Expression and agreement: some more arguments for the dynamic expression model