Identifying Levers for Improvement: Examining Proximal Processes and Contextual Influences on Preschool Language Development

ABSTRACT

Research Findings: The present study investigates both the proximal processes and contextual influences on children’s oral language development in preschool. We examine whether teacher language practices vary across activity settings and program type, which teacher language practices predict children’s oral language skills, and potential differences in children’s language learning trajectories. The instructional time in a diverse sample of 15 preschool classrooms was recorded and coded for teacher language practices hypothesized to support children’s language development, including conceptual and interactive talk. Children (n = 99) were assessed at the beginning and end of the school year with a comprehensive measure of oral language development. Results indicated that there was substantial variation across activity settings and program type, with more beneficial language practices used more often in large group settings and in preschools with extensive professional development and coaching support. The specific teacher language practices of vocabulary talk and elicitation practices were positively related to children’s language, controlling for pretest scores, child age, and program type. Practice or Policy: Findings indicate several leverage points for intervention, such as balancing time spent in different activity settings, maximizing the potential of small group time through elicitation strategies, and building in additional supports for interactive language during centers time.

Disclosure Statement

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

Notes

1. For the decomposition of CELF Fall into within and between, we used the software Mplus which created the corresponding latent variables that are known as more reliable than a group-mean centered variable (within) and group means (between) of CELF Fall (Lüdtke et al., 2008).

2. After the number of latent classes is determined, it is a common practice to introduce potential regressors to explain the extracted latent classes and investigate whether such variables are associated with the latent class membership. However, we did not conduct the conditional GMM with regressors of latent classes because we did not take into account the nested data structure in GMM due to sample size limitation. Because we conducted a single-level GMM (not multilevel GMM), students were grouped together into latent classes regardless of their classrooms and schools. However, the primary variables of this study were at the teacher level and thus could not explain the latent classes in which students from different classrooms and schools were mixed together. Multilevel GMM is appropriate with the nested data, but due to the small number of teachers, single-level GMM was conducted which is the limitation of this analysis.

Additional information

Funding

This work was supported by the University of South Florida New Researcher Grant to Elizabeth Burke Hadley.