The present thesis describes the development of a method for the analysis of human inferential communication processes in man machine systems. First an overview about the determinants of pure technical systems is given, to get a better understanding of those processes which would take place in human communication. Because the role and the power of sender and receiver is different in pure technical system from those in human communication, the implicatures are discussed to make clear what must happen to reach a higher standard in the usability of interfaces. Because methodological demands are fundamental in this field, the method called the Berner System is described because it is the basis of the IVA. After this the results of two studies, one a text editor (MS-Word) and a learning system (Lasersicherheit), are discussed. This studies make clear that using the method IVA the researcher is able to find those situations in the interaction between men and computer where problems inhibit the efficient use of the software. A special feature in the IVA was realised that enables the researcher to generate data of the subjects covert physiological processes. That was needed because inferential communication processes take place below the threshold of consciousness. Those data show which features of the interface disrupt the psychophysiological balance of the user and inhibit his computer based work. The exposure of those lacks of man computer interaction via IVA gives the software developer useful clues for what must be changed in the interface to make it more useful.