Analog audio doesn’t have a sample rate or a bit depth… it does have a noise floor and an usable frequency response range, but these are not hard limits and the distortion caused by moving beyond them is not as harsh or destructive as it is with digital audio.
When digitizing analog audio, you should record it as PCM with a bit depth of at least 16 at the absolute minimum; but using 24-bit is much better because the digital noise floor is well below anything a human could ever hear, and you have more breathing room in setting your recording levels without worrying about pushing the signal into clipping. With good quality converters, 48 kHz is quite acceptable as a choice of sample rate; with bad converters, a higher rate such as 96 kHz may be helpful since the converter artifacts will be pushed farther up into the ultrasonic range, where we can’t hear them. It can then be converted down to 48 with high quality software later on in order to fit with video standards.
If the bit depth is reduced to 16 on the final released copy to save space, dither must be used to preserve sound quality and avoid adding more digital distortion. Encoding lossy copies from a 24-bit PCM master may result in higher quality than encoding from a 16-bit version. Also be mindful that leaving at least 1 dB of headroom in the PCM is advisable so that clipping distortion is not added during a lossy encode, or by the conversion to analog during playback.