Demarcation and Difference: Language and Indigenous Self-Identity in Latin America

Abstract

Indigeneity remains a major axis of stratification in Latin America, making questions of when and why people identify as Indigenous central to understanding and addressing inequality in the region. Using representative Latinobarometer data from 16 countries, collected between 2007 and 2020, the authors analyze the two most widely used instruments for the identification of Indigenous people in Hispanophone America: self-identification and respondents’ mother tongue. Descriptive analysis shows that though the majority of respondents are nonspeakers and nonidentifiers, nonnegligible proportions of respondents are identifiers and speakers, identifiers and nonspeakers, and nonidentifiers and speakers. In multivariate analyses, the authors test factors associated with identification conditional on linguistic background. The findings support hypotheses that privilege and, in some cases, social mobility are negatively associated with Indigenous identification and of race as a “master category.” The authors also document that Indigenous identification has increased over birth cohorts. The results contribute to the literature on the complexity of measuring Indigeneity in the Latin American context.

Keywords

Latin America census self-identification demography response change

Indigeneity remains a major axis of stratification in contemporary Latin America (Amparo Cruz-Saco and Toor, 2018). Although “Indigenous” as a social category originated from a tributary status during the Spanish Conquista, by the independence period, Indigeneity was increasingly viewed as a biological and cultural status meant to be erased through acculturation. Following independence, Latin American national myths often celebrated the idea that racial mixture—mestizaje—dissolved racial distinctions, creating a unique national people—mestizos—who stood in sharp contrast to North America, which remained sharply divided along racial lines (Radcliffe and Westwood 1996). Mestizaje left little room for other racial identities, relegating Indigenous people—too numerous and symbolically significant to ignore—to represent a glorious but barbaric past (Hooker 2005, 2017; Marx 1997; Telles 2014). More recently, Indigenous has become category of legally protected but socioeconomically discriminated citizens (Loveman 2014; Martínez 2008; Yashar 2005).

Although Indigeneity has historically been an important social category, there has never been a clear consensus on who should be categorized as Indigenous. Early Latin American censuses counted Indigenous populations through questions on cultural practices and type of dwelling (Loveman 2014). In this way, Indigenous people were only those who the state could not recatalog as campesinos (peasants). The notion that Indigeneity is a practiced identity rather than an inherent racial category persists today, with social movements, national governments, and multilateral agencies often defining Indigenous people as members of distinct cultural communities engaged in nondominant linguistic practices (Hooker 2009; Wade 2010). Since the late 1980s, Indigeneity has also been increasingly measured through self-identity, based on the idea that individuals have the right to determine their racioethnic ascription (Hale 2006; Loveman 2014; Telles 2014). Loveman (2014) has documented the trend in Latin American censuses to shift from predominantly measuring Indigenous people by language spoken to self-identification. However, both measures are still prevalent: about 58 percent of countries that count their Indigenous populations include questions on mother tongue, while 66 percent use self-identification (World Bank 2015).

We still know relatively little about when and why people identify, or do not identify, as Indigenous in Latin America, particularly in cases where other markers of Indigeneity, such as language, are present. Drawing on this background, we develop several hypotheses about which social factors are correlated with Indigenous identification in contemporary Latin America, conditional on linguistic background. In our empirical analyses, we use representative Latinobarometer data from 16 countries in North, Central and South America collected between 2007 and 2020 to explore the relationship between those who report an Indigenous mother tongue (which we refer to as speakers) and those who self-identify as Indigenous (which we refer to as identifiers). We classify four distinct types of respondents: (1) identifiers who are also speakers, (2) identifiers who are not speakers, (3) nonidentifiers who are speakers, and (4) nonidentifiers who are not speakers. In what follows, first, provide a descriptive overview of how these different categories of identification have changed over time. Then, to test our hypotheses, we conduct multivariate analyses of the social factors associated with Indigenous identification, conditional on linguistic background. Our results contribute to literature on the complexity of measuring Indigeneity in the Latin American context.

When and Why Do People Identify as Indigenous in Latin America?

A lack of consensus on “who is Indigenous” (Canessa 2007)—or, more accurately, what constitutes a valid marker of Indigeneity—means there is no standardized method for counting Indigenous populations. Table 1 provides an illustrative overview of how both language and self-identification are commonly used to identify Indigenous populations throughout Latin America in censuses and household surveys. Across the region, the size of the Indigenous population varies depending on the method of measurement (Telles and Torche 2019). Evidence from countries that use both language- and self-identification-based questions in their censuses shows a widening numerical gap between speakers and self-identifiers (Fontana 2023). For instance, national surveys in Mexico from 2010 onward reveal a significant increase in the proportion of individuals identifying as Indigenous despite not speaking an Indigenous language. Similarly, in Bolivia, which accounts for a large proportion of the region’s Indigenous people, however defined, 20 percent of people self-identified as Indigenous without speaking an Indigenous language (Canessa 2007).

Table 1.

Variables Available for Identifying Indigenous Peoples in Censuses and Household Surveys.

	Self-Identification	Language
Household surveys	Bolivia, Brazil, Chile, Ecuador, Guatemala, Mexico, Nicaragua, Peru, and Uruguay	Bolivia, Chile, Ecuador, Guatemala, Mexico, Nicaragua, Paraguay, and Peru
Censuses	Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, and Venezuela	Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Mexico, Nicaragua, Paraguay, Peru, and Venezuela

Source: World Bank (2015).

The question of why Indigenous language speakers might not identify as Indigenous, and conversely why nonspeakers might, is closely tied to the social context in which these categories exist. Social identity theory suggests that individuals’ self-identifications are shaped by the penalties and rewards associated with different identity categories. If an identity is perceived as favorable, culturally or materially, individuals will seek strategies to affiliate with it (and vice versa) (Jiménez 2010; Leonardelli, Pickett, and Brewer 2010; Wimmer 2013). The UN’s Economic Commission for Latin America and the Caribbean reflects this sentiment in its discussion of using linguistic versus self-identificatory instruments:

In an environment of structural discrimination, Indigenous people may not declare themselves as such, especially in urban areas. In situations of ethnic revival, non-Indigenous persons may claim to be Indigenous by affinity or to gain access to specific policy benefits. (ECLAC 2006:157)

In the United States, a common finding in the racial classification literature is that higher economic status increases the likelihood of identifying as white rather than nonwhite (Daniel 2002; Golash-Boza and Darity 2008; Speight, Vera, and Derrickson 1996; Yancey 2003). The phenomenon of privilege “whitening” has also been observed in Latin America. In Brazil, among mixed-race (White-Black) students, higher family socioeconomic status is associated with lighter racial self-classification (Francis and Tannuri-Pianto 2013). In Latin America, privilege may also “de-indigenize” given that cultural definitions of Indigeneity have framed Indigenous populations as backward, either in need of civilizing or incompatible with modernity (Torres-Saillant 2010:297). Thus, opposition to Indigenous identity is not uncommon when Indigeneity evokes negative associations linked to a colonial past and a derogatory conception of contemporary Indigenous peoples (Fontana 2023). Telles (2014) noted that individuals who might be perceived as Indigenous or Black on the basis of skin color tend to self-identify as mestizo, suggesting a push against nondominant identities.

However, there have also been cases that do not conform to the “privilege whitens” thesis. In the United States, Morning (2001) found that higher socioeconomic status among South Asian immigrants correlates with a greater likelihood of self-identifying as South Asian. Additionally, qualitative studies suggest that middle- and upper-class minority groups may view their nonwhite status as a form of cultural capital (Lacy 2004; Tatum and Browne 2019; Zhou and Bankston 1994). In Latin America, Telles and Paschel (2014) found that the relationship between higher socioeconomic status and self-identification varies by context, with “whitening” occurring in Brazil, “mestizoization” in Colombia, “darkening” in the Dominican Republic, and no effect in Panama.

Indigenous self-identification is increasing over time across the region, despite the decline of most Indigenous languages (World Bank 2015:24). Scholars propose several explanations for this seemingly contradiction. Some argue changes in measurement tools account for the increase. Flores, Vignau Loría, and Martínez Casas (2023) argued that in Mexico a net gain of more than 11 million new Indigenous people can be attributed to changes in the phrasing of measurement tools such as avoiding essentialist language and treating Indigeneity as an individual (instead of a group-level) characteristic.

Others suggest that the symbolic value of Indigeneity has been on the rise (Hooker 2005; Van Cott 2005). In Bolivia, Venezuela, and Colombia, pro-Indigenous political discourse has led to an increase in Indigenous self-identification (Seawright and Barrenchea 2021). The regional tendency for people without previously regarded Indigenous markers, such as language, to identify as Indigenous has been framed as a “revitalization of the ethnic” or an “Indigenous awakening” resulting from cultural pride and safety in numbers enabled by increased Indigenous rights mobilization (Brysk 2000:6; Fontana 2023; Vázquez Sandrin and Quezada 2015).¹ Finally, Indigenous group membership allows access to targeted aid, associating material benefits with identification (Andolina, Laurie, and Radcliffe 2009; Brysk 2000).

Religious affiliation may also play a role in patterns of self-identification. Religion can play a “conservative role” in maintaining ethnic identity, or an “adaptive role,” in an acculturation process and a loss of ethnic identity (Mullins 1987). In Latin America, Catholicism and Protestantism (particularly evangelicalism) are the predominant faiths; Catholicism claims the majority of adherents, while Protestantism has made rapid inroads in the past few decades (Pew Research Center 2014). Research by Calvillo and Bailey (2015) has found that Catholicism is associated with the retention of Latino languages and identity among Latin American migrants to the United States, while Protestantism has the opposite effect. Along these lines, Kurien (2012) found that parishioners of U.S. Protestant “ethnic churches” undergo a process of “shedding ethnic languages” and other markers of ethnic background and adopting the practices of white counterparts. Differences in effects of Catholicism and Protestantism might reflect the latter’s lack of incorporation of local cultures or engagement with historical practices, leading to the shedding of minority ethnic practices (Min 2010). The Catholic Church in Latin America has historically integrated Indigenous practices, existing alongside pre-Columbian and African symbols and meanings (Parker G. 2014). Historically, protestant missionaries have worked to replace traditional forms of community in Latin America (Brysk 2000), though in some recent cases large Pentecostal movements have been shaped around local Indigenous knowledge imbued with Protestant theology and sensibility (Garrard and Doran 2020).

Research Hypotheses

Drawing on the literature discussed above, our multivariate analyses explore the following hypotheses about the factors that are correlated with Indigenous identification, conditional on linguistic background.

Hypothesis 1: Privilege Is Negatively Associated with Indigenous Identification

In correspondence with the perspective that “privilege whitens,” we expect markers of class privilege to be negatively associated with Indigenous identification among speakers and nonspeakers alike. This could be because privileged educational and occupational trajectories lead respondents to be less likely to identify as Indigenous, or because those who do not identify as Indigenous are more likely to be pursue privileged educational and occupational trajectories (or some combination of the two). We explore three markers of class privilege: (1) parents’ educational background, (2) respondents’ educational background, and (3) respondents’ occupation.

Hypothesis 2: Socioeconomic Mobility Is Negatively Associated with Indigenous Identification

As another extension of the “privilege whitens” hypothesis, we also expect that upward socioeconomic mobility will be negatively associated with Indigenous identification among speakers and nonspeakers alike. This could be because upward social mobility leads respondents to be less likely to identify as Indigenous, or because those who do not identify as such are more likely to have upward social mobility (or both).

Hypothesis 3: Evangelicalism Is Negatively Associated with Indigenous Identification

Drawing on literature suggesting that Evangelicalism is correlated with acculturation to dominant identities and loss of ethnic identity, we expect Evangelical speakers and nonspeakers to be less likely to identify as Indigenous compared with Catholics. This could reflect that Evangelicalism leads respondents to be less likely to identify as Indigenous, or that those who do not identify as Indigenous are more likely to join an Evangelical Church (or both).

Hypothesis 4: Indigenismo

In congruence with the idea that there has been a “revitalization of the ethnic” or “Indigenous awakening” we expect increasing Indigenous identification over subsequent birth cohorts among both speakers and nonspeakers.

Hypothesis 5: Race as Master Category

Building on mainstream U.S. sociology on race, we hypothesize that although speakers and nonspeakers differ with respect to an important marker of Indigeneity (e.g., linguistic ability) a similar set of social factors will be correlated with identification among both groups. This follows from Omi and Winant’s (2014) argument of race as a master category, that is, the theory that race is a form of human division that preconfigures and structures every other social domain (see also Bonilla-Silva 1997; Feagin 2013). Assuming the primacy of race as a sociocognitive frame, it follows that race would interact with other socioeconomic factors in a similar way for speakers and nonspeakers.

Data and Methods

Sample

For this study, we rely on the last eight releases of Latinobarometer data, covering almost every year between 2007 and 2020.² The Latinobarometer is a yearly public opinion survey, which annually conducts about 20,000 interviews in 18 Latin American countries. These are conducted with local partners and are representative at a national level. We focus on this time frame because before 2007, the survey did not include the race questions at the core of our analysis.

Our final sample pools respondents from 16 of the 18 sampled countries. Our pooled approach aims to establish key region-level trends in Hispanophone North, Central America and South America (we further discuss the implications of this pooled approach in the discussion). We exclude respondents from Brazil (the only country in the sample that was a Portuguese colony) and the Dominican Republic (the only island nation in the sample). Our sample is limited to people born after 1925 because the pre-1925 cohort was too small to capture cohort trends. We also limit the sample to adults older than 21 years because we are interested in exploring people’s identification after schooling processes have finished and individuals have entered the labor market. After completing a listwise deletion, this leaves us with a final sample of 117,605 respondents in 16 countries.

Measures

Indigenous self-identification is coded according to the response to the question “Ethnic group (wait for answers)” or “What ethnicity or race do you identify best with?” (the Latinobarometer switched from the latter to the former question in 2011).³ We code the answers as Indigenous or non-Indigenous and exclude nonresponses. Although interviewer classification would have been a useful consideration, this was not available in the survey.

Mother tongue is represented by an indicator to distinguish whether the respondent reports an Indigenous language as their mother tongue.

Education is measured as completed years of formal education. We record individuals’ education in years instead of using grade cutoffs because of national differences in types of degrees and time to completion. We also include an indicator for being in school at the time of the study.

Employment is captured by a categorical variable that indicates if a respondent works in a “white collar” job (an office or managerial job), in agriculture, or in the informal sector. We also include a categorical measure for if the respondent works for private, public industry, or is self-employed.

Religion is measured using indicator variables for Evangelical, Catholic, or other religious backgrounds.

Socioeconomic background is captured using parent’s education, also in years.

Social mobility is measured as changes in education between parents and children. Mobility is calculated as the difference in educational attainment between the respondent and the highest attainment of either parent. It is coded as “downward” in cases of a greater than a 1-year difference (noninclusive), “no mobility” if there is a difference between −1 and 2 years (inclusive), “upward” in cases of a difference greater than 2 but lower than 7 years (noninclusive) and “large upward mobility” in cases of a difference of 7 years or more.

Country fixed effects are included to account for time-invariant country-level differences to capture Latin America–wide trends.

Year fixed effects control for secular time trends over the period of study.

Methods

First, we provide descriptive information about patterns of self-identification and mother tongue at the regional level. We specify four distinct classes of respondents: (1) identifiers who are also speakers, (2) identifiers who are not speakers, (3) nonidentifiers who are speakers, and (4) nonidentifiers who are not speakers. We provide a descriptive overview of the prevalence of each of these categories and how this has changed over time.

Next, we explore the social factors that correlate with Indigenous identification given the presence (or absence) of linguistic markers of Indigeneity. To this end, we conduct a logistic regression analysis of the social factors that correlate with identification, conditional on linguistic background (i.e., being a speaker or a nonspeaker). We condition on linguistic background to ensure that we are assessing patterns of identification within socially coherent categories of respondents. This allows us to understand if the factors that correlate with Indigenous identification are similar or different for speakers and nonspeakers alike. We orient the discussion of our multivariate results upon a series of research hypotheses, as discussed above. It is important to emphasize that our analyses represent associations only and not causal effects. This is important for our interpretation of results, as we cannot disentangle the causal ordering of explanatory and outcome variables in our analyses. Indeed, we expect that the explanatory variables in our analyses both predict and are predicted by Indigenous identification.

Because we aim to establish region-level trends, all multivariate models include country fixed effects to account for unobserved confounders at the country level that might affect identification (e.g., policies targeted to Indigenous populations, political movements that might affect Indigenous identification, etc.) and year fixed effects to account for secular time trends (e.g., declines in the number of speakers). All analyses use sampling weights provided by the Latinobarometer as appropriate.

Findings

Descriptive Analysis of Indigenous Identification by Linguistic Background

Figure 1 shows that speakers (in blue) represent a higher proportion of identifiers compared with nonidentifiers. Among identifiers, 27.9 percent of respondents are also speakers whereas among nonidentifiers only 4.5 percent of respondents are also speakers. Only about 45 percent of speakers identify as Indigenous, with a surprising proportion of those who report Indigenous mother tongues not identifying as such. In fact, a preponderance of speakers, approximately 40 percent, identify as mestizo, 12 percent identify as White, and 3 percent identify as Asian, Black, Mulatto, or other. This makeup suggests that language alone cannot be used as a definitive indicator of racial identity, clarifying the need to treat language and racial identity as separate but related concepts.

Figure 1.

Self-identification and mother tongue.

Figure 2 breaks down the four categories of self-identification by mother tongue over different birth years and reveals several trends of interest. First, nonidentifier nonspeakers (black line) consistently constitute about 85 percent of the sample and exhibit very little variation over time. The other three categories (identifier nonspeaker, nonidentifier speaker, and identifier speaker) make up a much smaller proportion of the sample (about 15 percent cumulatively) and exhibit considerably more fluctuation by birth year. Whether this fluctuation can be accounted for by demographic phenomena (e.g., differences in mortality rates) or sociocultural forces (changes in self-ascription among older populations) remains open for further research.

Figure 2.

Self-identification and mother tongue by birth year.

The identifier speaker category (in red) is the least common (hovering at about 3 percent to 4 percent of respondents). The identifier speaker category also exhibits the greatest fluctuation in values, particularly for older birth cohorts born before 1940. The dramatic fluctuation in this category for pre-1940 cohorts (from about 1 percent to about 9 percent) may reflect the complicated dynamics of Indigenous identification among those who came of age during a time when Indigenous identification was particularly contested and marginalized, however estimates are less certain because of the smaller number of respondents.

The identifier nonspeaker category (in blue) is the second most common respondent category (ranging from about 5 percent to 8 percent of respondents), followed by the nonidentifier speaker category (in grey) (ranging from about 3 percent to 5 percent of respondents). From those born in the 1970s onward, there appears to be a small downward trend in both nonidentifier speakers and identifier speakers. These trends correspond with declines in Indigenous language speakers over time (World Bank 2015:24). Still, this decline is quite small and exhibits variation in more recent cohorts. The numbers of identifier nonspeakers and nonidentifiers speakers and their fluctuation over time help put to bed any assumptions about an equivalence between language and identity.

Logistic Regression Analysis of the Social Factors That Are Correlated with Indigenous Identification

Table 2 explores the social factors that are correlated with Indigenous identification, conditional on being an Indigenous language speaker, and Table 3 explores the factors that are correlated with Indigenous identification, conditional on being a nonspeaker. In this section we discuss these results in reference to the research hypotheses developed above.

Table 2.

Logistic Regression Analysis of the Factors That Predict Indigenous Self-Identification Conditional on Being a Speaker.

	Indigenous (Speaker Sample)
Age	.994** (.002)
Male	1.008 (.065)
Education	.976 (.019)
Employment
White collar
Agricultural	1.705*** (.182)
Informal	1.038 (.098)
No work	1.008 (.091)
Other	.790 (.132)
Student	.838 (.229)
Parent education	.935*** (.019)
Socioeconomic mobility
Downward	.672* (.105)
No mobility
Upward mobility	.802* (.085)
Large upward mobility	.713* (.123)
Faith
Catholic
Evangelical	1.259** (.101)
Other or none	1.376** (.141)
Country FE	Yes
Year FE	Yes
Constant	.486 (.205)
Log likelihood
χ²
n	8,645.000

Note: Results are presented as odds ratios with standard errors in parentheses. Weighted using sampling weights provided by the Latinobarometer. FE = fixed effects.

p < .05. **p < .01. ***p < .001.

Table 3.

Logistic Regression Analysis of the Factors That Predict Indigenous Self-Identification Conditional on Being a Nonspeaker.

	Indigenous (Nonspeaker Sample)
Age	.993*** (.001)
Male	1.089*** (.027)
Education	.954*** (.007)
Employment
White collar
Agricultural	1.579*** (.078)
Informal	1.135** (.045)
No work	1.031 (.036)
Other	.950 (.049)
Student	.891 (.084)
Parent education	.939*** (.007)
Socioeconomic mobility
Downward	.945 (.052)
No mobility
Upward mobility	.973 (.040)
Large upward mobility	.958 (.061)
Faith
Catholic
Evangelic	1.087** (.033)
Other or none	1.102** (.036)
Country FE	Yes
Year FE	Yes
Constant	.059*** (.007)
Log likelihood	−27,905.283
χ²	7,518.074
n	108,960.000

Note: Results are presented as odds ratios with standard errors in parentheses. Weighted using sampling weights provided by the Latinobarometer. FE = fixed effects.

p < .01. ***p < .001.

Hypothesis 1: Privilege Is Negatively Associated with Indigenous Identification

Results from multivariate regressions lend support to the hypothesis that privilege is negatively associated with Indigenous identification among speakers and nonspeakers alike. Tables 2 and 3 show that parents’ educational attainment is negatively correlated with the odds of identification among both speakers (odds ratio [OR] = 0.994, p < .05) and nonspeakers (OR = 0.993, p < .01). Respondents’ education is also negatively correlated with the odds of identification in speakers (OR = 0.976) and nonspeakers (OR = 0.954, p < .001), though the education coefficient in the model with speakers is not statistically significant at p < .05.

The magnitudes of these coefficients are potentially sizable if we take into the cumulative nature of education (i.e., the difference in years between completing middle and high school). Figure 3 shows that among speakers, those without any education have a 0.45 predicted probability of Indigenous identification compared with 0.32 among those with a university degree. Among nonspeakers, the predicted probability of identification is also significantly higher among those with no education (0.12) compared with those with a university degree (0.07), though the overall probability of identification among nonspeakers at all levels of education is dramatically lower compared with that of speakers.

Figure 3.

Predictive probability of identifying as Indigenous by years of education for speakers versus nonspeakers.

In further support of this hypothesis, we find that working in agriculture, as opposed to a white-collar job, is associated with 71 percent and 58 percent higher odds of identifying as Indigenous among speakers and nonspeakers respectively (p < .001). Given that Indigenous populations are heavily represented in agricultural spheres, this suggests that identification may be higher when respondents are heavily involved in lower-status occupations with large Indigenous representation.

Hypothesis 2: Social Mobility Is Negatively Associated with Indigenous Identification

Multivariate results provide partial support for the idea that social mobility is negatively associated with Indigenous identification. Among speakers, upward educational mobility is associated with 20 percent lower odds of Indigenous identification (compared with no mobility), and large upward mobility is associated with 29 percent lower odds of Indigenous identification (compared with no mobility) (p < .05). However, our initial set of hypotheses did not anticipate how downward social mobility may also be important for identification. Downward socioeconomic mobility is also associated with 33 percent lower odds of Indigenous identification (compared with no mobility) among speakers (p < 0.05). In contrast, among nonspeakers, the relationship between social mobility and identification is null across all categories of mobility.

Hypothesis 3: Evangelicalism Is Associated with Decreases in Indigenous Identification

In contrast to the hypothesis that Evangelicalism would be negatively associated with Indigenous identification, we find that Evangelicals are associated with 26 percent higher odds of Indigenous identification than Catholics among speakers (p < .01). Results are similar for nonspeakers, whereby Evangelicals are associated with 9 percent higher odds of Indigenous identification than Catholics (p < .05). Respondents who report “other or no religious identification” are also associated with higher odds of Indigenous identification compared with Catholics among both speakers and nonspeakers.

Hypothesis 4: Indigenismo

In support of the notion that Indigenous identification has grown over time, age is negatively correlated with Indigenous identification among speakers (OR = 0.99, p < .01) and nonspeakers (OR = 0.99, p < .001). As was the case with education, the small magnitude of the age coefficient may belie larger cumulative changes. To better understand these dynamics, Figure 4 shows the predicted probabilities of Indigenous identification over subsequent birth cohorts for speakers (in red) and nonspeakers (in blue). Figure 4 shows that there have been substantial changes over time among both speakers and nonspeakers (although overall identification remains much higher among speakers). For example, for those born in the mid-1920s, the predicted probability of identification was about 0.31 among speakers and .07 among nonspeakers. In contrast, among those born in the late 1990s the predicted probability of identification was about 0.40 among speakers and 0.10 among nonspeakers.

Figure 4.

Predictive probability of Identifying as Indigenous by birth year for speakers versus nonspeakers.

Hypothesis 5: Race as Master Category

By and large, the findings in Tables 2 and 3 support the race as a “master category” hypothesis, as the factors that correlate with Indigenous identification among speakers also correlate with Indigenous identification among nonspeakers. However, there are a few exceptions. For example, social mobility is significantly associated with Indigenous identification among speakers, but not nonspeakers. Likewise, males are associated with a 9 percent higher odds of Indigenous association (OR = 1.089, p < 0.001) among nonspeakers, though the association between gender and identification is null among speakers (OR = 1.008). Still, the comparability of findings across samples of speakers and nonspeakers for many variables in the analysis provides overall support to the race as master category hypothesis.

Discussion

Using representative Latinobarometer data collected between 2007 and 2020, our descriptive analyses showed that though the majority of respondents in Latin America did not speak an Indigenous language or identify as Indigenous, nonnegligible proportions of respondents were identifier speakers, identifier nonspeakers, and nonidentifier and speakers. The latter two groups in particular are important because they challenge a binary distinction between Indigenous and non-Indigenous group membership at the regional level. In this respect, our findings complement Telles and Torche’s (2019) findings that multiple varieties of Indigeneity exist in Mexico and Peru.

In multivariate analyses, we tested several hypotheses about the social factors that correlate with identification, conditional on linguistic background. Our findings lent support to the hypothesis that privilege is negatively associated with Indigenous identification. Our findings on social mobility and identification were more complicated: we found both upward and downward social mobility were associated with lower odds of Indigenous identification among speakers, but there was no significant association between social mobility and identification among nonspeakers. In contrast, our hypothesis that Evangelicalism would be negatively associated with Indigenous identification was not supported; in fact, we found that that Catholicism, compared with Evangelicalism, is negatively correlated with Indigenous identification among respondents.

Our finding that Indigenous identification has increased over birth cohorts (net of other observed characteristics) parallels other scholarship that suggests that Indigenous identification has increased over time in the region (Seawright and Barrenchea 2021; World Bank 2015). There remains a debate, however, whether increasing patterns of identification reflect changes in measurement criterion, as opposed to increased recognition of Indigenous heritage related to renewed attention to Indigeneity in national imaginaries. Importantly, the Indigenous identification question we use in our survey has remained stable over time. The stability of the Indigenous identification question employed in our analyses suggests that wording and/or measurement changes are not driving the changing patterns of identification that we observe across birth cohorts. Nonetheless, we interpret these results with some caution because of the difficulties of distinguishing cohort effects from age effects.

With a few exceptions (such as the social mobility results), we found that find that most of the correlates of Indigenous identification were similar among both speakers and nonspeakers. We argued that this similarity in findings across different linguistic groups supports the “race as a master category” argument. Scholars tend to conceive of Indigeneity as so closely tied to cultural elements to make it contextually contingent and “always in flux” (Harris, Carlson, and Poata-Smith 2019:5). Nevertheless, an implication of our findings might be that Indigenous self-identification is less contingent than previously thought. However, it is important to note that our study relied on a single, albeit significant, cultural marker of Indigeneity (i.e., language). Future research should incorporate a wider array of empirical tools, such as interviewer classification, for a more comprehensive analysis.

Despite these insights, it is important to recognize certain data limitations. First, our cross-sectional data prevented us from exploring how identification changes over time within the same individuals. Second, our data counted only with self-identification and did not include ascribed identity, which is a consequential factor. After all, racial disadvantages derive largely from ascribed traits, not from self-identification (Flores and Telles 2012; Del Popolo 2008; Telles and Lim 1998). Furthermore, our results represent associations only and not the causal effect of social variables on identification. It could be that those who select into certain social categories (e.g., Evangelicalism, higher education) are different from those who do not, in ways that contribute to differences in identification. Finally, our region-level approach had both pluses and minuses that are important to acknowledge. By pooling across countries, we provided a broad-stroke overview of regional trends that might be obscured when focusing on individual countries. However, in doing so, we may, miss out on some country-level heterogeneity. Exploratory analysis (available upon request) suggested that while there might be differences in size and significance, the direction and magnitude of the relationships here discussed hold at the country level. Ultimately, we see our findings as important complements to country-specific work that can be used for comparative purposes.

The increasing gap between speakers and identifiers coupled with the region-wide growth in policies targeted to Indigenous people make clear that scholars and policymakers need a deeper understanding of the complexity of measuring Indigeneity in the Latin American context. Although our results are descriptive, they underscore a critical issue: different measures strategies capture different populations which differ in important ways. This has important implications for research and policy. First, consistent measurement is essential for enabling meaningful regional and temporal comparisons. Second, the choice of measurement strategy can significantly influence outcomes, a factor that must be carefully considered when designing targeted policies. Our results thus contribute to a literature that problematizes the use of language and self-identification as interchangeable markers of Indigeneity in research and policy. As Del Popolo (2008) argued, cultural markers and self-identity are different concepts that reflect different experiences of Indigeneity, operating in conjunction but as somewhat separate processes. We thus agree with the conclusions drawn by the Mexican National Institute of Statistic and Geography (INEGI 2015:30–31), which recommends approaching subject of Indigeneity both though language and a self-identification to better understand how these to processes interrelate.

Footnotes

ORCID iDs

Andrés Rodríguez-Cáceres

Julia Behrman

1

Some suggest that this gap between speakers and identifiers could also be the consequence of loss of language.

2

Except for 2012, 2014, and 2019, which were unavailable. We also drop respondents in 2013 and 2010 who were not queried about mother tongue.

3

We use self-reported identity rather than interviewer ascribed identity and it is important to distinguish between these concepts. We discuss these in the limitations section.

Author Biographies

Andrés Rodríguez-Cáceres is a PhD candidate in the Department of Sociology at Northwestern University and a Mellon Comparative Historical Cluster Fellow with particular interest in institutional change. His current works span Indigenous rights in Latin America and working hours in rich democracies.

Julia Behrman is an assistant professor in the Department of Sociology at Northwestern University and is a faculty fellow at the Institute for Policy Research at Northwestern University. Her research and teaching interests are in gender and family, social demography, and immigration.

References

Amparo Cruz-Saco

Maria

Toor

Joanne

. 2018. “Indigenous Communities and Social Inclusion in Latin America.” Presented at the United Nations Expert Group Meeting on Families and Inclusive Societies, New York, May 15–16, 2018.

Andolina

Robert

Laurie

Nina

Radcliffe

Sarah A.

2009. Indigenous Development in the Andres. Durham, NC: Duke University Press.

Bonilla-Silva

Eduardo

. 1997. “Rethinking Racism: Toward a Structural Interpretation.” American Sociological Review 62(3):465–80.

Brysk

Alison

. 2000. From Tribal Village to Global Village: Indian Rights and International Relations in Latin America. Cambridge, UK: Cambridge University Press.

Calvillo

Jonathan E.

Bailey

Stanley R.

2015. “Latino Religious Affiliation and Ethnic Identity.” Journal for the Scientific Study of Religion 54(1):57–78.

Canessa

Andrew

. 2007. “Who Is Indigenous? Self-Identification and Claims to Justice in Contemporary Bolivia.” Urban Anthropology 36(3):195–237.

Daniel

G. Reginald

. 2002. More Than Black? Multiracial Identity and the New Racial Order. Philadelphia: Temple University Press.

Del Popolo

Fabiana

. 2008. “Los Pueblos Indígenas y Afrodescendientes en las Fuentes de Datos: Experiencias en América Latina” [“Indigenous and Afro-Descendant Peoples in Data Sources: Experiences in Latin America”]. New York: UNESCO.

ECLAC (United Nations Economic Commission for Latin America and the Caribbean). 2006. “Social Panorama of Latin America.” Santiago, Chile: United Nations.

10.

Feagin

Joe R

. 2013. The White Racial Frame: Centuries of Racial Framing and Counter-Framing. 2nd ed. New York: Routledge.

11.

Fontana

Lorenza B

. 2023. Recognition Politics: Indigenous Rights and Ethnic Conflict in the Andes. Cambridge: Cambridge University Press.

12.

Flores

René

Telles

Edward

. 2012. “Social Stratification in Mexico: Disentangling Color, Ethnicity, and Class.” American Sociological Review 77(3):486–94.

13.

Flores

René D.

Vignau Loría

María

Martínez Casas

Regina

. 2023. “Transitory versus Durable Boundary Crossing: What Explains the Indigenous Population Boom in Mexico?” American Journal of Sociology 129(1):123–61.

14.

Francis

Andrew M.

Tannuri-Pianto

Maria

. 2013. “Endogenous Race in Brazil: Affirmative Action and the Construction of Racial Identity among Young Adults.” Economic Development and Cultural Change 61(4):731–53.

15.

Garrard

Virginia

Doran

Justin M.

2020. “Pentecostalism and Neo-Pentecostalism in Latin America: Two Case Studies.” in The Oxford Handbook of Latin American Christianity, edited by Orique

D. T.

Fitzpatrick-Behrens

Garrard

Oxford University Press.

16.

Golash-Boza

Tanya

Darity

William

. 2008. “Latino Racial Choices: The Effects of Skin Colour and Discrimination on Latinos’ and Latinas’ Racial Self-Identifications.” Ethnic and Racial Studies 31(5):899–934.

17.

Hale

Charles R.

2006. Mas Que un Indio: Racial Ambivalence and Neoliberal Multiculturalism in Guatemala. Santa Fe, NM: School of American Research Press.

18.

Harris

Calrson

Poata-Smith

E. S.

2019. “Indigenous Identities and the Politics of Authenticity.” Pp. 1–9 in The Politics of Identity: Emerging Indigeneity, edited by Harris

Nakata

Carlson

Sydney, Australia: University of Technology Sydney E-Press.

19.

Hooker

Juliet

. 2005. “Indigenous Inclusion/Black Exclusion: Race, Ethnicity and Multicultural Citizenship in Latin America.” Journal of Latin American Studies 37(2):285–310.

20.

Hooker

Juliet

. 2009. Race and the Politics of Solidarity. Oxford, UK: Oxford University Press.

21.

Hooker

Juliet

. 2017. Theorizing Race in the Americas: Douglass, Sarmiento, Du Bois, and Vasconcelos. New York: Oxford University Press.

22.

INEGI (Instituto Nacional de Estadística y Geografía). 2015. “Encuesta Intercensal 2015.” Encuestas xii–234.

23.

Jiménez

Tomás R.

2010. “Affiliative Ethnic Identity: A More Elastic Link between Ethnic Ancestry and Culture.” Ethnic and Racial Studies 33(10):1756–75.

24.

Kurien

Prema

. 2012. “Decoupling Religion and Ethnicity: Second-Generation Indian American Christians.” Qualitative Sociology 35(4):447–68.

25.

Lacy

Karyn R.

2004. “Black Spaces, Black Places: Strategic Assimilation and Identity Construction in Middle-Class Suburbia.” Ethnic and Racial Studies 27(6):908–30.

26.

Leonardelli

Geoffrey J.

Pickett

Cynthia L.

Brewer

Marilynn B.

2010. “Optimal Distinctiveness Theory. A Framework for Social Identity, Social Cognition, and Intergroup Relations.” Advances in Experimental Social Psychology 43:63–113.

27.

Loveman

Mara

. 2014. National Colors: Racial Classification and the State in Latin America. Oxford, UK: Oxford University Press.

28.

Martínez

María Elena

. 2008. Genealogical Fictions: Limpieza de Sangre, Religion, and Gender in Colonial Mexico. Stanford, CA: Stanford University Press.

29.

Marx

Anthony W.

1997. Making Race and Nation: A Comparison of South Africa, the United States, and Brazil. Cambridge, UK: Cambridge University Press.

30.

Min

Pyong Gap

. 2010. Preserving Ethnicity through Religion in America. New York: New York University Press.

31.

Morning

Ann

. 2001. “The Racial Self-Identification of South Asians in the United States.” Journal of Ethnic and Migration Studies 27(1):61–79.

32.

Mullins

Mark

. 1987. “The Life-Cycle of Ethnic Churches in Sociological Perspective.” Japanese Journal of Religious Studies 14(4):321–34.

33.

Omi

Michael

Winant

Howard

. 2014. “The Theory of Racial Formation.” Pp. 105–36 in Racial Formation in the United States, 3rd ed., edited by M. Omi Winant

New York: Routledge.

34.

Parker G.

Cristian

. 2014. “Pluralismo Religioso y Cambio de Paradigma Identitario en el Campo Politico Latinoamericano Actual” [“Religious Pluralism and Change of Identitarian Paradigm in the Current Latin American Political Field”]. Pp. 39–71 in Simbolos, Rituales, Religiosos e Identidades Nacionales [Symbols, Rituals, Religious and National Identities], edited by A.R. Ameigeiras. Buenos Aires, Argentina: Clacso.

35.

Pew Research Center. 2014. “Religion in Latin America: Widespread Change in a Historically Catholic Region.” Pew Research Center. Retrieved February 22, 2022. https://www.pewforum.org/2014/11/13/religion-in-latin-america/.

36.

Radcliffe

Sarah A.

Westwood

Sallie

. 1996. Remaking the Nation: Place, Identity and Politics in Latin America. London: Routledge.

37.

Seawright

Jason

Barrenchea

Rodrigo

. 2021. “Shaping the People: Populism and the Politics of Identity Formation in South America.” Pp. 491–518 in The Inclusionary Turn in Latin American Democracies, edited by Kapiszewski

Levitsky

Yashar

D. J.

Cambridge, UK: Cambridge University Press.

38.

Speight

Suzette L.

Vera

Elizabeth M.

Derrickson

Kimberly B.

1996. “Racial Self-Designation, Racial Identity, and Self-Esteem Revisited.” Journal of Black Psychology 22(1):37–52.

39.

Tatum

Katharine

Browne

Irene

. 2019. “The Best of Both Worlds: One-up Assimilation Strategies among Middle-Class Immigrants.” Poetics 75:101317.

40.

Telles

Edward Eric

. 2014. Pigmentocracies: Ethnicity, Race, and Color in Latin America. Chapel Hill: University of North Carolina Press.

41.

Telles

Edward E.

Lim

Nelson

. 1998. “Does It Matter Who Answers the Race Question? Racial Classification and Income Inequality in Brazil.” Demography 35(4):465–74.

42.

Telles

Edward

Paschel

Tianna

. 2014. “Who Is Black, White, or Mixed Race? How Skin Color, Status, and Nation Shape Racial Classification in Latin America.” American Journal of Sociology 120(3):864–907.

43.

Telles

Edward

Torche

Florencia

. 2019. “Varieties of Indigeneity in the Americas.” Social Forces 97(4):1543–70.

44.

Torres-Saillant

Silvio

. 2010. Introduction to Dominican Blackness. New York: CUNY Academic Works.

45.

Van Cott

Donna Lee

. 2005. “Building Inclusive Democracies: Indigenous Peoples and Ethnic Minorities in Latin America.” Democratization 12(5):820–37.

46.

Vázquez Sandrin

Germán

Quezada

María Félix

. 2015. “Los Indígenas Autoadscritos de México En El Censo 2010: ¿Revitalización Étnica o Sobreestimación Censal?” [“The Self-Ascribed Indigenous People of Mexico in the 2010 Census: Ethnic Revitalization or Census Overestimation?”]. Papeles de Poblacion 21(86):171–218.

47.

Wade

Peter

. 2010. Race and Ethnicity in Latin America. 2nd ed. London, UK: Pluto Press.

48.

Wimmer

Andreas

. 2013. Ethnic Boundary Making. Oxford, UK: Oxford University Press.

49.

World Bank. 2015. “Indigenous Latin America in the Twenty-First Century.” Washington, DC: World Bank.

50.

Yancey

George A.

2003. Who Is White? Latinos, Asians, and the New Black/Nonblack Divide. Boulder, CO: Lynne Rienner.

51.

Yashar

Deborah J.

2005. Contesting Citizenship in Latin America: The Rise of Indigenous Movements and the Postliberal Challenge. Cambridge, UK: Cambridge University Press.

52.

Zhou

Min

Bankston

Carl L.

III . 1994. “Social Capital and the Adaptation of the Second Generation: The Case of Vietnamese Youth in New Orleans.” International Migration Review 28(4):821–45.