechanisms like metonymy and othering also decrease the costs of racial communication by reducing complex meanings into a succinct code. Metonymy allows speakers to use a low information-cost concept to access high information-cost concepts. Othering further lowers these costs by reducing the complexity, fluidity and temporality of a person or a group to a few objectified traits (ibid: 247). As a code word gets exposed, or repeated use unravels its congealed meaning, or a new circumstance demands a new code, compounding allows speakers to creatively reconfigure racial code words to convey new and evolving racial meanings. Techniques of syntactic stacking, grafting or remixing re-fix an unraveling meaning (e.g., inner city crime) or reinvent meaning for new purposes (e.g., political thug), fueling their evolution (Dudas Reference Dudas2005). The creation of racial codes from tropes, the acceptance of codes as commonsense, the unraveling or delegitimization of coded meanings and the eventual destruction and recreation of codes form the life cycles of racial code words.

Our research informs and extends empirical studies of racially coded language. Theoretically, we narrow a disciplinary gap between sociolegal studies of race and CRT. Our interpretive framework reflects sociolegal scholars’ calls to study race and racism as processes in social systems (Gómez Reference Gómez2012; Murakawa and Beckett Reference Murakawa and Beckett2010) by showing how use of racial code words in micro-instances constructs race and perpetuates epistemic and systemic racism in society. Our decoding tool integrates insights of CRT scholars (Lawrence Reference Lawrence1987; Rich Reference Rich2004; Onwuachi-Willig and Barnes Reference Onwuachi-Willig and Barnes2005) with our findings to assist researchers, legal professionals and social justice advocates in examining multiple dimensions of suspected words and placing them on a scored spectrum of possible coding. Practically, our tool can aid legal professionals and social justice advocates in confronting covert racism and uncloaking its various disguises, as recent retrenchments in civil rights by the Supreme Court and some state legislatures and courts exemplify the current battles over meaning and authority of interpretation.

We stress that our framework and tool are only the first steps in a long journey of developing more rigorous methods to detect racial code words, discern their mechanisms and understand how they contribute to epistemic and systemic racism. Discerning meaning from language – particularly indirect, contextually dependent and plausibly deniable language like racial code words – is difficult. This task is made harder in our research by our use of secondary data from scholarly literature, data that has been filtered at least once and sometimes twice or more by others’ perceptions. This quality leads to an iterative loss of contextual information valuable in studying race talk and increases the risk of misrepresenting and misinterpreting it. Our data also are biased by practical challenges encountered in comprehensively capturing and conveying racial code words in a large sample of scholarly literature. Our research is a necessary but incomplete attempt at muddling through these challenges to offer preliminary insights on a topic as timely and harmful as racial code words.

Our larger aim is for our framework and decoding tool to spark more research in sociolegal studies on how racial code words facilitate unequal power relations through epistemic and systemic racism. First, research is needed on how racially coded speech in distinct contexts shapes specific policies and practices, and how these in turn entrench racial power hierarchies and complicate remediation efforts. Researchers can use our tool to help discern which words might be racially coded in particular settings, such as legal education, jury deliberations, police officer testimony and local government decision making, and help refine the tool’s use for these settings. More case studies of racial code words in particular places, like Gonzalez Van Cleve (Reference Gonzalez Van Cleve2016)’s investigation of Chicago area criminal courts, would help to not only understand their racially stratifying functions but also conceptualize how they vary across different geographic, demographic, situational and discursive contexts. There also is an opportunity to uncover how these contexts relate to one another thematically and temporally within places (e.g., how code words are reinforced, complemented or contradicted among traditional and social media and public meeting contexts). Research on the ecologies and modes of functioning of racial codes in settings regulated by antidiscrimination law especially is warranted, as our sample features only a small number of code words in these settings.

Second, research into racial code words’ life cycles and continuums of harm is needed. Why do some retain their coding while others become decoded? Racial code words may have thresholds of efficacy; as they are widely adopted, their effectiveness may first grow but later shrink due to increased exposure and scrutiny. Why are some more effective in achieving racist outcomes than others? Discursive practices like othering may play a role. Code words that exhibit total and insurmountable forms of othering and thematic dimensions of Villain are most harmful, while those that convey thematic dimensions of Passing are less harmful, though there is harm in how they function to codify White normality and erase racial and ethnic differences. Why do some persist after decoding while others perish or evolve? While racial meanings and codes change over time, some may change faster than others, including in particular social contexts and historical periods. There is opportunity to marry historical, sociological and discourse analysis in building continuums of harm and tracing pathways of codedness in various settings, particularly those relevant to antidiscrimination law. Despite its limitations, antidiscrimination law can be a more powerful tool for positive social change if we invest in developing, applying and refining methods to expose racial codes and their functions in the reproduction of racism and racial stratification.