Oh, ok, my mistake.
But still, isn't the accidental lepton number symmetry a little different? Consider the charged current term in the lagrangian: \mathcal{L}_e^{\textrm{CC}} \propto \overline{\nu_e}\,\gamma^\mu(1-\gamma_5)\,e\,W_\mu.
Suppose we assign lepton number to the fields like so...