Relations between reading skill level in fifth grade and functional language skills at toddler age

ABSTRACT

This retrospective study, undertaken in Norway, examines how reading skills level in 851 fifth graders are related to how the children used their language skills in play and everyday activities as toddlers. Data were collected with the Norwegian National Reading Test and through structured observations by staff in Early Childhood Education institutions using the TRAS (Early Registration of Language Skills) instrument. Children with low reading skills had lower language skills than the group with high reading skills for both TRAS total and the different TRAS sections; effect sizes varied from small to moderate. The group with moderate reading skills had better results for TRAS total than the group with low reading skills but weaker results than the group with high reading skills. In the TRAS sections, whether the moderate skills group differed from the other groups varied. The results imply the need for early intervention to start at toddler age.

KEYWORDS:

• Reading related to early language
• low reading skills
• level of reading
• authentic assessment
• early childhood education
• toddler’s language

Introduction

Reading is essential for learning and gaining information in school and has overarching significance for later academic success (Bleses et al. 2016; Longoria et al. 2009; Snow et al. 2014). It is frequently reported that language skills at an early age influence later reading skills (e.g. Bleses et al. 2016; Hjetland et al. 2017; Hjetland et al. 2019; Lee 2011; Muter et al. 2004). Weak reading skills are often posited as a later manifestation of previously weak language development (Nation 2005; Snowling et al. 2011). There is a large concern that weak language skills in the toddler period are significant for the acquisition of reading, which in turn is the most important factor for educational success (National Early Literacy Panel 2008). Most studies that examine relations between early language skills and later reading start when children are preschoolers (from 4 years of age) and follow their development into school age (Hjetland et al. 2017). However, the present study, undertaken in Norway, includes even younger children and is retrospective. It examines how the reading skill level in 851 fifth graders is related to how the children used their language skills in play and everyday activities (functional skills) as toddlers (children younger than 3 years of age). In this way, we contribute with knowledge about what characterises early functional language skills in children who develop low reading skills compared to what characterises early functional language skills in children who succeed in reading.

Background

It is well known that several language skills are important for developing reading skills. The meta-analysis by Hjetland et al. (2017) showed that the development of language comprehension and code-related skills in preschool age serve as a foundation for later reading comprehension in school. In addition, they found that earlier research often is limited to one or a few aspects of language development (such as vocabulary or language comprehension) and how these particular aspects influence later reading skills. It is well known that language production skills (vocabulary skills) are predictive of later reading skills (Lee 2011; Muter et al. 2004), and language comprehension at an early age is found to predict vocabulary growth (Fernald, Perfors, and Marchman 2006). Language comprehension represents the semantic aspect of language and develops prior to other language skills such as word production and linguistic awareness (Goswami 2008).

Marchman and Fernald (2008) found that language comprehension at 25 months predicts language and cognitive outcomes in school at age eight. Language awareness, at an early age, is another important aspect related to learning to read that involves reflecting on language, especially the skill to draw attention to sound structures in words (phonological awareness) and participation in language awareness activities (Frost 2003). Phonological awareness among preschoolers has been found to be positively correlated with reading in first grade (Dickinson, Nesbitt, and Hofer 2019). However, phonological skills are developed only to a small extent in toddlerhood, and prior research has shown that participation in language awareness activities is the language awareness skills most commonly mastered by toddlers (Stangeland, Lundetræ, and Reikerås 2018).

Language is important in children’s general development, especially for the development of social competence in children older than three years (Bishop 2014; Longoria et al. 2009) and in toddlers (Stangeland 2017). Social competence is often understood as the ability to build positive relationships and interactions with siblings, peers, parents, and other adults, including playing, sharing, and showing concern and sympathy with playmates (Bretherton et al. 2014; Warren, Yoder, and Leeww 2002). Aro et al. (2012) stressed that successful social functioning includes skills in communication and self-regulation. In addition, research has found that social skills are crucial for language acquisition (Beitchman et al. 1996; Henrichs et al. 2013; Kuhl, Tsao, and Liu 2003; Mackie and Law 2010). The study of Hjetland et al. (2019) shows that oral language skills are highly stable and are a critical foundation for the development of both decoding and reading comprehension skills from four to ten years of age. There is less research on the relation between reading skills and language skills in children younger than four years.

Children with weak language and later reading skills

Much research on the relation between early language skills and later reading has focused on children with language difficulties (Hjetland et al. 2017). Many children with language difficulties are not identified until they begin formal education Sim et al. (2013), although 5–7% of the children in Norwegian kindergartens are reported to have specific language difficulties (Helland 2012). Dale et al. (2003) found that delayed language in toddlerhood does not necessarily mean difficulties later. In their study, only 40% of the preschoolers who had vocabulary scores below the 10th percentile as toddlers met the criteria for persistent language difficulties two years later. However, it has been confirmed that toddlers with language delays have great variability in reading outcomes in school (Rescorla and Dale 2013).

It is found that late-talking toddlers (with vocabulary falling within the lowest 10% of the age level) are at a considerable risk for later reading problems (Bleses et al. 2016). However, according to Dale et al. (2014), approximately half of late-talking toddlers will have moved into the normal range at the end of the preschool period, at least on measures of vocabulary and grammar. These children are nevertheless reported to be at a continuing risk after this catch-up because their performance often remains below the mean (Dale et al. 2014). The longitudinal study of Bleses et al. (2008) examined how a large group of toddlers’ early productive vocabulary skills predicted later educational outcomes. They found only a weak degree of prediction according to Cohen’s guidelines (Cohen 1988) when examining the whole sample. For late talkers, however, they predicted academic achievement ten years later and showed that children with early delays are at elevated risk for later reading problems. On the other hand, Bartl-Pokorny et al. (2013) found no correlation between vocabulary skills and later reading. This might, however, be due to the relatively small sample size in the study.

Rationale

In the meta-analysis of the relation between early language skills and reading by Hjetland et al. (2017), only two of the studies started when the children were three years old, and none included toddlers. However, several researchers have reported that progress in language skills during the toddler period is characterised by complexity and extensive variation (Bloom 1993; Bretherton et al. 2014; Goswami 2008). Consequently, more studies about the relations between reading skills and language skills in toddler age are needed.

Earlier research on the relation between reading skills and early language skills is often linked only to one or a few aspects of language development (Hjetland et al. 2017), and there is a need for more studies including a wide range of language skills. This is taken into account in the present study, where interaction, communication, attention, language comprehension, linguistic awareness, pronunciation, word production, and sentence production are examined. This may reveal whether other aspects in early language development are important for later reading. Toddlers develop language skills in interaction with caretakers (Goswami 2008). Staff in ECEC institutions have a special responsibility to encourage language development as a base for later reading, as these caretakers are particularly involved in young children’s daily life situations. In Norway, 92.2% of children aged one to five years attend ECEC institutions (Statistics Norway 2019). Many countries have a tradition to test children at an early age to prevent later difficulties, however, in Norway there are reported resistance to examine and do interventions on children at an early age in ECEC institutions (Haug 2014). This is contrary to official Norwegian policy and expert recommendations for early intervention (Ministry of Education and Research 2006/2007, 2017), and underline the need to reveal eventual relations between level of reading skills and early language skills in a Norwegian context.

Earlier research on toddlers’ language skills is frequently based on information from parents (Dieterich et al. 2006; Suggate et al. 2018; Torppa et al. 2007) or from a mixture of parental reporting and testing conducted by researchers unknown to the children (Bleses et al. 2016; Uccelli et al. 2018). Few studies have been published with observations from staff in ECEC institutions (Hindman et al. 2008). Observation by persons familiar with the child is the best practice to get hold of how language skills are used in play and everyday activities (Bagnato et al. 2014), and inthis way the present study has a socio- cultural approach. This Norwegian study examines functional language skills for all children, not only children with weak language skills. As previously mentioned, several studies that refer to the link between language skills in young children and later reading skills have mainly focused on children with weak language skills (Dale et al. 2003; Desmarais et al. 2008; Rescorla 2000, 2002, 2011; Rescorla and Dale 2013). It is essential to examine children without success in reading, but it is also important to examine what characterises the early language skills in children who do succeed in reading. The present study included children at all levels in reading in grade five. By taking the children’s reading level in fifth grade as a starting point and examining how their functional language skills were in toddlerhood, we contribute with knowledge about what characterises early language skills in children who develop low reading skills compared to what characterises early language skills in children who succeed in reading.

Research question

How are the relations between the level of reading skills in fifth grade and functional language skills in toddler age?

Method

Participants and recruitment

Children in the present study participated in the longitudinal and multidisciplinary Stavanger Project, – The Learning Child, conducted in collaboration with the municipality of Stavanger.

All 61 public ECEC institutions of Stavanger municipality were obliged to participate in the project by their owner, and privately owned institutions were invited. Twenty-five of the private institutions (approximately 50%) accepted this invitation. The children attended 32 schools.

The inclusion criteria for the children were that they had to have been born in the period from July 1st, 2005, to December 31st, 2007 and enrolled in a participating ECEC institution before the age of 30 months. For more information about the Stavanger project, see Reikerås, Løge, and Knivsberg (2012). The data used in the present study were collected when the children were toddlers (33 months) and in fifth grade when the children were between 9 ½ and 10 years old.

From the 1,347 children signed up in the Stavanger Project, there was data available at both assessment timepoints on 851 children. There were 418 girls and 433 boys, and 16.7% of the children were multilingual.

Authentic assessment

Staff in the ECEC institutions are in interaction with the toddlers for many hours a day and might be able to provide information of comparable validity on the children’s language skills. In the present study, familiar staff in the ECEC institutions observed how children used language skills in play and everyday activities over a three-month period. This provided data on children’s functional language skills in natural surroundings, which is uncommon (Bagnato et al. 2014). This method (Authentic Assessment) is gentle and nonintrusive, provides ecologically valid data (Bagnato 2007; Lambert, Kim, and Burts 2014) and is recommended for the assessment of children in this age group. This method assesses children’s functional skills and offers useful information on the children’s strengths and weaknesses (Keilty, LaRocco, and Casell 2009). Therefore, we assume that this way of collecting data will contribute with new or complementary information on how toddlers’ language is related to later reading skills.

Materials

The observation material TRAS (Early Registration of Language Development) (Espenakk et al. 2003) was used to observe key aspects of the children’s early language development. TRAS provides information in the age band of 2–5 years. The guidelines for TRAS consist of an instruction book and an observation scheme for each child (Espenakk et al. 2003). In addition, a booklet was developed for the project with a detailed description of each item and guidelines for scoring to support the staff in the observations and secure the quality of the data (Helvig and Løge 2006). TRAS is divided into eight sections: Interaction, Communication, Attention, Language Comprehension, Linguistic Awareness, Pronunciation, Word Production, and Sentence Production. There were 72 questions, nine questions on three levels of increasing difficulty in each of the eight sections. Level 1 is the easiest level, and Level 3 is the most difficult.

TRAS was developed to help teachers in ECEC institutions make systematic observations of children’s development. The reliability (Spearman correlations) for TRAS is calculated for each level of difficulty and stated as .54 (level 1), .40 (level 2), and .74 (level 3) (Espenakk et al. 2003). The material is widely used in Norwegian ECEC institutions and is available in Norwegian, Swedish, and Danish.

To assess reading skills for fifth graders, the standardised compulsory National Reading Test (NRT) was used (The Norwegian Directorate for Education and Training 2015). The internal consistency is .79 (Cronbach’s alpha). The NRT examines whether children can acquire information, understand, and evaluate a variety of texts in different structures and presentation forms. The items are classified into easy, average, and more demanding questions. Within each level, the items relate to three aspects of reading skills: 1) finding information in the text, 2) interpreting and understanding the text and 3) reflecting on and considering the form and content of the text. The test for fifth graders mainly assesses aspects 1) and 2).

The NRT consists of reading booklets for children and manuals for teachers with detailed descriptions on how to conduct and score the test. It is a question-and-answer test. The booklets include five reading texts followed by 29 questions. Twenty-four are simple multiple-choice items with four options for answers. Three are complex multiple-choice items consisting of five assertions, which the children must decide were correct or not. Two items are open-ended questions. The children’s total scores are divided into three achievement levels based on how many items they had mastered/not mastered. Level 1 is the lowest reading level, and Level 3 is the highest. At the national level, 25% of Norwegian 5th graders are placed at Level 1, 50% at Level 2, and 25% at Level 3. The present study used these levels to classify children at Level 1 as Low NRT, children at Level 2 Moderate NRT and children at Level 3 as High NRT.

Procedure

Information about The Stavanger Project was provided to parents orally by the staff in the ECEC institutions. In addition, detailed information about the study and what each individual child would be exposed to was distributed. The ECEC institutions and parents also received a brochure about the research project in its entirety. The staff in the ECEC institutions received courses with updated information on young children’s language development and were trained in using the TRAS observation scheme before the observation period started. Structured observations were made from the time the children were 30 months to when they turned 33 months old to ensure that their skills had been observed in different situations and at different times. During the observation period, two of the staff in each ECEC institution observed, independent of one another, the children’s level of mastery in natural play and everyday situations in the ECEC institutions. The same skill had to be observed twice for each of the observers before scoring. At the end of the observation period, the staff in the ECEC institutions marked the children’s mastery of each item in the scheme as completely mastered, partly mastered, or not yet observed mastering. The materials were returned to the researchers for statistical purposes. During coding of the results, three levels of achievement were established: 2 points for mastering, 1 for partly mastering, and 0 for not yet observed mastering. High scores indicate good language skills, whereas low scores indicate low language skills. Maximum score is 144 points. The NRT test is mandatory for all fifth graders in Norway and is administered by teachers in the classroom. The test material was forwarded to the schools directly from the Norwegian Directorate for Education and Training. The children had 90 min to read and answer questions related to the texts. The single multiple-choice items and the items for which the children had to give written answers were coded 1 point for mastering and 0 points for not mastering. The three complex multiple-choice items were coded 2 points for five correct answers, 1 point for partial mastery (four correct answers), and 0 points for 3 or fewer correct answers. This led to a maximum of 32 points for the 29 items. The schools received the results for each child from the Norwegian Ministry of Education and Research, and the NRT groups were defined by them. The teachers provided the results to the researchers.

Data analysis

The Statistical Package for the Social Science (SPSS) version 21.0 (IBM-Corporation 2013) was used for the statistical analyses. Two research assistants entered the data in the database. The first assistant entered the data, and the second assistant controlled the results; they alternated between these two tasks. After the data entry was completed, 10% of the participants were randomly selected, and the data were entered a second time to compare the degree of deviation. This control procedure revealed reasonable consistency between the datasets (nearly 100%). In addition, frequency analyses were performed for each variable to verify whether the values were within the range of possible values. The very few deviations discovered in this control procedure were corrected in the data set.

A normal distribution of data is considered to be a prerequisite for applying parametric statistical procedures. According to Tabachnick and Fidell (2014), a visual assessment of the shape of the distribution is the favourable method for evaluating deviations from normality with regard to larger samples, as the standard error and skewness will decrease with an increase in N. Therefore, we applied Q-Q-plots to evaluate the distribution of the data’s kurtosis and the skewness. Based on the Q-Q-plots for TRAS and the TRAS sections, we consider that these data do not exhibit significant deviation from a normal distribution. As the analysis of variance (ANOVA and MANOVA) is not substantially dependent on the normality assumption or assumptions of equal variance in compared groups (Norušis 2010), the prerequisites for applying parametric statistics should be satisfied. In the analyses, descriptive statistics were calculated for the total score for TRAS and for the sections of TRAS both for the whole sample and for the three NRT groups: Low NRT, Moderate NRT, High NRT. ANOVA and MANOVA were applied to examine whether the three NRT groups significantly differed in their mean scores at TRAS in total and at the TRAS sections.

Ethical considerations

The study was approved by the Norwegian Centre for Research Data. The local municipal authorities recommended the study, and parents gave written consent for their children to participate. Parents could withdraw their children from the research project at any time without giving a reason for their withdrawal. Special considerations must be taken when young children participate in research projects to treat them with respect and in a way that does not expose them to physical or psychological risk of harm. Special requirements on the assessment situation are of importance, as small children are vulnerable and have less power in child–adult relations. The familiar staff used daily life routines and did not disturb play or other activities when conducting the observations in the ECEC institutions. The children were guaranteed anonymity in the data analyses and in presentations of the project.

Results

The research question asks how the relations are between the level of reading skills in fifth grade and functional language skills in toddler age. The three NRT groups (low, moderate and high mean scores at NRT) results at NRT, TRAS Total and the sections in TRAS are presented in Table 1.

Table 1. The results for the NRT groups on TRAS Total and the sections on TRAS when the children were toddlers.

	Low NRT (n = 124) Mean (SD)	Moderate NRT (n = 441) Mean (SD)	High NRT (n = 286) Mean (SD)	All (N = 851) Mean (SD)
TRAS Total	60.45 (20.09)	69.05 (22.60)	75.56 (23.62)	69.98 (23.11)
Interaction	8.12 (2.97)	9.01 (3.38)	9.56 (3.59)	9.07 (3.43)
Communication	8.55 (3.24)	9.54 (3.54)	10.19 (3.46)	9.61 (3.50)
Attention	8.34 (3.04)	9.50 (3.40)	10.20 (3.57)	9.56 (3.46)
Language comprehension	6.97 (2.48)	7.95 (2.96)	9.07 (3.15)	8.18 (3.04)
Linguistic Awareness	7.26 (2.29)	7.92 (2.44)	8.49 (2.59)	8.02 (2.50)
Pronunciation	7.27 (3.99)	8.73 (4.34)	9.90 (4.36)	8.91 (4.38)
Word Production	7.11 (2.97)	8.39 (3.24)	9.23 (3.18)	8.48 (3.25)
Sentence Production	6.84 (2.87)	8.01 (3.05)	8.93 (3.09)	8.15 (3.11)

As seen from Table 1, the mean scores for the three NRT groups on Total TRAS and on all sections in TRAS increased with the level on NRT. The group with low NRT scores had the lowest language skills as toddlers, the group with moderate NRT scores had better language skills as toddlers than the group with the lowest NRT scores, and the group with the highest NRT scores in fifth grade had the best language skills as toddlers.

To examine whether the NRT groups differ significantly from each other in language skills in toddler age, a one-way between-groups analysis of variance (one-way ANOVA) was conducted. In addition, a post hoc Bonferroni test was performed to determine where eventually differences were. There were statistically significant differences at the p < .05 level for the three NRT groups: F(2,848) = 20,08, p < .01 on Total TRAS scores. The effect size, calculated using eta square, was .05 and considered by Cohen (1988) to be small. The post hoc test showed that all the NRT groups differed significantly, p < .01. To examine whether the NRT groups differed on any of the TRAS subscales as toddlers, a one-way between-groups multivariate analysis of variance (MANOVA) was performed. The Levene’s test showed nonequal variance in three of the sections: Interaction, Language Comprehension and Language Awareness. Therefore, the alpha level was adjusted to .025 for all sections, and the post hoc test used was Dunnett C. The results for the sections can be found in Table 2.

Table 2. Differences in language skills between NRT groups calculated by MANOVA.

			Mean differences
TRAS sections	F (2, 848)	Eta square	Low NRT and Moderate NRT	Low NRT and High NRT	Moderate NRT and High NRT
Interaction	7.87*	.02	−.89	−1.44**	−.55
Communication	9.89*	.02	−.99	−1.64**	−.65
Attention	12.98*	.03	−1.16**	−1.86**	−.70
Language comprehension	24.52*	.06	−.99**	−2.10**	−1.11**
Linguistic Awareness	11.47*	.03	−.66	−1.23**	−.57
Pronunciation	16.95*	.04	−1.47**	−2.63**	−1.16**
Word Production	19.52*	.04	−1.27**	−2.11**	−.84**
Sentence Production	21.46*	.05	−1.17**	−2.09**	−.92**

* p < .01. ** Sig. p < .025.

As shown in Table 2, in all sections, grouping by NRT in fifth grade was significantly related to language skills in toddlerhood. The effects were small to moderate, with the largest effect on Language Comprehension. Post hoc Dunnett’s C test showed that on all sections of TRAS, there were significant differences between the low NRT and the high NRT groups. In the Attention section, there was also a significant difference between the Low NRT and Moderate NRT groups but not between Moderate NRT and High NRT. For the sections Interaction, Communication and Linguistic Awareness, no significant differences were found between the Low NRTP and Moderate NRT groups or between the Moderate NRT and High NRT groups. There were significant differences between all groups in the Language comprehension, Pronunciation, Word Production and Sentence Production sections.

Discussion

This study examined how different levels of reading skills (low NRT, moderate NRT and high NRT) in fifth grade were related to functional language skills in toddler age. The results showed that the three NRT groups differed significantly from each other in language in toddler age (TRAS Total). The group with high NRT had the highest results, the Low NRT group had the lowest results and the Moderate NRT group was in the middle.

The relations between reading and early language skills found in the present study are consistent with findings in earlier studies (Bleses et al. 2016; Hjetland et al. 2017; Lee 2011; Muter et al. 2004). However, most of these studies do not include children as young as toddlers, and none of them had a retrospective design, as in the current study. The results in the present study show that children with low reading skills had weaker language skills as toddlers than children with moderate or high reading skills. This emphasises the importance of early identification of children with weak language skills, which has been found to be of high importance to prevent difficulties in later reading (Rescorla 2009). The consistency between the present study and the study of Bleses et al. (2016) shows that weak language skills in toddler age are related to weak reading skills even though the methods differed. Bleses et al. (2016) used parental reports and testing, and the present study used systematic observation by the staff in ECEC institutions. A strength of the method used in the current study is that the teachers, in addition to identifying children with weak language skills, obtain a foundation for language interventions for the individual child.

When considering the different language skills separately, the children in the low NRT group had significantly weaker language skills in toddler age compared to the group with high reading skills in all eight sections of TRAS. The effect of grouping by NRT was largest on language comprehension (moderate effect). In addition to the difference between the groups with low NRT and high NRT in this section, there were also significant differences between the low NRT group and the moderate NRT group and between the moderate and high NRT groups. This is in line with earlier research showing that language comprehension skills at an early age serve as a foundation for later reading (Hjetland et al. 2017). However, earlier research on the importance of language comprehension uses data from preschool age. One of the few studies on relations between toddlers’ language comprehension and later reading, as examined in the present study, was by Marchman and Fernald (2008), who found language comprehension at 25 months to predict reading at age eight. Language comprehension at an early age is also found to predict vocabulary growth (Fernald, Perfors, and Marchman 2006) and is an example of different language skills not being strictly separate.

In the three sections of language production, Pronunciation, Word Production and Sentence Production, significant differences were found between all the NRT groups. These three sections, together with the Language Comprehension section, are the TRAS sections with the highest mean differences between the High NRT group and Low NRT group. This result is in line with earlier research on ‘late talkers’, which found early language production to be related to reading skills 10 years later (Bleses et al. 2016). However, the vocabulary predicted reading to a very modest degree in Bleses’ study. The effect sizes are rather small in the present study, and others, e.g. Bartl-Pokorny et al. (2013), found no correlation between early vocabulary skills and later reading. However, Dickinson, Nesbitt, and Hofer (2019) found vocabulary in preschool age to be the language ability most strongly associated with early reading.

For the section Linguistic Awareness, the only significant difference was between the groups with High NRT and Low NRT. Most of the items in this section mastered by toddlers are mainly about awareness related to interest in language activities, and the phonological aspect develops mainly in preschool age (Frost 2003). There is much research finding strong positive relations between awareness and reading in children in the age group 5–7 (e.g. Dickinson, Nesbitt, and Hofer 2019), whereas in the present study, the children are older and have a higher level of reading skills, which may have influenced the results. Although the effect of language awareness is rather low, the present study contributes knowledge about a possible relation between language awareness and reading over an extended age span.

In the social language aspect, as represented in the Attention, Communication, and Interaction sections, the low NRT group and the high NRT group differ significantly. However, the effect sizes were among the lowest in the study. Although earlier research has found that social skills are crucial for language acquisition (Henrichs et al. 2013; Kuhl, Tsao, and Liu 2003; Mackie and Law 2010), they may be of more importance for children’s wellbeing and participating in play than for later reading.

Limitations and need for further research

The use of TRAS provides room for interpretation in assessing the toddler’s language skills. However, these measurement errors would not have a systematic effect. This is supported by the fact that the data revealed a normal distribution of the results for the toddler’s language, with sufficient variance. In addition to the training of the staff, the reliability was strengthened by two staff members conducting all observations independently. It may be a limitation that the two observers did not deliver independent observations but had a discussion when seeking agreement on how to mark the scheme. However, we did this to heighten the competence in the staff in the ECEC. It may be the case that individual testing routines conducted by trained research assistants would lead to more reliable assessment scores than those carried out by staff. However, our initial assumption that ECEC teachers tend to evaluate children’s skills in a positively biased way due to positive attitudes towards children in general does not appear to be the case. In addition, with children as young as toddlers, it is necessary to take special precautions to make a safe research environment. Using familiar staff and observing the children in natural surroundings will secure this. Even though the staff were trained in applying the instrument, the large number of data collectors still might be a problem.

The present study only uses three levels of reading in 5th grade to measure reading skills. Further studies are needed to examine whether different language skills in toddler age are related to aspects of reading such as understanding different text types or different reading skills such as decoding, vocabulary, and comprehension. The effect sizes found in the present study are small to moderate, which leaves room for alternative interpretations, and therefore we had to be careful not to draw too strong conclusions. The study is not designed to show causal relationships. This would require a different study design. Studies that can reveal causal relationships would be desirable and necessary in further research.

Contributions and pedagogical consequences

The present study contributes to expanding the knowledge about how the level of reading skills is related to early language skills. The study includes a wide range of language skills, unlike earlier research, which is mostly based on one skill or a few partial skills. In this way, the study contributes to a more comprehensive understanding of the relation between the level of reading and early language. In addition, the sample is large and consists of children at all skill levels. Our findings of the relations to early language skills provide useful information for the staff in ECECs and have pedagogical consequences. The children with weak reading skills in 5th grade had weaker language skills as toddlers than children with moderate or high reading skills. Therefore, our study emphasis the urgent need to examine and do interventions on children already at toddler age, to prevent later low reading skills. This is an argument against the resistance in Norwegian ECEC institutions, as has been reported by Haug (2014).

As shown in the present study, it is possible for the staff in ECECs, by using authentic assessment, to identify children with weak language already in toddler age. This is of high importance to be able to support children’s language development, both for children’s everyday functioning and play and for their later reading.

Disclosure statement

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