Poster #41 - Preschool Speech to Peers is a Paired Process: Dyadic Variance Predominates in Children’s Classroom Vocalizations

• In Event: Poster Session 20

Abstract

Developmental theorists emphasize the role of dyadic processes in language development and empirical research has focused on the interaction of caregiver-child dyads in home settings. Studying caregiver-child dyads, however, does not readily allow us to tease apart dyadic from individual processes. The preschool classroom is one understudied naturalistic developmental context in which the same child participates in many dyadic interactions. Preschool children’s speech could be primarily a factor of who is talking (target child variance), who they were talking to (partner child variance), or dyadic variance (the interaction of target and partner). Moreover, levels of speech might also vary by classroom and observation date. Here, we disentangled these sources of variance using objectively measured peer speech in preschool classrooms.

Participants included 132 children (57 girls; M=44.85 months) enrolled in 13 classrooms. Ninety-eight were Hispanic/Latinx (89 White, 1 Black, 3 multiracial, 5 not reported); 34 were not Hispanic/Latinx (20 White, 9 Black, 4 Asian, 1 multiracial). During monthly school-day-long observations (n=80; M=6 per class, SD=3), children wore lightweight vests equipped with Language ENvironment Analysis (LENA) recorders to measure vocalizations and Ubisense Radiofrequency Identification tags to measure location. Vocalization and location data were synchronized to determine each child’s vocalizations to and from each of their peers while in social contact (determined by proximity and orientation). All children were included as both target children and partners. Each dyad was given a unique dyad identifier such that e.g., Child A and Child B were always Dyad 1 both when A was talking to B and B was talking to A.

A linear mixed effects model predicted the rate of vocalizations each child made to each of their peers. The model included random intercepts of target child, partner child, dyad, classroom, and observation date. Fixed effects were not included to focus on variance in vocalizations associated with random effects, calculated as interclass correlations (ICCs). As can be seen in Figure 1, the dyad effect accounted for the most variance, ICC=.26, SD=8.04, 95% CI(7.48, 8.63); followed by classroom, ICC=.15, SD=6.07, 95% CI(3.71,9.53); date, ICC=.07, SD=4.29, 95% CI(3.44, 5.44); and target child, ICC=.02, SD=2.10, 95% CI(1.50, 2.76). The partner effect did not account for any variance. Model comparison revealed inclusion of dyad, classroom, date, and target child intercepts significantly improved the model ps<.00001, but inclusion of partner intercept did not p>.99, when the other intercepts were included. Figure 2 shows example data from 8 dyads in 1 classroom including the same child.

Results indicate variation in children’s vocalizations at both the dyadic and the individual level. At the individual level, some children speak more to their peers than others, a reflection of individual differences in sociality. However, there was no evidence that some children were preferred over others as recipients of this speech. Instead, dyadic variance was the largest contributor to differences in children’s speech indicating that dyadic interactions are more than the sum of their parts. The results buttress the finding that there are strong dyadic currents in the ocean of preschool classroom speech.

Authors

• Lynn K Perry, University of Miami
  Presenting Author

• Daniel Messinger, University of Miami