Poster #41 - The longitudinal predictive value of the MacArthur Communicative Development Inventories III, the Finnish version

• In Event: Poster Session 07

Abstract

Background and aims: From the different versions of the MacArthur Communicative Development Inventories (CDI; Fenson et al., 2007), the CDI III has been used and its validity has been investigated the least. The CDI III is targeted to measure language development of the children aged between 2,6 and 4;0 years (Fenson et al., 2007; Eriksson, 2017). As other versions of the CDI, the CDI III too is based on the parent’s views on their child’s language skills. Different versions of the CDI III are available. The present study focuses on the predictive validity of the Finnish version of the CDI III (FinCDI III), modified based on the Swedish version of CDI III (Eriksson, 2017).
The research questions were:
1. What kind of predictive validity the FinCDI III has in terms of the language ability at 5 years of age?
2. Do the different sections (lexicon vs. grammar) of the FinCDI III provide comparable information on language skills of children at 5 years of age?

Methods: Information on language skills was collected using the FinCDI III from Finnish speaking children aged between 2;6 and 4;0 (N=99). At 5;0, Children’s Communication Checklist (CCC-2; Bishop, 2015; sections: speech, syntax, semantics, coherence) was used to collect information on children’s language skills. The FinCDI III includes the following sections: lexicon, phonology, morphology, syntax and metalinguistic awareness. The sum score for the sections measuring grammatical knowledge (phonology + morphology + syntax) was modified for the needs of the present study. For the CCC-2, the sum score of the 4 sections used in the present study was calculated. Associations were investigated using the Pearson’s correlation co-efficient values. Two linear regression models were modified to test the predictive value of the lexicon and grammar parts of the FinCDI III together with the background factors (gender, maternal education). The sum CCC-2 score was used as a dependent variable both models.

Results: Clear variation was detected in all variables. From the 100 words included in the FinCDI III, children know roughly 68 words (SD 17, min.-max: 7-100). The sum FinCDI III grammar score varied between 0-42 (Mean 27, SD 9). At 5;0, the sum CCC-2 score varied between 0-23 (Mean 8, SD 6). Both the lexicon and the grammar sections of the FinCDI III correlated significantly with the total sum score of the CCC-2 (lexicon: r= - .412, p<.001; sum grammar score: r= -.527, p<.001). Correlations were negative since children with better skills receive lower scores in the CCC-2. Both regression models were significant (<.001) and between 17% (lexicon; F=7.68) and 27% (grammar; F=13.03)) of the variation in language skills measured using the CCC-2 could be explained. From the explaining factors, only the language variable was significant in both models (lexicon β=-.339, grammar β=-.495).

Conclusions: Findings support the validity of the parental estimates when assessing language skills of preschool aged children. Results also suggest that grammatical skills provide more representative information on the developing language capacity of preschool aged children than lexical knowledge.

Author

• Suvi Stolt, University of Helsinki
  Presenting Author