Bilingualism and achievement in the Spanish second generation: A longitudinal study

ABSTRACT

The current study contributes to the continuing debate about the advantages of bilingualism on life chance opportunities by examining whether bilingual children of immigrants achieve more or less in adulthood compared to monolinguals. Using longitudinal data from Spain, we find significant advantages among bilinguals in measures on educational attainment, occupational prestige, and familial income. We build on the existing literature of status attainment using structural equation modelling to show how the effect of bilingualism is longitudinal and reinforcing of adult outcomes. The study concludes with a discussion about the theoretical and practical implications of our findings, particularly about adaptation and incorporation of children of immigrants in receiving countries.

KEYWORDS:

• Bilingualism
• status attainment
• Spain
• ILSEG
• children of immigrants
• second-generation

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Past debates concerning the presence and status of foreign languages in Spain in general, and Catalonia in particular, have been reviewed in prior publications by the first author. See, in particular, Medvedeva and Portes (2018).

2 ILSEG was deliberately patterned after the Children of Immigrant Longitudinal Study (CILS) that followed large samples of second generation youths in two major American cities (Miami and San Diego) in the 1990s, interviewing them at average age 14 and re-interviewing them three and seven years afterwards, Like ILSEG in Europe, CILS was the only study of its kind in the United States by the time of its completion in 2002 (Rumbaut 2005; Fernández-Kelly and Konczal 2005).

3 Tests of cognitive skills are commonly lengthy and difficult to administer, especially in the context of self-completed surveys administered in schools. This is the main reason for their absence in ILSEG’s first survey, as well as most other studies of second generation adaptation in the U.S. and Europe.

4 The large loss of cases during the third survey is attributable to the dispersal of the sample in early adulthood, with members moving to other cities or abroad. The ILSEG field teams attempted to compensate for this dispersal with full use of social media by tracing names in the platforms most commonly used by adolescents and young adults in Spain: Facebook and Tuenti. The procedure made possible the location and interview of respondents who had moved to other cities in Spain and even abroad. Yet limitations of time and resources presented continuing this effort indefinitely.

5 Family SES is a composite index consisting of the standardized sum of father’s and mother’s education (in complete years); father’s and mother’s occupational status (in prestige scores); and home ownership divided by five. When one or two indicators were missing, the index was constructed with the remaining ones divided by their number.

6 The ILSEG first survey does include data on the legal status of immigrant parents in Spain. It was omitted among predictors because the vast majority of the parental sample was in the country legally. Past analysis indicated that variations in legal status did not have significant effects on any major outcome of the adaptation process by late adolescence, with exception of country of birth. This last variable is included as a predictor in the following analysis. See Portes, Aparicio, and Haller (2016).

7 The selection equation includes age, gender, city of residence, home ownership, and languages spoken at home. For space reasons, selection equation coefficients are omitted but the resulting rho and lambda coefficients showing the effects of sample attrition are included.

8 The PRESCA scale of occupational prestige was constructed following procedures similar to those used for the development of Treiman’s Occupational prestige scale in the U.S. PRESCA’s original range is 0-270. See Carabaña and Bueno (1996).

9 A just-determined model employs the full information provided by the matrix of correlations between N variables, which in this case is: N(N-1)/2=45. The model in Figure 1 uses only 10 variables. See Heise (1969) and Maruyama (1997).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.