Indirect Effects of Early Shared Reading and Access to Books on Reading Vocabulary in Middle Childhood

ABSTRACT

Purpose

This study investigated the effects of early shared reading and access to books on reading vocabulary in middle childhood and the pathways associated with later reading success.

Method

Path analysis of data from four waves of a national longitudinal study of children (N = 7,751) was used to investigate direct and indirect effects of early home literacy practices at age 3 on reading attainment at age 9, controlling for demographic and family characteristics. Children were aged 3, 5, 7, and 9 years at each respective wave of data collection, and 49.1% of the sample were girls. The majority of caregivers in the study were White (94.3%).

Results

Early shared reading and availability of books predicted higher reading scores at age 9, controlling for covariates at age 3. Sixty-four percent of the association between early shared reading and reading attainment in middle childhood was explained by several paths incorporating language and literacy variables at age 5 and reading for pleasure at age 7. The association between books at age 3 and vocabulary at age 9 was fully accounted for through an increase in the number of books at age 5 and greater likelihood of reading for pleasure at age 7.

Conclusion

Both early shared reading and availability of books predicted reading attainment in middle childhood though through different pathways. The findings support reading interventions that provide books and shared reading opportunities in early childhood.

Keywords:

• Early Childhood
• Literacy
• Shared Reading
• Books
• Reading Attainment

Exposure to storybooks in early childhood is an effective way to promote early vocabulary skills (Bus et al., 1995) and is associated with reading comprehension skills in middle childhood, both indirectly (Sénéchal, 2006) and directly (Shahaeian et al., 2018). Early storybook exposure has been conceptualized as incorporating both shared book reading (i.e., when an adult reads to a child) and the number of children’s books in the home and predicts vocabulary scores in grade one (age seven) and later reading for pleasure in grade 4 (age 10), as well as indirectly predicting reading comprehension in grade 4 (age 10) (Sénéchal, 2006). Availing of nationally representative data in Ireland, this study investigated whether early storybook exposure through shared book reading and the number of children’s books in the home at age 3 predicted vocabulary six years later at age 9 through effects on children’s language and literacy skills at school entry and reading for pleasure at age seven.

Shared book reading in early childhood and reading in middle childhood

Shared book reading is well researched as an informal home literacy experience and is a well-established predictor of early language and literacy skills (Frijters et al., 2000; Levy et al., 2006; Raikes et al., 2006); these skills, in turn, are important predictors of educational attainment (Duncan et al., 2007; Ritchie et al., 2015). Early shared book reading is an informal literacy experience which increases opportunities for learning novel vocabulary (Cunningham & Stanovich, 1997; Fernald & Marchman, 2012), as well as helping to develop phonological awareness, print knowledge, asking questions and developing oral language skills and reading for pleasure (Tucker-Drob & Harden, 2012). There is some evidence that early shared reading is directly associated with educational attainment in middle childhood (Shahaeian et al., 2018) as well as indirectly through effects on early literacy and language skills (Shahaeian et al., 2018; Sénéchal, 2006). The primary mechanism by which early shared reading is thought to support later reading comprehension is primarily through increasing vocabulary skills and metalinguistic awareness, as well as providing a sense of comfort with the “culture of literacy” (Snow et al., 1995).

A key component of shared book reading is the social interaction between the adult and child, and it is this interaction that is thought to account for the learning of new words and to increase children’s interest in books (DeTemple & Snow, 2003; Mol et al., 2008). Reading to infants as young as 9 months has been found to uniquely predict vocabulary skills at age 3, even after controlling for concurrent literacy practices and SES (Leech et al., 2022) and recent findings from the Longitudinal Study of Australian Children (LSAC) suggested both direct and indirect effects of shared reading at age 2 to age 3 on a range of educational achievement measures at age 8 and age 9 (Shahaeian et al., 2018). One limitation of the longitudinal data in the LSAC study is that a measure of reading for pleasure among early readers (at age 7/8 years) was not included. Reading enjoyment is correlated with reading fluency (De Naeghel et al., 2012; Froiland & Oros, 2014) and there is evidence that children’s motivation to read in early to middle childhood is affected by their early literacy skills. For example, a meta-analysis of longitudinal literacy studies found bidirectional effects of reading ability and motivation, with early literacy skills predicting later motivation to read However, there is a growing research literature highlighting that shared book reading may reflect masked genetic effects of parental reading ability on children’s language and literacy skills (Hart et al., 2021; Puglisi et al., 2017). Overall, the evidence on the direction of the effects between children’s reading ability and reading for pleasure is mixed, with empirical support for viewing early reading ability as a predictor of later reading motivation (Toste et al., 2020)

Access to children’s books in early childhood and reading in middle childhood

The association between access to children’s books in early childhood and vocabulary and literacy outcomes has been explored through studies on book gifting programmes. In a systematic review, De Bondt et al. (2020) found that book giveaway programmes promoted the home literacy environment and children’s outcomes on literacy-related skills. Programmes which provided a small number of books to children, but which also incorporated advice to parents on shared reading with young children, were found to be the most effective in supporting children’s outcomes (De Bondt et al., 2020; Dowdall et al., 2020). Other studies show that the number of books available to children is an important environmental variable in supporting children’s literacy skills (van Bergen et al., 2017). A large international study of number of books in the home and the academic performance of adolescents, as measured using the OECD’s Programme for International Student Assessment (PISA’s) combined reading scale, found that the number of books in the home predicted reading skills more than parental educational attainment (Evans et al., 2014) providing further support for access to books as having an important, if modest (van Bergen et al., 2017), effect on children’s literacy development.

Access to books and the provision of a print-rich environment has long been considered important in supporting literacy development (Neuman & Celano, 2001). However, less attention has been given to the availability of books in the home as a unique predictor of reading attainment (McQuillan, 1996). Availability of books in the home has been found to account for a substantial proportion of differences in children’s early vocabulary skills by levels of maternal education (McNally et al., 2019). In a mediation analysis of the association between maternal education and early vocabulary using data from the same national longitudinal cohort study that is reported in this study, the number of children’s books in the home at age 3 accounted for a greater proportion of differences in vocabulary skills (25%) than shared reading at age 3 (16%; McNally et al., 2019). Further, in a familial effects study, access to books was the truly environmental literacy factor that predicted reading ability whereas the effect of shared reading was considered a masked genetic effect after controlling for parental reading fluency (van Bergen et al., 2017).

Investigating the role of early access to books in the home in children’s later reading attainment is often confounded by differences in how the availability of children’s books in the home is measured across studies. For example, the number of children’s books in the home has been analyzed as part of a storybook exposure factor (along with frequency of shared reading; Sénéchal, 2006) and conversely as a separate home learning variable (Shahaeian et al., 2018). However, literacy researchers have posited that access to books is a unique component in literacy development, with Krashen proposing that access to books is pivotal in supporting literacy development through increasing the likelihood of “free voluntary reading,” where children “read because they want to” (Krashen, 2004, p. 1). The unique effect of access to books on reading outcomes above and beyond other environmental literacy variables and genetic factors (van Bergen et al., 2017) also highlights the importance of investigating the unique impact of the number of books in early childhood on children’s later reading attainment.

Bio-ecological and transactional models of reading and child motivation

Most research examining the association between literacy practices at home and early childhood literacy outcomes assumes the former drives the latter. However, bio-ecological (Bronfenbrenner & Morris, 2007) and transactional (Sameroff, 2009) models of development consider parent and child practices to be reciprocal and mutually reinforcing. For instance, elementary-school children who demonstrate greater interest in literacy elicit more literacy-related experiences from parents, which in turn promotes children’s literacy skill development (L. Baker et al., 1997; Wigfield & Guthrie, 1997). In a bio-ecological model of development, shared book reading can be viewed as a proximal process in young children’s microsystems through which language development is supported; in this process children also have the potential of influencing parental behavior such as shared reading (Grolig, 2020). A bio-ecological model of oral language development through shared book reading thus posits that child and adult characteristics play a role in this proximal process (Bronfenbrenner & Morris, 2006 in Grolig, 2020). Transactional models not only consider proximal processes such as parent-child interaction but also consider more distal processes such as family sociodemographic factors, highlighting the fact that developmental outcomes are consistently impacted by both individual and contextual factors (Sameroff, 2010). In the current study, we adopt such theoretical constructs to explore relations between early literacy practices, sociodemographic factors such as family socioeconomic status, and children’s reading vocabulary through middle childhood.

It is also possible that genes and gene-environment correlations are associated with the ways in which children benefit from early shared book reading. However, low heritability of oral language skills, such as vocabulary, has been reported in early childhood, prior to school entry (Tosto et al., 2017), and shared environment plays a greater role in explaining oral language differences in early childhood than genetic factors (Hayiou‐Thomas et al., 2012). Regarding interest in reading, recent research has shown a longitudinal relation between parents’ book reading practices when children are preschool-age and measures of reading interest and reading ability in middle childhood (Demir‐Lira et al., 2019). In the current study, we add to this growing literature by using longitudinal path analyses to determine if children’s literacy interest at age 7 mediates the relation between early childhood literacy practices and children’s literacy outcomes at age 9.

Moderating effects of SES and gender

Main effects of SES on literacy outcomes are well-studied (Bus et al., 1995), yet moderating effects have received less attention. Focusing on potential moderating factors allow one to understand whether the home literacy environment is more strongly related to later literacy outcomes for some children more than others. Such a finding has practical importance as it highlights for whom early intervention may be most beneficial. Indeed, there is evidence of moderating effects of SES on the association between literacy experiences and children’s outcomes, such that shared reading has been shown to be more strongly related to early literacy indicators among low- to middle-SES families (Shahaeian et al., 2018). However, studies rarely have representative samples of different levels of SES to robustly test potential moderating effects (Sénéchal, 2006). The strength of the relations among home literacy experiences and later reading may also vary as a function of children’s gender: on average, girls score significantly higher on reading tests than boys (Lynn & Mikk, 2009; Reilly et al., 2016) and a rich home literacy environment may support later reading ability differently for boys and girls. As such, examining gender differences in the composition of the home literacy environment within a nationally representative sample may provide some explanation for later differences in literacy attainment between boys and girls.

Current study

The current study addressed five research questions. The first question explored literacy practices at age 3 and investigated the child and family characteristics that were associated with these practices. Our second research question asked whether shared book reading and availability of books at age 3 predicted reading vocabulary in middle childhood at age 9 (six years later). To address this question, we included data from parental reports of reading practices and children’s books in the home at age 3 and a direct assessment of children’s reading vocabulary scores at age 9.

Our third research question asked whether these associations would hold when accounting for important covariates at age 3, including parental (SES) and child characteristics (vocabulary skills) and the environment (cognitively stimulating activities such as rhymes, singing and drawing). We hypothesized that both shared reading and availability of books in the home at age 3 would predict reading vocabulary even after taking account of early known predictors of educational attainment.

In our fourth research question, we investigated potential mediators of the relations between our two predictors – shared reading and number of books – and reading vocabulary at age 9. Our prediction was that early shared reading would be associated with reading vocabulary in middle childhood through an increase in language and literacy skills at school entry, an increase in the likelihood of being read to at 5 years and the number of books in the home, and through enhancing children’s motivation as indicated by reading for pleasure at age 7. We predicted that access to children’s books at age 3 would affect reading vocabulary through an increased likelihood of being read to at 5 years, a greater number of books in the home at 5 years and enhanced reading for pleasure at age 7.

In our fifth and final question we asked whether associations would be moderated by children’s gender or by SES. Based on the limited literature available, we anticipated that gender and SES would moderate the relation between early shared reading and access to books and later reading vocabulary.

Method

Participants

Data from four waves of the national longitudinal study of children and youth, Growing Up in Ireland (GUI), a nationally representative longitudinal study of children residing in the Republic of Ireland were analyzed in this study. Infants were aged 9-months at the start of the study and data collection. In the current research we utilize information from the second wave (age 3), third wave (age 5), fourth wave (age 7/8), and fifth wave (age 9) of the study. Written informed consent was obtained from the child’s primary caregiver at each wave of the study. Materials and procedures for GUI were reviewed and approved by a dedicated Research Ethics Committee set up by the Department of Children and Youth Affairs in Ireland. Procedures relating to child protection were informed by the Children First: National Guidance for the Protection and Welfare of Children (Department of Children and Youth Affairs, 2011) as well as relevant Acts in Irish legislation, including the Data Protection Acts 1988, 2003 and the Statistics Act, 1993.

Data were collected throughout a series of interviews with the children’s primary caregiver and the child conducted in the children’s home by trained interviewers via Computer Assisted Personal Interviewing (CAPI), and Computer Assisted Sensitive Interview (CASI) for sensitive questions (McNamara et al., 2019). The child’s mother self-identified as the primary caregiver in 99.7% of cases. The majority of caregivers were White (94.3%), 2.4% were Black, 2.6% were Asian, and 0.7% self-reported as “other” or more than one ethnicity. The primary caregiver provided data on the frequency of shared book reading and the number of books in the home at age 3 (n = 9,789). Reading attainment was directly assessed by trained personnel at age 9 (n = 7,751). The sample for the current study consisted of participants with available reading test scores at age 9 (n = 7,751) and survey weights and maximum likelihood estimation are used to account for missing data, as described below.

Sample and sampling design

A full description of the GUI cohort and study design can be found elsewhere (e.g., Williams et al., 2013). Briefly, 11134 of the 41,185 infants born in Ireland between 1^st December 2007 and 30^th June 2008 were recruited using the Child Benefit Register (CBR) as a sampling frame. The CBR is highly suitable for use as a sampling frame as it includes close to full population coverage of all children residing in Ireland and up-to-date information on characteristics that can be used to model non-response. A simple systematic selection procedure based on a random start and constant sampling fraction was employed to sample participants (Williams et al., 2010). Data were re-weighted to account for any imbalances between the recruited sample and the overall population (for further information on weights see Quail et al., 2011 and supplementary file S1).

Materials and procedures

Reading vocabulary at age 9-years

When children were aged 9-years they completed the 40-item Reading Vocabulary test from the Drumcondra Primary Reading Test-Revised (DPRT-R; Educational Research Centre, 2007). The test was administered by trained interviewers during the in-home visit. The DPRT-R is a revised version of a long-standing reading attainment scale developed for Irish school children and linked to the national curriculum. Scores on the DPRT-R have been shown to correlate well with measures of cognitive ability such as the Drumcondra Reasoning Test (r = 0.63; McNamara et al., 2019) and other ability measures such as digit span and letter/number sequencing tasks (Hayes & Stewart, 2016). Internal consistency for the Reading Vocabulary Test was reported in a technical GUI report as Kuder-Richardson Formula 20 (KR20) = 0.93 (Williams et al., 2010). Children were administered the grade-appropriate version of the DPRT-R. A small portion of participants (3.5%) opted not to complete the test. Of those who completed the test, a third completed “Level 2” and 67% of children completed “Level 3” which corresponded to 2^nd and 3^rd class in the Irish primary school system respectively. The 40-item vocabulary measure took 20 minutes to complete and involved reading a sentence and selecting the meaning of an underlined word from four alternatives. For example, after reading the sentence “They had an anxious wait” children selected the meaning of the word “anxious” in this context from four options: (a) a lengthy, (b) an uneasy, (c) an unusual, and (d) a relaxed (McNamara et al., 2019). Raw scores from the test (i.e., the number of items correctly described from 0 to 40) constitute the unit of analysis in this paper.

Shared reading at age 3

Our main explanatory variable was the frequency with which the child was read to at age 3. Reading frequency was asked as part of a set of questions on activities that children do with their families at home. Primary caregivers were asked to reflect on the child’s usual patterns of activities and answer, “On how many days in an average week does anyone at home read to him/her?.” Responses were provided on an 8-point scale ranging from 0 to 7 days.

Number of children’s books in the home at age 3

We also examined the primary caregiver’s estimate of the number of children’s books the child currently has access to in their home, including library books. The number of books was assessed using a five-point scale, where 1 = None, 2 = Less than 10, 3 = 10 to 20, 4 = 21–30, and 5 = More than 30 books available to the child.

Covariates assessed at age 3

Child’s Sex. Binary variable categorized as male or female.

Parents in the Household. Whether the child is living in a one or two parent family was reported by the primary caregiver (binary variable categorized as 1 or 2).

Child’s First Language. The primary caregiver reported whether the child’s first language was English, Irish, or another language. This item was recoded into a binary variable indicating whether or not the child’s first language is English.

Parental Stress. The primary caregiver completed the six-item Parental Stressors Subscale of the Parental Stress Scale (PSS) (Berry & Jones, 1995). Parents responded to items including “the major source of stress in my life is my child” and “having a child leaves little time and flexibility in my life” using a 5-point Likert scale ranging from (1) “strongly disagree” to 5 “strongly agree.” Scores range from 6 to 30 with higher scores indicating higher levels of stress.

Primary Caregiver Education. The primary caregiver reported their highest level of educational attainment on a 13-point ordinal scale ranging from “no formal education” to “Doctorate.” This scale was reduced to four categories: lower secondary schooling or less (a maximum of 11 years of formal education, similar to a GED in the United States), higher secondary schooling (13 to 14 years of formal education, equivalent to a high school degree or diploma), certificate/diploma (14 to 15 years of formal education, equivalent to an associate’s degree), degree or postgraduate (a minimum of 16 years of formal education, equivalent to a four-year college degree or higher).

Household Income. Disposable family income was calculated as total household income less statutory deductions of income tax and social insurance contributions. In the current study, we used equivalized household income, calculated as the disposable household income divided by equivalized household size (i.e., accounting for differences in size and composition of households in terms of the number of adults and children per household) (Quail et al., 2011). Data were converted to income quintiles for analysis.

Vocabulary. Expressive Vocabulary was measured using the Naming Vocabulary test from the British Abilities Scales II (Early Years) (Elliott et al., 1997). Children were shown pictures of everyday objects and asked to name the object. The test authors (Elliott et al., 1997) report high internal reliability (α = .86) for the Naming Vocabulary scale at ages 3–3.5 years and a correlation of .68 with the Verbal IQ score on the Wechsler Preschool and Primary Scale of Intelligence – Revised (Weschler, 1989) (Elliott et al., 1997).

Home Learning Activities. Primary caregivers were asked to reflect on the child’s usual pattern of activities at the moment and to indicate the frequency with which anyone at home does the following activities with the child, which also support language and literacy development: (i) help him/her learn the ABC or alphabet, (ii) help him/her learn numbers or counting, (iii) help him/her learning songs, poems, or nursery rhymes, or (iv) play games [e.g. board games, jigsaws, card games] with him/her. Responses were provided on an 8-point scale ranging from 0 to 7 days.

Mediator variables

Shared Reading at Age 5-Years. The frequency of reading to the child was assessed by the question, “How often would you do any of the following?” followed by “Read to him/her.” Responses were provided on a 5-point scale ranging from 1 = Never, to 5 = Every day.

Number of Children’s Books in the Home at 5-Years. The primary caregiver indicated the number of children’s books currently available to the child in their home, including library books. Answers were provided on a five-point scale, ranging from 1 = None, to 5 = More than 30.

Teacher Ratings of Child’s Language and Reading Skills at 5-Years. Teachers were asked to rate the study child’s skills as part of the Teacher-on-Child questionnaire. The language and reading subscales were derived using questions from the Millennium Cohort Study, Age 5 survey and based on the Foundation Stage Profile in England, and assess the study child’s school achievement as reported by their teacher. Scores from both scales were closely related (r = 0.53, p < .001) and were combined to form a single indicator of teacher-rated language and reading skills.

Reading for Pleasure at Age 7/8-Years. The primary caregiver indicated how often the child “Reads for pleasure by themselves” on a five-point scale, ranging from 1 = Never, to 5 = Every day.

Statistical analysis plan

Analyses were carried out in Mplus (Muthén & Muthén, 2017) (see supplementary analysis). Sample characteristics are described using weighted estimates of means and standard deviations, or proportions for each of the variables as appropriate at age 3. The associations of shared book reading and the number of children’s books in the home with covariates at 3-years (i.e. primary caregiver characteristics, child’s sex, first language, vocabulary assessment scores, cognitively stimulating parent-child activities) were first estimated using linear regression. The associations between shared book reading at age 3 and the number of children’s books in the home and performance on the DPRT-R Reading Vocabulary scale scores at 9 years were then estimated using a series of linear regression models. First, we adjusted for background sociodemographic and family background factors: child’s sex, child’s first language, parental stress, education level, and household income. Next, we included an adjustment for the child’s expressive vocabulary score as assessed at age 3. Finally, we adjusted for the frequency of engagement in home learning activities (e.g. board games & jigsaws, songs & nursery rhymes, alphabet, counting) at age 3. To facilitate the interpretation of effect sizes, the DPRT-R Reading Vocabulary scores were standardized to have a mean of zero and standard deviation of one in all analyses.

Formal mediation analysis was used to test the indirect effects of shared reading and the number of books in the home at age 3 on reading vocabulary scores at age 9 via four potential mediators: teacher ratings of the child’s language and reading skills at age 5, the frequency of shared reading at age 5, the number of books in the home at age 5, and the frequency with which the child reads for pleasure at age 7/8 (see Figure 1). A non-parametric bootstrapping approach with 95% bias-corrected bootstrapped confidence intervals estimated using bootstrap resampling (N = 10,000 bootstrap samples) to test whether indirect effects were statistically different from zero.

Figure 1. Conceptual diagram of a path model linking the frequency of reading to the child and the number of children’s books in the home at 3 years to reading vocabulary at 9 years via language and reading skills, frequency of reading to child, and the number of books accessible in the home at 5 years, and reading for pleasure at 7/8 Years.

Finally, as a sensitivity test, we examined whether the association between shared reading at age 3 and vocabulary at age 9 was moderated by background socioeconomic status (i.e., parental education level and household income at age 3) or by child gender.

Missing data

The current study sample was participants with available reading vocabulary test scores at age 9 (n = 7,751). Missingness due to non-response and inter-wave attrition was handled using GUI survey weights which takes into account differential nonresponse patterns within population subgroups and differential attrition between survey waves. Application of the age 9 survey weights has been shown to ensure the distribution of key child and family characteristics (e.g., child gender, primary caregiver’s education level, number of parents present, income quintiles) matches that of the target population (McNamara et al., 2019). While survey weights were used to ensure the representativeness of the data was maintained, there was also a relatively small portion of missing data on the baseline and mediator variables (6.1% of data missing in total). Missing data on these variables was handled using full-information maximum

likelihood estimation (FIML). Rather than impute missing values, this approach utilizes values from all data including partially completed cases to fit a statistical model to incomplete data. FIML operates as if missing data were replaced by probable values implied based on the observed data from across the study variables. This information is incorporated into the likelihood function and used to estimate point estimates and standard errors. FIML has been shown to perform equivalently to multiple imputation in handling missing data (Lee & Shi, 2021) and to yield less biased and more efficient estimates than traditional approaches such as listwise and pairwise deletion (Enders & Bandalos, 2001).

Finally, we examined interactions between shared reading and books in the home at age 3 and child gender and background socioeconomic status (i.e., maternal education level and household income at 9 months) in predicting vocabulary at age 9. This provided a test of whether these the associations between shared reading and books in the home and subsequent vocabulary scores differed for males and females and those from low/high socioeconomic status families.

Results

Sample characteristics

Baseline child and primary caregiver characteristics are displayed in Table 1. In line with the Irish population, the first language of almost all children in the sample (93.3%) was English. Almost 16% of households were one-parent families. A minority of primary caregivers reported having lower second-level education or less (15.1%) and in a sizable portion of families (30.4%) the primary caregiver had an undergraduate degree or higher. Primary caregivers indicated that household members frequently engaged with the child in a range of stimulating activities such as learning the alphabet, numbers, or songs/poems/nursery rhymes, or playing board games/jigsaws/card games (between 3.9 and 5.2 days per week on average).

Table 1. Sample baseline characteristics at 3 years.

Variable	Category	% or mean (SD)	n for variable
Shared reading (days per week)		5.5 (2.01)	7,499
Number of children’s books	None	0.6%	7,501
in the home	Less than 10	7.2%
	10 to 20	20.3%
	21 to 30	18.3%
	More than 30 books	53.7%
Child’s Sex	Girl	49.1%	7,502
	Boy	50.9%
Parents in the home	Two parents	84.2%	7,502
	One parent	15.8%
Parental Stress	PSS Parental Stressors Subscale	12.4 (4.15)	7,407
Primary Caregiver Education	Lower secondary or less	15.1%	7,491
	Upper secondary	33.1%
	Associate’s degree	21.5%
	Degree/Postgrad	30.4%
Equivalised Household Income Quintiles	Lowest	20.3%	7,138
	2^nd	20.6%
	3^rd	20.5%
	4^th	19.2%
	5^th	19.4%
English is child’s first lang.		93.3%	7,501
Expressive Vocabulary	BAS Naming Vocabulary Test	74.96 (19.4)	7,187
Cognitively stimulating activities (days per week)
Learn the ABC with child		3.9 (2.4)	7,499
Learn numbers with child		5.2 (1.9)	7,499
Learn songs with child		5.2 (2.1)	7,497
Play board games/jigsaws /card games with child		4.4 (2.2)	7,497

Note: Weighted estimates are presented.

Literacy practices at age 3

First, we examined household engagement in literacy practices at age 3. Primary caregivers reported that shared reading practices were common with someone reading to the child 5.5 (SD = 2.01) days per week on average (see Table 1). While engagement in shared book reading was high on average, occurring every day in 57.2% of households, there was a sizable minority of households (19.5%) where the study child was read to 3 days a week or less. Similarly, the number of children’s books in the home was high on average (M = 4.2 [SD = 1.0] on a 5 point-scale) with 53.7% of caregivers reporting more than 30 books in the home. In 8.4% of households the primary caregiver reported less than 10 children’s books in the home.

Predictors of literacy practices at age 3

Next, we estimated linear regression models of the association between child and family background characteristics and shared reading and the number of children’s books in the home at age 3. The results of this analysis are reported in Table 2. Child gender and parental stress were unrelated to shared reading frequency or the number of children’s books in the home. Shared reading frequency and the number of books in the home were positively related to the education level of the primary caregiver. Notably, the child was read to approximately 0.9 more days per week (B = 0.88, SE = 0.12, p < .001) where the primary caregiver had a degree compared to lower secondary education or less. Similarly, household income was positively related to the frequency of shared book reading (B = 0.15, SE = 0.02, p < .001) and the number of books in the home (B = 0.09, SE = 0.01, p < .001). Children whose first language was English were read to more frequently (B = 0.41, SE = 0.14, p < .01) and had a larger number of books in the home than others (B = 0.40, SE = 0.07, p < .001). The frequency of shared reading and the number of books in the home were higher among children with higher BAS Naming Vocabulary scores and in households where members engaged more frequently in cognitively stimulating activities such as learning numbers, songs, games with the child.

Table 2. Associations between sample characteristics and shared reading and the number of children’s books in the home at 3 years (N = 7,502).

	Shared book reading (days per week)	Num. of children’s books in home^a
Variable	B(SE) (SE)	B (SE)
Female	.02(.05) (0.05)	0.03 (0.02)
Single-parent household	.03(.10) (0.10)	0.28*** (0.05)
Parental stress	−.01(.01) (0.01)	−0.01 (0.004)
Primary caregiver education^a
Upper secondary	.30*(.12) (0.12)	0.32*** (0.06)
Associate’s degree	.59***(.12) (0.12)	0.51*** (0.06)
Degree/Postgrad	.88***(.12) (0.12)	0.60*** (0.06)
Household Income quintiles	.15*** (.02)	0.09*** (0.01)
English first language	.41**(.14) (0.14)	0.40*** (0.07)
Expressive Vocabulary	.01***	0.01***
	(.002)	(0.001)
Cognitively stimulating parent-childactivities^c
Alphabet/ABC	.05***(.01) (0.01)	−0.02* (0.01)
Numbers/counting	.13***(.02) (0.02)	0.02* (0.01)
Songs/poems/rhymes	.12***(.02) (0.02)	0.03*** (0.01)
Play games	.15***(.01) (0.01)	0.06*** (0.01)

Weighted estimates are presented.

^aAssessed using a five-point scale: 1 = None, 2 = Less than 10, 3 = 10 to 20, 4 = 21–30.

and 5 = More than 30 books available to the child.

^bReference category for comparison is “lower secondary or less”.

^cMeasured in days per week activity is engaged in.

* p < .05. ** p < .01. *** p < .001.

Shared reading and number of books at age 3 as predictors of reading vocabulary at age 9

Using a hierarchical linear regression model, we estimated the association between shared book reading and the number of children’s books in the home at age 3 and reading vocabulary at age 9. Both the frequency of shared reading (B = 0.05, SE = 0.01, p < .001) and the number of books in the home at age 3 (B = 0.08, SE = 0.01, p < .001) were positively associated with vocabulary scores at age 9, after adjustment for primary caregiver education level, household income and other child and family background factors, as shown in Table 3. Further adjustment for naming vocabulary at age 3 did not markedly lessen the positive association between shared reading at age 3 and vocabulary scores at age 9 (B = 0.04, SE = 0.01, p < .001). Adjustment for vocabulary scores at baseline did attenuate the link between the number of books in the home at age 3 and vocabulary scores at age 9 (B = 0.05, SE = 0.02, p < .01). Further adjustment for cognitively stimulating parent-child activities at age 3 did not attenuate the association between vocabulary at age 9 and shared reading and number of books at age 3 (see Table 3).

Table 3. Associations between shared reading and the number of children’s books in the home at 3 years and reading vocabulary scores at 9 years (N = 7,751).

	Reading vocabulary at nine years (z-score)
		+ Age 3 vocabulary score	+ Age 3 cognitive activities
Variable	B (SE)	B (SE)	B (SE)
Shared reading^b	0.05*** (0.01)	0.04*** (0.01)	0.04*** (0.01)
Num. children’s books	0.08*** (0.02)	0.05** (0.02)	0.05* (0.02)
Female	0.04 (0.03)	−0.02 (0.03)	−0.02 (0.03)
Single-parent household	0.24*** (0.06)	0.21*** (0.06)	0.21*** (0.06)
Parental stress	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)
Primary caregiver education^b
Upper secondary	0.32*** (0.07)	0.29*** (0.06)	0.29*** (0.06)
Associate’s degree	0.47*** (0.07)	0.42*** (0.07)	0.42*** (0.07)
Degree/Postgrad	0.60*** (0.07)	0.54*** (0.06)	0.54*** (0.06)
Household Income quintiles	0.06*** (0.01)	0.05*** (0.01)	0.05*** (0.01)
English first language	0.35*** (0.06)	−0.02 (0.07)	−0.02 (0.07)
Expressive Vocabulary	–	0.01*** (0.001)	0.01*** (0.001)
Cognitively stimulating parent-childactivities^a
Alphabet/ABC	–	–	−0.01 (0.01)
Numbers/counting	–	–	0.00 (0.01)
Songs/poems/rhymes	–	–	0.00 (0.01)
Play games	–	–	0.01 (0.01)

Note: Weighted estimates are presented.

^aMeasured in days per week activity is engaged in.

^bReference category for comparison is “lower secondary or less”.

* p < .05. ** p < .01. *** p < .001.

Estimates from our most stringently controlled model indicate that vocabulary scores were 0.26 standard deviations (SD) higher at age 9 among children who were read to everyday (57.2% of participants) compared to those who were not read to at all at age 3 (2.7% of participants). Similarly, we estimate that children in households with 30 or more children’s books available at age 3 went on to have vocabulary scores 0.18 SD higher at age 9 than those with no children’s books available in the home at age 3.

The association between reading frequency at age 3 and vocabulary at age 9 was not moderated by child gender (B = −0.00, SE = 0.02, p = .84). Similarly, the link between the number of books in the home at age 3 and subsequent vocabulary scores at age 9 was not moderated by child gender (B = 0.00, SE = 0.04, p = .93). There was also no evidence that the links between either reading frequency or the number of books in the home and vocabulary scores at age 9 were moderated by household income or the education level of the primary caregiver (all interactions p > .05).

Direct and indirect effects of shared book reading and number of books at age 3 on reading vocabulary at age 9

Next, we examined the potential pathways through which shared book reading and the number of books in the home at age 3 are associated with vocabulary scores at age age 9. The proposed pathways are detailed in Figure 1. Table 4 shows that neither shared reading nor the number of books in the home were directly related to vocabulary levels at age 9 after adjustment for the proposed mediating variables (language/reading skills at age 5, frequency of shared reading and number of books in the home at age 5, frequency of reading for pleasure at age 7/8 years).

Table 4. Role of mediating variables in explaining the association between shared reading and number of books in the home at 3 years and reading vocabulary at 9 years (n = 7,751).

	Shared book reading at 3 years and vocabulary at 9 years		Number of books at 3 years and vocabulary at 9 years
Coefficient	B	95% CI	B	95% CI
Total effect^a	0.037***	[0.019, 0.056]	0.045*	[0.008, 0.078]
Direct effect^b	0.013	[−0.002, 0.033]	−0.002	[−0.041, 0.035]
Indirect effect^c	0.023***	[0.014, 0.033]	0.048***	[0.022, 0.071]
via Language/reading skills at 5y	0.006*	[0.001, 0.010]	0.002	[−0.007, 0.011]
Shared reading at 5y	0.000	[−0.005, 0.004]	0.000	[−0.001, 0.001]
Number of books at 5y	0.003**	[0.001, 0.005]	0.028**	[0.008, 0.046]
Reading for pleasure at 7/8y	0.010***	[0.005, 0.017]	0.014*	[0.001, 0.026]
Language/reading at 5y and reading for pleasure at 7/8y	0.000	[−0.001, 0.002]	0.000	[−0.001, 0.001]
Shared reading at 5 yr and reading for pleasure at 7/8y	0.004***	[0.002, 0.005]	0.001	[0.000, 0.002]
Number of books at 5 yr and reading for pleasure at 7/8y	0.000	[0.000, 0.001]	0.004	[−0.002, 0.010]

Note: All estimates are derived from a single path model and are adjusted for covariates at 3 years (child’s sex, primary caregiver education level, stress level, household income, whether English is the child’s first language, whether one or two parent household, BAS Naming Vocabulary Score, frequency of engagement in cognitively stimulating parent-child activities (e.g. board games & jigsaws, songs & nursery rhymes, alphabet, counting).

^aTotal association between shared reading and number of books and vocabulary scores.

^bPart of the association between shared reading and number of books and vocabulary scores not explained by mediators.

^cPart of the association between shared reading and number of books and vocabulary scores that is explained by the mediating variables.

* p < .05. ** p < .01. *** p < .001.

The total indirect effect of the mediating variables in explaining the association between shared reading and vocabulary scores was highly statistically significant (B = 0.023, SE = 0.005, p < .001) and explained 64% of this association. A series of pathways combined together to explain this overall indirect effect. Children who were read to more frequently at the age of 3 went on to be more likely to read for pleasure at age 7/8 (B = 0.05, SE = 0.01, p < .001), which in turn predicted higher vocabulary levels at age 9 (B = 0.21, SE = 0.02, p < .001). This indirect effect of reading for pleasure (B = 0.010, SE = 0.003, p < .001) explained 28% of the total association between reading at age 3 and vocabulary at age 9 (see Table 4).

The frequency of shared reading at age 3 was also positively related to teacher-ratings of language/reading skills at age 5 years (B = 0.04, SE = 0.02, p < .05) which directly predicted higher vocabulary scores at age 9 (B = 0.14, SE = 0.01, p < .001). The indirect effect of language/reading skills at 5 years was statistically significant (B = 0.006, SE = 0.002, p < .01) and explained approximately 17% of the association between shared reading and vocabulary scores. There was also a complex statistically significant indirect effect (B = 0.004, SE = 0.001, p < .001) of both shared reading at 5 years and reading for pleasure at age 7/8 years that explained 11% of the overall link. Specifically, shared reading at age 3 predicted higher levels of shared reading at age 5 (B = 0.11, SE = 0.01, p < .001), which in turn predicted reading more often for pleasure at age 7/8 (B = 0.15, SE = 0.03, p < .001), which was associated with higher vocabulary scores at age 9 (B = 0.21, SE = 0.02, p < .001). Finally, households that engaged in higher levels of shared reading at age 3 went on to have a larger number of books in the household at age 5 (B = 0.04, SE = 0.01, p < .001) and this predicted high vocabulary scores at age 9 (B = 0.07, SE = 0.02, p < .01). The indirect effect of the number of books at age 5 (B = 0.003, SE = 0.001, p < .01) explained 8% of the overall shared reading – vocabulary association.

The association between the number of children’s books in the home at age 3 and vocabulary scores at age 9 was fully accounted for by the mediating variables (indirect effect: B = 0.048, SE = 0.012, p < .001). The indirect effect was due to two paths. First, there was a positive link between the number books in the home at age 3 and the number of books at age 5 (B = 0.42, SE = 0.02, p < .001) which predicted reading vocabulary levels at age 9 (B = 0.07, SE = 0.02, p < .01). This indirect effect (B = 0.028, SE = 0.009, p < .001) explained 61% of the link between the number of books in the home and subsequent reading vocabulary levels. Second, those with a larger number of children’s books in the home at age 3 were more likely to read for pleasure at age 7/8 (B = 0.07, SE = 0.03, p < .05) which positively predicted vocabulary scores at age 9 (B = 0.21, SE = 0.02, p < .001). This indirect effect (B = 0.014, SE = 0.006, p < .05) could explain a further 30% of the association between the number of books and subsequent vocabulary scores.

Discussion

In a nationally representative longitudinal study, we found that early shared reading and availability of books predicted higher reading scores at age 9, even after accounting for important child and family characteristics at age 3. This finding supported our first hypothesis. However, through mediation analysis we showed that shared reading and the number of books in the home at age 3 were not directly related to vocabulary levels at age 9 after adjustment for four key mediating variables: children’s language and reading skills at age 5, frequency of shared reading and number of books in the home at age 5, and frequency of reading for pleasure at age 7/8. These mediating variables explained 64% of the association between early shared reading and later vocabulary, and just two variables (number of books at age 5 and reading for pleasure at age 7/8) fully accounted for the association between early availability of books and later vocabulary. Such findings complement prior work with nationally representative samples in other countries by examining associations over at least a five-year period and identifying potential moderating factors which can explain the association between the home literacy environment and future reading achievement (e.g., Brown et al., 2022; Shahaeian et al., 2018). Our findings thus point to related but unique pathways through which early shared reading and early access to books are related to later reading vocabulary, as discussed in detail in the following section.

Early shared reading and pathways to later reading vocabulary

A large proportion of the association between early shared reading and reading vocabulary in middle childhood was explained by a series of combined pathways incorporating language and literacy variables at age 5 and reading for pleasure at age 7/8. Over a quarter of the association (28%) between early shared reading and later vocabulary was explained by an increased likelihood of reading for pleasure at age 7/8 which in turn predicted higher reading scores at age 9. There was also a complex effect whereby early shared reading predicted greater frequency of shared reading at age 5 that in turn predicted reading more often for pleasure at age 7/8, and which predicted higher vocabulary scores at age 9. This pathway accounted for a further 11% of the association. Together these findings indicate the importance of early shared reading in supporting later reading vocabulary through an increased interest in and motivation to read books, supporting and extending similar findings in smaller, in-depth longitudinal studies of early reading and language development (Demir‐Lira et al., 2019). These findings confirm prior studies using nationally representative samples, which have highlighted the importance of early shared reading for academic skills (Brown et al., 2022; Farrant & Zubrick, 2013; Shahaeian et al., 2018). Importantly, these results extend prior work by offering interest and motivation as a potential pathway through which the home environment is related to long-term literacy outcomes. Given that motivation to read for pleasure declines during primary school, and may thus exacerbate individual differences in reading outcomes (Wigfield & Guthrie, 1997), our findings support a closer look at shared book reading and the way in which children engage in this proximal process to identify ways to support motivations for reading in early childhood.

Increased motivation to engage in reading offers further multiple opportunities for learning new words and complex language necessary for supporting literacy skills (National Early Literacy Panel, 2009). Indeed, this was also a pathway through which early shared reading predicted later reading vocabulary, that is, through more sophisticated language and literacy skills at age 5 (which is the average age of school entry in Ireland). The types of joint attention opportunities afforded by shared reading have been noted for some time in the research literature (Dunham et al., 1993): through shared reading, parents and educators are afforded a highly motivating context through which to sustain children’s attention in language-rich interactions, as well as to support the acquisition of vocabulary and complex language (Akhtar et al., 1991). Almost a fifth of the association between early shared reading and later vocabulary was accounted for through positive effects on vocabulary and reading skills at 5 years. Linguistic input in early shared reading activities has been shown to be more complex and sophisticated than in non-reading interactions (Demir‐Lira et al., 2019; Montag et al., 2015), supporting the acquisition of novel words. Building vocabulary is a key process by which early shared reading has been shown to support reading development: vocabulary knowledge is essential for both reading comprehension and developing phonological awareness (Metsala, 1999, 2011). Our findings show that the importance of early shared reading both in supporting language and literacy skills at school entry and in increasing children’s interest in reading for pleasure which in turn predict children’s reading vocabulary scores at age 9.

Finally, early shared reading was indirectly associated with later reading vocabulary via an increased availability of children’s books in the household at age 5, with eight percent of the association explained by an increased access to books at age 5. This pathway, in which early shared reading predicted children’s reading vocabulary six years later through increased access to books, further supports the application of transactional models (Sameroff, 2010) to reading motivation: parents who read to their young children are more likely to provide greater access to books at school entry, in turn supporting children’s interest in books and motivation to engage in early literacy activities. We discuss early access to books at age 3 next as an important predictor for reading vocabulary in middle childhood.

Access to books and reading attainment in middle childhood

The association between access to books at age 3 and reading vocabulary scores at age 9 was fully accounted for through an increased number to books at age 5 and a greater likelihood of reading for pleasure at age 7/8. The finding that 61% of the association was explained through an increased number of books at age 5 suggests that having books available to young children at age 3 increases the likelihood of access to books at school entry and that this access to books is a positive and effective way to support reading in middle childhood. A book-rich environment may affect reading attainment through a “scholarly” culture in the home which has been found to be more predictive of adolescents’ academic performance (as assessed using a combined measure of reading skills) than parental education (Evans et al., 2014). In this view of access to books at home, books are seen as indicative of valuing learning, further enhancing children’s motivation to read throughout childhood (Evans et al., 2014). This finding suggests that the large-scale book distribution programs in other countries such as the United States and Great Britain may also be beneficial in Ireland. For instance, a recent meta-analysis found children whose families received books via book distribution programs scored higher on later measures of reading motivation and literacy during early elementary school (De Bondt et al., 2020). Moreover, our results suggest that these programs (e.g., BookStart; Reach Out and Read) should target families with very young children – under age 3 – for maximum benefits on later literacy outcomes.

The transactional motivational model is further supported by our finding that an additional 30% of the association between early access to books and later reading vocabulary was explained through increased reading for pleasure at age 7/8. Availability of books at age 3 predicted children’s reading for pleasure four years later which in turn predicted reading attainment two years later at age 9. Within a transactional motivational model, access to books serves to enhance motivation to read which in turn offers more opportunities to develop literacy skills through increased exposure to books and print. Thus, access to books as early as age 3 has downstream effects for reading achievement and motivation.

Limitations

We relied on parental self-report of reading practices and number of books in the home which is likely to contain some bias as these activities are positively viewed by parents such that they are known to be important and valued. The limitations of single-item measures which are negatively skewed should also be highlighted: single-item measures are likely to have lower content validity, sensitivity, and reliability. However, this is a problem faced by large cohort studies which often rely on single-item, self-report measures where representative samples of the population are sought. We observed variation in parental self-report on the home literacy practices which indicates that, although skewed, these measures indicate parental practices at home on these key variables.

Similarly, this large cohort study does not include observations of the quality of shared reading interactions. Given the importance of reading for pleasure as a key variable in explaining the associations between early reading and later attainment, a measure of the responsivity of parents in this type of activity would be especially valuable and should be included in future studies. Data on the quality of the linguistic input during these activities is also not available and would be captured in observational studies of reading activities. Observations of the quality of reading interactions are time-intensive to collect and process, making them difficult to include in cohort studies with large samples. However, balancing these limitations, cohort studies offer a nationally representative picture of early literacy practices in the home and includes children from all SES, allowing for control of important characteristics and to robustly identify mechanisms which partly explain the positive impact of early practices on reading attainment in middle childhood. The national standardized assessment of reading attainment in this study is also a limitation as it is specific to Ireland and is not easily generalizable. However, it is a robust measure of reading vocabulary skills for the cohort under study.

Finally, the extent to which the home literacy variables measured in this study are truly environmental must also be noted as a potential limitation. As reviewed in the Introduction, recent familial and twin studies suggest that shared reading effects on literacy outcomes may mask genetic effects of parental reading ability (Hart et al., 2021; van Bergen et al., 2017). Nevertheless, studies that control for parental reading ability report that the effect of the number of books in the home on children’s reading attainment is likely to be a truly environmental effect, and the number of books in the home may serve as a proxy for a culture that values books and thus supports literacy outcomes (van Bergen et al., 2018). To address this potential limitation, we conducted supplementary analysis to control for stable individual characteristics over time. Taken together, these supplementary analyses suggest that time-invariant differences between children, such as genetic differences, are unlikely to explain the link between storybook exposure and reading and language skills and that there is evidence that storybook exposure predicts subsequent changes in vocabulary over time.

Conclusion

Our findings support investigating shared reading (activity) and availability of books (environmental resource) as unique predictors that are positively associated with reading attainment through different pathways. Common to both pathways is an association with an increase in reading for pleasure at age seven/eight and an increased number of books in the home at age five. However, the effect of early shared reading was also explained through sustained parental engagement in shared reading at 3 and 5 years which positively impacted children’s reading for pleasure and then reading attainment and through supporting vocabulary and early literacy skills. Our findings thus highlight the provision of books in early childhood as an effective way of supporting reading attainment but highlight the added value of early shared reading in increasing the likelihood of later shared reading and reading for pleasure, and the provision of multiple opportunities for supporting early vocabulary and literacy skills.

Acknowledgments

The Growing Up in Ireland (GUI) data have been funded by the

Government of Ireland through the Department of Children and Youth Affairs and have been collected under the Statistics Act, 1993, of the Central Statistics Office. The authors would like to thank the joint ESRI-TCD Growing Up in Ireland study team who designed and implemented the project, and the children and families who took part.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.