Development and Validation of Modified Basic English Lexicon Sentences for Use With a Pediatric Population

This article describes a new sentence corpus developed for use with children 5 years of age and older that is based on the Basic English Lexicon (BEL) sentence corpus and is referred to as the BabyBEL.

Sentences were constructed using words found in the expressive vocabulary of native English-speaking kindergarten and first-grade children and organized into 20 lists of 20 sentences each. A female talker recorded all sentences, and speech-shaped masking noise was created based on these recordings. First, psychometric properties of each list as a function of signal-to-noise ratio (SNR) in a cohort of adults were evaluated. Second, speech recognition in noise by list at -2 dB SNR in a cohort of 5- to 7-year-olds was evaluated. Participants had normal hearing and were native speakers of English.

Results indicate some differences between lists. The speech recognition threshold associated with 75% correct in adult data varied by no more than ±0.5 dB for 17 of 20 lists. Percent correct scores for children varied by no more than 3 percentage points for 12 of 20 lists. The pattern of performance across lists was similar for adults and children.

The vocabulary and sentence structure of the BabyBEL sentences are appropriate for use with children as young as 5 years of age. Recordings of the BabyBEL corpus are freely available for download. When designing future experiments, it is important to consider performance patterns both across lists and between groups of child and adult listeners.

https://doi.org/10.23641/asha.32715975.