It has been shown [J. C. Lilly, Man and Dolphin (Doubleday & Co., Garden City, N. Y., 1961); Science, 15 (1965)] that Tursiops truncatus can be induced to produce airborne sounds in response to human utterances. These responses [“humanoids”: J. C. Lilly, Proc. Am. Phil. Soc. 106, 520 (1962)] bear certain acoustic and visual‐acoustic resemblance to the vocal stimuli. Under effective operant conditioning, the number of Tursiops humanoid emissions in a given response train precisely matches the number of human vocal emissions in the corresponding stimulus train. Tabulations based on the analysis of hundreds of such sonic interchanges demonstrate that Tursiops is capable of accurate matching of number of events for runs of over 40 stimulus‐response interactions, in which the number of events in each train ranged as high as 10. In some experiments, the human was replaced through the use of tape‐recorded stimuli and an automatic fish‐dispenser, and, in some, the food reward was eliminated entirely to prevent its function as a signal. The incidence of either interpolation or overlap of response events with respect to stimulus events disappears with training. Differing vowel color and intonation contour of stimuli are reflected in humanoid variations of structure and contour. The correlation of humanoid and stimulus utterance is evident in acoustigraphic form. Tape‐recorded examples are given, and apparent errors and inter‐species considerations are discussed. [Work supported by the National Institute of Nervous Diseases and Blindness and the National Institute of Mental Health, National Institutes of Health, U. S. Department of Health, Education, and Welfare.]

This content is only available via PDF.