“Multimodality” is a popular current topic, but is understood quite differently: in terms of different sensory systems, semiotic resources and media, for example. The term “gesture” is no less ambiguous, also in the terminology of one the leading scholars in the field, David McNeill. In this talk, I focus on a phenomenon that is even more specific – iconic gestures. By reviewing work by Acredolo & Goldwin (1985,2000), McNeill (2005, 2012), and Andrén (2010), I show how different claims about the emergence of iconic gestures in children are related to different definitions of the concept. I then summarize our own empirical study (Zlatev 2012) of the emergence of iconic gestures on the basis of mimetic schemas, and propose a possible synthesis of the different theoretical claims, following a level-based definition of iconic gestures.