It is generally acknowledged that speech addressed to infants (baby talk) is modified in a variety of ways (Ferguson, 1977) which appear to facilitate acquisition and structure the input for the infant. It is also widely recognized, however, that speech produced under natural, rather than laboratory, conditions is complex and contains extensive reductions and coarticulations (Lindbiom, Brownlee, Davis, & Moon, 1992; PerkeIl & Klatt, 1986). Are the acoustic properties of baby talk (ßT) different from those of spontaneous adult speech? If not, the infant's learning of specific sound categories would seem to pose a significant research problem.