A novel method and microprocessor-controlled apparatus are provided for improving research and/or communication interaction between human and animals, such as dolphins and the like, or between normal humans and speech-impaired humans, by playing back audible pre-recorded human-language phrases corresponding to and descriptive of the meaning of sounds and/or gestures or responses produced by the animal or the speech-impaired human and in response to stimuli, conditions or environmental events to which the animal (or impaired human) is subjected, wherein the pre-recorded phrases represent what the human would say or describe if subjected to such stimuli, conditions or environmental events; such that during actual real-time interacting with the animals (or such impaired humans) one actually hears "from" the animal or speech-impaired human, spoken language phrases descriptive of their condition.