A note on complexity in efficient communication analyses of semantic typology

Abstract

Recently the principles of efficient communication have provided useful characterizations of semantic typology: the diversity of attested languages can be described by competing pressures for simplicity and informativeness. While this approach has achieved success in several semantic domains, the formalizations used to define complexity across domains vary. In this note, we list the conditions under which the two main approaches of defining complexity: channel rate and description length, unify and, thus, conclusions about near-optimal communicative efficiency generalize across formalizations. We illustrate this equivalence using simulations of communicative efficiency for Boolean concepts. We round out this note discussing the (un)importance of description languages and the limits on generalizing this equivalence for other behavioral targets for explanation.

Publication
Proceedings of the 46th Annual Conference of the Cognitive Science Society