Stevertigo wrote:
LD and I were talking about this at length. In attempting to sort out some basic rules for phonetic schemes, it occurred to us that though phonetics are important, they would tend to default to majority accents - which are irreconcilable ethno-political fodder.
For giving the pronunciation of English words to English speakers, I prefer abandoning phonemic transcription for morphophones. A morphophone is a concept such as .long-i., which doesn't specify what sound .long-i. is, only that it's the same sound as in "lie", "fight", "file", "fire", etc. To get a specific sound (either phoneme or phone), you'd have to specify an accent (a dialect or even idiolect) and possibly also a context (the surrounding morphophones). But the translation should be unambiguous for any combination.
For example, in transcribing "fire", we can give .f.long-i.r., and we don't have to argue whether it's pronounced /fai@/ or /fair/, because the rule is that .r. between a long vowel and the end of a word is /@/ for some people but /r/ for others, and that's just fine. (And for some people, it depends on how the next word begins, whether a vowel or a consonant, and that's fine too.)
Of course this has the same problem as a phonemic transcription, that ASCII isn't much up to the challenge of rendering .long-i., but we'll likely have this problem with essentially any solution.
I challenge anyone to show us here a scheme that is both easy to read (SAMPA eugh) and gives all the sonic description that these attempt to.
Well, I still prefer Evan Kirschbaum's ASCII IPA to SAMPA's. ^_^
-- Toby