A Model of Prenatal Acquisition of Speech Parameters
World Scientific series in 20th century physics · 1995
An unsupervised neural network model induc- tively acquires the ability to dis h categorically the stop consonants of English, in a manner c ent with prenatal and early postnatal auditory experience, and without reference to any spcialz knowledge of ligisic structure or the proper- ties of speech.This argues against the common assumption that inguisic knowledge, and spee perception particular, can- not be learned and must therefore be innately specified.