Individual English Proficiency,Community Language Resources,and Immigrants’ Employment in Western New York: Evidence from the American Community Survey

Abstract

This study examines associations between the ability to speak English, community language resources, and employment among immigrants in Western New York, United States. Building on existing literature that demonstrates the importance of individual human capital (English proficiency), social networks, and ethnic community resources in immigrants’ labor market outcomes, we introduce the concept of community-level language resources as a facilitator of immigrants’ economic integration. Using data from the American Community Survey and a measure of community language resources (the percentage of bilinguals among people speaking the same language in a community), we find that greater community language resources are associated with a significant increase in the likelihood of being employed for immigrant men with limited English proficiency. Regression analyses also show that neither individual English proficiency nor community language resources are significantly associated with immigrant women's probability of being employed. This study calls for attention to community language resources and individual human capital when designing research on and developing policies for immigrant populations. Findings also show the need for a gender-aware approach to research and practice with immigrant communities.

Keywords

economic integration language proficiency community resources

Introduction

Immigrants’ socioeconomic integration is a crucial policy issue in the United States as the number and percentage of immigrants in the population have steadily grown for the last several decades. The number of immigrants increased from 9.7 million (5.4%) in 1960 to 44.8 million (13.7%) in 2018. Furthermore, the size of the immigrant population is projected to double by 2065, indicating its continued growth (Budiman et al. 2020). Accordingly, immigrants’ success in the labor market and financial system is critical for the U.S. economy and society.

Numerous studies have investigated what facilitates or obstructs immigrants’ successful integration into the U.S. labor market (Chiswick and Miller 2014; Congressional Budget Office 2004). The majority of these studies identify immigrants’ individual English proficiency as a critical facilitator for favorable labor market outcomes (Adserà and Pytliková 2016; Chiswick and Miller 2014; Potocky-Tripodi 2004), supporting the human capital theory's argument that an individual's education and skills play critical roles in labor market outcomes (Becker 1964). Invaluable as they are, these studies may be limited in explaining immigrants’ economic integration because they focus mainly on an individual's ability to communicate in English without considering community-level language resources: assistance and resources available in the community to individuals.

Existing literature on social networks demonstrates the importance of community resources for immigrants’ economic integration. Ethnic social networks improve immigrants’ labor market outcomes (Waldinger 1994; Tegegne 2015), especially among those with limited English proficiency, by providing language assistance and job information in their ethnic language (Mouw and Chavez 2012). However, these studies focus mainly on ethnic community resources without recognizing that these resources may not be identical to community language resources. They equate ethnic resources with the size or concentration of co-ethnic populations in a specific geographic area (Hao and Kawano 2001; Beaman 2012). However, ethnic community resources may not be identical to community language resources. An ethnic community with many co-ethnic members may provide emotional and material support but may not have a high level of community language resources if only a few members are bilingual. Furthermore, existing studies do not account for the fact that ethnic community boundaries are not always identical to linguistic community boundaries. Some languages are shared by multiple ethnic groups (e.g., the Arabic language spoken by multiple ethnic groups in Middle Eastern and North African countries). In contrast, some ethnic groups are composed of subgroups speaking different languages (Schlee 2015).

Building on human capital and social network theories, this paper investigates the roles of community language resources in facilitating immigrants’ economic integration. In this paper, we define community-level language resources as the percentage of bilingual adults among those speaking the same native language in the community. We assume that an individual may receive assistance with finding a job or having other needs met through informal sources like a bilingual community member. Community members may broker job information for those with limited English proficiency. By examining these relationships, this article fills gaps in economic integration research by focusing on community-level language resources. This study goes beyond the individualist perspective that centers on an individual's ability to communicate in English. At the same time, this study sheds light on community language resources. Accordingly, this study differs from existing studies that identify community language resources with ethnic resources. This study asks the following questions: First, do community language resources increase immigrants’ employment chances? Second, does the effect of community language resources on employment differ by individual immigrants’ ability to communicate in English? Third, does the effect of community language resources on employment differ by gender? We ask this question because labor market experiences differ between immigrant women and men (Silles 2018) and ethnic social networks are often gendered (Allen 2009).

Background

Based on human capital theory (Becker 1964), the extant literature identifies English proficiency as a critical determinant of immigrants’ labor market success (Adserà and Pytliková 2016; Chiswick and Miller 2014; Isphording 2015). The ability to communicate in English increases immigrants’ chances of getting and maintaining jobs because it is a medium through which premigration human capital is transferred to the host country's labor market (Isphording 2015). English proficiency also facilitates immigrants’ job search by enhancing their access to job information (Guerrero and Rothstein 2012). At the same time, the ability to speak English increases immigrants’ cultural knowledge related to work and enables them to acquire new skills in the U.S. labor market (Guerrero and Rothstein 2012; Adserà and Pytliková 2016). Numerous empirical studies show that English proficiency has a positive association with employment status (Adserà and Pytliková 2016; Potocky-Tripodi 2004), earnings (Adserà and Pytliková 2016; Chiswick and Miller 2014), and the quality of jobs obtained (Connor 2010; Guerrero and Rothstein 2012).

Not every study, however, agrees that English proficiency is a prerequisite for immigrants’ labor market participation, especially for low-skilled jobs (Hagan 1998; Mouw and Chavez 2012). Based on social capital theory (Portes 1998), these studies show that immigrants with limited English proficiency often find and maintain their jobs through social networks, especially within their ethnic communities. They often get jobs using information passed through ethnic networks or referrals to open positions by co-ethnic community members (Cavello 2019; Hagan 1998; Waldinger 1994). In addition, chain employment is common among immigrant communities: once an employer hires a member of an ethnic community, additional job information spreads through the community, resulting in additional hires (Waldinger 1994; Cavello 2019; Mouw and Chavez 2012).

Insightful as it is, previous literature on ethnic social networks (Waldinger 1994; Cavello 2019; Mouw and Chavez 2012) often overlooks the possibility that community language resources may not be identical to ethnic resources. Ethnic communities with large numbers of co-ethnic members may not have rich community language resources if they only have a small number of bilingual members (Nawyn et al. 2012). These communities may be unable to build and maintain ethnic job networks because chain employment starts with bilingual “seedbed” immigrants who were first hired by a firm without connections. To these “seedbed” immigrants, English proficiency is essential in getting and maintaining jobs because they must obtain information about available jobs, have successful job interviews with potential employers, and understand supervisors’ instructions without language assistance (Cavello 2019; Hagan 1998). In addition, bilingual members of large ethnic communities with a low level of community language resources are often overburdened and, therefore, may avoid connections within their ethnic communities (Nawyn et al. 2012).

Furthermore, the boundary of an ethnic community is not always identical to that of the linguistic community, although a common language is often a key marker of an ethnic boundary (Schlee 2015). Multiple ethnic groups may share a common language. For example, Spanish is the first language among people from Central and Latin America and Spain. Alternatively, some ethnic groups consist of people speaking different languages, as shown in the example of the Garre in Northeastern Africa (Schlee 2015). It is plausible that linguistic networks, not ethnic networks, sometimes link new immigrants of multiple ethnicities with job opportunities. According to Pfeffer and Parra (2009), a Mexican undocumented immigrant obtained job information from non-Mexican Spanish-speaking people who crossed the border with him. Using this information, he got his first job in the United States at a farm that had not yet hired someone from Mexico (Pfeffer and Parra 2009). Immigrants with limited English proficiency often receive language assistance outside their ethnic community, provided they speak a common language.

The existing literature also suggests that individual English proficiency may have distinct impacts on labor market outcomes by gender. Although many previous studies show that English proficiency is positively associated with employment status for both immigrant men and women (Miyar-Busto, Díaz and Gutiérrez 2020; Silles 2018), one study finds that language ability significantly increases the chance of being employed only among immigrant men, and not among immigrant women (Dustmann and Fabbri 2003).

To our knowledge, no empirical study has examined whether the level of community language resources has distinct impacts on employment between immigrant men and women. Existing literature on social networks, however, suggests that gender influences how community language resources affect one's probability of being employed due to the gender-divided nature of social networks in ethnic communities: in general, immigrant men's ethnic social networks provide richer resources (e.g., the quantity and quality of information) than those of immigrant women, even within the same ethnic community (Hagan 1998). Traditional values in many ethnic communities may foster gender role ideology, obstructing or discouraging immigrant women from obtaining jobs outside the home while assigning breadwinner roles to immigrant men (Allen 2009; Gowayed 2019). For this reason, immigrant women with strong ties to their ethnic community are more likely to conform to traditional values, beliefs, and norms, and less likely to have gainful employment than their counterparts without these ties (Allen 2009). Considering the overlap between ethnic and linguistic community boundaries, community language resources may affect men and women differently in the labor market.

Built on the existing research, this study investigates the role of language in immigrants’ chances of employment. This study expands our understanding of immigrants’ economic integration by considering (1) individuals’ ability to communicate in English, (2) community-level language resources, and (3) the interaction between individual English proficiency and community-level language resources.

Study Setting: Western New York State

Western New York (WNY) is a relatively new destination for immigrants. A rust-belt area composed of five counties in New York state (Allegany, Cattaraugus, Chautauqua, Erie, and Niagara), WNY experienced population decline from the 1950s until the 2000s. However, this population loss was halted in the 2000s, mainly due to the arrival of immigrants and refugees (Office of the New York State Comptroller 2016). The proportion of the foreign-born population increased from 3.9% in 2000 to 5.8% in 2019 (authors’ own calculation using information from the Census Bureau located at https://data.census.gov).

The newly arrived foreign-born population consists mainly of two groups: professionals and refugees. The region's strategic development plans focus on higher education, healthcare, and advanced manufacturing industries, and have attracted highly educated and highly skilled professional immigrants to the area (Western New York Regional Economic Development Council 2011). At the same time, the number of refugees continuously grew after WNY was designated a refugee resettlement area by the Office of Refugee Resettlement: between 2003 and 2019, approximately 16,000 refugees were accepted into Erie County, the most populous county in WNY (Kim, Kim and Kim 2020; Adelman, Balta Ozgen, and Rabii 2019). Reflecting the two different types of foreign-born populations, WNY's immigrant population's socioeconomic characteristics show polarized distributions. In terms of formal educational attainment, the foreign-born population has higher percentages of people with either low or high levels of education than their native-born counterparts: 16% of immigrant men and 20% of immigrant women do not have any formal educational degree, whereas the comparable statistics are 7% and 6% among native-born men and women. Individuals with a graduate or professional degree compose 23% and 21% of immigrant men and women, respectively, and 10% and 15% of native-born men and women, respectively. Similarly, household income distribution is polarized: while mean household income is significantly higher among immigrant households than native-born households ($75,000 versus $67,000), immigrants’ poverty rate is higher than that of the native-born population (18% versus 15%) (Nam, Richards-Desai and Holler 2019).

Due to their brief tenure in WNY, immigrant communities have few institutionalized resources, including language assistance. Erie County has four refugee resettlement agencies, several nonprofit organizations providing healthcare and small-business support, and grassroots ethnic community organizations (Adelman et al. 2021). While metropolitan areas with larger, more established immigrant populations may have many non-English-language newspapers, radio stations, and broadcasting companies, WNY has only a few non-English newspapers, including Panorama Hispano News in Spanish and Karibu News published in six languages (Arabic, Burmese, Somali, Karen, Swahili, and Spanish). The scarcity of institutionalized language assistance programs likely forces immigrants with a language barrier to rely on informal assistance from bilingual community members.

Additionally, many immigrants in WNY speak languages that cross ethnic or national boundaries. According to our analyses of the 2012–2016 American Community Survey (ACS) data, six out of the top ten non-English languages are used by people from multiple countries. For example, Spanish, the most common non-English language, is a language spoken at home by people from 23 countries (e.g., Mexico, Ecuador, and Cuba), while Arabic, the second-most common non-English language in the area, is spoken by people from 12 countries throughout the Middle East and North Africa (e.g., Yemen, Iraq, and Egypt). French, the eighth-most popular language, is used by people from various continents, including Africa (e.g., Congo and Guinea), Asia (e.g., Myanmar), Europe (e.g., France), Central America (e.g., Belize), and North America (Canada). The language composition of the local immigrant community makes WNY an ideal location to study community language resources that often cross-ethnic or nationality boundaries.

Methods

Data and Sample

This study uses five-year compiled data from the ACS Public Use Microdata Sample (PUMS) collected from 2012 to 2016. As data from a nationally representative sample, the ACS offers reliable information on the U.S. population's social, economic, housing, and demographic characteristics. The ACS also contains data related to immigration, such as place of birth, year of entry to the United States, and citizenship status (U.S. Census Bureau 2020). As one of the largest surveys in the United States, the ACS data provide accurate estimates for small geographic areas (e.g., counties) when the five-year compiled data are used (U.S. Census Bureau 2016). In sum, the ACS is a suitable data source for studying small population subgroups in a relatively small geographic area (e.g., immigrants in WNY).

The sample in this study consists of immigrants living in five counties in WNY. We define immigrants as those born outside the United States whose parents are not U.S. citizens. We restrict our sample to prime working-age individuals (aged 25–54 years) who are not currently enrolled in school. The sample excludes those living in group quarters because their labor market participation is often limited. The final analysis sample comprises 608 immigrant men and 692 immigrant women.

Measures

The dependent variable of this study is current employment status. We assign the value of 1 to those currently employed either in the civilian labor force or in the armed forces, and 0 to those unemployed or not in the labor force.

The main independent variables are three language-related variables: Individual-level English language barrier, community-level language resources, and the interaction between these two language variables. Individual English language barrier is an ordinal variable created using two survey items in the ACS data: “Does this person speak a language other than English at home?” and “How well does this person speak English?” (U.S. Department of Commerce, 2012A). First, this study assigns the value “0” to people who do not speak a language other than English (“speaking English only (0)”). Second, for those who speak a non-English language, we allocate values based on their English proficiency: Very well (1), Well (2), Not well (3), and Not at all (4). Accordingly, the language barrier variable has five response categories, with a higher value indicating a more severe English language barrier.

The community language resource variable measures the percentage of bilingual adults among those speaking the same language in a community. In this measure, we use the Public Use Microdata Area (PUMA) as the unit of community. A PUMA is the smallest geographic unit available to the public in the ACS's PUMS dataset, and each PUMA contains data for no fewer than 100,000 people (U.S. Census Bureau 2022). In creating the community language resources, we first count the number of adults who speak English very well in each PUMA for each non-English language. Then, we divide it by the number of adults in a community for each non-English language, generating the percentage of adults fluent in English for each language. This aggregated information is matched with individuals in the sample based on their languages. Individuals who speak only English have a value of 100% because all individuals in this group speak English proficiently. Accordingly, every individual in a specific language group has the same value for the community language resource variable. We create the community language resource variable in this way based on the assumption that immigrants with a language barrier may be able to obtain language-related assistance from those who speak both English and their own language in their community. Existing studies on ethnic networks and language proficiency use a minority language concentration measure defined in this way (Chiswick and Miller 1996). Existing qualitative studies show that bilingual individuals help their community members with limited English proficiency, often sacrificing their own personal interests, for example, by missing work to assist (Nawyn et al. 2012). The interaction variable is created by multiplying the variable for the individual-level English language barrier by the community language resource variable.

Control variables include demographics, household characteristics, human capital, immigration-related variables, and other factors. (Please refer to Table 1 for the list of control variables.) Demographics include age and race/ethnicity. Age is a continuous variable, indicating an individual's age in the year the survey was administered. Race/ethnicity has five categories: Non-Hispanic White, Non-Hispanic Black, Non-Hispanic Asian, Hispanic, and Other. Household characteristics include household size, marital status, the number of children, and an indicator of having a young child. Household size indicates the number of people in the household. Householder's marital status has three categories: Currently married; Widowed, separated, or divorced; and Never married. The number of children is a continuous variable that counts the number of children in the household. An indicator of the presence of a young child is a dichotomous variable (a value of 1 to households with at least one child under the age of five years and 0 to others). Human capital has two variables: Education and Disability status. The education variable consists of five categories: No degree, High school diploma or General Education Diploma, Associate's degree, Bachelor's degree, and Graduate degree. Disability status indicates whether or not an individual has a disability. Immigration-related variables include the number of years in the United States and their citizenship status. The number of years in the U.S. is created by subtracting the year of entry into the U.S. from the survey year, yielding the duration someone has resided in the United States. The citizen variable is a dichotomous variable separating naturalized citizens from noncitizens. Other factors include the unemployment rate, county indicator, and survey year. The unemployment rate is a continuous variable of county-level yearly unemployment rates for that survey year, constructed using unemployment information from the New York State Department of Labor's Labor Force and Unemployment Data website (https://dol.ny.gov/labor-data). Since the ACS does not provide separate geographic indicators for Allegany and Cattaraugus counties, we use the combined unemployment rate in the Allegany/Cattaraugus Local Workforce Investment Area for these two counties. We use the following dummy variables for the county of residence: Allegany and Cattaraugus; Chautauqua; Erie; and Niagara. The survey year variable denotes the year the survey data were collected and consists of five categories (2012, 2013, 2014, 2015, and 2016).

Table 1.

Sample Characteristics by Gender.

	Men	Women
Language-related
English language barrier **
English only	25.01%	24.51%
Very well	38.32%	33.13%
Well	22.41%	19.05%
Not well	12.09%	16.98%
Not at all	2.17%	6.34%
Community language resources, Mean	64.29	63.27
Demographics
Age, Mean	39.72	39.45
Race/Ethnicity
White	39.95%	35.87%
Black	13.34%	10.98%
Asian	35.14%	38.74%
Hispanic	7.79%	11.06%
Other	3.78%	3.35%
Household characteristics
Household size, Mean	3.7	3.9
Marital status
Married	74.59%	75.24%
Widowed, separated, divorced	8.01%	10.51%
Never married	17.41%	14.25%
Number of children, Mean ⁺	1.13	1.30
Young child	30.67%	31.35%
Human capital
Education
No degree	15.50%	19.86%
High school diploma	36.76%	30.85%
Associate degree	5.95%	8.11%
Bachelor's degree	19.17%	20.01%
Graduate degree	22.62%	21.16%
Disability	4.03%	5.10%
Immigration-related
Number of years in the U.S.
5 or fewer	20.58%	20.82%
6–20	46.73%	50.05%
21 or more	32.69%	29.13%
Naturalized citizen	52.83%	49.93%
Other factors
Unemployment rate	7.16	7.15
County
Allegany and Cattaraugus	3.54%	3.31%
Chautauqua	3.55%	3.96%
Erie	82.59%	81.55%
Niagara	10.32%	11.18%
Survey year
2012	18.50%	18.18%
2013	18.87%	19.36%
2014	18.60%	17.98%
2015	23.35%	23.14%
2016	20.68%	21.35%
Sample size	608	692

⁺

p < .1.

^**

p < .01.

Statistical Models

This study utilizes logit regressions because the dependent variable, employment status, is dichotomous. We run separate regression analyses for immigrant women and men. As described earlier, immigrant women are likely to have different economic experiences than immigrant men (Silles 2018). Each regression model includes three language-related variables, demographics, household characteristics, human capital, immigration-related information, and other factors. In consideration of a nonlinear relationship between age and one's probability of being employed, the analysis model contains “age squared” in addition to the age variable. The regression analyses contain county and survey year dummy variables to control for unobserved contextual differences among these geographic areas and survey years. Since interpretation of results from regressions with an interaction term is not easy, we present predicted probabilities of being currently employed. For both descriptive and regression analyses, we use the weighted data using a person-weight variable created by the ACS. The analysis model is expressed as follows:

Y i = α + β_{1} * L_{i} + β_{2} * C_{i} + β_{3} * (L_{i} * C_{i}) + β_{4} * X_{i} + ε_{i}

(1)

where Y _i is the log-odds of being currently employed of an immigrant i; L _i indicates individual-level English language barrier of an immigrant i; C _i denotes community language resource of immigrant i; X _i is a vector of control variables of an immigrant i; and ε _i represents a random error.

In equation (1), β₁, the coefficient of individual-level English language barrier, indicates an immigrant's probability of being employed as explained by one's ability to speak English when the community language resource level is zero. We expect this coefficient to be negative: the more severe an immigrant's level of English language barrier, the lower their probability of being employed. The coefficient of community language resources (β₂) estimates how the likelihood of being employed changes as community language resources increase among immigrants who speak only English. Because a proficient English speaker's chance of being employed is not likely affected by the presence of community language resources, we expect this coefficient to be not significantly different from zero. This study's primary interest is β₃, the coefficient of the interaction term between the two language variables. The interaction term estimates changes in employment probability among those with an English language barrier as community language resources increase. We expect the interaction term coefficient to be significantly positive, as one's chances of being employed would be expected to improve among those with an English language barrier when community language resources increase.

We run a series of supplemental analyses to check the robustness of this study's findings. First, this study runs probit regressions in place of logit regressions. Second, we run regressions using a continuous variable describing an immigrant's number of years in the United States instead of the categorical variable used in the main model. Third, we run a regression model that adds a refugee indicator, as the increasing number of resettled refugees is a driving force of immigrant population growth in WNY (Partnership for the Public Good 2018), and labor market experiences may differ between refugees and other immigrants (Chiswick, Lee and Miller 2006; Chin and Cortes 2015). Because the ACS does not contain a refugee status indicator in the survey, we generate a proxy measure of refugee status (those from refugee-sending countries), using the method developed by Passel and Clark (1998). Fourth, we run analyses after excluding those who speak only English. Results from these supplemental analyses do not differ substantively from those from the main analysis model. When supplemental analyses generate different results from those reported in this article, we describe them in the Results section. (Full results of these analyses are available from the authors.)

Results

Table 1 reports sample characteristics by gender. As shown in Table 1, many immigrant men and women face an English language barrier. At the same time, immigrant women have more severe language barriers than their male counterparts. Among immigrant women, 6% cannot speak English at all, and 17% do not speak English well, while comparable statistics are 2% and 12% among immigrant men. A gender difference in individual English language proficiency is statistically significant at the 0.01 level. Community language resources do not differ by gender: on average, both immigrant men and women live in communities where 63–64% of adults who speak the same language are fluent in English.

In terms of demographics, the average age is about 40 years for both men and women. Regarding racial and ethnic composition, 35% of men and 39% of women are Asian, 13% and 11% are Black, 8% and 11% are Hispanic, and 40% of men and 36% of women are non-Hispanic white people. Results on household characteristics show that both women and men, on average, live in households of four. About three-quarters are currently married. Immigrant women have a slightly larger number of children than immigrant men (1.3 versus 1.1, p < 0.1). About 30% of both men and women have at least one young child in their household. Analyses of the two human capital variables do not show significant differences between immigrant men and women. Although the percentage of immigrant women with no formal educational degree is slightly higher than among men (20% versus 16%), the percentages of women and men with a higher education degree (associate, bachelor, or graduate degree) are comparable. The percentage of people with a disability is similar between genders. Results from immigration-related variables indicate that about 20% of immigrants of both genders have lived in the United States for five or fewer years, and about half of immigrant men and women are naturalized citizens. The average unemployment rate is around 7% for both immigrant men and women. Table 1 also shows that most immigrant men (83%) and women (82%) in WNY live in Erie County.

Figure 1 presents current employment rates by gender and the level of the individual English language barrier. We categorize those who speak “English only” and those who speak English “very well” as fluent in English, and the others as those with a language barrier, following Zong and Batalova (2015). Figure 1 clearly shows different levels of current employment status by gender and by individual English language barrier. Immigrant men's employment rates are higher than women's, regardless of individual language proficiency. Among immigrant men, the percentage who are currently employed is 86% for those fluent in English and 78% for their counterparts with a language barrier, showing an eight percentage-point difference (p < 0.01). The comparable statistics among immigrant women are 69% and 42%, respectively (p < 0.01).

Figure 1.

Employment Rate by Gender and Language Barrier.

Table 2 presents analysis results from two logit regressions: one from the sample including only immigrant men and the other including only immigrant women. Results from the control variables are as expected. Immigrants with disabilities are significantly less likely to be employed than those without disabilities (p < 0.01), and long-term residents of the United States are more likely to be employed than recent immigrants (p < 0.05). The association between marital status and employment probability varies across genders. Never-married immigrant men are significantly less likely to be employed than their married counterparts. In contrast, never-married immigrant women are significantly more likely to be employed than married immigrant women. The employment probability of widowed, separated, or divorced immigrant men does not significantly differ from that of married immigrant men. However, among immigrant women, widowhood, separation, and divorce are significantly and positively associated with an individual's employment probability.

Table 2.

Logit Regression Results: Current Employment Status.

	Men		Women
	Coeff.	S.E	Coeff.	S.E
Language-related
Individual language barrier	−0.93*	0.39	−0.16	0.32
Community language resources	−1.33	1.04	0.12	0.79
Interaction, 2 language variables	0.93⁺	0.50	−0.46	0.44
Demographics
Age	−0.13	0.19	0.26⁺	0.16
Age squared	0.00	0.00	0.00⁺	0.00
Race/Ethnicity
Black	−0.34	0.41	0.83⁺	0.47
Asian	0.55	0.35	−0.02	0.26
Hispanic	0.77	0.52	0.48	0.43
Other	−0.94	0.74	−0.02	0.51
Household characteristics
Household size	−0.16*	0.08	0.04	0.09
Marital status
Widowed, separated, divorced	−0.19	0.44	1.26**	0.41
Never married	−0.75⁺	0.40	1.83**	0.42
Number of children	0.16	0.16	−0.20	0.12
Young children	0.20	0.39	−1.02**	0.31
Human capital
Formal education
High school diploma	−0.37	0.42	0.42	0.35
Associate degree	0.10	0.73	0.24	0.42
Bachelor's degree	−0.58	0.50	0.22	0.40
Graduate degree	−0.19	0.54	1.38 **	0.42
Disability	−2.23**	0.47	−2.02 **	0.62
Immigration-related
Years in the U.S.
6–19 years	0.78*	0.37	1.13**	0.33
20 or more years	0.43	0.47	1.46**	0.39
Noncitizen	–0.16	0.30	–0.20	0.24
Other factors
Unemployment rate	0.66	1.34	–1.48	1.13
Constant	0.72	12.54	8.09	10.52
F-Statistics	F(30, 578) = 2.43**		F(30, 662) = 3.24**
Sample size	608		692

Note: Logit regressions include county and survey year dummy variables.

* p < .05.

** p < .01.

+ p < .1.

Table 2 shows that the three language variables have the expected results for immigrant men but not for immigrant women. As predicted, the coefficient for an individual's English language barrier is significantly negative, indicating that immigrant men with a more severe language barrier are less likely to be employed than those without barriers (p < 0.05). The community language resource variable has a coefficient that is not statistically significant, implying that the level of community language resources has little impact on the employment status of immigrant men who speak only English. The interaction term of individual and community language variables has a significantly positive coefficient (p < 0.1), suggesting that a high level of community language resources increases employment probability for immigrant men with an English language barrier. The regression analyses generate different results for immigrant women: none of the three language variables are statistically significant, suggesting that neither an individual language barrier nor community language resources are associated with immigrant women's chance of being employed in a statistically significant way.

We estimate the predicted probabilities of being employed by gender, individual English language barrier, and community language resources. Predicted probabilities are calculated for a typical immigrant in the sample with the following characteristics: the immigrant is 40 years old, white, and married; living in a household of four members with two children and no young child; has a high school diploma and does not have a disability; has lived in the United States for 5–19 years; has received citizenship through naturalization; lives in Erie county; the county's unemployment rate is 7.0%; and the data were collected in 2015.

Figure 2 presents the predicted probabilities of being employed. Figure 2 demonstrates the roles of both individual language barriers and community language resources among immigrant men. For all levels of community language resources, employment probabilities are higher among immigrant men who are more fluent in English. The association between community language resources and employment probability differs by individual-level English proficiency. Employment rates improve to the largest extent among immigrant men who cannot speak English at all as community language resources increase: the predicted probability of employment jumps from 60% to 91% as the level of community language resources increases from 10% to 90%, showing a 31 percentage-point increase. The impact of community language resources is predicted to be smaller among immigrant men who speak English “not well” and “well”: an increase in community language resources from 10% to 90% raises the employment rate by 14 and four percentage points, respectively. On the contrary, the predicted employment rates among immigrant men who speak English “very well” show little change as community language resources increase. As a result, the difference in the employment probability between immigrant men fluent in English and those with no knowledge of English declines as the level of community language resources increases: while the gap is estimated to be 35 percentage points when a community language resource is 10%, it declines to 2 percentage points when a community language resource reaches 90%.

Figure 2.

Predicted Probabilities of Being Employed by Individual English Proficiency & Community Language Resources.

Predicted probabilities of employment among immigrant women show different patterns than those for immigrant men. As in the case of immigrant men, immigrant women with greater English proficiency have a higher chance of being employed. However, community language resources are not associated with improved employment probabilities among immigrant women with a language barrier. Instead, a higher level of community language resources is associated with decreased employment probability among immigrant women with a language barrier (a regression coefficient is not statistically significant, as shown in Table 2). For example, the predicted probability of being employed among those who cannot speak English drops from 20% to 6% as the community language resource level increases from 10% to 90%.

Supplement analyses generate substantively similar results to those reported in this paper, with a few exceptions. When a regression is conducted on the sample without immigrant men who speak only English, the interaction term of the two language variables is not statistically significant at the 0.1 level. However, its coefficient is positive (as in the main analysis) and statistically significant at the 0.2 level (p = 0.16). In addition, its coefficient size (1.19) is larger than that in the main analysis. A decrease in sample sizes (from 608 to 448) may explain the loss of statistical significance in the supplemental analysis excluding immigrant men speaking only English.

Discussion

This study investigates the role of community language resources in facilitating immigrants’ employment using ACS data collected in WNY. This study contributes to the literature by considering the possibility that immigrants with an English language barrier may obtain and maintain a job if they receive assistance from bilingual individuals in their community and that an individual's chances of getting help are higher if the community has more language resources (a large number of individuals who can speak both English and their own language). In this way, this study differs from existing literature. This study is different from human capital theory by considering community characteristics in addition to individual-level English proficiency (Adserà and Pytliková 2016; Connor 2010, Guerrero and Rothstein 2012; Dustmann and Fabbri 2003; Silles 2018). This study differs from the existing literature on ethnic social networks, which does not differentiate community language resources from ethnic community resources (Maouw and Chavez 2012; Waldinger 1994). This study incorporates human capital theory (an individual's ability and skills valued in the labor market) and social capital theory (community resources) while investigating immigrants’ economic integration.

Using the percentage of individuals fluent in English among adults speaking the same non-English language in a community as the measure of community language resources, this study finds that a higher level of community language resources increases the chance of being employed among immigrant men with an individual-level language barrier. This finding supports this study's hypothesis of the effect of community language resources. At the same time, this study shows the gendered nature of community language resources by demonstrating the distinct effect of community language resources among immigrant women: the level of community language resources is not associated with a higher employment probability among immigrant women with a language barrier in a statistically significant way. Furthermore, predicted probabilities suggest that a higher level of community language resources may reduce the chance of being employed among immigrant women with an individual language barrier. The opposite patterns of association between community language resources and employment status by gender are consistent with existing studies on the gendered nature of ethnic social networks (Hagan 1998; Allen 2009). What remains unexplained by this study is whether or not the gendered nature of community language resources operates similarly to that of ethnic social networks.

This study has the following limitations. First, this study investigates only current employment status, leaving other labor market outcomes unstudied. We cannot rule out the possibility that community language resources may have distinct associations with other economic indicators (e.g., earnings and job quality). For example, relying on other people for language assistance may delay the acquisition of English proficiency, which may deter immigrants’ upward mobility in the labor market. Further research on other labor market outcome measures is warranted. Second, this study cannot explain why the associations between community language resources and employment outcomes differ by gender. Future studies should explore the gendered mechanism through which community language resources facilitate or hinder immigrants’ entry into and subsequent mobility in the labor market. Third, this study uses data from a relatively small geographic area, WNY. Accordingly, findings from this study are not generalizable to other areas. Additional research should explore the role of community language resources in geographic areas with different immigration histories and service infrastructures for immigrants. Fourth, the measure of community language resources may not be comprehensive enough. Our measure is a proxy for informal language resources bilingual community members provide. Accordingly, the measure in this study cannot capture formal and institutional community language resources, such as mandated language services or interpretation assistance, including language phone lines. Finally, we cannot rule out the possibility that the undercounting of immigrants in the survey data may have affected the results of this study. Survey data, including those collected by the U.S. Census Bureau, tend to undercount immigrants and refugees (Bernstein and DuBois 2018; Kissam 2017). Several reasons immigrants may be undercounted in survey data include fear of discrimination and deportation, limited English proficiency, and unstable housing situations (Bernstein and DuBois 2018; Garcia 1992; Kaneshiro 2013; Kissam 2017). Some immigrants may also fear discrimination, deportation, and other grave consequences due to anti-immigrant rhetoric and, therefore, do not fill out the survey (Kaneshiro 2013; Kissam 2017). The underrepresentation of immigrants with limited English proficiency in the data may have influenced our analysis results.

This study has implications and suggestions for future research and policy development. First, community language resources may be distinct from ethnic community resources, as linguistic communities differ from ethnic communities. Most existing studies combine community language resources and ethnic community resources assuming that linguistic communities are identical to ethnic communities (Beaman 2012; Nawyn et al. 2012). This is understandable, considering that a common language is one of the significant markers of an ethnic community (Schlee 2015). However, ethnic communities are not always congruent to linguistic communities (Schlee 2015; Pfeffer and Parra 2009). Accordingly, it is essential to deepen our understanding of community language resources that differ from ethnic community resources. Second, service providers and policymakers should develop programs and policies that mobilize community language resources to promote immigrants’ economic integration. This study's findings suggest that informal community language resources may promote employment among some immigrants with a language barrier. Relying solely on informal language resources is unsustainable as bilingual individuals are often overburdened with family members’ and friends’ requests for language assistance (Nawyn et al. 2012). Accordingly, it is urgent to develop and fund formal and institutionalized language assistance programs and policies (e.g., paid interpretation services and translation of public documents) to ensure meaningful access to services and benefits among immigrants with a language barrier. This study's findings call for enacting and implementing language access legislation (e.g., mandated language services in the Affordable Care Act) at the local, state, and federal levels. Third, it is critically important to recognize gender differences in research, service provision, and policy development. As shown in this study, community language resources have different associations with employment status by gender. Accordingly, a lack of attention to gender differences will hamper future research, service provision, and policy development.

Footnotes

Acknowledgments

This study has been supported in part by a grant from New York State Office of New Americans (ONA) Workforce Community Education, Community Navigator, and Welcoming Communities Program. The authors are grateful to Ms. Michelle Holler at Journey's End Refugee Services, Inc. for valuable partnership for the ONA grant program. The authors thank Harmony Neal for their wonderful editing assistance.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the New York State Office of New Americans (ONA) Workforce Community Education, Community Navigator and Welcoming Communities Program.

ORCID iD

Yunju Nam

References

Adelman

Robert M.

Cooke

Abigail

Kim

Wooksoo

Nam

Yunju

. 2021. “Another Voice: New refugee policy would bring benefits to Buffalo.” The Buffalo News, January 12, 2021. https://doi.org/10.1111/cico.12412 https://buffalonews.com/opinion/another-voice-new-refugee-policy-would-bring-benefits-to-buffalo/article_0f872256-54c8-11eb-ac20-27b89ddf2f34.html

Adelman

R. M.

Balta Ozgen

Rabii

. 2019. “Buffalo's West Side Story: Migration, Gentrification, and Neighborhood Change.” City & Community 18 (3): 770–91.

Adserà

Mariola

. 2016. “Language and Migration.” In The Palgrave Handbook of Economics and Language, edited by Ginsburgh

Weber

, 342–72. Basingstoke, UK: Palgrave Macmillan.

Allen

2009. “Benefit or Burden? Social Capital, Gender, and the Economic Adaptation of Refugees.” International Migration Review 43 (2): 332–65. https://doi.org/10.1111/j.1747-7379.2009.00767

Beaman

L. A.

2012. “Social Networks and the Dynamics of Labour Market Outcomes: Evidence from Refugees Resettled in the U.S.” Review of Economic Studies 79 (1): 128–61. https://doi.org/10.1093/restud/rdr017

Becker

G. S.

1964. Human Capital. New York: National Bureau of Economic Research.

Bernstein

DuBois

. 2018. Bringing Evidence to the Refugee Integration Debate. Washington, D.C.: Urban Institute. https://www.urban.org/sites/default/files/publication/97771/bringing_evidence_to_the_refugee_integration_debate_0.pdf

Budiman

Tamir

Mora

Noe-Bustamante

. 2020. Facts on U.S. Immigrants, 2018: Statistical Portrait of the Foreign-Born Population in the United States. Washington, D.C.: Pew Research Center. https://www.pewresearch.org/hispanic/2020/08/20/facts-on-u-s-immigrants/

Cavello

S. M.

2019. Rust Belt Renaissance? The Experience of Refugees from Burma in Buffalo, New York.” Department of Geography, State University of New York at Buffalo.

10.

Chin

. and Cortes

K. E.

. 2015. “The Refugee/Asylum Seeker.” In Handbook of the Economics of International Migration, Volume 1, edited by Chiswick

R. B.

Miller

P. W

, 585–658. Amsterdam: North-Holland.

11.

Chiswick

B. R.

Lee

Y. L.

Miller

P. W.

. 2006. “Immigrants’ Language Skills and Visa Category.” International Migration Review 40 (2): 419–50. https://doi.org/10.1111/j.1747-7379.2006.00023

12.

Chiswick

B. R.

Miller

P. W.

. 1996. “Ethnic Networks and Language Proficiency among Immigrants.” Journal of Population Economics 9 (1): 19–35. https://www.jstor.org/stable/20007488

13.

Chiswick

B. R.

Miller

P. W.

. 2014. “International Migration and the Economics of Language.” In In Handbook of the Economics of International Migration, edited by Chiswick

R. B.

Miller

, 211–69. Elsevier Science & Technology.

14.

Congressional Budget Office. 2004. A Description of the Immigrant Population. Washington, D.C.: The Congress of the United States.

15.

Connor

2010. “Explaining the Refugee Gap: Economic Outcomes Of Refugees Versus Other Immigrants.” Journal of Refugee Studies 23 (3): 377–97. https://doi.org/10.1093/jrs/feq025

16.

Dustmann

Fabbri

. 2003. “Language Proficiency and Labour Market Performance of Immigrants in the UK.” The Economic Journal 113 (489): 695–717. https://doi.org/10.1111/1468-0297.t01-1-00151

17.

Garcia

1992. Counting the Uncountable, Immigrant and Migrant, Documented and Undocumented Farm Workers in California: Results from an Alternative Enumeration in a Mexican and Mexican American Farm Worker Community in California and Ethnographic Evaluation of the Behavioral Causes of Undercount. Ethnographic Evaluation of the 1990 Decennial Census Report #12. Washington, D.C.: Bureau of the Census.

18.

Gowayed

2019. “Diverging by Gender: Syrian Refugees’ Divisions of Labor and Formation of Human Capital in the United States.” Gender & Society 33 (2): 251–72. https://doi.org/10.1177/0891243218819753

19.

Guerrero

Rothstein

M. G.

. 2012. “Antecedents of Underemployment: Job Search of Skilled Immigrants in Canada.” Applied Psychology 61 (2): 323–46. https://doi.org/10.1111/j.1464-0597.2011.00470

20.

Hagan

J. M.

1998. “Social Networks, Gender, and Immigrant Incorporation: Resources and Constraints.” American Sociological Review 63 (1): 55–67.

21.

Hao

Kawano

. 2001. “Immigrants’ Welfare Use and Opportunity for Contact with Co-Ethnics.” Demography 38 (3): 375–89. https://doi.org/10.1111/j.1747-7379.2009.00783

22.

Isphording

I. E.

2015. “Language and Labor Market Success.” In International Encyclopedia of the Social and Behavioral Sciences, edited by Wright

J. D.

, 260–5. London, UK: Elsevier.

23.

Kaneshiro

2013. “Missing Minorities? The Phases of IRCA Legislation and Relative Net Undercounts of the 1990 vis-à-vis 2000 Decennial Census for Foreign-Born Cohorts.” Demography 50 (5): 1897–919. https://doi.org/10.1007/s13524-013-0212-6

24.

Kim

. 2020. Resettlement Trends for Refugees in Erie County (2003–2019). Buffalo, NY: Immigrant and Refugee Research Institute, The State University of New York at Buffalo School of Social Work.

25.

Kissam

2017. “Differential Undercount of Mexican Immigrant Families in the US Census.” Statistical Journal of the IAOS 33 (3): 797–816. https://doi.org/10.3233/sji-170388

26.

Miyar-Busto

Díaz

Fco. J. M.

Gutiérrez

Rodolfo

. 2020. “Immigrants’ Educational Credentials Leading to Employment Outcomes: The Role Played by Language Skills.” Revista Internacional De Organizaciones 23: 167–91. https://doi.org/10.17345/rio23.167-191

27.

Mouw

Chavez

. 2012. “Occupational Linguistic Niches and the Wage Growth of Latino Immigrants.” Social Forces 91 (2): 423–52.

28.

Nam

Yunju

Richards-Desai

Sarah

Holler

Michelle

. 2019. “Immigrants’ Economic Integration in Western New York: Evidence from the American Community Survey.” Report to New York State Office of New Americans. Buffalo, NY: University at Buffalo.

29.

Nawyn

S. J.

Gjokaj

Agbenyiga

D. L.

Grace

. 2012. “Linguistic Isolation, Social Capital, and Immigrant Belonging.” Journal of Contemporary Ethnography 41 (3): 255–82. https://doi.org/10.1177/0891241611433623

30.

Office of the New York State Comptroller. 2016. A Portrait of Immigrants in New York. Albany, NY: Office of the New York State Comptroller.

31.

Partnership for the Public Good. 2018. Immigrants, Refugees, and Languages Spoken in Buffalo. Buffalo, NY: Partnership for the Public Good.

32.

Passel

J. S.

Clark

R. L.

. 1998. Immigrants in New York: Their Legal Status, Income, and Taxes. Washington, D.C.: Urban Institute.

33.

Pfeffer

M. J.

Parra

P. A.

. 2009. “Strong Ties, Weak Ties, and Human Capital: Latino Immigrant Employment Outside the Enclave.” Rural Sociology 74 (2): 241–69. https://doi.org/10.1111/j.1549-0831.2009.tb00391

34.

Portes

1998. “Social Capital: Its Origins and Applications in Modern Sociology.” Annual Review of Sociology 24: 1–24. https://doi.org/10.1146/annurev.soc.24.1.1

35.

Potocky-Tripodi

2004. “The Role of Social Capital in Immigrant and Refugee Economic Adaptation.” Journal of Social Service Research 31 (1): 59–91. https://doi.org/10.1300/J079v31n01_04

36.

Schlee

2015. “Language and Ethnicity.” In International Encyclopedia of the Social and Behavioral Sciences, edited by Wright

J. D.

, 251–5. London, UK: Elsevier.

37.

Silles

M. A.

2018. “The Effects of Language Skills on the Economic Assimilation of Female Immigrants in the United States.” Manchester School 86 (6): 789–815. https://doi.org/10.1111/manc.12231

38.

Tegegne

M. A.

2015. “Immigrants’ Social Capital and Labor Market Performance: The Effect of Social Ties on Earnings and Occupational Prestige.” Social Science Quarterly 96 (5): 1396–410. https://doi.org/10.1111/ssqu.12212

39.

U.S. Census Bureau. 2016. Fact Sheet: Differences Between the American Community Survey (ACS) and the Annual Social and Economic Supplement to the Current Population Survey (CPS ASEC). Suitland, MD: U.S. Census Bureau. Last modified May 16, 2016. Accessed January 10, 2021, https://www.census.gov/topics/income-poverty/poverty/guidance/data-sources/acs-vs-cps.html

40.

U.S. Census Bureau. 2020. Understanding and Using the American Community Survey Public Use Microdata Sample Files: What Data Users Need to Know. Washington, DC: U.S. Government Printing Office.

41.

U.S. Census Bureau. 2022. Public Use Microdata Areas (PUMAs). United States Census Bureau. Last revised August 22, 2022. Accessed October 10, 2022. https://www.census.gov/programs-surveys/geography/guidance/geo-areas/pumas.html

42.

U.S. Department of Commerce. 2012. The American Community Survey . https://www.census.gov/programs-surveys/acs/about/forms-and-instructions.2012.html#list-tab-9466845

43.

Waldinger

1994. “The Making of an Immigrant Niche.” The International Migration Review 28 (1): 3–30. https://doi.org/10.2307/2547023

44.

Western New York Regional Economic Development Council. 2011. A Strategy for Prosperity in Western New York. Buffalo, NY: Western New York Regional Economic Development Council.

45.

Zong

Batalova

. 2015. The Limited English Proficient Population in the United States in 2013. Washington, D.C.: Migration Policy Institute.