Mothers were recorded singing two versions of the same song, one to their i
nfants and the other to their preschool children. In Experiment 1, naive ad
ult listeners accurately identified the infant-directed versions from each
pair of mothers' songs. Pitch was higher for the infant-directed versions b
ut tempo and intensity did not differ across contexts. In Experiment 2, nai
ve listeners judged which version of each pair sounded more "loving" and wh
ich had dearer enunciation of lyrics. The versions sung to preschoolers wer
e rated as more clearly enunciated than those sung to infants. Although the
re were no differences in "loving" tone of voice, "loving" ratings of infan
t-directed versions predicted identification accuracy.