Examining the robustness of input-outcome associations in children’s language development and evaluating the effectiveness of a novel assessment in rural Ghana

Recent years have seen an explosion of interest in children’s language learning in the home, particularly in how, by talking to their children, parents can help them reach important linguistic milestones. In this dissertation, we address three significant questions that have emerged in the wake of this growing interest. Firstly, how large and consistent are the effects of parent speech on children across a range of measures, populations, and study designs? In our first paper, we conducted a meta-analysis on 75 studies relating child-directed speech to language outcomes. We found that associations between input and outcomes were reliable and roughly the same size across all tracked measures, with no evidence of publication bias. We also found that certain variables, such as child age and study duration, moderated these effects for certain input measures. Secondly, can we disentangle the effects of parental speech on language outcomes from the influence of genes that parents pass on to their children? In our second paper, we examined whether speech from adoptive mothers predicts vocabulary growth in 29 preschool-aged children and 17 toddlers adopted from China and Eastern Europe. We found that vocabulary was positively associated with features of maternal input, suggesting that input effects persist even in the absence of genetic confounds. Finally, can we track language development and other significant cognitive outcomes outside of urbanized Western settings, where children are often raised in learning environments radically different than our own? In our third paper, we evaluate the efficacy of a developmental assessment battery tracking language, executive function, numeracy, spatial reasoning, and mental state reasoning in four age groups of children between 1.5- to 7.5-years designed for use in primarily Akan-speaking households in Southern and Central Ghana. We find good reliability and validity of measures in our older age groups (3.5 years and older) and in language, executive function, numeracy, and spatial reasoning, but poor performance in our 1.5-year-old cohort and in our mental state reasoning tasks. In summary, we find that 1) the effects of parent input are reliable across studies, 2) input effects persist after controlling for genetic confounds, and 3) individual differences in language and other cognitive outcomes can be reliably tracked in a lower middle-income country using our assessment battery.