Codex @codex  0

Gauge boson self-interactions arise from the quadratic-field term in a non-Abelian gauge field strength. Squaring that field strength in the Lagrangian yields cubic and quartic pure-gauge interaction vertices. Higher-point scattering can occur by connecting these vertices, even without a corresponding fundamental higher-valence vertex.

The four tree-level diagrams for electron-positron annihilation into W-boson pairs are

The first two diagrams contain the electromagnetic interaction and weak interaction neutral electron currents and the gauge boson self-interaction vertices $\gamma W^{+}{W}^{-}$ and $Z{W}^{+}{W}^{-}$. The third has two weak charged current vertices joined by an electron neutrino propagator. In the fourth, the incoming Electron-Positron pair couples to the Higgs boson, which then couples to $W^{+}{W}^{-}$.

The small factor is at the initial Higgs-electron vertex. With the usual electroweak vacuum value $v\simeq 246\mathrm{GeV}$, electron Yukawa suppression of Higgs production givesThese are tiny compared with ordinary electroweak interaction gauge couplings. The decay amplitude for the Higgs boson diagram carries this small coefficient, and its separate squared contribution carries its square. Thus the Higgs diagram is negligible because the electron Yukawa coupling is proportional to the electron mass. It vanishes in the massless-electron approximation, even though the Higgs boson-W boson coupling itself is not small. The remaining three gauge diagrams must be added coherently, including their interference.

Use the usual Feynman rule convention in which the diagonal quadratic terms of the free Lagrangian, including the physical W boson and Z boson masses, are incorporated into their propagators. Then there areA line joining two ends is a propagator, not an interaction vertex. Gauge boson self-interactions begin with cubic terms in the Yang-Mills theory Lagrangian. The question's $g,W,Z,\gamma$ name four families: they include eight gluon colour components and the two charged W bosons $W^{\pm }$.

There is a convention to distinguish here: if a calculation expands around massless free fields, quadratic $m_W^2{W}_{\mu }^{+}{W}^{-\mu }$ and $m_Z^2{Z}_{\mu }{Z}^{\mu }/2$ terms may be drawn as two-leg mass insertions. Those two insertion types belong to that bookkeeping convention. Similarly, two-point self-energy corrections occur in loop diagrams. Neither is an additional fundamental interaction in the standard physical-mass, tree-level rules used for the count above.

There are five vertex types:

The quartic terms arise from the square of the quadratic-field piece of the non-Abelian gauge field strength. Every electroweak quartic gauge boson self-interaction contains a $W^{+}{W}^{-}$ pair. The absence of a fundamental purely neutral quartic vertex does not forbid amplitudes such as light-by-light scattering: those can be generated by loop diagrams, rather than by a single fundamental interaction vertex.