In music the microphone gain is set at the beginning of the session during a sound check. After ensuring a healthy signal-to-noise ratio, the gain is left alone.
In dialogue recording for dubbing the input levels will vary widely and constantly not only with each actor, but also with the same actor depending on the scene. Speech is very dynamic, and dubbing actors tend to be more expressive during a performance than during an ordinary conversation.
As a starting point, the actor can give you an example of the voice they are going to use so that you can set your gain. During a take you may have to do some fader riding to compensate for low level detail such as breathing, but if you set your gain correctly at the start you shouldn’t have to move more than 3 dB in either direction for normal speech. To ride the faders more than that will result in audible dips and boosts that will need to be corrected down the line. Note that "fader" refers also to pots, knobs, or whatever continuous gain controllers are available on your preamp. With time, overuse of those may cause grackles and distortions, which is where the Mic and Bus method below may come in handy.
ADR, Animation: As with compression, fader riding for ADR and animation is discouraged.
The Mic and Bus method
This is a technique specific to dialogue and foley recording, and is used to give the recording mixer flexibility right at the fingertips. Consider this signal flow diagram:
In this scenario, after setting the initial gain, the mic fader is used to set the overall level and determines how much signal goes to the compressor. The actual riding - or fine-tuning - before printing to track is done on the bus fader. This method is used when time is of the essence because it affords control over a wide dynamic range without requiring the recordist to constantly reach over for the mic gain pot or to mess with compression settings.