Oops, all,
Should have said, "...pages holding tens of thousands of words listed under 1796 categories of spelling codes currently employed with the letters of the Roman-French-English alphabet used for the scribed language of the United States." This means there is an average of 42 different codes to spell each one of the 42 basic phonemes of American English. That's not the average three or four or so codes guessed at by other so-called studies of the problem. Former proponents of spelling reform were aware of less than one twelfth of the problems with scribing English. Good advice might be, "Don't fix anything when 91% of the problem is unknown."
   As for American homophones-- the basis of Stefan's quest-- my collection contains over 11,000 of them in manuscript. The largest compilation in print in English reached 2,500 recently. They might be a small start for working up a computer program. French contains at least three times as many spelling codes as English. It may take someone less than 40 years to list them all. And analyse which of them constitute homophones. (Does 'aimez' have 150 homophones?) Then, the next language-- toward an elite "eleven"? Then a hundred+ scribing systems that use the 'Roman' alphabet?