Cognitive Distancing Language in Parent-Child Book Sharing Among Low-Income Mexican American and Chinese American Families: Cultural Group Comparisons and Links to Children’s Executive Functions

Abstract

Research Findings:

In the context of parent-child book sharing, cultural influences on cognitive distancing language and its associations with child executive function (EF) have been understudied. This study examined cultural group similarities/differences in parent and child book-sharing distancing language among preschool-aged dual language learners from low-income Mexican American (MA) and Chinese American (CA) families. We further tested concurrent relations between parent/child distancing language and children’s EF. The sample consisted of 88 children (45 MAs, 43 CAs, age = 38-68 months) enrolled in Head Start preschools and their parents. To assess distancing language, utterances were coded for cognitive demand and classified as referential (low demand), behavioral (moderate demand), or inferential (high demand). Cognitive tasks tapping inhibitory control and cognitive flexibility were used to assess child EF. Results showed parents and children from both cultural groups produced comparable numbers of story-related utterances. MA dyads used higher proportions of parent/child inferential statements and a lower proportion of parent referential questions. The proportions of parent behavioral questions and child behavioral statements were positively related to child cognitive flexibility.

Practice or Policy:

Findings highlight heterogeneity in parent-child verbal interactions among low-income multilingual families and have implications for promoting preschoolers’ cognitive/language development in diverse populations.

Introduction

During the preschool period, book sharing is an important avenue through which parents and children exchange rich and complex language (Adams, 1990; Luo et al., 2014; Noble et al., 2019). Cognitive distance, which reflects the cognitive demand of engaging with a given concept or idea (Sigel, 1993, 2002), has received attention as an aspect of book-sharing language that may be associated with children’s language and cognitive development (e.g., Ribner et al., 2020). Despite the long-theorized links between language and executive function (EF; e.g., Munakata, 2006; White et al., 2017; Zelazo, 1999), relations between book-sharing distancing language and children’s EF skills remain understudied, particularly in culturally diverse samples.

The present study aimed to fill this gap by examining cultural group similarities and differences in cognitive distancing language during parent-child book sharing among dual language learners (DLLs) from low-income Chinese American (CA) and Mexican American (MA) families. We further tested the concurrent relations between distancing language and children’s EF (inhibitory control and cognitive flexibility). Findings can inform the development of culturally competent interventions for promoting school readiness in young DLLs from socioeconomically disadvantaged backgrounds.

Cognitive Distancing Language During Parent-Child Book Sharing

Adult-child dyadic joint attention activities such as book sharing provide children the opportunity to produce and receive complex language, as well as to practice verbal reasoning skills (Arnold et al., 1994; Baker et al., 2015; Bus, 2003; Dexter & Stacks, 2014). Experimental research suggests that the language exposure and verbal reasoning practice that occur during book sharing can spur cognitive growth, making book sharing a well-recognized, low-cost intervention strategy for promoting child language and literacy skills (e.g., Manz et al., 2010).

Not all book sharing is created equal. Children are thought to benefit most from book sharing when the activity is interactive (Mol et al., 2008; Raikes et al., 2006). Interactive book sharing is often facilitated via active adult questioning to which children may respond (Huebner & Meltzoff, 2005; Mol et al., 2008). When adults ask children questions during book sharing, adults invite an increased number of child contributions, encouraging a deeper, more effortful processing of the language being uttered (Dalton-Puffer, 2011; Swain, 2000). Question-rich book-sharing interactions, also known as dialogic reading, are known to promote children’s language and academic development (Cristofaro & Tamis-LeMonda, 2012; Melzi et al., 2011; Mol et al., 2008; Sénéchal, 1997; Whitehurst et al., 1994). Thus, interactive reading (and dialogic reading in particular) is considered an important dimension of book-sharing quality.

Beyond interactive or dialogic emphasis, other aspects of book-sharing language quality have been studied. One such dimension is cognitive distance (Sigel, 1993, 2002). According to Sigel’s cognitive distancing theory (1993, 2002), the conceptual distance of a behavior, event, or idea from a child’s immediate context is associated with the level of cognitive demand placed upon the child in engaging with that behavior, event, or idea. Based on this perspective, book-sharing utterances lie along a low-to-high cognitive demand continuum (Luo & Tamis-LeMonda, 2017). Thus, the present study uses the term “cognitive distancing language” to refer to the classification of language units along the low-to-high cognitive demand continuum. Theoretically, the greater the distance, the more mental effort required to engage (Sigel, 1993, 2002). For example, children’s inferences about a book character’s emotional state have greater cognitive distance from their immediate context than descriptions of visual features of a pictured book character and are therefore considered more cognitively demanding (Luo & Tamis-LeMonda, 2017). Consistent with the theory, cognitive distance has shown positive associations with children’s reading and math skills (Ribner et al., 2020).

Researchers have also employed alternative terminologies (e.g., literal vs. inferential utterances, levels of abstraction, etc.) to characterize the cognitive difficulty entailed in engaging with varying types of book-sharing language (e.g., Tompkins et al., 2017; Zucker et al., 2010). Studies from these related literatures have likewise shown differences in child behavior and outcomes based on book-sharing language cognitive demand. For example, in the context of teacher-child book sharing, researchers found that children used longer utterances to respond to (comparatively more cognitively demanding) wh-style questions than to (comparatively less cognitively demanding) yes/no-style questions (Deshmukh et al., 2019). Additionally, a study of preschool teachers’ read-aloud practices in Chile found that teachers’ use of cognitively challenging talk (e.g., analysis of characters or events, summaries, defining word meanings, and making predictions) predicted children’s vocabulary growth over a school year (Gómez et al., 2017).

When studying book-sharing cognitive distancing language, separate consideration of questions and statements can shed light on whether relations between language cognitive demand and child outcomes are contingent on utterances’ communicatory function. Past research has broadly shown that questions and statements lead to different childhood outcomes (e.g., Willard et al., 2019). Specifically, Tompkins and colleagues (2017) found that inferential statements and yes/no questions but not inferential wh- questions predicted children’s receptive vocabulary growth. These findings that parent questions and statements at similar cognitive demand levels are differentially predictive of child language development illustrate the value of separately examining questions and statements during book sharing.

Parent-Child Book Sharing in Spanish- and Chinese-Speaking Families

Book-sharing cognitive distancing language among linguistically and culturally diverse American demographic groups merits attention given substantial and increasing childhood multilingualism in the U.S. One out of three children under age 9 in the U.S. is exposed to a non-English language at home (Park et al., 2017). These children are referred to as DLLs (U.S. Department of Health and Human Services, 2017). The most common home languages for this rapidly growing population are Spanish (59%) and Chinese (3.3%; Park et al., 2017).

Researchers have documented cultural variations in book-sharing practices. Research among Peruvian, European American, and MA mothers showed that Latina adults were more likely than European American ones to treat children as attentive listeners rather than narrative co-constructors (Luo et al., 2014; Melzi & Caspe, 2005; Melzi et al., 2011; Rodríguez et al., 2009). Similarly, Luo and colleagues (2014) found in a low-income sample that MA mothers used a greater proportion of statements and a lower proportion of questions compared to CA mothers. Cross-cultural studies showed that CA mothers were less likely than MA mothers and those from other ethnic groups to refer to thoughts and emotions during book sharing (Doan & Wang, 2010; Luo et al., 2014). Researchers have speculated that this might be due to Chinese/CA parents placing greater value on emotional restraint to fulfill social responsibilities, which is associated with a reluctance to talk directly about emotions (Chen et al., 2012; Luo et al., 2013, 2014; Wang et al., 2000). Furthermore, because a given language’s structural features could influence how speakers express themselves (e.g., the comparative availability in Spanish of a substantial emotion lexicon could promote emotional expression; Llabre, 2021), language-specific factors may also contribute to cultural group differences. In summary, previously documented cultural differences in general book-sharing practices justify thorough investigation of cultural differences in book-sharing distancing language specifically.

Despite recent attempts to quantify distancing language (or related language concepts) in adult-child book-sharing interactions (e.g., Blewitt et al., 2009; Kuchirko et al., 2016; Ribner et al., 2020; Tompkins et al., 2017), cultural group differences have rarely been explored. A notable exception is a study by Luo and Tamis-LeMonda (2017), which examined ethnic differences in book-sharing distancing language and parent-child language reciprocity (i.e., how parents and children verbally respond to each other) among low-income African American, Dominican American, MA, and CA participants. Luo and Tamis-LeMonda (2017) classified utterances as referential (low cognitive demand), behavioral (moderate cognitive demand), or inferential (high cognitive demand; see Method section for further information). Controlling for children’s expressive vocabulary and maternal education, they tested differences in parent questions and child statements. They found that CA mothers asked a greater proportion of inferential questions than MA mothers but did not observe other significant differences in maternal questions. CA children also contributed a greater proportion of referential statements but a smaller proportion of inferential statements than MA children. In contrast, no ethnic differences were found in reciprocity: mothers and children actively prompted and adapted to one another’s book-sharing behaviors similarly across groups (Luo & Tamis-LeMonda, 2017).

Further research on book-sharing distancing language is needed to better characterize cultural group variations. Yet, identifying cultural group differences in socioeconomically unmatched samples is rendered difficult by the overlap of race/ethnicity and socioeconomic status within the United States (Reeves et al., 2016). The present study aimed to overcome this pitfall by testing cultural group differences in parent/child distancing language among a low-income sample. Because Luo and Tamis-LeMonda’s (2017) cognitive distancing coding scheme had been used to specifically test cultural group differences, we adapted it for use in the present study. By doing so, we sought to replicate the findings in a low-income sample from a different geographical region of the U.S. We also endeavored to extend Luo and Tamis-LeMonda’s (2017) study by examining parental statements in addition to parental questions and child statements.

Links Between Book-Sharing Language and Children’s Executive Functions

EF is a self-regulatory construct linked to multiple developmental domains. It is generally conceptualized as a unitary construct with multiple, interrelated cognitive subcomponents (Garon et al., 2008). Two commonly studied subcomponents in research on early childhood are inhibitory control (i.e., one’s ability to inhibit a prepotent response) and cognitive flexibility (i.e., the degree to which one can fluidly adjust behavior in response to shifting contextual demands; Davidson et al., 2006; Diamond, 2013; Miyake et al., 2000). EF abilities rapidly develop during the preschool period (Zelazo et al., 2003). Moreover, preschool EF development predicts long-term academic achievement and socioemotional functioning (Best et al., 2009; Diamond, 2013; Zelazo & Carlson, 2020). Investigating EF among DLLs from low-income families is therefore consequential given individuals from this demographic group are more likely to have lower self-regulation in early childhood compared to socioeconomically similar, English-speaking monolinguals and more socioeconomically advantaged DLLs (Wanless et al., 2011).

Researchers have proposed various theories linking book-sharing language and child EF. For instance, information processing theory (Munakata, 2006; White et al., 2017) contends that children call on EF in storing and organizing the language they encounter, which might be one mechanism underlying the link between adult book-sharing language and child EF. Thus, a child’s EF may be differentially taxed depending on the complexity of the language to which they are exposed. Cognitive complexity and control theory suggests children’s ability to engage in EF-dependent problem-solving is driven by their ability to create representations of the problem through language (Doebel & Zelazo, 2016; Zelazo, 1999). Therefore, the more complex language children have at their disposal, the better they may be at EF-associated problem solving, thereby accounting for positive associations between child book-sharing language complexity and child EF.

Such theories have motivated empirical investigations of associations between language complexity and child EF while also shaping operationalizations of language complexity. Simple, global indicators of language complexity in studies of early childhood have been ubiquitous. For example, Gueron-Sela and colleagues (2018) found that maternal book-sharing language complexity at 36 months of age, operationalized as the mean length of utterances, positively predicted a task-based EF composite at 48 months. In another longitudinal study, the number of different words used by mothers during book sharing at 24 months positively predicted children’s EF at 48 months (Daneri et al., 2019). The same study also found that maternal mean length of utterance at 36 months similarly predicted children’s EF at 48 months. Although positive associations between early childhood independent narrative complexity and EF have been documented (e.g., Friend & Bates, 2014), specific relations between child narrative contributions to book sharing and EF have rarely been explored. One study that did investigate these relations found that the mean length of child utterances at 60 months was concurrently and positively associated with children’s performance on EF tasks (Kuhn et al., 2016). Moreover, the changes from 15 to 36 months and from 36 to 60 months in the number of different words children used were positively related to EF at 60 months (Kuhn et al., 2016). A methodological limitation of this line of research is the reliance on global indices of language complexity (e.g., mean utterance length, number of different words), which might fail to capture nuances in book-sharing content. These complexity operationalizations and the positive associations they have shown with child EF provide limited explanatory value beyond the proposition that “more language is better.”

To our knowledge, no previous study has investigated associations between book-sharing distancing language and child EF. Yet, we propose that cognitive distancing theory adds to other theoretical accounts of relations between child language and child EF by suggesting that the cognitive distance of language is associated with the level of EF needed to produce it. We also submit the theory suggests that a child’s EF skills are differentially taxed when comprehending language depending on its cognitive distance, establishing a link between parent language and child EF. In light of evidence suggesting that individuals improve on the EF skills they practice (Diamond & Ling, 2019), children may exercise and strengthen their executive “musculature” through comprehension and production of increasingly distant language. As others have previously theorized regarding parent book-sharing speech and child language and emotional development (e.g., Bailey et al., 2013; Mol et al., 2008), it is conceivable that parental language cognitive distance and child EF influence each other transactionally, consistent with Vygotsky’s zone of proximal development (1978; i.e., a problem-solving level a child cannot reach unaided but may be achieved with external guidance). Parents might modulate the complexity of their speech to approximately match the cognitive level of children while simultaneously spurring the development of EF by cognitively pushing them further. Evidence of mothers asking increasingly distant questions during book sharing as children age (Kuchirko et al., 2016) aligns with a bidirectional account of relations between book-sharing distancing language and child EF.

Although researchers have not tested the direct links between book-sharing distancing language and child EF, previous findings indirectly support the proposed relationship. Prior studies found positive relations between parental distancing language and other domains of child development including literacy, math development (Ribner et al., 2020), and symbolic play (Labrell et al., 2000), all of which are associated with EF (e.g., Kelly et al., 2011; Zelazo & Carlson, 2020). In sum, there is both theoretical rationale and indirect empirical evidence justifying the hypothesized links between book-sharing distancing language and child EF in the preschool period. The presence of direct relations could suggest that distancing language may facilitate EF development among DLLs in low-income families, a group at risk of early self-regulatory difficulties (Wanless et al., 2011).

Potential Covariates

Following Steiner et al.’s (2010) recommendation, we considered additional sociocultural/language variables that: a) are theoretically linked to both cultural group and book-sharing language when examining cultural group variations in book-sharing language, and b) are theoretically linked to both book-sharing language and children’s EF when examining the relations between book-sharing language and children’s EF. These factors included child age, gender, generation status, cultural group, family per capita income, at-home book sharing frequency, parent years of education, and child heritage language (i.e., Chinese/Spanish) expressive/receptive vocabulary. Child age and gender have been associated with EF (Grissom & Reyes, 2019; Huizinga et al., 2006). Existing evidence has also shown gender differences in book-sharing language specifically (Luo et al., 2014) and age differences in child-directed caregiver language generally (e.g., Huttenlocher et al., 2007). Previous research has linked ethnic/cultural group, family income, and parental education to differences in both EF and book-sharing language (e.g., Bernier et al., 2012; Luo & Tamis-LeMonda, 2017; Rea-Sandin et al., 2021). Child generation status may influence EF (e.g., Chen et al., 2015) and, as an acculturation indicator, may also relate to book-sharing cultural differences (e.g., Luo et al., 2014). Child verbal ability, additionally, has been found to relate to EF (e.g., Fuhs & Day, 2011) as well as book sharing (e.g., Raikes et al., 2006).

The Present Study

The present study examined book-sharing distancing language and its relation to child EF among MA and CA families with DLL children enrolled in Head Start preschool programs, which serve young children from low-income families in the U.S. The study had two aims. Aim 1 was to explore cultural group similarities/differences in parent and child distancing language. Given the relative lack of related empirical work – only Luo and Tamis-LeMonda (2017) have thus far tested ethnic/cultural differences using a comparable coding scheme – no directional hypotheses were associated with this aim. Aim 2 was to test the concurrent relations between parent and child distancing language and children’s EF. We expected the proportion of each parent distancing question type and each child distancing statement type to be positively associated with both inhibitory control and cognitive flexibility. Because higher parent statement proportions would necessarily entail lower parent question proportions (thereby signifying less dialogic book sharing), we expected the proportions of each parent distancing statement type to be negatively associated with child EF (inhibitory control and cognitive flexibility). We evaluated additional sociocultural and language variables (child age, gender, generation status, cultural group, family income and education, at-home book sharing frequency, and child heritage language) as potential covariates in the associations between cultural group and book-sharing language (Aim 1) and between book-sharing language and child EF (Aim 2). Altogether, there is tremendous practical importance to examining book-sharing distancing language and its relation to child EF among low-income MA/CA families with DLL children by virtue of the following: 1) the substantial size of the MA/CA DLL community (Park et al., 2017), 2) previous evidence of general cultural differences in book-sharing practices (and emerging understanding of cultural similarities/differences in book-sharing cognitive distancing language specifically; Luo & Tamis-LeMonda, 2017), 3) the importance of early childhood EF for future developmental outcomes (e.g., Zelazo & Carlson, 2020), and 4) the heightened risk for socioeconomically disadvantaged DLLs of lowered self-regulatory skills (Wanless et al., 2011).

Method

Participants

The sample consisted of 88 children (58% girls, age range = 38-68 months, M_age = 54.27 months, SD = 7.08) and their parents (one parent per child) drawn from a larger cross-sectional study of language and socioemotional development among low-income CA and MA DLLs (N = 90) enrolled in Head Start preschools in Northern California (Williams et al., 2019). Participants in the larger study were recruited from 15 Head Start centers. Two participants were excluded in the present study due to behavioral issues and procedural error in book-sharing task administration. Most children were 4 years of age (N = 46); the rest of the child participant sample was relatively evenly split between 3-year-olds (N = 19) and 5-year-olds (N = 23). Approximately equal numbers of CA (N = 43) and MA (N = 45) children participated in the present study. Of the participating children, 17% were born outside the U.S. (first generation), 77% were U.S.-born with at least one foreign-born parent (second generation), and 6% had two U.S.-born parents (third generation). The majority of children came from two-parent households (90.9%).

Continuous demographic variable descriptive statistics for the full sample and each cultural group (MA vs. CA) are displayed in Table 1. All but two participating parents were mothers. The participating parents’ countries of birth were China (45.5%), Mexico (43.2%), the U.S. (9.1%), or another country (2.3%). A plurality of participating parents reported working full-time as homemakers (28.4%), while another 10.2% and 22.7% reported having full-time and part-time employment respectively. 10.2% reported having occasional employment and/or working as day laborers. The remaining parents were either full-time students (3.4%) or unemployed (13.6% unemployed and looking for work; 11.4% unemployed and not looking for work). To calculate average per capita income, we divided the total income in the past year by the number of individuals living in the household.

Table 1.

Descriptive statistics for continuous demographic, executive function, and book-sharing variables

	Full Sample						CA Sample (N = 43)		MA Sample (N = 45)
Variable	Min	Max	Mean	SD	Skewness	Kurtosis	Mean	SD	Mean	SD
Age in Months (Child)	38	68	54.27	7.08	−0.11	−0.81	53.47	7.22	55.04	6.94
Age in Years (Parent)	21	46	34.48	6.41	−0.22	−0.72	38.05	4.79	31.00	5.89
Years of Education (Parent)	0	18	11.11	3.78	−0.74	−0.23	11.19	3.84	11.03	3.76
Years in U.S. (Parent)	0	28	9.31	6.19	0.92	0.62	6.40	4.74	12.53	6.05
Per Capita Income in Dollars	1,000	24,166.67	5,225.83	3,676.01	2.51	8.82	5,696.84	3,430.81	4,731.83	3,897.69
Inhibitory Control (SSS)	0.03	1.00	0.64	0.28	−0.55	−0.81	0.54	0.31	0.74	0.21
Cognitive Flexibility (STS)	0.30	1.00	0.77	0.16	−0.52	−0.26	0.75	0.18	0.79	0.14
Proportion of Referential Statements (Parent)	0.04	0.42	0.16	0.08	0.99	1.18	0.17	0.08	0.16	0.07
Proportion of Behavioral Statements (Parent)	0.13	0.70	0.40	0.12	0.14	−0.11	0.39	0.13	0.41	0.12
Proportion of Inferential Statements (Parent)	0.00	0.43	0.21	0.10	−0.13	−0.68	0.16	0.08	0.26	0.08
Proportion of Referential Questions (Parent)	0.00	0.28	0.09	0.07	0.75	−0.22	0.12	0.07	0.06	0.06
Proportion of Behavioral Questions (Parent)	0.00	0.18	0.04	0.04	1.55	2.75	0.03	0.04	0.04	0.04
Proportion of Inferential Questions (Parent)	0.00	0.14	0.03	0.04	1.07	0.41	0.04	0.04	0.03	0.03
Proportion of Referential Statements (Child)	0.00	1.00	0.36	0.22	0.56	0.59	0.38	0.22	0.35	0.22
Proportion of Behavioral Statements (Child)	0.00	0.50	0.19	0.15	0.17	−1.05	0.20	0.15	0.18	0.14
Proportion of Inferential Statements (Child)	0.00	0.75	0.18	0.16	0.90	0.64	0.11	0.13	0.23	0.17

Note. SD = standard deviation, CA = Chinese American, MA = Mexican American, SSS = Silly Sounds Stroop, STS = Something’s the Same.

To compare the two cultural groups on demographic variables, independent-sample t-tests, Pearson chi-square tests of independence, and Fisher’s exact tests were computed. A significant cultural difference in generation status was found (Fisher’s exact test, p < .001, V = .50). Among CA children, 34.9% were first-generation and 65.1% were second-generation. The MA group consisted of 88.9% second-generation and 11.1% third-generation children. Likewise, a significant difference was observed in the average time parents had lived in the U.S. (t(78) = 5.06, p < .001, d = 1.13). Parents in the MA group had lived in the U.S. an average of 12.53 years (range = 5-28, SD = 6.05), whereas CA parents had lived in the U.S. an average of 6.4 years (range = 0-17, SD = 4.74). This is congruent with American immigration trends indicating Asian Americans immigrants are more likely than other immigrant groups to have arrived recently (Pew Research Center, 2012). The MA and CA groups significantly differed on parent age (t(85) = −6.12, p < .001, d = 1.31), with CA parents being older (M = 38.05 years, range = 27-56, SD = 4.79) than MA parents (M = 31 years, range = 21-42, SD = 5.89). No significant differences were found in child age, gender, parental education levels, per capita income, proportion of two-parent vs. one-parent households, parent employment status, or gender of the participating parent.

Procedures

To recruit the sample for the larger study, multilingual research assistants visited local Head Start centers serving high concentrations of CA and MA children. Once there, they explained the study to parents and handed out flyers (available in Chinese, Spanish, and English). Contact forms from 229 families were collected across the 15 Head Start sites. Child inclusion criteria were: age 36-71 months, current enrollment at Head Start center at least three days per week, ability to understand/speak at least some English and Spanish/Cantonese/Mandarin, both parents self-identify as either ethnically Chinese or Mexican. Exclusion criteria included diagnosis of speech/language disorder and receipt of speech/language services. Of the 132 families determined to be eligible during screening, 90 (44 CA, 46 MA) completed the assessment for the larger study.

The assessment lasted approximately 2.5 hours and took place either at the family’s home or at our university laboratory (depending on parent preference). The assessment included child EF and language tasks (counterbalanced for each participant), parent questionnaires, and parent-child interaction tasks. Informed consent and parent questionnaires were administered in parents’ preferred language. EF tasks were administered in the child’s dominant language (English, Mandarin, Cantonese, or Spanish). The child’s dominant language was initially reported by parents during the phone screening. It was subsequently confirmed during the assessment’s first 10 minutes through observation by two trained research assistants. The two research assistants provided the child with a set of toys (play-doh and a coloring kit) and asked the child questions (one research assistant asked in English while the other asked in the child’s heritage language). The language that elicited the most responses from the child was designated as the dominant language. Among the children with book-sharing language data, 35 MA children and 33 CA were heritage-language dominant, and 7 MA children and 6 CA children were English-language dominant. Children received small prizes and parents were paid $70 for their participation.

Measures

Book Sharing Task and Language Coding

Book-sharing language was elicited using Frog, Where Are You? (Mayer, 1969), a wordless picture book commonly used by researchers to study narrative development (e.g., Luo & Tamis-LeMonda, 2017; Melzi & Caspe, 2005). A wordless picture book was chosen because it allowed parents to engage in book sharing without being restricted by linguistic background or literacy skills. Parents were allowed to use whichever language they preferred (English, Cantonese, Mandarin, or Spanish) and were specifically told: “I have a book for you to read. This book has pictures but no words and we would like you to tell the story to [child’s name]. You can use whichever language you are comfortable with. You will have 5 minutes alone to tell the story to your child, then I will come back in.” Story-related talk from all but nine book-sharing sessions (6 MA, 3 CA) was predominantly in participants’ heritage language. Of the predominantly heritage language CA book-sharing sessions, only 10 were in Mandarin while the rest were in Cantonese. Book-sharing sessions were video-recorded. Although most dyads finished reading the book within 5 minutes, the dyads who did not finish within 5 minutes were allowed to continue reading until they were done with the book. Multilingual English/Spanish and English/Chinese research assistants subsequently used those recordings to transcribe verbal interactions in their original language.

An adapted version of Luo and Tamis-LeMonda’s (2017) cognitive distancing coding scheme was used to code transcribed parent/child talk (again, in its original language). An utterance, defined as an independent clause or an isolated verbal expression conveying communicative information, constituted the unit of analysis (Crookes, 1990; Czwenar, 2014). One English/Spanish bilingual research assistant coded the MA transcripts, and one English/Chinese multilingual research assistant coded the CA transcripts. Each utterance was classified as a question or statement and coded according to the following mutually exclusive categories (see Table 2 for examples of each utterance type). In accordance with Luo and Tamis-LeMonda (2017), referential (low cognitive demand) utterances were questions that asked for or statements that described the “name, feature, or location of objects in the pictures.” Behavioral (moderate cognitive demand) utterances were questions that asked for or statements that described “story characters’ actions in the pictures.” Inferential (high cognitive demand) utterances were questions that prompted or statements that drew upon “inferences, imagination, internal states, or social knowledge that can only be inferred from but not directly observed in the pictures.” Although not the focus of the present study because they are thought to be conceptually distinct from the other utterance types (Luo et al., 2022; Luo & Tamis-LeMonda, 2017), story-related questions that could be answered with or independent statements of “yes” or “no” (i.e., yes/no utterances) were also coded. Utterances not fitting any of these categories (such as those unrelated to the book) were coded as “other”. For inter-rater reliability across languages, 25 transcripts were randomly selected, translated into English by multilingual research assistants, and coded by the two coders (Cohen’s kappa = .77).

Table 2.

Distancing language types and examples

Utterance Type	Example
Referential (Low Cognitive Demand/Distance)	Question: Where is the dog?
	Statement: The boy’s shirt has stripes.
Behavioral (Moderate Cognitive Demand/Distance)	Question: What is the dog doing?
	Statement: The bees are chasing the dog.
Inferential (High Cognitive Demand/Distance)	Question: Why is the boy mad?
	Statement: The boy said, “The frog is gone!”

Note. Adapted from Luo and Tamis-LeMonda (2017).

To account for varying book-sharing session lengths, utterance proportions were calculated for coding categories of interest by dividing the number of utterances in each category by the total number of book-related (i.e., non-other) utterances produced by the same speaker. Yes/no questions and statements were not considered coding categories of interest as they were not the focus of the present study. Child referential (M = 0.52, SD = 1.01), behavioral (M = 0.15, SD = 0.43), and inferential (M = 0.54, SD = 1.12) questions were excluded as they occurred infrequently. However, these utterance types were included in the total number of utterances when calculating proportions. Thus, the final nine book-sharing variables used in analyses were: the proportions of parent referential questions, parent referential statements, parent behavioral questions, parent behavioral statements, parent inferential questions, parent inferential statements, child referential statements, child behavioral statements, and child inferential statements. No differences were found for these variables between the book-sharing sessions with predominantly heritage language, story-related talk and the nine sessions for which less than half the story-related utterances were entirely heritage language (as opposed to English or mixed language utterances; ps = .089 to .964).

Executive Function

Two tasks (assessing inhibitory control and cognitive flexibility) from the Preschool Executive Functions Assessment (Willoughby et al., 2010) were used to assess child EF. Inhibitory control and cognitive flexibility were the focus of the present study given their developmental sensitivity during the preschool period (Carlson, 2005). The tasks were administered in the child’s dominant language. These tasks have previously demonstrated validity in samples of Chinese-speaking and Spanish-speaking preschoolers (White & Greenfield, 2017; Xing et al., 2019). The performance of children who completed the EF tasks in English (N = 16) was compared to that of the children who completed the tasks in their heritage language (N = 72; ps = .707 and .908). No significant differences were found, indicating EF task language did not influence performance. For both tasks, the percentage of correct responses was used in data analyses.

The Silly Sounds Stroop (SSS) task assessed inhibitory control. During the task, children were shown two images of a cat and a dog. The assessor then informed participants that in the SSS dogs make cat sounds (“meow”) and cats make dog sounds (“woof”). The assessor presented 36 side-by-side images of a dog and cat, pointed to each, and asked the participant to vocalize the sound made by the animal in question. Responses were marked correct if children were able to follow the rules, inhibiting a previously learned response in favor of a novel one (i.e., cats “woof” and dogs “meow”). Other responses, including self-corrections, were marked incorrect. Cronbach’s alpha for responses on the SSS was 0.95 (0.96 for CA, 0.91 for MA).

The Something’s the Same (STS) task assessed cognitive flexibility. During each trial, the assessor displayed two images that were similar in color, size, or content and explicitly named the dimension of similarity. Then the assessor presented the same two images alongside a third image that matched one of the original images along a new dimension of similarity. The participant was instructed to identify which of the first two images matched the third. Responses were marked correct if the participant successfully identified the matching pair and were otherwise marked incorrect. Correct responses required participants to flexibly shift attention from one similarity dimension to another. Cronbach’s alpha for responses on the STS was 0.72 (0.75 for CA, 0.68 for MA).

Demographic Covariates

Parent responses on an adapted version of the Family Demographics and Migration History Questionnaire, a measure previously used with CA/MA immigrant families, provided information about demographic factors including child age, gender, ethnicity/cultural group, generation status, parent educational background, family income, and household size (Chen et al., 2014; Roosa et al., 2008). Parents also completed a child language history questionnaire (Leung & Uchikoshi, 2012), which included the paired items “Do you or another adult in your home read to [child’s name]?” (answered “yes” or “no”) and “If yes: How often?” (answered on a 5-point scale, 1 = every day, 5 = once a month). For analyzing at-home book sharing frequency, responses on these items were merged and recoded to produce a 6-point scale (0 = no at-home book sharing, 5 = daily book sharing).

In the larger project, all children were administered four vocabulary tests regardless of their dominant language: a receptive heritage language vocabulary test (Spanish for MA children and Chinese for CA children), an expressive heritage language vocabulary test, a receptive English vocabulary test, and an expressive English vocabulary test. The present study only used data on children’s receptive and expressive heritage language vocabulary. The Woodcock Language Proficiency Battery-Revised, Spanish Form’s Vocabulario Sobre Dibujos subtest was used to assess MA children’s heritage language (i.e., Spanish) expressive vocabulary (Woodcock & Muñoz-Sandoval, 1995). As had been done in previous research with CA DLLs (Chernoff et al., 2021; Uchikoshi, 2014), CA children’s heritage language (i.e., Cantonese/Mandarin) vocabulary was assessed using the Spanish measure’s stimuli. During the test, children were asked to name the contents of images in order of increasing difficulty. Assessors scored child responses dichotomously (either correct or incorrect). Raw scores were used in data analyses given the lack of norm-referenced scores for the CA participants.

Children’s receptive heritage language vocabulary was measured using the Chinese (Lu & Liu, 1998) or Spanish (Dunn et al., 1986) versions of the Peabody Picture Vocabulary Test-Revised (Dunn & Dunn, 1981). During each task trial, the assessor uttered a word, simultaneously presented four images, and asked the participant to select the image best matching the uttered word. As with the expressive vocabulary measure, raw scores of receptive heritage language vocabulary were used in analyses.

Results

Data analyses were conducted in three steps. First, correlation analyses, t-tests, ANOVAs, chi-square tests of independence, and Fisher’s exact tests were conducted to identify demographic variables related to both book-sharing distancing language and cultural group or those that related to both book-sharing distancing language and EF variables. Second, multiple regression models were computed to test book-sharing distancing language’s unique associations with cultural group and the two EF variables respectively (controlling for covariates). As a final step, we adjusted the p-values of the associations of interest in the multiple regression models to correct for multiple comparisons.

Study variable descriptive statistics for the full sample as well as each cultural group are displayed in Table 1. Study variables and potential covariates were screened for normality. The only variable to violate recommended skewness and kurtosis cutoffs of 2 and 7 respectively (West et al., 1995), was per capita income (Skewness = 2.51, Kurtosis = 8.82). Log-transformation reduced skewness to 0.00 and kurtosis to 0.27. Log-transformed per capita income was consequently used in analyses. All analyses were conducted using R Version 4.1.0 (R Core Team, 2021) and RStudio Version 1.4.1717 (RStudio Team, 2021).

Relations Between Study Variables and Potential Covariates

As recommended by Steiner et al. (2010), potential covariates were included in subsequent analyses if they significantly related to both predictor and outcome variables. To examine whether potentially confounding factors were associated with Aim 1’s variables of interest (cultural group and the nine book-sharing variables), we computed Pearson’s correlations, independent-sample t-tests, ANOVAs, Pearson chi-square tests of independence, and Fisher’s exact tests. Similarly, we computed Pearson’s correlations, independent-sample t-tests, and ANOVAs to test whether potentially confounding factors were associated with Aim 2’s variables of interest (the two EF variables and the nine book-sharing variables). Pairwise correlations of all study variables and continuous potential covariates are displayed in Table 3.

Table 3.

Zero-order correlations between continuous key predictors/outcomes and continuous potential covariates

	Child Age	Parental Education (Years)	Per Capita Income (Dollars)	At-Home Book Sharing Frequency	Child Heritage Language Expressive Vocabulary	Child Heritage Language Receptive Vocabulary
Inhibitory Control (SSS)	.23*	−.03	−.10	.14	.15	.01
Cognitive Flexibility (STS)	.50***	.03	.21	.19	.19	.21
Proportion of Referential Statements (Parent)	−.01	.01	−.13	−.22	−.18	.09
Proportion of Behavioral Statements (Parent)	.00	.01	−.09	−.32**	−.03	−.04
Proportion of Inferential Statements (Parent)	−.08	.00	−.08	.26*	.10	−.04
Proportion of Referential Questions (Parent)	.06	.10	.30**	.12	.00	.02
Proportion of Behavioral Questions (Parent)	.08	−.26*	.04	.23*	.13	.20
Proportion of Inferential Questions (Parent)	.14	.09	−.02	.31**	.13	.07
Proportion of Referential Statements (Child)	.00	.01	.25*	.21	.05	−.13
Proportion of Behavioral Statements (Child)	.14	−.22	.01	−.05	.33**	.35**
Proportion of Inferential Statements (Child)	.12	.01	−.28*	.14	.11	.03

Note. SSS = Silly Sounds Stroop, STS = Something’s the Same; the ns for the correlations ranged from 68 to 88;

^*p < .05,

^**p < .01,

^***p < .001.

As shown in Table 3, the following variables showed significant correlations with at least one book-sharing variable: parental education, per capita income, at-home book sharing frequency, heritage language expressive vocabulary, and heritage language receptive vocabulary. Independent-sample t-tests also revealed a significant cultural difference in book sharing frequency (t(85) = 3.20, p = .002, d = 0.69) such that MA parents reported a higher frequency than CA parents. Thus, book sharing frequency was included as a covariate in Aim 1 analyses. Regarding Aim 2, child age was positively correlated with child EF. Gender also significantly related to children’s inhibitory control (t(80) = 2.30, p = .024, d = 0.51) and cognitive flexibility (t(85) = 2.44, p = .017, d = 0.53), with girls performing better on both tasks than boys. However, cultural group was the only potential covariate significantly associated with both child EF (t(80) = 3.41, p = .001, d = 0.76) and a book-sharing variable (ts (dfs = 73 to 77) = 3.52 to 5.32, ps < .001, ds = 0.81 to 1.20). Specifically, the MA group performed better on the children’s inhibitory control task and had higher proportions of parent and child inferential statements than the CA group. But the MA group had a lower proportion of parent referential questions than the CA group. Thus, cultural group was included as a covariate in Aim 2 analyses.

Aim 1 Analyses: Associations Between Cultural Group and Book-Sharing Variables

As noted, preliminary analyses revealed cultural differences in proportions of parent referential questions, parent inferential statements, and child inferential statements. Nine multiple regression models (one for each book-sharing variable) were used to investigate unique associations between cultural group and the book-sharing variables of interest, controlling for book sharing frequency (Table 4). Cultural differences in book-sharing language remained significant in the regression models. Across models, the CA group displayed lower proportions of parent (p < .001) and child (p = .002) inferential statements, but a higher proportion of parent referential questions than the MA group (p < .001). Book sharing frequency uniquely and positively predicted proportions of parent referential questions (p = .005), parent inferential questions (p = .001), and child referential statements (p = .042). Book sharing frequency was negatively associated with parent behavioral statements (p = .001). To correct for multiple comparisons, we adjusted the p-values of the nine hypothesized cultural group differences using the Benjamini-Hochberg false discovery rate (FDR) correction (Benjamini & Hochberg, 1995). Cultural differences in the proportions of parent inferential statements (FDR adjusted p < .001), child inferential statements (FDR adjusted p = .002), and parent referential questions (FDR adjusted p < .001) remained significant after the adjustment. Two additional multiple regression models were used to test cultural group differences in overall numbers of story-related parental utterances (F(2,76) = 3.33, p = .041) and child utterances (F(2,76) = 2.18, p = .120). While book sharing frequency positively predicted parent (β = .29, p = .016) and child (β = .24, p = .046) utterances, no cultural group differences were found in either parent utterances or child utterances (ps = .928 and .831).

Table 4.

Multiple regressions testing cultural differences in book-sharing variables

	Proportion of Referential Statements (Parent)		Proportion of Referential Questions (Parent)		Proportion of Referential Statements (Child)
Predictors	B (SE)	β	B (SE)	β	B (SE)	β
Cultural Group (CA)	−.00 (.02)	−.03	.08*** (.01)	.54	.06 (.05)	.14
At-Home Book Sharing Frequency	−.01 (.01)	−.23	.02** (.01)	.30	.05* (.02)	.25
Total R2	.05		.28***		.06
N	79		79		75
	Proportion of Behavioral Statements (Parent)		Proportion of Behavioral Questions (Parent)		Proportion of Behavioral Statements (Child)
Predictors	B (SE)	β	B (SE)	β	B (SE)	β
Cultural Group (CA)	−.05 (.03)	−.19	−.00 (.01)	−.04	.02 (.04)	.05
At-Home Book Sharing Frequency	−.04** (.01)	−.38	.01 (.00)	.21	−.00 (.02)	−.03
Total R2	.14**		.05		.00
N	79		79		75
	Proportion of Inferential Statements (Parent)		Proportion of Inferential Questions (Parent)		Proportion of Inferential Statements (Child)
Predictors	B (SE)	β	B (SE)	β	B (SE)	β
Cultural Group (CA)	−.09*** (.02)	−.49	.01 (.01)	.19	−.12** (.04)	−.37
At-Home Book Sharing Frequency	.01 (.01)	.10	.01** (.00)	.37	.00 (.02)	.02
Total R2	.28***		.13**		.15**
N	79		79		75

Note. Child statement models have lower ns because proportions could not be calculated for child participants who did not speak during book sharing (as doing so would involve dividing by zero); CA = Chinese American, B = unstandardized regression coefficient, SE = standard error, β = standardized regression coefficient;

^*p < .05,

^**p < .01,

^***p < .001.

Aim 2 Analyses: Associations Between Book-Sharing Variables and Child Executive Functions

Inhibitory Control

To test the unique associations between children’s inhibitory control and distancing language, nine multiple regression models were computed (Table 5). Children’s inhibitory control was positively related to parent referential questions (p = .035). Parent and child statement variables showed no significant relations to child EF. Cultural group was a significant predictor of child inhibitory control in all models such that CA children showed lower accuracy on the inhibitory control task than MA children (ps < .001). After adjusting the p-values of the hypothesized associations between child inhibitory control and distancing language, the proportion of parent referential questions was no longer a significant predictor (FDR adjusted p = .210).

Table 5.

Multiple regressions testing relations of inhibitory control and book-sharing proportions

Inhibitory Control (Silly Sounds Stroop)
Predictors	B (SE)	β	Predictors	B (SE)	β	Predictors	B (SE)	β
Referential Statements (Parent)	.25 (.40)	.07	Behavioral Statements (Parent)	−.39 (.24)	−.17	Inferential Statements (Parent)	−.65 (.37)	−.22
Cultural Group (CA)	−.23*** (.06)	−.41	Cultural Group (CA)	−.24*** (.06)	−.42	Cultural Group (CA)	−.30*** (.07)	−.53
Total R2	.17**		Total R2	.20***		Total R2	.20***
N	73		N	73		N	73
Predictors	B (SE)	β	Predictors	B (SE)	β	Predictors	B (SE)	β
Referential Questions (Parent)	1.04* (.48)	.25	Behavioral Questions (Parent)	.69 (.74)	.10	Inferential Questions (Parent)	.95 (.86)	.12
Cultural Group (CA)	−.30*** (.07)	−.52	Cultural Group (CA)	−.23*** (.06)	−.40	Cultural Group (CA)	−.24*** (.06)	−.42
Total R2	.22***		Total R2	.18***		Total R2	.18***
N	73		N	73		N	73
Predictors	B (SE)	β	Predictors	B (SE)	β	Predictors	B (SE)	β
Referential Statements (Child)	.21 (.14)	.16	Behavioral Statements (Child)	.33 (.21)	.17	Inferential Statements (Child)	−.09 (.21)	−.05
Cultural Group (CA)	−.26*** (.06)	−.46	Cultural Group (CA)	−.26*** (.06)	−.46	Cultural Group (CA)	−.27*** (.07)	−.47
Total R2	.23***		Total R2	.23***		Total R2	.21***
N	71		N	71		N	71

Note. Child statement models have lower ns because proportions could not be calculated for child participants who did not speak during book sharing (as doing so would involve dividing by zero); all language variables are proportions, CA = Chinese American, B = unstandardized regression coefficient, SE = standard error, β = standardized regression coefficient;

^*p < .05,

^**p < .01,

^***p < .001.

Cognitive Flexibility

Nine multiple regression models were computed to test unique associations between children’s cognitive flexibility and distancing language (Table 6). Children’s cognitive flexibility was positively associated with parent referential questions (p = .017), parent behavioral questions (p = .008), and child behavioral statements (p = .014). Parent statement variables showed no significant relations. Cultural group significantly predicted children’s cognitive flexibility in the parent referential question model such that CA children showed lower accuracy on the cognitive flexibility task than MA children after controlling for parent referential questions (p = .033). Only parent behavioral questions (FDR adjusted p = .048) and child behavioral statements (FDR adjusted p = .042) remained significantly and positively related to child cognitive flexibility after FDR adjustment.

Table 6.

Multiple regressions testing relations of cognitive flexibility and book-sharing proportions

Cognitive Flexibility (Something’s the Same)
Predictors	B (SE)	β	Predictors	B (SE)	β	Predictors	B (SE)	β
Referential Statements (Parent)	−.29 (.25)	−.13	Behavioral Statements (Parent)	−.27 (.15)	−.20	Inferential Statements (Parent)	−.17 (.23)	−.10
Cultural Group (CA)	−.04 (.04)	−.13	Cultural Group (CA)	−.05 (.04)	−.14	Cultural Group (CA)	−.06 (.04)	−.19
Total R2	.04		Total R2	.06		Total R2	.03
N	78		N	78		N	78
Predictors	B (SE)	β	Predictors	B (SE)	β	Predictors	B (SE)	β
Referential Questions (Parent)	.72* (.30)	.30	Behavioral Questions (Parent)	1.21** (.45)	.30	Inferential Questions (Parent)	.65 (.53)	.14
Cultural Group (CA)	−.09* (.04)	−.27	Cultural Group (CA)	−.03 (.04)	−.10	Cultural Group (CA)	−.05 (.04)	−.15
Total R2	.09*		Total R2	.11*		Total R2	.04
N	78		N	78		N	78
Predictors	B (SE)	β	Predictors	B (SE)	β	Predictors	B (SE)	β
Referential Statements (Child)	.03 (.09)	.04	Behavioral Statements (Child)	.32* (.13)	.28	Inferential Statements (Child)	−.10 (.13)	−.09
Cultural Group (CA)	−.05 (.04)	−.16	Cultural Group (CA)	−.06 (.04)	−.17	Cultural Group (CA)	−.06 (.04)	−.19
Total R2	.03		Total R2	.10*		Total R2	.03
N	75		N	75		N	75

Note. Child statement models have lower ns because proportions could not be calculated for child participants who did not speak during book sharing (as doing so would involve dividing by zero); all language variables are proportions, CA = Chinese American, B = unstandardized regression coefficient, SE = standard error, β = standardized regression coefficient;

^*p < .05,

^**p < .01,

^***p < .001.

Discussion

The present study examined cultural group similarities and differences in parent-child book-sharing distancing language among low-income CA and MA preschool-age DLLs and their parents using a single picture book. Though similar in many respects (including in the number of story-related utterances produced by both parents and children), results revealed that MA and CA parents differed in the proportion of referential questions they asked and that both parents and children differed in the proportion of inferential statements they uttered. These cultural group variations remained significant after applying an FDR correction. Furthermore, we examined distancing language’s associations with child inhibitory control and cognitive flexibility. After applying the FDR correction, we found the proportions of parent behavioral questions and child behavioral statements to be positively associated with child cognitive flexibility.

Regarding cultural group variations in parents’ book-sharing language, MA parents’ language can be said to have been more cognitively distant than that of CA parents. This is because the only cultural difference in parental language we observed across the levels of cognitive distance was that MA parents’ language contained proportionally more high distance utterances (i.e., a higher proportion of inferential statements). Likewise, MA parents’ language can be said to have been less dialogic than that of CA parents because the only cultural difference observed in parental questions was that MA parents displayed a lower proportion of referential questions; this interpretation is further supported by the fact that a higher parent question proportion mathematically equates to a lower parent statement proportion. These findings align with previous evidence of MA mothers’ talk being less dialogic (Luo et al., 2014), which is, in turn, congruent with the proposition that Latina adults tend to treat children as attentive listeners rather than narrative co-constructors during book sharing (Melzi & Caspe, 2005; Melzi et al., 2011; Rodríguez et al., 2009). Because all mental-state talk (including all references to both cognition and emotion) qualified as inferential, our findings are also congruent with research that showed CA mothers were less likely to refer to thoughts or feelings (Doan & Wang, 2010; Luo et al., 2014). Indeed, these findings are consistent with recent research analyzing cultural differences in book-sharing emotion talk between CA and MA mothers drawing from the same book-sharing recordings as the present study (Chan et al., 2022). Chan et al. (2022) observed that CA parents engaged in less elaborate emotion talk (i.e., talk related to feeling states), used fewer negative emotion words, and employed less emotion reasoning than MA parents, suggesting that CA parents’ observed usage of a lower proportion of inferential statements in the present study may indeed reflect a culturally specific reluctance to talk directly about emotions (Luo et al., 2013, 2014; Wang et al., 2000). These findings also align somewhat with those of Luo and Tamis-LeMonda (2017). Though they found no MA/CA referential question difference when comparing proportions of each question type out of all parental questions, the proportion of mothers’ referential questions out of all parental utterances was higher among the MA group (though this difference was not formally tested). In contrast to the present study, Luo and Tamis-LeMonda (2017) found that the proportion of inferential questions out of all parental questions was higher among CA mothers than MA mothers. We have two explanations for this discrepancy. First, unlike Luo and Tamis-LeMonda (2017), we compared parent questions using the number of each question type divided by the total count of book-related parent utterances (instead of dividing only by the number of parental questions). Therefore, we calculated utterance proportions differently. Second, as has been suggested (e.g., Kuchirko et al., 2016), parents may adjust their speech complexity to match children’s abilities. Thus, it is possible that MA parents produced more cognitively distant speech to match the comparatively higher EF of our MA child participants.

Our finding that CA children used a smaller proportion of inferential statements more straightforwardly aligns with previous results showing CA children’s statements being more referential and less inferential than those of MA children (Luo & Tamis-LeMonda, 2017). Again, because the inferential category contained mental-state talk, CA children’s speech patterns may reflect a culturally specific reluctance to directly discuss emotions (Luo et al., 2013, 2014; Wang et al., 2000). Moreover, the comparatively lesser cognitive demand of CA child speech may reflect the lower EF seen among the CA children in our sample.

Considering previous findings of cross-cultural similarity in the frequency with which children respond to parental questions (Luo & Tamis-LeMonda, 2017) in conjunction with our finding of MA parents’ lower proportion of referential questions, one might expect us to have found MA children producing fewer utterances. Yet, we found no cultural difference in the number of story-related child utterances. One potential explanation for the seeming discrepancy is that cultural group differences exist in the frequency with which children respond to parental statements.

Regarding EF, we found evidence that utterance-level indicators of parent/child cognitive distancing language are associated with higher cognitive flexibility among preschool-age DLLs from low-income families. These findings are consistent with previous research incorporating alternative language complexity measures (e.g., Gueron-Sela et al., 2018; Kuhn et al., 2016). Though associations between multiple levels of cognitive distancing language and both EF subcomponents initially emerged, only relations between moderately distant (i.e., behavioral) utterances and cognitive flexibility remained significant after applying an FDR correction. Therefore, the findings are preliminary and need to be replicated in larger samples. Since children’s responses tend to occur at the same cognitive distancing level as the parental questions that elicited them (Luo & Tamis-LeMonda, 2017), it remains unclear whether child behavioral statements mediated the parent behavioral question – cognitive flexibility relation or parent behavioral questions mediated the child behavioral statement – cognitive flexibility relation. Taken together, we found weak support for the hypothesis that more distant parent/child language is associated with higher EF among DLLs from low-income families.

Importantly, our results indicate that certain types of distancing language (namely behavioral utterances) are positively associated with child EF in this age and sociodemographic group. They should not be interpreted as suggesting that distancing language more broadly is positively associated with child EF. Though the present study – the first to test direct links between book-sharing distancing language and child EF – provides evidence for an association between behavioral (moderate cognitive demand) utterances and child cognitive flexibility, more research is needed to examine whether cognitive distancing language is related to child EF more broadly.

The significant association between cognitive flexibility and parent behavioral questions (alongside the null parent statement relations) suggests that relations between parent language cognitive demand and child EF are contingent on utterances’ communicatory function. While previous studies have reported that parent questions and statements at similar cognitive demand levels are differentially predictive of other child outcomes (e.g., Tompkins et al., 2017), the present study is the first to show this pattern of results in relation to child EF. Further research is needed to examine whether parental cognitive distancing language is linked to growth in children’s EF using larger samples and longitudinal approaches.

Given the cross-sectional, correlational study design, the present study cannot test whether complex parental language prompts child EF development, whether children’s EF abilities prompt parents to complicate/simplify their speech, or both. To expand on the bidirectional interpretation, that both significant associations between EF and distancing language specifically involved behavioral utterances might suggest this utterance type is most cognitively age-appropriate for children in our sample according to the zone-of-proximal-development perspective (Vygotsky, 1978). As with previous assertions about the interactions between parent book-sharing speech and child development in other domains (e.g., Bailey et al., 2013; Mol et al., 2008), this interpretation would support the transactional conceptualization of the effects in which parents simultaneously modulate their speech complexity to better match children’s EF while also pushing EF development forward. Such a conceptualization aligns with previous book-sharing research showing that child word learning was aided by a scaffolding-type procedure in which adults asked low cognitive demand questions at the introduction of novel words and subsequently followed up with high cognitive demand questions later (Blewitt et al., 2009). Similarly, others have suggested EF subcomponents may relate to language dimensions differently at various developmental times (e.g., Friend & Bates, 2014). It may be the case our cross-sectional snapshot captured a moment during which relations between cognitive flexibility and language were most appreciable in our sample.

Study Limitations and Future Directions

Several limitations are worth discussing. First, we cannot determine the direction of the EF effects because of the study’s cross-sectional design. Our study’s cross-sectional design also prevents us from examining the degree to which parent-child book-sharing language is associated with age-related growth in EF, which deserves consideration due to the cognitive growth that occurs from ages 3 through 5 (e.g., Diamond, 2013). Future longitudinal work is needed to test the direction of relations between book-sharing distancing language and child EF as well as whether book-sharing language is associated with age-related growth in EF. Second, like most studies in this area, parent participants were almost all mothers. Research has shown that mothers’ and fathers’ book-sharing language is differentially associated with child outcomes (e.g., Baker et al., 2015). Due to the small number of fathers, we were unable to examine differences in book-sharing behaviors between mothers and fathers or test whether the relation between book sharing and child EF differed by family member. Third, due to our focus on MA and CA DLLs from low-income families, the findings from our study might not generalize to other cultural/socioeconomic groups in the U.S. Others should attempt to replicate the findings in samples composed of different demographics. Fourth, it remains unclear how book-sharing distancing language may relate to other EF subcomponents such as working memory. The generalizability of the present study may also be limited by our focus on a single book-sharing session involving one specific book. We were unable to evaluate how different types of books or repeated readings of the same book may impact results. Finally, the small sample size limited the study’s statistical power to test more complex relations.

Using a socioeconomically matched sample, the present study helps characterize similarities and differences in book-sharing language among the two largest DLL groups in the U.S. Our results provide preliminary evidence that moderately distant parent questions and moderately distant child statements might be associated with higher child cognitive flexibility. While the findings need to be replicated in future longitudinal and experimental research, future studies incorporating alternative book types and examining teacher-led book sharing can further broaden the implications and shed light on when, how, and why distancing language and EF are associated. Overall, this line of research can inform the development of interventions for promoting cognitive and language development in children from diverse socioeconomic and cultural backgrounds.