Abstract
In the United States, children of immigrants face strong pressures to shift to English. We examine how the retention of Spanish-language skills affects the academic achievement of English-proficient Latino/a children of immigrants and how this varies by gender. Further, we examine the role that family interaction may play in mediating the impact of gender and language on achievement. We find that biliterate boys significantly outperform boys who have little Spanish proficiency. However, for girls there is no significant advantage or disadvantage to biliteracy in terms of GPA (grade point average). Our results suggest that, for Latino boys, the academic advantage of biliteracy is explained by strong family social cohesion. Our results also suggest that, while within-family social capital provides a scholastic benefit from family social cohesion in the case of biliterate boys, strong family ties can also have academic disadvantages.
Acknowledgements
We thank Roger Waldinger for inspiring this research. We also thank Dalia Abdelhady for comments on a previous version of this article.
Notes
1. However, from a different perspective, Williams, Alvarez and Andrade (2002) suggested that stereotypes and gender expectations associated with young, female, Mexican immigrants hinder the achievement of these young women because they experience more negative interactions with teachers and less access to career and graduation counselling.
2. The number of cases of students whose overall English skills are less than ‘well’ is too small to include students with limited English as a separate category. Among the parent-matched sample of Latino children less than 2 per cent of the children reported overall English skills that are less than ‘well’.
3. Fishman and Terry have found that self-reported language skills have high levels of validity. In their research, they compared self-reports based on questions that are very similar to those used in CILS:88 to proficiency measures used in linguistic and psychological research on language and find the validity of the self-reports ‘tends to be rather substantial and consistent’ (Citation1969, p. 643; see also Fishman Citation1969).
4. In a very small number of cases this variable refers to a language other than Spanish.
5. Although not primary to our analysis, we also note significant differences in GPA across ethnic groups, particularly as controls are included in the model. Net of a vector of controls, including socioeconomic status, GPAs for Cubans, Central Americans and Other Latinos are significantly lower than those of Mexicans. This is interesting particularly because, prior to controls, mean GPAs for English-speaking Cubans (2.3869), Mexicans (2.3711) and Other Latinos (2.3685) are strikingly similar, while mean GPAs for South Americans are notably higher (2.4502) and those of Central Americans (2.0673) are notably lower.