Modality effect in interactive alignment: Differences between spoken and text-based conversation

A distinctive feature of dialogue is the convergence of linguistic choices. The Interactive Alignment Model posits that this convergence is obtained by a mechanism of priming that operates at all levels of linguistic representation. Under the model, priming is supposed to foster mutual understanding. Experimental research has confirmed that priming increases language production and comprehension ease. However, the influence of constraints caused by the medium of communication on alignment remains unexplored. In this paper, we look at structural alignment in face-to-face and synchronous text-based communication. In our study, we analysed the influence of different factors on structural alignment in two datasets, one of computer-mediated conversation and the other of spoken conversations, and found that the likelihood of structural alignment in both cases is determined by the same factors. However, when comparing the overall magnitude of structural alignment in the two datasets, we found a greater magnitude of structural alignment in text-based conversations. In our view, this result suggests that structural alignment is used to trade off the constraints of the text-based medium.