Linguistic Life Expectancies: Immigrant Language Retention in Southern California Author(s): Ruben G. Rumbaut, Douglas S. Massey, Frank D. Bean Source: Population and Development Review, Vol. 32, No. 3 (Sep., 2006), pp. 447-460 Published by: Population Council Stable URL: http://www.jstor.org/stable/20058899 Accessed: 22/04/2009 12:03 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=popcouncil. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.
Population Council is collaborating with JSTOR to digitize, preserve and extend access to Population and Development Review.
http://www.jstor.org
Linguistic
Life Expectancies: Language
Immigrant
in Southern
Retention California G. Rumbaut
Rub?n Douglas
S. Massey D. Bean
Frank
research
in the view that Latin Americans supports generally from US data the Studies assimilate using linguistically. Census and other official statistics (Alba and Nee 2003; see also Alba et al. as well as investigations based on longitudi 2002; Bean and Stevens 2003), of immigrants the children nal surveys conducted among (Portes and Empirical
the United
States
if not out reveal a quick shift to English 2006), fluency, and the second first between Nevertheless, generations. right dominance, retention research on immigrant has been hampered language by a lack of
Rumbaut
data
2001,
on
use language the children
on
focus between Bureau no
first and eliminated
or ability broken that down by generation. Surveys a contrast of immigrants, by definition, permit only because the US Census second generations. Moreover, on parents'
the question to distinguish
generations possible into crude native-born/foreign-born In his controversial book Who Are We?
longer searchers
place
of birth
using
census
after data,
1970,
it is
forcing
re
comparisons. to America 'sNa The Challenges argued that the arrival of Latin
tional Identity, Samuel P. Huntington (2004) in numbers American of the immigrants large during the last three decades core of American in twentieth threatens the and culture century identity to immi Latin American the twenty-first century. According Huntington, are much
less likely to speak English than earlier generations of Eu a common because all ropean they speak immigrants language; they are concentrated and within residentially regionally segregated Spanish-speak in linguistic and cultural assimilation; ing enclaves; they are less interested in this lack of interest by activists who and they are encouraged foment
grants
identity
politics.
POPULATION
AND
He
is particularly
DEVELOPMENT
pessimistic
REVIEW
32(3):
about
447-460
the prospects
(SEPTEMBER
2006)
of Mexi
447
448
Life
Linguistic
Expectancies
can American
"If the second generation assimilation: does not re linguistic out to is of the third also be hand, ject Spanish likely generation bilingual, in both languages of fluency is likely to become and the maintenance insti in the Mexican-American tutionalized (2004: 232). According community" is no Americano dream. There is only the American an will share society. Mexican-Americans by Anglo-Protestant in that dream and in that society only if they dream in English" (ibid.: 256). was thesis dismissed in scholars, Huntington's widely Although by of authors and the this Bean, Brown, 2004; (see Massey study cluding it nonetheless achieved Rumbaut and 2006), widespread public diffusion to Huntington, dream created
"There
use of two in many circles. In this article we make tacitly accepted to test assertion of retention surveys among Huntington's linguistic as well as Asian of Latin American persons origin, by far the two largest sources of immigration to the United States over the past 40 years. Data on
has been new
to which the degree immigrants tional cohorts are able to speak used
to derive
"survival
linguistic
and
in different
their descendants
their mother
and actually
genera do so are
tongue across the generations. These sur we can apply life table meth to which
curves"
curves yield "mortality rates" to develop life of gen average number expectancies"?the "linguistic can a mother to in survive the States be erations United tongue expected
vival ods
after
the arrival
of an immigrant.
In doing so, we hope of understanding way
an intuitively appealing a threat to the continued constitutes
lic with
way the United
States.
This
exercise
is not
carried
predominance
to provide the pub in no that Spanish of English
within
out as a technical
analysis of a survival pro it adapts a well-known demo
sense. Rather, demographic a to make heuristic that those who worry about technique point: graphic from because of balkanization immigration heavy Spanish-speak linguistic use of Spanish dies out rapidly to fear, because have nothing ing countries con even in the area of highest Hispanic across the generations, immigrant cess
in the usual
centration
in the United
States.
a region California, the country's only adjacent largest net but one that also con receiver of immigrants 1970-2005, during the period of Mexican and tained more persons origin than any other Spanish-speakers a area of level and residential segre rising Hispanic displayed megalopolitan and Steinmetz 1987; Iceland, Weinberg, 2002). (Massey and Dent?n gation The
we
surveys to the Mexican
use were
border
conducted that was
in Southern
not
in the United States resided in By the year 2000 one of every five immigrants Los counties the region's six contiguous Ventura, (San Diego, Orange, Angeles, communities of Mexi San the and Riverside, Bernardino), including largest cans, Salvadorans, Koreans, Taiwanese, Guatemalans, Filipinos, Vietnamese, of origin. In the Los An outside of their countries Iranians, and Cambodians area alone, according to Current Population Survey es geles metropolitan 5 mil exceeded timates, by the year 2000 the Mexican-origin population
Rub?n
G.
Rumbaut
/ Douglas
S. Massey
D.
/ Frank
449
Bean
some 2.2 million lion persons, born in Mexico, including the United States of Mexican-born and another parents, or higher. third generation In the huge television market
2 million million
born
who
of Greater
in
were
Los An
in the summer
of 2005, nine of the ten most-watched in Spanish by KMEX, the Uni vision broadcast
geles
pro prime-time channel (see L?pez reasons our analysis offers a "hard test" of Huntington's hy If pothesis. speaking Spanish does not persist across immigrant generations in the urban corridor from San Diego on the Mexican border to stretching grams were 2005). For these
Los Angeles, then it probably out the United States.
Data We
not persist
will
in other
communities
through
and measures draw
our data
from two sources:
the Immigration and Intergenerational inMetropolitan Los Angeles Mobility (IIMMLA) survey, and the third wave of Immigrants of the Children Longitudinal Study (CILS) in San Diego. For of this two were the sets data since purposes analysis merged (N=5,703), on are based they representative samples of respondents evenly divided by sex, of the same approximate and age (28.6 years for IIMMLA respondents 24.2 years for CILS) and national Guatema Salvadorans, origin (Mexicans, and Koreans make of lans, Filipinos, Vietnamese, Chinese, up 78 percent the merged and other Latin American and Asian nationalities 10 sample, at about who were the same time (IIMMLA in 2004, percent), surveyed CILS in 2001-03) in the same metropolitan South region (the six contiguous ern California The sets data reflect the diversity of con counties). merged and refugees, laborers and profession temporary immigration (immigrants als, documented the least-educated
and undocumented) and and poorest immigrants
include
of significant subsamples from Latin America (particularly and Southeast Asia (especially
from Mexico, El Salvador, and Guatemala) from Laos and Cambodia). The focus in both surveys was on patterns of adult children guage assimilation, who were born abroad and arrived generation) ents (the
and those who second
were
generation).
of adaptation, lan including of contemporary those immigrants?both in the United States as children (the 1.5
born The
in the United
two
surveys
States
used
of immigrant par measures of
identical
and non-English and preference, and of other language proficiency two variables. we sets the data thus gain larger sample By merging for significant and greater precision and reliability for our subgroups
English relevant sizes
estimates
of linguistic
The
IIMMLA
The
IIMMLA was
random
life expectancies
by group
and generation.
survey
samples
a telephone in 2004 among survey conducted targeted of 1.5, 2nd, and selected 3rd and higher generation adults
450
Life
Linguistic
in the
Los Angeles metropolitan San Bernardino, Orange, Riverside, Angeles, et al. 2005). For purposes of sample design, five-county
area, which and Ventura
Expectancies
Los encompasses counties (Rumbaut
were eligible adult immigrants as "1.5 generation" if they came to the United States to live before if they were the age of 15; as "2nd generation" born in the United States as and had at least one parent who was and "3rd+ genera foreign-born;
defined
if both
tion"
they
and
their
foreign-born Before
parents
were
US-born
but had
one
or more
grandparents. the start of interviewing, for eligible respondents established
were
area, placing
special
emphasis
for ten ethnic strata targeted quotas to 20 in 40 years the five-county aged on the largest and most significant group?the
The IIMMLA also sampled a strategic handful of Mexican-origin population. and refugee origin-groups to that were other large immigrant differ expected in their modes of incorporation into US society, including Chinese, Filipinos, and Vietnamese, with and Guatemalans Salvadorans taken Koreans, along a separate stratum for 1.5 and assigned together. All groups were sampling 2nd generation and of 3rd+ respon respondents targeted quotas generation forMexicans, dents were also established whites non-Hispanic (hereafter sim and non-Hispanic blacks since they (not used in this analysis, ply whites), consist
overwhelmingly
speak English closed-ended dents: with
about
only).
of fourth or higher The final design called
telephone 3,500 with
interviews
with
African Americans who generation for completing 4,700 approximately random samples of eligible respon
1.5 and 2nd generation
3rd+ generation respondents. Multi-frame sampling procedures
were
respondents used
and around
to improve
1,200
the chances
of finding and interviewing of targeted populations. members The first stage screen to in the used random and households (RDD) sample digit dialing was com to IIMMLA and this the able area, approach using five-county For for and blacks of all Mexicans, whites, generations. plete sample quotas were
rates using RDD until the incidence compiled more At of eligible became low. this spe point, respondents prohibitively were RDD frames cific geographic and race-ethnic used, targeting sampling areas and those on lists of to households in high-density Asian residential other
groups,
samples
surnames. and Vietnamese Chinese, Filipino, Korean, in English or Spanish using a computer The surveys were administered com In total 4,655 interviews were assisted system. interviewing telephone in April 2004 and its con interviewing were from derived 2,822 (61 percent) RDD while interviews 1,833 (39 percent) using solely first-stage sampling, To achieve from interviews the augmented resulted this, using samples. were numbers dialed at least once, different 263,783 telephone including frame and 140,799 from 122,984 listings from the first-stage RDD sampling in These the identification of calls resulted the augmented 10,893 samples.
the start of full-scale pleted between 2004. Of these, in October clusion
Rumbaut
G.
Rub?n
/ Douglas
S. Massey
/ Frank
D.
451
Bean
of one of the ten targeted sample adults meeting the eligibility requirements were to Efforts made interviews with 8,815 of these adults complete subgroups. cases the for the had quota (in 2,078 subgroup already been filled). The num an average in ber of questions asked varied by generational status, yielding length of 27 minutes in the 2nd generation,
terview for those
The CILS
32 minutes for 3rd+ generation respondents, and 34 minutes for those in 1.5 generation.
survey
For more
than a decade the Children of Immigrants Study fol Longitudinal the progress of a large panel of youths several dozen representing areas in two main of immigrant in the United States: nationalities settlement lowed
Southern
California
(San Diego)
and The
Lauderdale
South
Florida
baseline
(the Miami conducted
and
Fort
in spring
survey, area). metropolitan students in the eighth and ninth grades enrolled 1992, interviewed eligible in the San Diego Unified of all schools School District (N=2,420). (A paral
lel sample was drawn from the Dade and Broward County Unified but we the South The Florida Districts, ignore sample.) sample was in the junior high school grades, where out school is of dropping rates between avoid the potential bias of differential ethnic dropout that are found
at the senior
school
level.
Students
were
School drawn rare,
to
groups to en eligible
high if they were US-born but had at least one immigrant (for or were if they themselves and had come parent, eign-born) foreign-born to the United States before bal age 13. The resulting sample was evenly anced between males and females and between and US-born foreign-born ter the
sample
of immigrants. the principal
the geographical of recent im Reflecting clustering nationalities in the San Diego migration, represented sample are Mexican, Viet (as is largely the case in the IIMMLA sample) Filipino, and smaller groups of other chil namese, Laotian, Cambodian, Chinese, children
dren of immigrants from Asia and Indian) and Japanese, (mostly Korean, Latin America of the countries of Central and South (most Spanish-speaking America and the Caribbean). Three immigrants interviewed
of later, a second years survey of the same panel of children was were this conducted. time the who By youths, originally 14 or 15 years old, were now when most were 17 to 18 years
old and had
reached the final year of high school out of (or had dropped The follow-up in San Diego in reinterviewing succeeded survey school). of the baseline identical propor almost 2,063 or 85.2 percent sample, with tions of males and females, of native-born and foreign-born of US youth, citizens and noncitizens, and of main nationalities. There was a slight ten for children from intact families to be over dency (both parents present) represented insignificant
in the follow-up survey; all other differences 2001: 25-31). and Rumbaut (Portes
were
statistically
452
Linguistic
a decade
2001-03, During conducted. The
was
after now
Life
Expectancies
the original survey, a final follow-up 23 to 27 years of age, from ranged
respondents or resi to be contacted in their places of work individually dence. Tracking the sample after an interim period of six to seven years was in our data files of infor made possible by two factors: first, the availability on mation Social Security numbers, birth dates, and last known addresses and most
had
of respondents and their parents; to conduct confidential searches
the rise of Internet second, on the basis of this type of methods. Mailed questionnaires
services
able
information,
(which in by other retrieval supplemented on were and preference) cluded detailed questions language use, proficiency, source of completed the principal data in this third survey. Respondents were when also interviewed visited trained interviewers by phone possible; no telephone were numbers for whom respondents or known address that of their parents was available.
last but whose known, a more Over period of or of fieldwork, retrieved CILS-III in San Diego than 24 months complete on 70 82 of the and of information percent percent original sample partial the first follow-up sample. focus on
We have
the
1,502 cases from the San Diego sample for which we data over the span of a decade. Unlike the first follow
survey complete the time elapsed effects of sample attrition were negligible, indicate last two surveys and the significant sample attrition for adjusting results for sample selection bias. Family composition
up, where tween the
be
need
and
the
were of presence/absence in the chief predictors early academic performance runs that in San Diego. the CILS-III sample indicate, however, Preliminary for this from those unadjusted do not differ significantly averages adjusted outcomes source of error, specifically with that are the respect to language focus of this analysis. 2005.) (For details on CILS-III, see Portes and Rumbaut we defined across the generations, To analyze gen linguistic variation the approach of Rumbaut erational (2004). Those born following categories the United
outside horts
on
based
arrived
as adults
of those who the United vided
into
States
comprise at arrival: age we (whom ignore
their
to those who in arrived (here restricted is also di second generation States before age 15). The US-born were born in the two cohorts: members of the 2.0 generation arrived
United
States
of two
eration
were
born
US-born
into two co the first generation, divided of immigrants who the 1.0 generation in this analysis), and the 1.5 generation
parent.
as children
of the 2.5 gen members parents, whereas States of one foreign-born parent and one two consists of US-born third generation persons with
foreign-born in the United
The
we distinguish the 3.0 cohort among whom (those with parents, 3.5 from the cohort three or four foreign-born (with only grandparents) one or two foreign-born those in the fourth genera Finally, grandparents). were all born in the and grandparents whose tion are respondents parents States. United US-born
G.
Rub?n
/ Frank
S. Massey
/ Douglas
Rumbaut
D.
453
Bean
to generation 1.0 versus 1.5, or 2.0 some to exact in a stan is identical another age for heuristic that gen dard life table. Rather, purposes, we adopt the notion as used here constitute erational of intervals representations meaningful do not
We
versus
2.5
time
in the
assess
how
life of a foreign language. We use of foreign languages long
States
United
that belonging to moving from
claim
were
intervals
if these
1 shows
except
sense
knowingly adopt retains its historical
as exact
years to fiction
a useful
reputation
as a
and CILS respondents (N=5,703) in this used groups by generation analysis (all and European non-Hispanic blacks). For all groups except Mexicans is so recent that sampling is infeasible the 2.5 immigration beyond
Table
whites,
in the same
equal
of age or precise birth intervals. We a point: States make that the United for languages." "graveyard broken
to apply life table methods to in last the could be expected then
of IIMMLA
the number
for the main
down
more for those seven groups without than exception, in the United total population States is foreign-born, to the US-born and of the remainder all belong second generation. nearly For those groups and their descendants in Southern of California, members Indeed, of their
generation. 70 percent
have not yet been born, and members of the third generation or are in and in small number still childhood. Thus, Mexi generation infancy cans offer the strongest test of Huntington's and clearly, by his hypothesis; of their situation size in the United and population States, frequent mention the
fourth
In total, the merged the group most salient in his mind. IIMMLA they were and CILS data set used in this analysis contains of Mexi 1,642 respondents can origin above are All of the cell sizes for Mexicans the 1.0 generation. large enough
TABLE sample
to provide
1 Numerical by
population
robust
estimates
distribution and
group
of the IIMMLA-CILS
White European Mexican Salvadoran-Guatemalan
81 423
83 578
181
182
93
Filipino Vietnamese
411
446
Chinese Korean
235
Other Asian
232
434 257
Total 2,345 NOTE: For definitions
of generational
cohorts
merged
cohort 2.0
91
Other Latin American
life expectancies.
generation
Generational 1.5 Group
of linguistic
2.5
18117 240
3.0
3.5
4.0
108 164
291
698
48
189
1,642
66
272
480
Total
380 17
240 56
126 983 148 590 8 170 28 433 18 133 408 70 32927
1,903 see text.
637
5,703
454
Linguistic
Linguistic We
survival
measure
to two
tongues using answers a respondent spoke the answer did not "very well"
of immigrants' mother first asked how well
The
survey questions. of his or her ancestors, and those who language were of a linguistic the equivalent death assigned sense in the the mother "dead" the that tongue to speak
ability
then "English," cause itwas no Those
two
was
longer criteria
the intimate
also
We
consider
has
lost
the the
language If the respondent answered to have considered "died" be
the mother
tongue used within
certificate.
respondent asked which
it fluently. The second question to speak in the household. preferred
respondent
Expectancies
curves
"survival"
the
Life
of family life. of language death. Other
confines
are reasonable
predictors of immi show that the children surveys more or a non are to to much lose their read write grants likely ability once to than their and in a that, ability speak it, English language literacy more ismuch level of fluency dies, the remaining language likely to dimin rare. Moreover, ish over time and bilingualism becomes it is in increasingly data
from
the home
the IIMMLA
where
and CILS
a non-English
with
mother who
tongue arrived
parents immigrant especially in Southern and higher generations California, and children ers, close friends, and even spouses in place English. 1 shows Figure criterion. The x-axis
linguistic
survival
is most
to be used, the 1.5 among
likely
as adults; communication
curves
cowork
with
is far more
likely
to take
to the first according in the United States in in
defined
spent specifies generations of group members of 0.5, and the y-axis indicates the proportion still speaking the mother well?that the is, among tongue very proportion Given the sheer number and den whom language fluency has "survived." crements
in Southern and the long his California, sity of Spanish-language speakers survival curves settlement, we would expect the generational tory of Mexican to of and white for Spanish-speaking be above Asians those groups Europe from 1.5 to 2.5 ans, and this is indeed the case. At each generational point is higher than the proportion the proportion Spanish speaking any speaking 29 In generation of Mexicans, 2.5, 35 percent tongues. 13 Latin and of other and Guatemalans, percent percent In contrast, the proportion Americans still speak Spanish very well. speak 2.5 not 6 per in does exceed well the mother very tongue generation ing of the other mother
of Salvadorans
cent
for any other group. In the third generation and beyond, we can only compare Mexicans of the mother the strong retention whites. and European tongue Despite 2.5 with among Mexicans percent (35 just 3 generation compared through curves con to thereafter the survival of white percent Europeans), begin verge. At generation ish and at 3.5 the fourth
generation,
17 percent of Mexicans still speak fluent By the time we arrive drops to 7 percent.
3.0 only
figure the proportion
of Mexicans
who
speak
Spanish
Span at the
very well
G.
Rub?n
/ Douglas
1 Proportion
FIGURE mother
Rumbaut
)ther
of immigrant well
very
tongue
S. Massey
by
/ Frank
D.
who
group members
455
Bean
speak
generation
Latin Americans
-Salvadorans-Guatemalans
0.4 0.3 Koreans
0.2
Vietnamese
0.1 0
2 2.5 Generation
is just speak death today fluent
3
in the United
States
5 percent, to around 1 percent of white who compared Europeans In other words, their mother the linguistic tongue very well. given rates prevailing in Southern California, Mexican arriving immigrants can expect
only
5 of every
100 of their great
to speak
grandchildren
Spanish.
Linguistic ability scendants of Mexican
is not
linguistic use, however, retain the basic immigrants
some de and although to ability speak Spanish, If they prefer to speak En
in most they may prefer to use English settings. at in other for example, home, glish they are not likely to prefer Spanish to the social situation appears settings and probably will only use it when shift. Figure 2 presents survival require a linguistic a occurs of the mother when tongue respondent
curves where
in a particular at home. Although,
to speak English that he or she prefers to this definition, Mexicans and Central Americans cording curves other groups, elevated survival play compared with states
the "death" generation even ac
continue
to dis
they no longer Latin Americans dis
as visually in the graph, and other distinct or Koreans. that is indistinguishable from that of Vietnamese play 96 percent Even among Mexicans, by the third generation prefer to speak at home. English stand
out
a curve
to ex the graph levels off, with just 3 percent continuing a Put another way, for Spanish. is 97 per the probability press preference cent that the great grandchildren of Mexican immigrants will not speak Span cannot of Mexicans in Southern retain California ish. If the vast majority Thereafter
456
Life
Linguistic
FIGURE 2 Proportion of immigrant group members mother tongue at home by generation
White
who
Expectancies
speak
Europeans
Filipinos Salvadorans-Guatemalans Mexicans
2 2.5 Generation
3.5
3
in the United
States
or a preference use beyond in Spanish for its household in the United then its survival prospects elsewhere generation, fluency
probably nation's
at least as dim.
Contrary
largest
to Huntington's enclave, within
In order
third are
States
even in the assertions, a border region that his
Spanish-speaking to be well to Mexico, appears Spanish torically belonged of US residence. natural death by the third generation
Generational
the
on
the way
to a
life expectancies
to compare
the
survival
among groups using a simple to compute life table methods in Figures vival curves shown
and easily
for different
mother
tongues metric, we employed interpretable based on the sur life expectancies
prospects
linguistic 1 and 2. In doing so, we follow a hypotheti as they "age" across the generations cal cohort of ethnic group members rates prevailing in Southern Califor the linguistic mortality and experience to the IIMMLA and CILS data. Rather than a person nia according aging incre in terms of half-generation here is measured year to year, duration ments. A language is "born" in the United States with the arrival of first over time to the extent that and it then survives generation immigrants, in subsequent generations people their households. and use itwithin
continue
to retain
the ability
to speak
it
G.
Rub?n
We
Rumbaut
/ Douglas
can extend
guistic Preston, ancies"
life and death
course,
are for Mexicans
/ Frank
S. Massey
the analogy
between the classic
D.
457
Bean
human
life and
formulas
death
of the
by applying to compute and Guillot 2000) Heuveline, "generational for the mother tongues spoken by different immigrant Southern California. The only complete survival generational life expectancies recent the most threshold
segment. erations and
fined
(see life expect in groups
and once
the curve
falls below
a survival
out the life table in the next half-generational of gen life number resulting expectancies give the average a foreign can to survive be within the cultural expected language of 0.05,
close
linguistic milieu The bar chart
of contemporary Southern California. 3 shows in Figure the life expectancies to this point. of the various studied origin groups
for the mother For each
group, is de life expectancy when death computed no occur to to speak the lan when the respondent longer prefers at home, and the right-hand bar shows the life expectancy computed or can be cannot it As the respondent he she well. very reports speak bar
shows
the
no mother of which is considered, definition tongue irrespective to survive beyond be expected the third generation given the linguistic now prevailing in Southern vival probabilities California. seen,
3 Linguistic by generation FIGURE 3.5
of
curves,
To permit of Europeans. computations we immigrant origins, linearly extrapolate
The
tongues the left-hand
guage when
lin
and white
other among half generation; we
and
life table
life expectancies
for selected
immigrant
groups
can sur
458
Linguistic
The most
Life
Expectancies
liberal
definition of linguistic to the ability life?retaining a language as opposed to a preference a life for its daily use?yields of 3.1 generations for Mexican 2.8 generations for the expectancy Spanish, Guatemalans and 2.6 and for that Salvadorans, Spanish spoken by spoken current Under the ability conditions, therefore, by other Latin Americans. to speak Spanish very well can be expected to disappear sometime between speak
the second
and third generation for all Latin American in Southern groups are even lower when Life expectancies life is defined by a pref erence use at In terms of daily use, Spanish can be expected for its home. to die out after 2.0 generations 2.1 generations Gua among Mexicans, among
California.
temalans
and Salvadorans, Asian groups, Among
and the
1.7 generations two definitions
for other
Latin Americans.
life and death linguistic do not yield very different life expectancies, and in some circum generally stances a at the mother home" tongue "speaks expec yields slightly higher no matter tation of life than the ability to "speak it very well." Nonetheless, can be which is considered, the average Asian group or definition language to die out at or near the second generation. The lowest life ex expected are observed
a former from the Philippines, among immigrants is The life for average English widely spoken. expectancy of is only around 1.3 genera tongue Filipinos (usually Tagalog) for the preference-based definition and 1.6 generations for the abil
pectancies US colony the mother tions
of
where
In general, life expectancies for Asian lan however, and in Chinese, Vietnamese, guages among (including Korean) immigrants Southern California 1.3 and 2.0 genera vary in the narrow range between ity-based
tions
definition.
of US
residence, observed
among is considered).
expectancies which definition
A graveyard
which
to the range of linguistic is comparable white (1.5 to 2.0, depending Europeans
life on
for languages
In this analysis we have tested Huntington's that Spanish assertion is un to other the of in States the United go way likely immigrant languages by to English-language across the generations. dominance South succumbing ern California offers an ideal test of his hypothesis it is the largest because some of the old in the United enclave States and houses Spanish-speaking as well est and largest Mexican as the in the country, neighborhoods sur of immigrants. We defined country's largest concentration linguistic a preference a mother vival in two ways: for speaking the tongue within household
and
the ability
Our
to speak that language very well. contradict assertions. Huntington's
directly findings States has aptly been described historical mother
ability tongues
as a "graveyard" for languages because of its the millions and their immigrants by extinguish a few generations and and Rumbaut (Portes 2006),
to absorb within
The United
Rub?n
G.
Rumbaut
S. Massey
/ Douglas
/ Frank
D.
459
Bean
to offer no threat to this reputation. to the number Owing Spanish appears in and density of Spanish Southern California, Mexi metropolitan speakers cans and other Latin American a retain greater ability to speak immigrants their mother with other but, by the very well groups, tongue compared at the latest, third generation for white the pattern observed
drops
Europeans. for speaking Spanish Latin American groups
as a preference Mexicans and other fined
Asians
ability
sharply However, at home,
and
toward converges is de when survival
the survival more
look much
curves
for
like those
of
the life expectancy of Spanish may in Southern its ultimate California,
and white
Europeans. Although be appreciably greater among Mexicans seems assured by the third generation. demise nonetheless Like taxes and seems a sure to in be death the United States, death, biological linguistic thing even
a city with in Los Angeles, in the world. populations
for Mexicans
living
urban ish-speaking This analysis carries estimated life table period
the
behavior
guistic
caveat
of the largest
Span
as any other study based on a that the lin data: it assumes
of today's second, forecasts the behavior
accurately Spanish
same
from cross-sectional
one
will
be
retained
more
third, and fourth generation immigrants It is possible of future generations. that no use in its is the future because readily
in schools; because continuous will create immigration longer stigmatized more to one's in the future; opportunities compatriots speak Spanish with or because At media will become Spanish-language increasingly prevalent. this point, however, after at least 50 years of continuous Mexican migra tion
into Southern
California,
Spanish
appears
to draw
its last breath
in the
third generation. The death
in the United of immigrant States is not only an languages a can as be of but also considered fact, part empirical larger and widespread or not this is of death" Whether global process (Crystal 2000). "language is another To the extent of course, that lan desirable, question altogether. is an asset and that knowledge of a foreign tongue represents guage fluency a valuable resource in a global economy, efforts to maintain immigrants' and pass it on to their children this part of their cultural heritage should not be discouraged. Without chances of sustaining fluent slim. Our
conclusions
thus
social structural the supports, however, seem in American communities bilingualism reverse the concerns and alarms often found in strong
to the proliferation call attention of foreign literature, which to to and the threat they pose Historical and languages English dominance. evidence indicates that English has never been contemporary seriously as the dominant States and that?with of the United threatened language the popular
well
over
200 million
threatened stead with
is the them
monolingual English speakers?it even in Southern California. What
today, not of the non-English survival to the United States.
languages
that
is certainly
not
is endangered
in
immigrants
bring
Linguistic
460
Life
Expectancies
Note of rethe support gratefully acknowledge search grants from the Russell Sage Foundation for the two studies on which this analysis
We
is based: tional
the
Immigration in Metropolitan
Mobility
and
and survey, (IIMMLA) Children of Immigrants
the
third wave
in San Diego. (CILS-III) ried out in Southern California
IntergeneraLos Angeles
of
the
Longitudinal Study car The surveys were 2001
during
04.
References Richard
Alba,
D.,
John
Stults. Lutz, and Brian Logan, Amy of the mother and preservation tongue 39(3): 467-484. immigrants," Demography
2002.
Loss
generation? temporary D. Richard
"Only English by the grandchildren
among
and Victor Nee. 2003. Remaking the American Mainstream: Assimilation MA: Harvard Press. temporary Immigration. Cambridge, University 2006. and Rub?n G. Rumbaut. "Mexican Frank D., Susan K. Brown, immigrant on Politics 4(2): and economic 309-313. incorporation," Perspectives
Alba,
Bean,
D.
Frank
Bean,
York:
Iceland,
and
the Dynamics
political
of Diversity.
New
Press. York: Cambridge University toAmerica's National The Challenges Identity. New
York:
2002. Racial and Ethnic Residential H. Weinberg, and Erika Steinmetz. Seg DC: US Census States 1980-2000. Bureau. Washington, "A Spanish cleans up in L.A.," The Los Angeles Times, 3 Septem soap opera
Daniel
John,
Massey,
Newcomers
and Con
& Schuster.
in the United
regation Steve. L?pez, ber,
America's
Language Death. New P. 2004. Who Are We?
Samuel
Huntington, Simon
2003.
of con
Foundation.
Sage
2000.
David.
Crystal,
Stevens.
and Gillian
Russell
third
the
2005.
p. Bl. S. 2004.
Douglas
of Who
"Review
Review
Development
Are We?
by
Samuel
P. Huntington,"
and
Population
30: 543-548.
1987. "Trends in the residential S. and Nancy A. Dent?n. of blacks, segregation 52: American Review 802-825. and Asians," Hispanics, Sociological G. Rumbaut. 2001. Second Gen and Rub?n Portes, Alejandro Legacies: The Story of the Immigrant of California Press and Russell and New York: University eration. Berkeley Sage Founda tion.
Massey,
Douglas
-.
"The
2005.
second
and Racial
Ethnic -. 2006.
Heuveline,
Processes. Population Modeling G. 2004. Rub?n "Ages,
Rumbaut,
first and
second
grant 1160-1205. Rumbaut,
Rub?n
Los Angeles,"
G.
and
the
Children
983-999. 28(6): America: A Portrait. 3rd edition.
Immigrant Samuel P., Patrick
Preston,
generation
of
Immigrants
Longitudinal
Study,"
Studies
York:
of California Press. University Berkeley: 2000. and Guillot. Demography: Measuring
Blackwell.
life stages, and generational in the United States," generations
et al. 2005.
Report
New
and Michel
"Immigration to the Russell Sage
cohorts:
Decomposing
International
and intergenerational Irvine: Foundation.
Migration
mobility University
the Review
immi 38(3):
in metropolitan of California.