Native and non-native speech database for children.

Abstract

Language acquisition, performance in school, and social interactions all require the ability to compensate for the enormous amount of variability in the speech signal. Foreign-accented speech represents one common, real-world source of speech variability, which can be particularly challenging for accurate speech perception. Although maintaining perceptual constancy in the face of variability is necessary for effective communication, little is known about how listeners develop the skills to compensate for speech variability. As a resource for investigating this issue, I am developing a new database of foreign-accented speech with materials appropriate for children. The database includes 28 speakers with 4 from each of the following language backgrounds: American English, Spanish, Mandarin, French, German, Japanese, and Korean. In each language group, half of the talkers are female. Speakers were recorded reading words, sentences, and paragraphs in English, which have been used in previous research and clinical settings with children. The database also includes assessments of objective intelligibility, perceived comprehensibility, and strength of foreign accent by adult native listeners. With this database, I plan to assess the development of the cognitive-linguistic skills that underlie a listener’s ability to compensate for variability in the speech signal. [Work supported by NIH-NIDCD R21-DC010027-01.]