Petronila: To be complete, I'll give you a few options for an assembled machine as well as a disassembled machine.
Assembled:
With enough knowledge of the internal geometry a current-signature approach can be used where a known voltage is applied to the stator winding and the rotor rotated through at least 360 mechanical degrees while current waveform is monitored. Different fault conditions will exhibit different current signatures. Be warned - there's a lot of detail required of the original design/construction/geometry to make truly accurate conclusions on the rotor condition. The same method can also be used for determining stator faults as well.
Detail necessary includes: number and geometry of both stator and rotor slot, stator coil turns/coil, stator coil pitch, rotor bar pitch, number bars-per-pole, number poles, whether rotor bar slot is "open" or "closed", rotor bar material - if multiple materials are used, method of joining bar to shorting ring, etc.
Disassembled:
If bars remain shorted (which may be possible for generator applications, but is less likely for motor applications), do a modified "growler" test. Modified in this case means applying enough current to create a noticeable magnetic field - which, due to the fewer number of bars, is going to be higher than a comparable induction rotor test. If bars are not shorted, the growler test still applies - but a sensitive enough instrument can read the impedance of individual bars and realistically determine the condition of the bar integrity (cracks and other deformations will show up as higher impedance, usually). Given the large cross-section of the bar and the fact that it is (often) a single conductor, the device needs to be very sensitive to make accurate measurements. If measuring impedance, remember to correct for ambient temperature as well!
Also if disassembled - ultrasound may or may not pick up a problem, but die penetrant most likely will - at least for visible sections of the bar extending past the pole piece. To be honest, I've only used the die penetrant approach when the bar and ring is still connected and I suspect the joint itself is the issue.
Converting energy to motion for more than half a century