Relationships Between Home and School Language Exposure and Changes in Bilingual Oral Language Performance From Kindergarten to First Grade

Abstract

Purpose:

With over one third of students in the United States exposed to languages other than English at home, there is a growing need to examine the role of bilingual exposure across settings on bilingual language performance. This study explored the longitudinal relationship between language exposure at home and school, and changes in bilingual oral language performance from kindergarten to first grade.

Method:

One hundred sixty-four Spanish–English participants who had different levels of bilingual exposure at home and school were assessed on bilingual oral language performance using a standardized language assessment in kindergarten and first grade. Caregivers and teachers reported bilingual exposure at home and at school, respectively, for both time points.

Results:

The results indicated that Spanish home exposure at kindergarten uniquely predicted Spanish semantics in first grade, while Spanish school exposure at kindergarten uniquely predicted Spanish morphosyntax in first grade, after controlling for kindergarten performance. No differential contributions were observed for English exposure and performance. Children with high Spanish home exposure (> 60%) had significantly higher gains in standard scores for Spanish semantics and overall Spanish performance compared to the low Spanish exposure group, but there was no significant effect on changes in English standard scores across groups.

Conclusions:

The current study suggests that language exposure at home and school had different contributions to bilingual home language development. This supports the importance of home and school Spanish language exposure in promoting heritage language development without compromising English.

More than one third of the students in the United States have exposure to one or more languages other than English at home (United States Census Bureau, 2019; Vespa et al., 2020), emphasizing an increasing need for understanding the relationship between language exposure across settings and direct measures of language performance. Language use in home and school contexts may have distinct influences on children's language experiences. Research has shown that individual differences in bilingual children's language experiences and exposure are related to differential learning outcomes, in both oral language and literacy skills (Beauchamp & MacLeod, 2022; Thordardottir, 2011). In this study, we focused on Spanish–English bilingual children who had different levels of bilingual exposure at school, and specifically examined their language development from kindergarten to first grade. This time frame was selected because this period marks the beginning of formal instruction, resulting in a major shift of language experiences for children. As children progress through elementary school, they encounter increasingly complex academic language, particularly in morphosyntax and literacy. Maintaining Spanish language and literacy remains important, as strong oral proficiency in the home language enhances reading outcomes. Instead of adding to the literature that explores the relationship between current exposure and current performance, we aim to examine whether language exposure at kindergarten predicts performance 1 year later. Children are exposed to varied input across contexts, such as a kitchen scenario at home and a science experiment at school, each involving different vocabulary and linguistic structures. Thus, we further examined how exposure across these contexts (home vs. school) differentially contribute to children's performance in the domains of morphosyntax and semantics.

Language Exposure and Language Performance in Bilingual Children

Past studies have shown that language exposure is related to bilingual children's language development, including vocabulary (Bowers & Vasilyeva, 2011; De Houwer, 1996; Pearson et al., 1997; Umbel et al., 1992), morphosyntax (Lam, Wang, et al., 2025; Paradis, 2010), and communication discourse (Hoff, 2006; Lam, Leachman, et al., 2025). Language exposure is associated with language development in bilinguals, which can be measured by language input and by language output (Bedore et al., 2016). Language input refers to what children hear from external sources. On the other hand, language output refers to what children say to others. Capturing language exposure (input and output) at different time points represents different aspects of children's language experiences. Current exposure is the measure of language input and output that the child is experiencing at the present time or within a recent period of time. Measurements of exposure are linked to different time periods as reference points. For example, Crespo et al. (2019) asked parents to fill out a survey detailing the hours their children were awake and exposed to English, Spanish, or both languages during their typical daily activities on both weekdays and weekends. This information helped researchers interpret the effects of dual exposure on children's language performance.

In addition to current exposure, researchers have also examined bilingual children's previous exposure, which reflects how long children have heard and used each language over a particular time period or cumulatively over a number of years (e.g., from birth to the current age; Unsworth, 2013). To measure cumulative exposure, researchers may ask parents to detail the amount of each language the child uses and hears during each year of the child's life, including in day care or school settings. By gathering this information, researchers are able to calculate and sum the total amount of exposure to a language use measure in years over time (Unsworth, 2013). Age of first English exposure, which is also referred to as age of acquisition, is an indication of past exposure that has been examined in research (e.g., Jia et al., 2006; Kan & Kohnert, 2008). Age of first English exposure gives information about length of exposure to the second language (i.e., English in this study). Bedore et al. (2012) found that both age of first English exposure and current exposure significantly contributed to the variance in children's language outcomes.

In this study, we focus on the role of both age of first English exposure and current exposure in kindergarten on the language performance 1 year later (first grade). In kindergarten, children traditionally receive early education that is less formal and structured. However, upon entering first grade, they receive formal, planned instruction from teachers, exposing them to a more structured academic language environment. As children enter the school system, their current use and exposure change as a function of school programming, so it is important to assess changes in exposure and performance during the initial stages of schooling (from kindergarten to first grade). Understanding the relationship between kindergarten exposure and first grade performance could also inform school programming to support later language performance. In addition, from an educational and clinical perspective, we can group bilingual students based on experience: According to Bedore et al. (2018), children are considered to have high Spanish exposure if they used less than 39% English, mixed exposure if they used 40%–59% English, and high English exposure if they used over 60% English. Prior studies showed significant differences in language performance in semantics and morphosyntax across language exposure groups (Bedore et al., 2018; Jasso et al., 2020; Pérez et al., 2010). In the current study, we were interested in how children's performance changed relative to their language experience grouping at baseline. The Unified Competition Model (MacWhinney et al., 2005) posits that in the development of two languages, cross-language transfer can either positively or negatively support the learning of the other language. The authors explain language transfer in bilingual development by emphasizing the role of competing linguistic cues across languages. According to this model, bilinguals rely on the strength and reliability of cues in each language when processing and producing linguistic structures, which can lead to transfer effects in both semantics and morphosyntax. When a particular cue (e.g., word order, case marking, or verb agreement) is strong in one language but weaker in the other, bilinguals may apply structures from the dominant language system to the weaker one. This transfer can be observed in areas such as semantics, where meanings from one language shape interpretations in the other, and in morphosyntax, where bilinguals may apply familiar grammatical patterns from one language to the other. This model highlights how bilinguals navigate and integrate linguistic systems based on input frequency, cue reliability, and competition between their two languages, and it is relevant to the observations of bilingual development in this study.

Language Learning Context

Bilingual children are primarily exposed to languages in home and school settings. Home language exposure reflects hearing and producing language at home, while school language exposure refers to hearing and producing language during the school day. The vocabulary used in home settings is expected to be more conversational and colloquial and more likely to be here-and-now, reflecting everyday language use. For example, during dinner children often observe and engage in narrative discourse as family members recount their activities during the day (Hoff, 2006). In addition, home exposure might include community activities and multimedia exposure. In contrast, the vocabulary that children hear and use in school settings tends to be more decontextualized and formal (Hoff, 2006). Additionally, school language exposure is more systematic with reference to the curriculum. It usually involves explicit instruction from teachers that contains complex words and sentences that are tied to the curriculum. Huttenlocher et al. (2002) found that teachers tend to use more syntactically complex language, which is associated with syntactic growth in children over the school year. Schleppegrell (2012) emphasized that academic language in classroom settings is characterized by its explicit and structured nature, including features such as dense noun phrases, logical connectors, and complex clause structures. This suggests that, compared to home language exposure, which is typically more context-dependent and informal, classroom discourse is explicitly structured.

Being exposed to varied learning contexts results in different learning outcomes in terms of language development. Palermo et al. (2013) have found that exposure to English at home was uniquely and positively correlated with children's receptive and expressive English vocabulary. However, Snow (1990) found that exposure to English at school was a better predictor of word definition skills than home exposure among elementary school children from diverse home language backgrounds. Our study examines whether home and/or school language use contributes differentially to children's language performance in two domains, semantics and morphosyntax.

Use of Raw and Standard Scores

In order to evaluate whether the amount of language exposure predicts children's language performance, we focus on two types of scores. Raw scores on a standardized test reflect the child's knowledge of the items, while standard scores that are converted from raw scores reflect a child's performance in relation to same age peers (Foster et al., 2018). Changes in raw scores provide a measure of how much language skill the child has gained, which is a direct reflection of their language performance in real terms. In contrast, changes in standard scores indicate the extent to which the child's language skills are improving or declining relative to their same-age peers. Another advantage of using standard scores is that they account for age differences among participants in the same grade.

Research Questions

1. Do the age of first English exposure and bilingual language exposure in home and school contexts at kindergarten (Year 1) predict children's bilingual language performance in first grade (Year 2)?

   For this research question, we use raw scores for regression analysis and examine how language exposure at home and school bilingual children's actual performance in Year 2.

2. Does bilingual language exposure at home and school in kindergarten (Year 1) predict changes in children's bilingual semantic and morphosyntax standard scores from kindergarten to first grade?

   For this research question, we use standard scores to investigate how exposure is related to changes from kindergarten to first grade on bilingual semantics and morphosyntax standard scores.

Method

Participants

Data for the current study were drawn from an existing dataset of 164 unique participants in the Diagnostic Markers of Language Impairment project (DM; Peña et al., 2006). DM was a three-phase longitudinal study with language screening in Phase 1. The research team recruited participants from three different school districts in Utah and Texas, and they had different levels of bilingual exposure at school. Children who scored below the 30th percentile on two or more subtests on the Bilingual English–Spanish Assessment (BESA) in either Spanish or English, and who used English and Spanish at least 20% of the time per caregiver report were recruited to the next two phases. Participants were included in this study if they had completed both the Bilingual Input–Output Scale (BIOS) and bilingual language assessment in both Phases 2 and 3. Data from 164 participants who completed Phases 2 (Year 1; kindergarten) and 3 (Year 2; first grade) were used. Table 1 presents the participants' demographics. Ethics approval was obtained from the University of Texas at Austin and Utah State University (Project Number: 2005-09-0096). Informed written consent was obtained from the caregivers of all participating children prior to testing.

Table 1.

Demographics of the participants.

Participant characteristic	n	%	M	SD
Age in months at Year 1			69.3	4.7
Age in months at Year 2			81.9	4.7
Sex
Female	81	49.4
Male	83	50.6
Language ability status
Typically developing	144	87.8
Developmental language disorder	20	12.2
Free and reduced lunch	126	76.3
Nonverbal intelligence			97.9	11.6
Socioeconomic status			24.7	11.2
Age of first English exposure			2.1	1.7

Note. Socioeconomic status is calculated by the Hollingshead's Four-Factor Index (Hollingshead, 1975).

Language Ability Status

Three bilingual clinicians independently evaluated the language abilities of children by analyzing a variety of both formal and informal language data collected during the children's first-grade year. They were asked to provide a comprehensive overall rating for each language based on these assessments on a 6-point scale (see Wang et al., 2025). Based on these ratings, there were 20 participants with developmental language disorder (DLD) and 144 typically developing participants in the dataset. All 20 of these participants were retained in the subsequent analyses.

Measures

School and home language exposure. School language exposure was measured by the BIOS (Peña et al., 2018). Teachers report the amount of English and Spanish children heard (i.e., input) to and produced (i.e., output) in a typical school day every half-hour from 7 am to 5 pm. Teachers responded “English only,” “Spanish only,” or “English and Spanish” for each 30-min timeframe for input and output, respectively. For the response of “English and Spanish,” each language was credited for 15 min. Then, school language exposure was calculated to represent a school day based on the information from teachers. The formula for calculating Spanish exposure at school was

$$\text{Spanish Exposure}\mathrm{at}\text{School}=\frac{\left(\mathit{\text{hours of Spanish input}}\right)+\left(\mathit{\text{hours of Spanish output}}\right)}{2\times \left(\mathit{\text{total number of hours}}\mathit{at}\mathit{\text{school}}\right)}\times 100\%$$ (1)

The Spanish school exposure was the inverse of English school exposure.

Home language exposure was also measured by BIOS (Peña et al., 2018). Parents or guardians reported the amount of English and Spanish their children were exposed (i.e., input) to and used (i.e., output) on a typical weekday and weekend outside of schools, hour by hour from 7 am to 10 pm. Parents or guardians could respond “English only,” “Spanish only,” or “English and Spanish” for each timeframe and for input and output, respectively. The formula for calculating Spanish exposure at home was

$$\text{Spanish Exposure}\mathrm{at}\text{Home}=\left\{\frac{\left[\left(\mathit{\text{hours of Spanish input for weekday}}\right)+\left(\mathit{\text{hours of Spanish output for weekday}}\right)\right]}{2\times \left(\mathit{\text{total number of hours awake}}\mathit{on}\mathit{\text{weekday}}\right)}\times 5+\frac{[\left(\mathit{\text{hours of Spanish input for the weekend}}\right)+\left(\mathit{\text{hours of Spanish output for the weekend}}\right)]}{2\times \left(\mathit{\text{total number of awake}}\mathit{on}\mathit{\text{weekend}}\right)}\times 2\right\}÷7$$ (2)

The Spanish home exposure was the inverse of English home exposure.

Bilingual language performance. The BESA (Peña et al., 2018) was administered. Four different subtests of BESA were used for the current analysis: English semantics, Spanish semantics, English morphosyntax, and Spanish morphosyntax. Semantic subtests measured children's understanding of word meanings and relationships in both English and Spanish, including category, similarities and differences, analogy, linguistic characteristics, function, and characteristic properties. In both languages, the concepts were brought out through both receptive and expressive elements incorporated within three illustrated narratives. The 25 English and 25 Spanish items targeted parallel concepts. The reported coefficient alpha for Spanish and English semantic items were .88 and .86, respectively.

The morphosyntax subtest included cloze and sentence repetition tasks in English and Spanish, assessing a range of basic and complex morphosyntactic structures. Cloze items consisted of an elicitation frame with a sentence completion targeting different morphosyntactic structures. The second task, sentence repetition, required the child to repeat sentences verbatim. Sentence repetition was scored based on pre-identified target words or phrases. The reported coefficient alphas for the morphosyntax subtest were .96 in Spanish and .95 in English.

Age of first English exposure. This information was collected in the BIOS (Peña et al., 2018). Caregivers indicated what language(s) the child heard and spoke on a yearly basis from birth to identify year of first English exposure.

Socioeconomic status. Socioeconomic status was computed using Hollingshead's Four-Factor Index (Hollingshead, 1975). The total score represents both parent education and parent occupation as reported by caregivers.

Nonverbal intelligence. The Universal Nonverbal Intelligence Test (UNIT; Bracken & McCallum, 1998) was administered at kindergarten. The UNIT is administered nonverbally.

Procedure

Consent was obtained from caregivers before the testing session. Caregivers' interviews were conducted to collect socioeconomic status, percentage of English exposure (input and output), and age of first English exposure information, using the BIOS (Peña et al., 2018). Percentage of English exposure at home was computed using the formula in the above section. Teachers also completed the school language exposure interview from which percentage of English exposure at school was computed. Nonverbal IQ task and BESA were administered by American Speech-Language-Hearing Association–certified speech-language pathologists, master's students enrolled in speech-language pathology programs, or Bachelor of Arts–level trained research assistants from related fields (e.g., psychology or education). All examiners were proficient in Spanish and English, and the testing was conducted in both languages. The examiner prompted the students to answer in the target language if the student responded in a nontarget language. The nonverbal IQ task and BESA were administered according to the manual (Bracken & McCallum, 1998; Peña et al., 2018).

Data Analysis

School language exposure and home language exposure were computed based on the formula above. For bilingual language performance, raw scores of the four subtests were calculated, and Spanish and English standard scores were obtained using the manual of BESA (Peña et al., 2018).

For the first research question on the role of language exposure at home and school contributing to bilingual language performance in Year 2, correlation analysis was first performed. Four sets of stepwise regression models with two alternative orders were performed. The order of stepwise regression models was determined by prior literature. Demographic variables (age, sex, and socioeconomic status), language ability status, and nonverbal intelligence were entered first to examine the effect of individual differences. Initial performance was entered as the second step to account for difference in Year 1 (Dalecki & Willits, 1991). Age of first English exposure was entered in the third step to account for the cumulative language exposure. Separate alternative orders were applied for the fourth and fifth steps. To examine the role of language exposure at home, school language exposure at kindergarten was entered as the fourth step, while home language exposure at kindergarten was entered as the fifth step. To examine the role of language exposure at school, home language exposure at kindergarten was entered as the fourth step, while school language exposure at kindergarten was entered as the fifth step. Dependent variables were the raw scores on English semantics, English morphosyntax, Spanish semantics, and Spanish morphosyntax at Year 2.

For the second research question on the relationship between standard score change and language exposure at home and school, two sets of the same statistical procedures were performed. For home language exposure, three separate groups were created based on home language exposure: high Spanish home exposure group (Spanish exposure > 60%), mixed home exposure group (60% ≥ Spanish exposure ≥ 40%), and low Spanish home exposure group (Spanish exposure < 40%) with reference to prior studies (Jasso et al., 2020; Pérez et al., 2010). One-way multivariate analysis of variance (MANOVA) was performed to determine whether there was a difference between home exposure groups on standard score changes in Spanish semantics, Spanish morphosyntax, English semantics, and English morphosyntax. Post hoc analysis with a Bonferroni correction was performed if the main effect of the exposure group was significant. In addition, four different groups (i.e., English and Spanish gain, English gain and Spanish loss, Spanish gain and English loss, and English and Spanish loss) were created based on the standard score changes between Year 1 and Year 2. A chi-square test was conducted to examine the difference in group membership across different home language exposure groups. Afterwards, the same statistical procedure was used with reference to school language exposure.

Results

Table 2 presents the descriptive statistics and correlation of language exposure and bilingual language performance in Year 1 and Year 2. Age of first English exposure significantly correlated with bilingual language performance only in Year 1 (rs = −.18–.25, ps < .05), but not in Year 2 (rs = −.03–.15, ps > .05). Language exposure was significantly and moderately correlated with Spanish language performance in both time points (rs = .33–.53, ps < .001) and significantly but modestly correlated with English language performance in both time points (rs = .16–.52, ps < .05).

Table 2.

Correlation among bilingual exposure and bilingual oral language performance at both time points.

Variable	M	SD	1	2	3	4	5	6	7	8	9	10	11	12
1. Age of first English exposure	2.10	1.74
2. Y1 Spanish exposure at home	51.27	20.62	.32***	—
3. Y1 Spanish exposure at school	39.38	30.74	.08	.80***	—
4. Y1 Spanish semantics	21.73	7.79	.20*	.48***	.46***	—
5. Y1 Spanish morphosyntax	24.38	15.89	.25**	.40***	.33***	.76***	—
6. Y1 English semantics	21.67	8.55	−.23**	−.31***	−.25**	.15	−.01	—
7. Y1 English morphosyntax	40.21	18.71	−.18*	−.36***	−.33***	−.04	.00	.70***	—
8. Y2 Spanish exposure at home	48.07	18.51	.05	.71***	.71***	.49***	.44***	−.30***	−.44***	—
9. Y2 Spanish exposure at school	33.80	30.92	.08	.72***	.85***	.44***	.31***	−.19*	−.27**	.69***	—
10. Y2 Spanish semantics	31.49	6.64	−.03	.55***	−.50***	.72***	.71***	.01	−.12	.53***	.51***	—
11. Y2 Spanish morphosyntax	38.53	15.56	−.01	.50***	−.45***	.74***	.89***	−.12	−.12***	.51***	.43***	.79***	—
12. Y2 English semantics	28.58	8.66	.10	−.23**	−.16*	.17*	.13	.67***	.61***	.25**	−.16*	.18*	.09	—
13. Y2 English morphosyntax	47.13	18.75	.15	−.43***	−.40***	−.05	.06	.60***	.83***	.52***	−.36***	−.09	−.05	.66***

Note. Y1 = Year 1 (kindergarten), Y2 = Year 2 (first grade).

^*p < .05.

^**p < .01.

^***p < .001.

Role of Home and School Language Exposure in Bilingual Language Change

Table 3 presents the hierarchical regression for predicting Spanish semantic and morphosyntax at Year 2. For predicting Spanish semantics at Year 2, the effect of socioeconomic status, DLD, and nonverbal IQ were significant. Spanish semantics at Year 1 were entered in Step 2, and it uniquely predicted 33% of the variance. Age of first English exposure did not significantly predict Spanish semantics at Year 2, p = .06. Spanish exposure at school was entered in Step 4, and it uniquely predicted 3% of the variance, β = .05, SE = 0.02, t = 3.23, p = .002, 95% confidence interval (CI) [0.02, 0.09]. Spanish exposure at home was entered in Step 5, and it uniquely predicted 2% of the variance, β = .09, SE = 0.04, t = 2.28, p = .02, 95% CI [0.01, 0.16]. In the alternative model, Spanish exposure at home was entered in Step 4, and it uniquely predicted 5% of the variance, β = .10, SE = 0.03, t = 3.96, p < .001, 95% CI [0.05, 0.15]. However, when Spanish exposure at school was entered in Step 5, it did not significantly predict Spanish semantics at Year 2, p = .59. Thus, Spanish exposure at home had a unique contribution to the variance in Spanish semantics at Year 2 compared to Spanish exposure at school.

Table 3.

Regression models for predicting change in Spanish oral language.

Variable	(a) Semantics							(b) Morphosyntax
	Regression coefficients					Fit of model		Regression coefficients					Fit of Model
	B	β	SE	t	95% CI	R	R 2	B	β	SE	t	95% CI	R	R 2
Step 1						.50	.25						.56	.31
Age	0.03	.02	0.13	0.24	[−0.23, 0.29]			0.03	.01	0.26	0.13	[−0.49, 0.55]
Female	1.57	.09	1.23	1.27	[−0.87, 4.00]			3.48	.09	2.51	1.39	[−1.48, 8.45]
Socioeconomic status	−0.21	−.26	0.06	−3.70***	[−0.32, −0.10]			−0.54	−.32	0.11	−4.70***	[−0.76, −0.31]
Developmental language disorder	−9.78	−.37	1.92	−5.09***	[−13.59, −5.98]			−24.00	−.43	3.92	−6.13***	[−31.74, −16.27]
Nonverbal intelligence	0.11	.15	0.06	2.07*	[0.01, 0.22]			0.17	.10	0.11	1.48	[−0.06, 0.39]
Step 2
Spanish at Year 1	0.68	.67	0.06	10.76***	[0.56, 0.81]	.76	.58	0.86	.86	0.05	19.09***	[0.77, 0.95]	.89	.80
Step 3
Age of first English exposure	0.54	.11	0.28	1.93	[−0.01, 1.09]	.77	.59	0.85	.08	0.41	2.06*	[0.04, 1.66]	.90	.80
Step 4a
Spanish exposure at school	0.05	.19	0.02	3.23**	[0.02, 0.09]	.78	.61	0.10	.17	0.02	4.53***	[0.06, 0.15]	.91	.83
Step 5a
Spanish exposure at home	0.09	.21	0.04	2.28*	[0.01, 0.16]	.79	.63	0.03	.04	0.06	0.58	[-0.07, 0.21]	.91	.83
Step 4b
Spanish exposure at home	0.10	.24	0.03	3.96***	[0.05, 0.15]	.79	.63	0.14	.15	0.04	3.77***	[0.07, 0.21]	.91	.83
Step 5b
Spanish exposure at school	0.01	.05	0.02	0.54	[-0.04, 0.06]	.79	.63	0.09	.14	0.04	2.47*	[0.02, 0.16]	.91	.83

Note. CI = confidence interval.

^*p < .05.

^**p < .01.

^***p < .001.

For predicting Spanish morphosyntax, the effect of socioeconomic status and DLD was significant. Spanish morphosyntax at Year 1 was entered in Step 2 and it uniquely predicted 49% of the variance. Age of first English exposure was entered in Step 3 and uniquely predicted 1% of the variance. Spanish exposure at school was entered in Step 4 and it uniquely predicted 2% of the variance, β = .10, SE = 0.02, t = 4.53, p < .001, 95% CI [0.06, 0.15]. Spanish exposure at home was entered in Step 5, and it did not significantly predict Spanish morphosyntax at Year 2, p = .56. In the alternative model, Spanish exposure at home was entered in Step 4, and it uniquely predicted 2% of the variance, β = .14, SE = 0.04, t = 3.77, p < .001, 95% CI [0.07, 0.21]. In addition, Spanish exposure at school significantly predicted 1% of the variance, β = .09, SE = 0.04, t = 2.47, p = .02, 95% CI [0.02, 0.16]. Thus, Spanish exposure at school had a unique contribution to the variance in Spanish morphosyntax at Year 2 compared to Spanish exposure at home.

Table 4 presents the hierarchical regression for predicting English semantics and morphosyntax at Year 2. For predicting English semantics at Year 2, the effect of age and DLD were significant. English semantics in Year 1 was entered in Step 2, and it uniquely predicted 30% of the variance. English exposure at school and home did not significantly predict English semantics at Year 2, ps > .05. The alternative model showed the same result. Thus, language exposure at home and school did not significantly predict English semantics at Year 2, controlling demographics and other variables.

Table 4.

Regression models for predicting change in English oral language.

Variable	(a) Semantics							(b) Morphosyntax
	Regression coefficients					Fit of model		Regression coefficients					Fit of model
	B	β	SE	t	95% CI	R	R 2	B	β	SE	t	95% CI	R	R 2
Step 1						.48	.23						.46	.21
Age	0.26	.18	0.10	2.56*	[0.06, 0.45]			0.49	.15	0.24	2.06*	[0.02, 0.95]
Female	−0.65	−.05	0.95	−0.69	[−2.53, 1.23]			−1.30	−.04	2.25	−0.58	[−5.74, 3.15]
Socioeconomic status	0.07	.11	0.04	1.58	[−0.02, 0.15]			0.33	.23	0.10	3.18**	[0.12, 0.53]
Developmental language disorder	−7.13	−.36	1.48	−4.81***	[−10.06, −4.20]			−16.55	−.35	3.51	−4.72***	[−23.48, −9.62]
Nonverbal intelligence	0.06	.11	0.04	1.43	[−0.02, 0.14]			−0.08	−.06	0.10	−0.83	[−0.28, 0.12]
Step 2						.73	.53						.85	.72
English at Year 1	0.53	.64	0.05	9.90***	[0.43, 0.64]			0.81	.05	0.83	16.36***	[0.71, 0.91]
Step 3						.73	.53						.85	.72
Age of first English exposure	−0.14	−.04	0.22	−0.63	[−0.59, 0.30]			−0.14	−.02	0.41	−0.35	[−0.95, 0.66]
Step 4a						.73	.53						.86	.74
English exposure at school	0.01	.07	0.01	1.03	[−0.01, 0.04]			0.09	.18	0.03	3.63***	[0.04, 0.14]
Step 4b						.73	.53						.86	.74
English exposure at home	0.02	.07	0.03	0.70	[−0.04, 0.09]			0.09	.13	0.06	1.70	[−0.15, 0.20]
Step 5a
English exposure at home	0.03	.08	0.02	1.24	[-0.02, 0.07]	.73	.53	0.14	.19	0.04	3.83***	[0.07, 0.22]	.86	.74
Step 5b
English exposure at school	0.00	.02	0.02	0.16	[−0.04, 0.04]	.73	.53	0.05	.09	0.04	1.24	[−0.03, 0.12]	.86	.74

Note. CI = confidence interval.

^*p < .05.

^**p < .01.

^***p < .001.

For predicting English morphosyntax, the background variables, including age, sex, socioeconomic status, DLD, and nonverbal IQ, were entered in Step 1, and these variables predicted 21% of the variance. English morphosyntax at Year 2 was significantly predicted by age, socioeconomic status, and DLD. English morphosyntax at Year 1 was entered in Step 2, and it uniquely predicted 50% of the variance. Age of first English exposure was entered in Step 3, and it did not significantly predict English morphosyntax at Year 2, p = .73. English exposure at school was entered in Step 4, and it uniquely predicted 3% of the variance, β = .09, SE = 0.03, t = 3.63, p < .001, 95% CI [0.04, 0.14]. English exposure at home was entered in Step 5, and it did not significantly predict English morphosyntax at Year 2, p = .09. In the alternative model, English exposure at home was entered in Step 4, and it uniquely predicted 3% of the variance, β = .14, SE = 0.04, t = 3.83, p < .001, 95% CI [0.07, 0.22]. However, when English exposure at school was entered in Step 5, and it did not significantly predict English morphosyntax in Year 2, p = .22. Thus, English exposure at home and school had similar contributions to English morphosyntax in Year 2.

Language Exposure at Home and School and Standard Score Changes on Bilingual Oral Language Performance

Home Language Exposure Groups

Participants were grouped with reference to their home Spanish exposure. There were 64 participants in the high Spanish home exposure group (Spanish exposure > 60%), 41 participants in the mixed home exposure group (60% ≥ Spanish exposure ≥ 40%), and 59 participants in the low Spanish home exposure group (Spanish exposure < 40%). Table 5 presents the standard scores for both years and changes in Spanish semantics, Spanish morphosyntax, English semantics, and English morphosyntax across home exposure groups. A one-way MANOVA was performed to determine whether there is a difference between school Spanish exposure groups on standard score changes in Spanish semantics, Spanish morphosyntax, overall Spanish, English semantics, English morphosyntax, and overall English. There was no significant difference in bilingual oral language standard score changes with reference to home exposure groups, F(8, 316) = 0.60, p = .78, Wilk's lambda = 0.970.

Table 5.

Bilingual oral language standard score and changes across home exposure groups.

Variable	High Spanish home exposure (n = 64)				Mixed home exposure (n = 41)				Low Spanish home exposure (n = 59)
	Year 1	Year 2	Change	% of positive change	Year 1	Year 2	Change	% of positive change	Year 1	Year 2	Change	% of positive change
	M (SD)	M (SD)	M (SD)		M (SD)	M (SD)	M (SD)		M (SD)	M (SD)	M (SD)
Spanish semantics	83.57 (21.51)	90.79 (20.91)	7.21 (13.27)	70.31	81.10 (20.24)	88.00 (20.46)	6.90 (16.17)	73.17	86.35 (20.74)	96.53 (18.69)	10.18 (15.70)	76.27
Spanish morphosyntax	80.81 (13.29)	83.46 (12.77)	2.64 (6.27)	67.19	77.23 (12.15)	81.56 (11.81)	4.32 (6.07)	75.61	79.66 (13.07)	84.04 (18.86)	4.38 (5.80)	79.66
Spanish	82.19 (16.65)	87.12 (15.92)	4.93 (7.72)	75.00	79.16 (14.69)	84.78 (14.99)	5.61 (8.67)	78.05	83.01 (15.64)	90.29 (13.87)	7.28 (9.34)	86.44
English semantics	77.08 (20.34)	89.66 (17.90)	12.58 (17.72)	82.81	77.19 (20.79)	89.01 (18.97)	11.82 (19.19)	70.73	78.35 (17.63)	89.65 (18.86)	11.31 (16.78)	71.19
English morphosyntax	61.65 (11.02)	68.10 (10.91)	6.45 (5.84)	87.50	61.93 (11.49)	68.56 (10.93)	6.63 (6.44)	90.24	60.18 (10.01)	66.48 (9.91)	6.30 (6.28)	86.44
English	69.36 (14.42)	78.88 (12.88)	9.52 (9.87)	85.94	69.56 (14.89)	78.79 (13.63)	9.22 (10.78)	78.05	69.27 (12.20)	78.07 (12.83)	8.80 (9.84)	84.75

Changes in Spanish and English standard scores of bilingual oral proficiency were computed, and children were classified into four groups (English and Spanish gain, English gain and Spanish loss, Spanish gain and English loss, and English and Spanish loss). Table 6 shows the distribution of group memberships by Spanish school exposure groups. The chi-square test showed no significant difference in group membership change between home language exposure groups, χ²(6) = 6.73, p = .35.

Table 6.

Group membership distribution of bilingual oral language standard score changes across home exposure groups.

Group	Count
	High Spanish school exposure (n = 64; %)	Mixed school exposure (n = 41; %)	Low Spanish school exposure (n = 59; %)
Spanish and English gain	43 (67.19)	28 (68.29)	41 (69.49)
Spanish Gain and English loss	5 (7.81)	4 (9.76)	8 (13.56)
Spanish loss and English gain	12 (18.75)	4 (9.76)	9 (15.25)
Spanish and English loss	4 (6.25)	5 (12.20)	1 (1.69)

School Language Exposure Groups

Participants were grouped with reference to the school Spanish exposure. There were 51 participants in high Spanish school exposure group (Spanish exposure > 60%), 32 participants in mixed school exposure group (60% ≥ Spanish exposure ≥ 40%), and 81 participants in low Spanish school exposure group (Spanish exposure < 40%). Table 7 presents the standard scores of both years and changes in Spanish semantics, Spanish morphosyntax, English semantics, and English morphosyntax across school exposure groups. One-way MANOVA was performed to determine whether there is a difference between school Spanish exposure groups on standard score changes in Spanish semantics, Spanish morphosyntax, overall Spanish, English semantics, English morphosyntax, and overall English. There was a significant difference in bilingual oral language standard score changes with reference to school exposure groups, F(8, 316) = 2.30, p = .02, Wilk's lambda = 0.893, η_p² = .055. There was a significant effect of school Spanish exposure groups on standard score changes in Spanish morphosyntax, F(2, 161) = 3.80, p = .03, and overall Spanish, F(2, 161) = 3.43, p = .04. There was no significant effect of school exposure groups on standard score changes in Spanish semantics (p = .14), English semantics (p = .06), English morphosyntax (p = .30), and overall English (p = .09). Post hoc analysis with Bonferroni correction showed children from high Spanish school exposure group had a significantly higher standard score change in Spanish semantics compared to children from low Spanish school exposure group, mean difference = 2.88, p = .02. In addition, children from high Spanish school exposure group had a significantly higher standard score change in Spanish morphosyntax compared to children from low Spanish school exposure group, mean difference = 3.89, p = .03.

Table 7.

Bilingual oral language standard score and changes across school exposure groups.

Variable	High Spanish school exposure (n = 51)				Mixed school exposure (n = 32)				Low Spanish school exposure (n = 81)
	Year 1	Year 2	Change	% of positive change	Year 1	Year 2	Change	% of positive change	Year 1	Year 2	Change	% of positive change
	M (SD)	M (SD)	M (SD)		M (SD)	M (SD)	M (SD)		M (SD)	M (SD)	M (SD)
Spanish semantics	91.38 (13.89)	103.03 (10.95)	11.65 (14.24)	78.43	91.37 (13.00)	97.74 (12.52)	6.38 (13.97)	64.52	76.35 (24.25)	83.11 (22.91)	6.76 (15.46)	72.84
Spanish morphosyntax	83.79 (10.27)	89.13 (6.87)	5.34 (5.51)	86.27	82.73 (10.13)	86.92 (8.22)	4.19 (5.68)	77.42	75.53 (14.26)	77.98 (13.21)	2.45 (6.34)	64.20
Spanish	87.59 (10.25)	96.08 (7.43)	8.49 (8.15)	84.31	87.05 (10.17)	92.33 (9.30)	5.28 (7.78)	83.87	75.94 (18.32)	80.54 (16.96)	4.61 (8.88)	72.84
English semantics	72.78 (21.32)	89.88 (20.19)	17.10 (22.68)	76.47	75.92 (16.36)	85.49 (18.30)	9.57 (12.02)	77.42	81.23 (18.74)	90.84 (17.25)	9.61 (15.30)	72.84
English morphosyntax	56.85 (9.31)	63.12 (10.51)	6.27 (6.12)	88.24	61.29 (8.62)	66.41 (9.46)	5.11 (6.09)	87.10	63.89 (11.54)	70.96 (9.89)	7.07 (6.11)	87.65
English	64.82 (13.88)	76.50 (13.74)	11.68 (12.66)	84.31	68.61 (10.90)	75.95 (12.48)	7.34 (7.54)	83.87	72.56 (13.87)	80.90 (12.40)	8.34 (8.77)	82.72

Changes in the standard scores from Spanish and English measures of bilingual oral proficiency were computed, and children were classified into four groups (English and Spanish gain, English gain and Spanish loss, Spanish gain and English loss, and English and Spanish loss). Table 8 shows the distribution of group memberships by Spanish school exposure groups. The chi-square test revealed no significant difference in group membership change between school language exposure groups, χ²(6) = 9.61, p = .14.

Table 8.

Group membership distribution of bilingual oral language standard score changes across school exposure groups.

Group	Count
	High Spanish school exposure (n = 51; %)	Mixed school exposure (n = 32; %)	Low Spanish school exposure (n = 81; %)
Spanish and English gain	35 (68.62)	25 (80.65)	52 (64.20)
Spanish gain and English loss	8 (15.69)	2 (6.45)	7 (8.64)
Spanish loss and English gain	8 (15.69)	2 (6.45)	15 (18.52)
Spanish and English loss	0 (0.00)	3 (9.68)	7 (8.64)

Discussion

This study examined the relationship between language exposure at home and school and bilingual oral language proficiency in 164 children (144 who were TD kids and 20 with DLD) from kindergarten to first grade. All children had different levels of bilingual exposure at school. Age of first English exposure only significantly predicted Spanish morphosyntax in first grade, after controlling Spanish morphosyntax in kindergarten. Past studies have shown that language exposure is associated with bilingual children's language performance in different domains (Bowers & Vasilyeva, 2011; De Houwer, 1996; Pearson et al., 1997; Umbel et al., 1992). School and home language exposure contributed differently to bilingual children's language performance. Specifically, we found that both school and home language exposure contributed significantly to English morphosyntax, suggesting that children utilized both contexts to build morphosyntax knowledge. However, home Spanish exposure at kindergarten uniquely predicted Spanish semantics in first grade, controlling for demographic variables and school language exposure, while school Spanish exposure at kindergarten uniquely predicted Spanish morphosyntax at first grade, controlling for demographic variables and home language exposure. No differential contributions of home/school English exposure were observed on English morphosyntax and semantics skills. The high school Spanish exposure group had significantly higher standard score changes in Spanish morphosyntax and overall Spanish than the low school Spanish exposure group, without any significant difference on standard score changes in English. Home Spanish exposure groups did not significantly differ on standard score changes in Spanish and English.

Differential Contribution of Home and School Language Exposure

The results indicated that school and home language exposure contributed in different ways to bilingual oral language change, specifically in the child's home language, Spanish. Home Spanish exposure significantly and uniquely predicted change in Spanish semantics, whereas school Spanish exposure significantly and uniquely predicted change in Spanish morphosyntax. Both home and school language exposure had significant yet shared predictions of change in English morphosyntax.

The differential contribution of language exposure to language changes in Spanish is consistent with the characteristics of home and school input. As suggested earlier, school language exposure usually focuses on literacy and formal language skills and contains more syntactically complex language structures (Huttenlocher et al., 2002). Home language exposure, on the other hand, is more conversational, colloquial, and more likely to focus on the here-and-now, reflecting everyday language use during mealtime, family-based activities. Less complex syntactic structures are included in daily conversational input. Moreover, in bilingual families, cultural activities usually support language-specific vocabulary in home language environments. The results summarized by Kan et al. (2020) suggested that the first language was primarily used at home, including watching TV, meal time, reading activities and playing video games. On the other hand, our study suggests that changes in morphosyntax are associated with academic language at school, which contains more complex syntax and academic vocabulary (Lichtman, 2013). We believe that exposure to academic language in children's first and second language in formal school settings provided structural learning opportunities that enhanced grammatical understanding. However, the differential contribution of language exposure at home and school on changes in Spanish oral language proficiency suggests the importance of home language exposure across settings.

The finding that home and school language exposure significantly predicted changes in English morphosyntax implies that there was no unique contribution of context for learning English morphosyntax. It is likely that both environments provided opportunities for bilingual children to be exposed to and practice different English sentence structures. At the same time, home and school language exposure did not significantly predict performance on measures of English semantics. This finding may be explained by cross language transfer (Chung et al., 2019). Specifically, the transfer of semantic skills between two languages may be relatively direct compared to the transfer of morphosyntax skills, meaning that bilingual children can utilize semantic knowledge from their Spanish and apply it to English. Recall that bilingual children in the current study only had 1–2 years of school experience. They could utilize their home experience and knowledge in their home language to support school learning, but their English may still be in a phase of early development compared to Spanish. Therefore, both home and school inputs contribute to its growth, and it is difficult to disentangle them in this stage. This interpretation is in line with the Unified Competition Model (MacWhinney et al., 2005), which suggests that the transfer of lexicon is rapid because it is related to the difference of codes. This might explain why language exposure did not significantly predict change in English semantics.

Age of first English exposure did not predict Spanish semantics, English semantics, nor morphosyntax in first grade after controlling for demographic variables and performance in kindergarten. But age of first English exposure did significantly predict Spanish morphosyntax. This aligns with Pham's (2016) longitudinal study on bilingual development, which showed that the age of first English exposure did not have a significant association with English semantics and morphosyntax measures. Given that their initial bilingual language performance was controlled in their regression models, the effect of age of the first exposure might have been included already, which would explain their nonsignificant results.

Standard Score Change Across School Exposure Groups

Another major finding of the current study was that school Spanish exposure groups were significantly related to changes in standard scores on Spanish morphosyntax and overall Spanish, without significant effect on changes on English standard scores. This finding is consistent with prior research suggesting that children in bilingual classrooms surpassed children in predominantly English classrooms on measures of Spanish language development, including receptive and expressive vocabulary, literacy skills, and phonological awareness (Durán et al., 2022). English-only immersion programs, which focus exclusively on English language and literacy skills, provide the least academic instruction in the heritage language for bilingual children. In contrast, bilingual programs emphasize the continued use and academic support of Spanish, fostering greater academic success in both languages. In addition, the current findings were consistent with prior research indicating that high Spanish exposure at school enhances Spanish language development without compromising their English language development (Barnett et al., 2007). Specifically, the higher standard score changes in Spanish morphosyntax in children with high Spanish school exposure suggests that Spanish instruction and the use of Spanish at school provides more chances to practice Spanish. That children made gains in bilingual oral language development in a high Spanish school environment is crucial for students because it forms the foundation of biliteracy and is beneficial for classroom learning and socialization.

Our research was able to examine the transition from early education to formal education, testing children's language performance at different times and grades. Early education primarily focuses on oral language development, whereas formal education during first grade shifts its focus toward literacy. By examining language exposure at home and school, our study showed that bilingual education had no negative effects on English oral language and had positive impacts on Spanish oral language, laying a foundation for biliteracy development in later formal education. The patterns observed in the first grade can inform hypotheses about language development in later elementary years and beyond. As children progress into upper elementary grades, they encounter increasingly complex academic language demands, particularly in morphosyntax and literacy. Research shows that strong oral proficiency in heritage language provides a foundation for literacy development in both Spanish and English (Chung et al., 2019; Miller et al., 2006), suggesting that maintaining Spanish can enhance reading comprehension and academic outcomes across languages. Future studies may draw on these findings to hypothesize about how older bilingual children's oral competencies evolve with different school and home language exposure.

We used an assessment tool that was developed for Spanish–English bilinguals, compared to previous research which relied on measures of receptive vocabulary or other standardized language measurements that were originally developed for monolingual students (Stockman, 2000). The current study measured children's language abilities across four subtests, offering a more comprehensive evaluation of bilingual language performance. Moreover, we included children with a range of language exposure levels to represent the wide heterogeneity of bilingual students. Our results show that children with high Spanish school exposure benefitted differently compared to those with low Spanish school exposure. This differentiated approach enhances our understanding of the distinct impacts of language exposure contexts (school vs. home), aiding parents and guardians in making informed educational decisions. The opposition to bilingual education policies often arises from the fact that its rationale seems to run counter to widely held beliefs that a person learns another language by using it frequently and by avoiding use of one's native language (Ovando, 2003). Our findings challenge this misconception by demonstrating that Spanish exposure at home and school enhances heritage language development without compromising English proficiency.

Clinical Implications

The findings of this study suggest that it is important for clinicians to collect language exposure information for both home and school settings. Bilingual language history and contexts of use are important factors to consider when conducting assessments for bilingual children (Kohnert, 2012). The current study provides evidence that language exposure information is important for evaluating language performance. Furthermore, the differential impact of language exposure at home and school further suggests that clinicians should collect information about language exposure across settings during assessment. Clinicians can use this information for understanding children's oral language competence across languages and can consider different sources of input (Albudoor et al., 2024). Even though this study did not control for different types of language exposure, such as screen time, the findings still demonstrate that both home and school exposure uniquely contribute to different aspects of bilingual language development, and that Spanish exposure does not hinder English proficiency.

Additionally, clinicians may use language exposure information to track oral language performance changes in bilingual children. There were differences in standard score changes between our school exposure groups. Clinicians can use this finding to explain the bilingual language performance changes over time with reference to the school language exposure. Finally, the differential prediction of language exposure across settings suggests that assessment and treatment can be context-based. For semantic goals, integrating activities in home-like, conversational, and play-based contexts may be more effective. For morphosyntactic goals, structured and explicit instruction could be more beneficial.

Limitations and Future Directions

This study has two major limitations. First, the measurement of language exposure only relied on quantitative reports from teachers and parents. Parents from immigrant families may not speak the children's school language, making it difficult to report language exposure (Castilla-Earls et al., 2022). Likewise, teachers may find it challenging to accurately report the extent of language exposure among a whole class of children. In addition, in this study, language exposure was only measured by the amount of time, not considering the context of exposure. We did not examine qualitative aspects of language development, such as language richness, which plays an important role in shaping bilingual students' language abilities, especially in the domain of vocabulary (Cheung et al., 2019). Finally, we considered only two types of language exposure (home and school) and did not collect data on other forms, such as screen time, which is becoming an increasingly significant source of exposure to different language contexts.

To address the limitation of this study, future research should consider examining activities-based language exposure and language performance change. Prior research shows some home activities, such as television-watching, book reading, and dining, might contribute to semantic development (Cheung et al., 2019). In addition, as bilingual children get older, media exposure and literacy may increase in both languages. Albudoor et al. (2024) found that English media consumption grew at a faster rate than Spanish media consumption, whereas English literacy activities increased 2.5 times more than Spanish literacy activities. This trend highlights the increasing dominance of English exposure in children's screen time. Future research should also examine the relationship between different activities and morphosyntax development. Moreover, investigations should focus on the longitudinal effects of language shifts on bilingual language proficiency. Bedore et al. (2016) indicated that children often experience a language-dominance shift, particularly in their language use from home-language dominance to English-dominance. Thus, it is important to examine the relationship between language attrition and language exposure with reference to prior research showing children in bilingual immersion language programs experience Spanish loss from preschool to kindergarten (Hiebert & Rojas, 2021).

Ethics Statement

The study was approved by the institutional review boards at the University of Texas at Austin and Utah State University.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author, J. H. Y. Lam, upon reasonable request.

Funding Statement

This research was supported by National Institute on Deafness and Other Communication Disorders (NIDCD) Grant R01DC007439 (Principal Investigator: Elizabeth D. Peña). This article does not necessarily reflect the views or policy of the NIDCD.