Distributional similarity is the idea that the meaning of words can be understood from their context.

This should not be confused with the term distributed representation, which refers to the idea of representing information with relatively dense vectors as opposed to a one-hot representation.