Language is created and used by networks of people. The size of the network can influence ease of communication. For example, communication is harder in larger networks because there is greater variability in larger communities, it takes longer for information to spread, and there is less shared knowledge. In this talk I will show how the communicative challenges that larger groups encounter shape the language they create. In particular, using cross-linguistic analyses, experimental data, and computational simulations, I will show that the communicative challenges that larger communities face lead them to create languages that are easier to learn and use. I will illustrate this with evidence that languages spoken by larger communities have more systematic grammar, a communicatively-preferred word order, and more iconic words. I will also show how the size of the community influences how the community will categorise semantic domains. Together the findings show that taking a networks perspective can shed light on why languages evolve along certain trajectories and account for patterns of cross-linguistic differences.