Speech and Language Technology in Education (SLaTE 2013)
This talk will review a range of studies illustrating the ways in which speech technology has enabled richer analyses of corpora of children's speech and non-speech vocalizations. One set of studies involves VLAM, an articulatory synthesis system that models the transfer functions of infant- and child-proportioned vocal tracts. VLAM has been used to evaluate cross-language differences in the early emergence of contrasting vowel categories. Another set of studies involves the paidologos corpus of utterances elicited in a picture-prompted word-repetition task from 2- through 5-year-old child speakers of a variety of languages. Tools for tagging the corpus on multiple tiers allow researchers to extract target sounds for spectral analysis as well as for calculating transcribed accuracy rates. The tag sets also can be used to extract stimuli for perception experiments, yielding naive-listener responses that can then become another layer of tags.
Index Terms: child speech corpora, articulatory synthesis, perceptual evaluation of child productions
Bibliographic reference. Beckman, Mary E. (2013): "Enriched technology-enabled annotation and analyses of child speech", In SLaTE-2013, 20-23.