Influential models in psycholinguistics and cognitive linguistics have shown that successful dialogue involves a process of synchronization across speakers at different semiotic levels. One such model, the Interactive Alignment Theory developed by Pickering & Garrod (2004, 2006), posits that the use of comparable linguistic forms allows speaker to construe comparable mental representations. In this paper, we zoom in on two aspects of this process, using a high-quality video corpus of dyadic interactions. First, we inquire into the sequential organization of interactive alignment, with a specific focus on the question whether synchronization is a gradual or digital process. The analysis reveals that alignment increases as discourse unfolds, leading to (near-)perfect synchrony across speakers. Second, we investigated whether the same temporal dynamics can be measured in different semiotic channels. The analysis indeed reveals similar patterns at the lexical and gestural level, hence providing an argument for a multimodal approach to interactive alignment.