Language Evolution and Change 3 4 Morten

6 No direct evidence remains from before the emergence of writing 7 systems to inform theories about the evolution of language. Only 8 as evidence is amassed from many different disciplines can theo9 rizing about the evolution of language be sufficiently constrained 10 to remove it from the realm of pure speculation and allow it to 11 become an area of legitimate scientific inquiry. To go beyond ex12 isting data, rigorously controlled thought experiments can be used 13 as crucial tests of competing theories. Computational modeling has 14 become a valuable resource for such tests because it enables re15 searchers to test hypotheses about specific aspects of language evo16 lution under controlled circumstances (Cangelosi and Parisi, 2002; 17 Turner, 2002). With the help of computational simulations, it is 18 possible to study various processes that may have been involved 19 in the evolution of language, as well as the biological and cultural 20 constraints that may have shaped language into its current form 21 (see EVOLUTION AND LEARNING IN NEURAL NETWORKS). 22 Connectionist models have played an important role in the com23 putational modeling of language evolution. In some cases, the net24 works are used as simulated agents to study how social transmis25 sion via learning might give rise to the evolution of structured 26 communication systems. In other cases, the specific properties of 27 neural network learning are enlisted to help illuminate the con28 straints and processes that may have been involved in the evolution 29 of language. This article surveys this connectionist research, start30 ing from the emergence of early syntax and continuing to the role 31 of social interaction and constraints on network learning in sub32 sequent evolution of language and to linguistic change within ex33 isting languages.