We design experimental games to evaluate the predictive power of the first cheap‐talk refinement, neologism‐proofness. In our first set of treatments designed to evaluate the refinement with its usual emphasis on literal meanings, we find that a fully revealing equilibrium that is...