Extracting Word Frequencies From Child Language Corpora: A New Tool for Developmental Psycholinguistic Research

We introduce childeswordfreq, an R-based tool for extracting frequency data from the CHILDES (Child Language Data Exchange System) database. This tool provides researchers with the ability to efficiently analyze word and phrase frequencies across multiple languages and speaker roles within the CHILDES database. The resulting frequency measures can be used for stimulus selection in experimental studies and for analyzing patterns in child language development. To illustrate the tool's utility, we conducted a case study using items from the Expressive One-Word Picture Vocabulary Test-Fourth Edition (EOWPVT-4), comparing their frequencies in CHILDES with adult-based Subtitle-based Word Frequency Database for American English estimates to assess how input frequency differs between child and adult linguistic environments.

The childeswordfreq package provides an accessible route for incorporating information on how frequently words are used in child-directed speech into research workflows. The results of our case study highlight why such access is critical. The CHILDES and SUBTLEX-US frequencies diverged in patterned ways tied to developmental progression, with earlier items on the EOWPVT-4 being overrepresented in child-directed speech and later items being more prominent in an adult corpus. These findings demonstrate how adult norms can diverge from frequency distributions relevant for acquisition. By grounding analyses in child-based input patterns, childeswordfreq strengthens developmental interpretations and supports more ecologically aligned research.

https://doi.org/10.23641/asha.32273607.