Interaction is a significant and dynamic aspect of human language use, however, investigations into the emergence and evolution of language do not adequately consider how interaction facilitates such processes. In this dissertation, interaction is considered both broadly- as in among the “causes” behind language emergence- and specifically- as between co-participants in a language use context. First, the complex adaptive systems approach is applied to the many theories of language origins and evolution, proposing that multi-causality brought about through interacting forces can lead to language emergence in humans. Looking at how children learning language interact with caregivers, the ability to pull together resources to make meaning, even without full language, becomes clear. This ability from natural language use and learning is what informs the experimental investigation of interaction’s affect on language emergence and evolution. In the lab, we have used the iterated learning paradigm (Kirby, Cornish, & Smith, 2008) but adapted it to face-to-face interaction involving a gradual turn-over of participants to simulate transmission over generations. Using silent gesture (Goldin-Meadow et al, 2008; Schouwstra, 2012), participants in the first experiment communicated with and matched gestures to a selection of target images involving a ball moving in a specific manner and path. Over generations, gesture time and diversity decreased (participants’ gestures became more aligned). Moreover, lineage-specific eye gaze patterns evolved, which, when deviated from, indicated a need for repair on the gestured form. These repairs, often in the form of clarifications, made elements of the gesture more salient, leading to their fixation in the system. A second study used the same interactive paradigm as the first, but incorporated a condition in which repair could be performed in a third-turn, as it typical of natural conversation. Having to disambiguate easily confusable noun-verb pairs using silent gesture, participants were allowed a “do-over” repair turn or not; both conditions developed a systematic noun marking system, though repair condition chains did so at a quicker rate. More importantly, the increased interactivity, namely via negotiation and repair, drove systematicity more rapidly than previous non-interaction studies.