Third Culture Kids and Five Centuries of Mixed Heritage

Short answers. Six questions the paper was written to answer, each with the paper's confidence in its own answer: strong, moderate, or weak. The sections named carry the evidence, the population it was measured on, and its limits. Most of that evidence comes from lives written down because they were exceptional, or from comfortable samples; section 4.1 says what that does to each answer, and a fifth anchor, the children of marriage migrants in Korea and Taiwan, carries the ordinary case.

1. What happened to mixed people in history, and over the generations? In the paper's library of 28 historical groups, more were absorbed into a majority or a parent group (9) than persisted as named groups (5); 9 dispersed abroad and 5 could not be traced. The five that persist each held a territory, a language, or a state category (sections 10 and 11).^[18,155,359] Confidence: moderate; the library is not a sample, one coder read it, 22 of its 28 rows carry a weak code, and 9 is an upper count (two pairs of rows overlap, and the pardos hold a state category).
2. Is being unmixed worth more than being mixed? No, not as a trait: every present population is a past mixture,^{[74,131,133,191,203,271,289]} and a premium on being unmixed exists only where a majority or a state builds it.^[205] The one practical cost found is in the bone-marrow registry, which rewards being typical rather than being unmixed, and the measured costs of being mixed are social, tracking discrimination aimed at mixedness; the largest genetic studies measure a cost of close kinship, not a bonus for distance (section 11.1).^{[25,93,130,166]} Confidence: moderate.
3. Stay or leave: which family line thrives? Leaving buys income, 263 percent more within a year in the one lottery experiment, and, in group averages, schooling; it loses the heritage language in about two generations, and by the third 42 percent of Asian-origin children (55 percent with one Asian side) are no longer identified as Asian.^{[101,106,218,284]} Leavers arrive healthier than the destination population, partly by selection; the advantage erodes with time, is weaker in childhood, and fails on some Asian outcomes; well-being is mixed, belonging is unmeasured on both sides, and no study measures what the line that stays keeps (section 12).^{[7,10,116,135,350]} Confidence: moderate for income, weak to moderate for the first generation's health, weak for the rest.
4. How do third culture kids develop, where do they struggle, and how do they use their edge? Difference registers early and the label moves in adolescence: infants prefer a native accent at five to six months, and about half of the Black-white and Asian-white teenagers in one national sample, 57 and 53 percent, kept their label six years on (sections 5 and 6).^[98,180] The paper reads the measured harm as tracking invalidation and discrimination more than mixedness, with the household (family structure, poverty) explaining part of what remains, and the best-supported edges are creativity and the bridging seat, which pays and burdens whoever holds it; both are context effects open to anyone, whether in-between people hold more bridging seats is an inference, and the bilingual brain is not among them (sections 7 and 14).^{[47,93,200,207]} The edge has a life cycle: early arrival and strength in both languages build it, bridging seats and diverse teams' creativity pay it rather than team performance, the loss of the heritage language thins it, a growing supply of speakers may thin it further on one cross-section, and a clash between the two countries turns it on its holder (section 8).^{[3,17,38,86,310]} Confidence: moderate on development and costs, moderate on building the edge, weak on using it.
5. How do today's Euro-Asian children compare with the historical Eurasians? Many of the old Eurasian groups had a legal category or a protected niche from a colonial patron, and most lost them with the empire (India's Anglo-Indian seats lasted to 2020); Hong Kong's families also married among themselves. Today's children have no legal category of their own (England and Wales offer a census box), no niche, and no patron, and their label thins by the third generation (section 10.9).^{[101,199,364,367]} In England, where they are counted at school, they outscore the White British majority inside and outside free school meals, and by 19 enter higher education at 56 against 38 percent.^[87,89] Confidence: moderate on the history; on the present, moderate for schooling (one country's official series) and weak for identity and work (88 US respondents, no labour-market split).
6. Are they becoming a new people? No. By four tests (a label most members use, a boundary, institutions, transmission) in-between kids today are a category with a vocabulary and a few organisations, not a group: 61 percent of mixed-background Americans do not call themselves multiracial, and 55 percent of third-generation children with one Asian side are not identified as Asian (section 11).^[101,262] Confidence: moderate for the United States (two single surveys and the paper's judgement), weak elsewhere, where no comparable count exists.

Some children differ from the people around them in how they look, how they speak, and how they think, and so they cannot blend in. This paper calls them in-between kids; the core case is a child of mixed heritage raised across cultures. It asks what happens to such a child across one life and across three generations of her family, and what that tells a family deciding whether to stay or to leave. It reads the evidence through Pollock and Van Reken's two questions (do you look like the locals, do you think like them) plus a third the paper adds (do you sound like them). The method is a research synthesis in which each finding names the population it was measured on and whether it replicated, joined to a coded library of 28 historical mixed groups, read by one coder and not a sample. Four lives run through the paper: Barack Obama from a Honolulu classroom in 1971, Ariana Miyamoto from a Sasebo schoolyard, the Ho Tung family of Hong Kong across three generations to a camp gate in 1942, and two Vietnamese Amerasians, one who left in 1985 and one who stayed; each is on record because of what made them exceptional. A fifth anchor is a cohort with no memoir: the children of marriage migrants in Korea and Taiwan, many with Vietnamese mothers, from families their ministry describes as tending to low incomes, measured in national surveys. Scale is unknowable but bounded: 304 million people live outside their birth country, and no census counts third culture kids. The findings, in short: in the paper's library more mixed groups were absorbed (9 of 28) than persisted as named groups (5), with 9 dispersed; leaving buys income, loses the language in about two generations and thins the label by the third, and what staying keeps is unmeasured; in-between kids today are a category with a vocabulary, not a people; and no recommendation in the playbook reaches grade A for mixed-heritage children. The paper begins where most of these lives begin to be recorded: on the first day at a new school.

1. Introduction

In 1971 a ten-year-old boy walks into a fifth-grade classroom at Punahou School in Honolulu. The teacher reads his full name from the roll, and titters cross the room. A classmate says, "I thought your name was Barry." Asked about his father, the boy says the family tribe is Luo, and another boy repeats the word in a loud hoot, like the sound of a monkey. A girl asks to touch his hair. A boy asks whether his father eats people. The only other Black child in the grade is a girl named Coretta, and the two of them keep apart, as if contact would sharpen the isolation. When a playground chant calls her his girlfriend, he shoves her away, and the shove buys him some room from the other children.^[248] In the frame this paper uses, Barack Obama on that day looks different from his classmates, sounds like them in the paper's reading (born in Honolulu and raised there to about 6; no card records his speech at Punahou), and thinks partly like them and partly not.

This paper is about people like that boy. It calls them in-between kids: people who differ from the locals in how they look, how they speak, and how they think, and so cannot blend in. The core case is a child of mixed heritage raised across cultures. Two comparison groups frame it: the plain third culture kid, who moves between countries with a parent's job but is not of mixed heritage, and the mixed child raised in one place. "In-between kid" is the paper's own working definition, not a term of art in any literature it reviews. Figure 1 shows the frame, with the four lives placed in their cells, one of them twice, before and after a move, and an arriving cohort beside them (Figure 1).

The paper follows four lives and one cohort from here to the conclusion, because a frame is not a story and the evidence on these children comes in pieces that only a life can hold together. Barack Obama, born in 1961 to a Kenyan father and a white American mother, was raised in Jakarta and Honolulu (Plate 1).^[248] Ariana Miyamoto, born in Sasebo in 1994 to a Japanese mother and an African American sailor, was crowned Miss Universe Japan in 2015.^[143,298] Robert Ho Tung, born in Hong Kong in 1862 to a Dutch merchant and a Chinese mother, was reputed the richest man in the colony, and is followed through his daughter Jean and the grandchildren.^[58,325,364] Kien Nguyen, born in Nha Trang in 1967 to an American father, left Vietnam in 1985; Bi Thi Loan, a Black American soldier's daughter, had not left when interviewed at about 39, in about 2005.^{[243,290,349]} Each of the four is on record because of what made them exceptional: a presidency, a crown, a fortune, a memoir and a wire feature (section 4.1). The fifth anchor is a cohort, not a person: the children of marriage migrants in Korea and Taiwan, many with Vietnamese mothers, about 152,000 children of international marriages among Korea's 210,838 multicultural students in 2026, 138,736 of them born in Korea, from families the Ministry of Education describes as tending to low incomes, and measured in national surveys and panels rather than in memoirs.^[54,103] Table 1.1 introduces them once. After it the text names the person, the age, and the year, and goes on.

Nobody knows how many in-between kids there are, but the number has bounds. The United Nations counted 304 million people living outside their country of birth in 2024, up from 154 million in 1990, while the share of the world's population rose only from 2.9 to 3.7 percent.^[341] No census anywhere counts third culture kids; the "220 million TCKs" that circulates online traces to a 2013 remark by the writer Pico Iyer and matches the UN's 2010 migrant stock, a match that is this paper's inference.^[109]

The paper asks eight questions, each a piece of the one question above. Who in-between kids are and how many there are (RQ1) is the ground, sections 2 to 4. How they develop (RQ2), what it costs them (RQ3), which advantages survive replication (RQ4), and what helps them accept their identity and use their edge (RQ5) are the one life, sections 5 to 9. How earlier in-between groups fared across generations (RQ6) is section 10. Whether they are becoming a new group (RQ7) is section 11, and whether the line that stays or the line that leaves thrives more (RQ8) is the family's decision, section 12.

Contributions. This paper makes five contributions.

1. A three-axis frame: Pollock and Van Reken's look and think, plus the paper's own sound axis. The third axis adds what the two-axis box cannot show: a move changes the cell. Miyamoto in Sasebo sounded local and in Arkansas at about 14 to 16 did not, and 69 percent of the Amerasians who had reached the United States by March 1992 arrived with Vietnamese and no English (section 2, Figure 1).
2. An evidence review in which each finding names the population it was measured on and whether it replicated, with counter-findings beside the findings they challenge (sections 3 to 9).
3. A coded library of 28 historical mixed groups, each scored on the same five variables, with cross-tabulations (section 10 and Appendix A).
4. Four tests for whether a category has become a people, applied to in-between kids today (section 11).
5. A playbook for the person, parents, schools, and employers, with an evidence grade on each recommendation (section 14).

The road is the spine. Sections 2 to 4 set the terms, the theories, and the literature. Sections 5 to 9 follow one life, stage by stage. Sections 10 and 11 follow three generations and ask whether anyone has made a people out of this position. Section 12 is the decision, and sections 13 to 15 are what follows from it. Before the evidence, the words, because the words were waiting for each of these children before they arrived.

2. Who they are: terms, the model, size

June 2015. Ariana Miyamoto, 21, has held the title of Miss Universe Japan for three months, and strangers online have said that a hāfu should not represent Japan. She tells the BBC why she uses their word. "If I say I am 'Japanese' the reply would be: 'No, you can't be'. People will not believe that. But if I say I am 'hafu', people agree".^[360] She was born and raised in Japan and speaks unaccented Japanese.^[221,360] In the frame she looks different, sounds local, and thinks local. The word she is allowed came before her. This section defines the words, builds the frame, and bounds the numbers.

2.1 Terms

"Third culture" came from fieldwork. In 1952 the sociologists Ruth and John Useem went to India for a year, returned about five years later to study Americans working there, and began to use "third culture" for the style of life of people who relate two societies to each other.^[346] The phrase is in print by 1963, in a paper about adults.^[344] Ruth Useem's writing on the children begins in the 1960s; no primary text of the first use of "third culture kid" was found, so the term was not "coined in 1963".^[346]

The standard definition is Pollock and Van Reken's: a third culture kid is "a person who has spent a significant part of his or her developmental years outside the parents' culture".^[264] Van Reken's wider term, cross-cultural kid (CCK), covers anyone raised among two or more cultural environments, mixed-heritage children raised in one place included.^[264] "Global nomad", Norma McCaig's 1984 label for the same group, is a third name.^[109]

2.2 Local words

Every host society has its own word for the mixed child, and each word carries its history and its period (Table 2.1). In Japan the word is hāfu, from the English "half". It replaced the older konketsuji ("mixed-blood child"), and it has rivals, daburu ("double") among them.^[293] Hapa, Hawaiian for mixed or part, was adopted on the US mainland over some Native Hawaiians' objection.^[26] Luk khrueng, "half-child", is Thailand's word.^[5] "Amerasian" is a legal category: a 1982 US law named five countries and told officers to "consider the physical appearance of the alien" as evidence of an American father.^[268] Kien Nguyen and Bi Thi Loan carried that word; one Amerasian in Vietnam said people called her "Lai", the word for mixed, instead of a name.^[198] Lai dai han names the children of Vietnamese mothers and South Korean soldiers.^[330] In Singapore "Eurasian" is a census category inside the box called "Others".^[301] Korea's older slur is twigi (section 6).^[15] The colonial words in the lower half of the table were legal categories first (section 10).

2.3 The model

Pollock and Van Reken sort a third culture kid in a given place by two yes-or-no questions: do you look like the locals, and do you think like them. The answers give four cells. The foreigner looks different and thinks differently. The hidden immigrant looks alike and thinks differently. The adopted child looks different and thinks alike. The mirror is alike on both.^[126,264] Life is simplest in the foreigner and mirror cells, where what others expect matches who the child is. Miyamoto at school sits in the adopted cell; Obama at Punahou sits between adopted and foreigner.

This paper adds a third question: do you sound like the locals? Three lines of evidence say speech is its own axis. First, code-switching breaks the two-axis box: the writer Nathania Gilson, a third culture kid asked on a Melbourne tram whether she was "from here", wrote that the identity box "is rendered useless alongside an event of code-switching".^[126] One writer has already relabelled the four cells by language, with no data.^[126] Second, accent steers social preference before any culture is learned: infants of five to six months look longer at a native-accented speaker (section 5.1).^[180] Third, speech and culture pull on each other: for eight Korean third culture kids, English was an edge and weak Korean a job barrier (section 14).^[297] Kazakh has a word for the cell: shala-Kazakh names a person with one Kazakh parent and, in online use, a Kazakh who does not speak Kazakh (section 10.7).^[186] One anchor shows the axis moving under a child, and one cohort shows it at scale (Figure 1). Miyamoto sounded local in Sasebo and, at about 14 to 16 in Arkansas, did not.^[298] Of the 13,060 Amerasians who had reached the United States by March 1992, 69 percent knew Vietnamese and no English at class placement: a cohort arriving in the does-not-sound-local square.^[343] Obama's Jakarta speech is not on record. The teasing he met there was over his pronunciation in religion-class Koran recitation, which is Arabic, and the one report on his Indonesian, a biographer's read only as a snippet, says his classmates took him for Indonesian.^[140,210] His Honolulu placement is the paper's inference from a birth and early childhood there; no card records his speech at Punahou. Kien Nguyen and Bi Thi Loan were raised in Vietnam, and so in the paper's reading sounded local; no card records Kien's English when he reached the United States at 17, so the figure places him only there.^[243,290]

The look axis needs its own line too, because looking different and looking the same each block a different thing. In 29 interviews in Japan, visibly mixed people could not pass as Japanese, while invisibly mixed people could pass but could not claim the mixed label.^[252] Miyamoto is the first case; Robert Ho Tung, who looked European in one historian's words and was allowed to claim the Chinese label, is the rarer exception (section 10.3).^[127,202] The paper found no peer-reviewed paper that adds a speech axis to the box; "none found in our searches" is the claim, not "none exists".^[126]

2.4 Size: what can and cannot be counted

The world's migrant stock doubled from 154 million in 1990 to 304 million in 2024 (Figure 2).^[341] Asian hosts grew faster: Korea's foreign-born population went from 18,499 to 1.8 million.^[341] These are outer bounds.

Mixed births in East Asia can be counted, within a definition (Figure 3). In Japan, babies with one Japanese and one foreign parent rose from 0.7 percent of births in 1987 to 2.2 percent in 2007, and were 15,511, or 2.2 percent, of 709,051 births in 2024.^[228] In Korea, births to multicultural families peaked at 22,908 in 2012 and were 13,416 in 2024, 5.6 percent of a shrinking total;^[66] in Taiwan, births with a new-immigrant parent fell from 13.4 percent in 2003 to 7.6 percent in 2018.^[225] Korea's share still rises because total births fell faster.

Intermarriage series tell the same story from the parents' side (Figure 4). Singapore's inter-ethnic marriages rose from 5.8 percent of marriages in 1984 to 19.2 percent in 2025.^[302] Japan's marriages with a foreign spouse went from 4,156 in 1965 to 44,701 in 2006, then fell to 18,420 in 2024.^[228] Korea's international marriages peaked at 13.4 percent of marriages in 2005 and, under a wider definition, were 9.6 percent in 2024.^[66] US newlyweds who married across race or ethnicity went from 3 percent in 1967 to 17 percent in 2015.^[204] The US "Two or More Races" count went from 2.4 percent in 2000 to 10.2 percent in 2020, a jump the Census Bureau attributes largely to a redesigned question.^[165] Vietnam sits on the sending side: Taiwan held 121,153 Vietnam-origin spouses in 2024, and one estimate put Vietnamese women married to Taiwanese or Korean men at up to 170,000 by 2015.^[182,229] Kien Nguyen's cohort was the earlier Vietnamese mixing, made by a war: 68,558 Amerasians and relatives had arrived in the United States by August 1993 (section 10.5).^[343]

Hong Kong shows what a count measures when the label hurts: 272 self-reported Eurasians in 1897, 42 in 1911, and 837 in 1931, while the Ho Tung family alone had twelve children (one adopted) and, through Clara, eighteen grandchildren.^[58,199] In 2021 the same city counted 66,732 residents of "Mixed" ethnicity (section 10.9).^[53]

International schools are the last bound, on the mobile side (Plate 2; Figure 5). ISC Research counts 2,584 English-medium international schools with 969,000 pupils in 2000, on its revised back-series, and 15,075 with 7.6 million pupils in 2025.^[111,158] In thirty years, by one relayed estimate, international school places went from 80 percent expatriate children to 80 percent local children.^[44] Enrolment is a loose ceiling on the mobile population.

2.5 A size is a definition

The size of a mixed group is a definition, not a count. One series puts the mestizo share at 23 percent in 1808, 43 percent in 1885, and 59 percent in 1921, with no source for the first figure,^[55,238] and Chance found colonial census sources that "frequently contradict one another".^[55] When Mexico's genome institute studied "the Mexican Mestizo population", it sampled 300 people who called themselves mestizo, so the label came before the result.^[300] Pew found 6.9 percent of US adults mixed by background once parents and grandparents count, against 2.1 percent by the census.^[262] Singapore's census has no mixed box at all, and a Eurasian sits in "Others".^[301] The words are the host society's. The next section gives the ideas the researchers brought, and asks how far any of them would have predicted Robert Ho Tung.

3. Ground knowledge: the identity theories

Hong Kong, 1928. Robert Ho Tung is 65, "generally believed to be the wealthiest man in Hong Kong".^[364] He is barred from the Hong Kong Club and was the first chairman of the Chinese Club that the barred men formed, in 1899 by Carroll's date.^[364] He is the only non-European allowed to live on the Peak, despite a 1904 law written to keep Chinese off it, and he has just built Ho Tung Gardens there.^[11] In Chicago that year the sociologist Robert Park writes that migration produces a person, "who may or may not be a mixed blood", living in two cultural groups. "The effect is to produce an unstable character," he writes; this is "the 'marginal man'".^[256] The man and the theory sit side by side, and the theory has no cell for him. This section gives the ideas the rest of the paper leans on, each with its evidence and its critics (Figure 6).

3.1 The marginal man

Park's essay had no sample. Milton Goldberg argued in 1941 that a marginal situation need not make a marginal personality, because a "marginal culture" shared with similar people makes them at home; Hong Kong's intermarried Eurasian families were one (section 10.3).^[128,199] A meta-analysis of 83 studies and 23,197 people found people engaged with both their heritage and the dominant culture better adjusted than people with only one, results that "clearly invalidate early sociological accounts" of biculturals as marginal.^[242] A review of third culture kid studies says they "are not confused by the multiplicity of their identity".^[96] A 2021 reanalysis finds the link much weaker (section 7.1).^[34]

3.2 Erikson and Marcia

Erik Erikson framed a life as eight stages, each a crisis to resolve; the fifth, "identity vs. role confusion", "becomes more prominent during adolescence".^[148] James Marcia turned the stage into four measurable statuses: diffusion, foreclosure, moratorium, achievement. A meta-analysis of 124 studies found that identity does mostly move forward in adolescence, but "large mean proportions of samples were not identity achieved by young adulthood".^[185]

3.3 Phinney

Jean Phinney carried Erikson and Marcia into ethnic identity, a person's sense of belonging to an ethnic group, and built the scale most studies still use. Across 184 correlational studies of people of colour in North America, ethnic identity and well-being correlated at r = .17, a weak link.^[304]

3.4 Berry

John Berry's acculturation model asks how a person engages with the heritage culture and with the larger society, and sorts the answers into four strategies: integration, assimilation, separation, and marginalisation. Integration, "being engaged in both their heritage culture and in the larger society", tends to go with better adaptation.^[27] Floyd Rudmin's critique says the model fails on logic and on evidence: two cultures in contact allow 16 types, not four, and "there is no robust evidence that biculturalism is most adaptive".^[281] Rudmin and the Nguyen and Benet-Martinez meta-analysis disagree, and the disagreement is live.^[242,281] Berry's boxes also assume a free choice; Ho Tung, a dual Chinese and British subject, chose Chinese, though China's rule by the father would have classed him European (section 11).^[230]

3.5 Biracial models

The biracial models came in the 1990s, built for mixed people. Maria Root, a clinical psychologist born in the Philippines to a Filipina mother and a white father, who heard slurs from classmates as a child, wrote a charter in 1993, printed in 1996, for feelings mixed people had long had but not organised.^[280] Her Bill of Rights claims the right "Not to justify my existence in this world", "To identify myself differently in different situations", and "To change my identity over my lifetime--and more than once".^[280] Rockquemore and Brunsma surveyed 177 Black-white biracial people and found four options: singular, border, protean, and transcendent identity.^[56,278] The field has judged its own theories: theory-building on multiracial identity has been "largely disconnected from empirical data".^[279] What the data show is movement: about half of the Black-white and Asian-white teenagers in one national sample, 57 and 53 percent, gave the same label six years later (section 5.3).^[98]

3.6 Bicultural identity integration and frame switching

Bicultural identity integration (BII) is the degree to which a person sees their two cultures as compatible or in conflict. Shown Chinese primes, high-BII people made more external attributions, the Asian pattern; low-BII people did the reverse.^[24] The underlying finding is cultural frame switching: bicultural people switch the frame they use to read events when shown culture-laden symbols, in priming experiments first run in Hong Kong.^[147] A 2025 study replicated the Hong Kong result 25 years on.^[253] In three studies of multiracial US adults, people who identified with different races in different settings reported lower well-being, less so if they tolerated ambiguity in the self; the data are correlational.^[288]

3.7 Why the question is never neutral

One more idea sits under every scene here. When Tajfel and colleagues split people into groups on a trivial basis, the subjects favoured their own group in handing out real rewards.^[320] Social identity theory grew from that result. "Where are you from?" is a categorisation, and categorisation alone is enough to move rewards. Each theory was tested on somebody. Section 4 asks whether anyone like the four people in Table 1.1 was ever in a sample.

4. Related work and gaps

Put the four anchors against the two literatures and each falls through. Multiracial research would count Obama and Miyamoto as subjects but not Jakarta or Arkansas; it measures mixed heritage in one place. Third culture kid research would count the moves but not the face; it has no race variable. The Ho Tung family predates both fields.^{[127,202,256]} The Vietnamese Amerasians have a small literature of their own, the nearest thing to a study of the core case.^[175,215]

The first literature is multiracial research. A review of 125 US articles (133 studies) from 1990 to 2009 found Black-white samples in 55 percent, a West Coast tilt, and non-US studies excluded by design.^[56] A 2024 systematic review of identity in non-Black-white mixed groups screened 1,846 articles and kept 18, all in the United States.^[371] A narrative review of well-being found maladjustment only in clinical samples.^[295] The second is third culture kid research, which is small and does not see race. A systematic review of adjustment during relocation screened 9,433 records and kept 14 studies, none longitudinal; a search of its full text finds no hit for "racism", "discrimination", "multiracial", or "mixed-race".^[164] A review of identity and belonging screened 509 articles and kept 6.^[96]

A few studies sit near the overlap. In London, 79 teenagers, half born abroad, gave the pattern "no gain without pain": higher open-mindedness and cultural empathy, lower emotional stability.^[91] Two 2023 studies split: in 489 adults, TCKs had weaker ethnic identity than non-TCKs but more resilience,^[94] while 399 students in the UAE did better with integrated rather than compartmentalised identities.^[94] A US master's thesis interviewed six non-white TCKs at one college, five of them biracial, and its author wrote that "a study dedicated to understanding the biracial identity of non-white TCKs would be beneficial".^[60]

The gap, stated as a search result: this paper found no study that measures mixed-heritage children raised across cultures as such. Title and abstract searches on 2026-10-26 for "multiracial third culture kids", "mixed-race third culture kid identity", and four related queries returned no paper whose focus is both; Appendix C lists them so the search can be repeated.^[60] "No study found" is the claim; "no study exists" is not. So nearly every finding in sections 5 to 9 was measured on one of the two comparison groups, and each finding says which. No study follows a child like these four from the first day of school to old age. The next four sections do it with what exists, one life stage at a time, and say at each stage which comparison group the evidence came from.

4.1 Whose lives this paper sees

The lives this paper can see are the ones somebody wrote down, and people get written down for being exceptional. The four anchors are a president, a national titleholder, a colony's reputed richest man and his family, and two Amerasians whose stories reached a memoir and a wire feature; most side stories are celebrities. The comparison literatures have the same tilt. Useem and Cottrell's adult TCKs came from families overseas "with the military, the State Department, business enterprises, religious organizations, and a miscellany of other sponsors", with "highly educated parents" and good overseas schools, and they volunteered.^[345] The TCK label, one sociologist of international schools writes, "is usually applied to those who are relatively economically privileged", by a literature "unreflexive of its own American-centric approach",^[323] and young people who fit the frame "do not want to be identified as being privileged".^[80] The wider term, cross-cultural kid, folds in children of immigrants, refugees and minorities (section 2.1),^[264] and the research under it is thin. The mixed-race qualitative literature says the same of itself: a leading British study drew its 65 interviewees from universities and colleges and wrote that they "may have been insulated from the harshest manifestations of racism";^[307] a Birmingham study recruited a class-balanced sample because such studies "often report high proportions of middle class participants";^[50] even a study of Japanese-Filipino children with poorer mothers landed on private-school and university students, and its author calls class "often overlooked".^[336] No TCK study the paper found samples poor migrant-worker families.

What that does to every reading: the advantages of section 7 were measured on people with the schooling and the networks that turn a position into a seat; the costs of section 6 were measured on people a study could reach, so the worst-off are under-counted on both sides; and the stay-or-leave verdicts of section 12 lean on leavers who had somewhere to go. The paper's answer is a fifth anchor that no one wrote a memoir about: the children of marriage migrants in Korea and Taiwan, counted in national surveys and panels (Table 1.1; sections 5.4, 6.4 and 12). Where a finding below comes from a comfortable sample, the text says so; where the ordinary cohort gives a different number, it is printed beside it.

5. Development across the life course

Jakarta, about 1970. A nine-year-old boy waits in the library of the US embassy, where his mother works, and reads Life magazine. He finds a man whose skin has been chemically lightened, and learns that thousands of Black Americans had bought the treatment "in response to advertisements that promised happiness as a white person". "I felt my face and neck get hot." He says nothing to his mother.^[248] Afterwards he began to notice that "there was nobody like me in the Sears, Roebuck Christmas catalog", and kept it to himself so as not to show her that her lessons had failed.^[248] This is Barack Obama a year before the Punahou classroom of section 1, the earliest moment any anchor records about the look axis.

This section walks from the first months of life to old age. Figure 7 is the master figure: beneath the comparison-group evidence, an anchor band pins each scene in the four lives at the age it happened, and the empty stretches are gaps in the record (Figure 7). Children react to look and speak years before they can think about either.

5.1 Early childhood

No anchor has a scene before school age that touches difference; the earliest, Kien Nguyen's fifth birthday in 1972, records a roasting pig and nothing about his blond hair.^[243] The evidence below is on infants in general.

Newborns show no preference for faces of their own group. Three-month-olds do, so the preference is learned from the faces around the crib.^[174]

Sound arrives next. At five to six months, 18 of 24 infants looked longer at a speaker with the local accent; at five, eight children "tended to choose" the local-accented child as a friend. The cells are tiny.^[180] The child who sounds foreign loses friends before anyone asks where she is from.

The doll studies measure the look axis. In 1947 the Clarks tested 253 Black children aged three to seven; the share who said the brown doll "looks like you" rose from 36 percent at three to 87 percent at seven,^[68] while the share who chose the white doll to play with peaked at four, at 76 percent, and fell to 60 percent by seven (Figure 8).^[68] One caveat carries: a forced choice between two dolls is not self-esteem, and a 1995 re-test found Black preference more likely where children showed any.^[46]

Sorting people by race comes later, and whether race is a category at all depends on place. Children can sort faces into racial groups between six and eight.^[258] In 136 children, those in Massachusetts believed more firmly than those in Hawaii that race cannot change, and that belief explained their higher stereotyping.^[260] The review that collects this work adds that multiracial children have been "almost entirely excluded" from it.^[258]

5.2 School age

All four anchors are on record at school, and all four were marked there. Obama entered first grade in Jakarta in 1968 under his stepfather's surname; later, at the Besuki school (Plate 4), a classmate remembers him "laughed at because of his funny pronunciation" during Koran readings, which are recited in Arabic, so the teasing says nothing about his Indonesian;^[140] a biographer who interviewed classmates reports, in a passage read only as a snippet, that by the end of his first year they took him for Indonesian.^[210] His mother (Plate 5) could not afford the International School, kept him in Indonesian schools, and woke him at four, five days a week, to teach English before school: "I was an American, she decided, and my true life lay elsewhere".^[248] At ten he landed in Honolulu to live with his grandparents, "I realized that I was to live with strangers", and walked into the Punahou classroom of section 1.^[248] That winter he told the boys that his father, due from Kenya, was a prince.^[248]

Ariana Miyamoto, at school in Sasebo (Plate 6), was called "kurombo", a slur for a dark-skinned person. When the teacher told the class to hold hands, children said, "Don't touch me", afraid her colour would rub off; a board eraser was thrown at her.^[92,139,298] Most mixed children in Sasebo, she says, lived inside the US base; she lived outside it, where "everyone around me was a so-called 'pure Japanese'".^[298] Her coping: "In the end I think I made myself stop thinking about the words".^[139,298]

Kien Nguyen was seven when Saigon fell in 1975. "Because her children were Amerasian", his mother was not allowed to work, and the family took refuge with relatives who, in one biographical entry's words, "abused them no less than did the government".^[243] Bi Thi Loan, given to her grandmother at two months because her mother's husband disliked the child, was pulling a plough at 12, and an uncle who had fought for the Viet Cong beat her head with a stick each time she failed.^[290] Alex Oi, a Mexican-Japanese boy filmed at a Nagoya school in 2013, showed "physical symptoms of stress due to the teasing he receives from his classmates for being hafu".^[321]

The school decides what the same child is. Among Japanese-Indonesian children, hāfu pupils were "deemed inferior" at the Japanese school in Indonesia and "superior" at the Indonesian one; the abstract gives no sample size.^[322] Parents in Japan's 2016 foreign-resident survey named teasing "for name or nationality" as their top school worry, at 30.1 percent, a worry rather than a rate.^[150] No survey measures bullying of hāfu children; Miyamoto's eraser and Alex's stomach are cases.^[82]

Kikokushijo, the returnee children Japan has counted since 1977, are the one in-between group it does count: 10,000 to 13,000 a year from 1985 to 2000.^[171] Their public image swung from "educational orphans" to "symbols of internationalization", and many hid their years abroad.^[171]

In Pew's 2015 survey of 1,555 multiracial adults, 15 percent had been teased as children for being mixed, 30 percent among Asian-white backgrounds.^[262] The largest school survey is Add Health, a US study that followed teenagers from 1994 into adulthood. In its first wave, 7.4 percent of 83,135 students chose more than one race; their odds of smoking were about twice those of single-race peers for Black and Asian mixed teens and 1.2 times for white, yet "most mixed-race adolescents are at low risk".^[337] Fryer and colleagues found Black-white teens in the middle on most measures and "stark outliers" on risky behaviour.^[114]

Korea now counts its mixed children in school. Students from multicultural families, the government's term for children with a foreign-born parent, rose from 46,954 (0.70 percent) in 2012 to 168,645 (3.19 percent) in 2022. By 2026 they were 210,838 (4.31 percent), while the school population fell (Figure 9).^[173]

5.3 Adolescence

Obama at 12 or 13 stopped telling people that his mother was white, because he began to suspect that doing so was ingratiating himself to whites.^[248] At 14, on a 1975 roll call, he told a teacher, "Just call me Barry".^[361] Of himself at about 15 he wrote, "I was trying to raise myself to be a black man in America, and beyond the given of my appearance, no one around me seemed to know exactly what that meant".^[248] One school friend, also mixed, remembers it differently: "He made everything out like it was all racial"; and Hawaii, where everyone was some kind of minority, is a weaker test of the look axis than Sasebo.^[361]

Miyamoto, in her mid-teens, went to live with her father in Jacksonville, Arkansas. In Japan people had said, "You're hafu, so why can't you speak it?"; in Arkansas friends said of course she could not, having grown up in Japan.^[298] The same missing language cost her status at home and nothing abroad. Loan at 15 begged her mother to take her back, and her stepfather refused.^[290] For the fifth anchor, adolescence brings a share of the mother's Korean reading and writing, usually behind the father: among ten Korean-born teenagers with immigrant mothers, four of them Vietnamese, language brokering for the mother was "an inevitable part of family assistance for all the participants", since eight mothers spoke Korean well and two read or wrote it well, though the father usually handled the official paperwork and the teens called their share minor.^[160]

In adolescence the label moves. Of Add Health's 20,743 Wave I students, aged 12 to 18, 5,548 (27 percent) were lost by Wave III, at 18 to 25; Doyle and Kao analyse the non-Hispanic remainder: single-race white and Black youth kept their label 96.3 percent of the time, Black-white youth 57.0 percent, Asian-white youth 53.1 percent, and Native American-white youth 18.6 percent (Figure 10).^[98] About 12 percent of the same survey's teenagers gave different answers to near-identical questions about their race.^[98] A looser count, from a national sample of 14- to 18-year-olds re-measured five years later, in which adding or dropping any one category counts as a switch, found youth who ever called themselves multiracial four times as likely to switch as to keep a consistent multiracial label;^[144] it measures a different thing from Doyle and Kao's shares, and the two numbers do not combine. Interviewers recorded 45 percent of the Asian-white teenagers as white: the look axis measured by a stranger with a clipboard.^[98]

The choice follows how the teenager thinks others read her face. Among 177 Black-white adults, identity followed "an actor's assumption of how others perceive his or her appearance", not skin colour.^[278] In Japan the constraint runs both ways, as the 29 interviews of section 2.3 showed.^[252] Fusae Miyako, in the film Hafu, is the invisible case none of the four anchors shows: she grew up in Kobe believing she was entirely Japanese and learned in her teens that a parent was Korean, a hidden immigrant.^[321] In Pew's survey, 61 percent of adults with a mixed background do not call themselves multiracial.^[262] A 27-year-old respondent: "Sometimes I just get tired of explaining who I am".^[262] The Amerasian cohort's teenagers were measured only as adults, on opposite sides of the Pacific; section 12 holds that comparison.^[198]

5.4 Young adulthood

Kien Nguyen left Vietnam in 1985 at 17, after several failed attempts, under the Orderly Departure Program, with his mother, brother and sister: "I was the first kid to get on the plane." The family spent 22 days in Bangkok and three months in a camp on Bataan.^[349] Miyamoto was crowned Miss Universe Japan on 12 March 2015, at 20, the first biracial winner, and placed in the top ten at the world final.^[92,298] Jean Gittins, born Ho Tung, at about 20 left university to marry Billy (Bill) Gittins in March 1929, over her father's objection to any daughter "marrying into an Eurasian family which did not subscribe to his view and adopt Chinese nationality".^[127]

Obama's twenties are the one stretch between 18 and 30 that any anchor records in detail. At Occidental College, about 1980, a mixed classmate who refused the Black label made him "question my own racial credentials all over again".^[248] The same year friends began calling him Barack; his father had taken "Barry" in 1959 to fit in, and the son took the full name back as "an assertion of being comfortable with the fact that I was different".^[361] A friend said, "I thought your name was Barry", the sentence that had reached him at ten.^[248] At Columbia, at 20 to 22, he understood himself as a black American, but "that understanding remained unanchored to place".^[248]

Most in-between young adults get a move "home" instead. In Useem and Cottrell's survey of 680 adult American TCKs, read here in the authors' own 1993 newsletter articles, 81 percent held at least a bachelor's degree against 21 percent of Americans, and in a later article nine in ten of nearly 700 said they were "out of synch" with their age group for life; the sample volunteered, and came from what the authors call "elite communities", so section 7.8 weighs it.^[76-77,345]

The fifth anchor's young adulthood is measured in two national series, and it is ordinary. In Korea, 61.9 percent of multicultural-family youth were in higher education in 2024 against 74.9 percent of all Koreans, a gap of 13 points, down from 31 in 2021; part of the jump may be the survey's older age mix.^[226] In Taiwan, adult children of new immigrants aged 18 to 40 showed no significant gap with the general population in college attendance or income in one self-selected survey, and family income decided who climbed;^[333] the "ethnic dividends" of a Southeast Asian mother reached mostly those with a university place.^[190] Harry Chan and Sean Hsu, 19, in Taipei (section 8), are two of them.

Reverse culture shock, the distress of returning to a country that is supposed to be home, has only small studies behind it: 785 Turkish returnee students, 18 percent with clinical depression against 11 percent of classmates who never left;^[120] four Japanese returnees "polarized toward one language and culture" as teenagers;^[170] and a life-history study in which returnees "scrape off" the overseas self or "graft" Japanese patterns on top of it.^[171] Miyamoto's return from Arkansas is the anchor case, unmeasured.^[298]

Adult TCK samples show one pattern with two faces: weaker ethnic identity, more resilience and less comfort in close relationships in 489 adults;^[94] cultural homelessness, the feeling of belonging to no culture, with lower self-esteem in 475;^[145] normal maturing of personality and well-being in 700.^[1]

5.5 Adulthood

Jean Gittins at about 32, in 1941, sent her children to Melbourne. Elizabeth was ten and John five. The evacuation order of July 1940 had applied, she wrote, "only to those of 'pure' British descent", and her father's "claim to Chinese nationality would automatically rule out the possibility of my entry into Australia". The children went with family friends on their own passports: "Gittins was an English name and they looked foreign enough".^[127] Kien Nguyen at 31 took his dental degree in 1998; nightmares followed, a therapist told him to write everything down, and the diary became a memoir. By 38 he had given up his solo practice for writing: "I'm a writer, I love what I do, I find this is my calling".^[349] Loan at about 39 lived in a house inside a Saigon cemetery with her husband, four children and in-laws, ten people on about $38 a month, and was going back to school after fifth grade.^[290] Obama in his mid-twenties, organising in Chicago (Plate 7), found that church women traded stories with him, and the stories "helped me bind my world together".^[248] At 46, in 2008, he opened a speech on race with "I am the son of a black man from Kenya and a white woman from Kansas".^[247] Miyamoto at 29 had two sons and lived in the United States.^[298]

What adults pass on is mostly language. In 1,899 families with a minority-language parent, every child spoke the majority language, and the "one parent, one language" rule was neither necessary nor sufficient for the minority one.^[83] In the United States, immigrant languages are nearly gone by the third generation, Asian ones faster than Spanish (section 9).^[284] Mixed parents also talk more than their parents did: 47 percent had discussed the mixed background with their children, against 32 percent who recalled their own parents doing so.^[262] Miyamoto's sons are the one next generation on record in a parent's words: on American forms she ticks Black and Asian for them, and at a kindergarten open day in Japan one was greeted with "a foreigner has come".^[298]

For the Amerasians who reached the United States as young adults, adulthood meant low-paid work, and Kien, a dentist from a wealthy family, is not typical. In a 1994 GAO sample of 100, 60 percent were employed;^[343] of 13,060 Amerasians processed by March 1992, 7 percent had no schooling and 65 percent some primary school.^[343]

5.6 Old age

The only anchor old age on record is a rich man's. Robert Ho Tung at 72, in about 1935, received the US vice-president at Ho Tung Gardens, in Chinese dress (Plate 8). He spent the Japanese occupation mostly in Macau, and his loyalties came under question afterwards.^[199] His daughter Irene lived in his household for his last ten years.^[58] He was knighted a second time in 1955, at 92, and died in 1956.^[11] No card follows the other anchors further.

No study follows mixed-heritage TCKs into old age. Figure 7 goes empty after 45 and after 1956 for lack of data, not of people. The nearest evidence comes from adult TCKs. The Useem and Cottrell respondents ran to age 87, and a majority, "including those over 65", still reported re-entry difficulty.^[78,345] Four lives, one pattern: at every age somebody asked what they were. Section 6 prices the question.

6. Challenges

Tokyo, March 2015. Ariana Miyamoto, 20, has held the crown for a week. A post in Japanese circulates online: "Even though she is Miss Universe Japan, her face is foreign no matter how you look at it!" Others write that Japan should be represented by "a pure blood". She answers: "But I am a Japanese".^[92,221] Two months later she tells AFP she had expected it, and that it hurt. "I'm Japanese." "I stand up and bow when I answer the phone".^[143] She speaks unaccented Japanese; what fails the test is the face.^[221] The cost of that test lands on the child, and from both sides.

6.1 "What are you?": invalidation and forced choice

Identity invalidation means that others deny the identity a person claims. Miyamoto's "No, you can't be" is its plainest form.^[360] Obama describes the subtler one: people who learn his background make "split-second adjustments", and "They no longer know who I am".^[248] Forced choice means a form, or a person, that allows one box. Invalidation has three measurable forms, tested in 497 multiracial adults: behaviour ("you do not act Asian"), phenotype ("you do not look Black"), and discrimination aimed at an identity the person does not claim.^[110] In the lab, denial raised self-reported stress in 126 bicultural and 119 biracial participants, and the bicultural group recovered more slowly in cortisol, the stress hormone; that design is causal.^[6] In 317 multiracial adults, frequent forced-choice dilemmas predicted more depressive symptoms through lost identity autonomy. Identity autonomy means feeling free to identify as you like.^[287]

The harm is specific. In 193 multiracial college students, discrimination aimed at being mixed predicted distress (beta = 0.26), while general discrimination did not.^[93] For a child like Miyamoto, the measured harm is in the sentence "you cannot be".

6.2 Mental health: the findings disagree, and the paper says why

Read the large surveys and mixed teenagers are at risk. Read the matched studies and they are not (Figure 11).

The risk side begins with Udry's 2003 Add Health paper. Teenagers who named more than one race had higher odds on most items in the printed tables, which carry no controls, yet "most mixed-race adolescents are at low risk".^[337] The authors also re-ran the analysis with controls for age, sex, vocabulary score, grades, family structure and parents' education, and report that very few odds ratios crossed below 1, though many changed in significance; the controlled tables were not printed, and income was not among the controls.^[337] For White-Asian teenagers, the group nearest this paper's case, the excess was modest and uneven: 1.50 for fair or poor health and 1.58 for sleep problems against white teenagers, with several items not significant.^[337] In four Seattle middle schools in 1997, 50.1 percent of the 454 multiracial pupils had ever smoked, against 32.8 percent of white pupils; 55 percent of the multiracial pupils were low-income against 11 percent of the white ones, and the higher risk held in models with age, gender and low income controlled.^[64] A systematic review of 22 studies from 2016 to 2022 found that multiracial people "tended to have worse mental health, with notable exceptions", and reported no pooled effect.^[250]

The no-gap side uses the same or better data. Cooney and Radina, also in Add Health, found differences on fewer than half the measures.^[75] Shih and Sanchez's review (section 4) found maladjustment only in clinical samples.^[295] In the UK Millennium Cohort of 16,330 children, mixed three-year-olds had fewer behaviour problems than non-mixed children of the same minority group, 8.9 against 13.1 for Pakistani-mixed children on the Strengths and Difficulties Questionnaire, though some mixed groups' trajectories rose after three.^[375] In a matched sample of 309 multiethnic and 508 monoethnic US teenagers, the multiethnic group had slightly fewer symptoms,^[176] and across 47,491 teenagers in two national surveys, biracial substance use often sat between the two parent groups.^[328]

One case sits beside these rates and is not one of them. Miyamoto entered the pageant after a close friend, half Japanese and half white, died by suicide; the BBC wrote that he killed himself "in part" because he could not face being treated as an outsider, and in her words, "Japanese people didn't accept him and because he didn't speak English he wasn't like a foreigner".^[139,360] It is one account of one death.

The competing explanation is not mixedness but the household and the school around the child: family structure, father absence, poverty, context. In CILS, in 1998 the largest study of its kind, intact families did "much better" and high-conflict broken ones worst, whatever the nationality;^[282] the New Orleans authors tie Vietnamese delinquency to families split by war and resettlement;^[374] and inside the multiracial group, household dysfunction predicts depression and anxiety, with standardised coefficients of 0.50 to 0.61 in 1,231 teenagers.^[357] The Amerasian record is soaked in both: Loan given away at two months, Kien's family refused work (sections 5 and 6.4); and in the Amerasian case the household hardship was itself imposed for the children's mixedness, Kien's mother barred from work because her children were Amerasian, so the two readings overlap.^[243] The studies above control for these unevenly. Udry reports that with family structure and parents' education held constant very few odds ratios fell below 1; the controlled tables were not printed, income was not among the controls, and his White-Asian teenagers had more college-educated parents than white teenagers and fewer two-parent homes, 73.5 against 78.4 percent.^[337] The Seattle study held income but not family structure.^[64] The two direct tests of the poverty reading disagree: in the UK cohort, adjusting for household income narrowed the mixed children's advantage, with family structure not held;^[375] in the United States the multiracial excess in depression and drug use appeared only in high-income families.^[134] Context changes the gap too: in poor inner-city neighbourhoods African American teenagers took fewer risks than white or mixed-race ones, and in one Alaskan city mixed-race students carried fewer protective assets.^[40,119] No study found matches mixed-heritage teenagers with single-race teenagers on father absence.

The paper's reading is the authors' synthesis, not a tested result: where a gap shows, it tracks invalidation and discrimination more than mixed heritage itself, and the household explains part of what remains. Against the household reading: the odds stayed above 1 in Udry's unprinted controlled re-run, and the US excess appears in high-income families where poverty cannot carry it. For it: the UK gap moved with income, household dysfunction predicts symptoms inside the multiracial group, and no study controls for household and discrimination at once. The studies that find a gap are large and cross-sectional; the studies that match on background or follow a cohort find no gap or a small one in the other direction.^[375] The one study that separates the two kinds of discrimination (section 6.1) points the same way.^[93]

6.3 Belonging, restlessness, grief

Obama and Coretta kept apart in the Punahou playground "as if direct contact would only remind us more keenly of our isolation".^[248] The TCK literature measures that cost under another name: a thin sense of home. In 19 interviews with third culture adults, 6 said they belonged to no culture, and 15 called the life mainly positive; one said, "I feel like a tourist when I go to my home country".^[232] The London teenagers of section 4 showed more open-mindedness and empathy and less emotional stability;^[91] the 489 adults, more resilience and less comfort with closeness.^[94] Nearly half of 696 adult American TCKs "do not feel central to any group".^[78] Obama at Columbia named it from inside: a self "unanchored to place".^[248]

Side story Pico Iyer, Oxford, California and Japan, 1957 to 2013 A plain third culture kid who calls belonging nowhere "a terrific emancipation".

Pico Iyer (Plate 9) was born in Oxford in 1957 to Indian parents; the family moved to California when he was seven, and he was schooled in England and the United States. He is the comparison group, a mobile child without mixed heritage, and his account of belonging is a sunny one. Of India he told a 2013 audience, "I've never lived one day of my life there", and said he could not speak "even one word of its more than 22,000 dialects". In England he was "the only kid in all my classes who didn't begin to look like the classic English heroes represented in our textbooks". In the United States he carried a "permanent alien" card, "this funny little pink card with green lines running through my face". In Japan, where he had lived as much as he could for the previous 25 years, he was on a tourist visa. He calls being part of many places and wholly of none "a terrific emancipation".^[246] Of the 19 interviewees above, 6 belonged to no culture and 15 called the life mainly positive; Iyer's account sits in both answers, and he has written books from the position.

The loss has a name in the literature but no count. Forty-three TCKs described losses of persons, places, pets, and possessions, and the loss of a self. Others did not recognise the grief, so the literature calls it disenfranchised grief.^[125] Ruth Van Reken, co-author of the field's main book, wrote down her 1958 departure from Nigeria as the child she had been: "It's sort of like a death, to lose your whole world in one moment".^[348] The child literature behind all this is 14 studies, none longitudinal.^[164]

Two claims need deflating. Code-switching between cultural frames is a measured skill (section 3.6), and no measured "code-switching fatigue" was found.^[24,147] The "migratory instinct" is a book term with no survey behind it;^[264] Cottrell's own 604-person data say most respondents "experience some stress due to conflicting desires for stability and mobility", a tension, not a need to keep moving.^[77]

6.4 Rejected from both sides: Korea, Japan, Vietnam, Thailand

Korea first (Plate 10), through the one visitor who learned the word. Hines Ward, born in Seoul to a Korean mother and an African American soldier and raised in Atlanta, went back in 2006 and learned that mixed-race Koreans were called "twigi" and that the army took only "pure-blood Koreans".^[15] Korea measures the cost in national surveys. In the 2015 Korea Youth Risk Behavior Survey, 795 multicultural teenagers met school violence at 10 percent, against 1.9 percent of their peers.^[192] The three-yearly National Survey of Multicultural Families shows discrimination among children rising from 6.9 percent (2015) to 9.2 percent (2018), then falling to 2.1 percent (2021) (Figure 12); the drop may reflect a changed question or sample.^[227] The higher-education gap in 2021 was 31 points: 40.5 percent against 71.5 percent.^[227] By the 2024 round it was 13: 61.9 percent of multicultural-family youth in higher education against 74.9, a jump the survey's older age mix may partly explain.^[226] High-school dropout rose from 1.36 percent (2020) to 2.22 percent (2023), above the national 2.0,^[193] and in 2021 it ran above the national rate at every school level, which the Ministry of Education blamed on income.^[103] These are the fifth anchor's families: about one household in three earned under 3 million won a month in 2024, down from half in 2021, with no national comparison published, and the Ministry of Education says they tend to low incomes.^[103,226] Inside the group the mother's origin and schooling matter more than the mix. Chronic bullying lowered life satisfaction cumulatively for children of Southeast Asian mothers and for a shorter time for those of Chinese or Japanese mothers, in a panel of 1,375;^[255] the mother's education predicted teenage depression where her birthplace did not, in 1,147;^[121] and among 586 fifth-graders with Vietnamese mothers, life satisfaction tracked self-esteem, stress and parental support rather than the family's rated wealth.^[63]

Taiwan counts the same cohort as children of "new immigrants" (section 2.4).^[225] In a 10 percent sample of 162,219 tenth graders from 2012 to 2018, children of mainland Chinese mothers matched native pupils on grades, class rank and school type; children of Thai mothers did best; children of Vietnamese and Indonesian mothers did worst, and their families were poorer and more often single-parent, which the author reads as the cause.^[332] A 2026 meta-analysis of 77 studies finds the gap fell from medium to small over 25 years and is negligible in junior and senior high.^[73] One Vietnamese mother in a 2022 study recorded another child telling her son to "go back to Vietnam"; two others had been mocked in public as having been "bought".^[57]

Japan has no mixed category, so its baseline is people read as foreign. The Ministry of Justice surveyed 4,252 foreign residents in 2016: 39.3 percent of house-hunters and 25.0 percent of job-seekers had been refused for being foreign, and 29.8 percent had been insulted.^[150] Hāfu bullying stays case literature (section 5.2): one girl asked her foreign father to stay away from school.^[82] Scholars call the frame behind Miyamoto's post, and behind the reply she quoted in section 2, the homogeneity discourse: the story that Japan has one people.^[178] Inside the hāfu category the test is uneven: a mixed-race playwright says there would have been "less brouhaha" had Miyamoto been part white, an opinion no Japanese survey has tested.^[143] Japan's ordinary count is children who need Japanese instruction: 69,123 in 2023, double the 2008 figure, with foreign nationals taking most of the growth and Japanese-national pupils, dual nationals and mixed children among them, more than doubling from 4,895 to 11,405.^[222] They drop out of high school at 8.5 percent against 1.1, go on to college at 46.6 against 75.0 percent, and 38.6 percent of those who take jobs hold non-regular ones against 3.1; the group mixes language, poverty and recent arrival, and is not a mixed-heritage count.^[222] In the one cohort study of mixed parentage, children with one Japanese and one immigrant parent attended lower-ranked high schools than both native and immigrant children, East Asian parentage the exception; an immigrant mother brought more disadvantage than an immigrant father, a gap socioeconomic factors largely explain.^[159]

Vietnam's in-between children were the enemy's. Amerasians, the children of American fathers, were called "Lai", the word for mixed, instead of a name. One school refused a girl because "we don't know her roots".^[198] Loan put the reason in one line: "They see Amerasians as children of the enemy".^[290] In a 1994 GAO sample of 100, 71 percent reported harsh discrimination in Vietnam.^[343] Thu Thuy Le, born in 1971 to a US soldier, stopped going to school at nine and "can only write my name".^[290] The 1987 Amerasian Homecoming Act opened a door (section 10.5); about 25,000 Amerasians and 60,000 relatives had left by 2009, within a consensus range of 21,000 to 30,000 and 55,000 to 70,000, and one second-hand estimate puts suicide attempts at 14 percent.^[175] The children of Korean soldiers carry their own word, lai dai han; Tran Van Ty "was beaten up regularly on his way to school for having a South Korean father".^[330]

Outside Asia the ordinary mixed child is not a success story on average either. In England in 2025/26, mixed pupils on free school meals scored 37.4 Attainment 8 points against 32.0 for poor White pupils, but the Mixed White and Black Caribbean group sat at 32.3, level with poor White British pupils, and reached university at 21.8 percent against 34.1 for Mixed White and Asian pupils and 16.9 for poor White British pupils; where the White and Black Caribbean group trails is outside free school meals, 45.0 against 48.8 Attainment 8 points and 38.9 against 42.1 percent at university. The mixed advantage of section 10.9 is a White-and-Asian result.^[87,89] The US adult children of immigrants, mixed or not, in 2010 to 2012 data, are ordinary: a median household income level with the national figure, 36 percent with a degree, 10 percent without a high-school diploma.^[263]

Thailand is the region's one acceptance story, and it is thin: a 1996 AP feature says luk khrueng went from "society's outcasts" to "fashionable and hip" in twenty years, and ties the change to class.^[362] The same position that draws the question also sits between two markets. Section 7 asks what that is worth, and starts with the man who made the most of it.

7. Advantages, with replication status

Hong Kong, 1880, by one account. Robert Ho Tung, 17, raised as Chinese by a mother whose Dutch partner had left when his business failed, joins Jardine Matheson, the largest British trading house in the colony, in its compradore department: the bilingual broker's post between a British firm and its Chinese market.^[325] By thirty he is a millionaire; later he was "generally believed to be the wealthiest man in Hong Kong".^[364] He looked European, in one historian's words, sounded Chinese, and had chosen to live as Chinese; the position between two markets was the one the colony paid for.^[127,202] Nine edges have been claimed for people like him. For each, the text says which population the study measured and whether the result survived a second look, and ends in a one-line verdict. Figure 13 ledgers the four edges with meta-analyses (Figure 13).

7.1 Bicultural adjustment

Adjustment means psychological well-being plus getting along in daily life. The 2013 meta-analysis of section 3.1 found "a significant, strong, and positive association" between the two.^[242] In 2021 Bierwiaczonek and Kunst reanalysed the same 83 studies and added 19 longitudinal ones with 6,791 people: the link "is much weaker than previously assumed", and the longitudinal evidence "is minuscule at best".^[34] A 2023 follow-up put the true effect of integration at negative "almost 30% of the time".^[35] The counter-finding: multiracial adults who changed their racial label across settings reported lower well-being, unless they tolerated inconsistency in the self.^[288]

Side story Naomi Osaka, New York and Japan, 2018 to 2024 A present-day Japanese case beside Miyamoto: told her "black card" was revoked, she called her two heritages her strength.

Naomi Osaka was born in Osaka to a Japanese mother and a Haitian father and raised in the United States. At the 2018 US Open (Plate 12) she told reporters that she grew up in a Haitian household in New York, living with her grandmother. Her mother told the Wall Street Journal that year that the girls "have always felt Japanese". Years after she chose to play for Japan, she was told online that her "black card" had been revoked; she answered that people "really don't know the difference between nationality and race". In 2024 she told USA TODAY Sports that her competitiveness comes from her Haitian side, and that "Being of Japanese and Haitian heritage is my source of strength".^[112] Set beside the counter-finding above, she is a person who kept two labels at once and was asked by the public, not by herself, to drop one. The measured harm in section 6.1 goes with being denied the choice. She refused the denial in public, as Miyamoto did in 2015, and no study has measured what that costs or saves.

Verdict: direction positive, size contested, cause unproven.

7.2 Integrative complexity

Integrative complexity is the habit of weighing several viewpoints and linking them. Tadmor and Tetlock proposed in 2006 that life under two cultures forces the habit, with no data.^[319] Bicultural professionals who had lived abroad "achieved higher promotion rates and more positive reputations", with complexity as the mediator, in observational work from one research network.^[318]

Verdict: one lab network, observational, no independent replication found.

7.3 Creativity

Five studies in 2009 found that living abroad, but not travelling abroad, went with creative performance; these are adults who moved, not children born in between.^[207]

One study reaches mixed-heritage people, and it cuts both ways. Writing about their racial identity raised multiracial people's creativity and did nothing for monoracial people; but reminding anyone of their own multiple identities worked too, so the effect is a mind-set, not ancestry.^[117] An independent lab, n = 596 and 174, found multilingualism and deep immersion abroad the most robust predictors.^[115]

Verdict: one of the two best-supported edges, with brokerage (7.7), but it is a mind-set and a context effect, not a trait of mixed people; the one study on mixed people shows the mind-set can be switched on in anyone.

7.4 Cultural intelligence

Cultural intelligence (CQ) is the ability to work well across cultures, built by Earley and Ang from three facets and later four.^[102] A 70-study meta-analysis finds "incremental validity" beyond personality;^[291] a review of 15 papers finds the link from international experience to CQ inconsistent.^[223]

Verdict: a real and measurable construct, mostly self-report, partly downstream of personality; growing up abroad does not reliably raise it.

7.5 The bilingual cognitive advantage

This edge shrank most, and no study of it measured mixed or mobile children. Executive function is the mental control that ignores distraction, switches tasks and holds things in mind. In 2004 Bialystok and colleagues found that bilingual adults paid smaller costs on a test of ignoring a misleading cue.^[32] Then the counts came in. Across 152 adult studies, "No evidence for a bilingual advantage remained after correcting for bias",^[200] and about 23,000 children gave a standardised difference of .08 before and -.04 after bias adjustment.^[206] The proponents now argue for cognitive reserve in old age, having conceded in 2017 that young-adult studies "largely fail to show these effects";^[33] on dementia the meta-analyses split.^[9,235]

Verdict: the executive-function advantage is small or absent and should not be listed as a proven edge;^[200,206] the dementia question is split; the communication benefit is not in dispute.

7.6 Less essentialism, less stereotype threat

Essentialism is the belief that race is a fixed inner essence. Stereotype threat is the performance drop that follows a reminder of a negative stereotype. In four studies, Asian-white and Black-white multiracial people "were less susceptible to racial stereotypes than monoracial individuals".^[296] Teaching that race is socially constructed gave anyone the same buffer.^[296] The developmental side is in section 5.1: Massachusetts children essentialising more than children in Hawaii, a two-place comparison.^[260]

Verdict: consistent small studies from two lab networks, no independent large replication.

7.7 The bridge: network brokerage

Brokerage is the position of a person whose contacts do not know each other. Burt called the gaps "structural holes". Among managers at one American firm, pay, good evaluations, promotions and good ideas were "disproportionately in the hands of people whose networks span structural holes".^[47] Burt says nothing about heritage; the link is this paper's inference.

The closest measurement is three studies of 1,196 Canadian employees.^[108] Among 819 health-care workers, those with more cultural identities interpreted more often and scored higher on adaptability;^[108] among 31 hotel staff, the multicultural ones were rated higher by supervisors.^[108] The same people felt "significantly more overburdened with cultural translation work" (Figure 14).^[108] The cost has one face the anchors cannot show: a software developer whose role as his team's "bridge to India" took 70 percent of his time, on top of his job.^[108] Untested: people who keep their identities apart may bridge more than people who integrate them.^[108]

The two anchors who bridged show the benefit and the bar together. Obama wrote of his Honolulu teens: "I learned to slip back and forth between my black and white worlds, understanding that each possessed its own language and customs and structures of meaning".^[248] The sentence answers a Black friend's "Speak for your own self", and the memoir goes on: "So why did such comments always set me on edge?".^[248] Ho Tung's bridge made him, by reputation, the richest man in the colony and still did not open the Hong Kong Club; he became the first chairman of the Chinese Club, founded in 1899 by Carroll's date, instead.^[364] When he reached San Francisco in 1908 with two wives, the New York Times greeted him with the headline in Plate 13.^[325]

Verdict: brokerage advantage is well replicated for people who hold bridging positions; that mixed-heritage people hold them more often is an inference, and the cost arrives with the benefit.

7.8 Careers

The Amerasian cohort gives the two ends of one in-between career: Kien Nguyen, from a wealthy family, a dentist at 31; Loan, who stayed, on about $38 a month for a household of ten.^[290,349] The middle is in the group data: 60 percent employed at an average wage of $6.54 an hour in the GAO's 1994 sample;^[343] three in four with fewer than eight years of school in a California survey of 275.^[65] Thirty years on, in the Homecoming wave as a whole (Amerasians, siblings and other refugee youth), those who had arrived at 14 to 17 earned about 11 percent less than US natives, and those who arrived at 18 to 21 about 24 percent less.^[175]

Useem and Cottrell surveyed American adult TCKs (ATCKs) in the early 1990s, the only adult-TCK dataset with attainment figures: 81 percent of 680 with at least a bachelor's degree against 21 percent of Americans.^[345] The later 604-person report gives 98 percent with post-secondary education, and the authors' own explanation is "a culture of highly educated individuals" in "elite communities".^[77] The sample was self-selected, so the gap is largely selection.^[345] Table 7.1 carries the rest, none of it with a comparison group. Military ATCKs had the least re-entry difficulty and missionary ATCKs the most.^[78]

Other career data are thin: a thesis finding that TCK schooling and identity shaped careers more than mixed parentage;^[299] the eight Korean TCK students of section 2.3.^[297] Expatriate research gives the one solid negative. Across 66 studies adjustment drives expatriate outcomes,^[30] and across 30 studies four Big Five traits predicted performance while "openness was not" among them.^[231]

Verdict: careers data are selected and thirty years old; openness does not map onto assignment success.

7.9 Attractiveness

Low confidence. Across 1,205 faces, mixed-race faces drew "a small but highly significant" advantage, with a hybrid-vigour explanation that is untested;^[201] Chinese raters showed no preference among composites.^[273]

Verdict: a perception, not a skill; small, uneven, and it sits beside exoticisation. Ho Tung chose a side and a club. Miyamoto chose a word. Section 8 asks when the edge is built, where it pays, when it fades and when it turns; section 9 asks what a family can choose on purpose, and section 14 grades it.

8. The edge across a life

Taipei, 2026. Harry Chan and Sean Hsu, both 19, both sons of Vietnamese mothers and Taiwanese fathers, have enrolled in the Vietnamese track of National Chengchi University's Department of Southeast Asian Languages and Cultures, a department mocked online as the "migrant worker department". Chan was born in Vietnam and grew up speaking Vietnamese with his mother and brother; he found he could understand it and not read it. Hsu's family spoke Mandarin at home, and he could manage little beyond greetings: "My mother was busy working and didn't teach me much about Vietnam's cultural or historical background." As children, both recall, speaking Vietnamese in public drew stares, and a market vendor "looked differently" at Chan and his mother. Now a vendor says "Wow, you can speak Vietnamese", and about half their cohort comes from immigrant families.^[189] Two ordinary lives, one edge, bought at nineteen with the literacy a childhood had not supplied. This section asks when the edge is built, where it pays, when it fades, and when it turns on its holder, and grades each answer by the evidence, which is thinner than the question deserves.

8.1 Building it

The host language is bought by age at arrival, and it buys schooling. Among immigrants who reached the United States as children, each step up a four-point English scale raised wages by about a third, 33 percent for "well" and 67 percent for "very well" against "poorly", with arrival at 11 or younger as the instrument; controlling for schooling cut the effect by a factor of three or more, so schooling carries most of it.^[38] The chance of dropping out of high school rises with each year of arrival after eight.^[22] The window for native-level grammar stayed open to about 17.4 in one test of 669,498 English speakers, later than the old puberty rule, though a reanalysis the paper did not read argues for a steady decline.^[136] The same early arrival that gives the host language pulls the child toward the host network: better English from early arrival raised intermarriage and divorce and lowered fertility and, for some, enclave residence,^[37] and in Sweden each five years' later arrival added 12 points to the chance of an immigrant spouse and 2 points to the immigrant share of neighbours and coworkers, with employment 6 points lower for ten years' delay.^[14] A bridge needs two live networks. Early arrival secures the host one; the heritage one survives only if the family and the community keep it alive (section 9).

A second language pays little by itself. Among US college graduates it earned a 2 to 3 percent wage premium;^[286] in a national adult-literacy sample bilinguals' higher pay "largely reflects" their higher schooling, and the premium was not significant once human capital was held constant.^[113] What pays is being strong in both. In two longitudinal samples, balanced bilinguals among children of immigrants earned more as young adults than English-dominant peers from the same minority groups, and the cost of complete linguistic assimilation was put at $2,100 to $3,300 a year after controls;^[3] in Southern California, fluent bilingualism lowered dropout and raised job status and earnings after controls.^[283] Whether the premium needs literacy in both languages is untested in the studies found: none isolates reading and writing from speech, and heritage literacy is the part that dies first (section 9).^[284] Chan could speak Vietnamese and not read it, the usual state of a second generation; the department is where he bought the rest.

The two identities pay when they fit together. In three studies of multiracial US adults, switching identity by setting went with lower well-being (section 3.6);^[288] people high in bicultural identity integration shift frames with the cue and people low in it shift against it (section 3.6);^[24] and among people who had lived abroad, those who identified with both cultures were promoted more than those who picked one, in observational data.^[318]

8.2 Where it pays

The edge is a seat, not a trait, and the seat is contextual. Brokerage pays whoever sits between contacts who do not know each other (section 7.7);^[47] Ho Tung's compradore post was that seat.^[364] In teams, cultural diversity raised creativity (0.16, on five effect sizes from 160 teams) and satisfaction (0.15), raised task conflict (0.10), lowered social integration (minus 0.07), and left performance unchanged (0.02), across 108 studies and 10,632 teams; the cohesion loss appears in student samples and not in field studies of managers.^[310] Those are teams of different cultures, not teams with one bicultural member, so the result reaches the in-between person by inference: the gains run through process, and process is what the bridge does. A university place can also make the seat. In Taiwan, 57 adult children of cross-border marriages said a bicultural identity had become "a likely option" for children of Southeast Asian mothers, with "ethnic dividends" mostly reaching university students with academic capital.^[190] For eight Korean third culture kids, English was an edge and weak Korean a job barrier (section 14).^[297] The seat pays where someone needs the bridge.

8.3 When it fades

Three things erode it. The heritage language goes first: in Southern California, Chinese, Vietnamese and Korean last 1.5 to 2.0 generations (section 9),^[284] and by the third generation 42 percent of Asian-origin children are not identified as Asian (section 11).^[101] Using it in an enclave costs the host language and income, independently of each other, in 1990 census data on adult men.^[62] Scarcity goes next: in India, fluent English raised men's hourly wages 34 percent, but among young workers only the educated kept a premium, which the authors list as one sign of a supply of English speakers that "has expanded so much", a cross-section of cohorts rather than a time series.^[17] Restlessness is the third, and it rests on small work: nine in ten adult TCKs said they were "out of synch" with their age group for life (section 5.4),^[76] and the paper found no study that measures job-change rates. What does not fade, in the one longitudinal test, is the appearance premium: facial attractiveness still predicted men's earnings in their early 50s.^[292] No study found tests whether the mixed-face rating advantage of section 7.9 pays in earnings or changes with age, so the paper prints no exotic premium.

8.4 When not to use it

The seat between two groups is exposed in four ways. First, the middleman-minority trap. The classic theory casts trading minorities as useful and resented at once, and its best-known case fails the test: American Jews had no sojourner outlook, did not shift into the intermediary positions the theory anticipates, and stayed a working-class group before the Second World War;^[59] across 107 countries from 1984 to 2003 there was no worldwide effect of market-dominant minorities on ethnic violence, only in Sub-Saharan Africa.^[29] The theory describes a niche, not every mixed person; the library's compradors and Chinese mestizos held it (section 10). Second, the clash. When the two countries fight, partial ancestry becomes full ancestry in the state's eyes. In 1942 the US Army incarcerated orphans "of Japanese ancestry" at Manzanar's Children's Village, 101 children by 1945; before the war the Shonien, one of three Japanese American orphanages in California, had taken in nineteen orphans of mixed heritage;^[86] a priest's question about orphans of part Japanese descent drew the reported answer that if any "have one drop of Japanese blood in them they must all go to camp", and a blond six-year-old was taken from a "white" orphanage.^[48] In Jakarta in May 1998 the joint fact-finding team interviewed 85 victims of sexual violence, 52 of them raped, in a report that files them under the rape of ethnic Chinese women; the death tolls disagree and the paper prints none.^[340] In the United States, anti-Asian hate crimes reported to the FBI rose from 279 in 2020 to 746 in 2021, on incomplete reporting,^[335] and two in three of nearly 11,500 incidents reported to a community centre over two years were harassment rather than crimes.^[316] The library's rule holds here (section 10.6): the premium a patron or a peace confers goes with it. Third, the cost of switching. Black professionals who adjusted themselves to fit in saw more career success and more burnout, in a survey of about 300;^[214] Black-white multiracial students at a mostly white university described "chronic codeswitching" to answer "too Black" and "too White";^[366] and Asian Americans, seen as less American and knowing it, answered identity denial by performing American knowledge, in five studies.^[61] Fourth, tokenism: the costs of being one of few at work are documented for token women, and the paper read that literature only in summaries, so the transfer to ethnic mix is an analogy and nothing more.^[172,369]

Side story Dennis Tojo, Los Angeles to Manzanar, 1942 A blond six-year-old who did not know he had Japanese heritage, taken from a "white" orphanage to the camp's orphanage.

Dennis Tojo, also printed Togo and later Bambauer, lived his first six years in a Children's Home Society orphanage in Los Angeles without knowing he had Japanese heritage. In 1942 Army authorities removed him from the "white" orphanage and sent him to the Children's Village at Manzanar, the one orphanage inside the camps. The Wartime Civilian Control Administration's director, Colonel Karl Bendetsen, asked by a Maryknoll priest what would happen to orphans of partial Japanese descent, reportedly replied that if any "have one drop of Japanese blood in them they must all go to camp"; the National Park Service report that carries the line marks it second-hand and adds that it "was not rhetoric". Sixty-one children arrived by bus in June 1942, and 101 lived in the Village until September 1945. Dennis later gave an oral history for the park's project, and the report reproduces his identification card.^[48,86] Plate 14 shows the Village's superintendent, Harry Matsumoto, with some of its children in 1943; it is not a portrait of Dennis. He is the paper's clearest case of the edge turning: a child housed as white in peacetime, counted as the enemy's the year the two countries went to war.

8.5 The life cycle, graded

Table 8.1 grades the section's advice on the scale of Table 14.1: A means pooled or causal evidence in the right population; B means causal or replicated evidence in a nearby population (minority students, immigrants, bicultural adults); C means observational or small-sample evidence; D means expert opinion only. Nothing here was measured on mixed-heritage third culture kids, and two of the questions this section set have no evidence at all: no study found tests whether the mixed-face rating advantage pays or changes with age, and none tests whether heritage literacy, as against heritage speech, pays.

The grades say what the evidence can say: the edge is real where someone needs a bridge, it is built in childhood by arrival and by two kept languages, it thins with the heritage language and with the supply of people who can do the same, and in a clash it is a liability first. What a family can choose on purpose comes next.

9. Embracing it: what the evidence says works

Tokyo, 2015. Ariana Miyamoto, 20, had entered the pageant after her friend's death, and she said so.^[139] She kept the title through the posts saying it belonged to a "pure" Japanese, and she took the word "hāfu" because the other word was refused her.^[143,360] She said she did not need to learn the tea ceremony or be better at calligraphy "just because I look different": "I don't feel like I need to compensate".^[219] "When I was small I stood out and always felt I had to fit in with everyone," she told a wire reporter; "now I say what I feel".^[143] That is identity autonomy, the freedom to name oneself, claimed in public by a woman who had spent her school years in Sasebo making herself stop thinking about the words.^[139,298] This section asks what a family, a school or a person can do on purpose.

One case shows the speak axis doing the work the look axis refuses, more cleanly than any anchor: Trevor Noah, born in South Africa to a Black mother and a white father, learned as a boy that "if I spoke like you, I was you".^[244] Obama's return to his full name in his late teens or early twenties (his sister dates it to Christmas 1980), "not a racial assertion" but a statement that he no longer needed to fit in, is the same move on the think axis.^[361]

Side story Trevor Noah, South Africa, 1980s and 1990s The chameleon: a child whose language changed how people read his colour.

Trevor Noah (Plate 15) was born in South Africa to a Black mother and a white father; his memoir's title, Born a Crime, names what their relationship was under the law of the time. In its fourth chapter, "Chameleon", he describes a group of Zulu-speaking men who relaxed once they heard him speak their language. He drew a lesson from the moment: "Maybe I didn't look like you, but if I spoke like you, I was you." If someone addressed him in Tswana, he answered in Tswana. He became, in his own word, a chameleon, able to change how people read his colour by changing his language, and the habit became his trade as a writer and performer.^[244] The paper's three axes meet in his sentence. The look axis asked the question, and the speak axis answered it, which is what the speak axis of section 2.3 is for. No anchor shows the move this cleanly: Obama's "slip back and forth" (section 7.7) crosses the think axis, and Miyamoto's unaccented Japanese could not pass the test her face failed (section 6).

Parents' heritage talk. Ethnic-racial socialization is what parents tell children about their heritage and about bias. Across 102 studies and 27,221 participants, such talk was "positively, albeit modestly" linked to self-perception, relationships and internalising behaviour, and not significantly to externalising behaviour.^[353] The samples were children of colour; mixed-heritage children are not reported separately. Obama's mother taught him pride in being Black where he saw nobody like himself, and he kept his doubts from her; no study tests that case.^[248]

Heritage language. Among 414 ninth graders of Asian and Latin American background, proficiency in the heritage language, not the habit of using it, went with closer parents and a stronger ethnic identity.^[251] The arrow may run the other way: in over 620 teenagers followed for two years, closeness predicted later language use and not the reverse.^[331] An 18-study meta-analysis finds a moderate correlation between heritage-language skill and ethnic identity.^[234] The language itself does not last. In Southern California, Chinese, Vietnamese and Korean mother tongues have a life expectancy of 1.5 to 2.0 generations, against 3.08 for Spanish,^[284] and even Spanish falls to 17 percent fluent speakers by the third generation.^[284] In the Children of Immigrants Longitudinal Study, at age 24, 34.8 percent of the Asian-language group spoke the language very well and 7.5 percent wrote it very well; the 2.5 cohort, with one US-born parent, the nearest proxy for a mixed home, spoke very well at 22.1 percent (Figure 15).^[284] The Amerasian cohort had no father's language to keep and arrived with 7 percent unschooled, so these series describe their host community, not them.^[343]

What keeps a language alive is input. The 1,899 families of section 5.5 succeeded most often where both parents used the minority language.^[83] One Vietnamese weekend school was linked to some language skill and "had little impact on integrated cultural identity".^[209] The best school evidence is a lottery study: children who won a dual-language immersion place read 7 to 9 months ahead in English by grades 5 and 8, with no identity outcome measured.^[313] Pew reports that 72 percent of Vietnamese Americans speak Vietnamese at home (Plate 16).^[152]

Identity autonomy. Let people name themselves. Section 6.1 gave the harm side.^[287] The other side holds: among 182 mixed-heritage high schoolers, those who claimed several groups reported "equal or higher" well-being.^[36] Root's Bill of Rights (section 3.5), a charter rather than a study, claims the right "To have loyalties and identification with more than one group of people".^[280] In 399 TCK students in the UAE, integration went with higher well-being, in cross-sectional data.^[94] The one randomised trial, in a general rather than mixed-heritage sample: 215 ninth graders assigned to a short identity-exploration course showed higher self-esteem, lower depressive symptoms and better grades a year later.^[338]

Stories people tell. People whose life stories find meaning in hardship and feature agency "tend to enjoy higher levels of mental health, well-being, and maturity".^[212] All of it is correlational and none of it is on in-between people. The stories exist anyway: Kien Nguyen wrote his memoir on a therapist's instruction,^[349] and a Vietnamese Amerasian who returned in 1996 wrote, "I felt I had finally come home".^[198] Japan's returnee children went from "educational orphans" to "symbols of Japan's internationalization" (section 5.2), a shift in the watchers more than in the children.^[171]

Peers like you. Families in Global Transition began at Ruth Van Reken's kitchen table in 1997 and held its first conference for 85 people in 1998.^[105] A systematic review of 14 TCK studies found that close friendships buffer adjustment and reviewed no programme evaluations.^[164] Six adult TCKs wanted friends with migrant backgrounds.^[187] Miyamoto's generation built its own: Edward Sumoto, Venezuelan-Japanese and raised in Kobe, founded Mixed Roots Kansai.^[321] No study shows that any of these groups improves an outcome.

Transition practice. RAFT (Reconciliation, Affirmation, Farewells, Think destination) is widely taught in international schools and tested nowhere; it is expert opinion.^[264] Everything so far is one life or one generation. Jean Gittins, in 1941, had to decide for the next one.

10. History as natural experiments: the coded case library

Hong Kong, February 1942. The colony surrendered on Christmas Day. Dr Jean Gittins, 33, born Ho Tung, daughter of the man reputed to be its richest, is free: Eurasians are not interned, and she holds two nationalities.^[364] If she claims the Chinese one she can "return to the family home and remain free". She asks herself: "Could I on moral grounds now use my Chinese nationality merely to escape internment?" At a bank she tests the claim, "in name and appearance I was a foreigner", and the Japanese finance controller takes her word. On 21 February she walks into Stanley camp with the British (Plate 20), hoping for an exchange that would take her to her children in Melbourne.^[127] Her husband Bill is a prisoner of war; his all-Eurasian volunteer company lost all its officers and 70 percent of its men.^[199] One historian adds a detail her memoir does not: that a doctor advised her to appear as "Miss Jean Gittins" on the hospital staff list, with "Hotung" dropped.^[199] Either way, the second generation of the Ho Tung line chose its label under a regime shock, and two of the third had already left for Melbourne. That is the pattern this section tests on 28 groups.

10.1 Method

The library has 32 rows (Appendix A, Table A1): 28 mixed-heritage groups with a social history, plus four deep-history rows on ancient admixture that enter no count. Each group carries five codes, written from the sources alone: a bridge role (did the group work as go-betweens for rulers and ruled?), legal security at the group's height, a regime shock with its main year, majority acceptance as the sources describe it, and a coping mode, the dominant response across generations: niche, pass, merge, people, leave, or mixed. A sixth variable, the third-generation outcome, is free text collapsed into four classes: dispersed, absorbed, persists as a named group, or untraceable (Appendix A.2). The groups were chosen in an afternoon, Asia first; famous cases are over-represented; 22 rows carry a weak code; and with n = 28 the paper runs no test (Appendix A.1).

10.2 The library at a glance

The map is dense in Asia by design (Figure 16). The timeline shows one thing the paper did not plan (Figure 17): ten of the 25 dated shocks fall in the twenty years from 1941 to 1961, eight of them in Asia, from the Japanese conquest to the end of empire, and the Hong Kong surrender, behind Jean Gittins's camp gate, is the first entry (Table A1). Seven groups have a count that can be drawn over time (Figure A1, Appendix A; section 10.8).

Before the first colonial row, every present population was already a mixture: about 15 percent Jomon ancestry in present-day Japanese;^[74] the Kinh today closest to Hmong-Mien and Tai-Kadai groups, with no Hoabinhian signal.^[203]

10.3 One family through the codes

The Ho Tung line is the one group the library can follow person by person.

Bridge role: yes. Robert's compradore post at Jardine Matheson, section 7's opener, was the go-between job the codes name.^[364] Legal security: partial. He held dual Chinese and British status.^[199] He was the first non-European allowed to live on the Peak, despite a 1904 ordinance meant to keep Chinese off it, through "an English friend" who bought houses for him; the Hong Kong Club still barred him.^[11,58,364] Majority acceptance: low, on both sides. In 1901 the Registrar General remarked that most Eurasians dressed and lived as Chinese and would call themselves Chinese; "Eurasian" was a reproach, and Cantonese had its own slurs.^[199] The census count of section 2.4 fell to 42 in 1911 and rose to 837 in 1931, a jump one historian reads as a generation born to marriages rather than cohabitations.^[199] Irene Cheng, writing in 1976: "the majority therefore identified themselves with the Chinese and tried in every way to be as 'Chinese' as possible".^[58]

Coping mode: mixed, and visible inside one family. Robert chose Chinese. A cemetery historian and his daughter Jean both describe it as a choice: "In spite of his pale skin, blue eyes and red hair, Sir Robert Hotung chose to live the life of a Chinese gentleman"; he and "others who thought likewise threw in their lot wholeheartedly with the Chinese".^[127,202] China's rule by the father would have made him European; the American rule, by the non-white parent, would have made him Chinese; he chose Chinese.^[230] His children married into the Eurasian web and out of it.^[58,199] Jean sent her children out on their own passports, with an English surname, and then declined to claim Chinese nationality for herself; others swapped European surnames for Chinese ones to stay out of the defence force,^[199] and in July 1940 some Eurasian evacuees "not of pure British descent" were "weeded out" at Manila.^[199]

Regime shock: 1941. Under Japan the Eurasians were "Third Nationals". Sir Robert Kotewall dropped his British title and took the name Lu Kuk Wo (Plate 19); M. K. Lo, Robert's son-in-law, was held in solitary confinement until he agreed to join the new government.^[199] Robert spent the occupation in Macau (section 5.6).^[199]

Third generation: dispersed, and this line documents it. "Oddly enough," Irene Cheng wrote in 1976 of Clara's eighteen grandchildren, "none of the eighteen grandchildren married Eurasians"; the endogamous web describes her own generation, not her children's.^[58,199] After 1949 most Eurasians left China, and the 1967 riots "marked the first wave of Eurasian ... diaspora from Hong Kong to the US, UK, Canada and Australia".^[199] Jean later lived in Melbourne.^[127] Of the grandchildren, Sir Joseph Hotung died in London in 2021; his obituary does not say how he described his heritage.^[305] The heritage board lists a general, a philanthropist and a casino magnate's grandfather among the descendants.^[11]

10.4 Five coping modes, one group each

Five groups stand for five of the six modes, one paragraph each; Table 10.1 carries the rest.

Leave: the Sri Lankan Burghers. The census counted 46,000 Burghers and Eurasians in 1953 and 37,100 in 2012, and their share fell from 0.65 percent in 1881 to 0.18 percent in 2012.^[309] The law defined them through the father's line from 1883.^[236] Their niche was English, and the Sinhala-only Act of 1956 changed the language of the state: "We were ruined by the Sinhala-only Act, because by then English had become our mother tongue," the president of the Dutch Burgher Union said in 1988.^[81]

Mixed: the Anglo-Indians. The last census to count them separately, in 1931, found 138,895; estimates for 1947 run from 200,000 (Blunt) to 300,000 to 400,000 (community leaders), since claiming to be British was a way out of prejudice.^{[31,39,233,367]} India's constitution let the President nominate two Anglo-Indians to parliament; in 2011 a census with no Anglo-Indian column recorded 296 of them, and the nominations lapsed in 2020.^[31,367] On 15 August 1947 a troopship reached Fremantle with more than 700 Anglo-Indians aboard; about a third of the community had left by the 1970s.^[39]

Merge: the Chinese mestizos of the Philippines. Spain taxed the mestizos de sangley, children of Chinese fathers and local mothers, at double the indio tribute.^[359] Wickberg puts them near 120,000 in 1810 and near 240,000 by mid-century.^[359] The niche was conversion, marriage, money-lending and land.^[359] The category was legal, so it could be changed: Jose Rizal's family was re-registered as indio, though by ancestry he was a fifth-generation Chinese mestizo. By the late 1800s the mestizos "increasingly identified themselves with the indios in a new kind of 'Filipino' cultural and national consensus". The group did not vanish. Its name did.^[359]

Side story Jose Rizal, 1861 to 1896 By one historian's reckoning a fifth-generation Chinese mestizo, registered as indio, who used a surname of his own at school and became the hero of the nation that absorbed his group.

Jose Rizal's paternal ancestor Lam-co, a Chinese man from the Chinchew district of Fujian, was baptised in the Parian church of San Gabriel in June 1697. He was 35, and took the name Domingo. The family took the surname Mercado. "Sangley, Mercado and Merchant mean much the same", his 1913 biographer Austin Craig wrote; the name kept the Chinese line in view while avoiding its prejudice. On his mother's side his great-grandmother came from the Florentina family of Chinese mestizos of Bulacan. After an 1849 decree that every Filipino take a surname, Rizal alone used "Rizal" at school while the family stayed Mercado (Plate 21).^[79] The tax register classed the family as indio.^[359] Craig: "He was a typical Filipino, for few persons in this land of mixed blood could boast a greater mixture than his." He was executed in 1896. Craig wrote that his life prepared "the union of the various disunited Chinese mestizos, Spanish mestizos, and half a hundred dialectically distinguished 'Indians' into the united people of the Philippines".^[79] Merge, in the library's code, has a face here. The category changed on paper and the name at school, and the man who carried both became the symbol of the people that absorbed his group.

People: the Canadian Metis. The census counted 204,115 Metis in 1996 and 624,220 in 2021, and Statistics Canada's own reading is that most of the rise is people reporting their identity differently, not births.^[312] In 1982 the Constitution named "the Indian, Inuit and Métis peoples of Canada" and defined none of them; in R v Powley (2003) the Supreme Court set three "indicia" of Metis identity: self-identification, ancestral connection, and community acceptance.^[72]

Pass: the United States and the one-drop rule. Virginia made a "mulatto" (1785) anyone with one-fourth or more "negro blood", moved the line to one-sixteenth in 1910, and in 1924 defined a white person as one with "no trace whatsoever" of other blood.^[205] Thirty states banned interracial marriage at the peak from 1913 to 1947; sixteen still did when Loving v. Virginia struck the bans down in 1967.^[205] Homer Plessy, "seven-eighths Caucasian and one-eighth African blood" in the Court's words, said in his petition that the mixture "was not discernible in him".^[205] Jean Gittins's children passed, on their own passports, under the same logic in reverse.^[127]

Table 10.1 carries the other coded groups in one line each. The Can Tho children at its foot are a present-day case the library did not code.

10.5 Children of soldiers

The Amerasian cohort's door was built by a photograph: a 1985 picture of Le Van Minh, an abandoned Amerasian boy with polio, set off a petition, and the congressman who flew to Vietnam in 1987 found dozens of Amerasians camped in a church, saying, "I want to go to the land of my father".^[175] The Amerasian Homecoming Act of 1987 followed. It covered children born in Vietnam from 1962 to 1975 and fathered by a US citizen, and asked for a face-to-face interview; the 1982 Act of section 2.2 had named "physical appearance".^[268] The GAO funnel runs 131,814 signed up by August 1993, 77,577 approved, 68,558 arrived, most of them relatives; approvals rested partly on appearance in practice (Figure 18).^[343] Kien Nguyen left two years before the Act, under the Orderly Departure Program.^[349]

The cousin cohorts had no such door. Japan's occupation-era children, the konketsuji, numbered about 5,000 in two 1952 surveys, against an earlier guess of 150,000.^[334] Miki Sawada opened a home for them at Oiso in 1948, and the exit was American law: the acts of 1952 and 1953 opened adoption.^[104,334] Korea's Amerasians were never counted; estimates in 1981 ran from 1,500 to 10,000.^[270] Those who stayed were filed in welfare reports as "social handicaps";^[363] Insooni, born in 1957 to a Korean mother and an American soldier, became one of Korea's best-known singers.^[261]

Side story Miki Sawada and the Elizabeth Saunders Home, Oiso, 1948 The Mitsubishi founder's granddaughter, who sold her property to house the occupation's children.

Miki Sawada (Plate 23) was born on 19 September 1901, the daughter of Hisaya Iwasaki, eldest son of the founder of Mitsubishi. A founding committee met at St Luke's International Hospital in Tokyo on 1 October 1947. It named the planned home for Elizabeth Saunders, an English Anglican who had died the year before and left US$170 to Anglican social work. The Home opened on 1 February 1948, with Sawada as its first director and two children; it stands on the Iwasaki family's former villa at Oiso. She sold her remaining property to run it, became the first chair of its corporation in 1953, and opened a school the same year. Wikipedia, citing her autobiography, tells of a Black baby's body falling from a train rack onto her lap in Gifu in early 1947, the moment that moved her. The paper could not confirm the scene outside her own telling. She died in Palma, Mallorca, on 12 May 1980.^[104] The children's exit was the one the cousin cohorts had: American adoption law, from 1952.^[334] Of the roughly 5,000 children the 1952 surveys counted,^[334] the sources read trace none into adulthood.

The Vietnamese leavers are the one cohort with outcome data at a few years and at thirty. Before they left, 161 Amerasian youth were measured in Vietnam and 95 again in a Philippine camp: more risk factors at home meant more symptoms abroad, and a sample of 147 replicated the line.^[217,354] A few years after arrival, 140 Amerasians carried more trauma and distress than their 71 non-Amerasian siblings yet adapted as well as other Vietnamese immigrants of their age.^[216] Thirty years on, those of the Homecoming wave, Amerasians, siblings and other refugee youth together, who had arrived at 14 to 17 matched US natives on college completion, while those who arrived at 18 to 21 kept a gap; an estimated 400 to 500 Amerasians never left.^[175,198] Loan, when interviewed, was one of them.

10.6 Five patterns

The five cross-tabs behind this section are printed in Appendix A as Tables A2 to A6, each a table of names.

Five patterns are worth printing, each with its count and its limit (Figure 19). None is a cause, and the last two are true by selection: they describe how the library was built, not what happened to mixed groups in general.

First, absorption beat persistence. "Mixed" is the modal coping mode (12 of 28), and absorption into a majority or a parent group beats persistence as a named group, 9 against 5. A mixed-heritage group more often became part of someone else's nation than its own: the Chinese mestizos came to identify with the indios and then as Filipinos, the Minh Huong "all but disappeared", the castas became mestizos, the pardos became Brazil's largest category.^{[155,238,358-359]} The five groups that persist as named groups (Metis, Krio, Garifuna, Cape Coloured, Kristang) each had a territory, a language, or a state category to hold the boundary, which is Barth's point (section 11).^[18] The limits: five cases cannot test it, and 9 is an upper count, since two pairs of rows describe overlapping populations (the two Mexico rows, the two US Black-white rows) and the pardos hold a state category of their own.

Second, being the bridge did not decide the third generation. Dispersal (9 of 28), absorption (9) and persistence as a named group (5) each occur under every bridge code (Table A3). The limit is that the four classes are one reader's coding of free text, and 5 groups could not be traced.

Third, the go-between niche and the legal shield went together. Eight of 10 bridge groups had a secure legal status; 5 of 10 non-bridge groups had none (Table A6). In the analyst's reading, both came from the same patron, a colonial state that used a group as clerks and interpreters and wrote it into law; Ho Tung's Peak exemption was a family privilege inside a "partial" code, not the group's. The limit is that 6 of the 8 codes in the top-left cell are weak.

Fourth, true by selection: this is a decolonisation library. Twenty-five of 28 groups met a change of rulers or state policy, and 10 of the 25 dated shocks fall between 1941 and 1961 (Table A2; years in Table A1 and Figure 17). The workers chose groups that had a shock, so the count describes the library and says nothing about what shocks do.

Fifth, true by selection: every group that left had a door. All five "leave" groups had a receiving state with a legal route into it: Burghers, Anglo-Burmese, Indochina metis, Vietnamese Amerasians under the 1987 Act, Japanese konketsuji under the laws of 1952 and 1953. A group coded "leave" could hardly have left without one, so the pattern is close to a definition, and the informative question, whether a group with a door and no shock left, or a group with a shock and no door stayed, cannot be asked of 28 rows in which only three lack a shock (Table A2). The konketsuji code is the weakest of the five and now carries the weak mark (Table A1): the sources count about 5,000 children in 1952, record adoption abroad under the US laws of 1952 and 1953 as the exit, with a 1954 magazine report of 300 American couples adopting Japanese orphans, mixed or not, and no total, and trace none into adulthood, so "leave" there means that the recorded exit ran abroad, not that most of the cohort went; for the same reason its third-generation class is untraced, not dispersed.^[334] Legal security at home did not predict staying (Table A4). The limit is five cases.

One table is printed and not counted: acceptance against coping (Table A5), because the same reader coded both from the same texts.

The historical ledger, in two sentences. The pro was a niche with a legal shield: compradors, clerks and interpreters whose status a patron state wrote into law. The con was that the niche and the shield died with the patron, and what followed ran from a changed state language and an exodus to removal, sterilisation and a camp gate.^[81,199,356]

10.7 Outside the colonial coast: a test of the pattern

Most of the 28 rows share a setting: a colonial or occupying state, a foreign father, a local mother, and a law that named the child. A seventh research pass asked which steps of the library's sequence, a niche under one regime, a shock, then leave, merge, pass or persist, appear where that setting is absent, in seven more cases coded on the same five variables. Four (Koryo-saram, Zainichi, Nikkei, Joseonjok) are co-ethnic diasporas, not mixed-heritage groups: they test the sequence, not the mixed child. They are an out-of-sample check, read against the 28 and never added to their counts (Table 10.2).

Kazakhstan shows every step. The niche was the Russian-speaking city. In 1989 the census had more than 98 percent of Kazakhs naming Kazakh as mother tongue, while a leading Kazakh scholar put the share who knew it poorly or not at all near 40 percent.^[107] The shock of 1991 was the end of the Union, and the new state's lever was language.^[107,272] Russians and Germans then left by the million, and by 2004 four in five of those Germany admitted came as spouses, descendants or other relatives.^[45,365] In 2022 focus groups some older Kazakhs called poor Kazakh shameful, Russian remained the route up, and participants of mixed origin tended to define themselves by job or family rather than by nation. A Shymkent entrepreneur, Russian on her papers, would rather call herself "a citizen of Kazakhstan".^[157]

The Koryo-saram show every step too: deported, then an educated niche, then Russian as first language inside one generation and marriage out.^[177,329] In Korea they look local and speak Russian, the hidden-immigrant cell, and the homeland admits them as low-wage workers: "Chinese think that I am Chinese, Uzbeks that I am Uzbek, only Hanguks (South Koreans) never think that I am one of them".^{[161,294,370]} The Zainichi show shock and exit without a niche, passing and merging at the scale of a community. A treaty stripped their nationality, and naturalisation, marriage to Japanese and a 1985 nationality law drained the category with little emigration.^[153,245]

The Nikkei Brazilians and the Joseonjok show neither the niche nor the regime shock. Both were pulled in by a visa for descendants, as factory labour (Plate 24).^{[142,294,368]} The speak axis was wide open: at the start of the return more than half of Japanese Brazilians could not read Japanese.^[179,368] One observation cuts against the look axis: a dekasegi who looked wholly Japanese described a "problem of similarity", while a cited study found that those of mixed descent, from whom nobody expected the norms, reported smoother relations.^[179]

Black Amerasians add a ladder by skin colour inside a group the library holds; Loan's cohort is where its sources place the sharpest stigma.^[194,198] The tally by steps: two show the full sequence (Kazakhstan, the Koryo-saram); two a shock and an exit without a niche (the Germans, the Zainichi); one an exit with no niche or shock carded (the Black Amerasians); two a visa and no niche or regime shock (the Nikkei, the Joseonjok). The limits are sharper than the library's: few sources per row, most of them secondary, nationality counts that hide mixed people inside the categories, and no third-generation data for any of the seven. The 28-case counts of section 10.6 stand unchanged.

10.8 What the library cannot say

The library supports no causal claim, for the reasons section 10.1 gave, and its codes capture look and speak better than think. The size question, asked of history, gets the answer section 2.5 gave for the present: a definition, not a count (Figure A1). Anglo-Indians run from 296 in the 2011 census to 400,000 in community claims,^[31] and Hong Kong's 42 Eurasians in 1911 against 837 in 1931 is a count of who was willing to tick the box.^[199]

10.9 Euro-Asians then and now

Set the 837 Hong Kong residents willing to tick "Eurasian" in 1931 against the 66,732 who ticked "Mixed" in 2021, 63 percent of them born in the city, and the question is whether the second group is the first one's heir.^[53,199] Singapore holds the same question in one Association, which widened its definition to anyone of Asian and European heritage and, one member says, recruits "New Eurasians"; some old members resist: "Pure Eurasians no more," said one 62-year-old.^[277] The Euro-Asian of the present has no patron. Daniel Henney, son of a white American father and a Korean adoptee mother, grew up in Michigan leaning "heavily to my white side" and now calls himself "a Korean actor who works in Hollywood".^[303]

Side story Merle Oberon, Bombay to Hobart, 1911 to 1979 The daughter of an Anglo-Indian mother, who passed into stardom on a Tasmanian birth story and never put it down in public.

Merle Oberon (Plate 25) was born Estelle Merle O'Brien Thompson in Bombay in 1911. Her grandmother raised her as her own and told her that her teenage mother was her half-sister. At a good Calcutta girls' school, where she held a scholarship, classmates drove her out with overt racism. When the producer Alexander Korda took her up, a Tasmanian birth was chosen for her. It was picked, in the film-maker Maree Delofski's words, "because it was so far from the US and Europe and generally considered to be 'British' to its core". She also said her birth records had burned in a fire. In 1965 she cut short an Australian trip when journalists pressed her about her background. In 1978 she attended a "welcome home" reception in Hobart; the accounts of what she said there conflict. Her biographer Mayukh Sen has her telling a driver, "I was actually born in India." She died in 1979, never having admitted it in public. Sen describes her mother as half-Sinhalese and half-white, and Anglo-Indian by her own account.^[21] Table 10.3 has a cell for this: "claimed British or European where it paid". Oberon is what the claim cost one person to keep for a working life, and the school in Calcutta is where the paying began.

The largest present-day dataset on Euro-Asian children's outcomes that the paper found is England's. Pupils recorded as Mixed White and Asian, 10,198 of them in the 2025/26 school-leaving cohort, averaged 52.3 Attainment 8 points against 44.4 for White British pupils and 53.3 for all Asian pupils, a lead that has held in every comparable year since 2018/19.^[87] It holds inside poverty as well as outside it: 39.5 against 31.2 among pupils on free school meals and 57.1 against 48.8 among the rest.^[87] Of the cohort aged 15 in 2020/21, 56 percent were in higher education by 19 against 38 percent of White British pupils, and 34 against 17 percent among those on free school meals;^[89] and among UK graduates of 2008/09, those recorded as Mixed White and Asian earned a median of GBP 36,600 ten years on, against 32,200 for White graduates, though fewer were in sustained work or study, 82 against 86 percent, and more had activity the tax records did not capture.^[88] The lead is the White-and-Asian group's, not the mixed label's: poor Mixed White and Black Caribbean pupils scored 32.3, level with poor White British pupils, reached university at 21.8 percent against their 16.9, and outside free school meals trailed White British pupils, 45.0 against 48.8 (section 6.4).^[87,89] Three caveats. The 2025/26 figures are the newest release and may be revised; free school meals are a crude poverty proxy and parents' education is not controlled; and no official labour-market table splits the group from the broad Mixed category, which sits level with White British on pay and below on employment, so the adult who is not a graduate is unmeasured.^[87,90]

Four differences carry the comparison. The category: a legal name that followed the father then; a self-ticked box or parents' nationalities now.^{[165,228,249,367]} The community: the Ho Tungs married inside a web;^[199] the new are dispersed couples whose children drift toward the majority label.^[101,269] Who the mix is: then a European father with colonial power; now the Western partner is the minority of the mixing, 2,038 births in Japan with a US or UK parent against 7,326 with a parent from Korea, China, the Philippines or Thailand in 2024.^[228] The patron: the old niche came with a legal shield and died with the empire; what remains is a status image in which half-white ranks above half-Asian, by one filmmaker's account, with no wage or hiring study behind it.^[285] Singapore's double-barrelled race lets such a child carry both labels but still ranks one first (section 11).^[276-277] Verdict: today's Euro-Asian children lack the colonial patron, and with it both the protected niche and the regime shock; what they inherit is the question of which side will claim them, now answered more often by the majority than by a community of their own, and, in England, where the school census counts them, a lead over the majority that the old niche never measured. The Welfare League, founded in 1930, proclaimed "no gulf between a Chan and a Smith amongst us" and could not fill its membership list.^[199] Section 11 asks whether anyone has done better since.

11. A new group? Tests for people-hood

Hong Kong, 1930. Charles Anderson's Welfare League has a manifesto, dated to 1929 in Teng's history, that there is "no gulf between a Chan and a Smith amongst us". The League's own record laments a short membership list, and Anderson's daughter later recalled four leading figures who were members in secret and Chinese in public.^[199] Teng's history reads the manifesto as a call for unity and argues against a simple story of marginalisation.^[325] The anchor family's own community tried to make a people of its position and could not fill a list. Whether anyone has done better is this section's question.

The theory cards give three tests and the courts a fourth. Weber: an ethnic group rests on "a subjective belief in their common descent".^[355] Barth: a group persists through the boundary its members keep up.^[18] Brubaker: do not assume a group from a category; "groupism" is the analyst's error.^[43] Powley, from section 10.4: self-identification, ancestral connection, community acceptance.^[72] The paper folds these into four tests: a self-label most members use, a boundary against both parent groups, institutions that act for the group, and something that passes to the next generation.

Test 1, the label. Among Americans with two or more races in their background, 39 percent call themselves mixed race or multiracial and 61 percent do not, in one survey of 1,555 adults.^[262] In Japan the label exists and is fought over: hāfu against daburu (section 2.2).^[293] Miyamoto accepts it because the alternative is refused her.^[360] Singapore's "Eurasian" is an old label the census still counts (Plate 27), and the group got younger between 2010 and 2020.^[301] A label used by a minority of those it could cover fails the first test.

Test 2, the boundary. In the United States the boundary dissolves outward. Twenty-one percent of second-generation adults and 42 percent of third-generation children of Asian origin are not identified as Asian; the figure is 55 percent when only one side of the family is Asian.^[101] Alba's argument is that the children of mixed unions join the mainstream rather than a new minority.^[342] Hong Kong's "endogamous web" is the clearest marriage boundary in the sources, and it held for one generation: none of Clara Ho Tung's eighteen grandchildren married a Eurasian.^[58,199] Mixed children in the United States do not harden into a group; they thin into the majority (Figure 20).

Test 3, institutions. The Association of MultiEthnic Americans was organised in Berkeley in 1988 by 14 member groups. It first wanted a "multiracial" box, then backed the rule the Office of Management and Budget adopted in 1997: mark one or more races, no separate category.^[97] Hong Kong's Welfare League had its short list.^[199] Families in Global Transition grew from a 1998 conference of 85 people to affiliates on three continents, by its own account.^[105] One state built a category with a hyphen. From 1 January 2011 a Singaporean child of parents of different races could carry both in the recorded race at birth, "dominant race" first; the first component counts for housing quotas and mother-tongue classes.^[276] In January 2010, the week the rule was announced in Parliament, Carol Loo, 28, and her Indian Singaporean husband decided their children would be "Indian-Chinese": "It's good to open our minds to the idea that Singapore is not made up of just four races".^[324] The institutions are small, and the American one asked the state for a way to be counted as several things, not as one.

Test 4, transmission. Heritage languages in Southern California last 1.5 to 2.0 generations for Chinese, Vietnamese and Korean, 3.08 for Spanish (section 9).^[284] In the case library, 9 of 28 groups were absorbed, an upper count (section 10.6), and 5 persist as named groups, each holding a territory, a language, or a state category (Appendix A.2). The Ho Tung line passed a surname and a fortune, not the label.

By these tests, and in the paper's judgement, in-between kids today are a category with a vocabulary and a few organisations. They are not a people. The label is used by a minority; the boundary runs outward into the majority; the institutions are small; and what passes to the next generation is mostly the mainstream language. The historical groups that persisted as named groups had what today's do not: a place, a tongue, or a line in law. A category, not a people, so each family decides alone.

11.1 Is unmixed worth more?

No, not as a trait. Neither state is worth more in itself; unmixed status pays only where the majority and the state reward it, and where that reward has been dismantled it went fast: US approval of Black-white marriage went from 4 percent in 1958 to 94 percent in 2021 (Figure 22).^[213] The reasons follow.

"Original" is a definition, not a fact. Every present population is a past mixture, from the 15 percent Jomon ancestry in present-day Japanese to the Kinh, closest today to Hmong-Mien and Tai-Kadai groups (section 10.2).^[74,203] Where being unmixed carried a premium, a state built it. China gave a child the father's ethnicity and the United States the non-white parent's, and Robert Ho Tung, a dual Chinese and British subject, chose Chinese.^[230] Virginia's 1924 rule admitted "no trace whatsoever";^[205] a Korean army, Hines Ward said, took only "pure-blood Koreans";^[15] Japan's one-people story told Miyamoto she could not be Japanese.^[178,360] The costs of being mixed are social: in 193 students, distress tracked discrimination aimed at being mixed, not general discrimination (section 6.1).^[93] The one practical cost is in the bone-marrow registry. A stem-cell transplant needs a donor whose HLA type, the set of immune markers on cells, matches the patient's, and HLA types run in populations. In 2012 a white European patient had a 75 percent chance of an available full match in the US registry, a Black South or Central American patient 16 percent, and Asian groups 27 to 42 percent (Figure 21).^[130] A patient with one white parent often does better than a patient from a single minority group; Asian-African mixes do worst.^[25] Race is a poor proxy even here: among donors who ticked two or more races, 28 percent of men but 9 percent of women carried HLA from two origins.^[146] Nick Glasgow, a Californian with one Japanese grandparent, was told a match was a long shot, "zero chance" in his mother's recollection; two donors were found within weeks, and the economists who analysed his case blame his rare haplotype pair.^[25] The registry rewards being typical, not being unmixed. The gains of mixedness are small and contextual (section 7), and the "hybrid vigour" reading of mixed faces is untested.^[201]

The biological reading of the question also has an answer, and it is small. The largest genetic studies measure inbreeding, not mixing. In up to 354,224 people from 102 cohorts, more genome-wide homozygosity, the signature of related parents, went with less height, schooling and cognitive ability: about 1.2 cm, 10 months and 0.3 standard deviations for the child of first cousins against the child of unrelated parents, effects the authors themselves call small.^[166] The effect shrank by 17 to 35 percent once schooling was added; later samples found it smaller once background was controlled, and one earlier sample found it driven by educated parents who moved and partnered across distance.^[70,163,166] A 1.4-million-person follow-up confirmed the inbreeding cost, strongest on fertility.^[67] None of this compares mixed with unmixed children: Joshi analysed each continental ancestry group apart,^[166] and neither study compares outcomes of mixed with unmixed children or ranks groups.^[67,166] What they show is a cost of close kinship, not a bonus for distance; whether a child of two populations gains anything over a child of two unrelated parents from one population is untested in the studies found.

Side story Homer Plessy, New Orleans, 7 June 1892 Seven-eighths white by his petition's arithmetic, arrested because the other eighth was the law.

Homer Plessy, a shoemaker of mixed descent, bought a first-class ticket on the East Louisiana Railroad from New Orleans to Covington and sat in the whites-only car. The conductor, J. J. Dowling, asked, "Are you a colored man?" and told him to move. He refused, and Detective Christopher C. Cain arrested him (Plate 28). The eighteen founders of the Comité des Citoyens, Louis A. Martinet and Rodolphe Desdunes among them, had him released on bond that evening. The Supreme Court ruled 7 to 1 against him on 18 May 1896; he pleaded guilty eight months later and paid a $25 fine. He worked as a labourer, a warehouse worker and a clerk, then as a collector for a Black-owned insurance company, and died in 1925. Louisiana pardoned him on 5 January 2022.^[220] The court record calls him "seven-eighths Caucasian and one-eighth African blood" and says the mixture "was not discernible in him"; in the historian K. W. Medley's line, he was "white enough to gain access to the train and black enough to be arrested for doing so".^[205,220] Where being unmixed carried a premium, a state built it. Plessy's case is a state drawing a line that one eighth placed a man across.

A category, not a people, and neither state worth more as a trait, so each family decides alone. Two children of American fathers, born in South Vietnam during the war, show the two answers.

12. Stay or leave: which line thrives?

Vietnam, 1985. Kien Nguyen, 17, boards a plane under the Orderly Departure Program, "the first kid to get on the plane"; by 1998 he holds a dental degree, and by 2005 he is a published writer in California.^[349] Bi Thi Loan, born in the same war to a Black American soldier, stays. At about 39 she lives in a cemetery in Saigon, ten people on about $38 a month.^[290] One cohort, two doors, and the whole question in two lives. The pair is an illustration, not a comparison; Kien's family was wealthy and his mother a banker.^[243] The comparison is the 377 narratives from 286 Amerasians, relatives and community members collected in 2017, the first study to put stayers beside leavers, whose authors note that "no comparative data has been collected" before,^[197-198] and the thirty-year follow-up of the Homecoming wave.^[175]

"Thrive" here means six things: income, schooling, health, well-being, belonging, and identity continuity, meaning language and label. The premise needs a check first: staying does not keep a family unmixed, since 2.2 percent of births in Japan have a foreign parent and 5.6 percent in Korea are born to multicultural families (section 2.4), and leaving does produce a mixed line that fades within two or three generations (sections 9 and 11).

Income. The only experiment is a lottery. New Zealand drew Tongan applicants by ballot, so winners and losers differ only by luck. Winners who moved earned 263 percent more within a year and almost 300 percent more a decade on; ordinary comparisons overstate such gains by 20 to 82 percent, because leavers are selected.^[124,218] Place alone is worth a multiple: a Vietnamese man with nine to twelve years of schooling earned about six to eight times as much in the United States as an equivalent worker at home, in 2000s data.^[69] The second generation keeps the gain: sons of immigrants from nearly every sending country, Haiti and Jamaica the named exceptions, climb further in the US income ranks than sons of the US-born who started at the same income.^[2] Verdict: leaving wins on income, by less than raw comparisons say; the gain reaches the household that goes, and those left behind lost on balance.^[123-124]

Side story Eddie Van Halen, Amsterdam to Pasadena, 1962 A leaver's child: the crossing came before the stage, and the first day at school was the price he named.

Eddie Van Halen's mother Eugenia came from the Dutch East Indies and met his father Jan, a Dutch musician, in Indonesia under Dutch rule. After the war the couple moved to the Netherlands, where Eddie and his brother Alex were born. In 1962 the family crossed to the United States by boat in nine days and settled near Pasadena, sharing a house with two other families. His mother worked as a maid and his father as a janitor while keeping up his music. Eddie spoke no English and was treated as a "minority" student at a school that was still segregated; he called his first day "absolutely frightening". "It was actually the white people that were the bullies," he said in 2017. He became the guitarist of the band that carried the family name (Plate 29) and died of throat cancer in October 2020, aged 65.^[239] If, as other accounts say, Eugenia was Indo-Dutch, her move is one of the 254,000 departures of Dutch-status people from Indonesia between 1946 and 1967,^[347] and the family part of a diaspora put at 124,000 outside the Netherlands by 2001.^[347] He is one selected case, not evidence about families: a leaver's child who thrived in the second country, and who named what the first day there cost.

Schooling. No experiment or family-line study reaches it; the lottery's children were too young to measure.^[124] The group-level evidence says leavers' children do better, partly through selection: the more selected the immigrant group, the higher its children's college rate.^[106] The standard account of how immigrants' children fare, segmented assimilation, says the path depends on what the parents brought and how the country received them, with three outcomes: into the middle class on the parents' schooling, upward through a tight co-ethnic community, or downward into the lower strata.^[265-266] The split shows beyond language, in national census-survey data and in two merged Southern California surveys, one of them the Children of Immigrants Longitudinal Study that supplies the paper's language data (section 9): children of Indian, Chinese, Korean and Filipino parents finished college at 64 to 70 percent against 18 to 19 percent for children of Mexican and Central American parents, with arrest and teen-birth rates two to five times apart.^[265] In its high-school wave, Vietnamese-origin pupils sat between the Chinese and the Mexican-origin pupils, a 3.03 grade average, 48 percent expecting an advanced degree and 5.5 percent dropping out, and children of immigrants as a whole dropped out at a third (San Diego) to a half (Miami-Dade) of their districts' rates.^[282] Reviews of the newer literature find little of the downward path and more explanatory power in class than in race.^[99]

The theory's proof case for the community path is Vietnamese. In Versailles, a poor New Orleans neighbourhood where over half the Vietnamese families were in poverty and the high school had 13 percent of pupils at or above the national median, Zhou and Bankston found most Vietnamese children doing well, and credited the ethnic community and the family as well as, and beyond, the parents' money or schooling.^[373-374] The same authors print the cost. In Add Health, Vietnamese teenagers were the least likely of three groups to agree that they had a lot to be proud of, 70 against 90 percent for white teenagers, and the most likely to report feeling depressed and fearful;^[374] and they name a bifurcation, with dropout rising and delinquency high, and a family split by war and resettlement one of three routes into it.^[374] The stayers' side has one measured cohort, the fifth anchor, mixed by birth and staying by default: Korea's college gap narrowed from 31 to 13 points between 2021 and 2024, partly perhaps by an older sample, and Taiwan's adult gap is not significant in the one study, with family income doing the sorting (sections 5.4 and 6.4).^[226,333] Verdict: leavers' children do better in group averages, with selection and the place they land doing part of the work, and the one stayer cohort is narrowing its gap at home; the achieving Vietnamese second generation reports the lowest self-evaluation of the three groups compared, a gap that cultural norms against self-praise may inflate.

Health. The lottery winners' children grew taller within a year, and also fatter.^[314] The wider literature is the immigrant health paradox. Across 96 studies and more than 15 million migrants, all-cause mortality ran about 30 percent below the destination population's, and 50 percent below for refugees.^[7] Part of it is selection, since people who move are healthier than people who stay: measured against US-born people who had also moved, the self-rated health advantage shrank by 28 percent.^[135] Part of it erodes: immigrant women reach American body weights within ten years of arrival, and men close a third of the gap in fifteen.^[10] In Canada's 78-study review the advantage is weaker in childhood and adolescence, the ages at which this paper's children move,^[350] and it is not universal: Asian migrants to western countries had more feto-infant deaths and preterm births than majority women,^[116] and among Asian Americans the mental-health advantage of the foreign-born was no longer significant once material and psychosocial factors were held constant.^[162] Verdict: mixed. The first generation arrives healthier than the destination population, partly by selection; the weight advantage erodes within ten to fifteen years in the country, the advantage is weaker at the ages a child moves, and it fails on some Asian outcomes.

Well-being. The experiment cuts both ways. Mental-health items improved, yet four years on the winners rated themselves less happy and less respected than they would have been at home; a decade on, the ladders showed no difference and mental health stayed better.^[124,315] Across 117 countries, immigrants' life evaluations track the country they live in, not the one they left, with a correlation of 0.96, in cross-sectional data.^[138] The stayer's baseline moves too: Vietnam's own score rose from 5.06 in 2015 to 6.43 in 2025, so the gap to the United States and Australia shrank from about two points to under half a point.^[137] Verdict: mixed; happiness falls at four years, a decade on the welfare ladders show no difference and mental health stays better, and Vietnam's gap to the destination has shrunk.

Identity continuity. The leaver's grandchildren speak the majority language (section 9) and drop out of the ethnic label (section 11); the Anglo-Indians who left India after 1947 integrated so well that one review calls it "invisibility", while noting much of the culture kept.^[275] The label also moves with the climate: among immigrants' children in California, the plain "American" label fell below 2 percent in the months after Proposition 187 while national-origin labels rose from 32 to 48 percent.^[282] Verdict: leaving loses the language in about two generations and thins the label by the third, 42 percent of Asian-origin children, 55 percent with one Asian side, no longer identified as Asian;^[101,284] no study measures the stayers' side, which the paper assumes keeps both.

Belonging. No study measures belonging on both sides; the 2017 narratives are the nearest comparison, and they are weak. The indirect evidence points both ways. The mixed child who stays in a society that prizes homogeneity carries the worst measured costs in this paper: children of multicultural families in Korea, a group that includes mixed children, met school violence at 10 against 1.9 percent and trailed by 31 points in higher education in 2021, 13 in 2024 (section 6.4).^{[192,226-227]} In the 377 narratives, those still in Vietnam tied their stress to poverty and those in the United States to ethnicity; post-secondary schooling appeared in 13.8 percent of the Vietnam narratives against 50 percent of the US ones, from self-selected samples that included relatives.^[198] One leaver wrote, "I grew up longing to be what I wasn't"; back in Vietnam in 1996, she "felt I had finally come home". A stayer in Quy Nhon wrote, "No one dishonoured me because I was an Amerasian. My life has become comfortable".^[198] Among the Homecoming wave, Amerasians, siblings and other refugee youth together, age at arrival decided how far the gap closed, and the newly arrived Amerasians carried more distress than their own siblings yet adapted as well as other Vietnamese.^[175,216] One stayer rose where he was stigmatised: Denny Tamaki, born in Okinawa in 1959 to an Okinawan mother and a US Marine who left before he was born, became the prefecture's governor in 2018.^[274] Verdict: no evidence either way.

Three generations, one family. The Ho Tung line is the paper's one reading of a family that stayed through the shock. The first generation held the niche.^[11,364] The second chose its label under the shock: Jean sent her children out on English-name passports and declined to claim her Chinese nationality for herself.^[127] The third dispersed: none of Clara's eighteen grandchildren married a Eurasian, the exodus of 1949 and 1967 followed, and a grandson died in London.^[58,199,305] Dispersal ties absorption as the library's most common third-generation outcome, 9 of 28 each (Table A3), and the family of Hong Kong's reputed richest man did not escape it.

What no study can answer. The evidence reaches a decade for lottery-assigned leavers and a century for group averages.^[2,124] Nothing follows one family line, stay against leave, for 75 years. The leavers' group averages are also biased by who still answers to the name, and the direction depends on the group. Among Mexican Americans the children of better-educated parents drift out of the label, so the group's averages understate it.^[100] Among Asian Americans the selection runs the other way, negative on education, so Asian-origin averages likely overstate the group, and correcting for attrition lifts Hispanic outcomes relative to Asian ones.^[101] For a Vietnamese or any other Asian line, the published Asian group averages are more likely an upper bound than a floor. The plain verdict: leaving buys income and, in group averages, schooling, loses the language in about two generations and thins the label by the third; health and well-being are mixed; what staying keeps, and belonging on either side, are unmeasured. In the paper's judgement from the Korea and Japan cases, the mixed child pays most where the majority treats mixing as a flaw. Loan's children "get picked on for looking black". Whether that sentence is still true in 2075 is the next section's question.

13. Futures to 2075: three scenarios

Two of the anchors have spoken about the next generation. Loan, in Saigon: "My children hate going to school here, because they get picked on for looking black".^[290] Miyamoto, to the BBC in 2015: "There will be more and more international marriages and there will be more mixed kids in the future".^[360] In a 2024 interview she said one of her sons had been greeted at a Japanese kindergarten open day with "a foreigner has come".^[298] Forecasts of this kind have a record. In London in 1911 the director of Brazil's National Museum told the First Universal Races Congress that within a century Brazil's population would be "in the main" white and its métis extinct.^[188] By the 2022 census pardos were 45.34 percent of Brazilians and the largest group.^[155] This section sketches three futures to 2075 as lives (Figure 23), each tied to the anchor line it most resembles, and attaches no number of the paper's own to any of them.

The official projections are few. The US Census Bureau's 2017 series put Two or More Races at 6.2 percent of the population by 2060; the 2020 census then counted 10.2 percent, because the race question changed, so that projection is already overtaken.^[165,342] The United Nations counts 304 million international migrants in 2024, 3.7 percent of humanity.^[341] East Asia starts low: Korea's foreign-born share was 3.5 percent in 2024 and Japan's 2.8.^[341] Japan's own institute projects 87 million residents in 2070, with about 10.8 percent non-Japanese by the paper's subtraction of two printed figures.^[237] The paper found no projection of mixed-heritage children outside the United States.

Scenario A, the mainstream widens. Mixed children fold into the majority, as Alba argues; the category grows in births and shrinks in self-label.^[342] If 55 percent of third-generation children with one Asian side keep not identifying as Asian, each cohort of mixed births loses about half its members from the Asian column within two generations.^[101] This is the future of Clara Ho Tung's eighteen grandchildren, none of whom married a Eurasian, and, if group labels fade as they do elsewhere, of the Amerasian cohort's unmeasured grandchildren.^[58,100] For the child it is the quiet future: an international school whose pupils are now, by one relayed estimate, 80 percent local,^[44] and a local answer to "where are you from?" that the questioner accepts.

Scenario B, the category hardens. A label, a box on a form, and organisations keep a boundary. The Metis are the model: 204,115 in 1996 to 624,220 in 2021, a tripling that births cannot explain.^[312] Singapore's double-barrelled race and an Association that, one member says, recruits "New Eurasians" are the smaller versions (sections 10.9 and 11).^[276-277] For the child this is the future with a name and a club, Miyamoto's "hāfu" turned from a concession into a banner.

Side story Louis Riel, Red River and Regina, 1869 to 1885 The Metis leader hanged for treason in 1885 and named Manitoba's first premier in 2023: Scenario B across 138 years.

Louis Riel (Plate 31) was born on 22 October 1844 in the Red River Settlement. On 8 December 1869 he established a provisional government at Red River; the Manitoba Act of 1870, negotiated by the delegate Abbé Ritchot, made Manitoba a province. In March 1885 he formed a second provisional government at Saint-Laurent, Saskatchewan. The rebellion ended at Batoche on 12 May; he surrendered on 15 May, and on 1 August a Regina jury found him guilty of high treason "with a recommendation of mercy". He was hanged at the North-West Mounted Police barracks in Regina on 16 November 1885. His last words, as Charles Boulton recorded them, were "Remerciez, Madame Forget", thanks to the woman who had lent him a crucifix. By the Manitoba Metis Federation's account, the province's Louis Riel Act, which recognises him as its first premier, received royal assent in December 2023.^[327] The Metis count that tripled between 1996 and 2021 runs alongside that arc (section 10.4 and Scenario B above).^[312] A boundary held, through a court test and a line in law, long enough for a hanged man to become a premier.^[72] Of the five library groups that persist as named groups, the Metis show Scenario B most plainly, and none of the mixed categories of today has its 1870.

Scenario C, the pendulum. The marriage-migration booms of East Asia have already reversed: Japan's international marriages peaked in 2006 and Korea's share in 2005 (section 2.4), and Taiwan's foreign marriages peaked near 54,600 in 2003.^[66,225,228] The mixed cohort ages as a one-generation bulge, as the Amerasian cohorts did. The return leg is already visible in Can Tho: 206 children of returned marriage migrants in 45 families, 136 holding only a foreign passport.^[182] For the child this is Loan's children's future with a better school: in Okinawa, five mothers founded one in 1998 so that half-Japanese, half-American children could "explore their identities free of judgement".^[8] The stayer's country also changes under her.^[137]

Three indicators would tell the scenarios apart by 2040: the self-label share in a census, falling in A and rising in B; the intermarriage rate of the mixed generation itself, high in A and lower in B; and heritage-language survival in mixed homes, under two generations in A and C and longer in B. None is measured today for mixed-heritage children outside the United States, and none of the three futures is chosen by a census. Families choose, one at a time, with the evidence below.

14. Playbook for the person, parents, schools, and employers

The four families already made their choices, and each can be graded against the table below. Miyamoto's family let her go to her father in Arkansas in her mid-teens, and she came back and named herself (row 1, grade C).^[92,298] Robert Ho Tung funded a school and club for Eurasian children in East London (row 9, grade C).^[199] Loan went back to school at about 39 (no row).^[290] Obama's mother taught him English before dawn, and he started at Punahou (no row).^[248] Table 14.1 lists fourteen recommendations, graded. A means pooled or causal evidence in the right population; B means causal or replicated evidence in a nearby population (minority students, immigrants, bicultural adults); C means observational or small-sample evidence; D means expert opinion only. No row reaches A for mixed-heritage children (Figure 24).