Research

Most research on children's language acquisition occurs in households where children hear the most speech from their parents. However, in many parts of the world, it is not uncommon for children to be cared for by other relatives, or alloparental caregivers. My postdoc focuses on language use in societies where peers, typically older siblings or cousins, provide significant amounts of care.

From a theoretical perspective, this entails that children must be prepared to learn from speakers who may still not have adult proficiency in their language. From a practical perspective, when caregiving is thus distributed, chances for parent behavior to impact children is lower, reducing the potential effects of interventions.

In many of these societies, it is difficult to conduct language acquisition research at a large enough scale to guarantee statistical power. We address these obstacles through the use of a novel speech processing technology, which combines cutting-edge speech processing software with affordable recording hardware, to collect and analyze long-form audio recordings from thousands of hours of child-centered audio.

Currently, we employ these technologies in the Solomon Islands, Timor-Leste, and the Philippines. We have also conducted similar work with a different system in Ghana.

Children's early learning environments differ greatly across households, communities, and countries. As a result, it is vital that the tools we use to assess children's development are sensitive to individual differences in varied contexts.

Partnering with economists working in Ghana, we developed a direct testing battery for children aged 1 to 8 to evaluate the effects of an educational intervention on children's early cognitive development. We used insights from developmental cognitive science to target areas of core knowledge that have been found to emerge across cultures in infancy and predict later performance on related skils in adulthood.

Using data from 1256 children collected longitudinally, we find consistent, above chance performance on tasks from 3 years onward. We also find that early performance on language, numeracy, executive function, and spatial reasoning tasks predicts performance years later on the same tasks. The battery was found to work effectively when used outdoors and in multiple languages.

We are also involved in ongoing work in Timor-Leste and the Philippines, where we are using child-worn recording devices to evaluate the impact of parent-targeted educational interventions on the frequency of child-directed speech in the home.

Children's exposure to speech and conversation in the home is associated with how quickly they learn language. As a result, many early life educational interventions focus on increasing the amount and quality of child-directed speech in the home. To better set expectations for the effectiveness of these interventions, it is important to know how robust this relationship is across different groups of children.

We reviewed 71 studies examining the relationship between parent speech input and children's later performance on tests of language ability. These studies collectively comprised 4760 participants between the ages of 1 and 5. Using a meta-analytical statistics, we found moderate associations between parent speech and children's outcomes across a variety of populations and study designs. 

A key finding was that the number of different words parents used was more strongly predictive of older children's outcomes. We also found that parent speech was just as predictive of outcomes for children who were rarely spoken as they were for children were frequently spoken to. However, we also found some evidence of publication bias, suggesting that studies that find no relationship might be under-reported.

Although many studies have found links between children's language development and their early home environments, few of these studies can disentangle these links from the contribution of children's genetic or maturational differences.

We addressed this problem by studying language acquisition in children adopted internationally into the U.S. Because adoptees are not related to their caregivers, any association between language development and parent speech must be environmental in nature.

We observed interactions between 17 toddlers adopted from China and their adoptive mothers. We found that differences in the growth of their vocabularies was linked to the average length of their mother's sentences. We found the same relationship for 29 preschool-aged adoptees from China and Russian. In addition, we also found that the number of different words mothers produced predicted their vocabulary growth.

Because international adoptees are immersed in English and rapidly undergo attrition of their first language, they can also be studied to distinguish between the effects of their cognitive maturation and gaining language-specific knowledge. 

A previous study of ours found that children first learn words that are both frequently encountered and easily imagined. However, it was unclear if this was because of their immature cognitive or conceptual limitations, or because these words were easier to learn without any other knowledge of the language.

Examining 53 preschool-aged adoptees, we found the same learning patterns: older children also learned more frequent and imageable words first. However, adoptees were less affected by how imageable words were, suggesting that differences in conceptual knowledge or early life experiences may still play an important role.