Abstract
A paradigm for experimental studies of conversation training is developed. This paradigm offers a way of interpreting conservation training studies as well as suggesting several fertile areas for future studies. Several possible uncontrollable sources of error are noted.