Not sure I understand the question but maybe the distinction could be along these lines:

Acquisition bandwidth = sweepwidth i.e. 1/dw were dw = acquisition dwell time for a given experiment (assuming 'normal' as opposed to 'digital' signal acquisition. Think Nyquist.

Receiver bandwidth = the maximum possible sweepwidth corresponding to the shortest dw that you can set on your spectrometer.

They may be similar in scope, but technically they are different terms. Simply put, there is a digitiser and a Fourier transform in between them.

NMR receivers are relatively narrow band devices. They see only the Intermediate Frequency (IF) after the NMR signal is heterodyned down in frequency by a mixer. That's how we observe more than one nucleus with a single receiver. Receiver bandwidth needs to be only wide enough to encompass the frequencies in the FID.

Acquisition BW is 1/PW, is it not? Compared to the hundreds of MHz in the original signals, the observed spectral bandwidth of a few KHz is also pretty narrow (ppm). This might deserve correction, but wouldn't the maximum spectral bandwidth available be determined by the speed of the digitizer divided by 2.2?

The bandwith should be interpreted in conjunction with the term carrier frequency. The spectrometer frequency is the carrier frequency and the NMR information is contained in about a few Kilohertz range (audio frequency) around the Carrier frequency. Intermediate Frequency is also a carrier RF frequency obtained from the original spectrometer frequency for electronic considerations. The NMR information is obtained by the Phase Sensitive detection (demodulating) of the carrier frequency with appropriate reference frequency. Till the end stage of the detection this bandwidth consideration is determined by the NMR signal spread from the carrier frequency which is termed the receive bandwidth. The digitization is by the A/D converter at the final stage of receiving the analogue signal before storing it in the computer memory.

Aravamudhan