Maternal question use and child language outcomes: The moderating role of children's vocabulary skills and socioeconomic status

Highlights

• Mothers’ Y/N and referential questions with 2-year-olds predicted child receptive language 1 year later.

• Stronger association between referential questions and child receptive language for children with larger vocabulary.

• Stronger association between referential questions and child receptive language for children from lower-socioeconomic status families.

• The association between Y/N questions and child receptive language did not vary by children's concurrent vocabulary and socioeconomic status.

Abstract

Question use is a key feature of high-quality language input and supports language development. The current study examined how different types of maternal questions were associated with children's language outcomes, and how the associations were moderated by children's existing vocabulary skills and family socioeconomic status. Participants were 165 mothers and their 2-year-olds from the study of Early Child Care and Youth Development. Mother-child interactions during book-reading and toy play were video-recorded and coded for three types of questions: Yes or No (Y/N; e.g., “Is that a doggy?”), referential (e.g., “What's that?”), and advanced questions (e.g., “What will happen next?”). Child vocabulary at age 2 and receptive and expressive language skills at age 3 were assessed using standardized measures. Results suggested that Y/N questions and referential questions uniquely contributed to children's receptive language 1 year later. Yet, the benefit of referential questions did not generalize to children with small vocabularies at age 2. The association between referential questions and children's receptive language was stronger for children from families with lower-socioeconomic status, highlighting the possibility for intervention in this area. Together, these findings reveal the critical role of individual and contextual factors in language-learning mechanisms.

Introduction

Early language learning occurs in the context of social interaction (e.g., Bruner, 1981; Snow, 1999; Vygotsky, 1978). While the quantity of language input matters, quality plays a greater role in language development (e.g., Hirsh-Pasek et al., 2015; Rowe, 2012). One important characteristic of high-quality language interactions is parents’ question use. Questions help maintain children's attention (Robinson et al., 2009), facilitate comprehension (Weinstein et al., 2010), encourage critical thinking, and allow children to practice using language (Honig & Wittmer, 1981; Mol et al., 2008). Questions also allow parents to evaluate children's knowledge and adjust their behaviors accordingly to meet children's changing needs (see Bailey et al., 2013). Extensive evidence suggests that parental questions facilitate children's vocabulary (Blake et al., 2006; Blewitt et al., 2009; Fletcher et al., 2008; Tompkins et al., 2017), narrative (Kuchirko et al., 2016), and cognitive (Cristofaro & Tamis-LeMonda, 2012; Hubbs-Tait et al., 2002) development during infancy and the preschool years. While it is critical to examine the questioning behaviors of both parents (Rowe et al., 2017), here we focused on maternal questions because of the predominance of mothers as primary caregivers.

Although the benefits of maternal questions have been demonstrated for children from various socioeconomic (e.g., Cristofaro & Tamis-LeMonda, 2012; Tompkins et al., 2017) and ethnic/cultural contexts (e.g., African American, Cristofaro and Tamis-LeMonda, 2012, Quiroz et al., 2010), most studies have examined fairly homogeneous samples. The current study aimed to examine the association between maternal questions and children's language development in families from diverse socioeconomic background and in children with different levels of vocabulary skills. The bioecological theory posits that the influences of language interactions on developmental outcomes may differ by individual characteristics and contextual factors (Bronfenbrenner & Morris, 2006). Indeed, early interventions aiming to enhance question use are found to be less effective for children from lower socioeconomic status backgrounds and for children at high risk for language delay (Manz et al., 2010; Mol et al., 2008). Yet it remains unclear whether these discrepancies result from group differences in the effects of question-asking, training effectiveness, or both. More nuanced examinations are needed to understand the generalizability of the effect of maternal questions. To do so, the current study examined how different types of maternal questions predicted children's language outcomes 1 year later. Furthermore, we examined whether the association between maternal questions and child language outcomes varied by children's vocabulary skills and family socioeconomic status (SES). Findings from this study will shed light on language learning mechanisms and inform early language interventions for populations diverse in SES and child language skills.

The first aim of the study is to understand how different types of maternal questions relate to children's language development. Question use is a key feature of high-quality language interactions between caregivers and children. Studies have suggested that the frequency and proportion of maternal questions predict children's vocabulary size (Blake et al., 2006; Blewitt et al., 2009; Tompkins et al., 2017), emergent literacy (Cristofaro & Tamis-LeMonda, 2012), narrative skills (Kuchirko et al., 2016), and cognitive skills (Hubbs-Tait et al., 2002), concurrently and longitudinally. Although limited, studies with fathers and both parents have yielded similar findings (Leech, Salo, Rowe, & Cabrera, 2013; Rowe et al., 2017). Intervention work further demonstrates the benefits of question use. Early language and literacy programs, such as dialogic reading, which focus on facilitating parents’ question use, successfully improve children's language outcomes (Mol et al., 2008; Sénéchal, 1997; Whitehurst et al., 1994).

Questions are generally beneficial, but not all are created equal (Leech, Salo, Rowe, & Cabrera, 2013). The cognitive distancing theory differentiated questions with various level of cognitive demand (Sigel, 1993). At a basic cognitive level, there are closed-ended questions, also referred to as Yes-No (Y/N) questions. These questions only require children to verbally or non-verbally affirm or negate a statement (Leech, Salo, Rowe, & Cabrera, 2013). In contrast, open-ended questions (e.g., what, when, where, who, why, and how) can have many different answers and are therefore more cognitively challenging than closed-ended questions (Leech, Salo, Rowe, & Cabrera, 2013). At a moderate cognitive level, referential questions, also referred to as prompt wh-questions (Leech, Salo, Rowe, & Cabrera, 2013) or simple labeling questions (Britto et al., 2006), ask for locations, descriptions, and labels of objects in the here and now context (e.g., “What's this?” “What color is it?”; Luo & Tamis-LeMonda, 2017). At a high cognitive level, advanced questions, also referred to as informative wh-question (Leech, Salo, Rowe, & Cabrera, 2013), ask about behaviors, inferences, predictions, or internal states (e.g., “What will happen next?” or “Why are you doing this?”).

In line with the cognitive distancing theory, there is evidence that referential questions can better support vocabulary learning than Y/N questions (Leech, Salo, Rowe, & Cabrera, 2013). Additionally, studies have suggested that advanced questions used by mothers and fathers prompt children to imagine and think abstractly beyond the here and now context, and are more likely to elicit multiword, syntactically complex responses (Danis et al., 2000; Leech, Salo, Rowe, & Cabrera, 2013; Luo & Tamis-LeMonda, 2017). Studies using sequential analysis further demonstrate contingencies between maternal questions and children's responses at the same cognitive level (Luo & Tamis-LeMonda, 2017; Tompkins et al., 2017). For instance, a referential question (e.g., “What's this?”) from a mother is likely to elicit a simple response from her preschool child (e.g., “a tree”), whereas an advanced question (e.g., “Why is he scared?”) is likely to get a more advanced response (e.g., “he is scared because the owl is chasing him”; Luo & Tamis-LeMonda, 2017).

Some types of questions may be more beneficial than others for a specific developmental stage. The current study focused on toddlerhood, when children experience rapid vocabulary growth, start to produce word combinations and simple sentences, and increasingly engage in conversations (Pence Turnbull & Justice, 2017). Open-ended questions might be especially beneficial during this period of time, because these questions invite children to practice vocabulary words and produce verbal responses in a sophisticated manner (Leech, Salo, Rowe, & Cabrera, 2013). Indeed, one study with middle-SES and low-SES families found that mothers who used a higher proportion of “what” questions during book-sharing with their 2-year-olds had children with better vocabulary skills (Ninio, 1980). Likewise, fathers’ use of open-ended questions with toddlers in low-income households predicted children's vocabulary skills at age 2 and verbal reasoning skills at age 3 (Leech, Salo, Rowe, & Cabrera, 2013; Rowe et al., 2017).

Unlike open-ended questions, there are mixed findings on the benefits of Y/N questions during toddlerhood. Although evidence suggests that Y/N questions with 1-year-old infants support language skills (Muhinyi & Rowe, 2019; Smith et al., 2019), researchers have argued that Y/N questions might not be challenging enough for older toddlers (Leech, Salo, Rowe, & Cabrera, 2013). One study found that fathers’ Y/N questions with 2-year-olds did not relate to children's vocabulary skills (Leech, Salo, Rowe, & Cabrera, 2013). However, other studies have combined parental Y/N and open-ended questions and found that parents’ total question use was associated with 2-year-olds’ attention (Fletcher et al., 2008), vocabulary skills, and language complexity (i.e., mean length of utterance; Blake et al., 2006). Although Y/N questions do not require complex responses from children, they may help maintain mother–child joint attention, thereby supporting language learning (Fletcher et al., 2008).

Together, these previous findings have highlighted the importance of open-ended questions for toddlers’ language development. Yet, less is known about which type of open-ended questions is most helpful and the role of Y/N questions. The current study filled in this gap by differentiating questions at three levels of cognitive demand: Y/N questions, referential questions, and advanced questions. We examined the joint and unique contributions of different types of questions to children's language outcomes.

The second aim of the study is to understand whether different types of questions relate to later language outcomes differently for children who start with varying levels of language skills. According to the sociocultural theory, maternal questions are most beneficial when they are slightly beyond children's concurrent language skills and fall within the zone of proximal development (Vygotsky, 1978). Thus, children's existing skills may moderate the associations between maternal questions and language outcomes.

Studies comparing typically developing children to those with language impairments or delays yield mixed findings. Some studies suggest that maternal questions predicted language outcomes in children with language impairments (Yoder, 1989) and late talkers (Levickis et al., 2014). Two-year-old late talkers were also as likely as their typical developing peers to answer questions from their mothers (Rescorla et al., 2001). However, other studies have shown that children with language impairments had difficulty processing questions (Marinis & van der Lely, 2007). Furthermore, maternal questions did not elicit verbally complex responses from children with language impairments in the same way they did from typically developing children (McGinty et al., 2012). Meta-analyses also suggest that children who are at higher risk of language delays benefit less from dialogic reading than those children who are at lower risk (Mol et al., 2008).

Additional research is needed to understand how the developmental role of questions varies based on children's language skills. As mentioned above, children with lower language skills, like younger children, benefit more from simple questions, whereas those with higher language skills, like older children, benefit more from advanced questions. Examining how different types of questions relate to language development for children with a wide range of normative language skills can inform the design and individualization of early interventions.

The third aim of the study is to understand whether SES moderates the association between different types of maternal questions and child language development. Here we focused on maternal education and family income-to-needs ratio as two indicators of SES. On average, mothers from low SES backgrounds, as measured by income, maternal education, and/or occupation, tend to ask fewer questions than mothers from middle or high SES backgrounds when interacting with their children (e.g., Chang & Huang, 2016; Rodríguez et al., 2009). However, there are inconsistent findings on whether the differences are seen in Y/N questions, open-ended questions, or both (Chang & Huang, 2016; Heath, 1982; Rodríguez et al., 2009; Vernon-Feagans et al., 2020). These differences might be due to disparities in mothers’ own language proficiency (Hoff et al., 2018), their knowledge of child development (Rowe, 2008), contextual factors such as limited resources or high stress level (Froiland et al., 2013; Lovejoy et al., 2000), or cultural variations in parenting styles and expectations (Anderson-Yockel et al., 1994).

The SES-related disparities in maternal questions have led efforts to increase the number of questions mothers in low-SES homes use with their children. However, meta-analyses have found that the effects of dialogic reading and other early interventions vary significantly by family income, with a moderate effect size for children from middle or high-income households but only a small, non-significant effect size for children from low-income households (Manz et al., 2010; Mol et al., 2008). Similarly, the intervention effects are stronger as mothers’ educational level increased (Mol et al., 2008). Why do children from lower-SES families benefit less from dialogic reading and other early interventions than those from higher-SES families? One possibility is that the use of questions does not facilitate child language outcomes as much for children from low-SES households as for those from middle- or high-SES households. Question asking and answering is a very common and highly-valued routine within a formal schooling context. However, children and parents from low-SES families are often less familiar with this teaching and learning style than those from higher-SES families (see Fannin et al., 2018). Although low-SES parents increase the sheer number of questions they ask as a result of the intervention, the quality and fluency of their interaction with children might still need to be improved. Alternatively, some researchers have argued that comprehending and answering questions may be especially challenging for low-SES children, who as a group have relatively low language skills (Mol et al., 2008). In this case, the differential effect of questions could be due to disparities in children's existing language skills, rather than SES differences directly. Once children's existing language skills are controlled for, children from low-SES families might benefit as much from questions as their middle- or high-SES peers. Finally, it is also possible that some interventions had low sociocultural validity and failed to adapt to the specific needs of low-SES families (Mol et al., 2008). In addition to questioning, low-SES parents may use other ecologically valid strategies, such as storytelling, teasing, and rhyming, to facilitate early language development (Gilmore, 1986; Miller, 1986; Vernon-Feagans et al., 2001). Yet, these culturally valued teaching strategies have not been fully documented in existing literature or reflected in the design of early interventions (Larson et al., 2020). To date, empirical tests of these possibilities are missing. In this study, we examine the first and second possibilities by asking whether SES moderates the relation between maternal questions and child outcomes, while controlling for children's existing vocabulary.

There is a well-established association between maternal questions and children's language outcomes. However, less is known about which types of maternal questions are most beneficial for toddlers from SES-diverse families. Additionally, more work is needed to understand individual and contextual factors that moderate the association between maternal questions and child language outcomes. The current study examined the association between maternal questions with 2-year-olds and children's language skills 1 year later and asked whether the strength of the association varied by children's existing vocabulary skills and family SES (i.e., maternal education and income-to-needs ratio). In particular, we reduced the confound between SES and children's existing language skills by oversampling children with low SES and high language skills and those with high SES and low language skills. We asked three research questions:

• Do different types of maternal questions predict children's language skills 1 year later? We focused on three types of questions with cognitive demands changing from low to high: Y/N, referential, and advanced questions. We expected referential and advanced questions to be more beneficial than Y/N questions during toddlerhood.

• Does children's concurrent vocabulary moderate the associations between maternal questions and children's language skills? Maternal questions at moderate or high cognitive level (i.e., referential and advanced questions) were expected to be more strongly associated with child outcomes for those children with larger vocabularies. In contrast, maternal questions at a lower cognitive level (i.e., Y/N questions) might be more strongly associated with child outcomes for those children with smaller vocabularies.

• Does SES moderate the associations between maternal questions and children's language skills? We expected the associations to be stronger for higher-SES families, based on findings from previous intervention work. However, the strength of the associations might not vary by SES after controlling for child language skills.