The study intended to test a number of implications of the alleged communicative functions of nonverbal behaviours. Four sets of specific hypotheses were defined under the general expectation of important effects of the suppression of visible behaviours from a channel of communication. Pairs of subjects were requested to interact either face-to-face or through a wooden screen depriving them of reciprocal visibility. Dependent variables comprised nonverbal behaviours, indices of paralinguistic, syntactic, grammatical and content aspects of the speech, and ratings of the partner and of the interpersonal situation. In spite of the number and variety of measurements, differences in line with the predictions were extremely rare. It is concluded that, better than the communicative ones, other functions such as assistance to the speech encoding processes, probably account for the abundance of nonverbal behaviours among speaking subjects.