Talking in Timor‐Leste: Urbanization, Peer Input, and Other Influences on Children's Verbal Environments

Children's early experiences with language are influenced by socioeconomic and cultural factors that vary greatly across societies. By expanding the geographical diversity of language acquisition studies, researchers are able to answer broader questions about mechanisms linking children's speech input to their learning process. We present a large-scale study of 277 children between 11 and 58 months old growing up in Timor-Leste, an island nation in Southeast Asia. Children were drawn from households in 72 villages comprising both rural and urban communities. We used child-worn audio recorders combined with an open-source speaker diarization software to quantify their verbal input and production throughout the day. Our analysis found that children were exposed to 15 min of verbal input per hour total, mostly from other children and female adults. Of total input, 6 min/h were produced during conversational exchanges with the child. Children from rural communities heard more adult vocalizations and vocalized more often than those sampled from urban communities. Older children and children with more siblings were exposed to more input from other children and vocalized more often themselves. Higher maternal education was associated with fewer child vocalizations. After adjusting for speaker misclassification, child vocalizations were significantly associated with input from other children and female adults. Our study indicates that verbal input from other children may be especially important in Timorese homes and suggests that a positive relationship between education and adult input may not be universal. These results affirm the usefulness of long-form audio analysis for educators and policymakers interested in children's early learning environments. SUMMARY: Timorese children are most commonly exposed to vocalizations from other children and female adults, but only vocalizations from other children increase with child age. Children from more educated urban communities hear less adult input and produce fewer vocalizations than children from less educated rural communities. Verbal input from all sources is associated with child vocalizations, but input from other children is the strongest predictor. The accuracy, affordability, and flexibility of long-form audio analysis make it ideal as a tool for addressing geographical and cultural gaps in language acquisition research.