Varieties of Ethnic Self-Identities:Children of Immigrants in Middle Adulthood
This mixed-methods longitudinal study examines ethnic self-identity change from mid-adolescence to middle adulthood among a representative sample of adult children of immigrants first surveyed in 1992 in San Diego. Findings reveal the complexities of ethnic identity. Ethnic identity labels are often used interchangeably yet usually stabilize by middle adulthood. While the importance of ethnic identity often diminishes, immigrants' children in their late thirties express distinct ethnic identity formations, ranging from strong ethnic attachments to indifference, that vary within and between nativity and national-origin groups. Ethnic identities relate to political views and behaviors, interethnic friendships, and cultural practices, but not interethnic unions. Consistent with life course theory, results show how identities develop across nearly a quarter century, influenced by sociohistorical contexts and relationships with others.
ethnic identity, second generation, immigrants, middle adulthood, life course
The study of the new second generation, framed as the most consequential legacy of contemporary immigration to the United States, is now more than twenty years old, and has generated a vibrant field of study.1 Although the incorporation trajectories of the adult children of this immigration have been the subject of vigorous debates, the complexities of ethnic-panethnic identity formations in adult life are understudied.2 The ethnic-panethnic identity outcomes of the adult children of immigrants—their meaning, importance, stability, or change—remain open questions, as does the nature of their interplay with other key aspects of incorporation processes, such as politics, family formation, and cultural practices. [End Page 26]
Ethnic self-identities can be understood as "definitions of the situation of the self" (Rumbaut 2005). For children of immigrants, they can be conceptualized to emerge from the interplay of racial and ethnic labels and categories imposed by the external society and the ancestral attachments asserted by the newcomers. They may be hypothesized to vary across social situations, across developmental stages throughout the life course, and indeed across historical contexts. Previous studies have shed light into the ethnic self-identities of the growing second generation in adolescence and early adulthood, but we know little about how ethnic identities shift as individuals age beyond early adulthood, particularly in an extraordinarily diverse population whose immigrant parents have come largely from Asia and Latin America, differing significantly in their national, cultural, and class origins from the white Europeans whose incorporation in American life had dominated conventional narratives. No studies to date have examined ethnic identity change as children of immigrants transition from their teens and mid-twenties to their late thirties, and especially as they form unions (often ethnically mixed) and have children of their own to raise and socialize. Our longitudinal data from mid-adolescence into middle adulthood allowed us to apply a life course perspective to the development of ethnic identities among the 1.5 and second generations of children of immigrants, in which we considered how individuals' constructions of their identities are embedded within particular historical and sociogeographic contexts and linked to relationships with others (Elder, Johnson, and Crosnoe 2003).
Moreover, contextual influences on the definition and presentation of the self suggest that mixed-method work on ethnic identity should be an essential part of a comprehensive analysis. Survey instruments alone can explore neither the situational contexts in which ethnic self-identities may be differentially deployed nor the subjective meanings behind ethnic labels; yet how an ethnic identity is expressed in one's presentation of self can differ depending on the audience, and public expressions of identity may not correspond with private self-conceptions (Brunsma 2006). By combining quantitative and qualitative approaches to a longitudinal study encompassing a life span of more than two decades, we aim to enhance and contextualize our understanding of ethnic identity formations throughout the life course, and their connections to inter and intra-ethnic relationships, political orientations and behaviors, and cultural practices.
ETHNIC IDENTITIES AMONG CHILDREN OF IMMIGRANTS
Because ethnic identities necessarily invoke subjective feelings of group belonging and thus reveal boundaries between insiders and outsiders, the ethnic identifications of immigrants' children provide insight into the long-term incorporation trajectories of immigrant groups. According to Milton Gordon's influential framework, identiflcational assimilation, which he defines as the "development of a sense of peoplehood based exclusively on the host society," is one of the later stages of the assimilation process—indeed, arguably, its end point (1964, 71). Conventional accounts of ethnic identity shifts among descendants of European immigrants, conceived as part of a larger, linear process of assimilation, have pointed to the "thinning" of ethnic self-identities in the United States such that ethnic identity, for this population, has become an optional, leisure-time form of symbolic ethnicity (Alba 1990; Gans 1979; Waters 1990). However, such an outcome is possible only when accompanied by the absence of prejudice and discrimination in the core society (Gordon 1964; Waters 1990). In a context of perceived threats, persecution, discrimination, and exclusion, ethnic identity may not erode, but rather rise, in the form of reactive ethnicity (Rumbaut 2008). Reactive ethnicity is consistent with a psychological process of rejection-identification, when perceptions of prejudice lead to hostility toward the dominant group and identification with the minority group (Branscombe, Schmitt, and Harvey 1999). Indeed, Portes and Rumbaut found such a "thickening" of Mexican identity among adolescents in response to the passage of Proposition 187 in 1994, which, though never implemented, would have denied health care, public education, and social services to undocumented immigrants and their children in California, [End Page 27] and required government employees to report suspected undocumented immigrants to the authorities (2001).3
However, assessing such possibilities among children of contemporary immigrants is complicated because no universally accepted definition of ethnic identity, nor shared understanding of its multiple dimensions, exists (Ashmore, Deaux, and McLaughlin-Volpe 2004; Phinney and Ong 2007). Self-labeling is one dimension of ethnic identity used widely in research on immigrants' children (Feliciano 2009; Fuligni, Witkow, and Garcia 2005). Based on responses to an open-ended question from the first wave of the Children of Immigrants Longitudinal Study, Rubén Rumbaut identified four types of self-labels among adolescents: national, hyphenated, plain American, and pan-ethnic (1994). Although previous research on such outcomes reveals key insights, self-labeling captures just one dimension of a broad conceptualization of ethnic identity as "membership in a social group together with the value and emotional significance attached to that membership" (Taijfel 1981, 225). Moreover, studies find that ethnic minorities often use multiple labels in different situations, and that panethnic identities are often overlapping rather than distinct identities (Kasinitz et al. 2008; Dowling 2014).
ETHNIC IDENTITY CHANGE ACROSS THE LIFE COURSE
Adolescence is a crucial stage of life during which identities develop (Erikson 1968); and ethnic identity is assumed to stabilize by early adulthood (Phinney 1993). However, although previous studies show considerable change in ethnic identity among children of immigrants from early adolescence through early adulthood, researchers have not yet examined how ethnic identities shift from early to middle adulthood (Smith 2014; Feliciano 2009). From a life course perspective, lives are patterned by transitions (Elder, Johnson, and Crosnoe 2003). Thus, self-labels in middle adulthood, as well as the meaning and importance of ethnic identification, should be linked to those in earlier life stages, but may also shift as individuals transition further into adulthood and take on additional roles and identities as workers, spouses, and parents.
Moreover, a life course perspective emphasizes how development occurs within particular historical and social contexts (Elder 1998). Our focus is on immigrants' children born in the late 1970s, who were coming of age in San Diego during a time of relative inclusion. Unlike today's context, in which a high proportion of children of immigrants grow up in mixed legal status families, most parents in our sample arrived in the late 1970s and 1980s, including some who may have migrated without papers but were able to legalize through the Immigration Reform and Control Act of 1986. Thus, the overwhelming majority of our respondents who were not citizens by birth became naturalized U.S. citizens. Moreover, they grew up in a period when diversity was increasingly the accepted norm. From 1980 to 2015, California's Asian population grew from 5 percent to 14 percent, and the Latino population from 19 percent to 39 percent (Bohn et al. 2018). At the national level, the U.S. Census Bureau's decision in 2000 to allow identification as more than one race reflected an emerging consensus that individuals should be able to define themselves as they choose (Csizmadia et al. 2012).
Yet the political context was not uniformly inclusive. As mentioned, Mexican-origin immigrants' children often formed reactive ethnic identities in response to Proposition 187, passed during a core developmental period for the study respondents, from early to late adolescence (Portes and Rumbaut 2001). Moreover, a "Latino threat narrative" fueled by fears about Mexican immigrants and their descendants invading, taking over, refusing to assimilate, and posing a danger to the nation has long been evident in public discourse in the United States (Chavez 2008). Thus, we might expect children of Mexican immigrants' ethnic identities in middle adulthood to be more reactive than others, perhaps rejecting an American identity in favor of a panethnic or Mexican identity (Massey and Sánchez 2010). On the other hand, [End Page 28] the respondents have diverse ethnic backgrounds, and the vast majority spent most of their lives in a multicultural California context, where anti-immigrant views do not dominate. Given the relatively accepting political and social climate for diversity, from a life course perspective, we may expect ample room for individual agency in ethnic identity development, leading to wide variations in ethnic identity formations even within ethnic groups.
ETHNIC IDENTITIES AND POLITICS, RELATIONSHIPS, AND CULTURAL PRACTICES
Ethnic identities are important because they signify boundaries between groups and thus potentially enduring divisions in society. However, as subjective attachments, ethnic identities may be only personally meaningful, unrelated to behaviors or beliefs. Few studies investigate how ethnic identities among children of immigrants are associated with attitudes or practices in other social realms, perhaps because existing studies focus on adolescents or young adults but the outcomes of most interest become most salient only in later stages of adulthood.
Political views and behaviors may be associated with ethnic identities. Although many suggest that the demographic shifts creating a more diverse electorate may lead to a more liberal and Democratic Party-affiliated population, Taeku Lee cautions against assuming that shared racial and ethnic identity labels correspond to group-based politics, arguing that the identity-to-politics link evident among African Americans may not apply to Asians or Latinos (2008). Others suggest that with identificational assimilation, differences in political views and behaviors by ethnic origins should decline (Alba, Beck, and Sahin 2017; Gordon 1964). Considering variation in ethnic identity attachments within ethnic groups may matter: immigrants' children who are less attached to their ethnic origins may differ in political affiliations, attitudes, and behaviors from those with strong ethnic attachments.
Like ethnic identity formations themselves, cross-ethnic and cross-racial friendships and romantic relationships offer insights into the strength of social boundaries (Kao and Joyner 2006; Qian and Lichter 2007). However, only limited research has examined how ethnic identity—conceived broadly as the strength of attachments and not only self-labeling—is associated with relationship formation. Jessica Vasquez's work is an exception; she finds that though Mexican Americans with "thinned ethnic attachments" often marry whites, bicultural identity often persists even within intermarriages (2011; Vasquez-Tokos 2017). Nevertheless, life course theory's emphasis on how individuals' lives are linked to significant others suggests that those with strong ethnic identities may be most likely to limit both close friendships and cohabiting or marital relationships to ethnic group members.
Few studies examine how ethnic identities relate to cultural practices among the second generation. Given that ethnic identity signals attachment to cultural heritages, we expect strongly ethnically identified respondents to incorporate cultural practices related to their parents' homeland into their lives, including passing on foreign languages to their children. However, Vasquez finds that Mexican Americans with strong emotional attachments to their ancestry often do not maintain cultural practices and language (2011). These questions have not yet been explored among second-generation adults from diverse ethnic backgrounds.
This study addresses three main questions. First, how do ethnic identities among immigrants' children change across the life course from adolescence to middle adulthood? Second, what types of ethnic identity formations are found in middle adulthood and how do these vary by gender, immigrant generation, and national origin? Third, are ethnic identities in middle adulthood associated with politics, relationships, and cultural practices? If so, how?
METHODS, SAMPLES, AND DATA COLLECTION
We rely on a mixed-methods analysis of survey and qualitative data drawn from a sample of original respondents from the Children of Immigrants Longitudinal Study (CILS), a longitudinal study of foreign-born and U.S.-born children with at least one immigrant parent and [End Page 29] attending eighth or ninth grades in the San Diego City schools in the fall of 1991 (see Portes and Rumbaut 2001).4
The study used a school-based sampling frame to accurately represent the population of children of immigrants in San Diego younger than the age that they can legally drop out of school. Because the data are limited to the baseline sample drawn in San Diego, the findings cannot be generalized to the United States as a whole. However, San Diego is a large immigrant destination and a principal site of contemporary immigrant and refugee settlement.
Respondents were surveyed during four periods from early 1992 to early 2016: the first in spring 1992 (14.2 years old on average), the second in spring 1995 (17.2 years old), the third between 2001 and 2003 (mid-twenties), and the fourth between 2014 and 2016 (late thirties). The third phase included surveys from 1,480 respondents averaging 24.2 years old and in-depth, open-ended qualitative interviews conducted about a year after the survey with a representative subsample of 134 respondents. More than twelve years later (2014 to 2016), that subsample was tracked, and a full fourth wave of surveys and in-depth qualitative interviews conducted. We refer to each wave of data collection as T1, T2, T3, and T4, respectively. At T4, we completed 111 in-depth interviews (83 percent of the 134) with respondents in California and across the country (from Alaska to Atlanta and New York City), as well as Mexico, ranging in age from thirty-six to thirty-nine, with an average age of 37.2 years.5 Table A1 provides a descriptive portrait of these respondents.
The interview format at T4 was flexible enough to delve deeply into the most important aspects of each person's experiences, while also allowing us to collect standard survey data to permit direct comparisons to their earlier responses. We analyzed the data using multiple methods. The data collected through closed-ended responses were combined with the existing CILS longitudinal data and analyzed using descriptive statistics. We analyzed the interview data with the assistance of Dedoose, a software program for analyzing qualitative and mixed-methods data, using the constant-comparison method.
Ethnic Identity Measures
Ethnic identity measures at T1, T2, T3, and T4 are based on the same open-ended question asked in each wave: "how do you identify? That is, what do you call yourself?" Examples include Asian, Hispanic, American, Mexican, Mexican American, Chicano, Latino, Filipino, Filipino American, Vietnamese, Vietnamese American, Lao, Lao American, Cambodian, and Cambodian American. In the surveys at T1, T2, and T3, respondents wrote in an open-ended response. In the T2 and T3 questionnaires, this question was followed by another, "how important is this identity to you?" with three options: "not important," "somewhat important," or "very important." In the in-depth interviews at T3 and T4, the same initial questions about ethnic self-identity and importance were asked but were followed by more in-depth discussion about identities, such as "do you always use this term to describe yourself?" and "What does being [ethnic identity label] mean to you?"
Following Rumbaut, we initially coded responses into five mutually exclusive ethnic identity labels: American includes those who responded only as American; national includes those who referred only to their or their parents' country of birth, such as Mexican; hyphenated includes those who combined the home country with an American identity, such as Filipino American; panethnic includes those who used a panethnic or racial label, such as Hispanic or Asian; and a last heterogeneous group included some who asserted mixed identities, such as Mexican Filipino or black Filipino, and others who did not respond in racial or ethnic [End Page 30] terms at all, such as human being (1994). Placing respondents in one category was not always straightforward. In open-ended interviews, respondents often mentioned more than one ethnic self-label (a finding we elaborate on later). In cases with multiple labels, we coded the most important or preferred label as the primary label. If this was unclear, we used the more specific label. For example, some stated that they were Hispanic and Mexican. Unless Hispanic was more important or preferred, their primary identity label was the national identity, Mexican. However, this coding scheme was used only for the first set of analyses to make comparisons across earlier waves of the study. We also conducted an inductive analysis of the qualitative data on ethnic identities, examining how and when multiple labels were used, and the subjective meaning of ethnicity. Through this process, we identified four distinct ethnic identity formation patterns, discussed in the following section.
Respondents were asked several questions about politics (see table A1). First was whether they considered themselves politically conservative, moderate, or liberal. Most chose one of these options; several responded that they were socially liberal and fiscally conservative; and some refused to answer. Respondents were also asked whether they considered themselves Republican, Democratic, Independent, or Other; the four respondents who stated Other were recoded into a residual other-no answer category. U.S. citizens (93 percent by T4) were asked whether they voted in the 2012 election and for whom. All respondents were asked about political engagement beyond voting.
Respondents also provided their relationship histories and the ethnic background of their partners. For those currently or previously in a cohabiting or marital relationship, we coded whether the partner was a member of the same national-origin group, same panethnic group, white American, or another non-white ethnic group (for example, a Filipino married to a Chinese American would be coded as panethnic, and a Filipino married to a Mexican American would be coded as other non-white); we used the most recent partner for those with multiple partners. Respondents also provided the ethnic backgrounds of their close friends; we created a binary variable indicating whether they had close friends outside their panethnic group.
We examined cultural practices in two main ways. First, we coded thirteen categories from an open-ended question asking "what cultural practices of [your ethnic background] do you practice?" The most common responses were food (39 percent), language (20 percent), and holidays (19 percent). Only 18 percent replied "none." We created a categorical variable indicating whether respondents mentioned no practices, one, two, or three or more. We also asked a series of close-ended questions about language use with others and focus here on language with children (for those with children). We created a dichotomous measure indicating whether respondents speak about the same amount or more in a non-English language with their children. This outcome provides a window into language transmission to the next generation.
Our first set of analyses focus on changes in ethnic identity labels; however, we find that outward expressions of identity fail to capture the complexities of ethnic identification. Further analyses of ethnic identity and salience revealed four types of ethnic identity formations, which vary by demographic characteristics and relate to politics, relationships with others, and culture.
Change in Primary Ethnic Identity Labels Across the Life Course
Like previous research at earlier life stages, we find that immigrants' children continue to most commonly use national (51 percent) and hyphenated (20 percent) ethnic labels as they age into middle adulthood. Panethnic (12 percent), mixed or other (13 percent), and especially plain American identity labels (3 percent) are used less often (see Portes and Rumbaut 2001; Tovar and Feliciano 2009). Figure 1 shows the percentage of respondents who changed their ethnic identity label across time (see also figure A1). About half changed ethnic identity labels in adolescence (T1 to T2) or the transition [End Page 31] to adulthood (T2 to T3), but only 30 percent changed from early to middle adulthood, suggesting a crystallization of ethnic identity labels for most in adulthood. Across all four periods, the modal response was to use the same ethnic identity label. Twenty-six percent of respondents (n = 27) self-identified using the same ethnic identity label in response to an open-ended question from early adolescence to middle adulthood, revealing remarkable stability across more than twenty-two years. Of those who used the same ethnic identity label, 56 percent used a national term, 22 percent a hyphenated term, 19 percent a mixed term, and only one consistently a panethnic label. Another 33 percent used the same ethnic identity label at three of the four time points, again suggesting that stability is more common than fluidity. Only one respondent chose a different ethnic identity label each time.
Multiple Ethnic Identity Labels and Substantive Versus Nominal Change
Analyzing in-depth interview data at T3 and T4 allows us to examine whether different ethnic identity labels correspond to substantive differences in ethnic self-identification from early to middle adulthood. This analysis indicates that only 17 percent of the ethnic identity label change suggested by the quantitative data from T3 to T4 is meaningful, whereas nearly half of the change in ethnic labels is in name only (indicated by the shaded area of figure 1). Respondents often use ethnic identity labels interchangeably. For example, thirty-year-old Enrique, who migrated to San Diego as a four-year-old, explained: "I never changed my identity as far as I'm Hispanic, I'm Latino, I'm Mexican, I'm Mexican American."6 Similarly, Anh, who migrated from Vietnam at age five, used some terms interchangeably. As a twenty-three-year-old woman, she identified as Vietnamese American, but at age thirty-six as Vietnamese. When asked whether she ever used different terms in different contexts, Anh responded, "Sometimes I'll just do Vietnamese and sometimes I'll do Vietnamese American."
Some respondents clearly articulated the contexts in which they used different terms. Enrique explained:
"It just depends on who's asking me and how they're going about asking me . . . At my work . . . it's like, "Hey . . . what are you?" "I'm Mexican American." I wanna make sure they know I'm an American . . . If I was . . . just socializing, "Hey what are you?" "Oh, I'm [End Page 32] Mexican." Because the conversation's not about your citizenship [or] your status."
Some Asian respondents similarly explained that they reserved hyphenated labels for white Americans and used national terms with Asians. For example, 1.5-generation Brian, who identified as Chinese American, said, "if it's . . . an Asian person, I'll just say Chinese . . . . I'm more likely to say Chinese American if it's a . . . a white . . . person." He explained his reasoning by sharing a recent incident in which three older white people assumed that he did not speak English. For Brian and other Asians, a hyphenated American ethnic identity label served to combat a stereotype of Asian foreignness (Tuan 1999), whereas for Enrique and other Latino respondents, a hyphenated American identity combatted an undocumented immigrant stereotype.
The minority of respondents (16 percent) whose ethnic identity label in middle adulthood reflected a substantive change from early adulthood moved away from identities rooted in their or their parents' origin countries and toward Americanized identities, especially pan-ethnic. For example, Kham, a thirty-six-year-old man who migrated at age twelve and previously identified strongly as Lao, identified solely as Asian. When asked, "what does it mean to say that you are Asian?" Kham explained, "the longer we live here . . . we look at ourselves as being American now. . . . Everybody in our . . . household became U.S. citizens . . . we consider ourselves American."
Although Kham clearly distinguished between a panethnic identity (Asian) and his previous identity as Lao, it was more common for respondents to invoke panethnic labels in conjunction with national or hyphenated labels: only 12 percent used panethnic labels as their primary label, but another 34 percent mentioned panethnic labels. Several respondents noted that specific ethnic identity labels were rarely options on forms, or they did not think others were familiar with specific ethnic groups, so they would use a panethnic term, although their personal identity was more specific. For example, second-generation Vanna said, "I would just say Asian . . . and . . . in details I would say Cambodian." The use of panethnic terms on forms is consistent with our finding from T3 that 31 percent of respondents used a different ethnic identity label in the qualitative interview than they had on the written survey, and most who identified in panethnic terms on the survey used a national or hyphenated term in the interview.
Changes in Ethnic Identity Salience Across the Life Course
Ethnic self-labels are one aspect of identity, but we also considered whether ethnic identity changes in importance as immigrants' children age further into adulthood. Figure 2 shows that ethnic identity often becomes less important later in the life course. Although 64 percent of adolescents (T2) and 57 percent of young adults (T3) described their ethnic identity as very important, this proportion dropped to 41 percent by their late thirties (T4). These findings hold even among the fifteen respondents who consistently identified in national terms across all four waves: the number describing this identity as very important declined from eight in late adolescence, to six in early adulthood, to only three in middle adulthood (not shown).
Our qualitative findings suggest that the declining importance of ethnic identity reflects a shift in priorities as respondents age into new roles and identities. For example, thirty-seven-year-old Brian explained the declining importance of his Chinese American identity this way:
when we were younger . . . our . . . sense of self or identity is . . . tied with what the outside world sees you as. . . . And so, we all—the majority of my friends are Asian—had to figure out what that actually meant. It was a much more relevant and important topic back then. Now that we're all established professionals and . . . we're married with kids and all this stuff. . . I'm never gonna say it's inconsequential or insignificant. But it's not one of the driving things. It doesn't define us anymore.
Brian's explanation is consistent with life course theory, which emphasizes how "changing lives alter developmental trajectories" (Elder 1998, 1), and how development is a lifelong [End Page 33] process shaped by relationships with significant others (Elder, Johnson, and Crosnoe 2003).
Ethnic Identity Formations in Middle Adulthood
Despite an overall pattern of decline in importance and the common use of multiple ethnic labels, inductive analyses of the qualitative interview data revealed substantial variation in the subjective meaning and importance of ethnic identification among adult children of immigrants in their late thirties. This analysis focused less on the label and more on the attachment to an ethnic identity, revealing four ethnic identity formations, as shown in table 1: a strong ethnic identity, a moderately important ethnic identity, an American-oriented ethnic identity, and a de-ethnicized identity (for variations by ethnic label, see table A2). Most respondents identified with an ethnic or pan-ethnic group and held this identity to be at least somewhat important.
Strong Ethnic Identity
The largest group of respondents (36 percent) expressed a strong ethnic identity, indicating that their ethnic background and associated culture was central to their personal identity. Vinh, a 1.5-generation woman, explained it this way: "I think being Vietnamese American is such a specific . . . experience that I think it is a defining part of—of my identity." Similarly, second-generation Elaine responded that her identity as Filipino is "very important . . . it's just the whole legacy thing . . . that's what I am. And that's all I've known to be." These respondents often described their ethnicity as a source of pride and a core part of themselves.
A few respondents, particularly Mexican-origin men, connected their pride in their ethnic identity to an awareness of negative stereotypes about their ethnic group and a hostile political climate—vivid examples of reactive ethnicity (Rumbaut 2005, 2008; Portes and Rumbaut 2014). Leo, who migrated as an infant and by age thirty-six had completed a master's degree, articulated how his strong ethnic identity was in part a response to hostile outside forces:
For me it's very important to label me Mexican American because I wanna show people . . . two sides. I wanna show my parents [End Page 34] that . . . I'm proud of being Mexican. And the second aspect is, I wanna show . . . conservative people . . . we don't just come to this country and take advantage of it . . . I'm an immigrant . . . and I succeeded.
Expressions of strong ethnic identities do not correspond to rejections of American identities. Most with strong ethnic identities also clearly identify as American. For example, thirty-six-year-old Isabella emphasized her American identity along with her Mexican identity: "[My identity as Mexican American] is very important because my family is from Mexico and . . . they came here to be better, you know, for that American Dream . . . I'm proud, you know, to say that I'm Mexican, but I was born here, so I am American" (emphasis added).
Identifying as American, along with maintaining strong ethnic attachments, was not just common among the second generation who preferred hyphenated or American-based labels such as Mexican American or Chicano. For instance, 1.5-generation Anh recognized her dual identities: "[Identifying as Vietnamese] is very important because it's a huge part of me . . . [being Vietnamese American] is really important too . . . I do all the stereotypical American things."
Moderate Ethnic Identity
Other adult children of immigrants described their ethnic identity as somewhat important, but not central to who they are or their daily life. For example, second-generation Tara, whose parents migrated from Mexico and whose grandmother had migrated to Mexico from China, said, "I think it's important [to identify as Chinese Spanish] . . . I like to identify with my cultures . . . I'm not, like, adamant about it. But, you know, I think it's . . . important." Similarly, second-generation Jimmy said, "I don't think [being Indian] necessarily defines who I am . . . but it's something I want to always have some connection to."
Respondents with strong or moderate ethnic identities accounted for 69 percent of the sample, indicating that a substantial majority of immigrants' children maintain an attachment to an ethnic identity rooted in their homeland by middle adulthood. Nevertheless, 31 percent were not attached to their parents' homelands or their ethnic origins.
American-oriented respondents may respond to questions from others about their background with an ethnic identity label, but personally feel they are really American. This category, making up 14 percent of the total (sixteen cases), includes three respondents who identified solely as American as well as some who responded with varied ethnic labels. Mike, for example, whose father is white American and mother is from China, said that he is "Chinese and Irish" but that it is not important because "I don't really know either of those sides, so I'm American." He explained, "It's more that I identify myself as a courtesy to others." Mike suggests it is only because others want to classify him that he mentions ethnic origins beyond American.
Similarly, Trung, who came to the United States from Vietnam as an infant, explained that he identifies as Asian because "when you look at me, I'm not . . . the—physical embodiment of . . . an American . . . But I still consider myself an American since I lived here all my life and . . . this is where I call home." Trung and others like him suggested they would prefer to identify only as American but cannot because they do not appear white.
Others, like Thanh, who also came from Vietnam as a young child, asserted that a common American identity outweighs ethnic origins. Despite stating a Vietnamese identity in response to the question, Thanh also asserted that "[being Vietnamese is] not important . . . we're all Americans, I think. . . . Everyone's coming from different countries, so, the fact that I happen to be from Vietnam, not . . . a big deal, I don't think." Thanh alluded to how context—in this case, one in which diverse ethnic origins are common—shapes her identity. Similarly, Kim Cuc, who identified as plain "American," said that "living in San Diego you see a multicultural group of people." However, Kim Cuc's awareness of contextual variation is clear when she said, "in San Diego I'm part of [the American mainstream.] Outside of it, I'm not. [End Page 35] I've been to states like Texas and Florida where they . . . look at me as an alien and they make it known."
Like American-oriented respondents, those with de-ethnicized identities (16 percent, eighteen cases) are not attached to identities rooted in their origin country. Unlike those with American-oriented identities, however, these respondents expressed indifference toward ethnic and national identities in general. Although these respondents would respond (often with a specific national term) when asked, they said that ethnic background did not matter to them. For example, thirty-nine-year-old 1.5-generation Vanessa responded to a question about the importance of her identity this way: "[Being Filipino] is not really important. Like, I don't care. Like, I'm not like 'I'm proud! I have the—the sticker.' I'm like, no . . . I'm a person . . . it's not important to me." Individuals in this category suggested that their common humanity was more important than their ethnicity. For example, 1.5-generation Kye will tell people he is Chinese, but explained that
your question . . . it's kind of hard for me to answer because . . . I kinda feel like the whole classification is . . . unnecessary. If we classify ourselves as humans, it makes everything so much easier . . . we don't need to pick apart our . . . identities . . . we're all one, one kind.
Similarly, Santi, a thirty-six-year-old second-generation male whose parents were from Mexico, identified as Hispanic but said, "I think it's more important to say I'm human . . . I think we're all mixed a little bit with something nowadays." Santi's comment suggests that growing up in a context in which mixed ancestries are commonplace has shaped his identity development.
Differences in Ethnic Identity Formations
Table 1 shows that ethnic identity formations do not vary significantly by gender, but, as expected, do vary by immigrant generation and [End Page 36] national origin. However, contrary to linear assimilation theories predicting that individuals who have lived in the United States longer would be most oriented toward American identities and away from ethnic identities, we find that in middle adulthood, U.S.-born respondents with two immigrant parents were the most strongly attached to their ethnic identities: 49 percent had a strong ethnic identity relative to the 32 percent of those who migrated as children.7 These findings differ from earlier CILS research on adolescents and young adults showing ethnic identity was more salient for the foreign-born (Tovar and Feliciano 2009; Rumbaut 2005). Among the 2.5 generation (with one U.S.-born parent), the pattern is more consistent with a linear assimilation trajectory in that fewer expressed strong ethnic identities (21 percent), and more expressed American-oriented identities (29 percent).
With the caveat that the sample sizes for many groups are small, we note significant national-origin differences in ethnic identity formations. First, respondents with origins in Mexico (50 percent), the Philippines (50 percent), and Vietnam (43 percent) were most likely to express strong ethnic identities in middle adulthood, and the Chinese (10 percent) and East Indians (none) the least likely to do so. The Vietnamese sample was split: 43 percent also expressed American-oriented identities, versus only 4 percent of Mexicans, 10 percent of Filipinos, and 6 percent of Cambodians, Lao, and Hmong. Also, East Indians (50 percent) and Chinese (40 percent) were most likely to express de-ethnicized identities. Notably, none of the sixteen respondents whose parents came from different countries expressed de-ethnicized identities, and the same percentage (38 percent) expressed American-oriented or moderate ethnic identities.8 Despite these national-origin differences, nearly all types of ethnic identity formations are found within each national-origin group, a finding that reinforces life course theory's principle of individual agency within structural contexts (Elder 1998).
Ethnic Identities and Politics, Relationships, and Culture
Are ethnic identity formations associated with other outcomes of interest, such as politics and family and cultural practices in middle adulthood? Table 2 considers whether these outcomes are significant correlates of ethnic identity formations. In terms of politics, the findings do not line up neatly with existing assumptions. First, we do not find that those with the strongest ethnic identities are the most liberal politically. The biggest differences are in the socially liberal, but fiscally conservative category: 31 percent of respondents with American-oriented identities held this political orientation versus only 6 percent of those with de-ethnicized identities and 8 percent of those with strong ethnic identities. Contrary to conventional assumptions of identity politics, more strongly ethnically identified respondents said that they were somewhat conservative (30 percent) than those with de-ethnicized (11 percent) or American-oriented (6 percent) identities did. Finally, those with de-ethnicized identities were much more likely to not express a political orientation at all (28 percent, versus only 0 to 3 percent among those with other identities), suggesting that individuals indifferent toward ethnicity are similarly apathetic about other aspects of identity and social life.
Ethnic identity formations and political party affiliations are also related. Somewhat as expected, individuals who express at least moderately strong ethnic identities are more likely than are those with de-ethnicized or American-oriented identities to identify as Democrats (58 and 70 percent versus 39 and 25 percent, respectively). American-oriented respondents are [End Page 37]
[End Page 38] especially likely to identify as independents (44 percent, versus 5 to 28 percent of others).
In terms of political behaviors, we do not find significant differences in whether one voted in the 2012 presidential election. However, among those who voted in 2012, American-oriented respondents were least likely to vote for Barack Obama (64 percent versus 82 percent or higher among those with other ethnic identity formations). Although not statistically significant, that those lacking ethnic attachments were less likely to participate in politics beyond voting (6 percent versus 24 to 28 percent of others) is consistent with the earlier interpretation that a lack of attachments to ethnic identities extends to indifference in other social realms.
Beyond politics, ethnic identity may influence family and friendship formations and cultural practices. In particular, we might expect those who are strongly ethnically identified to marry and form friendships primarily within their ethnic group, practice cultural traditions, and pass on cultural practices, such as languages, to their children. However, differences by ethnic identity formations in interethnic unions are not substantial or statistically significant, nor always in the direction of theoretical expectations. For example, although a higher percentage of those with strong ethnic identities had partners who share their ethnic background than those with American-oriented identities (43 percent versus 31 percent), 20 percent of those with strong ethnic identities married or cohabited with white Americans, whereas only 6 percent of American-oriented respondents did. In terms of interethnic friendships, the findings are more consistent with existing theories. Respondents with strong or moderate ethnic identities were less likely than those with American-oriented identities to have close friends outside their panethnic group, although those with de-ethnicized identities were least likely to have close interracial friendships.9
The association between ethnic identities and cultural practices is consistent with existing theories: those who were not attached even moderately to their ethnic groups (especially American-oriented respondents) were less likely to mention any cultural practices related to their ethnic backgrounds. Those with strong ethnic identities were most likely to mention three or more cultural practices; moderately ethnic identified respondents fell in the middle.
Finally, for respondents who were parents, we considered whether ethnic identity formations related to speaking foreign languages with their children. Here we see a clear pattern: those who expressed strong (54 percent) and [End Page 39] moderate (32 percent) ethnic identities were more likely than those with de-ethnicized (18 percent) or American-oriented (11 percent) identities to speak to their children at least half of the time in a language other than English. When we limited the sample to those who considered themselves bilingual, these differences are even more pronounced: 72 percent of bilin-guals with strong ethnic identities spoke a non-English language to their children versus only 14 percent of American-oriented bilinguals. A strong ethnic identity relates to maintaining at least some cultural practices, including passing on a foreign language to children.
DISCUSSION AND CONCLUSIONS
Previous research has focused on self-identity expressions in the often volatile developmental stages of adolescence or early adulthood. This study is among the first to examine ethnic identity among immigrants' children as they age into the relatively stable life course stage of middle adulthood. Through our mixed-methods approach, we show the importance of incorporating in-depth qualitative interview data to understand the subjective meanings behind ethnic identity labels. Our data, drawn from a more representative sample than studies limited to snowball or convenience samples, suggests a wide range of ethnic self-identity formations among immigrants' children in middle adulthood. It is most common to have a strong ethnic identity; these adult children of immigrants indicated that identifying with their heritage or culture was a source of pride and a defining part of themselves. On the other hand, a significant portion of adult children of immigrants in our sample did not feel even moderately connected to an ethnic group, revealing de-ethnicized or American-oriented identities, which, in some cases, they lamented they could not fully claim because others wanted to classify them based on their nonwhite phenotype. Thus, although many adult children of immigrants fully embrace their ethnic identities, others feel constrained by a U.S. racial structure, which limits how much choice they have to assert a plain American identity or to not identify in ethnic or racial terms at all (Waters 1990; Song 2003; Waters 2001). Such diversity, which is evident even among those who respond with the same outward ethnic identity label (such as Filipino or Mexican), supports the life course perspective; previous research suggests that when structural constraints are loosened, in this case a multicultural historical context in which self-expression is encouraged, individuals can exert agency in choosing how to ethnically self-identify, but that structural constraints within historical contexts also shape development (Csizimadia et al. 2012; Elder et al. 2003).
We distinguish between ethnic identity labels (outward expressions of self-definitions) and ethnic identity formations (a subjective sense of ethnic belonging and its importance). We find that ethnic identity labels are used more consistently in adulthood than in adolescence, and overall, are much more stable than fluid. Moreover, measuring fluidity in ethnic identities based on survey responses (even to open-ended questions), misses the degree to which ethnic labels are used interchangeably. Consistent with previous research, our qualitative data show that immigrants' children often express different labels to different audiences in strategic ways, for example, to emphasize Americanness to a white native-born American, or to clarify one's national origin to an Asian American (Dowling 2014; Kasinitz et al. 2008). Like prior studies, we also find that adult children of immigrants tend to view panethnic terms, such as Hispanic or Asian, as broader groups, corresponding to official racial categories, which they layer over more specific ethnic identities (Itzigsohn and Dore-Cabral 2000; Kibria 2002; Lien, Conway, and Wong 2003; Okamoto 2014; Padilla 1985; Dowling 2014). Overall, our findings suggest that though the presentation of self-identifications to others may be contextually contingent, internal ethnic identities remain fairly consistent more often than not. However, whereas stability in ethnic self-labels was the more common pattern from early to middle adulthood, the importance of ethnic identity tended to diminish over this part of the life course as new social identities emerged.
Existing assimilation theories suggest two modes of ethnic self-identity formation. One view posits that the forging of a reactive ethnicity in the face of perceived threats, discrimination, [End Page 40] and exclusion may account for the rise rather than the erosion of ethnicity over time (Rumbaut 2005, 2008; Portes and Rumbaut 2001, see also Massey and Sánchez 2010). This contrasts with the conventional account maintaining that descendants of European immigrants experienced a linear (if at times bumpy) process of assimilation, in which their ethnic self-identities thinned over time in the United States, such that ethnic identity became an optional, passive, leisure-time form of symbolic ethnicity (Alba 1990; Gans 1979; Waters 1990; see also Child  1970; Nahirny and Fishman 1965). Yet, our findings suggest that most adult children of Asian and Latin American immigrants in their late thirties developed ethnic identities in varied ways that fall somewhere in between these two poles. For instance, only a handful of respondents, all Mexican-origin men, conveyed a sense that their ethnic self-identities were primarily reactive, shaped by continuing awareness of discrimination, negative stereotyping, and a hostile political context. A much larger proportion of adult children of immigrants identified strongly with their parents' national origins or a combination of national origins and American identities because they maintained strong, positive, attachments to family and cultural practices that they associated with their ethnic origins. We must keep in mind, however, that respondents maintained strong ethnic attachments within a historical and social context they perceived as welcoming diversity and self-expression. Reactive ethnic identity formations may be more common in different historical and social contexts and among different ethnic groups facing more hostile political contexts—such as Sikh Americans facing post 9/11 backlash or the children of undocumented immigrants in today's political climate (on Sikh Americans, see Kurien 2018).
Although we find that ethnic identity formations vary significantly by national origin (for example, more Mexican-origin than Chinese-origin respondents identify strongly with their ethnicity), within each ethnic group we also find individuals attached to their ethnic identities and others not, such that the variation within each national-origin group is greater than the variation between them. Across all national-origin groups, we find a common pattern of difference by immigrant generation: those born in the United States with two immigrant parents being more likely than either the 1.5 or 2.5 generation to maintain strong ethnic identities into middle adulthood, and least likely to have American-oriented identities. Further, 1.5-generation adults in their late thirties were most likely to have de-ethnicized identities. The reasons behind these patterns are difficult to discern, but it may be that those who migrated as children are more aware of being disconnected from their origin country, and thus, by middle adulthood, are less likely to maintain those ethnic attachments. In contrast, the second generation may cling strongly to ethnic attachments to overcompensate for feeling they lack ethnic authenticity having been born in the United States. It may also be that immigrants who have crossed national borders themselves, even as children, may be more likely to downplay or "see through" the borders and boundaries implied by ethnic labels distinguishing us versus them. Exploring the sources of such variation further is a question for future research.
By examining ethnic identity among adult children of immigrants in their late thirties, who have often completed schooling, married, and formed families, we can consider the consequences of ethnic identity formation in ways previous studies have not. Of particular importance in today's context is how ethnic identity formations relate to political orientations and behaviors. We find that political views vary by ethnic identity formations, but in complex ways. For instance, those with strong ethnic identities were more conservative than those with American-oriented identities, which may relate to conservative religious views tied to ethnic backgrounds. However, this does not currently translate into affiliations with the Republican Party. In addition, respondents with American-oriented identities were more likely to express socially liberal, fiscally conservative political views than those with other ethnic identity formations, and more likely to identify as political independents. Ethnic identities do not appear to coincide with political engagement, including voting, except that adult children of immigrants with American-oriented [End Page 41] identities were more likely to vote for someone else over Barack Obama in 2012 than those with any other ethnic identity formation. These findings challenge assumptions that the growing population of Latinos and Asians who are descendants of immigrants will constitute a monolithic voting block (Immigration Policy Center 2010), and instead reveal that political views, affiliations, and behaviors vary across the range of ethnic identity formations within these populations.
Like recent research on Mexican Americans, our findings do not support linear assimilation theories positing that those with stronger ethnic identities are more likely to marry within their ethnic group, or, conversely, that intermarrying leads to weaker attachments to ethnic identities (on Mexican Americans, see Vasquez-Tokos 2017). However, we do find that adult immigrants' children with stronger ethnic identities are more likely to have close coethnic friends, maintain cultural practices associated with their ethnic background, and speak with their children in a language other than English. This is not to say that retaining strong definitions of self as ethnic relate to patterns of behavior that are markedly distinct from those of nonethnic Americans. Indeed, most respondents also said that their everyday lifestyles and behaviors made them part of mainstream America. Nevertheless, our strongly ethnically identified respondents conveyed that their ethnicity and cultural practices made them unique, and as parents, some are at least attempting to transmit an important cultural element, language, to the next generation. Whether those efforts will be successful remains to be seen. Moreover, even though most bilingual adult children of immigrants with strong ethnic identities spoke a language other than English with their children, they constituted less than one-quarter of all parents in our sample.
Overall, this study suggests that ethnic identity formations among the second generation in middle adulthood are complex and varied. Although for some, attachments to an ethnic identity remain strong into their late thirties, for many, ethnic identity attachments begin to wane through the life course, as their social identities as parents, workers, or spouses become more central. Moreover, ethnic identities relate to political and social views and behaviors in complex, and often modest, ways. Future research should further explore the consequences of varied ethnic identities for the resilience or decline of ethnic boundaries into the next generation, as well as how ethnic identity formations vary in different historical and social contexts. [End Page 42]
Cynthia Feliciano is professor of sociology at the University of California, Irvine.
Rubén G. Rumbaut is Distinguished Professor of Sociology at the University of California, Irvine.
We thank Dr. Linda Borgen and Alma Nidia Garza for helpful research assistance and the Russell Sage Foundation for funding this research.
1. The new second generation refers to children of immigrants who have come to the United States since the 1960s. Although the term most often refers to U.S.-born children of immigrants, we use it here to also include those who migrated as children to the United States (see Portes and Rumbaut 2001).
2. We use the term incorporation trajectories to refer to adaptation paths and patterns over time.
3. Proposition 187 was halted by multiple lawsuits after passage and found unconstitutional by a federal court in 1997.
4. The original CILS study included a South Florida sample, who were not followed beyond early adulthood.
5. Systematic comparisons between the 111 interviewed in-depth and the full, original T1 survey respondents, along key variables (including age, gender, grade point average, family socioeconomic status, and so on) showed no sample attrition bias on any characteristic except, in one instance, national origin, which was by design because the T3 in-depth interviews intentionally included a larger sample of Chinese respondents to facilitate comparisons with other ethnic groups.
6. All names are pseudonyms.
7. Analyses restricted to only respondents with parents born in the same country reveal the same pattern of differences between the 1.5 and the second generation, as do analyses focused only on the two largest groups, Mexicans and Filipinos.
8. Those in the mixed category included seven with a white American parent and an Asian immigrant parent, four with a black American father and a Filipino immigrant mother, two with Filipino immigrant fathers and Mexican immigrant mothers, two with white American fathers and Mexican-born mothers, and one with a Mexican American father and Filippino immigrant mother.
9. Further analyses show that those with American-oriented identities were both more likely to have white American friends and more likely to have nonwhite friends of another panethnic group than those with strong or moderate ethnic identities, whereas de-ethnicized respondents more often had no close friends.