Home language and societal language skills in second-generation bilingual adults

Abstract

Aims:

Research aims were (a) to test competing predictions regarding the levels of heritage and societal language proficiency among young adults who experienced early exposure to both languages, and (b) to identify sources of individual differences in degree of bilingualism.

Design/methodology/approach:

Participants comprised 65 Spanish–English bilinguals who reported using both languages on a weekly basis, 25 native English monolinguals, and 25 native Spanish monolinguals. Language and literacy skills were assessed with a battery of standardized and research-based assessments. Degree of bilingualism was calculated for the bilingual participants.

Data and analysis:

Paired sample t-tests compared the bilinguals’ skills in English to their skills in Spanish. Hierarchical regression evaluated factors related to their degree of bilingualism. Independent sample t-tests compared bilinguals’ single-language skills to monolinguals.

Findings/conclusions:

The bilinguals’ English skills were stronger than their Spanish skills on every measure. Thus, degree of bilingualism was largely a function of level of Spanish skill and was associated with concurrent Spanish exposure. Bilinguals’ English skills were not different from the monolinguals except in speed of lexical access. The bilinguals’ Spanish skills were significantly lower than the Spanish monolinguals on every measure except in accuracy judgments for grammatically correct sentences.

Originality:

Previous studies of bilingual adults have focused on sequential bilinguals and previous studies of heritage language speakers have focused on their grammatical skills. This is the first study, to our knowledge, to assess a wide range of functionally relevant skills in adults with early exposure to a heritage and societal language.

Significance/implications:

These findings demonstrate that early exposure to and continued use of a home language does not interfere with the achievement of societal language and literacy skills, while also demonstrating that even a widely and frequently used home language may not be acquired to the same skill level as a societal language used in school.

Introduction

At one time true bilingualism was defined as the ability to speak two languages, each with the same linguistic competence as a monolingual (Volterra & Taeschner, 1978). That view has largely been replaced with the acknowledgment that bilingualism takes many forms and that bilinguals can have many different patterns of linguistic competence. However, the specifics of what sort of bilingual competences are associated with different types of early dual language experience remain to be described. The aim of the present study is to describe the dual language skills observed in young bilingual adults who were exposed to a heritage and societal language from an early age. The adult bilinguals are US-born children of Spanish-speaking immigrant families who were exposed to Spanish at home from birth and who also experienced early exposure and subsequent schooling in English.

Three bodies of work are relevant to the expectations one might have for the bilingual proficiency of adults who were exposed to both a majority and heritage language at home, in early childhood. One is research on the critical period hypothesis, which finds that children can, if they are young enough when first exposed, achieve a level of proficiency in a second language that is indistinguishable from that of monolingual native speakers (Abrahamsson & Hyltenstam, 2009). However, this evidence of the important role of early language experience in language acquisition does not mean that dual language exposure from birth will result in monolingual-like levels of proficiency in two languages. Abrahamsson and Hyltenstam (2009), and most studies testing the critical period hypothesis, measured only proficiency in the second language (Hakuta et al., 2003; Johnson & Newport, 1989). Studies that have tracked the development of both languages in children who are speakers of a minority language and are exposed to a majority language have consistently found that development in the majority language overtakes development in the first language (Gathercole & Thomas, 2009; Jia & Aaronson, 2003; Montrul, 2016). The ultimate outcome is not balanced, monolingual-like proficiency in two languages.

A second relevant body of literature comes from studies of adult heritage language speakers. This research has largely focused on grammatical measures of language knowledge or production and has found incomplete acquisition and even attrition of early-acquired skills (Montrul, 2002; Montrul & Bowles, 2009; Montrul & Ionin, 2010). Adult heritage language speakers tend to appear non-native-like in numerous aspects of grammatical proficiency (see Montrul, 2010 for a review). The few studies that have examined both majority language and heritage language skills among adult early bilinguals have typically found proficiency in the majority language to be stronger than proficiency in the heritage language (Ardila et al., 2017; Montrul, 2008; Schmidtke, 2017). However, the range of skills examined in these studies was narrow. It is not clear from the literature whether heritage language skills would also differ outside the domain of morphosyntax.

The third relevant literature consists of studies of the early bilingual development of children in immigrant families. Studies here consistently find that early exposure does not guarantee acquisition of a minority language at all (De Houwer, 2007; Eilers et al., 2006). For children who continue to acquire two languages into early childhood, studies of Spanish–English bilinguals in South Florida provide evidence that majority language development may outpace heritage language development, and that the rate of growth in the majority language may lag behind monolingual children (Hoff, 2018; Hoff & Ribot, 2017). These gaps between majority and minority language skills and between bilingual and monolingual skills in the majority language have been found not only in multiple samples of Spanish–English bilinguals in the US (Gathercole, 2002; Hoff et al., 2018; Oller & Eilers, 2002) but also in other language pairs and social contexts, including Finnish–Russian bilinguals (Silven et al., 2014), and French–English bilinguals (Thordardottir, 2011, 2014).

The data on when—or whether—bilinguals catch-up to monolinguals in majority language skills are inconclusive. In the domain of vocabulary, a study of 1738 children found that bilingually developing children had consistently lower levels of receptive vocabulary skill in English than did same-aged monolingual children (Bialystok et al., 2010). Studies of Spanish–English bilingual university students in the US have found bilinguals to score lower than monolinguals on English vocabulary (Pearson, 1993; Schmidtke, 2017). In contrast, Gathercole and Thomas (2009) reported that Welsh–English bilingual children in Wales no longer differed from monolinguals in English proficiency by 11 years of age on a measure of receptive vocabulary skill.

In sum, the literature provides conflicting predictions regarding the dual language skills of young adults with early exposure to a majority and heritage language. Studies that have tested the critical period hypothesis have found that individuals can acquire a second language with native-like proficiency if they were exposed sufficiently early. The literature on heritage language acquisition suggests the dual language outcomes of heritage language speakers might be strong skills in the majority language and weaker skills in the minority language. However, it is not clear if this pattern of proficiency will be consistent across a wide array of linguistic skills or confined to certain domains. The literature on bilingual development in childhood raises the question of whether the early gap between bilinguals’ majority language skill and monolinguals will persist into adulthood.

The current study is designed to address these open questions with data on the English and Spanish skills of second-generation adults in the US who were exposed to English and Spanish from a young age and who described themselves as bilinguals who used both languages on a regular basis. The aim of the present study is to build a comprehensive picture of the dual language skills of this population by employing assessments of vocabulary, grammar, broad language comprehension skills, and speed of lexical retrieval. The specific aims were (a) to compare the bilinguals’ proficiency in English to their proficiency in Spanish; (b) to describe the degree of bilingualism achieved by these speakers and to identify sources of variance in their degree of bilingualism; and (c) to compare the bilinguals’ skill levels in each language to levels achieved by native monolingual speakers with similar levels of education.

Method

Participants

The bilingual participants were 65 young adults (M_age = 19.5 years) who were currently residing in South Florida. The English monolingual participants were 25 young adults (M_age = 19.1 years) also residing in South Florida, and the Spanish monolingual participants were 25 young adults (M_age = 21.2) residing in Chile.

The bilingual participants met the following criteria: (a) they were born in the US; (b) they had at least one parent who was born in a Spanish-speaking country; (c) they were exposed to both English and Spanish before the age of 5; and (d) they currently reported using both English and Spanish on a weekly basis. Sixty of the 65 bilingual participants were college students, four were college graduates, and one had a high school diploma and was not in college. According to participants’ self-report on the Language Experience and Proficiency Questionnaire (LEAP-Q; Marian et al., 2007), the average concurrent relative amount of English exposure was 62% (SD = 12.1). The participants attended school in the US and for the most part received formal education only in English; however, some participants reported previous schooling and/or working in an environment where Spanish was spoken. The average number of years spent in a school and/or working environment where Spanish was spoken was 2.2 years. Separate questions measured the frequency of participants’ exposure to English and Spanish in different contexts; those data are presented in Table 1.

Table 1.

Bilinguals’ self-reported frequency of current English and Spanish exposure by context.

Context	English exposure mean (SD)	Spanish exposure mean (SD)
Interacting with friends	9.58 (.81)	4.35 (2.48)
Interacting with family	6.12 (2.8)	8.95 (1.73)
Watching TV	8.78 (1.87)	4.28 (2.4)
Listening to music/radio	8.43 (1.79)	5.22 (2.8)
Reading	9.18 (1.14)	3.58 (2.1)

Note: All scores are based on a 0 to 10 scale with 0 indicating “never” and 10 indicating “always”.

SD = standard deviation

The English monolinguals were born in the US, reported no foreign language use at home, and reported less than 5% exposure to any language other than English. All were college students. The Spanish monolinguals were born in South America, 24 in Chile and 1 in Colombia (the individual born in Colombia permanently moved to Chile at age nine). All resided in Santiago, Chile or Constitución, Chile, reported no foreign language use at home, and reported less than 5% exposure to any language other than Spanish. Twenty-two were college students or had a vocational degree, and three were college graduates. Characteristics of the samples are further described in Table 2. An additional two Spanish monolinguals and four English monolinguals were tested but excluded from analysis to create English monolingual and Spanish monolingual groups with comparable levels of performance on the Woodcock-Muñoz in English and Spanish, respectively. Data for all participants, including those excluded from analyses, are publicly available at https://osf.io/3wjue/.

Table 2.

Demographic information for Spanish–English bilinguals, English monolinguals, and Spanish monolingual participants.

	Bilinguals n = 65	English monolinguals n = 25	Spanish monolinguals n = 25
Age (mean, SD)	19.5 (2.1)	19.1 (1.5)	21.2 (2.8)
Female (%)	67.7	68	56
Years of education (mean, SD)	13.7 (1.7)	13.2 (.71)	14.8 (2.1)
Concurrent English exposure % (mean, SD)	62.3 (12.1)	99.6 (1.4)	3.6 (3.4)
Age began acquiring English (mean, SD)	2.7 (2.1)
Age began acquiring Spanish (mean, SD)	1.8 (1.8)
Participants with two Spanish-dominant parents (%)	76.9

SD = standard deviation

The bilingual and English monolingual participants were recruited in the South Florida community via fliers posted at local universities, the psychology department subject pool, and snowball sampling. Spanish monolinguals were recruited in Santiago, Chile and Constitución, Chile via fliers posted at local universities and snowball sampling.

Measures and instruments

Language experience and use.

All bilingual participants completed the LEAP-Q, a self-report instrument widely used in studies of bilinguals with established reliability and validity (Marian et al., 2007; Prior & Gollan, 2013). Participants reported on their concurrent language exposure, frequency of language use in different contexts, age of first exposure to each language, and history of exposure to each language at work and school.

Receptive vocabulary.

The Peabody Picture Vocabulary Test (PPVT) (Dunn & Dunn, 2007) and its Spanish counterpart, Test de Vocabulario en Imágenes Peabody (TVIP) (Dunn et al., 1986), provided measures of receptive vocabulary in each language. The PPVT was normed on individuals from the US (Dunn & Dunn, 2007). The TVIP was normed in Mexico and Puerto Rico and had a median internal reliability coefficient of .93 (Dunn et al., 1986).

Expressive vocabulary.

Expressive vocabulary skills were assessed using the Picture Vocabulary subtest from the Woodcock-Muñoz Language Survey-Revised (WMLS-R); (Woodcock et al., 2005) in which participants are asked to provide labels for pictures shown by the examiner. All subtests from the WMLS-R have an English and Spanish version that were co-normed and equated to produce a Rasch-based W score. These W scores are developmentally standardized, such that a value of 500 indicates the average score for a 10-year-old. They provide interval scale measures that are comparable across languages. Extensive research has documented strong internal consistency reliability and content validity (Alvarado et al., 2005; Woodcock et al., 2005).

Oral language comprehension.

Oral language skills were assessed using the Understanding Directions subtest from the WMLS-R. Participants listened to a series of oral commands and answered by pointing to images on a flipbook, often in a required sequence.

Story recall.

Story recall skills were assessed by the Story Recall subtest from WMLS-R. Participants listened to a recording of short stories and were immediately asked to orally recall the passage. As the test continued, the stories increased in complexity and length.

Reading comprehension.

Reading comprehension was assessed using the Reading Comprehension subtest from the WMLS-R. Participants read a short paragraph with a missing word and were asked to verbally produce the missing word.

Grammar.

Proficiency in English and Spanish grammar was assessed with different and non-comparable instruments. Each of these instruments was developed as part of different research projects aimed at examining the grammatical proficiency of adult second language learners and heritage speakers. The English task required the participants to judge the grammaticality of 27 grammatical and 27 ungrammatical sentences presented in random order (Johnson & Newport, 1989). These stimuli have been used to measure the grammatical skill of highly proficient L2 speakers (DeKeyser, 2000; Johnson & Newport, 1989). The task yielded two scores: mean levels of judged acceptability for grammatical sentences and mean levels of judged acceptability for ungrammatical sentences.

The Spanish grammaticality judgment task developed by Montrul and Bowles (2009) consisted of 25 grammatical and 25 ungrammatical Spanish sentences presented in random order, and participants rated their acceptability on a scale of one to five.

Speed of lexical retrieval.

In a picture-naming task participants were presented with line drawings of common objects and asked to name those objects as quickly and accurately as possible (Jia et al., 2009). Each image was presented on a computer screen for 2000 ms or until a label was provided. A 1000 ms fixation cross was presented in the center of the screen between each trial. Stimuli were presented and reaction times were recorded using e-prime and a serial box with a voice-activated timer. Each version of the test began with five practice trials, followed by 25 trials. “Cane,” “hand,” and “key” are examples of images from the task. Stimuli for this task were matched in lexical frequency across English and Spanish, and all images had at least an 80% agreement among English and Spanish speakers (see Jia et al., 2009, for a detailed description of task design). Therefore, performance on this task allows for direct comparison across languages.

Degree of bilingualism.

A degree of bilingualism score was calculated for each bilingual adult following the procedure used in Gollan et al. (2011). We created a single degree of bilingualism score by first calculating each participant’s mean W score for each language across the four Woodcock-Muñoz subtests and then dividing the lower mean score by the higher mean score. This yields a measure of degree of bilingualism independent of absolute level of proficiency that ranges from zero (monolingual) to one (perfectly balanced bilingual).

Procedure

The monolingual participants were tested in one 70-minute session. The bilingual participants were assessed in two sessions (one in English, one in Spanish); the order of language administration was counterbalanced across participants. Each assessment session lasted approximately 70 minutes with a minimum 15-minute break between sessions. Data collectors were trained research assistants who were fully proficient in each language. They exclusively spoke in English during the English session and in Spanish during the Spanish session. Participants completed the LEAP-Q (Marian et al., 2007) at the beginning of the first session, regardless of the language being evaluated.

Results

Bilinguals’ English and Spanish proficiencies

To ask whether the bilinguals’ levels of English and Spanish proficiency differed, paired sample t-tests were conducted on the speed of lexical retrieval task and on the four measures from the Woodcock-Muñoz battery: expressive vocabulary, understanding directions, story recall, and reading comprehension. Means, SDs, and t-test comparisons are presented in Table 3. On all tasks, English scores were significantly higher than Spanish scores. The size differences (i.e. Cohen’s d) are plotted in Figure 1.

Table 3.

Means, standard deviations, and t-test comparisons of bilinguals’ English and Spanish skill on Woodcock-Muñoz subtests and speed of lexical retrieval.

Measures	English score mean (SD)	Spanish score mean (SD)	t	p (2-tailed)
Picture vocabulary (N = 64)	522.94 (14.34)	487.19 (18.51)	12.2	<.001
Understanding directions (N = 64)	512.14 (13.39)	504.91 (8.81)	4.15	<.001
Story recall (N = 64)	502.61 (3.97)	496.41 (6.7l)	8.54	<.001
Passage comprehension (N = 65)	522.18 (9.0l)	499.51 (18.89)	9.52	<.001
Speed of lexical retrieval (N = 63)	800.45 (159.19)	1140.68 (192.8)	−13.24	<.001

Note: All measures, excluding speed of lexical retrieval, are from the WMLS-R (2005).

Figure 1.

Effect sizes and 95% confidence intervals for comparisons of bilinguals’ English and Spanish skills (expressed in Cohen’s d). Speed of lexical retrieval was measured in milliseconds where higher scores indicate slower performance. In all other measures higher scores indicate better performance. *** p < .001 by paired sample t-test comparison.

Correlates of bilinguals’ degree of bilingualism

Fifty-nine of the 63 bilingual participants who had complete data on the Woodcock-Muñoz battery had a higher average W score in English than Spanish. The degree of bilingualism measure was calculated by dividing the participants’ lower language score by their higher language score. Thus, for almost all the bilingual participants, their degree of bilingualism was the degree to which their Spanish skills approached the level of their English skills. To investigate the factors associated with degree of bilingualism we examined only the 59 participants who were English dominant. We excluded the four Spanish-dominant participants in these analyses on the logic that there were not enough Spanish-dominant bilinguals to yield meaningful information about influences on their degree of Spanish dominance.

A hierarchical regression predicting degree of bilingualism was conducted in which Step 1 included age, gender, and years of education as control variables; Step 2 included a measure of Spanish immersion experience that was a mean composite score of time spent working in Spanish and time spent in a Spanish-speaking country, and mother’s dominant language; and Step 3 included self-reported percentage of time currently exposed to Spanish. The rationale for the order of entry was to first control for demographic characteristics, then to include measures that were related to language exposure developmentally, and finally to include concurrent language exposure. Results of the hierarchical regression are presented in Table 4. In the final model, only the measure of current exposure to and use of Spanish was a significant source of variance in degree of bilingualism, accounting for 22% of the variance.

Table 4.

Hierarchical multiple regression predicting degree of bilingualism from Spanish immersion, mother’s dominant language and current Spanish exposure, controlling for age, gender, and years of education (N = 59).

Predictors	ΔR2	β
Step 1	.079
Age		.126
Gender		.277*
Years of education		.036
Step 2	.013
Age		.111
Gender		.300*
Years of education		.067
Spanish immersion		.068
Mother’s dominant language		.131
Step 3	.219***
Age		.218
Gender		.179
Years of education		−.029
Spanish immersion		.118
Mother’s dominant language		−.022
Current Spanish exposure		.516***
Total R2	.312
N	59

Note: Females were coded as 2 and males as 1. Spanish immersion was a composite measure of time spent working in Spanish and living in a Spanish-speaking country. Mother’s dominant language was coded as 1 for English and 2 for Spanish.

^*p < .05

^***p <.001, β = standardized beta coefficient.

Bilinguals’ single language skills compared to those of monolinguals

Mean scores on measures of English proficiency for the English monolingual and Spanish–English bilingual participants are presented in Table 5, along with the results of the statistical comparison of the two groups. The distributions of scores within each group for each measure are plotted in Appendix A. Statistical comparisons were accomplished with independent sample t-tests with the alpha value adjusted to .006 to account for multiple comparisons. The bilinguals’ English skills were not significantly different from the monolinguals’ English skills on the measures of receptive vocabulary, judgments of grammatical sentences, judgments of ungrammatical sentences, expressive vocabulary, oral language comprehension, story recall, and reading comprehension. There was a significant difference on the measure of speed of lexical retrieval; the bilingual participants were significantly slower at picture naming. The size of the differences between the bilinguals’ and English monolinguals’ language scores, expressed in Cohen’s d, are plotted in Figure 2.

Table 5.

English measures: Means, 95% confidence intervals, and t-test comparisons of measures of English skill in Spanish–English bilinguals and English monolinguals.

Measures	English monolinguals mean (95% CI)	Spanish–English bilinguals mean (95% CI)	t	p (2-tailed)
Receptive vocabulary	193.08 (188.24–197.92)	193.91 (189.89–197.93)	−0.23	.817
Picture vocabulary	526.48 (521.76–531.20)	523.06 (519.53–526.60)	1.1	.287
Understanding directions	514.2 (509.67–518.73)	512.14 (508.8–515.49)	0.68	.496
Story recall	502.3 (500.36–504.28)	502.6 (501.62–503.60)	−0.29	.771
Passage comprehension	525.6 (522.80–528.40)	522.2 (519.95–524.42)	1.7	.090
Judgments of ungrammatical sentences	4.12 (3.88–4.36)	4.29 (4.15–4.40)	1.45	.150
Judgments of grammatical sentences	4.72 (4.64–4.81)	4.61 (4.55–4.67)	2.0	.049
Speed of lexical retrieval (ms)	710.3 (673.83–746.78)	799.99 (760.53–839.44)	−3.38	.001

Note: Receptive vocabulary = PPVT (2007); picture vocabulary, understanding directions, story recall and passage comprehension = WMLS-R (2005); judgments of ungrammatical and grammatical sentences were selected on a Likert scale ranging from 1 to 5. Speed of lexical retrieval was measured in milliseconds with lower scores indicating faster retrieval; raw scores were used on all measures except those from the Woodcock-Muñoz language battery, which were converted into W scores.

Figure 2.

Effect sizes and 95% confidence intervals for bilingual–monolingual comparisons of English skill (expressed in Cohen’s d). Speed of lexical retrieval was measured in milliseconds where higher scores indicate slower performance. In all other measures higher scores indicate better performance. ** p < .01 by t-test comparison.

Mean scores on measures of Spanish proficiency for the Spanish monolingual and Spanish–English bilingual participants are presented in Table 6, along with the results of independent sample t-tests. The distributions of scores within each group are plotted in Appendix B. The bilinguals’ Spanish skills were significantly lower than the Spanish monolinguals’ on the measures of receptive vocabulary, judgments of ungrammatical sentences, expressive vocabulary, oral language comprehension, story recall, reading comprehension, and speed of lexical retrieval. There was no significant difference in judgments of grammatical sentences. The size of the differences between the bilinguals’ and Spanish monolinguals’ language scores, expressed in Cohen’s d, are plotted in Figure 3. Standard scores from the Woodcock-Muñoz language battery are presented for descriptive purposes in Table 7.

Table 6.

Spanish measures: Means, 95% confidence intervals, and t-test comparisons of measures of Spanish skill in Spanish-English bilinguals and Spanish monolinguals.

Measures	Spanish monolinguals mean (95% CI)	Spanish–English bilinguals mean (95% CI)	t	p (2-tailed)
Receptive vocabulary	118.36 (116.59–120.13)	96.55 (92.65–100.46)	10.2	< .001
Picture vocabulary	532.88 (527.69–538.07)	487.19 (482.56–491.81)	13.4	< .001
Understanding directions	514.96 (512.24–517.68)	504.85 (502.68–507.02)	5.3	< .001
Story recall	502.64 (500.87–504.4l)	496.14 (494.73–498.08)	5.2	< .001
Passage comprehension	529.48 (526.45–532.51)	499.51 (494.83–504.19)	10.8	< .001
Judgments of ungrammatical sentences	4.03 (3.78–4.29)	3.35 (3.15–3.55)	3.8	< .001
Judgments of grammatical sentences	3.74 (3.59–3.90)	3.56 (3.44–3.67)	1.76	.081
Speed of lexical retrieval (ms)	683.56 (642.63–724.49)	1140.68 (1092.13–1189.23)	−11.3	< .001

Note: Receptive vocabulary = TVIP (1986); picture vocabulary, understanding directions, story recall and passage comprehension = WMLS-R (2005); judgments of ungrammatical and grammatical sentences were selected on a Likert scale ranging from 1 to 5. Speed of lexical retrieval was measured in milliseconds with lower scores indicating faster retrieval; raw scores were used on all measures except those from the Woodcock-Muñoz language battery, which were converted into W scores.

Figure 3.

Effect sizes and 95% confidence intervals for bilingual–monolingual comparisons of Spanish skill (expressed in Cohen’s d). Speed of lexical retrieval was measured in milliseconds where higher scores indicate slower performance. In all other measures higher scores indicate better performance. *** p < .001 by t-test comparison.

Table 7.

Mean standard scores and standard deviations for monolingual and bilingual participants on subtests from the Woodcock-Muñoz language battery.

	Monolingual		Bilingual
Measures	English	Spanish	English	Spanish
Picture vocabulary	94.4 (8.6)	97.56 (8.4)	91.60 (9.7)	67.23 (12.5)
Understanding directions	101.48 (11.2)	105.52 (9.9)	100.39 (13.0)	88.4 (14.8)
Story recall	95.88 (9.7)	97.36 (9.3)	94.3 (10.7)	78.75 (17.0)
Passage comprehension	102.68 (8.3)	110.04 (14.4)	100.62 (11.1)	78.43 (17.5)

Note: Standard scores were designed to have a mean of 100 and a standard deviation of 15.

Discussion

The purpose of the present study was to describe the dual language proficiencies of young adult, US-born, Spanish–English bilinguals who were raised with early exposure to English as the majority language and to Spanish as their heritage, minority language. Previous literature on the positive effects of early exposure to language on ultimate language skill (Abrahamsson & Hyltenstam, 2009), on the incomplete acquisition of language skill in heritage language speakers (Montrul, 2016), and on the lags in single language growth experienced by simultaneous dual language learners (Hoff, 2018) yields inconsistent predictions regarding the dual language skills such adults might display. The current study comprehensively examined the English and Spanish skills of early bilinguals and compared their single language skills with those of monolinguals.

The results yielded the following clear findings: (a) as young bilingual adults, second-generation immigrants from Spanish-speaking families in the US, despite early exposure to both English and Spanish, are overwhelmingly English dominant; (b) on multiple measures of language skill these bilinguals are not different from native, English monolinguals in their English skills, but they have consistently and significantly lower Spanish skills than native, Spanish monolinguals; (c) the sole English task that robustly distinguished native monolinguals from bilinguals was speed of lexical retrieval; (d) as result of their reliably strong English skills and weaker Spanish skills, the degree of balanced bilingualism in this population is essentially a function of level of Spanish skill; and (e) the experience most strongly associated with Spanish skill and thus balanced bilingualism is not early exposure but concurrent exposure and use.

These findings have theoretical and practical implications. With respect to theory, the results provide new data on the dependence of language acquisition on not only early oral experience but also continued and school-based experience in later childhood. On a more practical level, a comprehensive description of the dual language skills of second-generation immigrants in adulthood addresses two questions that parents and educators frequently voice. One is the question of to what degree early simultaneous exposure to two languages provides children with the opportunity to become bilingual—something that many parents hope for and that evidence suggests may confer multiple cognitive and professional benefits, although the cognitive benefits associated with bilingualism are unclear and controversial (Kroll & Bialystok, 2013; Paap & Greenberg, 2013). A second is the question of whether exposure to and acquisition of a heritage language interferes with the achievement of skill in the societal language.

With respect to the question of whether heritage language acquisition interferes with majority language skills, the results of the present study are unequivocal: the bilingual adults in this study were not detectably different from monolingual English speakers on a range of language and literacy measures that other research has related to academic achievement (Alvardo et al., 2005). This finding provides evidence that the early majority language lag relative to monolinguals, which has been well documented in the developmental literature (Hoff, 2018), is no longer apparent in adulthood on a wide array of measures that tap functional language skills and skills related to academic success. The only evidence in the bilinguals’ performance on English tasks that they were not monolingual was in their speed of lexical retrieval.

With respect to whether early exposure to two languages results in bilingualism, the data provide a qualified affirmative answer. The qualification is of two sorts: first, these participants were bilingual in the sense that they used both English and Spanish on a regular basis. They were not balanced bilinguals, however. Second, the present study was not a prospective investigation of the outcome of early dual language exposure. We do not have a measure to indicate whether the bilingualism these participants display is a common or rare outcome of their early experience. Other research suggests 75% of children with early dual language exposure are functionally bilingual in adulthood (De Houwer, 2007; Eilers et al., 2006).

The present findings regarding the bilinguals’ English skills are consistent with other findings in the literature. Previous research similarly suggests that acquisition of the majority language is unproblematic for dual language learners (Gathercole & Thomas, 2009) and that dual language exposure does not hinder long-term majority language proficiency (Bylund et al., 2012). The sole difference observed with the present data, in speed of lexical access, is also consistent with findings from studies of adult bilinguals (Bialystok, 2009), although the underlying cause of the effect is not clear. Two, not mutually exclusive, explanations for this group-related difference can be found in the literature. First, the weaker link hypothesis suggests that due to a reduced frequency in use, bilinguals have weaker connections between words and concepts (Gollan et al., 2008) and therefore take longer to retrieve labels. Second, it has been suggested that both languages are always activated in the bilingual brain, and consequently the two lexicons compete for selection, slowing down retrieval time (Ivanova & Costa, 2008; Sunderman & Kroll, 2006). Selecting one language while inhibiting the other is further described by the activation threshold hypothesis, which proposes that higher levels of exposure to a language will result in less inhibition of the non-target language when retrieving words (Paradis, 2004).

The present findings regarding the bilinguals’ Spanish skill add to the growing body of research that finds that heritage language skills are vulnerable to incomplete acquisition despite early exposure to Spanish (Montrul, 2002, 2016). One potential explanation for the bilinguals’ relatively strong English and weaker Spanish skills is that they were educated in English. Furthermore, although they used both languages at least weekly, they used Spanish less frequently, and in a restricted range of contexts—at home and interacting with friends, according to their own report. Consequently, the communicative demands on Spanish for the bilinguals were different from and more limited than the communicative demands placed on Spanish for the monolinguals who exclusively used Spanish in their daily lives.

Conclusion

A classic paper in the field of bilingualism has in its title an admonishment to reject a simple view of bilingualism, stating “the bilingual is not two monolinguals in one person” (Grosjean, 1989, p. 4). The results of the current findings reported here are consistent with that admonishment, providing clear evidence that early dual language exposure does not guarantee the achievement of two monolingual-like levels of proficiency in adulthood. The current findings reveal that by young adulthood early bilinguals achieve language skills comparable to monolinguals in English for most functional purposes, but fall far short of monolingual-like competence in Spanish.

The present findings of substantially monolingual-like English skills and limited Spanish skills in the adult bilingual suggests a very close dependency between skill and concurrent levels and contexts of use. The bilinguals in the current study used Spanish primarily to interact with family, and their Spanish skills may be entirely adequate for the communicative purposes to which they are put. In contrast, these bilinguals went to school in English, were currently attending college in English and, even apart from literacy, they likely used English in a wider range of contexts and for a wider variety of communicative purposes. The difference between the English and Spanish skills observed among Spanish–English bilingual adults in the US in this study is entirely consistent with Grosjean’s (1989) explanation of why bilinguals are not two monolinguals in one person: the bilingual, Grosjean states,

has developed competencies to the extent required by his or her needs and those of the environment. The bilingual uses the two languages—separately or together—for different purposes, in different domains of life, with different people. Because the needs and uses of two languages are usually quite different, the bilingual is rarely equally or completely fluent in the two languages.

(Grosjean, 1989, p. 6)

Implications and future directions

There is an important implication in the present findings with respect to how language science experts communicate with parents who are raising bilingual children. It is that experts should not lead parents to expect that early exposure to two languages is sufficient to achieve adult levels of monolingual proficiency in both languages. Well-intentioned authority figures, who are rightfully aware of the importance of early exposure for ultimate language skill, have claimed that children with early and continual bilingual exposure will become native-like speakers in each of their languages. The current study provides strong evidence against this claim. Rather, it should be made clear that early exposure is necessary for monolingual-like proficiency, but it is not sufficient. If the end goal is monolingual-like language skills in two languages, then bilingual education is likely necessary. However, more research is required to shed light on factors related to the attainment of monolingual-like skills in two languages.

Future research should examine under what conditions individuals are able to acquire multiple languages with native-like competence. This line of research could examine the heritage language proficiency of young adults who received bilingual education and continue to take heritage language classes into adulthood. Continued formal education in the heritage language would likely result in stronger heritage language proficiency and more balanced dual language skills. A second line of research could examine bilinguals’ dual language skills in a society where their two languages are held to similar levels of esteem. One such instance could be the English and French skills of early bilinguals from the Quebec province. In conclusion, the results of the present study are not necessarily suggestive of the language proficiencies of early bilinguals under all conditions; however, they do provide a clear and comprehensive prediction of the dual language skills of second generation immigrants who grow up in a majority–minority language context.

Biographies

David Giguere is an Assistant Professor of Psychology at California State University, Sacramento. His research examines individual differences in language and literacy skills among individuals with early bilingual exposure.

Erika Hoff is Professor of Psychology at Florida Atlantic University. She is Principal Investigator of an NICHD-funded longitudinal study of Spanish-English bilingual children in South Florida. She is the author of numerous articles and chapters and the editor of multiple books on early language development, including Research Methods in Child Language and, with Peggy McCardle, Childhood Bilingualism: Research on Infancy through School Age.

Appendix

Appendix A.

English skill: Boxplot figures for bilingual and monolingual participants.

Appendix B.

Spanish skill: Boxplot figures for bilingual and monolingual participants.