Pragmatic communicators can overcome asymmetry by exploiting ambiguity

How can people communicate successfully when they perceive the world differently (asymmetry) and when words can mean different things (ambiguity)? Prior work has appealed to contextual information to blunt the impact of ambiguity, and explicit feedback to resolve misunderstandings caused by asymmetry. We demonstrate that even without contextual scaffolding and before resorting to interactive feedback, communicators using pragmatic inference can actually exploit ambiguity to overcome asymmetry. While intuition suggests that ambiguity only aggravates asymmetry, agent-based simulations show that to pragmatic communicators it is a ‘helping hand’: They can use ambiguity to trade computationally leaner pragmatic inference for costly interactional turbulence. This computational approach to a longstanding linguistic problem establishes a novel theoretical baseline for explaining natural language use.