"On the Evolutionary Dynamics of Meaning-Word Associations"

I've read Tom Lenaerts and Bart de Vylder's contribution to Game Theory and Pragmatics. The evolutionary model they have done computer simulations of is very similar to the one that I looked at from a game-theoretical perspective. Some differences are:

• They model a lexicon as an n x m matrix, not a square matrix. Potentially, lexica can thus have too few words to express all meanings, in which case I'm not sure right off the bat what happens.
• They penalize failure and reward success by increasing the contrasts between word-meaning associations by a fixed factor, and they do this for randomly selected speaker-hearer pairs. I looked at the incentives to change the lexicon given any particular profile of lexica in the speech community. I think that's better, because that allows me to quantify the stability of a particular profile.

Other than this, the differences are mainly technical, as far as I could see from a quick inspection.

Natural improvements would include (1) theoretical results in closed form (2) computable gradients for a player in a given landscape defined by a strategy profile (3) applications, such as the discussion of accidental gaps or redundancy in language.