The Contingent Relationships Between Maternal and Child Cognitive Demand Contributions in Shared Book Reading Among Low-income Preschool-Age Dual Language Learners: A Sequential Analysis

Abstract

This study employs sequential analysis to explore the contingent dynamics of cognitive demand in shared book reading across 152 low-income immigrant Mexican American (M age = 48.98 months) and Chinese American (M age = 47.67 months) dyads. Utterances were coded for cognitive demand (yes/no, referential, behavioral, and inferential). While both groups matched the cognitive demand level of each other’s speech, group-specific deviations through the interactions of cultural background and maternal education revealed distinct scaffolding patterns. Mexican American children tended to undermatch inferential questions by reverting to yes/no answers, while Chinese American children showed matching responses across all question types, and their mothers employed mid-level referential and behavioral questions for follow-up. These patterns show that cultural and demographic variations impact the contingent structure of scaffolding, resulting in functionally different instances of overmatching and undermatching. These findings underscore the need for culturally sensitive interventions focused on the process of contingent language interactions.

Introduction

Dual language learners (DLLs), defined as young children acquiring a second language alongside their first language (Espinosa, 2013), make up one-third of all children ages 0 to 5 in the United States (Giang & Park, 2022). Shared book reading is one of the most common activities to characterize home literacy (Dexter & Stacks, 2014). It involves an adult sharing or reading a book with a child, either individually or in a group of children (Noble et al., 2019). Studies have shown that shared reading practices predict various language and literacy skills (e.g., Gámez et al., 2017; Wasik et al., 2016). Beyond the quantity of shared literacy activities, the complex parent-child interactions during these activities, such as child engagement, sustained attention, and parental positive regard, mediate the relationship between shared literacy activities and children’s literacy and language skills (Meng, 2021). Additionally, the nature of these interactions is profoundly shaped by the family’s cultural and socioeconomic backgrounds (Escobar et al., 2017; Luo & Tamis-LeMonda, 2017).

The current study focuses on low-income immigrant Mexican American and Chinese American dyads, the two largest immigrant communities in the United States (Budiman, 2020), to examine the cognitive demands of their interactions. Although shared book reading interactions can involve different caregivers, including fathers and grandparents (Noble et al., 2019), focusing specifically on mother-child dyads aligns with the foundational literature on shared book reading and controls for variation in interaction styles (Tamis-LeMonda et al., 2004). The concept of cognitive demand is grounded in Sigel’s (2002) Psychological Distancing Model, which posits that speakers adapt the cognitive complexity of their communication to the other speaker’s understanding. Based on this model, the present study utilized sequential analysis to investigate how mothers and children calibrate their contributions to one another during shared book reading interactions. Understanding the contingent relationships in parent-child shared book reading interactions within low-income Mexican American and Chinese American families is critical for identifying shared patterns and unique cultural variations. The findings can inform the development of culturally competent shared book reading interventions to support children’s language development from diverse backgrounds, without overgeneralizing middle- or upper-class, monolingual norms (Hoff, 2013; Neuman et al., 2018).

Shared Book Reading and Cultural Variations

Parent-child shared book reading styles among immigrant families in the United States vary by culture and individual characteristics (e.g., Kibler et al., 2020; Sawyer et al., 2018). These variations are rooted in differing cultural values, beliefs about child development and communication styles (van Kleeck, 2006). For instance, Melzi and Caspe (2005) identified two distinct parental narrative styles compared between Peruvian and American mothers. Extending these between-group comparisons, Luo and colleagues (2014) added critical nuance to the cross-cultural discussion by highlighting the diversity in narrative styles found across and within African American, Dominican, Mexican, and Chinese mothers. Their findings showed that these cultural groups are not monolithic. For example, African American mothers frequently emphasized individual goals and personal achievement, while Mexican mothers focused on providing more storyline details and engaging in a higher level of questioning than their Chinese and Dominican counterparts.

Parental questioning strategies, including the frequency, types, and purposes of questions, vary across cultures. It is critical to recognize that question frequency and contingent sequences are two distinct dimensions of parental input. Previous research has primarily focused on frequency (e.g., Cutler & Palkovitz, 2020; Zibulsky et al., 2018). For example, Luo and colleagues (2014) found that Mexican mothers asked relatively more questions than Chinese mothers, yet mothers of all groups predominantly narrated. However, questioning may be highly effective and beneficial even at a low frequency if the questions consistently and predictably elicit a child response.

Cultural values, beliefs, and expectations play a significant role in shaping parent-child shared book reading practices. Individualistic cultures (often reflected in European American samples) may emphasize child autonomy and knowledge acquisition through more direct questioning and explicit instruction. On the other hand, collectivistic cultures (often reflected in East Asian samples) may prioritize social harmony and interdependence, which could lead to more indirect communication and a stronger focus on shared meaning-making (Chiu & McBride-Chang, 2010). Specifically, Chinese pedagogical styles often involve intensive reading, recitation, and a strong focus on knowledge transmission (Anderson et al., 2012; Li & Fleer, 2015). This can translate to a focus on referential or low-level comprehension questions. Conversely, Mexican American families often prioritize socioemotional development and shared meaning-making (Salinas et al., 2017; Chan et al., 2022), which can lead to a focus on inferential statements and storytelling (Caspe, 2009). Recent studies have found that the level of cognitive demand during parent-child shared book reading differs between Mexican American and Chinese American families (Mak et al., 2024; Mauer et al., 2024). Luo and Tamis-LeMonda (2017) also noted differences in the cognitive level and contingency of the questions asked across low-income groups.

Additionally, these cultural interaction patterns are often shaped or mediated by socioeconomic background (Hoff, 2013). Families from low-income backgrounds may navigate unique stressors that influence the frequency or complexity of home literacy practices, such as high parenting stress or limited access to literacy materials (Bracken & Fischel, 2008; Neuman et al., 2018; Salinas et al., 2017). Therefore, examining low-income immigrant populations enables a more localized understanding of how these families leverage cultural resources or strengths (Salinas et al., 2017). Further research is needed to explore the contingent language dynamics, beyond simple frequency counts, of parent-child shared book reading interactions among low-income Mexican American and Chinese American families in order to move toward culturally appropriate interventions. Moreover, the specific cognitive nature of language used is also a critical driver of development in addition to the frequency of these interactions (e.g., Mak et al., 2024; Rowe & Snow, 2020).

Cognitive Demand

Sigel’s (2002) Psychological Distancing Model explains the cognitive demand of parent-child interactions. The model posits a hierarchy of cognitive demands, from Level 1 (minimal demands, such as labeling) to Level 3 (high demands, such as inference and generalization). This model also suggests that parents may calibrate the cognitive demand of their language when interacting with their children based on what their children can understand. Multiple terminologies, including inferential, decontextualized, or non-immediate talk, describe concepts that are not directly observable in pictures (e.g., van Kleeck, 2003). These languages are often challenging for children to comprehend, requiring more cognitive capacity, abstract thinking, and vocabulary knowledge (Snow et al., 2001). For this study, we use “inferential language” to represent the most cognitively demanding type of language used in mother-child interactions during shared book reading.

Research with primarily monolingual samples has shown that engaging children in inferential discussions encourages them to process story content at a deeper level and make connections to their lives, thus promoting vocabulary skills (Dunst et al., 2012; Reese et al., 2003). Children are also able to make inferences about story characters’ emotions and thoughts and predict future events that will happen in the story through these high cognitive demand conversations (e.g., Dunst et al., 2012). Engaging in rich conversational exchanges, including inferential discussion, can significantly enhance young children’s emergent literacy skills and vocabulary (Conica et al., 2023). These inferential discussions are particularly recommended as higher abstractions are labeled as quality language input that is developmentally appropriate and challenging for preschool children (Rowe & Snow, 2020). However, it is also crucial to acknowledge that not all question types are equally effective for all children. For instance, Luo and colleagues (2022) found that less cognitively demanding maternal questions, such as yes/no and referential questions, positively predicted receptive vocabulary in toddlers from English-speaking families. This suggests that the relationship between cognitive demand and language outcomes is not always linear and may vary based on the child’s linguistic background (Hoff, 2006).

The impact of question types on monolingual children may not be the same for DLLs, depending on the context. While inferential questions during shared reading have been shown to positively predict heritage language vocabulary development in some dual language learners (Rydland & Grøver, 2023), other research suggests that contextualized questions may be more effective for enhancing word meaning and story comprehension in the heritage language, particularly when cognitive demand is a factor (Sun & Batra, 2024).

Recent studies have specifically examined cognitive demand in parent-child shared book reading among Mexican American and Chinese American families. Mak and colleagues (2024) investigated the relations among cognitive demand in shared book reading, children’s language output, receptive vocabulary, and expressive vocabulary outcomes in the home language of DLLs and found that parents’ use of cognitive demanding language was a significant predictor. Moreover, Mauer and colleagues (2024) explored the differences in cognitive distancing language in parent-child shared book reading and its links to children’s executive functions. These studies highlight the importance of considering cognitive demand when examining parent-child interactions during shared book reading in these cultural groups. Understanding this relationship between cognitive demand, developmental stage, and bilingual experience is essential for optimizing the benefits of these discussions for children of different ages.

It is also important to consider that both statements and questions facilitate language development through different mechanisms. Studies support the idea that questions that elicit children’s responses predict vocabulary and narrative skills (e.g., Cristofaro & Tamis-LeMonda, 2012; Walsh & Rose, 2013), while other studies demonstrate that statements can significantly predict vocabulary growth (e.g., Barnes et al., 2016). Since both forms of discourse are commonly observed and influence outcomes (Ard & Beverly, 2004), this study investigated the cognitive demand levels of statements and questions to provide a more comprehensive view of the shared book reading interactions.

Sequential Analysis

As suggested by the Psychological Distancing Model (Sigel, 2002), parents may adjust the cognitive demands of their language to their children’s language skills and comprehension abilities. This process is fundamentally aligned with Vygotsky’s (1978) social interaction theory, which describes scaffolding to support learner function within their Zone of Proximal Development (ZPD). This study used sequential analysis to investigate such contingent communication mechanisms and to identify cultural variations in scaffolding during shared book reading. Vygotsky’s theory posits that a speaker’s point of view is influenced by the listener’s point of view. Therefore, the cognitive demand of parents’ and children’s speech might be reciprocally related (e.g., Luo & Tamis-LeMonda, 2017; Tompkins et al., 2017).

Sequential analysis is ideal for this purpose, as it can inform us whether the likelihood of a specific parent-child conversation sequence (Lag 1) was more significant than by chance alone (Bakeman & Gottman, 1997). Previous sequential studies confirmed contingent relationships in adult and child discourse (e.g., Luo & Tamis-LeMonda, 2017; Tompkins et al., 2017). Tompkins and colleagues’ (2017) study demonstrated that children could respond to their mother’s inferential questions and statements at the same level of abstraction. In the only prior study that included both Mexican American and Chinese American families, Luo and Tamis-LeMonda (2017) found a significant sequential association between maternal questions and child responses. They observed cultural variations in which Mexican American children were more likely to respond to maternal inferential questions with low-level yes/no answers, whereas Chinese American children were more likely to match. These findings suggest that although there are some common patterns, there are important cultural variations in the contingencies of their interactions, including the moment-to-moment process of scaffolding, which warrants deeper investigation.

The Present Study

The extant literature demonstrates the potential contingent relationships between adult and child discourse during shared book reading. Given the benefits of engaging in higher levels of cognitive demand talk for their language development (Sun & Batra, 2024), understanding the specific sequential patterns in low-income DLL families is essential. The current study aims to determine whether DLLs’ responses would match their mothers’ cognitive distancing level and whether their mothers would adapt the level of their questions or statements to match the level of their children’s previous contributions, focusing on low-income immigrant Mexican American and Chinese American groups.

The first research question asks: What is the pattern of utterances (statements and questions) regarding cognitive demand level produced by mothers and DLL children in both the Mexican American and Chinese American groups? It was hypothesized that the DLLs would produce fewer inferential utterances than the lower levels of cognitive distancing utterances (Luo & Tamis-LeMonda, 2017). It was also hypothesized that mothers would produce more statements than questions because they were being instructed to tell a story and that there would be differences in the types of maternal utterances between the two cultural groups (Chen et al., 2022; Luo et al., 2014).

The second research question asks: What are the sequential associations and magnitude of effect of the mother-child sequences (child’s speech follows mother’s speech) and the child-mother sequence (mother’s speech follows child’s speech), including both statements and questions, in both the Mexican American and Chinese American groups? It was hypothesized that mother-child sequences (specifically maternal questions to child statements) would have a positive and more substantial magnitude of effect than child-mother sequences (Justice et al., 2002; Luo & Tamis-LeMonda, 2017), as the mothers in this study were instructed to lead the storytelling.

The third research question asks: What are the sequential associations among different cognitive demand levels of maternal and child utterances (statements and questions)? It was hypothesized that the cognitive demand level of the child’s utterances would be contingent on the cognitive demand level of the mother’s utterances and vice versa, but that there might be small to moderate sequential associations between nonmatching cognitive demand levels of mother’s and child’s utterances (Luo & Tamis-LeMonda, 2017; Tompkins et al., 2017). Such nonmatching exchanges are theoretically grounded in the concept of scaffolding; the mother may intentionally employ overmatching or undermatching strategies to stay within their child’s ZPD (Sigel, 2002; Vygotsky, 1978). Further, the Psychological Distancing Model (Sigel, 2002) suggests that child age may influence language use. Therefore, we also hypothesized that older children would be more likely to match their mother’s higher level of cognitive demanding languages. Specifically, we hypothesized that Mexican American children would show a distinct pattern of undermatching for inferential questions, a pattern less pronounced in the Chinese American group (Luo & Tamis-LeMonda, 2017). This hypothesis is grounded in cultural communication styles and socialization practices, in which Mexican American families often prioritize narrative flow and shared socioemotional experiences over direct instruction or knowledge testing (Caspe, 2009; Salinas et al., 2017). Mexican American children may respond to high cognitive demand level questions with simpler, low-level statements to maintain the conversation momentum, whereas Chinese American children may be more motivated to match the cognitive demand level of the mother’s prompts based on their cultural socialization goals, which often emphasize knowledge accuracy (Li & Fleer, 2015).

Materials and Methods

Participants

A total of 194 DLL children and their families (69 Mexican Americans and 125 Chinese Americans) were recruited from Head Start centers and state-funded preschools in Northern California in the fall of 2018. To be eligible for the current study, children had to (1) be between 36 and 59 months of age, (2) be enrolled in a Head Start or state-funded preschool programs that used English as one of the instructional languages for at least three days per week, (3) be living with at least one biological parent for at least 50% of time, (4) both biological parents identified as either ethnic Mexican or ethnic Chinese, (5) be exposed to Spanish or Chinese (Cantonese or Mandarin) at home, and (6) do not have language delays (or are not currently receiving speech and language services). Data were collected from fall 2018 to fall 2019. The descriptive statistics of the sample demographic information are presented in Table 1.

Table 1.

Descriptive Statistics of Study Variables

	Mexican Americans (n = 67)				Chinese Americans (n = 85)				Group Difference
	M	SD	Min	Max	M	SD	Min	Max
Demographic
Child Age (months)	48.98	6.66	36	63	47.67	7.51	36	61	U = 3102.0
Maternal Educational Year	11.32	2.84	0	16	12.68	2.54	8	18	U = 2021.0**
Per Capita Income	7345.24	4305.25	625	24166.67	8866.96	5563.23	1000	32500	U = 1862.0
Maternal Utterances
Yes/No Statements	0.63 (1%)	1.50	0	10	2.08 (2%)	2.08	0	9	U = 1487.5***
Referential Statements	14.61 (15%)	9.83	3	40	17.87 (15%)	8.62	3	44	U = 2531.0
Behavioral Statements	33.28 (38%)	12.92	9	70	42.22 (37%)	17.51	13	92	t = 0.57
Inferential Statements	30.97 (32%)	19.50	4	100	15.10 (13%)	8.18	0	35	U = 5110.5***
Yes/No Questions	4.36 (4%)	6.66	0	46	9.02 (8%)	6.44	1	35	U = 1438.0***
Referential Questions	5.37 (5%)	7.34	0	27	20.64 (19%)	10.30	1	57	U = 525.0***
Behavioral Questions	2.37 (2%)	4.56	0	24	2.77 (3%)	3.00	0	17	U = 2074.5**
Inferential Questions	2.61 (2%)	5.20	0	27	4.76 (4%)	4.44	0	20	U = 1656.5***
Child Utterances
Yes/No Statements	1.06 (6%)	2.45	0	17	4.43 (18%)	5.29	0	28	U = 1159.0***
Referential Statements	4.76 (34%)	5.12	0	20	9.87 (43%)	8.05	0	37	U = 1904.0**
Behavioral Statements	2.33 (17%)	3.30	0	16	4.72 (19%)	4.90	0	23	U = 1978.0*
Inferential Statements	3.01 (23%)	3.94	0	22	2.41 (9%)	2.80	0	15	U = 3468.0***
Yes/No Questions	1.16 (7%)	2.20	0	15	0.16 (1%)	0.43	0	2	U = 3415.0***
Referential Questions	0.78 (5%)	1.43	0	7	1.36 (6%)	2.10	0	14	U = 2158.5
Behavioral Questions	0.09 (1%)	0.34	0	2	0.22 (1%)	0.54	0	2	U = 2319.0
Inferential Questions	0.84 (5%)	2.21	0	11	0.65 (2%)	1.41	0	9	U = 2663.5

Note. M = mean, SD = standard deviation. The proportions of each coding category in relation to the total number of utterances of the given speaker are presented in parentheses. Group differences statistics for utterances were examined using the proportion of each coding category.

^*p < .05

^**p < .01

^***p < .001.

Participants were recruited during school drop-off and pick-up times as well as during parent meetings. Eligible families completed a home visit lasting two to three hours with bilingual research assistants. During the home visit, parents completed questionnaires in their preferred language, and children participated in assessment tasks related to socio-emotional development, executive function, and bilingual language skills (English and Spanish or Chinese). The present study used a subsample of the larger dataset with participants who had completed both the parent-child shared book reading activity and the parent questionnaires containing demographic information. The final sample size for the present study comprised 152 DLL children and their mothers (67 Mexican Americans and 85 Chinese Americans).

Procedure

During the home visit, the mother was instructed to tell a story to the child based on the wordless picture book, Frog, Where Are You?, by Meyer (1969), using any language they preferred in five minutes. This book was selected to allow mothers the freedom to share it in whichever way and language(s) they felt comfortable with, without being constrained by a specific written language. Beyond the linguistic flexibility, this book has a clear episodic structure that centers on the problem of losing a frog, a goal-directed search across various settings, and a resolution in which the boy finds that his frog has started a family, leading him to leave with a new baby frog. This specific narrative arc is conducive to eliciting diverse cognitive and linguistic behaviors, ranging from labeling to more complex causal inference (Heilmann et al., 2016). Although the book's familiarity was not formally measured, it is a classic research tool selected for its general novelty to families in this demographic; no parents reported prior exposure to the book during the home visit sessions. Furthermore, this book has been previously used with both Chinese- and Spanish-speaking families (e.g., Heilmann et al., 2016; Leung, 2012; Wu, 2022). Mothers were allowed to spend up to ten minutes finishing the story. This activity was audio recorded.

Spanish- or Chinese-speaking research assistants transcribed the book reading session using the audio files following the Codes for Human Analysis of Transcripts (CHAT) conventions from the Child Language Data Exchange System (CHILDES; MacWhinney, 2000). A second transcriber verified each transcription to ensure that the transcriptions were done accurately. One trained Spanish-speaking and one trained Chinese-speaking research assistant coded each mother and child utterances in the transcriptions using a coding scheme adapted from Luo and Tamis-LeMonda (2017). The four levels of cognitive distancing were yes/no, referential, behavioral, or inferential. These codes were mutually exclusive and exhaustive (see Appendix for the definitions and examples of the coding categories). Each utterance from the mother and the child was coded as yes/no, referential, behavior, or inferential. To maintain interrater reliability, the two research assistants coded 20% of all the randomly selected transcriptions translated into English. The overall Cohen’s kappa for coding of both Mexican American and Chinese American groups was .726 (Mexican American = .77; Chinese American = .706), demonstrating satisfactory interrater reliability. They then proceeded to code the transcriptions independently in the original language in Spanish or Cantonese, with regular meetings held throughout the coding process to prevent drift and ensure the coding definitions were applied consistently from the first to the final transcription.

Data Analysis

The frequency and event-based percentage of each cognitive distancing category occurrence were reported based on the total number of story-related utterances. The event-based percentage is the proportion of utterances in each cognitive distancing category (e.g., referential) relative to the total number of utterances produced across all questions or statements (e.g., all maternal questions). This percentage allows comparison of the qualitative distribution of talk while controlling for the individual's overall talkativeness. The normality of each variable was checked with the Shapiro-Wilk test. The maternal and child statement and question variables in all cognitive distancing categories for the Mexican American group were positively skewed, with kurtosis ranging from −0.30 to 24.18. For the Chinese American group, all cognitive distancing categories of statements and questions for both the mothers and children were positively skewed, with kurtosis ranging from −0.20 to 13.56. The majority of the variables were non-normally distributed, except for maternal behavioral statement proportions, so Mann-Whitney U tests were used to examine group differences for non-normally distributed variables, and an independent samples t-test was used for normally distributed variables. To answer the first research question, descriptive statistics for all cognitive demand levels of utterance from the mothers and children were reported separately for the Mexican American and Chinese American participants.

Sequential analysis was used to answer the second and third research questions by examining the bidirectional relationships between maternal and child contributions to the story using the generalized sequential querier (GSEQ; Bakeman & Quera, 1995). All maternal and child utterances were coded as event-sequential data and pooled across all participants into GSEQ for analysis. Yule’s Q, adjusted residuals, expected values, and odds ratios Q were calculated. The adjusted residual assumes that the two-code sequences occur independently and indicates that the probabilities are greater or less than chance, similar to a z-score (Bakeman & Gottman, 1997). Adjusted residuals greater than ±1.96 when the alpha level is at .05 would be considered significant. Yule’s Q (ranging from −1.0 to +1.0) demonstrated the magnitude of the effect of the sequential association of different levels of utterance pairs (0 to ±.29 is small; ±.30 to ±.49 is moderate; ±.50 to ±.69 is large; ± .70 to ±1.0 is very large; Bernard, 2000). In the Psychological Distancing Model (Sigel, 2002), these conventional benchmarks are interpreted in light of the theoretical context of scaffolding and the conversational nature of shared book reading, where even moderate sequential associations may indicate meaningful pedagogical scaffolding (Tompkins et al., 2017; Vygotsky, 1978).

The second research question was answered by examining the sequential associations between statements and questions of the mothers and children using the lag 1 sequential method. Lag 1 refers to the probability of a target event occurring immediately after a given event (Bakeman & Gottman, 1997). The lag 1 transitional probabilities of eight two-code sequences were calculated for each group and tested for significance using adjusted residuals and Yule’s Q. The eight two-code sequences include maternal statement to child statement, maternal statement to child question, maternal question to child statement, maternal question to child question, child statement to maternal statement, child statement to maternal question, child question to maternal statement, and child question to maternal question.

To answer the third research question regarding whether there was any systematic sequencing of a target utterance that followed a given type of utterance in terms of cognitive demand level, the lag 1 sequential method was used in both the Mexican American and Chinese American groups. The analysis of both cultural groups consisted of 64 two-code sequences, divided into 32 mother-to-child sequences and 32 child-to-mother sequences. For example, given an event of a maternal yes/no question, the lag 1 transitional probabilities for child yes/no statement, child referential statement, child behavioral statement, child inferential statement, child yes/no question, child referential question, child behavioral question, and child inferential question were computed and tested to determine if each probability significantly exceeded its expected value using sequential analysis. However, due to the low frequency of Mexican American child behavioral questions, Yule’s Q and odd ratios for their sequences with other cognitive demand levels were statistically impossible to calculate. Therefore, a single level of “child advanced questions,” combining child behavioral questions and child inferential questions from the Mexican American group, was used for additional sequential analyses. Additionally, each cultural group was divided into younger and older children based on the median age (Mexican American = 49 months; Chinese American = 48 months). This allowed us to explore if the sequential associations between mother and child speech across different cognitive demand levels were influenced by child age. These child-age subgroup analyses were exploratory; formal statistical tests for significant differences between the younger and older groups were not conducted.

Results

Table 1 displays the descriptive statistics for demographic information and the descriptive statistics for all cognitive demand levels of maternal and child utterances. The Mann-Whitney U tests revealed no significant differences in child age or per capita income between the Mexican American and Chinese American groups. However, maternal year of education was different between the two groups (U = 2021.0, p = .006, r = .225), where the Chinese American group reported higher levels of education.

Utterance Patterns

The relative event-based proportions of each cognitive demand level were compared between the Chinese American and Mexican American groups to control for the total number of utterances. On average, the Chinese American mothers produced significantly more yes/no statements than the Mexican American mothers during shared book reading (U = 1487.5, p < .001, r = .425). Mexican American mothers produced more inferential statements than Chinese American mothers (U = 5110.5, p < .001, r = .700). Chinese American mothers asked significantly more yes/no questions (U = 1438.0, p < .001, r = .420), referential questions (U = 525.0, p = .005, r = .230), and inferential questions (U = 1656.5, p < .001, r = .356) than Mexican American mothers. No group differences were detected for maternal referential or behavioral statements.

Further, examining the relative event-based proportions of each cognitive demand level within maternal utterances showed that mothers from both groups elicited more statements than questions. Mexican American mothers dedicated the most time to providing behavioral statements (38%) and inferential statements (32%), whereas Chinese American mothers focused on behavioral statements (37%), referential questions (19%), and referential statements (15%).

In terms of relative event-based proportions of different child utterances, the Mexican American children asked significantly more yes/no questions (U = 3415.0, p < .001, r = .367) and produced more inferential statements (U = 3468.0, p < .001, r = .311) than the Chinese American children. In contrast, the Chinese American children produced more yes/no statements (U = 1159.5, p < .001, r = .477), referential statements (U = 1904.0, p = .009, r = .217), and behavioral statements (U = 1978.0, p = .020, r = .194). Similar to their mothers, children in both groups produced more statements than questions. Among children’s contributions during shared book reading, Mexican American children contributed the most by providing referential statements (34%) and inferential statements (23%). In contrast, Chinese American children contributed the most by providing referential statements (43%), behavioral statements (19%), and yes/no statements (18%).

Mother-Child and Child-Mother Sequences in Questions and Statements

Among Mexican American mother-child dyads, the sequential analysis showed significant sequential associations between six pairs of two-code sequences of questions and statements (see Table 2). Specifically, positive and robust sequential associations were observed between maternal questions and child statements, maternal statements and child questions, child questions and maternal statements, and child statements and maternal questions. Among these sequential associations, maternal questions to child statements had the strongest magnitude (Yule’s Q = .73), followed by child questions to maternal statements (Yule’s Q = .39), child statements to maternal questions (Yule’s Q = .31), and maternal statements to child questions (Yule’s Q = .29). Further, the likelihood of a child statement to a maternal statement (Yule’s Q = −.19) and a maternal statement to a child statement (Yule’s Q = −.58) were significantly lower than chance, as evidenced by the negative Yule’s Q values. In other words, Mexican American children were more likely than chance to respond to maternal questions and ask questions following maternal statements, while Mexican mothers were likely to ask follow-up questions and answer their children’s questions.

Table 2.

Two-Code Sequential Associations of Statements and Questions Across Cultural Groups

Sequences	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio
Mexican Americans
Maternal questions → Child statements	0.73	22.17***	269/87.11	6.33
Maternal questions → Child questions	−0.12	−0.94	17/20.96	0.78
Maternal statements → Child statements	−0.58	−16.43***	317/492.22	0.27
Maternal statements → Child questions	0.29	2.85***	134/118.43	1.83
Child questions → Maternal statements	0.39	3.77**	142/120.84	2.30
Child questions → Maternal questions	−0.20	−1.53	15/21.50	0.66
Child statements → Maternal statements	−0.19	−4.15***	403/446.01	0.69
Child statements → Maternal questions	0.31	5.92***	126/79.37	1.88
Chinese Americans
Maternal questions → Child statements	0.59	25.63***	864/439.60	3.85
Maternal questions → Child questions	0.18	2.25*	58/45.06	1.44
Maternal statements → Child statements	−0.35	−13.83***	709/969.09	0.48
Maternal statements → Child questions	0.10	1.33	108/99.34	1.23
Child questions → Maternal statements	0.12	1.64	118/106.86	1.28
Child questions → Maternal questions	0.08	0.96	55/49.27	1.17
Child statements → Maternal statements	0.12	4.48***	1050/965.69	1.27
Child statements → Maternal questions	0.15	5.10***	530/445.26	1.34

^*p < .05

^**p < .01

^***p < .001

Among the Chinese American mother-child dyads, sequential analyses revealed significant associations in five pairs of mother-child sequences of questions and statements (see Table 2). Positive sequential associations were observed between maternal questions and child statements, maternal questions and child questions, child statements and maternal statements, and child statements and maternal questions. The magnitude of the effect from maternal questions to child statements was the strongest (Yule’s Q = .59), followed by maternal questions to child questions (Yule’s Q = .18), child statements to maternal questions (Yule’s Q = .15), and child statements to maternal statements (Yule’s Q = .12). Moreover, the likelihood of the sequence between maternal statements and child statements was significantly lower than chance (Yule’s Q = −.35). In other words, Chinese American children were likely than chance to respond to mothers’ questions, either by providing statements or asking further questions. Chinese mothers were also likely to ask follow-up questions and provide statements after their children’s statements.

Mother-Child and Child-Mother Sequences in Cognitive Demand Levels

Among the Mexican American dyads, the sequential analysis revealed significant sequential associations for seven of the 32 mother-to-child sequences in cognitive demand levels (see Table 3). Specifically, positive and strong sequences with matching cognitive demand levels were observed. They included maternal yes/no questions to child yes/no statements (Yule’s Q = .99), maternal referential questions to child referential statements (Yule’s Q = .81), and material behavioral questions to child behavioral statements (Yule’s Q = .91). However, one non-matching sequence, maternal inferential questions to child yes/no statements (Yule’s Q = .60), showed a moderately strong association. Additionally, positive sequences included maternal referential statements to child yes/no questions (Yule’s Q = .41) and maternal yes/no statements to child referential questions (Yule’s Q = .82).

Table 3.

Two-Code Sequential Associations of Statements and Questions in Cognitive Demand Levels Among Mexican Americans

Given Events: Maternal Questions	Target Events: Child Statements
	Yes/no				Referential				Behavioral				Inferential
	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio
Yes/no	0.99	36.28**	61/2.703	228.58	−0.4	−1.90	5/11.04	0.43	0.16	0.83	7/5.17	1.38	0.02	0.10	7/6.76	1.04
Referential	−1.00	−1.91	0/3.41	0	0.81	17.74**	77/13.95	9.39	0.52	4.65**	18/6.54	3.16	0.27	1.95	14/8.53	1.73
Behavioral	−1.00	−1.20	0/1.40	0	−0.19	−0.74	4/5.71	0.69	0.91	18.31**	32/2.68	20.63	0.07	0.28	4/3.50	1.15
Inferential	0.60	3.44**	6/1.65	4.00	0.02	0.10	7/6.76	1.04	−0.23	−0.67	2/3.17	0.62	0.78	10.54**	25/4.14	8.14
Given Events: Maternal Statements	Target Events: Child Questions
Yes/no	0.44	0.97	1/0.40	2.59	0.82	3.80**	2/0.22	9.99	—	−0.15	0/0.02	—	−1.00	−0.54	0/0.29	0.00
Referential	0.41	3.22**	18/9.04	2.38	0.19	0.93	7/5.05	1.48	—	−0.78	0/0.53	—	0.03	0.15	7/6.65	1.06
Behavioral	−0.07	−0.52	19/20.96	0.87	0.08	0.45	13/11.71	1.17	—	−0.25	1/1.23	—	−0.17	−1.05	12/15.41	0.71
Inferential	0.04	0.29	21/19.93	1.08	0.05	0.31	12/11.14	1.11	—	−0.19	1/1.17	—	0.24	1.67	20/14.66	1.61
Given Events: Child Questions	Target Events: Maternal Statements
Yes/no	1.00	44.51**	32/0.51	462.44	0.09	0.57	11/9.39	1.20	−0.06	−0.47	20/21.82	0.88	−0.60	−3.80**	7/21.74	0.25
Referential	−1.00	−0.55	0/0.30	0	0.61	4.78**	16/5.53	4.08	−0.05	−0.29	12/12.85	0.91	−0.37	−1.94	7/12.80	0.46
Behavioral	—	−0.17	0/0.03	—	—	−0.77	0/0.52	—	—	0.88	2/1.20	—	—	−1.30	0/1.19	—
Inferential	0.88	6.62**	4/0.32	15.21	−0.73	−2.14*	1/5.79	0.15	−0.55	−2.76*	5/13.45	0.29	0.50	3.80***	25/13.40	2.98
Given Events: Child Statements	Target Events: Maternal Questions
Yes/no	0.72	6.03**	11/2.23	6.16	0.22	0.85	4/2.66	1.55	0.37	1.09	2/0.95	2.16	0.40	1.46	3/1/34	2.34
Referential	0.25	2.02*	17/10.68	1.68	0.22	1.84	19/12.74	1.57	−1.00	−2.21*	0/4.57	0.00	0.12	0.65	8/6.42	1.27
Behavioral	0.17	0.85	7/5.14	1.40	0.21	1.20	9/6.13	1.52	−1.00	−1.51	0/2.20	0.00	−0.22	−0.63	2/3.09	0.64
Inferential	0.58	5.81**	21/6.59	3.75	0.01	0.05	8/7.87	1.02	0.18	0.72	4/2.82	1.45	0.51	3.63**	11/3.96	3.06

“—” indicates that calculations of Yule’s Q and odds ratios were impossible due to low occurrence of events

^*p < .05

^**p < .01

^***p < .001.

Child-to-mother sequences also showed significant sequential associations, with 12 out of 32 sequences showing positive contingencies (see Table 3). These sequential associations involved children’s questions followed by matching cognitive demand levels in mothers’ responses. They included child yes/no questions to maternal yes/no statements (Yule’s Q = 1.00), child referential questions to maternal referential statements (Yule’s Q = .61), child inferential questions to maternal inferential statements (Yule’s Q = .50), and child advanced questions, combining child behavioral and inferential questions, to maternal inferential statements (Yule’s Q = .42). The sequence from child inferential questions to maternal yes/no statements exhibited a strong positive association (Yule’s Q = .88). Moreover, the sequential associations of child advanced questions leading to maternal yes/no statements remained strong (Yule’s Q = .87).

Moreover, the likelihood of various maternal statements following child questions was significantly lower than chance. Three types of sequences of child questions to maternal statements showed negative sequential relationships, which means that these sequences were unlikely to occur. These sequences included child inferential questions to maternal referential statements (Yule’s Q = −.73), child inferential questions to maternal behavioral statements (Yule’s Q = −.55), and child yes/no questions to maternal inferential statements (Yule’s Q = −.60). Further, two additional positive contingent sequences at matching levels were found between child statements and maternal questions: child yes/no statements to maternal yes/no questions (Yule’s Q = .72) and child inferential statements to maternal inferential questions (Yule’s Q = .51). The sequence of child referential statements to maternal yes/no questions (Yule’s Q = .25) and child inferential statements to maternal yes/no questions was positive (Yule’s Q = .58). However, the likelihood of child referential statements to maternal behavioral questions was significantly lower than chance (Yule’s Q = −1.00).

Among the Chinese American mother-child dyads, the sequential analysis results in Table 4 revealed six significant sequential associations across 32 mother-to-child sequences at varying cognitive demand levels. All six sequential relationships involved maternal questions followed by child statements. Four of these demonstrated matching cognitive demand levels, exemplified by maternal yes/no questions to child yes/no statements (Yule’s Q = .73), maternal referential questions to child referential statements (Yule’s Q = .70), maternal behavioral questions to child behavioral statements (Yule’s Q = .84), and maternal inferential questions to child inferential statements (Yule’s Q = .80). All demonstrated substantial magnitudes of effects. The remaining two sequences involved non-matching levels: maternal referential questions to child behavioral statements (Yule’s Q = .18) and maternal inferential questions to child referential statements (Yule’s Q = .28).

Table 4.

Two-Code Sequential Associations of Statements and Questions in Cognitive Demand Levels Among Chinese Americans

Given Events: Maternal Questions	Target Events: Child Statements
	Yes/no				Referential				Behavioral				Inferential
	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio	Yule’s Q	Adjusted Residual	Observed Frequency/Expected Frequency	Odds Ratio
Yes/no	0.73	16.57**	97/22.37	6.34	−0.07	−0.87	42/47.57	0.87	−0.16	−1.30	17/22.93	0.72	−0.11	−0.71	10/12.38	0.79
Referential	0.02	0.32	54/51.93	1.05	0.70	24.40**	339/110.41	5.74	0.28	4.64**	84/53.21	1.79	0.12	1.27	35/28.74	1.27
Behavioral	−0.30	−1.23	4/7.22	0.54	−0.13	−0.90	12/15.35	0.77	0.84	18.76**	57/7.40	11.86	−0.15	−0.51	3/3.99	0.74
Inferential	0.19	1.52	17/11.93	1.47	0.18	2.02*	35/25.36	1.44	0.11	0.82	15/12.22	1.25	0.80	14.54**	43/6.60	9.00
Given Events: Maternal Statements	Target Events: Child Questions
Yes/no	−1.00	−0.39	0/0.15	0.00	0.34	1.23	3/1.51	2.04	−1.00	−0.46	0/0.21	0.00	−1.00	−0.87	0/0.74	0.00
Referential	−0.16	−0.30	1/1.32	0.73	−0.12	−0.74	11/13.50	0.79	−1.00	−1.46	0/1.85	0.00	0.24	1.41	10/6.62	1.64
Behavioral	−0.28	−0.74	2/3.08	0.56	0.12	1.21	37/31.40	1.28	0.11	0.40	5/4.31	1.25	0.20	1.42	20/15.39	1.50
Inferential	0.34	0.92	2/1.09	2.04	−0.06	−0.37	10/11.16	0.88	−0.23	−0.46	1/1.53	0.63	0.05	0.24	6/5.47	1.11
Given Events: Child Questions	Target Events: Maternal Statements
Yes/no	0.85	4.24**	2/0.19	12.68	−0.28	−0.56	1/1.68	0.56	−0.42	−1.19	2/3.97	0.41	0.18	0.46	2/1.47	1.43
Referential	0.55	2.81**	5/1.55	3.42	0.15	1.16	18/13.98	1.35	0.17	1.68	41/33.00	1.40	−0.45	−2.20*	5/12.21	0.38
Behavioral	−1.00	−0.50	0/0.24	0.00	−1.00	−1.59	0/2.20	0.00	0.69	3.59**	12/5.19	5.47	−0.34	−0.71	1/1.92	0.49
Inferential	0.47	1.47	2/0.74	2.78	−0.58	−1.96*	2/6.73	0.27	−0.42	−2.38*	8/15.89	0.41	0.59	4.88***	17/5.89	3.85
Given Events: Child Statements	Target Events: Maternal Questions
Yes/no	0.23	2.58*	34/22.38	1.61	0.21	3.16***	69/48.89	1.54	0.49	4.15**	16/6.04	2.89	0.04	0.25	13/12.15	1.08
Referential	−0.35	−3.69**	27/51.71	0.49	0.21	4.65***	157/112.95	1.54	−0.18	−1.11	10/13.96	0.70	0.02	0.19	29/28.07	1.04
Behavioral	−0.29	−2.17*	14/24.13	0.55	0.21	3.22***	74/52.72	1.53	0.17	1.00	9/6.52	1.41	−0.09	−0.60	11/13.10	0.83
Inferential	−0.32	−1.78	7/13.17	0.51	0.00	0.05	29/28.77	1.01	−0.09	−0.30	3/3.56	0.84	0.65	8.02**	28/7.15	4.66

^*p < .05

^**p < .01

^***p < .001.

Child-to-mother sequences also displayed significant reciprocal patterns in 15 of 32 instances (see Table 4). Seven of these involved child questions leading to maternal statements. Three positive sequences matched cognitive demand levels: child yes/no questions to maternal yes/no statements (Yule’s Q = .85), child behavioral questions to maternal behavioral statements (Yule’s Q = .69), and child inferential questions to maternal inferential statements (Yule’s Q = .59). Additionally, the sequence from child referential questions to maternal yes/no statements showed a positive association with a large magnitude (Yule’s Q = .55). Three sequences with non-matching cognitive demand levels showed negative associations, indicating a lower likelihood of occurrence than chance: child inferential questions to maternal referential statements (Yule’s Q = −.58), child inferential questions to maternal behavioral statements (Yule’s Q = −.42), and child referential questions to maternal inferential statements (Yule’s Q = −.45).

Eight significant contingent sequences were identified for child statements to maternal questions. Six showed positive associations with small to moderate magnitude of effects. Three of these matched cognitive demand levels: child yes/no statements to maternal yes/no questions (Yule’s Q = .23), child referential statements to maternal referential questions (Yule’s Q = .21), and child inferential statements to maternal inferential questions (Yule’s Q = .65). Other positive associations included child yes/no statements to maternal referential questions (Yule’s Q = .21), child behavioral statements to maternal referential questions (Yule’s Q = .21), and child yes/no statements to maternal behavioral questions (Yule’s Q = .49). Further, the likelihood of both maternal yes/no statements following child referential statements (Yule’s Q = −.35), child behavioral statements (Yule’s Q = −.29), and child inferential statements (Yule’s Q = −.32) was significantly lower than chance.

Effects of Child Age

The effects of child age were examined to test the hypothesis that older children would be more likely to engage in and match higher levels of cognitive demand talk, as suggested by the Psychological Distancing Model (Sigel, 2002). Children in each group were categorized into younger and older cohorts based on the media age split. The median age for Mexican American children was 49 months. Younger children (n = 32) ranged from 36 to 48 months, while the older children (n = 35) ranged from 49 to 63 months. Certain significant contingencies emerged only among older Mexican American children. The sequences involving maternal inferential statements to child inferential questions (Yule’s Q = .42, z = 2.34), maternal inferential statements to child advanced questions (Yule’s Q = .36, z = 2.02), maternal referential questions to child inferential statements (Yule’s Q = .53, z = 3.69), and child yes/no statements to maternal inferential questions (Yule’s Q = .65, z = 2.73) were moderately strong and positive only for the older children.

The median age for Chinese American children was 48 months. The younger group (n = 40) ranged from 36 to 47 months, and the older group (n = 45) ranged from 48 to 61 months. Among the older children, the sequence of child inferential questions to maternal yes/no statements showed a moderately strong and positive association (Yule’s Q = .61, z = 2.10). Further, the younger group displayed a moderately strong and positive association for the sequence of maternal behavioral statements to child inferential questions (Yule’s Q = .61, z = 2.79).

Discussions

This study examined trends in mother-child shared book reading, focusing on the levels of cognitive demand among low-income immigrant Mexican American and Chinese American families. The study used sequential analysis to examine the relationships between mother and child utterances, focused on cognitive demand in both cultural groups, and compared how mothers and preschool-age DLL children interact during shared reading. The results revealed unique and shared practices that challenge the notion of a monolithic DLL experience and confirmed that cultural background strongly influences the dynamic processes of language interactions during shared book reading.

Overall Talk Patterns and Cultural Priorities

While Chinese American mothers offered more narrative statements and inquiries than Mexican American mothers, the overall pattern of their interactions was quite similar. This finding aligns with prior research showing that mothers often prioritize storytelling over questioning during shared book reading (Luo et al, 2014; Mak et al., 2024), which suggests a universal baseline for narrative engagement. Consistent with the findings in Luo and Tamis-LeMonda (2017) and Mak and colleagues (2024), Chinese American mothers in this study asked more questions than Mexican American mothers. Both groups in this study displayed generally similar distributions across different cognitive demand levels in their statements and questions, except for Mexican American mothers, who provided more inferential statements, and Chinese American mothers, who favored referential queries. This finding partially aligns with previous research indicating a general preference for referential questions among mothers from both cultural backgrounds (Luo & Tamis-LeMonda, 2017).

Cultural expectations and pedagogical practices might explain the observed differences in the cognitive demand level of mothers’ contributions, particularly at the statement level. Mexican American mothers in this study placed greater emphasis on inferential statements than Chinese American mothers. Their reading strategies involved utilizing manufactured speech, discussing characters’ emotions, and educating children about consequences. It was evident by previous research that Mexican American mothers were more likely to engage in emotion talk and socioemotional meaning-making with their children during shared reading (Chen et al., 2022; Luo et al., 2014; Leyva et al., 2014; Salinas et al., 2017). This style suggests that book reading is often integrated into family time with a priority on establishing socioemotional connection and moral lessons, rather than solely academic content. The example below shows a Mexican American mother discussing the characters’ emotions to illustrate the focus on inferential statements among the Mexican American dyads.

Mother: El niño está preocupado buscando la tortuga. (The boy is worried about the turtle.) [inferential statement]

Mother: El niño se mira enojado abrazando a su perro. (The boy looks mad hugging the dog.) [inferential statement]

Mother: El niño está como asustado. (The boy is scared) [inferential statement]

On the other hand, Chinese American mothers favored asking referential questions during shared book reading to encourage their children to label objects within the illustrations. This emphasis on labeling and concrete identification is strongly linked to the cultural value placed on knowledge acquisition and intensive reading styles common in Chinese educational traditions (Anderson et al., 2012; Chang & Shih, 2023; Li, 2005). Chinese parents often prioritize vocabulary development while reading, thus prompting children to identify objects and readily provide the correct word if their child struggles. The following example illustrates this focus on referential queries.

Mother: 阿Johnny佢有一隻狗仔. (Johnny he has a doggie.) [referential statement]

Mother: 同埋有個呢個叫乜嘢啊? (And what is this called?) [referential question]

Child: Froggie. [referential statement]

Mother: 同玩呢個叫咩名話? (And then what is this called?) [referential question]

Child: Doggie. [referential statement]

Children from both groups mainly contributed by providing statements. Chinese American children leaned more towards yes/no, referential, and behavioral statements, and referential questions than Mexican American children. On the other hand, Mexican American children tended to ask more yes/no questions. Notably, both groups favored providing referential and behavioral statements overall, aligning with the findings from Luo and Tamis-LeMonda (2017). Children might adjust their contributions based on their perceived expectations about the shared reading activity with their mothers. Mexican American children might ask their mothers more yes/no questions to stay on track with the story. In contrast, Chinese American children might provide more statements to participate actively in the narrative. Additionally, the relationship between each child and their mother could influence their communication patterns during shared book reading. Chinese American children might feel more comfortable providing statements to their mothers who expect responses.

Although these patterns are discussed through a cultural lens, it is important to acknowledge the relationships between culture and socioeconomic factors. Maternal education is a robust predictor of the complexity and quantity of parent-child conversation in diverse households (Hoff, 2013; Shen & Del Tufo, 2022). In the current study, the Chinese American mothers reported significantly higher levels of education than the Mexican American mothers. Higher educational attainment may partially explain the observed preference for matching cognitive demand with pedagogical instruction (Li, 2005). Conversely, the international patterns observed among Mexican American dyads might reflect complex relationships between cultural socialization goals and the participants’ specific educational backgrounds (Luo et al., 2022). Therefore, the identified interactional styles likely reflect both cultural socialization and maternal educational experiences (Rydland & Grøver, 2023).

Contingent Dynamics: Matching, Undermatching, and Scaffolding

Sequential analyses revealed a moderate to strong likelihood of children in both groups responding to their mothers’ questions, and mothers in both groups followed up with questions following their children’s statements, confirming the findings by Luo and Tamis-LeMonda (2017). This bidirectional relationship confirms the contingent responsibility, which is a crucial component of effective scaffolding where a speaker’s response is specifically tailored to the preceding contribution to maintain the conversation flow within the child’s ZPD (Vygotsky, 1978). Mothers in both groups appear to be dynamically adjusting the scaffolding to provide the appropriate level of support for their child’s linguistic engagement through consistently following up on child statements. The excerpt below provides a representative example of this significant contingency in back-and-forth conversations observed in Chinese American dyads, in which the child responded to the mother’s questions and the mother continued to ask follow-up questions.

Mother: 咁佢唔開心噃係唔係啊? (He is not happy, right?) [yes/no question]

Child: 佢搵唔到青蛙. (He cannot find the frog.) [behavioral statement]

Mother: 佢邊度搵啊? (Where did he find it?) [referential question]

Child: 佢喺個喺個... (He is at…) [referential statement]

Mother: 你睇佢喺邊度搵? (You see where he found it?) [referential question]

Child: 喺喺喺個 shoes 嗰度. (In, in, in the shoes.) [referential statement]

Mother: 哦喺鞋嗰度搵. (Oh, he found it in the shoes.) [behavioral statement]

Consistent with Luo and Tamis-LeMonda’s (2017) work, the magnitude of the mother-to-child reciprocal relationship was more robust than that of the child-to-mother relationship. This asymmetry reflects the mother’s intentional scaffolding, in which they have the primary responsibility for maintaining the structure of the interaction. The present study identified distinct response patterns within each group that extend previous research. While children in both groups responded to maternal questions with statements, a significant sequential association for children responding with questions was observed only in Chinese American dyads. Chinese American children may seek clarification through both straightforward answers and additional queries. This pattern suggests a collaborative inquiry style of scaffolding, in which the child actively seeks information to support their understanding of the subject within the ZPD. For example:

Mother: 呢個係咪鹿仔鹿仔啊? (Is this a deer?) [yes/no question]

Child: 鹿仔啊? (A deer?) [yes/no question]

Child: 做咩變咗鹿仔嘅? (Why did it turn into a deer?) [inferential question]

Another divergence in the observed sequences was that Mexican American children showed a significant likelihood to initiate questions following their mothers’ statements, which was not observed in the Chinese American dyads. From the perspective of ZPD, this finding suggests that Mexican American children may take on a more active role in negotiating the cognitive demand of the conversation, and thus prompting their mothers to provide a higher level of cognitive distancing explanations or discussion (Sigel, 2002). An example of this is shown below.

Mother: Estaba triste. (She was sad.) [inferential statement]

Child: Uh. [yes/no statement]

Mother: La ranita. (The frog.) [referential statement]

Child: Por qué? (Why?) [inferential question]

Mother: Porque ella quería estar con su mamá. (Because she wanted to be with her mom.) [inferential statement]

Child: Por qué? (Why?) [inferential question]

Mother: Si a tí te llevan tú quieres estár lejos de tu mamá? (If someone takes you away, do you want to be far from your mom?) [inferential question]

Child: Uh-uh. [yes/no statement]

The instances of undermatching and overmatching are not behavioral occurrences but rather moment-by-moment negotiations of the child’s ZPD, where cultural practices and backgrounds may dictate whether the mother or the child initiates or adjusts the complexity of the speech or scaffolding.

Cognitive Demand and the Contingency of Scaffolding

A main finding of this study is that DLL children do not share a uniform developmental profile. Sequential analyses provided novel insights into the dynamics of the cognitive demand levels of the discussion during mother-child shared reading interactions through distinguishing between the availability of high-level talk and the child’s ability to contingently participate in it. The results challenge the idea of a uniform DLL experience by revealing distinct cultural patterns in how children and mothers navigate high levels of cognitive demanding languages.

The most striking difference in the significant sequential associations between the two groups was how DLL children navigated maternal inferential questions. A unique undermatching sequence was identified in the Mexican American group, in which children significantly transitioned to yes/no answers, whereas the Chinese American dyads showed significant matching sequences across all cognitive demand levels. The undermatching observed among Mexican American dyads may be due to difficulty grasping the inferential nature of the questions or a reluctance to engage in inferential talk in a non-academic setting (Hindman et al., 2014). However, this non-uniform experience is further explained by age, with older Mexican American children more likely to respond to their mothers’ inferential questions with inferential statements, aligning with previous literature indicating that children can match the complexity of their parents’ speech (Luo & Tamis-LeMonda, 2017). The Mexican American children’s systematic reversion to yes/no answers on inferential questions means they are contingently less likely to engage in and benefit from the most cognitively demanding exchanges (Sun & Batra, 2024). The sequential data provide a precise mechanism for how this disparity in linguistic exposure occurs moment-by-moment and reinforces the need for interventions to target this specific inferential-response contingency rather than assuming a DLL deficit.

In contrast, the matching observed in the Chinese American dyads demonstrates a more consistent ability to meet high cognitive demands, potentially reflecting the cultural emphasis on academic effort and the structured knowledge transmission observed in mothers’ style. Sequential analysis further showed that these children were also likely to respond at a lower cognitive demand level with referential statements, like their Mexican American counterparts, which suggests that high-level responses are not mandatory. Below is an example of a Chinese American child responding to their mother’s inferential question with an inferential statement.

Mother: 點叫啊? (How does he say?) [inferential question]

Mother: 你講俾媽咪聽佢點叫? (Tell mommy how he says it?) [inferential question]

Child: 唔好暱喺入面. (Don’t hide in there.) [inferential statement]

Below is an example of the Chinese American child responding with a referential statement to the mother’s inferential question.

Mother: 個蜜蜂追住個狗去. (The bees are chasing the dog.) [behavioral statement]

Mother. 點算呀? (What should he do?) [inferential question]

Child: 老鼠呀. (A mouse.) [referential statement]

Furthermore, there were a few occasions when children responded at a higher cognitive demand level than the initial questions or statements. Both Mexican American and Chinese American children were likely to answer maternal referential questions with behavioral statements. For example:

Mother: A quién andan buscando? (Who are they looking for?) [referential question]

Child: Ya la encontraron. (And they found it.) [behavioral statement]

This finding suggested that when children were asked to identify information in the book, they went beyond simply labeling the object and incorporated contextual information. This finding highlights preschool-age DLL children’s ability to integrate and apply existing knowledge and experiences into their responses. Furthermore, Mexican American children often followed their mothers’ yes/no answers with referential questions, which suggests their desire to gain a deeper understanding of the story. Additionally, younger Chinese American children were more likely to follow up with inferential questions following maternal behavioral statements. This finding suggests that younger children may be curious and eager to seek additional information from their mothers to support their understanding of the story. The example below demonstrates a younger Chinese American child who follows up with an inferential question after their mother’s behavioral statement, prompting the mother to continue the conversation using inferential language.

Mother: 個狗仔仲喺度跟住佢. (The dog is still following him.) [behavioral statement]

Child: 點解佢? (Why does he?) [inferential question]

Mother: 因為佢見到有個鹿喺度囉. (Because he sees the deer there.) [inferential statement]

Mother: 佢驚啊. (He is afraid.) [inferential statement]

Furthermore, the strategies mothers used to follow up on their children’s responses further illustrate the distinct cultural significance of scaffolding. Mexican American mothers primarily responded with yes/no, referential, and inferential statements, while Chinese American mothers responded with yes/no, behavioral, and inferential statements. Both groups of mothers were likely to use yes/no answers when responding to inferential or referential questions from their children. This tendency could be attributed to a preference for concise and direct communication when interacting with young children. Mothers might prioritize acknowledging their children’s questions with a quick yes/no response and leaving room for elaboration in subsequent responses. However, this study only examined the immediate responses (lag 1 sequences). It is plausible that the mothers provided more detailed answers and matched the cognitive demand level of their children’s questions later in the conversation (lag 2 sequences).

Mothers’ follow-up questions from both groups also mirrored children’s previous contributions in terms of cognitive demand levels. Mexican American mothers followed up with yes/no and inferential questions, while Chinese American mothers followed up with yes/no, referential, and inferential questions. This differential follow-up strategy highlights how mothers strategically adjust their scaffolding in response to perceived cultural norms and competence. Mexican American mothers followed up with yes/no questions after their children’s referential and inferential statements. As mothers were aware of adjusting the complexity of their questions to match their children’s understanding (Sigel, 2002), there is a possibility that Mexican American mothers might have used yes/no questions to confirm their children’s understanding before moving on, or a strategy to reduce the overall interactional load (Chan et al., 2022). Below is an example of a Mexican mother asking a yes/no question after their child’s inferential statement:

Child: Estàba enojado. (He was mad.) [inferential statement]

Mother: Estàba enojado el niño? (The boy was mad?) [yes/no question]

Child: Sì porque… (Yes, because…) [yes/no statement]

Mother: Sì estàba enojado porque se había metido al frasco. (Yes, he was mad because the dog's head was in the jar.) [inferential statement]

Mexican mothers also tended to ask inferential questions after child yes/no statements in the older age group. This finding demonstrates that mothers of older children adjusted their speech and encouraged their children to engage in inferential discussions based on their age and perceived abilities. This supports the findings from Luo and colleagues (2022), where developmental age plays a role in the cognitive demand levels of mothers’ speech.

On the other hand, Chinese American mothers typically followed up with referential and behavioral questions in response to their children’s yes/no statements. Chinese American mothers likely wanted to encourage exploration and elaboration from their children during shared reading, so they followed up with referential and behavioral questions to learn more about the object in question. This is a targeted scaffolding move to return the conversation to the narrative facts without immediately jumping to the highest cognitive demands (Huang et al., 2025). An example is shown below:

Mother: 睇個frog. (Look at the frog.) [referential statement]

Child: Mm. [yes/no statement]

Mother: 仲有個呢? (And one more?) [referential question]

Mother: 同邊個睇呀? (Who is there to watch with him?) [referential question]

Child: 嗰個doggie. (That doggie.) [referential statement]

Additionally, Chinese mothers were likely to give yes/no responses after older children's inferential questions. Older children might have asked inferential questions to seek information, and their mothers interpreted their questions as seeking confirmation rather than further exploration of the topic, as demonstrated by the following example:

Child: 點解佢要咁睇嘅? (Why does it have to look this way?) [inferential question]

Mother: 係囉. (Yes.) [yes/no statement]

These divergent patterns, in which one group maintains matching cognitive demand sequences and the other prioritizes different narrative flows, demonstrate that the DLL experience is not uniform. Both groups showed the ability to go beyond simple labeling and confirming, such as when children answered referential questions with behavioral statements. This highlights that preschool-age DLL children in this study were able to integrate and apply existing knowledge when prompted, but the contingent associations through which they expressed this skill were shaped by the interactions of their cultural, linguistic, and socioeconomic environments, including the varying educational backgrounds of the mothers.

Theoretical and Practical Implications

The findings of the current study refine the theories of bilingual language socialization by demonstrating that language learning through scaffolding is not a universal set of practices but a culturally contingent process (Ochs & Schieffelin, 2017). Although foundational theoretical frameworks like the ZPD emphasize the importance of adult scaffolding (Vygotsky, 1978), the current findings suggest that the ZPD can take different forms depending on cultural priorities, ranging from structured knowledge instruction to socioemotional discussion (Rogoff, 2003).

This study extends Sigel’s (2002) Psychological Distancing Model by demonstrating that undermatching cognitive demand is not necessarily a failure to provide high cognitive demand input, but rather a strategic language adjustment. In Mexican American dyads, the child’s reversion to a yes/no response following a maternal inferential question suggests a preference for maintaining the story narrative over abstract cognitive conversation (Caspe, 2009). Conversely, the consistent matching observed in Chinese American dyads demonstrates a language socialization goal focused on accuracy and knowledge transmission (Li & Fleer, 2015). Overall, these distinct scaffolding practices confirm that DLL experience is defined by diverse contingent associations rather than a monolithic developmental pathway.

A major practical insight from the current study is that interventions for DLL families must move beyond general suggestions for more talk and instead target specific response contingencies, especially inferential ones (Sun & Batra, 2024). These interventions should build upon existing cultural strengths. For Mexican American dyads, the intervention could focus on supporting mothers in scaffolding their inferential questions, with prompts that leverage their existing strengths in socioemotional discussion to encourage children’s inferential responses (Leyva et al., 2014). For Chinese American dyads, intervention could focus on encouraging mothers to gradually shift toward more inferential inquiry rather than only referential talk. Mothers could leverage their child’s strong factual knowledge to facilitate deeper abstract discussions (Anderson et al., 2012; Chang & Shih, 2023). These culturally responsive interventions should respect and leverage the existing strengths in these diverse mothers’ storytelling and cultural practices to foster literacy and cognitive development.

Limitations and Future Directions

Although sequential analysis highlights the temporal contingencies between two events, it does not establish causality between them. This study identifies specific and recurring patterns of contingent risk and success (e.g., child undermatching) and provides a mechanism for how differences in language environments emerge, but it cannot determine if these specific sequences cause later developmental outcomes. Additionally, the subgroup comparisons by child age were exploratory, focusing on identifying unique contingent patterns within cohorts rather than testing for statistical differences between age-group coefficients. Future longitudinal research is required to connect these observed contingent patterns to DLL children’s later language and literacy proficiency.

Further, as the present study focuses on mother-child shared book reading, it limits the generalizability of the contingent relationships between mothers and children’s speech in families of Mexican and Chinese heritage to other home contexts. Although this study focused on low-income immigrant families of Mexican and Chinese heritage, the results may not generalize to other subgroups within these populations with different socioeconomic status, origin, or immigration history. Specifically, the significant difference in maternal education between the two groups limits our ability to isolate “culture” as the primary driver of the interactional dynamics, as DLL variability can be influenced by demographic factors (Rydland & Grøver, 2023). Future research with a larger and more educationally diverse sample should examine how these demographic factors within diverse cultural groups exhibit interactional styles and determine the extent to which these contingent patterns are driven by cultural heritage rather than educational history. Given that cultural norms may prescribe different interactional styles across settings, future research should explore how cultural differences affect parent-child interactions, such as play activities and daily conversations.

Additionally, since the present study focused on the immediate responses (lag 1), examining multi-lag sequences (e.g., lag 2 or lag 3) is crucial for capturing the broader, multi-turn structure of scaffolding. For instance, examining the third turn (lag 2) in a sequence would clarify whether a mother eventually scaffolds a child’s undermatching response, such as by rephrasing an inferential question or providing an answer that bridges the child’s simple response back to a higher level of cognitive demand. These multi-lag sequential analyses would provide stronger evidence for whether undermatching functions as a strategic adjustment to maintain conversation flow within the child’s ZPD rather than a missed pedagogical opportunity. It is also important to acknowledge the potential limitations of measuring speech complexity using cognitive demand levels. Exploring alternative measures, such as linguistic complexity (e.g., mean length of utterance), may provide additional insights into the parent-child interactions during shared reading.

Finally, investigating specific cultural beliefs, values, language, and literacy-related expectations, and communication styles through qualitative interviews will provide a more nuanced understanding of the cultural perspectives on the practices employed by parents during shared book reading. Moreover, it is recommended that comparative studies with families from other cultural backgrounds be conducted. These studies should also go beyond identifying culturally specific but universal features that support language development. This will uncover shared and unique communication patterns in parent-child shared book reading to identify common mechanisms underlying language and literacy development through shared book reading.

Conclusion

The present study provides novel sequential evidence that cultural background not only influences the frequency of parental talk but systematically shapes the moment-to-moment dynamics of scaffolding during shared book reading. Consistent with Vygotsky’s (1978) Sociocultural Theory and Sigel’s (2002) Psychological Distancing Model, mothers and children in both Mexican American and Chinese American groups tended to adjust the complexity of their speech to match each other’s contributions.

However, the findings revealed culturally specific deviations from this principle of complexity match among low-income immigrant dyads that likely reflect the combined influence of cultural priorities and demographic variations, such as maternal education (Hoff, 2013; Rydland & Grøver, 2023; Shen & Del Tufo, 2022). Specifically, the Mexican American children’s systematic undermatching to high-level inferential questions, alongside Mexican mothers’ tendency toward low demand yes/no follow-up, highlights a pattern of contingent interaction that reflects a cultural prioritization of socioemotional connection and narrative flow over inferential discussion. This may limit sustained inferential exchanges in specific contexts. Conversely, Chinese mothers’ preference for mid-level (referential and behavioral) follow-up demonstrates a strategic approach to scaffolding that keeps the child within a productive zone of learning.

These results go beyond correlational findings by providing precise and dynamic indices (overmatching and undermatching) that show how cultural nuances and specific contextual factors during shared reading can lead to deviations from the match principle. This underscores the critical need for culturally responsive interventions that do not just promote more talk but target the contingent structure of the discourse. Future research should use these fine-grained sequential methods to develop culturally informed strategies that effectively support language and literacy development and foster positive and enriching parent-child interactions for DLL children across diverse cultural and socioeconomic contexts.

Paper Highlights.

• This study examined how cultural backgrounds influence the moment-to-moment scaffolding and cognitive demand during shared reading among dual language learners.

• Sequential analysis of 152 mother-child dyads revealed group-specific patterns of cognitive matching, overmatching, and undermatching during interaction.

• Findings suggest culturally responsive interventions should focus on the contingent structure of language interactions for dual language learners.

Appendix. Cognitive Demand Coding Definitions and Examples

Coding Categories	Definitions	Examples
Yes/No Questions	Questions that can be answered by yes or no that are unrelated to child’s lived experience.	Do you see the boy?
Yes/No Statements	Statements that consist of a yes (affirm) or no (deny).	Yes. That is right.
Referential Questions	Questions that ask for labels, descriptions, or locations of visible objects in the picture, including objects previously pictured in the book.	What is that? Where is the frog?
Referential Statements	Statements that describe the name, feature, or location of objects in the picture, including objects previously pictured in the book.	This is a frog. This is blue.
Behavioral Questions	Questions that focus on inferring unobserved actions of story characters in the picture.	What is the boy doing? He looked for the frog.
Behavioral Statements	Statements that describe the actions of the story characters in the picture, including actions that have been suggested by previous pictures in the book.	The bees are chasing the dog. The frog left.
Inferential Questions	Questions that prompt inferences, imagination, internal states, or social knowledge that go beyond the picture.	Why is the boy unhappy? He ran as fast as he could.
Inferential Statements	Statements about the book that require inferences, interpretations, internal states, manufactured speech, and/or deductive thinking.	The boy should not have put his clothes on the floor like that. The boy said, “The frog is gone.”

Data Availability Statement:

The data that support the findings of this study are openly available in the Open Science Framework (OSF) at https://doi.org/10.17605/OSF.IO/3UTYC.