Gary Shannon:
> In conclusion, it appears that my original concept of
> a dialect-neutral phonetic alphabet is not possible.
> In fact, phonetic alphabets in general would be
> useless except in a very small community of speakers
> who all use the same dialect.

An orthography such that the spelling can be fully
predicted from the pronunciation is not practicable.
But an orthography such that the pronunciation (in
a given accent) can be fully predicted from the
spelling is possible in principle, though (for
English) a more practical variant might be one where
some spellings will correspond to two possible
pronunciations, in cases where this will not impede

> That being the case, it appears that spelling of an
> international language like English more or less
> _must_ be somewhat arbitrary.  So those phonetic
> spelling fans, like the Shavians, are really wasting
> their time.  Arbitrary spelling in English (within
> limits, of course) is really the only system that is
> workable.  The best we could hope for in the way of
> reform is to get rid of some of the grosser oddities
> like the "gh" in "ghoti".

The spelling could at least be regularized. IMO the
main desideratum is that the orthography should
facilitate the process of learning to read, so
the grapheme-to-phoneme rules ought to be as
straightforward as is consistent with secondary
desiderata (such as having a common orthography that
treats all dialects equitably, and having an orthography
that does not disguise classical etymologies). The
upshot would be an orthography that contains very
little arbitrariness, relative to the principles
governing its design.