OK folks, here's my take on it. Traditionally vadose caves were known to be mainly influenced by physical erosion ("corrasion") and phreatic passages were thought to be formed essentially by solution. That was probably because few or none of the people researching speleogenesis studied active phreatic passages (quite reasonably because they weren't divers).
Cave diving became more common in the 70s but even then it was regarded as a relatively extreme technique. Those few who did it were more interested in their own survival than in studying the underwater passages - there was still a fairly widespread attitude around then that a cave dive wasn't actually justified unless there was a good chance of breaking airspace in new dry cave. All that changed with the quantum leap in techniques which came in the late 70s based around the discoveries at Keld Head. People psychologically accepted long submersions, which brought the possibility of spending time doing other things than just focussing on mere survival. So there started to be comments in CDG Newsletter dive logs which were of scientific interest. (One of these was Geoff Yeadon's mention of finding submerged stal at 4.5 m depth in the Marble Steps Branch of Keld Head, for example, circa 1976.)
This trend continued and nowadays certain UK cave dives are done where the main aim is scientific study of the passages. However, most cave diving is done when conditions are at an optimum; low flow and clear water. To see what's really happening with the sediments you have to go there in flood conditions. In many cases this isn't possible in safety. However, certain places sometimes allow us a glimpse of what really goes on in raging flood. Chapman's Rising in Chapel-le-Dale is one such place. After the entrance slot there's 5 metres of rift passage to a 5 m pot down to the main passage. The current howls up this pot in wet conditions and it can provide a fascinating spectacle if you can hang on to avoid being blown out by the fierce current. It's possible to watch great cobbles being blasted UP this pot. When they reach the top they go sideways and drop onto ledges, then rattle back down. Some get back down the pot, only to be blasted up again. Some are caught in the current before they can redescend the pot and are sent back round again, several times. All these cobbles make an incredible noise as they batter the rock walls.
So, not only are large lumps of abbrasive sandstone being transported along; they're going backwards and forwards all the time, almost like a giant file, when considered over the timescale of cave development. Another example is Hurtle Pot; you can't go there when it's in full flood or you'd probably not make it out again. But just downstream from the entrance there's a big (many metres across) pot rising from about -10 m up to about -4 m. In this pot are some enormous cobbles (known to some as the "dinosaur eggs"). We're talking 60, 70 or maybe 80 cm long cobbles here - far too big to pick up. They're always in a different position after every flood and the walls of the pot are pockmarked as the boulders get hurled against the limestone bedrock. I think this is what geologists call "corrasion".
Another example is the Ramp in the Leck Beck Head sump; here hundreds of tons of big pebbles sit precariously on a steep slope up from -30 m to -9 m. In flood they get shoved upslope, only to collapse and come roaring back down, wearing away lots of limestone in the process. (Exactly the same thing probably happens even today in high flood on the entrance ramp of Sleets Gill Cave, which is essentially the same sort of feature.) I could go on and quote many other examples. I could describe underwater rockmills ("moulins") in sumps, with great round pebbles being swirled around. I could mention what look almost like vadose canyons in sumps, perhaps caused by the sediment load being nearer to the floor (due to gravity).
Anyway, hopefully the above convinces anyone who is sceptical about the role of physical erosion in the phreatic zone!