Parental Reports on Children’s Language Development: Validation of the Slovenian CDI-III

Abstract

Parental reports have been shown to be an informative, cost-effective, and non-invasive method of assessing children’s early communicative and language development (up to the age of 3 years). However, less is known about their applicability in older preschool children after the age of three. The main aim of this study was to present the validation process of the Slovenian version of the Communicative Development Inventories (CDI-III), which is based on the Swedish and Croatian versions. The study was conducted on a sample of 337 parents whose children were aged between 2;6 and 4;0. The results indicate a significant intercorrelation between all subsections of the scale and a clear trend of progress related to age. Among the demographic factors, mother’s education was found to have a significant influence on children’s scores on the Grammar and Metalinguistic Awareness subscales. Gender and the presence of siblings did not appear to be a significant factor. In addition, scores on the language subscales (Vocabulary, Grammar, Metalinguistic Awareness) correlated significantly with scores in the Comprehension and Production Scales of the New Reynell Developmental Language Scale. The study confirms that the Slovenian version of the CDI-III is a valid and psychometrically suitable instrument for further research into language development in Slovenian.

Keywords

MacArthur-Bates Communicative Development Inventory III parental reports language development preschoolers Slovenian language

Introduction

A reliable and developmentally appropriate assessment of early language skills is necessary to understand the child’s overall cognitive development and to recognise possible developmental differences at an early stage. For several decades, scientists and clinicians have endeavoured to find the best methods, that is, those that would be comprehensive, reliable, and cost-effective, for assessing preschool children’s language development. Numerous studies have confirmed that standardised parent reports are a valuable source of information for researchers and clinicians on the development of children’s early language skills (Frota et al., 2016; Guiberson et al., 2011; Kas et al., 2022; O’Toole & Fletcher, 2010; Sachse & Von Suchodoletz, 2008; Stolt, 2023; Tulviste & Schults, 2020).

Of the various scales available for parental assessment of a child’s language development, the MacArthur-Bates Communicative Development Inventories (MB-CDI) are the most commonly used in both research and clinical practice. The first two versions – MB-CDI-I and MB-CDI-II – were designed to assess a broad range of communicative and language skills in children up to 30 months of age (Dale & Penfold, 2023; Fenson et al., 1993). Originally developed for assessment in American English, their simplicity of use and theoretical transparency have led to their adaptation in more than 100 languages so far (https://mb-cdi.stanford.edu/adaptations.html), including Slovenian (Marjanovič Umek et al., 2012).

Although the MB-CDI instruments have proven successful in assessing early language development up to the age of 30 months, there was still a need for instruments suitable for assessing the language abilities of children beyond this age range. To address this gap, the MB-CDI-III was introduced in 2007 (Fenson et al., 2007), which is aimed at children aged 30 to 37 months and consists of 3 subsections: Vocabulary Checklist (100 items), Syntactic Complexity (12 items), and Language Use (12 items). The American English version of the MB-CDI-III has been adapted for a few languages, including Slovenian (Fekonja et al., 2025). However, ceiling effects in older children have raised concerns about the sensitivity of the original version. To overcome these limitations, Eriksson (2017) developed a Swedish version of the MB-CDI-III (henceforth, CDI-III), which includes additional items and extends the age range of use up to 48 months. As Šmit Brleković and Kuvač Kraljević (2023) indicate, contemporary adaptations are devised in accordance with two different, albeit identically termed (i.e. CDI-III) directions. In other words, some of the authors have adapted the original American, while others have adapted the Swedish inventory (see the list in Šmit Brleković & Kuvač Kraljević, 2023). However, some authors have adapted only a part of the original scale. For example, the Portuguese version follows the Swedish original in the vocabulary section while featuring a new subscale for syntactic structures (Cadime et al., 2021). This study presents the validation of the Slovenian version of the CDI-III (CDI-III-SI), which is based exclusively on the Swedish version (Eriksson, 2017) and recognises the experience gained in adapting the Croatian version – the first version developed for a Slavic language (Kuvač Kraljević et al., in press).

CDI-III covers the period of language development in which the number of words exceeds 400 and the syntactic structures become longer and more complex, with a greater number of morphological markers. The child’s vocabulary gradually expands beyond concrete terms related to their immediate environment and daily routines, increasingly incorporating abstract words from different parts of speech. Thus, around the age of three, children begin to acquire an emotional and cognitive vocabulary – primarily to express their own feelings (Baron-Cohen et al., 2010). Many words that are acquired during this period are not only abstract but polysemous as well, that is, they have multiple meanings that need to be understood and differentiated. Mastering such a vocabulary not only expands the child’s lexical repertoire, but also improves their ability to use language flexibly and interpret the world more deeply. An ever-improving grammatical competence enables children to construct coherent sentence sequences that allow them to participate meaningfully in discourse. Children are capable of telling personal stories through discourse and thus share their personal experiences (Westby & Culatta, 2016) or invent fictional stories that cultivate their imagination. Good language knowledge is a prerequisite for the development of metalinguistic awareness – “language about language” (Berry, 2009). Thinking about how language works is a prerequisite for the development of early literacy skills. The rapid lexical development mentioned above significantly supports metalinguistic development, as children with a larger vocabulary begin to distinguish words more accurately in terms of both meaning and sound (Neuman, 2014). This marks the beginning of the development of phonological awareness, thus analysing a spoken word in its basic phonological units (syllable, rhyme, and phoneme) (Ouellette & Haley, 2013), which is an important foundation of orthographic knowledge, that is, knowing how a letter looks and sounds (Lonigan et al., 2000).

Standardised parental scales not only provide information about the language abilities of individual children, but also about the patterns of language acquisition in certain linguistic and cultural environments. This is particularly important for less-researched or under-resourced languages for which there are few formal assessment tools and empirical data. The Slovenian language belongs to the group of Slavic languages and is typologically notable for its tripartite grammatical number (singular, dual, and plural), gender division into masculine, feminine and neuter, while having a complex case system comprising six cases. Its verbal system includes two aspects and several conjugation types, features broadly shared across Slavic languages, although the use of the dual number is preserved in only a few, including Slovenian.

It has approximately 2 million speakers and the patterns of its mastery are insufficiently described. Under these circumstances, parental reports can be a valuable source of data on language development. Since the CDI-III has been shown to be valid and useful for research and clinical practice in other, more explored languages, it was decided to adapt this scale to Slovenian. Adapting the scale in another language provides additional evidence of its content validity (Jarůšková et al., 2023), that is, its ability to function in other languages.

The Present Study

The main objective of this study was to examine the psychometric properties of the CDI-III-SI. To this end, the following specific objectives were defined:

To examine how scores on the CDI-III-SI subsections reflect age-related language development of children aged 2;6 to 4;0 years.

To determine the internal consistency of the CDI-III-SI language subsections.

To analyse the correlations between the CDI-III-SI language subsections.

To assess the influence of various demographic factors (age, mother’s level of education, siblings, and gender) on scores on the CDI-III-SI language subsections.

To establish concurrent validity by comparing scores on the CDI-III-SI with those obtained from the New Reynell Developmental Language Scale (NRDLS).

Methods and Materials

Participants

In the standardisation study of the CDI-III-SI, data were obtained through standardised parental reports completed by caregivers for a total of 337 children aged between 2 years and 6 months and 4 years. The reports addressed 167 girls and 170 boys. Strict inclusion criteria were applied. According to the information provided in the demographic section of the CDI-III-SI, children included in the study had no diagnosed neurodevelopmental disorders (e.g. intellectual disability or autism spectrum disorder), no hearing impairment, and no other medical condition known to affect language development. Furthermore, none of the children had the official status of a child with special needs.

An additional inclusion criterion was monolingual language exposure. Only children whose first and only language was Slovenian, confirmed by parental report, were eligible for participation. All children who did not meet the inclusion criteria were excluded from the sample. In addition, three children who were undergoing speech-language therapy for speech sound disorders were also excluded. A total of 9 children were excluded, so that the final sample comprised 337 monolingual Slovenian-speaking children.

In order to identify differences in language development between the age groups for which norms will ultimately be set, the sample was divided into three age groups as shown in Table 1. There is no significant difference between age groups in terms of gender distribution (χ²(2) = .63; p = .73), sibling status (χ²(2) = .19; p = .91), or the structure of mother’s education (χ² (8) = 4.38; p = .82).

Table 1.

Sample’s Distribution by Age, Gender, Siblings and Mother’s Education.

Age groups	Gender		Siblings		Mother’s education*					Total
Age groups	Girls	Boys	Yes	No	1	2	3	4	5	Total
Age group 1 30–35 months	54	55	83	26	16	12	34	39	8	109
Age group 2 36–41 months	53	60	86	27	17	20	29	43	4	113
Age group 3 42–48 months	60	55	90	25	18	16	31	45	5	115
Total	167	170	259	78	51	48	94	127	17	337

1 = Primary or vocational education; 2 = Secondary general or technical education; 3 = Higher vocational or first Bologna cycle (Bachelor’s degree); 4 = University or second Bologna cycle (Master’s degree); 5 = Master of Science or PhD.

Mothers were the dominant respondents, representing 95% of all reports. Fathers completed the scales in 4% of all reports, while both parents participated in just three cases (1%).

For 91 sampled children whose parents completed the CDI-III, an additional assessment of language abilities was conducted using NRDLS-SI.

Procedure

To ensure both geographical and linguistic representativeness, participants were selected using stratified sampling across all 12 regions of Slovenia. In addition to ensuring regional representativeness, the approach was designed to include the broad range of dialects spoken throughout Slovenia. The selection process was conducted in collaboration with six speech and language therapists (SLTs), who played a central role in identifying kindergartens and individuals willing to participate in the study. In cooperation with the SLTs, kindergarten teachers selected children aged between 30 and 48 months who met strict inclusion criteria and whose parents were willing to participate in the study. Some parents were additionally invited to approach the language assessment of their children using the NRDLS-SI. Coordinated efforts between SLTs and kindergarten teachers were crucial to ensure parental engagement. The exact number of children and the quotas for girls and boys was determined for each region according to the population density ratio. Prior to data collection, written informed consent was obtained from all participating kindergartens and from the parents or guardians of the children whose language abilities were assessed by the scale. Participants were informed that their participation was voluntary and that they could withdraw from the study at any time without consequence.

Parents who only had to complete the CDI-III-SI at home, returned it to the kindergarten teachers within a few days. The kindergarten teachers collected the completed scales and forwarded them to the designated SLT. The parents who agreed to have their children assessed with the NRDLS-SI completed the CDI-III-SI during the SLT’s assessment. The SLTs then forwarded the scales for the entire sample to the research team, along with the NRDLS-SI test results for the subset of children who underwent additional assessment. All responses were anonymised prior to analysis.

Slovenian CDI-III – Adaptation

CDI-III-SI is based on the Croatian version (CDI-III-HR; Kuvač Kraljević et al., in press), which in turn is derived from the Swedish CDI-III (SCDI-III; Eriksson, 2017). The CDI-III-HR represents the first translation and adaptation of the Swedish original for a Slavic language and effectively transcends differences between Germanic and Slavic languages, especially in the morphological and syntactic parts of the scale. Slovenian, like Croatian, belongs to Slavic languages, which are characterised by a complex morphology. Although the two languages are similar in many respects, the characteristics and peculiarities of the Slovenian language had to be acknowledged when adapting and developing the CDI-III-SI (e.g. the grammatical number dual, which does not exist in Croatian).

The development of the Croatian version of the CDI-III began in 2018 and overcomes all typological differences between Swedish and Croatian, especially in the grammatical part of the scale. The final version of the CDI-III-HR follows the structure of the Swedish scale in terms of the number of items in the communication-level subsection (6 items) and the number of words (100 words) and 4 semantic categories in the vocabulary subsection. In the CDI-III-HR, the subsection originally labelled as Syntax in the SCDI-III was renamed Grammar, while the total number of items remained unchanged (18). The internal structure of the subsection was also retained and now consists of two parts: Grammar-morphology (formerly Grammar in the SCDI-III), featuring eight items, and Syntax complexity (formerly Language complexity), comprised of 10 items. CDI-III-HR has two more items in the metalinguistic subsection compared to the Swedish original (nine vs. seven) and further elaborates the final pronunciation subscale with five novel queries on top of the existing one general question, which is identical to the original version. These five additional items provide valuable descriptive information about the child’s sound production.

The Slovenian version of the scale was translated and linguistically adapted by the end of 2021 on the basis of the Croatian version. The first version of the scale was reviewed by five SLTs, who confirmed its content validity. A pilot study was then conducted with a sample of 54 children (27 girls and 27 boys) using a larger pool of items. The results of the pilot study showed a ceiling effect on a considerable number of vocabulary and syntactic complexity items. Based on statistical analyses, the most suitable items were selected for the standardisation phase. The data for standardisation were collected in 2023 in all regions of Slovenia on a sample of 337 children.

The structure of the final Slovenian version of the scale mirrors the structure of the Croatian scale, retaining the same number of items in all five subsections. In the Grammar subsection (in the Grammar-morphology and Syntax complexity parts), only a few changes have been made to the content based on the differences between the languages. For example, items related to the grammatical number “dual” were added.

Slovenian CDI-III – Structure and Scoring

The CDI-III-SI has five subsections which appear in the following order: Level of communication, three language-related subsections (Vocabulary, Grammar, and Metalinguistic Awareness) and Pronunciation.

In the first subsection, the parents rank their child’s level of communication into one of the proposed six levels. The statements are graded from the first to the sixth level as follows: He/she does not yet speak; His/her speech is unintelligible; He/she produces one-word utterances; He/she produces two- or three-word utterances; He/she produces simple sentences; He/she produces long and complex sentences. This part is not scored and contains one exclusion criterion. If parents define their child’s communication level as fitting the description under level one or two, they must stop completing the scale. If the child’s communication corresponds to levels three to six, the parents can continue with the scale.

The second subsection evaluates vocabulary. The word list (100 items) is divided into 4 semantic categories: Food words (16 items), Body words (26 items), Mental words (30 items), and Emotion words (28 items). Parents mark the words that their child can say. This section is scored by awarding 1 point for each marked word, with a possible score ranging from 0 to 100 points.

The third subsection addresses grammar and consists of two parts. It commences with Grammar-morphology whose 8 items verify the manner of used structures, where points are attained by the frequency of usage (never – 0 points, sometimes – 1 point, often – 2 points; so the possible score ranges from 0 to 16 points). It continues with Syntax complexity where they are presented with 10 pairs of sentences from which parents select one, depending on which syntactic structure their child is most likely to use. The less syntactically complex sentence is scored with 1 point, and the more complex sentence is scored with 2 points (this part has a possible score ranging from 10 to 20 points). The total score for the grammar subsection ranges from 10 to 36 points.

The fourth subsection relates to metalinguistic awareness and includes nine items about phonological (three items) and orthographic (six items) awareness. Each confirmed statement is rated with one point, with a possible score ranging from zero to nine.

The fifth subsection is the only one that assesses speech. It contains a general question asking parents whether they think their child speaks as well, worse, or better than their peers. This is followed by five items assessing specific aspects of speech: two questions on the pronunciation of individual sounds (the fricatives /ʃ/ and /ʒ/, the affricate /ʧ/ and the trill /r/), two on the occurrence of phonological errors (the omission of syllables and the substitution of phonemes within individual words), and one on the intelligibility of the child’s speech to unfamiliar listeners. For each item, parents select one of three response options (never, sometimes, always). Although these are not scored, they provide valuable descriptive information about the child’s speech and sound production.

Slovenian New Reynell Developmental Language Scale

A subset of the sample of children (n = 91) was additionally assessed by SLTs using the Slovenian version of the NRDLS-SI (Edwards et al., 2024), which was developed to assess the language abilities of children aged 2;0 to 7;5 years. The Slovenian version was translated, adapted, and standardised based on the Croatian version – NRDLS-HR (Edwards et al., 2019), which follows the structure of the original English version (Edwards et al., 2011).

The NRDLS-SI test battery consists of 2 independent scales: The Comprehension Scale (72 items) and The Production Scale (72 items). Both scales assess vocabulary, early two-word combinations, morphology, as well as simple and complex syntactic structures. The last section of the Comprehension Scale assesses inferencing, which is a component of pragmatic skills. The final section of the Production Scale assesses the child’s ability to make grammatical judgements, which is considered a component of metalinguistic awareness.

Normative data were collected from a sample of 500 typically developing Slovenian children from all regions of Slovenia. The strong psychometric properties of the test enable its application in both clinical practice and research. The correlation between the Comprehension Scale and the Production Scale is high (r = .89). The reliability coefficient, which was calculated using the split-half method, is .94 for the Comprehension Scale and .96 for the Production Scale.

Additionally, data were collected from children with developmental language disorder (DLD, n = 28) and compared with appropriately selected results from the standardisation sample to examine discriminant validity. The children from the clinical sample scored significantly lower on both scales (Comprehension Scale t = 5758; Production Scale t = 6230; for both df = 58; p < .001).

Data Preparation and Analysis

Based on the collected data, the raw scores for all subsections of the CDI-III-SI and both scales of the NRDLS-SI were calculated and used in all analyses to capture the whole variation between individual children. To address the first specific objective, the sample was divided into 3 age groups (30–35 months, 36–41 months, and 42–48 months), and the effect of age on all subsections of the CDI-III-SI scale was examined. The score distributions were calculated for each age group.

The reliability of the CDI-III-SI language subsections, vocabulary (all four semantic categories), grammar (Grammar-morphology and Syntax complexity), and metalinguistic awareness, was assessed using two internal consistency coefficients. Cronbach’s alpha, the most commonly used coefficient to assess internal consistency in the social, behavioural, and educational sciences (Ravinder & Saraswathi, 2020), is used to demonstrate the reliability of the scale. However, some authors argue that its use is not entirely appropriate as it relies on assumptions that are rarely met (Ravinder & Saraswathi, 2020). In addition, there are still many open questions about the precise meaning of internal consistency, whether the degree of internal consistency can be determined using the alpha value and, if so, what value is acceptable depending on the specific test requirements (Cho & Kim, 2015). Therefore, alternative reliability coefficients were recommended, including McDonald’s omega (Malkewitz et al., 2023; Ravinder & Saraswathi, 2020), which is also used in the present study. To provide a more robust measure of internal consistency, corrected item-scale correlations were also calculated, as this metric is independent of the number of items.

Intercorrelations between the CDI-III-SI language subsections were examined using Spearman’s rho correlation coefficient, as none of the distributions met the assumption of normality. Although the deviations from normality were not substantial, a non-parametric approach was chosen to ensure robustness. Three linear regression analyses were conducted to examine the effects of child’s gender, age, mother’s education, and presence of siblings on CDI-III-SI language subsection scores. To assess concurrent validity, Spearman’s rho correlation coefficients were calculated between the children’s scores on the CDI-III-SI language subsections and their scores on the Comprehension and Production Scales of the standardised NRDLS-SI test.

Results

Age-Related Progress in Language Development as Measured by Scores on the CDI-III-SI Subsections

The subsequent sections will illustrate the progression in language development across different age groups for all subscales of the CDI-III-SI, as reported by parents.

Level of Communication

In this part of the scale, parents rate their child’s level of communication. The first two levels serve as exclusion criteria. Parents who selected one of the remaining four levels were included in the sample. The members of one-word utterances are solely children of age group one (Figure 1); however, even these are few in number and represent less than 1% of the entire group. This is to be expected in a population of typically developing children over 30 months of age, as this is the age at which children should be able to form multi-word utterances.

Figure 1.

Chart of the communication level among three age groups (n = 337).

At the level of two-to-three-word utterances, the first age group dominates with 13.8%, while the other two groups amass approximately 2%. Communication at the fifth level, the level of simple sentences, was chosen by 43% of parents whose children are in the first age group, 32.7% of parents whose children are in the second age group, and by merely 14% of parents whose children belong to the third age group. At the highest level of communication, the level of complex sentences, the highest proportion of children came from the third age group (82%), slightly less from the second age group (64%), and the lowest from the youngest age group (42%). The results clearly show the expected developmental trend of progress in the communication skills of children aged 30 to 48 months.

Language Subsections – Vocabulary, Grammar, and Metalinguistic Awareness

The test results of all three language subscales – Vocabulary, Grammar, and Metalinguistic Awareness, are presented in Table 2. The results are further divided into three age groups. An upward trend in scores in age progression is evident across all language subsections. The standard deviations indicate a high dispersion of results across all age groups, suggesting that the sample included children with varying language abilities. Table 2 also presents the results of statistical tests comparing means between groups, that is, one-way analysis of variance (in case of homogeneity of variance) with Tukey’s post hoc test, and Welch’s t-test (for data with unequal variances) with Tamhane’s T2 post hoc test. Statistically significant differences are observed across all age groups in the vocabulary and metalinguistic awareness subsections. In the Grammar subsection, significant differences are found between the first and second, as well as the first and third age groups, whereas no statistically significant difference is observed between the second and third age groups.

Table 2.

Statistics for Language Subsections of the CDI-III-SI Across Age Groups.

Age groups (n = 337)	Vocabulary (0–100)		Grammar (10–36)		Metalinguistic Awareness (0–9)
Age groups (n = 337)	M (SD)	Min-Max	M (SD)	Min-Max	M (SD)	Min-Max
Age group 1 (n = 109)	61.98 (17.89)	11–100	27.20 (5.52)	10–36	3.51 (1.64)	0–7
Age group 2 (n = 113)	71.13 (15.39)	37–100	30.17 (4.59)	15–36	4.16 (1.93)	0–9
Age group 3 (n = 115)	77.53 (14.79)	26–100	31.29 (3.94)	13–36	5.52 (1.77)	1–9
F/Welch test	26.45		20.17		37.13
p	< .001		< .001		< .001
η²	.137		.118		.182
Post hoc test (p < .05)	1–2; 1–3; 2–3		1–2; 1–3		1–2; 1–3; 2–3

Note. CDI-III-SI = Communicative Development Inventories; SD = Standard deviation.

Pronunciation

In response to the first general question regarding their child’s pronunciation, 66% of parents stated that it was comparable to that of their peers. Five per cent of parents assessed that their child’s pronunciation was below expectations associated with age, while 29% considered it to be above expectations.

The following are the results of five specific pronunciation-related items. The first two items relate to the pronunciation of individual sounds. In Slovenian, the fricatives /ʃ/ and /ʒ/, the affricate /ʧ/, and the trill /r/ are among the latest-acquired phonemes, typically mastered by the age of five. According to parental reports, 52% of the total sample had acquired these sibilants, while only 21% had mastered the /r/ sound. A trend towards increasing acquisition was also observed across age groups. In the youngest age group, 40% of the children had acquired these sibilants, compared to 52% in the second age group and 61% in the third. The trill /r/ was acquired by only 13% of the children in the first age group, by 21% in the second, and by 29% in the third.

The third and fourth items relate to the omission of syllables and the substitution of phonemes within individual words. Parents reported that 12% of children in the youngest age group still omitted syllables within words, whereas this percentage decreased to 7% in the second group and 4% in the third. The differences between age groups were less pronounced for phoneme substitutions. Nine per cent of children in the first age group exhibited phoneme substitutions within words, compared to 8% in the second and 7% in the third group.

The final item concerns the intelligibility of the child’s speech to unfamiliar listeners. Sixty-four per cent of parents of the youngest children reported that their speech was understandable to strangers. This percentage increased to 75% in the second age group and 83% in the third.

Internal Consistency of the CDI-III-SI Language Subsections

The findings of the present study indicate no significant differences between the values of the Cronbach’s alpha and McDonald’s omega coefficients, as shown in Table 3. Given the number of items, the highest coefficient values are observed in the Vocabulary subsection (α = .96; ω = .97), followed by the coefficients for the Grammar subsection (α = .86; ω = .87). The lowest coefficient values are found in the Metalinguistic Awareness subsection (α = .66; ω = .65), although these values remain within an acceptable range. Due to the influence of the number of items on both coefficients, values for individual components within the language subsections were also calculated, as recommended by Jarůšková et al. (2023), and are presented in Table 3.

Table 3.

Internal Consistency of CDI-III-SI Language Subsections Measured by Corrected Item-Scale Correlations, Cronbach’s Alpha, and McDonald’s Omega.

Subsections of CDI-III-SI (number of items)	Corrected item-scale correlations			Cronbach’s α	McDonald’s ω
Subsections of CDI-III-SI (number of items)	<.30	.30–.50	>.50	Cronbach’s α	McDonald’s ω
Food words (16)	2 (12.5%*)	14	0	.73	.71
Body words (26)	4 (15.4%*)	14	8	.87	.87
Mental words (30)	3 (10%*)	10	17	.92	.93
Emotion words (28)	6 (21.4%*)	11	11	.89	.90
Vocabulary (100)	17 (17%*)	46	37	.96	.97
Grammar-morphology (8)	0 (0%*)	2	6	.81	.81
Syntax complexity (10)	1 (10%*)	7	2	.76	.77
Grammar (18)	1 (5%*)	9	8	.86	.87
Metalinguistic Awareness (9)	2 (22%*)	7	0	.66	.65

Note. CDI-III-SI = Slovenian version of the Communicative Development Inventories.

Proportion of correlations <.30.

As an alternative measure of internal consistency, corrected item-scale correlations were included, following the recommendation that no more than 10% of items should have a low correlation < .30 (Cohen & Swerdlik, 2005). The results indicate that the Grammar subsection exhibits high internal consistency, whereas the vocabulary subsection is considerably less consistent than suggested by the Cronbach’s alpha and McDonald’s omega coefficients. The same applies to the metalinguistic awareness subsection, which also exhibits generally lower coefficient values.

Correlations among the CDI-III-SI Language Subsections

All bivariate correlations pertaining to language subsections of the CDI-III-SI were found statistically significant. Table 4 shows a high correlation between the subsections Vocabulary and Grammar and a moderate correlation between Vocabulary and Metalinguistic Awareness, and between Grammar and Metalinguistic Awareness.

Table 4.

Correlation Coefficient Values (ρ) Between Language Subsections of the CDI-III-SI.

Language Subsection	Grammar	Metalinguistic Awareness
Vocabulary	.60**	.45**
Grammar		.49**

Note. CDI-III-SI = Slovenian version of the Communicative Development Inventories.

Correlation is significant at the .01 level (two-tailed).

Influence of the Various Demographic Factors

The three regression models explored the impact of demographic factors, such as age, gender, maternal education and the presence of siblings on children’s vocabulary, grammar, and metalinguistic awareness.

In the first model (Table 5), age was the only significant predictor of vocabulary scores, while mother’s education, presence of siblings, and gender had no significant effect. The results indicate that children’s scores improve with age, with an estimated increase of 1.23 words per month.

Table 5.

Linear Regression Model for Vocabulary.

Variable	β	SE (β)	Stand. β	p
Constant	18.52	7.65		.016
Age (months)	1.23	0.17	.37	< .001
Mother’s education	.41	0.76	.03	.587
Siblings	1.01	2.09	.02	.630
Gender	1.51	1.77	.04	.396
R square (adjusted)	.130

Note. SE = Standard error.

The second model (Table 6), which analysed Grammar, identified both age and mother’s education as significant predictors, whereas the presence of siblings and gender showed no significant impact. These findings suggest that older children and those whose mothers have a higher level of education tend to perform better in this language area.

Table 6.

Linear Regression Model for Grammar.

Variable	Β	SE (β)	Stand. β	p
Constant	14.52	2.22		< .001
Age (months)	0.32	0.05	.34	< .001
Mother’s education	0.72	0.22	.17	< .001
Siblings	−0.41	0.61	−.03	.502
Gender	0.48	0.51	.05	.354
R square (adjusted)	.132

Note. SE = Standard error.

Similarly, the third model (Table 7) for Metalinguistic Awareness found age and mother’s education to be significant predictors, while the presence of siblings and gender had no significant effect. The adjusted R-squared value of .201 suggests that the model explains a meaningful portion of the variance in said scores.

Table 7.

Linear Regression Model for Metalinguistic Awareness.

Variable	Β	SE (β)	Stand. β	p
Constant	−2.43	0.84		.004
Age (months)	0.16	0.02	.43	< .001
Mother’s education	0.30	0.08	.17	< .001
Siblings	−0.10	0.23	−.02	.663
Gender	−0.14	0.19	−.04	.464
R square (adjusted)	.201

Note. SE = Standard error.

Overall, age was consistently determined as a significant predictor across all three language abilities, while mother’s education influenced grammar and metalinguistic awareness but not vocabulary. In contrast, the presence of siblings and gender had no significant effect on any of the examined language skills.

Concurrent Validity of the CDI-III-SI

From the total sample, 91 were additionally assessed using the NRDLS-SI language test. As shown in Table 8, all language subsections of the CDI-III-SI exhibit statistically significant correlations with children’s performance on both the Comprehension and Production Scales of the NRDLS-SI. Notably, slightly higher correlations are observed with the Production Scale. The vocabulary and grammar subsections show moderately high correlations with the Production Scale, while the correlation with the metalinguistic awareness subsection is slightly lower.

Table 8.

Correlation Coefficient Values (ρ) Between Language Subsections of the CDI-III-SI and Comprehension Scale and Production Scale of the NRDLS-SI.

Scales of the NRDLS-SI	Language subsections of the CDI-III-SI
Scales of the NRDLS-SI	Vocabulary	Grammar	Metalinguistic Awareness
Comprehension Scale	.49**	.42**	.43**
Production Scale	.50**	.50**	.46**

Note. CDI-III-SI = Slovenian version of the Communicative Development Inventories; NRDLS-SI = Slovenian version of the New Reynell Developmental Language Scales.

Correlation is significant at the .01 level (two-tailed).

Discussion

This study describes the adaptation process of the Slovenian version of the CDI-III, based exclusively on the Swedish version (Eriksson, 2017), and its first adaptation for Slavic languages – the Croatian version (Kuvač Kraljević et al., in press). The study also presents basic psychometric data that provide a clear insight into the validity, appropriateness, and usefulness of the scale.

The descriptive data for all sections of the CDI-III-SI show the progression of language development with increasing age for all language variables. The greatest shift can be observed in lexical development, followed by grammatical development, and the smallest in metalinguistic development. A comparison of the lexical part of the CDI-III-SI with the original Swedish version (Eriksson, 2017) shows slightly higher average scores at the age of 30 months in the Slovenian version (SCDI-III: 51; CDI-III-SI: 60) and more uniform scores at the age of 4 years (SCDI-III: 80; CDI-III-SI: 81). Compared to the Finnish version (Stolt, 2023), the results are very similar at both ages (at 30 months: FinCDI-III: 59; at 4 years: FinCDI-III: 76). Same can be said upon comparing the Slovenian with the Croatian version (Šmit Brleković & Kuvač Kraljević, 2023), where average values for the three age groups regarding lexical variables are almost identical. Although small differences in the number of produced words arise between languages, especially for the youngest involved children, these distinctions may be explained by greater variability in lexical development at around 30 months of age (Eriksson et al., 2012), which becomes more uniform in later stages. The grammar section shows a linear increase in knowledge in the application of morphological rules and the formation of various syntactic constructions. The same tendency can be observed in other languages for which this scale is utilised, regardless of language typology (Croatian and Swedish as inflectional languages and Estonian and Finnish as agglutinative languages). The metalinguistic skills have the lowest value, which is to be expected, as the development of these skills is only expected in the following years. Low scores for metalinguistic awareness have been confirmed also in other CDI-III studies (see Eriksson, 2017; Šmit-Brleković & Kuvač Kraljević, 2023; Stolt, 2023). In addition to other cognitive abilities, metalinguistic awareness is an important factor in later-stage literacy development (Ouellette & Haley, 2013). Using the CDI-III, it would be interesting to analyse the emergence of early signs of metalinguistic awareness across languages that differ in orthographic transparency and in the initial development of pre-reading and pre-writing skills (Zaretsky et al., 2009). These data point to two important aspects: (a) the continuation of lexical and grammatical development observed at younger ages and the onset of metalinguistic skill and (b) the ability of parents to observe language development even after the age of 30 months, when language becomes much more complex. Additional group analyses confirm the importance of age in influencing first-language acquisition. Children in the middle and oldest groups perform better than the youngest group on all three language variables. The same pattern is visible in general communication – as age increases, children use more complex language in everyday life activities. Clear tendencies of progress with age were also observed in the last subsection of the scale, the only one related to speech. The results show that the correct pronunciation of sibilants (/ʃ/, /ʒ/, /ʧ/) and the sound /r/ increases with the age of the children. In addition, older children make fewer phonological errors (syllable omission and phoneme substitution) at the word pronunciation level compared to younger children. Parents also report that the intelligibility of their children’s speech for unfamiliar listeners increases with the age of the children.

Internal consistency as a measure of the agreement of responses across the individual items was excellent for the vocabulary section, good for grammar, and acceptable for metalinguistic awareness. The same data are confirmed in all other versions of the CDI-III based on the Swedish version (e.g. for the vocabulary section, r = .97 in the Croatian, Estonian and Norwegian CDI-III), confirming that all these scales are very well adapted from the original version. In this way, a very similar description of language development can be observed in these languages. Scales with almost identical reliability values enable further cross-linguistic comparisons, which provide more comprehensive information about general and language-specific features of language development than monolingual studies (Kuvač Kraljević et al., 2021).

The stable intertwined developmental pattern between lexical and grammatical development, which has been proven in many studies (Stolt, 2023; Tulviste & Schults, 2020), is also confirmed here. The moderate correlation between vocabulary and grammar indicates that a larger lexicon is accompanied by better grammatical knowledge and, conversely, better morphosyntactic knowledge is associated with progress in lexical development. In contrast, the correlation between vocabulary and metalinguistic awareness, as well as between grammar and metalinguistic awareness, is much lower. Unlike vocabulary and grammar, which are part of the language knowledge that a child uses daily, metalinguistic knowledge refers to the ability to consciously think about a language. As this knowledge is just beginning to develop at this age, the weak correlation of metalinguistic awareness with vocabulary and grammar is not surprising. However, it is to be expected that these values will increase after the age of four, when the child progressively develops prerequisites for reading and writing that are largely based on metalinguistic knowledge (e.g. phonological awareness).

Regression analysis was conducted to investigate the extent to which four predictor variables (age, mother’s education, the presence of siblings, and gender) influence lexical, grammatical, and metalinguistic development. In all three analyses, only age was confirmed as a robust factor. As children age, they progress in all three criterion variables. In all other languages in which the CDI-III is adapted, the age was confirmed as an important factor of language development and its variations (Eriksson, 2017, Holm et al., 2023; Šmit Brleković & Kuvač Kraljević, 2023; Stolt, 2023; Tulviste & Schults, 2020). Such stability is not achieved with other factors. In this study, only maternal education proved to be significant for grammatical and metalinguistic development. This could mean that mothers with a higher level of education are more likely to expose their children to more complex and varied grammatical structures during verbal interaction. Interestingly, the Finnish study (Stolt, 2023) confirms gender as an important factor in language development, but not maternal education. In contrast, the Estonian study found that being a girl, having more siblings, spending more time in day care and having a (highly) educated mother were associated with higher vocabulary scores. What these studies obviously depict is a diversity of results regarding the influence of various demographic factors on language development, with sample heterogeneity likely being a major common factor. Furthermore, it must be emphasised here that the overrepresentation of mothers with higher education in the sample prevents a generalisation of the findings. It can be concluded that higher maternal education contributes to better language skills of children, but nothing can be said about the children’s language development in relation to the other levels of mothers’ education. Interestingly, all studies that validated the CDI-III scale had the same problem related to mothers’ education (Holm et al., 2023; Stolt, 2023; Tulviste & Schults, 2020), which also prevented these studies from generalising their findings. It appears that mothers with higher education are more willing to participate in various research studies, regardless of the language and country in which the study is conducted.

A moderate correlation between the three language subscales of the CDI-III-SI and the Comprehensive and Productive Scales of the NRDLS-SI indicates adequate concurrent validity of the new scale, that is, agreement between the two language tools. Parental assessment of a child’s language was equally related to both language domains, thus for comprehension and production. This means that parents can reliably assess their child’s language production between 30 and 48 months.

Limitations and Further Research

A limitation of this study is that the sample included only a small proportion of parents with secondary general or technical education or less (29%), while a disproportionate number had higher vocational education or a first Bologna cycle degree or more (71%). This differs notably from the educational structure of the Slovenian population as of 1 January 2024, when, according to the Statistical Office of the Republic of Slovenia (n.d.), 73.8% of individuals had completed secondary general or technical education or less, and only 26.2% had attained higher vocational education or a first Bologna cycle degree or more.

Since most of the scales were completed by mothers (95%) and maternal education was found to be an important demographic factor influencing the results of the Grammar and Metalinguistic Awareness subscales, this should be taken into account when interpreting the results. Since various studies based on the CDI-III have investigated the influence of different factors on language development, it is difficult to compare the obtained data. For example, in this study the influence of siblings was only examined by sibling presence, while in other studies the same variable was observed via birth order. Future studies should therefore include as many factors as possible.

Furthermore, it would be worthwhile to investigate language development in other populations of children, for example, bilingual speakers or children with DLDs. This would not only broaden our understanding of language acquisition but also provide insights into parental perceptions of language development, which may differ from typical development.

Conclusion

The results of the present study show that the Slovenian version of the CDI-III is a linguistically appropriate and psychometrically robust instrument for assessing language development in children aged 30 to 48 months. The findings indicate a clear and systematic progression and significant relationship among all examined language domains. CDI-III-SI reflects the structural features and developmental patterns of the Slovenian language and is therefore suitable for use in research and clinical practice. The study contributes to existing knowledge about the validity of parent-report scales and their practical value in comparison to formal assessments of preschool children’s language skills, especially for the period after the age of three, which remains somewhat under-researched. It also adds to the findings on the validity of the CDI-III instruments in different languages and supports further cross-linguistic comparisons. Such studies are particularly valuable for languages with a small number of speakers – Slovene being one of them – as they facilitate integration into a broader scientific discourse and contribute to the understanding of universal and language-specific development patterns.

Footnotes

Acknowledgements

We thank Professor Mårten Eriksson for permission to adapt the Swedish version and the publisher Naklada Slap for technical–financial support in validating the CDI-III for the Slovenian context. We also thank the speech-language therapists who participated in the data collection in Slovenia and all participating parents and children.

ORCID iDs

Barbara Penko

Jerneja Novšak Brce

Damjana Kogovšek

Jelena Kuvač Kraljević

Authors Contributions

Barbara Penko: Conceptualisation; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Visualisation; Writing – Original draft; Writing – Review & editing.

Jerneja Novšak Brce: Investigation; Writing – Original draft; Writing – Review & editing.

Damjana Kogovšek: Investigation; Project administration; Writing – Original draft; Writing – Review & editing.

Jelena Kuvač Kraljević: Conceptualisation; Supervision; Visualisation; Writing – Original draft; Writing – Review & editing.

Ethical Considerations

The study was approved by the Ethical Commission of the Faculty of Education at the University of Ljubljana.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was funded by Naklada Slap as a part of the project Adaptation of Slovenian version of CDI-III.

Declaration of Conflicting Interests

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

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the corresponding author, without undue reservation.

References

Baron-Cohen

Golan

Wheelwright

Granader

Hill

(2010). Emotion word comprehension from 4 to 16 years old: A developmental survey. Frontiers in Evolutionary Neuroscience, 25(2), Article 109. https://doi.org/10.3389/fnevo.2010.00109

Berry

(2009). Making the most of metalanguage. Language Awareness, 14(1), 3–20. https://doi.org/10.1080/09658410508668817

Cadime

Santos

A. L.

Ribeiro

Viana

F. L.

(2021). Parental reports of preschoolers’ lexical and syntactic development: Validation of the CDI-III for European Portuguese. Frontiers in Psychology, 12, Article 677575. https://doi.org/10.3389/fpsyg.2021.677575

Cho

Kim

(2015). Cronbach’s coefficient alpha: Well-known but poorly understood. Organizational Research Methods, 18(2), 207–230. https://doi.org/10.1177/1094428114555994

Cohen

R. J.

Swerdlik

M. E.

(2005). Psychological testing and assessment: An introduction to tests and measurement (6th ed.). McGraw-Hill.

Dale

P. S.

Penfold

(2023). The MacArthur-Bates Communicative Development Inventories: Updates from the CDI Advisory Board. Frontiers in Psychology, 14, Article 1170303. https://doi.org/10.3389/fpsyg.2023.1170303

Edwards

Letts

Sinka

(2011). The New Reynell Developmental Language Scales (NRDLS). GL Assessment.

Edwards

Letts

Sinka

Kuvač Kraljević

Belić

Kologranić Hržica

Kovačević

(2019). Priručnik Nove Reynell razvojne jezične ljestvice [Manual of The New Reynell Developmental Language Scale]. Naklada Slap.

Edwards

Letts

Sinka

Kuvač Kraljević

Kologranić Belić

Hržica

in Kovačević

Penko

Kogovšek

Novšak Brce

Pavlič

Polutnik

Završnik

Vidovič

(2024). Priročnik Novi Reynell razvojni jezikovni lestvici [Manual of The New Reynell Developmental Language Scale]. Naklada Slap.

10.

Eriksson

(2017). The Swedish Communicative Development Inventory III: Parent reports on language in preschool children. International Journal of Behavioral Development, 41(5), 647–654. https://doi.org/10.1177/0165025416644078

11.

Eriksson

Marschik

P. B.

Tulviste

Almgren

Pérez Pereira

Wehberg

Marjanovič-Umek

Gayraud

Kovačević

Gallego

(2012). Differences between girls and boys in emerging language skills: Evidence from 10 language communities. British Journal of Developmental Psychology, 30(2), 326–343. https://doi.org/10.1111/j.2044-835X.2011.02042

12.

Fekonja

Hacin-Beyazoglu

Marjanovič-Umek

(2025). Language development between 30 and 48 months in monolingual Slovenian-speaking children: A study using the Slovenian Adaptation of the Macarthur-Bates Communicative Development Inventory CDI-III. Journal of Child Language, 53(3), 706–722. https://doi.org/10.1017/S0305000925000121

13.

Fenson

Dale

P. S.

Reznick

J. S.

Thal

D. J.

Bates

Hartung

J. P.

Pethick

S. J.

Reilly

J. S.

(1993). The MacArthur Communicative Development Inventories users guide and technical manual. Singular.

14.

Fenson

Marchman

V. A.

Thal

D. J.

Dale

P. S.

Reznick

J. S.

Bates

(2007). MacArthur-Bates Communicative Development Inventories. Brookes.

15.

Frota

Butler

Correia

Severino

Vicente

Vigário

(2016). Infant communicative development assessed with the European Portuguese MacArthur-Bates Communicative Development Inventories short forms. First Language, 36(5), 525–545. https://doi.org/10.1177/0142723716648867

16.

Guiberson

Rodríguez

B. L.

Dale

P. S.

(2011). Classification accuracy of brief parent report measures of language development in Spanish-speaking toddlers. Language, Speech, and Hearing Services in Schools, 42(4), 536–549. https://doi.org/10.1044/0161-1461(2011/10-0076)

17.

Holm

Hansen

P. B.

Romøren

A. S. H.

Garmann

N. G.

(2023). The Norwegian CDI-III as an assessment tool for lexical and grammatical development in preschoolers. Frontiers in Psychology, 14, Article 1175658. https://doi.org/10.3389/fpsyg.2023.1175658

18.

Jarůšková

Smolík

Chládková

Oceláková

Paillereau

(2023). How to build a Communicative Development Inventory: Insights from 43 adaptations. Journal of Speech, Language, and Hearing Research, 66(6), 2095–2117. https://doi.org/10.1044/2023_JSLHR-22-00591

19.

Kas

Jakab

Lőrik

(2022). Development and norming of the Hungarian CDI-III: A screening tool for language delay. International Journal of Language and Communication Disorders, 57, 252–273. https://doi.org/10.1111/1460-6984.12686

20.

Kuvač Kraljević

Blaži

Schults

Tulviste

Stolt

. (2021). Influence of internal and external factors on early language skills: A cross-linguistic study. Infant Behavior and Development, 63, Article 101552. https://doi.org/10.1016/j.infbeh.2021.101552

21.

Kuvač Kraljević

Matić Škorić

Eriksson

. (in press). Komunikacijske razvojne ljestvice – III [Communicative Development Inventories – III]. Naklada Slap.

22.

Lonigan

C. J.

Burgess

S. R.

Anthony

J. L.

(2000). Development of emergent literacy and early reading skills in preschool children: Evidence from a latent-variable longitudinal study. Developmental Psychology, 36, 596–613. https://doi.org/10.1037/0012-1649.36.5.596

23.

Malkewitz

C. P.

Schwall

Meesters

Hardt

(2023). Estimating reliability: A comparison of Cronbach’s α, McDonald’s ωt and the greatest lower bound. Social Sciences and Humanities Open, 7(1), Article 100368. https://doi.org/10.1016/j.ssaho.2022.100368

24.

Marjanovič Umek

Fekonja

Podlesek

. (2012). Characteristics of early vocabulary and grammar development in Slovenian-speaking infants and toddlers: A CDI-adaptation study. Journal of Child Language, 40, 779–798. https://doi.org/10.1017/S0305000912000244

25.

Neuman

S. B.

(2014). Explaining and understanding early literacy. Investigaciones sobre Lectura, 2, 7–14.

26.

O’Toole

Fletcher

(2010). Validity of a parent report instrument for Irish-speaking toddlers. First Language, 30(2), 199–217. https://doi.org/10.1177/0142723709359237

27.

Ouellette

G. P.

Haley

(2013). One complicated extended family: The influence of alphabetic knowledge and vocabulary on phonemic awareness. Journal of Research in Reading, 26, 29–41. https://doi.org/10.1111/j.1467-9817.2010.01486.x

28.

Ravinder

Saraswathi

A. B.

(2020). Literature review of Cronbach alpha coefficient and McDonald’s omega coefficient. European Journal of Molecular and Clinical Medicine, 7(6), 2943–2949. https://ssrn.com/abstract=4443362

29.

Sachse

Von Suchodoletz

(2008). Early identification of language delay by direct language assessment or parent report? Journal of Developmental and Behavioral Pediatrics, 29(1), 34–41. https://doi.org/10.1097/DBP.0b013e318146902a

30.

ŠmitBrleković

KuvačKraljević

. (2023). Parental reports on language development in toddlers and preschoolers based on the Croatian version of Communicative Development Inventories III. Frontiers in Psychology, 14, Article 1188550. https://doi.org/10.3389/fpsyg.2023.1188550

31.

Statistical Office of the Republic of Slovenia. (n. d.). Socioeconomic characteristics of population, 1 January 2024. https://www.stat.si/StatWeb/en/News/Index/13268

32.

Stolt

(2023). Internal consistency and concurrent validity of the parental report instrument on language in pre-school-aged children – The Finnish Communicative Development Inventory III. First Language, 43(5), 492–515. https://doi.org/10.1177/01427237231167301

33.

Tulviste

Schults

(2020). Parental reports of communicative development at the age of 36 months: The Estonian CDI-III. First Language, 40, 64–83. https://doi.org/10.1177/0142723719887313

34.

Westby

Culatta

(2016). Telling tales: Personal event narratives and life stories. Language, Speech, and Hearing Services in Schools, 47(4), 260–282. https://doi.org/10.1044/2016_LSHSS-15-0073

35.

Zaretsky

Kuvač Kraljević

Core

Lenček

(2009). Literacy predictors and early reading and spelling skills as a factor of orthography: Cross-linguistic evidence. Written Language and Literacy, 12(1), 52–81. https://doi.org/10.1075/wll.12.1.03zar