Clarifying Patterns in Team Communication Through Extended Recurrence Plot with Levenshtein Distance

In this study, we have analyzed the patterns and quantitative features in the verbal data of team communications and explore an indicator to assess the quality of responses to dynamic changes in task demands. We conducted collaborative-task experiments with three-person teams and collected and analyzed the data from these experiments. A coding scheme with twelve categories representing the contents and functions of the utterances in the communications was used to code the data. Then, a recurrence plot (RP) was used to visualize the sequential patterns with the verbal codes in the team communications. We applied the Levenshtein distance, a quasi-distance between two sequential codes, which converts discrete and categorical data into continuous data. We also applied recurrence quantification analysis (RQA) to quantify and analyze the characteristics of the RP. We compared the analysis results with those obtained using a regular RP for discrete and categorical data. The proposed RP that considered the Levenshtein distance visualized the sequential patterns more clearly and provided more comparable RQA measures—such as recurrence rate (RR) and percentage of determinism (DET)—than the typical RP did. The regular RPs were sparse with many single dots and thereby did not reveal clear patterns. This result suggested that the proposed RP could reveal hidden sequential patterns in qualitative data, such as communication and behavioral data, more efficiently than the existing RP could.