Alejandro Portes Alejandro Portes
Richard Schauffler Richard Schauffler
The Johns Hopkins University
The language adaptation of second generation children is explored in
the context of the history of linguistic absorption and bilingualism in
America. Strong nativist pressures toward monolingualism have com?
monly led to the extinction of immigrant languages in two or three
generations. Contemporary fears of loss of English dominance are
based on rapid immigration during recent decades and the emergence
of linguistic enclaves in several cities around the country. This article
explores the extent of language transition and the resilience of immi?
grant languages on the basis of data from south Florida, one of the areas
most heavily affected by contemporary immigration. Results from a
sample of 2,843 children of immigrants in the area indicate that: 1)
knowledge of English is near universal; 2) preference for English is
almost as high, even among children educated in immigrant-sponsored
bilingual schools; 3) preservation of parental languages varies inversely
with length of U.S. residence and residential locations away from areas
of ethnic concentration. Hypotheses about other determinants of bil?
ingualism are examined in a multivariate framework. The relationships
of bilingualism to educational attainment and educational and occupa?
tional aspirations are also explored.
u Where linguistic unity has broken down, our energies and resources flow into ten?
sions, hostilities, prejudices and resentments. These develop and persist. Within a
few years, if the breakdown persists, there will be no retreat. It becomes irrevocable,
irreversible. Society as we know it can fade into noisy babel and then chaos."
?U.S . English policy statement
This paper examines the process of linguistic adaptation and the extent and
determinants of bilingualism among children of immigrants, the new
second generation spawned by accelerated immigration during the last
decade. The setting of the study is south Florida, one of the areas most
heavily affected by recent immigration. We review the findings in the
context of the history of linguistic conflict and language assimilation that
have accompanied U.S.-bound migration over the life of the nation.
The data on which this article is based were collected by the project "Children of Immigrants:
The Adaptation Process of the Second Generation," supported by the Andrew W. Mellon
Foundation, the National Science Foundation (grant No. SES-9022555), the Spencer Founda?
tion, and the Russell Sage Foundation. The chapter was written while the senior author was in
residence as a fellow of the Russell Sage Foundation, whose support, is gratefully acknowledged.
The authors are exclusively responsible for the contents.
640 IMR Vol. xxviii, No. 4image 1
Language and the Second Generation 641
The present wave of immigration was triggered by the 1965 Immigration
Act, as well as by subsequent changes in American asylum and refugee
policies. The overall direction of the new policies was toward greater
universalism, eliminating previous discriminatory racial barriers and open? universalism, eliminating previous discriminatory racial barriers and open?
ing the doors of the country on the basis of uniform criteria. Since 1965, ing the doors of the country on the basis of uniform criteria. Since 1965,
occupational skills, family reunification and fear of political persecution
have been the guiding criteria of U.S. immigration policy. As is well known,
the result has been a rapid increase in immigration and an equally rapid
shift in its origins from Europe to the Third World. Most of today's children
of immigrants thus have parents who came from Latin America and Asia,
and about half are phenotypically nonwhite (Portes and Zhou, 1993).2
The size of the new immigration has given rise to sizable language enclaves
in several U.S. cities and parallel nativistic concerns about these enclaves'
resilience. In particular, the growing use of Spanish by Latin American
immigrants has triggered gloomy assessments of the future prospects of
linguistic unity, as illustrated by the opening quote. Clearly, the decisive issue
that would validate or refute the nativists' fears is language use and prefer?
ences among the second generation. In the past, almost every first genera?
tion's loyalty to their ancestral language has given way to an overwhelming
preference for English among their children (Lieberson, Dalto, and
Johnston, 1975). The extent to which this process will repeat itself today
represents a central question, from both sociological and policy perspectives,
south Florida is one of the areas most directly affected by contemporary
immigration and the growing use of Spanish. Our sample of second genera?
tion students from this region will allow us to address the current scholarly
and public debate about the future of language in America.
HISTORICAL OVERVIEW
Language Assimilation in Perspective
The current controversy over language is best understood in the context of
a cyclical trend in the history of the United States since colonial days.
Descendants of earlier immigrants who had "dropped the hyphen" and
considered themselves plain Americans have often looked upon later arri?
vals as the source of potential cultural disintegration. This was true even
prior to the Revolutionary War. Benjamin Franklin complained as early as
1751 that German immigrants in Pennsylvania "will shortly be so numerous
as to Germanize us instead of our Anglifying them, and will never adopt
our Language or Customs, any more than they can acquire our complexion"
(Franklin [1751], 1959:234). The notion of "one nation, one language" was
2This figure is arrived at by adding the percentage of Latin American immigrants estimated
to be nonwhite on the basis of the 1990 census figures on the racial composition of the Hispanic
population to Asian and black immigrants.image 2
642 International Migration Review
often idealized as a state of linguistic perfection to which the nation should
return. This idea was discussed at length by philosophers of the seventeenth
and eighteenth centuries, including President John Adams, who contended
that "language influences not only the form of government, but the temper,
sentiments, and manners of the people" (Adams [1780], 1856 7: 249-251).
During the colonial and early independence period, the notion that the During the colonial and early independence period, the notion that the
country and its citizens were defined by a common language was justified country and its citizens were defined by a common language was justified
on two grounds. First, along with incipient American nationalism came the
idea that American English both reflected and constituted the democratic
and rational nature of the country. Second, the acquisition and use of
English was seen as the litmus test of citizenship. Lacking a common culture
or common history, the use of English became the essential part of "real"
Americanism (Baron, 1990). The two rationales were related insofar as the
ability to think logically, seen as necessary for a democracy, was only possible
on the basis of fluency in English.
The perceived necessity for "Good English" has taken many forms
throughout American history. In 1902, for example, New Mexico's state?
hood was delayed until, in the words of one prominent politician of the
time, "the migration of English-speaking people who have been citizens of
other States does its modifying work with the Mexican element" (Baron,
1991: 8). Nebraska banned teaching any foreign language to students below
the ninth grade in 1919 and organized formal "Good English" campaigns
from 1918 to the early 1920s. At the time, language loyalty oaths were
commonly extracted from school children (Dillard, 1985; Marckwardt, 1980).
In this and other ways, in schools and public life, monolingualism was
linked to the idea of democracy, national unity and allegiance to the country.
Although many parents of upper and middle class backgrounds encouraged
their children to learn Latin, French or German, bilingualism on the part
of recent immigrants was frowned upon. As today, that attitude was prompted
by the existence of large ethnic communities which lay beyond the pale of
the English-speaking population, out of sight but never out of mind.
The Shifting Implications of Bilingualism
During the early twentieth century, opposition to bilingualism derived
strength from the then dominant scientific wisdom. Academic studies in the
fields of education and psychology argued that bilingualism created failure,
mental confusion and damaged the psychological wellbeing of immigrant
children. Two schools of thought existed at the time: one which argued that
lower intelligence caused the failure of children to acquire English, and
another which argued the opposite. The first school (low intelligence: low
English) based its conclusions on beliefs about genetic differences between
races, arguing that heredity limited the ability of immigrants to learn. Theimage 3
Language and the Second Generation 643
second school (low English: low intelligence) based its conclusions on beliefs
about environmental factors, in particular the use of a foreign language at
home. Intellectual failure was imputed by this school to the "linguistic
confusion" of children exposed to two languages.
It was not until 1962 that these views were convincingly disproved by a
methodologically sound study of the effects of bilingualism on cognitive
ability. French- and English-speaking children in Canada were studied by
Peal and Lambert (1962) who demonstrated that, if social class was taken Peal and Lambert (1962) who demonstrated that, if social class was taken
into account, true bilingualism was associated with higher scores on a variety into account, true bilingualism was associated with higher scores on a variety
of intelligence tests {see also Cummins, 1981; Lambert and Tucker, 1972).
True bilinguals, defined as those who could communicate competently in
two languages, were shown to enjoy a greater degree of cognitive flexibility
and an enhanced ability to deal with abstract concepts than their monolin?
gual peers. Instead of creating "confusion," having two symbols for each
object enhanced understanding.
Subsequent studies have generally supported the findings of Peal and
Lambert's pioneer study. An analysis of a national sample of high school
students in the United States, for example, found a positive correlation
between academic achievement and bilingualism among Hispanic youth
(Fernandez and Nielsen, 1986). More recently, a study of San Diego high
school students also showed significant differences in academic perform?
ance between true bilinguals and monolinguals, as well as between true
bilinguals (defined by the local school system as Fluent English Proficient)
and semibilinguals (defined as Limited English Proficient). Again, true
bilingualism was shown to have a positive effect on scholastic achievement
(Rumbaut and Ima, 1988).
Despite accumulating factual evidence on the advantages of bilingualism,
the United States is unique in the rate at which other languages have been
abandoned in favor of English. Lieberson, Dalto and Johnston (1975)
provide evidence showing that in no other country have foreign languages
been extinguished with such speed. In the past, the typical pattern has been
for the first generation to learn enough English to survive economically;
the second generation continued to speak the parental tongue at home, but
English in school, at work and in public life; by the third generation, the
home language shifted to English, which effectively became the mother
tongue for subsequent generations.
This pattern has held true for all immigrant groups in the past with the
exception of some isolated minorities. As in previous periods of high
immigration, the fear of nativist groups is that the pattern is about to be
abandoned. Our analysis will address this issue. However, growing research
evidence about the cognitive effects of bilingualism indicates that the
obverse of that question should also be examined. That is, to the extent that
knowledge of two languages has positive effects, it is also important to inquireimage 4
644 International Migration Review
about the determinants of preservation of foreign languages. We explore
both angles - English fluency and bilingualism - in the following sections.
THEORY
Possible outcomes of the clash of languages confronted by second genera?
tion youths are fairly clear. They can be arranged in a continuum ranging
from full language assimilation (English monolingualism) to fluent bilin?
gualism to full language retention (monolingualism in the parental lan?
guage). Recent theoretical developments in the sociology of immigration
can be brought to bear on the analysis of these outcomes insofar as they
emphasize the significance of social class and social context in the adapta? emphasize the significance of social class and social context in the adapta?
tion of immigrant groups. Clearly, newcomers from more advantaged tion of immigrant groups. Clearly, newcomers from more advantaged
educational and occupational backgrounds tend to do better on the aver?
age, but often individual resources interact with the social context that
receives them. Hence, immigrants who face an unfavorable governmental
or societal reception may find their human capital skills seriously devalued,
while those in the opposite situation may put their individual resources to
full use. In addition, those who arrive into large and economically diversi?
fied coethnic communities may advance rapidly through use of the social
capital that community networks make available (Massey, Goldring, and
Durand, 1994; Portes and Sensenbrenner, 1993).
These general notions translate into certain expectations concerning the
linguistic adaptation of second generation youths. Children growing up in
sociocultural contexts where the native English-speaking majority is domi?
nant or where immigrants from other linguistic backgrounds are most
numerous will experience a faster process of home language loss and a rapid
conversion to English monolingualism. Conversely, those raised in contexts
where a large conational concentration exists will have greater probability
of parental language preservation. In such instances, there will be a clear
economic incentive to retain proficiency in that language, along with
greater facilities for learning and practicing it within the community. The
predicted outcome will be widespread bilingualism. Parental socioeconomic
background will have contradictory effects on bilingualism because, while
educated and wealthier parents may wish to transmit their language, they
will also make available more opportunities for their children to enter the
cultural mainstream. The prediction, in this case, is of a positive effect of
family socioeconomic status on English proficiency along with an insignifi?
cant effect on parental language retention.
Finally, the passage of time will inexorably lead toward greater English
proficiency and English preference and gradual abandonment of the im?
migrants' tongues. In this case, we draw on the American historical record
to anticipate that, regardless of the size and economic power of the coethnicimage 5
Language and the Second Generation 645
community, the trend over time will be away from bilingualism. These
arguments can be summarized in the following three hypotheses:
1. Language assimilation (English monolingualism) among the second
generation will vary directly with demographic dispersion of the
immigrant group and with length of U.S. residence.
2. Bilingualism will vary directly with demographic concentration and
economic diversification of the immigrant community and inversely
with length of U.S. residence.
3. Parental status will lead toward greater English proficiency, but not
toward greater bilingualism due to its contradictory effects on chil?
dren's cultural adaptation.
SETTING AND METHOD SETTING AND METHOD
The site of our study, south Florida, has been so transformed by recent The site of our study, south Florida, has been so transformed by recent
immigration that several commentators have actually placed it as culturally
closer to Latin America and the Caribbean than to the rest of the nation
(Rieff, 1987). Miami, in particular, is home to more foreign-born residents
on a proportional basis than any other American city. Cuban exiles have
built a large and diversified ethnic community, which also serves as a cultural
resource for other Latin American immigrants. Haitians have sought to do
the same in Little Haiti, a neighborhood which lies directly adjacent to
Liberty City, Miami's main African-American area. Many native-born whites
have reacted to the immigrant influx and the emergence of the Cuban
enclave by leaving the city or by militantly supporting the English-only
movement. The result has been a debate over language more acrimonious
than in other American cities (Portes and Stepick, 1993).
In 1973, county commissioners voted to declare Dade County officially
bilingual. Seven years later, however, a grassroots-led referendum repealed
that ordinance and replaced it with a new one stipulating that public funds
could not be used to teach languages other than English or "promote a
culture other than the culture of the United States" (Boswell and Curtis,
1984:121). In early 1993, however, the newly elected Dade County Commis?
sion, where Cuban Americans now comprised a plurality, rescinded the
antibilingual ordinance mandating that public notices and brochures be
printed in Spanish and, in certain cases, in French Creole, as well as in
English. The decision triggered an immediate spate of lawsuits by opponents
who argued that the county could not countermand the English-only amend?
ment to the state constitution, passed two years earlier (Stewart, 1993).
There is little doubt that foreign languages, particularly Spanish, are
widely spoken by first-generation exiles and immigrants in south Florida.
This pattern parallels that followed by large immigrant groups in the past.image 6
646 International Migration Review
Italian, Polish and Jewish communities created by turn-of-the-century im?
migration also retained their home languages for a long time (Glazer, 1954).
The central theoretical and policy question, however, is the language shift
in the second generation and the effects on it of time, differential levels of
ethnic clustering, and parental status. It is possible, as some nativists argue,
that the extraordinary concentration of immigrants in this area is changing
the historical patterns and creating instead a permanent linguistic enclave
where Spanish is the predominant language. Alternatively, south Florida
may simply be in the early stages of absorbing a large foreign influx which,
in due time, will follow the time-honored pattern.
We examine this question on the basis of data from a survey of 2,843
eighth and ninth grade students in Miami (Dade County) and adjacent Ft.
Lauderdale (Broward County) schools. Inclusion of schools in Ft. Lauder-
dale was dictated by the need to compare highly clustered immigrant dale was dictated by the need to compare highly clustered immigrant
communities in Miami with a nearby area where immigrants and their communities in Miami with a nearby area where immigrants and their
children are far more dispersed among the native population. The sample
included children from the most diverse national origins although, reflect?
ing the composition of the immigrant population to the area, the largest
contingents come from Cuba, Nicaragua, other Latin American countries,
Haiti and the West Indies. The survey defined "second generation" as
youths born in the United States with at least one foreign-born parent or
children born abroad who had lived in the United States for at least five
years. The sample is evenly divided between boys and girls, and the average
age is 14.8 years. The sampling design used for the survey included both
inner-city and suburban schools and targeted schools where children of
particular immigrant groups were known to concentrate, as well as those
where immigrants of diverse nationalities were dispersed among a majority
native-born population.
The questionnaire included an extensive array of items on family and
individual characteristics such as the child's age, sex, national origin, place
of residence, length of residence in the United States, education of the
father and the mother, occupation of the father and the mother, home
ownership, and class self-identification. We explore how these charac?
teristics affect the children's proficiency in English, their knowledge of the
parental language, and their overall linguistic preference. The measure of
English proficiency is constructed from the students' reported ability to
speak, understand, read and write English. Previous studies have indicated
that self-reports of language ability, unlike other individual characteristics,
are both reliable and valid (Fishman, 1969; Fishman and Terry, 1969).
Students chose their ability rating from four categories (Not at All, Not Well,
Well and Very Well), which were converted into an overall proficiency score.
Other related questions asked about retention of the parental language and
language preferences. Knowledge of the parental language is measuredimage 7
Language and the Second Generation 647
with the same four self-reported indicators as knowledge of English, coded
identically. Language preference is tapped with an attitudinal question
asking what language the child mostly prefers to speak.
With a sample size this large, almost every relationship turns out to be
statistically significant. To discriminate between substantive and trivial relation?
ships, we use instead a coefficient of strength of association, Cramer's V2,
selected because of its fixed range between 0 and 1. Only associations where
V2 is greater than .12 (significant at the .001 level) will be considered of
empirical import. Multivariate analyses involving determinants and conse?
quences of language proficiency employ both least square and maximum
likelihood methods, described in the following sections.
RESULTS
Bivariate Relationships
The first question of interest is the extent to which today's children of
immigrants coming from different national origins become proficient in
English. On this point, the evidence is unmistakable. For the sample as a English. On this point, the evidence is unmistakable. For the sample as a
whole, 73 percent report that they are able to speak, understand, read, and whole, 73 percent report that they are able to speak, understand, read, and
write English "very well" and an additional 26 percent "well." This leaves
the sum total of those knowing little or no English at just 1 percent. Table
1 presents cross-tabulations of English proficiency with ten characteristics.
Only age, national origin and length of U.S. residence are significantly
related to English proficiency. It is important to note that such differences
exist only between the "well" and "very well" categories, signaling relative
minor variations in English knowledge. In agreement with the first hypothe?
sis, length of U.S. residence has the strongest association with this depend?
ent variable. Slightly over half of foreign-born children with less than ten
years in the country report knowing English very well; the figure climbs to
more than 80 percent among the native-born.
National origin also has a strong correlation with English ability. In this
area, the large Cuban-origin group is divided into those attending Latin-ori?
ented bilingual private schools in Miami and those attending public schools.
Differences between both groups on English knowledge are minimal. Over
70 percent of each category report knowing English very well. Highest
proficiency is associated with children of European and Asian origin, grouped
in the "Other Nationalities" category, and with those of West Indian parent?
age. The latter result is a natural consequence of the fact that most West
Indian parents come from English-speaking countries such as Jamaica,
Trinidad, Grenada, and the Bahamas. Second-generation Nicaraguans have
the lowest English proficiency. This result is related to the relative recency of
Nicaraguan migration. Very few of our Nicaraguan respondents are U.S.
born and most have been in the country less than ten years.image 8
TABLE 1
Language Knowledge and Preferences Among Second-Generation Youths in South Florida: Characteristic English Proficiency
Not at All/
Not Well Well Very Well
(%) (%) (%)
Foreign-Language Proficiency
Not at All/
Not Well
(%)
Well Very Well
(%)
Sex
Male
Female
Age
13 or younger
14
15
16 or older
National Origin
Cuban (private school)
Cuban (public school)
Nicaraguan
Other Latin American
Haitian
West Indian
Other nationality
Length of U.S. Residence
Rve to nine years
10 years or more
U.S.-born
Place of Residence
Miami (Dade County)
Fort Lauderdale (Broward County)
Father's Education
Not high school graduate
High school graduate
College graduate
Mother's Education
Not high school graduate
High school graduate
College graduate
1.30
1.30
0.70
1.20
0.70
5.90
1.20
0.60
3.10
1.00
4.60
0.80
1.20
2.90
0.80
0.80
1.30
1.50
1.70
1.00
0.70
1.50
1.40
0.30
28.80
23.10
19.50
23.70
31.10
35.80
27.30
23.90
41.70
26.60
25.70
17.90
18.60
4.10
28.30
18.20
26.90
18.30
3.00
25.00
21.50
3.70
24.30
22.10
69.90
75.60
79.80
75.10
68.20
58.30
71.50
75.50
55.10
72.40
69.70
81.30
8.20
57.00
7.90
81.00
71.80
8.20
68.30
74.00
77.80
67.80
74.30
77.60
0.066
(0.010)
0.120
(0.001)
0.126
(0.001)
0.157
(0.001)
0.063
(0.010)
0.061
(0.010)
0.062
(0.001)
36.00
33.40
34.20
33.00
36.30
39.20
11.00
27.50
22.10
27.00
67.80
7.60
58.10
24.50
35.40
38.90
31.50
57.50
3.90
35.50
32.90
33.50
34.60
32.70
36.90
33.60
37.90
36.20
32.50
31.90
51.20
4.00
36.40
38.20
2.40
15.60
24.90
32.70
35.20
36.30
36.90
22.40
36.80
35.50
35.20
35.10
35.70
36.00
27.10
33.00
27.90
3.70
31.20
28.90
37.80
32.50
41.40
34.80
11.80
13.70
17.00
42.70
29.40
24.90
31.50
2.10
32.30
28.90
31.90
31.40
29.70
31.30
0.064
(0.010)
0.038
(0.250)
0.265
(0.001)
0.120
(0.001)
0.177
(0.001)
0.031
(0.300)
0.015
(0.890)
1image 9
TABLE 1 (Continued)
Language Knowledge and Preferences Among Second-Generation Youths in South Florida: English Proficiency
Not at All/
Not Well Well Very Well Not Well Well Very Well
(%) (%) (%) V (%) (%) (%) V
Foreign-Language Proficiency
Not at All/
Not Well Well Very Well
(%) (%) (%)
Father's Occupational Statusc
Lower
Lower-Middle
Upper-Middle
Higher
Mother's Occupational Status0
Lower
Lower-Middle
Upper-Middle
Higher
Class Self-Identification
Working class or poor
Lower-middle class
Upper-middle class
Totals_
0.90
0.90
0.70
0.70
1.10
0.70
0.50
1.30
30.10
21.60
18.50
21.10
30.60
26.90
16.50
19.30
2.20 33.10
1.50 26.70
0.80 21.90
1.30 25.90
69.00
77.50
80.80
78.20
68.30
72.40
83.00
79.50
64.70
71.90
77.30
72.80
0.080
(0.001)
0.098
(0.001)
0.079
(0.001)
33.80 35.40 30.80
36.00 37.30 26.80
28.30 34.90 36.80
34.20 38.80 27.10
35.00 35.30 29.70
31.90 38.40 29.70
44.70 30.70 24.60
31.30 38.00 30.80
32.40 35.70 31.90
34.10 36.10 29.80
36.10 34.20 29.70
0.057
(0.03)
0.071
(0.010)
0.023
(0.550)
34.60 35.20 30.20
aCramer's V2. Probability levels in parenthesis.
''Table totals exclude missing data.
CSEI occupational prestige scores collapsed as follows: Lower 40%="Lower;" next higher 20%="Lower-middle;" next higher 20%="Upper-middle;" image 10
650 International Migration Review
Associations with father's education, mother's education and class self-iden?
tification are not significant by our criterion, but they consistently follow the
pattern predicted by Hypothesis 3. In every case, the higher the parental
position, the better the reported command of English. Though not supporting
the hypothesis, they provide some indication of a tendency in the expected
direction. More counterintuitive is the relationship with age since older chil?
dren show less proficiency. This pattern is attributable to the tendency of
recently arrived immigrant youths to enter school at grades lower than the
respective native-born age cohort. In this sample, older students generally
come from non-English-speaking countries and are among the most recent
arrivals. Nicaraguan children are heavily represented in this group.
However, the key story in Table 1 is the overwhelming dominance of English
knowledge among children of immigrants and its strong positive association
with length of residence in the United States. There is little variance in
widespread fluency among the second generation, and whatever variance exists
is highly responsive to the passage of time. A very different story emerges when
we consider preservation of parental languages. As indicated above, foreign
language proficiency was measured in an identical manner to English knowl?
edge. The second panel of Table 1 presents cross-tabulations of this variable
with the same set of individual and parental traits.
The bottom row of the panel indicates that one-third of students in the
sample are already English monolinguals. The absolute number of such sample are already English monolinguals. The absolute number of such
cases (N=984) far exceeds the number of children of West Indian and other cases (N=984) far exceeds the number of children of West Indian and other
English-speaking nationalities, indicating a rapid loss of parental language
among non-English-speaking groups.3 Yet a comparable proportion of
respondents report knowing parental languages "very well" and, hence, it
is worth examining possible determinants of this difference. Among the set
of potential predictors in Table 1, national origin has by far the strongest
association with foreign language fluency. There is a clear difference be?
tween Latin American nationalities, on the one hand, and Haitian, West
Indian and Asian/European nationalities, on the other. Reported English
monolingualism among West Indian-origin students and respondents
grouped in the "Other Nationalities" category is again a straightforward
consequence of many of their parents being English speakers. The same is
not the case, however, for Haitian-origin youths whose home language is
French or Creole. Almost 70 percent of this group reports little or no
3The cutoff points for the language categories in this case are slightly below those in the first
panel of Table 1, where the strong rightward skew of the English Proficiency Index led us to
classify into the "well" and "very well" categories only those reporting very high proficiency.
The more balanced Foreign Language Proficiency Index allows less extreme cutoff points. If
the same cutoffs were used, the proportions of those reporting themselves as English monolin?
guals (left-most column of Table 2) would increase by 7.4% and those declaring full command
of a foreign language (right-most column) would decline by 15%. The patterns of association
with potential predictors in Table 1 would hold.image 11
Language and the Second Generation 651
knowledge of these parental languages, and only 12 percent declare them?
selves proficient in either.
The opposite is the case among Latin American groups where foreign
language loss affects only about one-fourth of respondents and drops to
only 11 percent among Cuban students in private schools. Retention of the
parental language (Spanish) is in part a consequence of the recency of some
migrant flows such as Nicaraguans. More significantly, however, it reflects
the presence of a large and diversified ethnic enclave where Spanish is the
language of daily intercourse for all kinds of transactions. Respondents in
private bilingual schools are mostly the children of middle-class Cuban
exiles who represent the core of this ethnic economy. It is not surprising
that they have the lowest propensity to give up Spanish. Combined with the
pattern of responses in the first panel of Table 1, these results indicate that
Cuban and other Latin American-origin youth in south Florida are mostly
bilingual. These results lend support to Hypotheses 1 and 2 insofar as they
predict positive effects of immigrant concentration and a diversified ethnic
economy on language preservation.
The originally Cuban and now pan-Latin enclave is located in Miami
(Dade County). Hence, it is possible to predict that preservation of Spanish
will be significantly greater among second generation youths in this city
than in adjacent Ft. Lauderdale where no similar phenomenon exists. This
expectation is borne out by the results. Place of residence has the second
strongest association with home language retention, with Miami respon?
dents being almost twice as likely to be bilinguals (reporting knowing the dents being almost twice as likely to be bilinguals (reporting knowing the
parental language "well" or "very well") as those living in Ft. Lauderdale. parental language "well" or "very well") as those living in Ft. Lauderdale.
The very strong influence of ethnic concentration is counteracted, however,
by the passage of time. As shown in the second panel of Table 1, there is a
clear monotonic relationship so that the longer the child has resided in the
United States, the stronger the tendency toward English monolingualism.
Among recent arrivals, 43 percent report full command of a foreign
language, a figure that falls to just one-fourth among the native born. This
result again supports the first and second hypotheses' prediction of a
significant negative effect of time on bilingualism.
Parental education, occupational status and class self-identification have
essentially no association with foreign language fluency. This result supports
Hypothesis 3 which attributes it to the contradictory effects of family status on
linguistic adaptation. Interviews with a sample of immigrant parents of our
respondents in Miami indicate that they are consistendy in favor of English
language acquisition, but not at the cost of giving up their mother tongue.
Those with greater resources are in a better position to implement this bilingual
project, but their efforts are frequently neutralized by greater exposure of their
children to mainstream culture which the same resources make possible.image 12
652 International Migration Review
Overall, these findings are in close agreement with the theoretical argu?
ment outlined previously. Children of relatively isolated immigrants - such
as those living in Broward County or Asians and Europeans grouped in the
"Other Nationality" category - experience a faster language transition
toward monolingual English; children of relatively prosperous and highly
concentrated immigrants, such as Cubans, are far more likely to retain their
parental language. The passage of time significantly increases language
proficiency and undermines bilingualism. Education and occupational
status of immigrant parents, which could have reasonably been expected to
have the opposite effect, fail to do so because of seemingly contradictory
effects on linguistic adaptation.
A final variable of interest is the child's attitude toward speaking English
versus speaking the parental or other foreign language. Just because
children of immigrants know English well does not guarantee that they will
use it, given the choice. The evidence on this point is presented in the last
panel of Table 1. Preference for English is overwhelming: 80 percent of the
entire sample endorses it. Length of U.S. residence is strongly and positively
correlated with English preference, but even among the most recent arrivals
over 71 percent opt for English over their home languages.
National origin is also associated with language preference, but the trend
here differs from those found previously. Children of Haitian and West
Indian parents, as well as those grouped in the "Other Nationality" category,
lean strongly toward English in a fashion congruent with their weak reten?
tion of other languages. Cubans, however, also have a very strong preference
for English, in particular those attending private schools. Despite their
greater reported knowledge of Spanish, over 90 percent of Cuban-origin greater reported knowledge of Spanish, over 90 percent of Cuban-origin
youths prefer communicating in English. This result means that even youths prefer communicating in English. This result means that even
among youths educated in bilingual schools at the core of an ethnic enclave,
linguistic assimilation is proceeding with remarkable speed. Somewhat
lower attachment to English is found among Nicaraguans and other Latin
Americans, a probable consequence of their recency in the country, but even
among these groups three-fourths endorse their new country's language
over native Spanish.
No other predictor has a significant association with this final dependent
variable, although there is a clear tendency for children of better-educated
and higher-status parents to prefer English. Again, however, these differ?
ences take place in the context of overwhelming language assimilation. An
eloquent indicator of the trend is the absence of significant differences
between students in Dade and Broward schools. This finding indicates that,
whether second generation children live in an English-only environment
or in one where use of Spanish is widespread, their ultimate preference for
the language of the land will be the same.image 13
Language and the Second Generation 653
Determinants of Bilingualism
The principal difference observed in our sample pertains to parental
language retention rather than English acquisition. In other words, the
central difference among immigrant youths is not whether they know and
prefer English, but the extent to which they retain some command of the
parental language. As seen previously, past studies have reported benefits
of bilingualism in terms of cognitive development and academic achieve?
ment. Yet, the preceding bivariate results offer only a preliminary approach
to an analysis of determinants of bilingualism since several of the predictors
are themselves highly correlated.
To establish the net effect of each predictor controlling for others, we ran
multivariate regressions with two different versions of the dependent vari?
able. The first is the Foreign Language Proficiency Index, constructed in
this case as the logarithm of the sum of responses to the items measuring
different aspects of language ability.4 The effect of this logarithmic trans?
formation is to render unstandardized regression coefficients (when small)
interpretable as a proportional increase/decrease in language fluency, net
of other factors. The second indicator is a dichotomous variable where
"Bilingual" is defined as a respondent who is fully proficient both in English
and in a foreign language.5 This restrictive definition seeks to identify
"true" bilinguals, differentiating them from both monolinguals and those
with a lesser command of a second language.
Both indicators are regressed on the same set of predictors. The first
panel of Table 2 presents results of an ordinary least squares regression of
the logarithmic index. Raw regression coefficients indicate the net propor?
tional gain associated with each predictor. The second panel presents results
of a logistic regression of the dichotomous indicator of bilingualism on the
same array of independent variables. Coefficients in this case represent the same array of independent variables. Coefficients in this case represent the
net increase/decrease in the logarithm of the odds of being fully bilingual net increase/decrease in the logarithm of the odds of being fully bilingual
associated with each predictor. To clarify the meaning of these results, the
column labelled Ap presents the associated probabilities.
Results in the first panel of the table reveal clearly the forces arrayed for
and against bilingualism. The most significant factor in the pro-bilingual
side is national origin. Latin American nationalities display, without excep?
tion, a much greater probability of retaining their parental language. The
corresponding coefficients represent the approximate net gain/loss in for?
eign language proficiency relative to a reference category constituted, in
this case, by "Other Nationalities." Cuban-origin students in public schools
have, for instance, a significant advantage in foreign language knowledge
4The untransformed Index ranges from 1 to 12 (high).
?The variable is coded 1 for respondents falling into the "very well" categories in the English
Proficiency and Foreign Language Proficiency panels of Table 1; all others are coded 0.image 14
654 International Migration Review
TABLE 2
Determinants of Foreign Language Proficiency and Fluent Bilingualism,
Second-Generation Students in South Florida
Foreign Language Proficiency (Logged) Fluent Bilingualism
s.e. V
Intercept
Age
Sexc
City of Residenced
Father's Education
Mother's Education
Father's Occupational Status6
Class Self-Identification
Length of U.S. Residence
Immigrant Nationality
Cuban (private school)
Cuban (public school)
Nicaraguan
Other Latin American
Haitian
West Indian
R
R2
N
-2 Log Likelihood
Model Chi Square
df df
P P
1.600
-.003
.118
-.254
-.007
-.014
-.0001
.022
-.148
.741
.532
.453
.475
.020
-.683
.578
.334
2,840
.216
.014
.023
.043
.010
.010
.0007
.025
.016
.072
.045
.055
.047
.064
.057
-108
.014
-.029
-.003
.016
-.180
.219
.381
.221
.306
.006
-.272
.001
.001
.001
.001
.001
.001
-1.835
-.062
.467
-.076
.001
.650
-.0003
.041
-.196
.966
.526
.143
.486
-.959
-.720
2,840
2,837.19
127.42
13
.001
.849
.054
.094
.1744
.040
0.39
.003
.072
.062
.167
.096
.141
.099
.256
.187
n.s.
n.s.
-.03
.21
.11
n.s.
.10
-.13
-.11
"Standardized regression coefficients.
bProbability increase or decrease, computed on statistically significant coefficients only.
cFemale= 1; male=0.
dFort Lauderdale = 1; Miami=0.
eDuncan's Socioeconomic Index Scores.
relative to statistically comparable children of other nationalities. The
advantage increases among Cuban-American students attending bilingual
private schools. Nicaraguan-origin children and those of other Latin na?
tionalities are also strongly inclined to preserve their parents' Spanish. The
strong negative effect of West Indian origin is again interpretable as a direct
result of their coming from monolingual English households.6 Arrayed
against foreign language knowledge are time in the country and spatial
^he nationality coefficients in the first panel of Table 2 are not interpretable as percent change
in bilingualism for each group. This is because they are computed relative to a reference
category and because they are excessively large. Their appropriate interpretation is indicative
of the direction and relative strength of the effect associated with each national group.image 15
Language and the Second Generation 655
dispersion. Children living outside of Miami have a 25 percent disadvan?
tage in knowledge of a foreign language relative to comparable youths
residing in Miami's concentrated ethnic communities. The corresponding
disadvantage for the native born is 30 percent in comparison with more
recent arrivals.7
The only unexpected effect in these results is that of gender, which
indicates that girls have a significantly greater propensity for retaining the
parental language than do boys of comparable characteristics. We may
speculate that this result is attributable to the greater seclusion of female
youngsters in the home environment which exposes them to greater contact
with parents. That interpretation would represent an extension of the logic
of the argument about the linguistic effects of concentrated ethnic commu?
nities, underlying Hypotheses 1 and 2.
Overall, this multivariate analysis shows that the effect of national origin
is not a spurious consequence of other individual or family characteristics.
There is something unique about the character of specific immigrant
communities which leads to different propensities to retain their home
language. This influence persists even after controlling for length of U.S.
residence and location. On the other hand, as predicted, effects of parental
education and occupational status are consistently insignificant. These
tendencies are buttressed by results in the second panel of Table 2. They
may be interpreted as influences on "true" or fluent bilingualism, defined
by speaking both languages very well. Results confirm the overwhelming
importance of nationality, which again has the strongest effect on this
dependent variable.
In this last analysis, individual nationality effects are not computed
relative to the omitted category but relative to the average effect for all
categories. Looking at the probability column, we see that Cuban Americans
in private schools are 21 percent more likely to be fluent bilinguals, while
Haitian and West Indian-origin youths are 13 and 11 percent less likely to
be so relative to the average. The Haitian effect is actually stronger, possibly
reflecting less fluency in English. Similarly, Nicaraguan-origin children reflecting less fluency in English. Similarly, Nicaraguan-origin children
who, as seen above, are significantly more likely to retain their parents' who, as seen above, are significantly more likely to retain their parents'
language are no different from the average in their bilingual skills. As with
Haitian Americans, this result reflects their greater difficulties with English.
Another major difference in these results is the failure of place of residence
to affect the dependent variable. This finding indicates that, while living in
a less ethnic environment reduces the likelihood of retaining a foreign
language, it does not by itself reduce the probability of becoming fluently
bilingual. It is the character of the immigrant community - its internal
7As shown in Table 1, length of U.S. residence is coded as a 3-point scale: 1 = foreign born
with 5 to 9 years in the United States; 2 = foreign born with more than 10 years in the United
States; and 3 = native-born. Each point in this ordinal scale is associated with approximately
a 15% reduction in knowledge of a foreign language.image 16
656 International Migration Review
diversity, history and cohesiveness - that seems to hold the key on whether
second-generation children successfully combine two languages. On the
other hand, length of residence in the country again decreases the prob?
ability of bilingualism. Jointly, these results lend qualified support to
Hypothesis 2.
Finally, there is a resilient gender effect in the same direction as found
previously. Among children of similar national origin and length of U.S.
residence, girls are significantly more likely than boys to retain their parents'
language and to become fully bilingual. This strong and unanticipated
effect points to the potential importance of families and differential sociali?
zation by sex on the linguistic adaptation of immigrant youth.
Effects of Bilingualism
As seen previously, the past literature on bilingualism and its effects on
academic achievement registers considerable debate. While there has been
a shift from predominantly negative views toward a more positive stance,
the issue has generally been researched among older immigrant groups or
in other countries. There is so far little information about the potential
academic effects of bilingualism among the new second generation in the
United States. We approach this issue by considering the relationship
between fluent bilingualism and four dependent variables. The first two are
individual scores of our respondents on the Stanford mathematics and
reading performance tests, drawn from school records. The others are
educational and occupational aspirations measured by items in the survey
questionnaire.
All variables were measured contemporaneously, and hence it is not
possible to speak strictly of effects of bilingualism since the direction of
causation is not always clear. However, the key issue in the literature has
been the net sign of the relationship between communicating in two
languages and academic performance, and this can be addressed with the
data at hand. We approach the question through a multivariate framework
with a focus on the net relationship of each dependent variable with fluent
bilingualism, controlling for other predictors, including knowledge of
English. Table 3 presents the results of this analysis.
The first columns of the table present the relationship between the latest The first columns of the table present the relationship between the latest
available math scores for each respondent, in Stanford test percentiles, and the available math scores for each respondent, in Stanford test percentiles, and the
independent variables. For this analysis, the Knowledge of English Index is
allowed to vary through its full un transformed range (1 to 12). Bilingualism is
the same dichotomy described previously with 1 representing fluent bilinguals.
Other variables in the equation include age and sex, place and length of U.S.
residence, parental education, and paternal occupation measured in Duncan
SEI prestige scores. Because of its potential significance as a predictor, we alsoimage 17
TABLE3
Determinants of Academic Achievements and Aspirations, Second-Generation Youths in South Reading Scores Educational Sex1*
Length of U.S. Residence
City of Residence1
Father's Education
Mother's Education
Father's Occupational Status
Hours of Homework
Knowledge of English
Bilingual
Intercept
R
R2
JS_
-3.747 -3.747
-.187 -.187
.174
-9.809
1.023
1.914
.128
2.981
2.341
3.087
60.748
.345
.119
2,349
.672
1.100
.701
1.939
.483
.478
.033
.422
.468
1.347
12.824
.001
.04
.001
.001
.001
.001
.03
-2.000
-.659
2.077
3.745
1.308
1.753
.128
1.799
4.828
1.577
-23.466
.414
.171
2,349
.583
.954
.608
1.681
.419
.414
.029
.366
.406
1.168
11.124
.001
n.s.
.001
.03
.002
.001
.001
.001
.001
-.062
.151
.021
.087
.066
.068
.001
.129
.084
.151
2.764
.373
.139
2,349
aOrdinary least squares regressions.
,>Female= 1; male=0.
cFort Lauderdale=l; Miami=0.image 18
658 International Migration Review
include hours of homework as a control. This is a five-point measure
ranging from "less than 1 hour daily" to "five hours or more."
Most predictors have significant effects in the expected direction. In the
interest of brevity, we focus exclusively on those of language knowledge. The
effect of knowledge of English is very strong, exceeding four times its
standard error. Each extra point in this index yields a 2.3 percentile gain
in math scores. Yet even after controlling for English fluency, bilingualism
retains a significant positive relationship with the dependent variable.
According to these results, bilingual students who are comparable in all
other characteristics to those who have lost fluency in the parental language
have a small but significant advantage in math performance.
The same is not the case for reading ability. Since the test measures ability
to read in English, it is not surprising that the Knowledge of English Index
has the strongest effect, dwarfing those of other predictors. By itself, this
result essentially signals the validity of the Index. It is worth noting, however,
that the zero-order relationship between bilingualism and reading perform?
ance is positive and significant and that the net effect, although statistically
weak, retains a positive sign.
The last two panels of Table 3 present regressions of educational and
occupational aspirations. The first is based on responses to the question:
"What is the highest level of education that you realistically expect to get?"
Answers were coded along a five-point scale from "less than college" to
"graduate degree." The second is based on the question, "What occupation
do you plan to have as an adult?" Responses were transformed into
continuous SEI occupational prestige scores. Results indicate that English
knowledge and hours of school homework both have positive and sizable
effects on aspirations. Net of them, however, bilingualism retains a signifi?
cant and positive relationship with both dependent variables.
Past research in the sociology of education has found aspirations to be a
reliable predictor of subsequent achievement (Haller and Portes, 1973;
Sewell and Hauser, 1972; Kerckhoff and Campbell, 1977). To the extent
that this continues to be the case, fluent bilingualism represents a net
advantage for immigrant youth since it is associated with both higher
present achievements and more ambitious plans for the future. These
results lend support to the recent literature that views the ability to speak a
foreign language as adding to, not subtracting from children's chances for
educational success (Rumbaut and Ima, 1988; Figueroa and Garcia, 1994).
CONCLUSION CONCLUSION
English is alive and well in south Florida. Miami is the American city most English is alive and well in south Florida. Miami is the American city most
heavily affected on a proportional basis by recent immigration and, hence,
the one where the demise of English predicted by nativist organizationsimage 19
Language and the Second Generation 659
should be most evident. Our results indicate that such fears are exaggerated.
Children of immigrants not only possess widespread competence in Eng?
lish, but also demonstrate an unambiguous preference for it in everyday
communication.
Children raised in the core of the Spanish-speaking Miami community
(those attending bilingual private schools) are actually the most enthusiastic
in their preference for the language of the land. Moreover, the passage of
time in the country strongly influences linguistic assimilation leading to a
rapid shift toward English.
These results indicate that, contrary to nativist fears, what is at risk in this
area is the preservation of some competence in the languages spoken by
immigrant parents. Our results support those of prior research indicating
that fluent bilingualism is an intellectual and cultural resource. In this sense,
the rapid transition toward monolingualism represents a loss. Even highly
educated immigrant parents do not stand much of a chance of transmitting
their language to their children. Their illusions of communicating with their
children and grandchildren in their native language will come to naught
for the most part. Nativist fears that they will be able to do so to the
detriment of English dominance are entirely unfounded. Results of the
study indicate that only in places where immigrant groups concentrate and
manage to sustain a diversified economic and cultural presence will their
languages survive past the first generation. In the absence of policies
promoting bilingualism, even these enclaves will be engulfed, in all prob?
ability, in the course of two or three generations.
REFERENCES
Adams, J.
1856 Life and Works. Boston: Little, Brown. Originally published 1780.
Bailey, G., N. Maynor and P. Cukor-Avila
1991 The Emergence of Black English. Philadelphia: J. Benjamins.
Baron, D.
1990 The English-Only Question. New Haven, CT: Yale University Press.
Berrally, A. and T. L. McCarty
1990 "The Navajo Language Today." In Perspectives on Official English. Ed. K. L. Adams
and D. J. Brink. New York: De Gruyter. Pp. 234-246.
Boswell, T D. and J. R. Curtis
1984 The Cuban-American Experience. Totowa, NJ: Rowman & Allanheld.
Cummins, J.
1981 "Empirical and Theoretical Underpinnings of Bilingual Education," Journal of
Education, 163:16-29. Winter.
DillardJ. L. DillardJ. L.
1985 Toward a Social History of American English. New York: Mouton. 1985 Toward a Social History of American English. New York: Mouton.
Fernandez, R. M. and F. Nielsen
1986 "Bilingualism and Hispanic Scholastic Achievement: Some Baseline Results," Social
Science Research, 15:43-70.image 20
660 International Migration Review
Figueroa, R. A. and E. Garcia
1994 "Issues in Testing Students from Culturally and Linguistically Diverse Backgrounds,"
Multicultural Education, 2:10-23. Fall.
Fishman, J. A.
1969 "A Sociolinguistic Census of a Bilingual Neighborhood," American Journal of Sociology,
75:323-39.
Fishman, J. A. and C. Terry
1969 "The Validity of Census Data on Bilingualism in a Puerto Rican Neighborhood,"
American Sociological Review, 34:636-650.
Franklin, B.
1959 The Papers of Benjamin Franklin. Ed. Leonard W. Labarre. New Haven, CT: Yale
University Press.
Glazer, N.
1954 "Ethnic Groups in America." In Freedom and Control in Modern Society. Ed. M. Berger,
T Abel, and C. Page. New York: Van Nostrand. Pp. 153-178.
Greenblatt, S. J.
1976 "Learning to Curse: Aspects of Linguistic Colonialism in the Sixteenth Century." In
First Images of America: The Impact of the New World on the Old. Vol. 2. Ed. F Chiapelli,
M. J. B. Allen, and R. L. Benson. Berkeley: University of California Press. Pp.
561-580.
Haller, A. O. and A. Portes
1973 "Status Attainment Processes," Sociology of Education, 46:51-91. Winter.
Herskovits, M. J.
1990 The Myth of the Negro Past. Boston: Beacon Press.
Kerckhoff, A. C. and R. T. Campbell
1977 "Black-White Differences in the Educational Attainment Process," Sociology of Edu?
cation, 50:15-27. Jan.
Laird, C. G.
1970 Language in America. New York: World Publishing Co.
Lambert, W E. and G. R. Tucker
1972 Bilingual Education of Children: The St. Lambert Experiment. Rowley, MA: Newbury
House.
Lieberson, S., G. Dalto, and M. E.Johnston
1975 "The Course of Mother Tongue Diversity in Nations," American Journal of Sociology,
81:34-61. July.
Marckwardt, A. H.
1980 American English. New York: Oxford University Press.
Massey, D.S., L. Goldring, andj. Durand
1994 "Continuities in Transnational Migration: An Analysis of Nineteen Mexican Com?
munities," American Journal of Sociology, 99:1492-1533. May
Mintz, S. W
1970 "Foreword." In Afro-American Anthropology. Ed. N. E. Whitten, Jr. andj. F Szwed. New 1970 "Foreword." In Afro-American Anthropology. Ed. N. E. Whitten, Jr. andj. F Szwed. New
York: The Free Press. Pp. 1-16. York: The Free Press. Pp. 1-16.
Peal, E. and W E. Lambert
1962 "The Relation of Bilingualism to Intelligence," Psychological Monographs, 76(27): 1-23.
Portes, A. and R. G. Rumbaut
1990 Immigrant America: A Portrait. Berkeley: University of California Press.
Portes, A. andj. Sensenbrenner
1993 "Embeddedness and Immigration: Notes on the Social Determinants of Economic
Action," American Journal ofSociology, 98:1320-1350. May.
Portes, A. and A. Stepick
1993 City on the Edge: The Transformation of Miami. Berkeley: University of California Press.image 21