Used point-light displays (lights corresponding to the joints of the human body) to examine 3-year-olds' understanding of verbs. Found that children could extend familiar motion verbs (walking, dancing) to videotaped point-light actions shown in an intermodal preferential looking paradigm. Children watched the action matching the requested verb significantly more than they watched the nonmatching action. Additional results validated the findings by having children spontaneously produce verbs for these actions. (Author/KB)