It might help to think about things in terms of Weyl spinors. If we work in a chiral basis for the [itex]\gamma[/itex] matrices, then we can think of a Dirac spinor as two different objects, a right-chiral spinor and a left -chiral spinor-
[tex]\psi_{d} = \left(\begin{array}{c}\psi_L\\ \psi_R\end{array}\right)[/tex]
So the Dirac mass term connects right handed with left handed spinors.
[tex]\mathcal{L}_{md} = m\left(\bar{\psi}_R \psi_L + \bar{\psi}_L \psi_R \right)[/tex]
The goal of a majorana mass is to build a mass term with only one of these spinors.
[tex]\mathcal{L}_{mm} = \frac{1}{2} m\left(\bar{\psi}_L C \psi^*_L + H.C. \right)[/tex]
Here, C is the charge conjugation matrix.
From this it should be clear that a Majorana spinor requires fewer degrees of freedom than a Dirac spinor. You can build 3 mass terms with a Dirac spinor, the Dirac mass, a left Majorana mass and a right Majorana mass. You can only build one Majorana mass from a Majorana spinor.
As to the idea that a Majorana particle has to be sterile under electroweak symmetry, this depends on your model. There are GUTs (like SO(10)) that naturally admit a particle that is a singlet under the standard model interactions. In these models, you can give the right hand neutrino a mass related to the GUT breaking scale, which then mixes with the standard model dirac masses, as you suggest.
However- these GUTs are speculative. The majorana mass term for left-handed neutrinos in the standard model does not require such particles. Consider that we can make an SU(2) singlet out of any particle in an SU(2) doublet (call the doublet [itex]\phi[/itex]) as follows-
[tex]singlet = \phi^T\epsilon \phi[/tex]
where [itex]\epsilon[/itex] is a 2d levi-civita. Under an SU(2) transformation T, this transforms to det(T) = 1. So you can add a term to the standard model like
[tex]\phi^T\epsilon \phi (E_L^T\epsilon E_L)^*[/tex]
We can only do this with the left-handed lepton doublet, because of the hypercharge.
Here [itex]\phi[/itex] is the SM higgs and E the SM neutrino/lepton SU(2) doublet. This will give majorana masses to the neutrinos.
Its worth noting that this term has a mass dimension 5 (which breaks renormalization), and so the coupling can be written as [itex]\frac{g}{\Lambda}[/itex], where [itex]\Lambda[/itex] is a mass scale. The smallness of neutrino masses may well be related to the fact that this mass scale is large. In your see-saw models it would be the mass of the right handed neutrino.