Goldstone's theorem states that every continuously broken symmetry produces a gapless (zero-energy) excitation mode. Breaking translational symmetry yields phonons; breaking spin symmetry yields magnons. These Goldstone bosons are the long-wavelength fluctuations of the broken order parameter and appear at all temperatures below the transition.
From your study of symmetry breaking, you know that a phase transition can lower the symmetry of a ground state below the symmetry of the Hamiltonian. A ferromagnet provides the clearest example: the Hamiltonian is rotationally symmetric, but below T_c the spins align in some specific direction — the ground state picks a direction that the Hamiltonian did not prefer. The order parameter (the magnetization) points somewhere, breaking the continuous rotational symmetry. Goldstone's theorem tells you that this is not free: every continuously broken symmetry must produce a specific type of low-energy excitation.
The intuition comes from thinking about the symmetry the ground state broke. If rotational symmetry is broken by alignment along the z-axis, you can ask: what happens if you slowly rotate the magnetization? A uniform global rotation costs no energy — it just moves you to a different but equally valid ground state. But a spatially varying rotation — where spins gradually tilt from one direction to another over a long wavelength — costs only a little energy, and that cost vanishes as the wavelength goes to infinity. These long-wavelength, low-energy "twists" of the order parameter are the Goldstone modes (for magnets, they are called magnons or spin waves). The key is that the energy cost of a Goldstone mode goes to zero as its wavevector k → 0: they are gapless, meaning no minimum energy is required to excite them.
Contrast this with a discrete symmetry breaking, such as an Ising magnet where spins are either up or down. There, the broken symmetry is discrete — you cannot continuously rotate between the two ground states. No Goldstone theorem applies, and excitations typically have an energy gap. The theorem is specifically about continuous symmetries because only there can you continuously interpolate between ground states, generating the smooth long-wavelength deformations that become gapless modes.
Phonons are the canonical example in a crystal. A crystal breaks continuous translational symmetry: the atoms settle into a lattice that picks specific positions. Long-wavelength sound waves — coherent slow displacements of the lattice — are the corresponding Goldstone modes. Their dispersion relation ω ∝ k vanishes as k → 0, confirming the gapless character. In general, the number of Goldstone bosons equals the number of broken continuous symmetry generators, a counting rule that gives a powerful handle on the low-energy physics of any ordered phase without solving the full microscopic problem.