Sources of individual variation in Estonian toddlers’ expressive vocabulary

Abstract

The vocabulary size of 16- to 30-month-old children (N = 1235) was assessed using the Estonian adaptation of the MacArthur–Bates Communicative Development Inventory: Words and Sentences (ECDI-II). The relationship between children’s expressive vocabulary size and different factors of the child and his/her social environment was examined. Results confirm the findings of studies from other languages showing that girls have larger vocabularies than boys, and first-born children are at an advantage compared to later-born children. Children of highly educated parents have bigger vocabularies than children whose parents have a lower educational level; this effect was significant for both maternal and paternal education. It was found that children of less-educated fathers who spend a high amount of time per week (> 40 hours) in daycare have smaller vocabularies than children of fathers with higher educational attainment.

Keywords

Estonian language development MacArthur-Bates CDI parental report vocabulary

In this article we present normative data for the Estonian adaptation of the MacArthur–Bates Communicative Development Inventory: Words and Sentences (ECDI-II) and report on the relationship between Estonian toddlers’ vocabulary skills and various characteristics of the child and his/her social environment. The main focus is on the variability of expressive vocabulary, which has been thoroughly researched in various language and cultural contexts (e.g. Bates et al., 1994; Eriksson et al., 2012; Le Normand, Parisse, & Cohen, 2008; McNally & Quigley, 2014; Stokes & Klee, 2009; Trudeau & Sutton, 2011; Westerlund & Lagerberg, 2008). The studies demonstrate that across these contexts the developmental pathways are similar: vocabulary development is characterized by high variability, a vocabulary spurt and the subsequent vast growth of vocabulary. Expressive vocabulary is one of the most important indicators of development at the second year of life (Van Hulle, Goldsmith, & Lemery, 2004). Previous studies have shown that language development at these early stages is significantly associated with later cognitive and language development (Botting & Conti-Ramsden, 2000) as well as psychosocial development later in adulthood (Schoon, Parsons, Rush, & Law, 2010). This highlights the importance of studying the sources of individual variation.

It is widely acknowledged that certain characteristics of the child and the child’s social environment, such as the child’s gender (Gleason & Ely, 2002), birth-order (Pine, 1995) and parental educational level (Hoff, 2003), contribute to explaining this variation. Yet some of the findings regarding the effects of the characteristics of the social environment have been inconsistent across studies conducted in different language communities and socio-cultural contexts and vary based on the age of the children studied, as well as methods of assessment, which are discussed in more detail as follows.

Sources of individual variation

A number of studies have investigated gender-related variability in early vocabulary development. It has been consistently demonstrated that girls compared to boys are at an advantage in emerging language skills (Berglund, Eriksson, & Westerlund, 2005; Eriksson et al., 2012; Feldman et al., 2000; Fenson et al., 1994; Huttenlocher, Haight, Bryk, Seltzer, & Lyons, 1991; Hyde & Linn, 1988; Le Normand et al., 2008; Schults, Tulviste, & Konstabel, 2012; Simonsen, Kristoffersen, Bleses, Wehberg, & Jørgensen, 2014; Stokes & Klee, 2009; Wehberg et al., 2008; Zambrana, Ystrom, & Pons, 2012). Several studies with methodological approaches ranging from parental-report measures of child language to speech sampling of toddlers’ verbalizations show that from about the second year of life girls’ vocabularies tend to grow much more quickly than boys (Bornstein, Hahn, & Haynes, 2004). According to some researchers, this might be due to the differences in boys’ and girls’ physiological development and maturation (Le Normand et al., 2008). It is also explained by the differences in parental input to their sons and daughters (Hoff & Naigles, 2002; Huttenlocher et al., 1991). There have been some indications in previous research that gender differences at the first stages of vocabulary acquisition do not warrant the need for separate norms for boys and girls (Kern, 2007). For example, Fenson and colleagues (1994) argue that even though there is a slight advantage for girls, the variance in early language skills that is explained by gender is small (1–2%).

Another frequently analysed variable associated with variation in language growth is birth-order (Berglund et al., 2005; Fenson et al., 1994; Hoff-Ginsberg, 1998; Jones & Adamson, 1987; Oshima-Takane & Robbins, 2003; Oshima-Takane, Goodz, & Derevensky, 1996; Reese et al., 2015; Zambrana et al., 2012). The findings about the effect of birth-order on language outcomes are controversial. For example, it has been suggested that first-born children are at an advantage in some aspects of their speech (e.g. clausal diversity, Huttenlocher, Waterfall, Vasilyeva, Vevea, & Hedges, 2010), have bigger vocabularies (Gayraud & Kern, 2007; Hoff-Ginsberg, 1998) and learn 50 words a month earlier than their second-born siblings (Pine, 1995). By contrast, Oshima-Takane and her colleagues (1996) compared spontaneous language data of 32 children and found that second-born children are more advanced in pronoun production, while not differing in general language development compared to their first-born siblings. A reason for the divergent findings may lie in the various measures used to assess children’s language abilities in these studies. For example, Jones and Adamson (1987) found birth-order related differences in children’s language skills using parental-report measures but not with spontaneous speech samples.

Hoff-Ginsberg (1998) has shown that mothers use longer utterances with their first-born rather than later-born children. Berglund and her collaborators (2005) point out that, while later-born children have to compete with their siblings for their mothers’ attention, this can be alleviated with the attention they get from their first-born siblings, in addition to learning from the overheard conversations between caregivers and older siblings (Oshima-Takane et al., 1996). It is likely that there are differences in the social environment of the first-born and later-born child but it is uncertain how these differences in the environment influence language development.

Family socio-economic status (SES) – most frequently indexed by maternal education (Hoff, 2006) in studies of child language – is another widely examined predictor (Rowe, 2008). In their study, Huttenlocher and colleagues (2010) found that SES is a highly significant predictor of language, with either maternal education or family income as the SES measure. According to their explanation the influence of SES on language growth is mediated by the differences in caregiver speech. It has been suggested that mothers with higher educational levels offer their children a richer environment in which different words can be acquired (Hoff, Laursen, Tardif, & Bornstein, 2002). General knowledge and beliefs about children’s development as well as parenting and conversational style have been found to be the factor mediating the effect of parental education on language development (Hoff, 2003; Rowe, 2008). For example, it has been shown that highly educated mothers talk more and use a richer vocabulary, use less directive speech, ask more questions, answer their children’s questions more frequently and start conversations more often with their children as compared to mothers with lower educational attainment, who are more focused on regulating their children’s behaviour (Hoff et al., 2002; Hoff-Ginsberg, 1998). Similarly, Fernald, Marchman, and Weisleder (2013) found that at 18 months, children from lower-SES families had significantly lower vocabulary scores and they were less efficient in real-time processing of language than children from higher-SES families at the same age. Since different cultures vary based on their socio-economic structure, the extent of the influence SES has on early vocabulary development might vary across cultures (Berglund et al., 2005).

In recent years more attention has been paid to the second caregiver in the family – the father. Pancsofar, Vernon-Feagans, and the Family Life Project Investigators (2010) argue that the greater emphasis on fathers stems from the changing roles of fathers in the lives of their children, with less stereotyped roles of fathers, and increasing cultural and linguistic diversity among fathers (Pancsofar, Vernon-Feagans et al., 2010). This reflects in the social environment of the child, which is associated with children’s developmental outcomes. It is widely acknowledged that sensitive, nurturing and cognitively stimulating interactions between fathers and their children are associated with children’s linguistic and cognitive abilities (Cabrera, Fagan, Wight, & Schadler, 2011; Cabrera, Shannon, & Tamis-LeMonda, 2007) as well as social and emotional development (Cabrera et al., 2007). Tamis-LeMonda, Shannon, Cabrera, and Lamb (2004) explain that fathers who are more educated and economically secure have more positive exchanges with their children, thus providing an opportune environment for the child’s vocabulary to grow. For example, Tamis-LeMonda and her colleagues (2004) found that fathers’ as well as mothers’ supportive parenting predicted children’s outcomes, whereas fathers’ education and income were directly associated with child measures and indirectly via the quality of mother–child engagements.

Daycare facilities offer another different learning environment for young toddlers where their socio-communicative skills develop. From the point of view of emerging language skills it is necessary to receive a lot of child-directed and referential language at the early stages of language development. In studies of language development and non-parental care (Berglund et al., 2005; National Institute of Child Health and Human Development [NICHD], 2000), it has been proposed that childcare quality – indexed by number of employees per child, employee’s educational level – is the most important variable associated with language development (Berglund et al., 2005). High-quality care occurs when caregivers provide frequent warm, stimulating and responsive interactions with children, as well as clear intentional instruction (Li, Farkas, Duncan, Burchinal, & Vandell, 2013). It is uncertain if a daycare employee is able to pay enough personal attention to each of the 14 children (the average number of children per adults in Estonian daycare groups) and engage every child in one-to-one conversations throughout the day which would result in the growth of vocabulary. Previous studies have reported that children who experience high-quality care in early childhood have higher cognitive, language and pre-academic skills prior to school entry (Li et al., 2013) as well as a reduced risk for late talking (Lekhal, Zachrisson, Wang, Schjølberg, & von Soest, 2011). In a large-scale longitudinal study with Dutch preschoolers, Luijk and colleagues (2015) investigated associations between non-parental care and language development of preschoolers, using the Dutch CDI as one of the assessment methods. Their analyses showed that after controlling for the effect of other variables (i.e. ethnicity, SES, gender, parity) more time spent in non-parental care is associated with better language skills. They also found that more hours spent in non-parental care in the first year of life are associated with a drop in language proficiency at 1;0–1;6 years of age, but this effect was reversed at later age and followed by language acceleration in preschool children (Luijk et al., 2015). In another example, Bleses and her collaborators (2008a) analysed Danish 0;8 to 1;8 year olds’ daycare attendance, based on the amount of hours children spend at daycare in a week, in relation to children’s language comprehension scores on the Danish CDI. They found a high and consistent decrease in the comprehension scores from children who attend daycare a few hours per week to children who attend a medium number of hours per week and children who attend daycare a high number of hours per week, but no significant difference between children who do not attend daycare and those who do (Bleses et al., 2008a). These studies suggest that daycare attendance benefits some aspects of children’s language proficiency at certain stages of development.

Communicative Development Inventories

Development of parent-report measures (i.e. MacArthur–Bates Communicative Development Inventories; Fenson et al., 1994) in language acquisition research has provided an important tool for gathering data from large representative samples in a cost-efficient manner. Longitudinal studies have shown the predictive validity of the parental-report measures as well as the stability of the CDI scores across time and different assessment methodologies (Bates, Dale, & Thal, 1995; Gatt, Grech, & Dodd, 2014; Perez-Pereira & Resches, 2011). For example Ring and Fenson (2000) compared the parent-reported CDI score with laboratory assessments of children’s expressive and receptive language and found a significant correspondence between the two different approaches. In a more recent study, Gatt et al. (2014) reported that children’s CDI score is significantly related to word use elicited through picture naming and vocabulary sampled naturalistically through play. Fenson et al. (1994) reason that parents have broad knowledge of their children’s language abilities and parental-report measures can give a better overview of the range of children’s linguistic competencies, because they are based on parents’ knowledge and experience of their children’s behaviour across different situations. Although there are some limitations related to parental reports, such as report bias – some parents might overestimate their children’s skills, others underestimate – Bates et al. (1995) conclude that parental reports give an accurate view of the variation and developmental changes from 8 to 30 months of age.

Due to their high reliability and validity as well as applicability in research and clinical settings, the Communicative Development Inventories have now been adapted for up to 60 different languages across the world (Dale & Penfold, 2011). Data from a wide range of languages help understand early language development. The current study makes a contribution to the broader documentation of variation in early language development by compiling data from a less studied language.

The Estonian language belongs to the Finnic branch of the Finno-Ugric languages and is spoken by approximately 1 million people. Estonian is an agglutinative type of language and it is characterized by a large number of cases (14 productive cases), no grammatical gender (either of nouns or personal pronouns), no articles (either definite or indefinite) and differentiation between three quantities both in vowels and consonants (see Sutrop, 2004 for detail). The adaptation of the ECDI-II (Schults, Tulviste, & Haan, 2013; Tulviste, 2007) was based on the ECDI-I (Schults et al., 2012; Tulviste, 2007). Both adaptations followed closely the original American versions (Fenson et al., 1994), and the Polish (Smoczynska, unpublished material) adaptations of the CDI. In constructing the inventory the main purpose was to follow the example of the original CDI vocabulary checklist by keeping the same semantic categories and the amount of words roughly the same. Due to cultural differences a number of words were added and/or replaced across the vocabulary checklist. In piloting the ECDI-II, additional information was gathered from the parents about words their children use most frequently. Based on this information and data analysed following the CHILDES conventions (MacWhinney, 2000) from Estonian mother–child spontaneous speech samples, an adapted version of the ECDI-II was created, with 631 words in 21 semantic categories.

Current study

Based on the literature reviewed above, the main aim of our study is to determine the predictors of expressive vocabulary development in a sample of Estonian 16- to 30-month-old toddlers. Our first hypothesis assumes that age is the main predictor of vocabulary scores at ages 16–30 months. Our second hypothesis states that girls, first-born children and children of highly educated mothers and fathers have a bigger vocabulary compared to boys, later-born children and children of less-educated parents, recruited respectively. Lastly, the third hypothesis assumes that children who spend more time at daycare have smaller expressive vocabularies compared to children who do not attend daycare.

Method

Participants

This study is based on data drawn from a large number of children who made up the standardization sample for the Estonian adaptation of the MacArthur–Bates Communicative Development Inventory: Words and Sentences (ECDI-II; Schults et al., 2013; Tulviste, 2007). The distribution of the sample by age and gender is given in Table 1. Children with serious health problems were excluded from the sample. Children who have contact with other languages besides Estonian are included in the sample (N = 131). An ANCOVA with age as a covariate did not reveal significant differences in the vocabulary size of children who are in contact with other languages compared to children who are brought up in an Estonian-only language environment.

Table 1.

Distribution of participants by age and gender.

Age in months	Boys	Girls	Total
16	17	19	36
17	47	42	89
18	51	52	103
19	41	44	85
20	24	36	60
21	38	28	66
22	31	45	76
23	47	48	95
24	64	52	116
25	37	45	82
26	39	45	84
27	36	44	80
28	48	44	92
29	41	59	100
30	34	37	71
Total	595	640	1235

The sample consisted of 1235 16- to 30-month-old native Estonian speakers of whom 48% were boys and 52% girls. The distribution of the sample according to birth-order, parental education and daycare attendance is given in Table 2. Among the sample 60% of children were first-born (N = 718), whereas 474 children were born second or later (40%). With regard to parental education, 42% of the mothers (N = 498) and 32% of the fathers (N = 352) had acquired tertiary education. This is reflective of the educational distribution among the Estonian population. Based on the OECD adult education level indicator, 38.4% of 25- to 64-year-old Estonians have acquired tertiary education, which is one of the highest among the EU member states (Organisation for Economic Co-operation and Development [OECD)], 2015). The indicator is especially high for Estonian women (46.8%), while it is 29.4% for Estonian men.

Table 2.

Children’s observed and adjusted mean expressive vocabulary size by gender and factors of the child’s social environment.

Variables		N	Obs. mean (SD)	Adj. mean (SEM)
Gender	Male	595	192.08 (164.47)	194.74 (5.33)
	Female	640	258.19 (177.30)	255.72 (5.13)
Birth-order	First-born	721	230.69 (175.80)	232.03 (4.97)	40 missing
	Later-born	474	219.59 (173.07)	217.56 (6.13)
Maternal education	Low	321	193.47 (160.96)	189.56 (9.17)	61 missing
Maternal education	Primary	89	185.94 (160.79)	197.84 (13.73)
	Vocational	232	196.36 (161.27)	189.02 (8.51)
	Medium	355	224.00 (170.21)	217.32 (7.74)
	Secondary	200	211.89 (166.02)	193.81 (9.15)
	Post-secondary studies	155	238.51 (174.98)	245.07 (10.41)
	High	498	247.25 (179.48)	250.05 (6.33)
	Tertiary education	419	245.52 (180.32)	251.27 (6.31)
	Graduate degree	79	261.28 (180.44)	271.84 (14.57)
Paternal education	Low	388	194.42 (157.87)	207.71 (6.74)	123 missing
Paternal education	Primary	118	166.28 (148.09)	185.09 (12.01)
	Vocational	270	206.72 (160.67)	210.81 (7.93)
	Medium	372	221.53 (170.78)	215.49 (6.87)
	Secondary	246	216.08 (170.96)	206.07 (8.26)
	Post-secondary studies	126	231.72 (172.58)	236.05 (11.61)
	High	352	261.24 (185.72)	233.73 (9.56)
	Tertiary education	294	264.89 (186.24)	256.08 (7.59)
	Graduate degree	58	247.97 (186.75)	269.04 (17.11)
Daycare attendance	Not attending	401	167.53 (153.28)	219.47 (6.86)	48 missing
Daycare attendance	Attending	786	255.44 (176.46)	228.94 (4.81)
	1–24 hours per week	233	216.64 (178.42)	254.95 (8.68)	42 missing
	25–40 hours per week	374	288.27 (175.33)	232.93 (7.07)
	> 40 hours per week	137	255.73 (160.74)	188.52 (11.45)
Home language environment	Estonian	1055	226.95 (174.67)	228.41 (4.05)	49 missing
Home language environment	Other languages	131	220.68 (167.89)	208.95 (11.49)

Notes: Obs. mean = observed mean; Adj. mean = mean vocabulary adjusted by age.

Approximately 66% of the 16- to 30-month-olds attended daycare at the time of assessment (N = 783). On average the children started daycare at 18 months of age (SD = 7.5), they spend 4 days of the week (SD = 1.54) and on average 6.4 hours per day (SD = 2.93) in daycare; 33% of the children are at daycare for approximately 25–40 hours (N = 373). This is quite representative of the Estonian population. Based on the data from 2013, 75% of Estonian 1- to 6-year-olds attend daycare or preschool institutions with approximately 20% of them being 2 years of age or under and the distribution of boys to girls is fairly equal (48% vs 51%). (Statistical Yearbook of Estonia, 2014).

Material and procedure

The ECDI-II, covering 16–30 months of age, is a parental-report measure of children’s semantic and grammatical development (Eriksson et al., 2012; Schults et al., 2013; Tulviste, 2007). It consists of two major parts – expressive vocabulary checklist and early grammar.

The participants were recruited through childcare facilities, health clinics, parenting groups and websites by posting advertisements. Parents were asked to fill in the ECDI-II form and a background information sheet containing questions about children’s developmental history, socio-demographic characteristics of the family and language(s) spoken in the home.

Parents were presented with a list of 631 words in 21 semantic categories (see Table 3) and asked to mark the words their children produce on their own. The parent was instructed to mark the words even if the child had used a different pronunciation or a synonym for the word represented in the form. If the parent was unsure whether or not his/her child uses certain words they were instructed to observe their child for a couple of days and then fill in the form. The forms were mostly filled in by mothers, with a few instances where fathers or both parents had noted themselves as filling in the form. Table 3 shows the number of items in the semantic categories and the scale reliabilities (Cronbach’s alpha) for each category in order of appearance in the vocabulary checklist section of the ECDI-II.

Table 3.

Summary of number of words and scale reliabilities for each of the semantic categories in order of appearance in the vocabulary checklist of the ECDI-II.

CDI category	No. of words	Alpha	CDI category	No. of words	Alpha	CDI category	No. of words	Alpha
Sound effects	15	.856	Body parts	26	.961	Action words	82	.989
Animals	47	.971	Furniture	34	.967	Direction words	24	.954
Vehicles	14	.913	Adverbs of time	15	.929	Question words	7	.85
Toys	19	.929	Small household items	49	.978	Conjunctions	8	.845
Clothing	30	.959	Things outside	32	.965	Quantifiers	9	.827
People	23	.909	Places	22	.922	Descriptive words	67	.979
Food and drink	69	.982	Games and routines	27	.925	Pronouns	12	.898

Analyses

First, we examined the internal consistency of the ECDI-II vocabulary checklist. Descriptive statistics with observed and adjusted vocabulary scores (by age in months) for the whole sample are given in Table 2. Chi-square tests were used to analyse associations between the factors of the child and his/her social environment. Finally multiple regression analyses were carried out to determine the predictors of expressive vocabulary size. Age (in months) was treated as a continuous variable; gender was treated as a categorical predictor with two values (boys – 0, girls – 1). Birth-order was dichotomized into a variable with two values (first-born children – 0, second or later-born children – 1). Parental education (maternal and paternal separately) was coded into variables with three categories each (0 – primary/vocational, 1 – secondary/post-secondary studies, 2 – tertiary education/graduate degree) to represent low, medium and high educational level respectively. Daycare attendance was coded as a variable with four categories based on the average hours per week children spend at daycare (0 – children not attending daycare, 1 – at daycare 1–24 hours per week, 2 – at daycare 25–40 hours per week, 3 – at daycare more than 40 hours per week). For the regression equation parental educational level and daycare attendance were treated as categorical variables and were therefore dummy-coded into several binary variables. In addition to observing the main effects of these variables, we included a full set of interactions into the final regression equation.

Results

Reliability

Reliability of the ECDI-II vocabulary checklist was examined for the entire sample based on the internal consistency of scores for each of the semantic categories. Calculations of Cronbach’s alpha yielded a coefficient of .961 overall, which shows high internal consistency and is comparable to previously published values (Berglund & Eriksson, 2000; Bleses et al., 2008b; Simonsen et al., 2014) including the CDI normative study (Fenson et al., 1994). Thus, the ECDI-II appears to be a reliable measure of vocabulary development in normally developing Estonian toddlers 16–30 months of age (Figure 1). Percentile scores for boys’ and girls’ expressive vocabulary size across five age groups where age is scored in 3-month intervals are given in Table 4.

Figure 1.

Percentile scores of expressive vocabulary size by children’s age (grouped in 3-month intervals) and gender.

Table 4.

Percentile scores for boys and girls’ expressive vocabulary size on the ECDI-II across five age groups in 3-month intervals.

Boys% Rank	16–18 mo	19–21 mo	22–24 mo	25–27 mo	28–30 mo
N	115	103	142	112	123
Mean	54	106	191	240	352
SD	48.24	106.24	146.22	152.8	147.7
5	6	9	14	33	98
10	9	15	24	51	126
25	18	28	56	106	229
50	39	62	160	228	376
75	71	162	285	363	486
90	112	293	426	472	532
95	158	352	467	511	557
Girls% Rank	16–18 mo	19–21 mo	22–24 mo	25–27 mo	28–30 mo
N	113	108	145	134	140
Mean	87	154	244	341	412
SD	85.87	117.75	160.09	141.45	138.81
5	8	20	24	104	100
10	15	28	47	140	230
25	33	49	104	243	347
50	58	133	226	350	446
75	116	230	348	448	508
90	179	350	497	528	560
95	238	387	542	563	580

Predictors

Chi-square tests were used to examine the relations between age, gender, birth-order, daycare attendance and parental educational level (Table 5). Significant associations were observed between age and daycare attendance with the number of daycare attendees growing as children get older. Additionally, daycare attendance was significantly related to both maternal and paternal education. Among children who attend daycare and whose mothers have a high educational level most of the children are at daycare 1–24 hours per week (41%, N = 128). Among the low and medium maternal education categories, the proportion of children who are at daycare 1–24 hours is approximately 20% (N = 43 for low maternal education; N = 57 for medium maternal education). Based on paternal education, the distribution is similar, with highly educated fathers having children who spend less time in daycare (43%, N = 94), compared to children of fathers with low or medium educational levels where approximately 70% of the children are at daycare from 25 up to 40 or more hours per week (N = 171 for low paternal education, N = 172 for medium paternal education). Lastly we observed a significant association between maternal and paternal education, with mothers and fathers having spouses with similar educational attainment. While there are observed significant associations between the factors of interest, tests to see if the data met the assumption of collinearity indicated that multicollinearity was not a concern. The Variance Inflation Index (VIF) for all the factors of interest was close to 1.

Table 5.

Associations between predictor variables age, gender, maternal and paternal educational level, daycare attendance and birth-order (chi-square tests).

	Age	Gender	Maternal education	Paternal education	Daycare attendance
Gender	12.05
d.f.	14
Maternal education	23.75	0.19
d.f.	28	2
Paternal education	39.97	2.26	260.29*
d.f.	28	2	4
Daycare attendance	275.53*	6.60	26.52*	21.46*
d.f.	42	3	6	6
Birth-order	23.57	1.69	0.33	0.49	7.01
d.f.	14	1	2	2	3

p < .05.

Sources of individual variation

A hierarchical multiple linear regression analysis was carried out to ascertain the extent to which child’s age, gender, birth-order, parental educational level (maternal and paternal separately) and daycare attendance predicted children’s expressive vocabulary size. In the first step we included age, gender and birth-order in the regression equation, and obtained a statistically significant model explaining 46% of the variance in expressive vocabulary (R² = .46, F_(3,1190) = 335.4, p < .001). In the next step we included the dummy-coded categorical predictors maternal and paternal education and daycare attendance into the regression equation, which significantly improved the predictive power of the regression model (R² = .49, F_(10,1190) = 114.6, p < .001). In the next step we included a full set of interactions among age, gender, birth-order, maternal and paternal education as well as daycare attendance into the regression equation, out of which only age by gender and paternal high educational level by daycare attendance more than 40 hours per week appeared as significant predictors of children’s expressive vocabulary. Since the age by gender interaction introduced multicollinearity into the model, we opted to remove the age by gender interaction from the final regression model, because the effect of these variables is already represented by age and gender as separate predictors. The age by gender interaction indicates that gender differences become more pronounced as the children grow older and their vocabularies grow larger. There is a levelling off at the end of the observed age range, due to a ceiling effect among children at 30 months of age (see Figure 1). To facilitate interpretation, we only added the paternal education and daycare attendance interaction term into the final model. This yielded a model explaining 50% of the expressive vocabulary size (R² = .50, F_(12,1190) = 98.6, p < .001). The inclusion of the interaction term into the regression equation accounted for an additional 0.01% of the variance. The raw and standardized coefficients for the predictors in the three observed regression models with R² change statistics are presented in Table 6.

Table 6.

Summary of hierarchical multiple regression analysis examining the main and interaction effects of child’s age, gender, birth-order, maternal and paternal education and daycare attendance in predicting expressive vocabulary size.

Variable	Model 1			Model 2			Model 3
	B	SE	β	B	SE	β	B	SE	β
Constant	−431.37	21.53		−484.35	22.96		−483.97	22.78
Age	27.08	.89	.65***	27.77	.97	.67***	27.89	.97	.67***
Gender	61.76	7.45	.18***	62.89	7.27	.18***	64.94	7.22	.19***
Birth-order	−15.72	7.61	−.04*	−16.33	7.41	−.05*	−16.22	7.36	−.05*
Maternal education ^a
Medium				21.27	9.76	.06*	20.96	9.69	.06*
High				46.71	9.68	.13***	46.8	9.60	.13***
Paternal education ^a
Medium				4.11	8.95	.11	3.87	8.88	.01
High				28.04	9.78	.07**	15.1	10.12	.04
Daycare attendance ^b
1–24 h per week				15.96	10.22	.04	17.16	10.14	.04
25–40 h per week				9.47	9.55	.03	7.86	9.48	.02
> 40 h per week				−40.69	12.93	−.07**	−73.08	14.74	−.13***
Two-way interaction ^c
> 40 h per week in daycare × Paternal education (High)							112.29	25.16	.11***
R²	.459			.493			.501
F for change in R²	335.41***			11.26***			19.92***

Notes: *p < .05; **p < .01; ***p < .001. ^aMaternal and paternal education were represented as two dummy variables respectively, with low educational level (primary, vocational education) serving as the reference category. ^bDaycare attendance was represented as three dummy variables with children not attending daycare as the reference category. ^cOnly the significant interaction term is presented here.

Age was the strongest predictor of expressive vocabulary size, explaining most of the variation in the sample. However gender, maternal and paternal education and birth-order still accounted for a significant amount of the variance. Gender differences were statistically significant with girls scoring on average 65 words more than boys. In total gender accounted for approximately 2% of the variance in expressive vocabulary size. Children whose mothers have acquired a university degree have significantly bigger vocabularies than children whose parents have a low educational level even when age, gender and birth-order are controlled for. On average, high maternal education accounted for 1% of the total variance in expressive vocabulary size. Children who attended daycare more than 40 hours per week had on average a smaller vocabulary compared to children who do not attend childcare, which accounted for an additional 1% of the variance in expressive vocabulary size. Both results were highly statistically significant (p < .001). In addition to observing the main effects of our explanatory variables, we observed interactions among the predictor variables and found that daycare attendance for more than 40 hours per week predicts children’s vocabulary as a function of paternal education. Figure 2 presents the unstandardized predicted vocabulary scores for children according to their father’s educational level and daycare attendance. It can be observed that the biggest differences relating to paternal education are among children who attend daycare more than 40 hours per week, out of whom children of highly educated fathers score on average 350 words on the CDI (SD = 97, N = 38) and children of fathers with a low educational level score 131 words less (mean = 219, SD = 96, N = 98). These effects are significant after accounting for the effect of other explanatory variables.

Figure 2.

Predicted mean scores of expressive vocabulary size as a function of paternal educational level and daycare attendance.

Discussion

This study examined the relations between child’s age, gender, factors of the child’s social environment and children’s expressive vocabulary in a representative sample of Estonian 16- to 30-month-old toddlers. The results demonstrate a significant effect of age, gender, birth-order and parental education as well as daycare attendance on toddlers’ vocabulary size. Our data corroborate the findings of earlier studies showing that vocabulary grows rapidly from 16 to 30 months of age, indicating that the ECDI-II is a sensitive measure of communicative development at this early age range. Another important finding was that Estonian girls scored consistently higher than boys, first-born children had bigger vocabularies than later-born children and children of highly educated parents had bigger vocabularies than children whose parents had lower educational attainment; this effect was significant for both maternal and paternal education. Daycare attendance also revealed an interesting link with children’s vocabulary, showing that daycare attendance is associated with vocabulary size, more specifically spending a high amount of time (> 40 hours) in daycare per week seems to result in a smaller vocabulary compared to children not attending daycare.

Our results are in accordance with those of Trudeau and Sutton (2011): in their study with 16- to 30-month-old children acquiring Quebec French, age accounted for the highest amount of variance in children’s expressive vocabularies, but gender, maternal education and birth-order were also significant predictors. While the effect of daycare setting, paternal education and whether or not the child lives with one or two parents did not account for any additional variation in Trudeau and Sutton’s study (2011), paternal education and daycare attendance (as well as their interaction) were found to be significantly related to the variance in the present investigation of expressive vocabulary of Estonian children.

In our sample, girls had larger vocabularies than boys in almost every age group. This result is in accordance with a number of studies conducted with the same instrument in various language communities (Berglund et al., 2005; Eriksson et al., 2012; Le Normand et al., 2008). Although Fenson and his collaborators (1994) report an advantage in favour of girls in their analysis, they argue that gender accounts for only 1–2% of the variance in children’s vocabulary. Similarly, our data indicated a relatively small advantage for girls that accounted for about 2% of the variance in Estonian children’s expressive vocabulary size. Significant gender-related differences suggest the use of separate norms for Estonian girls and boys. For example, the 50th percentile of the expressive vocabulary score for a 22- to 24-month-old is approximately 193 words. But for girls the 50th percentile score at this age is at 226 words and for boys, 160 words. If we look at the 10th percentile, which is often used as a cut-off for children at risk of language delay (Dale, Price, Bishop, & Plomin, 2003), we see that for 22- to 24-month-old girls the 10th percentile score cut-off is at approximately 47 words, for boys at 24 words and for the total sample the 10th percentile is at 37 words. Using combined norms increases the risk of not noticing delay in girls’ early communicative development, therefore applying separate norms for ECDI-II-based assessments increases the sensitivity of ECDI-II and allows for more accurate estimation of the need for further detailed assessments of risk of language delay.

Our results also support the previously reported advantage of first-born children over later-born children in lexical productivity (Berglund et al., 2005; Pine, 1995). It is likely that it is because first-born children are subjected to more child-directed speech. In contrast, the conversational context of a later-born child might often include an older sibling, which results in fewer opportunities for joint attention between the caregiver and the child, which is facilitative of language growth at this early age.

Parental education, whether it be mothers’ or fathers’, appears an influential factor in these early stages of development. Our results indicate consistently that children whose mothers have acquired a post-secondary education are more advanced in their language abilities compared to children whose mothers have not attended university or college. Children whose mothers have secondary education or whose mothers are in the process of acquiring a university degree are also at an advantage compared to children whose mothers have primary or vocational educational levels.

Our results show that paternal education plays an important role in the early vocabulary development of their children. This result is in accordance with a wealth of studies on fathers’ involvement which show that interactions between fathers and their children are positively associated with child cognitive and language outcomes (Pancsofar, Vernon-Feagans et al., 2010; Sarkadi, Kristiansson, Oberklaid, & Bremberg, 2008; Tamis-LeMonda et al., 2004). While it has been consistently found that maternal and paternal educational levels explain the variation in child development, specifically language development (e.g. Hoff et al., 2002; Rowe, 2008), further analysis is needed on the specific measures of parents’ child-directed speech in order to determine which parental variables (e.g. speech measures, communicative style, the amount of talk, contingency of the speech and so on) are associated with child language outcomes.

Our findings revealed a significant interaction indicating that high daycare attendance per week relates to child’s expressive vocabulary as a function of paternal educational level. Children whose fathers have high educational attainment have bigger vocabularies, even if they spend a high amount of time per week at daycare, whereas children who spend the same amount of time in daycare but have less-educated fathers, have significantly smaller vocabularies. This result might indicate that families where fathers are highly educated can provide more stimulating at-home environments and communicative opportunities for their children which facilitate language acquisition, even after the long days children spend at daycare. It is likely that it is because their knowledge is more up to date with scientific approaches to development (see Hoff et al., 2002). This result is in accordance with several studies observing the associations of family SES and the early communicative environment of the child (Dollaghan et al., 1999; Huttenlocher et al., 1991; Zambrana et al., 2012). At the same time, the differences in expressive vocabulary related to parental education may also reflect the differences in how parents fill out the parental-report measure of their child’s vocabulary. Parental-report measures rely on the parents’ knowledge and abilities to recollect and notice their child’s growing vocabulary.

In order to analyse the effect of daycare on vocabulary development, we compared the vocabularies of children who do not attend daycare with those spending a different amount of hours per week in daycare. We looked specifically at the average time children spend at daycare per week and found that children who spend more than 40 hours per week at daycare have significantly smaller vocabularies. The findings differ from those of Luijk and colleagues (), who found in their study with Dutch 1- to 6-year-olds that more hours in non-parental care after the first year of life was associated with better language skills. In our study, no statistically significant differences emerged in the size of vocabulary between children who do not attend daycare and children who go to daycare 1–40 hours per week. At the same time, our findings imply that spending long days at daycare (over 40 hours) is a risk factor for children’s vocabulary development. This result is similar to what Bleses and her colleagues (2008a) found with Danish children, whose word comprehension decreased by the number of hours spent in daycare per week. Our study adds to an accumulating body of evidence indicating too much daycare is related to a smaller expressive vocabulary. Overall, these results show that parental decisions regarding their children’s daycare attendance might have far-reaching consequences. It is likely that these decisions are influenced by the family SES, which in turn determines the age of daycare admittance and the amount of time the child spends in daycare per week. Since the current study design is cross-sectional we cannot draw any conclusion on whether these effects will have a lasting impact on children’s language development. We also do not have data on the size of children’s vocabulary at daycare entry. It is possible that children with a bigger vocabulary at daycare entry interact more with their teachers and peers, and benefit more from attending daycare. Longitudinal designs are needed to study the long-term effects of early daycare attendance.

The current results have several implications. First, determining the importance of gender, birth-order and parental education factors in accounting for the variation of vocabulary development offers support to a substantive amount of previous research. Our findings indicate that different norms should be applied for Estonian boys and girls at ages 16–30 months. This can help childcare specialists assess children’s language abilities more accurately and determine the risk of possible language delays. The association between children’s language production and parental education has to be considered when developing strategies and interventions in order to educate parents from families with lower socio-economic status.

There are some limitations that have to be considered while interpreting the current results. For instance, the assessment of children’s expressive vocabulary development is based on parental reports, which have been shown to be a valid representation of children’s abilities (Bates et al., 1994; Gatt et al., 2014; Ring & Fenson, 2000) but they can be influenced by subjective components associated with the characteristics of the reporter, such as social desirability, value orientation and personality (Bornstein & Putnick, 2012). That is why it is beneficial to include various assessment methods to study child language at these early stages of development.

In conclusion, it can be stated that the vocabulary development of Estonian children has similarities with developmental trends observed in other languages and cultural contexts measured with the same instrument – the MacArthur–Bates Communicative Development Inventory. The results of our study extend the literature on the relationship between expressive language and factors of the child’s social environment by demonstrating the importance of age, gender, birth-order and parental education in predicting the growth of expressive vocabulary on a representative sample of Estonian toddlers. Furthermore, our study demonstrated the negative effect daycare attendance can have on expressive vocabulary size, by providing information on how children of fathers with lower education are at risk if they spend long days in daycare.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study received funding from the Estonian Science Foundation grant No. ETF9033.

References

Bates

Dale

P. S.

Thal

(1995). Individual differences and their implications for theories of language development. In Fletcher

MacWhinney

(Eds.), The handbook of child language (pp. 96–151). Oxford, UK: Wiley-Blackwell.

Bates

Marchman

Thal

Fenson

Dale

Reznick

J. S.

. . . Hartung

(1994). Developmental and stylistic variation in the composition of early vocabulary. Journal of Child Language, 21, 85–123.

Berglund

Eriksson

(2000). Communicative development in Swedish children 16–28 months old: The Swedish early communicative development inventory—Words and sentences. Scandinavian Journal of Psychology, 41, 133–144.

Berglund

Eriksson

Westerlund

(2005). Communicative skills in relation to gender, birth order, childcare and socioeconomic status in 18-month-old children. Scandinavian Journal of Psychology, 46, 485–491.

Bleses

Vach

Slott

Wehberg

Thomsen

Madsen

T. O.

Basbøll

(2008a). Early vocabulary development in Danish and other languages: A CDI-based comparison. Journal of Child Language, 35, 619–650.

Bleses

Vach

Slott

Wehberg

Thomsen

Madsen

T. O.

Basbøll

(2008b). The Danish communicative developmental inventories: Validity and main developmental trends. Journal of Child Language, 35, 651–669.

Bornstein

M. H.

Hahn

C.-S.

Haynes

O. M.

(2004). Specific and general language performance across early childhood: Stability and gender considerations. First Language, 24, 267–304.

Bornstein

M. H.

Putnick

D. L.

(2012). Stability of language in childhood: A multiage, multidomain, multimeasure, and multisource study. Developmental Psychology, 48, 477–491.

Botting

Conti-Ramsden

(2000). Social and behavioural difficulties in children with language impairment. Child Language Teaching & Therapy, 16, 105–120.

10.

Cabrera

N. J.

Fagan

Wight

Schadler

(2011). Influence of mother, father, and child risk on parenting and children’s cognitive and social behaviors. Child Development, 82, 1985–2005.

11.

Cabrera

N. J.

Shannon

J. D.

Tamis-LeMonda

(2007). Fathers’ influence on their children’s cognitive and emotional development: From toddlers to pre-K. Applied Developmental Science, 11, 208–213.

12.

Dale

P. S.

Penfold

(2011). Adaptations of the MacArthur–Bates CDI into non-US English languages. Retrieved from http://mb-cdi.stanford.edu/documents/AdaptationsSurvey2011.pdf

13.

Dale

P. S.

Price

T. S.

Bishop

D. V.

Plomin

(2003). Outcomes of early language delay in predicting persistent and transient language difficulties at 3 and 4 years. Journal of Speech, Language, and Hearing Research, 46, 544–560.

14.

Dollaghan

C. A.

Campbell

T. F.

Paradise

J. L.

Feldman

H. M.

Janosky

J. E.

Pitcairn

D. N.

Kurs-Lasky

(1999). Maternal education and measures of early speech and language. Journal of Speech, Language, and Hearing Research, 42, 1432–1443.

15.

Eriksson

Marschik

P. B.

Tulviste

Almgren

Pérez Pereira

Wehberg

. . . Gallego

(2012). Differences between girls and boys in emerging language skills: Evidence from 10 language communities. British Journal of Developmental Psychology, 30, 326–343.

16.

Feldman

H. M.

Dollaghan

C. A.

Campbell

T. F.

Kurs-Lasky

Janosky

J. E.

Paradise

J. L.

(2000). Measurement properties of the MacArthur communicative development inventories at ages one and two years. Child Development, 71, 310–322.

17.

Fenson

Dale

P. S.

Reznick

J. S.

Bates

Thal

D. J.

Pethick

S. J.

. . . Stiles

(1994). Variability in early communicative development. Monographs of the Society for Research in Child Development, 59, 1–185.

18.

Fernald

Marchman

V. A.

Weisleder

(2013). SES differences in language processing skill and vocabulary are evident at 18 months. Developmental Science, 16, 234–248.

19.

Gatt

Grech

Dodd

(2014). Early expressive vocabulary skills: A multi-method approach to measurement. First Language, 34, 136–154.

20.

Gayraud

Kern

(2007). Influence of preterm birth on early lexical and grammatical acquisition. First Language, 27, 159–173.

21.

Gleason

J. B.

Ely

(2002). Gender differences in language development. In McGillicuddy-De Li

De Lisi

(Eds.), Biology, society, and behavior: The development of sex differences in cognition (Vol. 21, pp. 127–154). Westport, CT: Ablex Publishing.

22.

Hoff

(2003). The specificity of environmental influence: Socioeconomic status affects early vocabulary development via maternal speech. Child Development, 74, 1368–1378.

23.

Hoff

(2006). How social contexts support and shape language development. Developmental Review, 26, 55–88.

24.

Hoff

Laursen

Tardif

Bornstein

(2002). Socioeconomic status and parenting. In Bornstein

M. H.

(Ed.), Handbook of parenting: Volume 2 – Biology and ecology of parenting (pp. 231–252). Mahwah, NJ: Lawrence Erlbaum.

25.

Hoff

Naigles

(2002). How children use input to acquire a lexicon. Child Development, 73, 418–433.

26.

Hoff-Ginsberg

(1998). The relation of birth order and socioeconomic status to children’s language experience and language development. Applied Psycholinguistics, 19, 603–629.

27.

Huttenlocher

Haight

Bryk

Seltzer

Lyons

(1991). Early vocabulary growth: Relation to language input and gender. Developmental Psychology, 27, 236–248.

28.

Huttenlocher

Waterfall

Vasilyeva

Vevea

Hedges

L. V.

(2010). Sources of variability in children’s language growth. Cognitive Psychology, 61, 343–365.

29.

Hyde

J. S.

Linn

M. C.

(1988). Gender differences in verbal ability: A meta-analysis. Psychological Bulletin, 104, 53–69.

30.

Jones

C. P.

Adamson

L. B.

(1987). Language use in mother–child and mother–child–sibling interactions. Child Development, 58, 356–366.

31.

Kern

(2007). Lexicon development in French-speaking infants. First Language, 27, 227–250.

32.

Le Normand

M. -T.

Parisse

Cohen

(2008). Lexical diversity and productivity in French preschoolers: Developmental, gender and sociocultural factors. Clinical Linguistics & Phonetics, 22, 47–58.

33.

Lekhal

Zachrisson

H. D.

Wang

M. V.

Schjølberg

von Soest

(2011). Does universally accessible child care protect children from late talking? Results from a Norwegian population-based prospective study. Early Child Development and Care, 181, 1007–1019.

34.

Farkas

Duncan

G. J.

Burchinal

M. R.

Vandell

D. L.

(2013). Timing of high-quality child care and cognitive, language, and preacademic development. Developmental Psychology, 49, 1440–1451.

35.

Luijk

M. P. C. M.

Linting

Henrichs

Herba

C. M.

Verhage

M. L.

Schenk

J. J.

. . . van IJzendoorn

M. H.

(2015). Hours in non-parental child care are related to language development in a longitudinal cohort study. Child: Care, Health and Development, 41, 1188–1198.

36.

MacWhinney

(2000). The CHILDES project: The database (Vol. 2). Mahwah, NJ: Lawrence Erlbaum.

37.

McNally

Quigley

(2014). An Irish cohort study of risk and protective factors for infant language development at 9 months. Infant and Child Development, 23, 634–649.

38.

National Institute of Child Health and Human Development Early Child Care Research Network. (2000). The relation of child care to cognitive and language development. Child Development, 71, 960–980.

39.

Organisation for Economic Co-operation and Development. (2015). Adult education level (indicator). Retrieved from https://data.oecd.org/eduatt/adult-education-level.htm

40.

Oshima-Takane

Robbins

(2003). Linguistic environment of secondborn children. First Language, 23, 21–40.

41.

Oshima-Takane

Goodz

Derevensky

J. L.

(1996). Birth order effects on early language development: Do secondborn children learn from overheard speech? Child Development, 67, 621–634.

42.

Pancsofar

Vernon-Feagans

, the Family Life Project Investigators (2010). Fathers’ early contributions to children’s language development in families from low-income rural communities. Early Childhood Research Quarterly, 25, 450–463.

43.

Perez-Pereira

Resches

(2011). Concurrent and predictive validity of the Galician CDI. Journal of Child Language, 38, 121–140.

44.

Pine

J. M.

(1995). Variation in vocabulary development as a function of birth order. Child Development, 66, 272–281.

45.

Reese

Ballard

Taumoepeau

Taumoefolau

Morton

S. B.

Grant

Perese

(2015). Estimating language skills in Samoan-and Tongan-speaking children growing up in New Zealand. First Language, 35, 407–427.

46.

Ring

E. D.

Fenson

(2000). The correspondence between parent report and child performance for receptive and expressive vocabulary beyond infancy. First Language, 20, 141–159.

47.

Rowe

M. L.

(2008). Child-directed speech: Relation to socioeconomic status, knowledge of child development and child vocabulary skill. Journal of Child Language, 35, 185–205.

48.

Sarkadi

Kristiansson

Oberklaid

Bremberg

(2008). Fathers’ involvement and children’s developmental outcomes: A systematic review of longitudinal studies. Acta Paediatrica, 97, 153–158.

49.

Schoon

Parsons

Rush

Law

(2010). Children’s language ability and psychosocial development: A 29-year follow-up study. Pediatrics, 126, e73–e80.

50.

Schults

Tulviste

Haan

(2013). Early vocabulary in full term and preterm Estonian children. Early Human Development, 89, 721–726.

51.

Schults

Tulviste

Konstabel

(2012). Early vocabulary and gestures in Estonian children. Journal of Child Language, 39, 664–686.

52.

Simonsen

H. G.

Kristoffersen

K. E.

Bleses

Wehberg

Jørgensen

R. N.

(2014). The Norwegian communicative development inventories: Reliability, main developmental trends and gender differences. First Language, 34, 3–23.

53.

Statistical Yearbook of Estonia. (2014). Retrieved from https://www.stat.ee/72570

54.

Stokes

S. F.

Klee

(2009). Factors that influence vocabulary development in two-year-old children. Journal of Child Psychology and Psychiatry, 50, 498–505.

55.

Sutrop

(2004). Estonian language (2nd ed.). Tallinn, Estonia: Eesti Instituut.

56.

Tamis-LeMonda

C. S.

Shannon

J. D.

Cabrera

N. J.

Lamb

M. E.

(2004). Fathers and mothers at play with their 2-and 3-year-olds: Contributions to language and cognitive development. Child Development, 75, 1806–1820.

57.

Trudeau

Sutton

(2011). Expressive vocabulary and early grammar of 16-to 30-month-old children acquiring Quebec French. First Language, 31, 480–507.

58.

Tulviste

(2007). Variation in vocabulary development among Estonian children as a function of child’s gender, birth order, child-care, and parental education. In Eriksson

(Ed.), Proceedings from the first European network meeting on the communicative development inventories (pp. 16–21). Gävle, Sweden: University of Gävle.

59.

Van Hulle

C. A.

Goldsmith

H. H.

Lemery

K. S.

(2004). Genetic, environmental, and gender effects on individual differences in toddler expressive language. Journal of Speech, Language, and Hearing Research, 47, 904–912.

60.

Wehberg

Vach

Bleses

Thomsen

Madsen

T. O.

Basbøll

(2008). Girls talk about dolls and boys about cars? Analyses of group and individual variation in Danish children’s first words. First Language, 28, 71–85.

61.

Westerlund

Lagerberg

(2008). Expressive vocabulary in 18-month-old children in relation to demographic factors, mother and child characteristics, communication style and shared reading. Child Care, Health and Development, 34, 257–266.

62.

Zambrana

I. M.

Ystrom

Pons

(2012). Impact of gender, maternal education, and birth order on the development of language comprehension: A longitudinal study from 18 to 36 months of age. Journal of Developmental & Behavioral Pediatrics, 33, 146–155.