Efficient Estimation of Children's Language Exposure in Two Bilingual Communities

Area of Science:

• Developmental Psychology
• Linguistics
• Speech-Language Pathology

Background:

• Early language input significantly impacts children's speech-language development.
• Daylong audio recordings offer rich, naturalistic data on children's language environments.
• Analyzing extensive audio data requires efficient computational and annotation tools.

Purpose of the Study:

• To develop and validate an efficient method for analyzing child language exposure from daylong audio recordings.
• To estimate the proportion of child-directed speech (CDS) and dual language exposure in bilingual environments.
• To compare sampling-based annotation with full-day annotation and parental reports.

Main Methods:

• Employed a sampling with replacement technique on 30-second segments from daylong audio recordings.
• Collected data from Spanish-English (US) and Quechua-Spanish (Bolivia) bilingual communities.
• Compared estimates from random sampling to full-day annotations and parental reports.

Main Results:

• Stable estimates of CDS and bilingual exposure were achieved using approximately 7% (49 minutes) of the total recording time.
• Strong correlations were observed between sampling-based and full-day annotation methods for bilingual exposure.
• Audio-based dual language estimates did not align well with parental reports.

Conclusions:

• Random sampling of daylong audio recordings is a viable and efficient method for estimating key language exposure characteristics.
• This approach can significantly reduce data processing time, making large-scale analysis more feasible in developmental research.
• Further research is needed to refine methods for capturing accurate dual language exposure estimates.