I while back I was playing with a signal generator and a cheap ADC running at 48 kHz and found that if I blasted enough 25 kHz into it I could indeed hear a 1 kHz tone (the expected aliasng product). --Frank
I'm pretty sure the expected aliasing product in that situation is 23 kHz. 1 KHz might be some sort of intermodulation distortion.
I remember seeing the effect with early DAT recorders (Sony PCM-2500). I was using a test set designed for analog recorders that generated and then measured and plotted a sine wave sweep. The sweep started at about 40 KHz, and then went down. As the sweep started, some random (or so I thought at first) noise would show up on the trace, before being erased by the lower frequency sweep. Doing a manual sweep confirmed that this "noise" was aliasing products.
As an aside, the notion that 48 kHz sampling must sound better than 44.1 was mostly a falacy as well. Early digital recorders (including Studers that cost well into the 6 figures), had a fixed multipole anti-aliasing filter that was the same regardless of the sampling frequency you selected. I believe the current state of the art is to over sample the analog signal, and then do the anti-aliasing in DSP before saving the resulting samples to a hard drive, so sampling frequency may actually have some bearing on frequency response (if you think you can actually hear that high anyway).