Tools for extracting word and phrase frequencies from the Child Language Data Exchange System (CHILDES) database via the 'childesr' API. Supports type-level word counts, token-mode searches with simple wildcard patterns and part-of-speech filters, optional stemming, and Zipf-scaled frequencies. Provides normalization per number of tokens or utterances, speaker-role breakdowns, dataset summaries, and export to Excel workbooks for reproducible child language research. The CHILDES database is maintained at < https://talkbank.org/childes/>.
Nahar Albudoor
childeswordfreq is an R package for extracting word and phrase
frequencies from the CHILDES database using the childesr API. It
enables users to:
word_counts().phrase_counts().The primary functions are:
word_counts() for type- or token-based frequencies by speaker role,
with optional stemming, MOR-tier filtering, wildcard matching,
normalization, and Zipf scaling.phrase_counts() for surface phrase counts in utterance text, with
optional wildcard patterns and per-utterance normalization.childeswordfreq does not require CLAN or local copies of CHILDES
corpora. All queries go through childesr. Optional on-disk caching can
be enabled to speed up repeated CHILDES queries.
# CRAN
install.packages("childeswordfreq")
# Or install the development version from GitHub
remotes::install_github("n-albudoor/childeswordfreq")
Then load the package:
library(childeswordfreq)
The package depends on childesr and its requirements. An active
internet connection is required as all queries are executed through the
TalkBank API.
For more detailed guidance on using the package, see the vignette:
browseVignettes("childeswordfreq-best-practices")
MIT.