The APlausE-MR project aims to gain a robust understanding of the factors influencing the audiovisual plausibility of shared experiences and communication in multi-party mixed reality environments. In this context, a study was conducted to elicit factors which contribute to the concept of plausibility. Building on previous work from the literature, we analyze communication behavior and evaluate the effect of spatial vs. non-spatial audio on plausibility and co-presence in a virtual face-to-face task in multi-modal IVEs and compare it to face-to-face communication in the real world. The study involves dyads completing an interactive conversation task with a shared visual reference to encourage gestural and verbal communication. The study task is carried out in three conditions: face-to-face, embodied VR with simple monaural audio reproduction, and embodied VR with position-dynamic binaural synthesis. This contribution presents results of the study based on questionnaires as well as behavioral and conversation analysis.