Speaking the Same Language? English Language Fluency and Violent Crime at the Neighborhood Level

ABSTRACT

In this article, we seek to extend knowledge of the relationship between immigration and crime. We synthesize existing theories and literatures and argue that one largely overlooked aspect of immigration—language use—may play critical roles in how and why immigration is related to macro-level violence rates. Specifically, there are theoretical reasons to expect that levels of lack of English fluency and of bilinguals in neighborhoods will be associated with violent crime rates. We test these relationships using data from the National Neighborhood Crime Study (NNCS). The multivariate results reveal that lack of English fluency has a nonlinear relationship with homicide and robbery, such that when lack of fluency is low, crime rates increase, but the positive effect diminishes as English nonfluent individuals come to represent a greater proportion of neighborhood population. In addition, levels of bilingualism moderate the effects of lack of English-language fluency for homicide but less so for robbery. This finding applies to the total sample as well as in traditional immigrant destinations, but not in new destinations.

KEYWORDS:

• Community
• crime
• immigration
• law and deviance
• urban sociology

Acknowledgments

In the current study we used data from the National Neighborhood Crime Study funded by a grant (SES-0080091) from the National Science Foundation. An earlier version of this study was presented at the annual meeting of the American Society of Criminology in Washington, DC (November 2015). We would like to thank the editors and anonymous reviewers for thoughtful comments and suggestions.

Notes

1. We obtained NNCS from The Inter-university Consortium for Political and Social Research (Peterson and Krivo 2006). Additional census data were obtained from the National Historical Geographic Information System (Minnesota Population Center 2011). Missing data were not a major issue in our study. The only variable with missing values in our study was homicide counts. Police departments in Philadelphia and San Antonio did not provide any homicide data to NNCS researchers, so we chose to exclude these cities in their entirety from our models predicting homicide because imputation of values for dependent variables is inappropriate.

2. We reestimated all of the models using (1) total violent crime counts, which include murder, robbery, forcible rape, and aggravated assault, and (2) following Ramey (2013), a sum of murder and robbery incidents as dependent variables. There were few substantive differences that we report in the sensitivity analysis section.

3. Census data that disaggregate tract population by the level of English-language proficiency and by a specific language spoken at home were publicly available for Spanish speakers only (see technical documentation for Summary File 3).

4. Note that some of the cities included in the NNCS were not classified by Ramey (2013:625) as either new or traditional destinations. In our study, these cities are included in analyses with the total sample but are excluded from the subsample analyses. Additionally, other scholars have distinguished “new” from “traditional” immigrant destinations using somewhat different classification schemes. As these have made use of Metropolitan Statistical Areas (Singer 2004) or county-level (Shihadeh and Barranco 2013) data and are different units of analysis than used in this study, we follow the classification scheme of Ramey (2013), who used the same data set. One alternative schema for classifying immigrant destinations used with NNCS data was utilized by Velez and Lyons (2012). We reestimated all models using this classification scheme to differentiate between new and traditional destinations and we report these results in the sensitivity analyses section.

5. We also calculated the inflection points for traditional and new destinations and found that in the former, the effects of English nonfluency on homicide become negative when English nonfluent individuals come to represent 32 percent of tract population, and in the latter the inflection occurs at 35 percentage points. Readers should note, however, that only a small number of neighborhoods in our study (164 in traditional and 20 in new destination cities) actually have a concentration of English nonfluent Spanish speakers above the noted levels.

6. We included the results for new destinations despite the fact that the interaction term is not statistically significant for comparison purposes.

7. While the null and negative results for the percentage of young males found here may seem counterintuitive, as McCall et al. (2013) discuss, prior city-level studies find a variety of patterns with regard to the macro-level age structure and violence (homicide) relationship, and their own initial findings show a negative relationship between the percentage young (15 to 29 years) and homicide rates.

8. The results of all sensitivity tests discussed are not shown but are available upon request.

9. One exception was the negative and significant interaction term between English nonfluency and total bilingualism in the total sample in models predicting total violence. This effect was not significant in our main analysis predicting robbery.

Additional information

Notes on contributors

Feodor A. Gostjev

Feodor A. Gostjev, PhD, is an Assistant Professor of Criminal Justice at Bridgewater State University in Massachusetts. Fed’s current research focuses on immigration and crime, immigration policy, and methodological issues in measuring and modeling ethno-racial diversity at the macro-level.

Amie L. Nielsen

Amie L. Nielsen, PhD, is an Associate Professor of Sociology at the University of Miami. Her research focuses on race/ethnicity, immigration, and crime and deviance, particularly at the macro level.