Developing Latent Semantic Similarity in Initial, Unstructured Interactions

Abstract

We investigated how same-sex strangers develop latent semantic similarity (LSS)—that is, how they come to use words in the same way—in their initial interactions. In a previous study, Babcock, Ta, and Ickes found evidence suggesting that dyad members’ talking, looking, and acknowledging are important behaviors for the development of dyad-level LSS. Using a different sample of initial interactions, we replicated the major findings of Babcock et al., but found that, in both data sets, only those behaviors that introduced words into the conversation were uniquely predictive of LSS. These findings suggest that “the words may be all you need,” and that LSS might develop as effectively in non–face-to-face (i.e., computer-mediated) conversations in which only words are exchanged as in face-to-face conversations in which nonverbal behaviors are exchanged as well.