Evaluation of Communication in Teleconferencing by Markov Analysis

Abstract

In this paper, a Markov analysis is applied for evaluation of differences between communication in teleconferences and face-to-face conferences. A discernment problem, which is that speakers are unable to discern the nonverbal cues, such as body language or eye contact, of their conversation partners in teleconferences is assessed with an experiment. In the experiment, two groups consisting of three subjects speak in a usual teleconference, improved teleconferences concerning discernment and a face-to-face conference, and the sequence of utterances which indicates who spoke after whom is obtained and coded. The sequence is regarded as a stationary Markov chain, and is tested for whether it is a first-order Markov process, a second-order one, or a third-order one. Moreover, mean first passage time, mean recurrence time and mean visiting time, which are typical quantities used in a Markov analysis, are calculated. These quantities showed that the conversation structure in the improved teleconferences concerning discernment was closer to that in a face-to-face conference than the usual teleconference.