The Simulated Evolution of Language paper (see below) describes the Natavist vs. Non-Natavist debate in linguistics. It also describes some theories of the evolution of languages. It seemed to me that the Kirby-Hurford model discussed in section 8 might be a good starting point for a series of Netlogo models testing various theories about language acquisition and evolution, which would be an interesting thesis topic.
The series of models I'm picturing all have the same basic framework. Each agent is equipped with a vocabulary and grammar for some random language (think about context free or regular grammars, for example). The simulation can be in one of several phases: learning, communicating, and mating. During the learning phase agents interact with random neighbors by trying to acquire a bit of the other's language—maybe a few words and a few grammar rules. During the communication phase agents generate phrases from their grammars and pass them to random neighbors. The neighbors attempt to parse the phrases. If successful, fitness points are scored. During the mating phase the fittest agents pair up and produce offspring whose languages are some combination of their parents' languages. Eventually, the old generation dies off and the cycle repeats with the new generation. (Lots of Netlogo models have this structure. In the paper the author refers to this as an A-Life model.)
· Emerging Artificial Societies
· Language Emergence Experiment
· Simulated Evolution of Language