In this study, the researchers introduced the linguistic style matching (LSM) algorithm for calculating verbal mimicry based on an automated textual analysis of function words. LSM is calculated using the Linguistic Inquiry and Word Count (LIWC) They applied this algorithm to language generated during a small group discussion in which 70 groups engaged in an information search task either face-to-face or via text-based computer-mediated communication. The researchers found that LSM predicted the cohesiveness of groups in both communication environments, and it also predicted task performance in face-to-face groups.
Moreover, the researchers found that other language features such as word count, pronoun patterns, and verb tense were related to the groups’ cohesiveness and performance. Overall, the study demonstrated the effectiveness of using language to predict changes in social psychological factors of interest, suggesting that this type of automated measure of verbal mimicry can be an objective, efficient, and unobtrusive tool for predicting underlying social dynamics.
