Investigating social alignment via mirroring in a system of interacting language models