Emergent Language Generalization and Acquisition Speed are not tied to Compositionality
This work questions foundational assumptions in emergent communication research, potentially shifting focus from compositionality to other factors for language learning in AI agents.
The paper challenges the assumption that compositional structure in emergent languages leads to faster acquisition and better generalization, showing that non-compositional languages can perform equally or better depending on the task.
Studies of discrete languages emerging when neural agents communicate to solve a joint task often look for evidence of compositional structure. This stems for the expectation that such a structure would allow languages to be acquired faster by the agents and enable them to generalize better. We argue that these beneficial properties are only loosely connected to compositionality. In two experiments, we demonstrate that, depending on the task, non-compositional languages might show equal, or better, generalization performance and acquisition speed than compositional ones. Further research in the area should be clearer about what benefits are expected from compositionality, and how the latter would lead to them.