John C: > Joe scripsit: > > > Indeed. But you need to represent every single phoneme in every > > single(widespread, native) dialect. That is the nature of the quest. > > Well, a practical diaphonematic orthography doesn't have to go that > far: it can allow for a few multiple spellings, or blur some distinctions > with low functional load. The familiar Wells lexical sets for > vowels (KIT, DRESS, TRAP, LOT, STRUT, BATH, CLOTH, NURSE, FLEECE, FACE, > PALM, THOUGHT, GOAT, GOOSE, PRICE, CHOICE, MOUTH, NEAR, SQUARE, START, > NORTH, FORCE, CURE, and the weak vowels HAPPY, LETTER, and COMMA) make > 27 distinctions, but that's almost certainly overkill. Underkill if we want to treat all accents equally. (Wells's sets don't cover all old surviving distinctions, let alone innovative ones.) > (For me the mergers are TRAP = BATH, LOT = PALM, NORTH = FORCE, CLOTH = > THOUGHT, FLEECE = HAPPY, and NURSE = LETTER, leaving 21 distinctions.) The problem is how to choose which distinctions to blur. Consider THOUGHT, GOAT, FORCE, NORTH, CURE: collapsing all 5 into one would result in too many homographs, but any partial collapse into fewer than 5 is going to run contrary to one accent or another. The usual answer is to ignore the accents of the speakers most ignored or looked down on by everybody else, and to give priority to the accents of the rich and powerful. --And.