Acoustics of clear speech from children with normal hearing and cochlear implants

Abstract

It is common among clinicians to ask children to produce their best speech during intervention. However, it is unclear whether children know how to respond to these directions and whether any contingent changes in their production actually result in better speech. Normal and best speech for children with normal hearing (NH) and cochlear implants (CI) were compared on a variety of acoustic measures. The children (7 to 14 years of age) were asked to read ten simple sentences to train up a fictional automatic speech recognizer for children. They subsequently produced the same sentences with their best and then very best, very clearest speech (ostensibly to test the recognizer). Sentence, pause, and vowel duration were calculated for all sentences as well as vowel intensity and fundamental frequency. Additionally, phonetic distinctiveness was evaluated through vowel space size and voice onset time differences between voiced and voiceless stops. NH children demonstrated patterns of best speech that resemble adult clear speech productions. CI children’s speech resembled NH best speech in all elicitation conditions, except that they slowed (their already slower) speaking rate for best speech. The results indicate that clear speech can be produced by children. [Work supported by NIH.]