swampy, there are various possibilities. One is that the recorder synched itself to the incoming bitstream and recorded all the samples as they came in (i.e. at 48 kHz) without skipping or burping--but then stored them in a file that it marked as being at 44.1 kHz, since that's the rate it was expecting.
A receiving circuit can lock onto an incoming signal simply by following its ups and downs, irrespective of whether or not it "knows" the nominal data rate of that signal. (Up is up; down is down; so who's counting?)
If that's what happened, then when you play that file back, it should play back nearly 10% slower (and lower in pitch) than the original sound. That's about one-and-a-half musical semitones--like the distance between, say, an F and a basically out of tune note sitting between E and E-flat. Does the recording sound OK except for the pitch and speed being slow to that degree? If so, various kinds of editing software (e.g. Sound Forge) can simply alter the WAV file header to indicate the correct (48 kHz) sampling rate, and then it should be fine.
Some equipment can accept a bitstream at one sampling rate and record it at a different rate; some equipment does a (possibly unnecessary) conversion even if the incoming bitstream is at the same nominal sampling rate as what you're going to store on disk. But I don't think that the iRiver does that.
--best regards