Comparing Manual and Automated Coding Methods of Nonverbal Synchrony

The manual coding of nonverbal cues by trained human coders, such as counting the occurrence of a specific behavior or subjective rating about a speaker, is still a major method in the study of nonverbal synchrony. However, its time-consuming nature has been a serious barrier to the development of the field and has made it difficult for new scholars to adopt the technique. Recent advances in automated coding techniques allow researchers to collect nonverbal behavioral data effectively and objectively, but it is unclear how comparable the manual and automated coding methods are. This study, therefore, directly compared both methods of coding in a face-to-face conversation experiment. In the manual coding, a software system C-BAS was employed to count the occurrence of synchrony in gesture, posture, nodding, and other cues. In the automated coding, a cross-correlation analysis and a cross-wavelet coherence analysis were separately performed for the participant’s movement data. The results showed that the manual and automated coding were moderately correlated, and yielded similar significant differences between experimental conditions of conversational involvement in the degree of synchrony. Further, synchrony measured via both coding techniques was significantly associated with post-conversation self-reports. The advantages and disadvantages of both techniques are discussed.