Some of the homophone pairs the author used seem pretty dubious to me: ant/aunt, choral/coral, air/err, awed/odd, veldt/felt. This is partly British versus Amer

Some of the homophone pairs the author used seem pretty dubious to me: ant/aunt, choral/coral, air/err, awed/odd, veldt/felt. This is partly British versus American English, but not wholly. My personal homophony group is not trivial (I think); in particular, I believe that for all pairs of words I regard as having the same pronunciation, the number of "v"s is the same. The counterexample alleged on the linked page is "veldt = felt", but I (like the Dutch, I believe) pronounce "veldt" with a "v" rather than an "f" sound. I can reduce everything else to the identity using only (what I think are) uncontroversial homophone pairs, so I claim that every native English speaker of large enough vocabulary has a homophony group that's either trivial or isomorphic to Z (and generated by "v"). (I did need some uncommon words, though I didn't try very hard to avoid them: od, gneiss, phlox, qat, flyte, lam. And some somewhat-uncommon ones: banns, rapt, wright.)