Audio Files

Audio recordings took place each month from 6 months to 17 months of age. Recordings were taken using LENA recorders placed in vests that the target child wore with pockets specifically made to hold the recording device.

There is no audio recording for subject 17 at month 6 (17_06), meaning that there are 527 unique audio recordings (one less than 44 subjects * 12 months = 528).

Audio recordings were intended to be 16 hours long (LENA's default) but due to individual circumstances, some files are shorter than 16 hours.

The length of each recording can be found in recordings.csv.

Audio files were annotated using CLAN software (https://dali.talkbank.org/clan/; note that for MacOS, files were annotated using CLAN for OS 10.14 and below, not CLANc).

CLAN files XX_XX_sparse_code.cha were then transformed into simple csv files XX_XX_sparse_code.csv with one row per annotated noun and columns for each element of the annotation. At this stage, the column for basic level was added.

Audio recordings are stored as .wav files and have been "scrubbed," or silenced post-hoc, in places where the trained coder identified personal information or sections where the file has not been annotated.