Philip,
Thanks for your analysis. --Steve
> STEVE: I am going to post an article extoling Shavian but for
now I would like for
> the group to attempt to explain why Shavian
notation is superior to Unifon
> and the IPA.
PN: Superior for
what purpose?
SB: A dictionary key or whatever. You can state your purpose in
your argument.
PN: I don't know Unifon, but my opinion is that the Shaw alphabet
is
better than the IPA _for representing the phonemes of
English_.
PN: Firstly, IPA is the International _Phonetic_ Alphabet and
is designed
to represent phones, not phonemes; secondly, even though it can
be --
and is -- used to represent a language's phonemes, it's not
tailored
to any particular language, and some sounds (particularly
dipthongs
and affricates) which are phonemes in a language have to
be
represented by sequences of IPA characters.
SB: Abercrombie, Sweet, Jones, and John Wells have used an IPA based
script for
representing BBC English and related dialects. The Association has
never endorsed
a particular mapping of the symbols. There are at
least a dozen in use and while there
may be a 90% overlap, they are all different in some respect.
PN:
For this reason, I believe schemes that are tailored to a
particular
language can do a better job of representing that language's
phonemes;
the Shaw alphabet is one such scheme which happens to have
been
created to represent English phonemes. (Or the phonemes of
a
particular dialect of English; for example, it distinguishes
between
the vowels of "cot" and "caught" but not between the beginnings
of
"which" and "witch", or "horse" and "hoarse" -- the last two
pairs
sound identical for me but I know that there are speakers
who
distinguish between them.)
SB: An IPA based phonemic alphabet has all of these capabilities.
The example is not ideal since there is considerable variation in the
pronunciation of *cot.
cat-cot-caught = kæt- kAt/kOt/kQt -
kO:t where O represents the mirrored c symbol.
Shavian:
kAt - kot kYt - kyt
PN:
On the other hand, I believe the Shaw alphabet is ill-suited to
represent
arbitrary sounds; for example, it's not a good fit to
represent, say,
German or French. The IPA would be better for this
purpose; a
language-specific phonemic transcription even better.
SB: The IPA represents phonemes when the symbols are used as a
dictionary key.
PN: Another advantage of the Shaw alphabet is that since it
represents
phonemes, not phones, it doesn't tie you down to a
particular
realisation of those phonemes. So, for example, people from
Australia,
New Zealand, Michigan, London, and other places, who have
different
phonetic vowel inventories, nevertheless have similar
phoneme
inventories. So while the vowel sound of "day" and "play" may
vary
from place to place, I believe all speakers rhyme those words, so
they
can use the same symbol _age_ for both.
SB: Pitman did something similar with the i/t/a. He insisted that
it was not
a pronunciation guide, only a reading guide. The sounds associated
with
particular letters were dialect specific.
PN: Strictly speaking, IPA should represent how each particular person
pronounces a phoneme, so someone whose _age_ phoneme sounds more like the way
I pronounce _ice_
(Australians?) should write not /EI/ but /aI/ (or
something like that)
in ASCII IPA. This masks the fact that the phonemes
are the same.
SB: So all of those dictionaries that use IPA for their key are
wrong;-)
> Shavian has an IPA inspired keyboard map but it also has
arbitrary
> symbols for some diphthongs.
PN: That's not part of
the Shaw alphabet, though, merely an arbitrary
convention used by certain
fonts which map Shaw alphabet letters to
ASCII symbols so that they can be
keyed in from a regular keyboard on
systems which do not support Unicode
fonts or changeable keyboard
schemes.
SB: I have argued that the keyboard map could be more user
friendly.
The keyboard map is not completely arbitrary but more readable
options
could have been used. Reforming the keyboard map is relatively
simple.
PN: "Proper" Unicode fonts -- which already exist -- do not
need to map
Shaw alphabet letters to English letters at all, but can place
them in
the Unicode Shavian block U+10450 - U+1047F.
PN: I wouldn't
place too much attention to the keyboard mapping.
PN: (If you wanted,
you could also create a font which maps Shaw alphabet
letters to ASCII
symbols in another way, for example -- say, 'c' to
_array_ rather than to
_church_. The important thing is the resulting
shapes; it's they which
represent the sound.)
Philip Newton
<[EMAIL PROTECTED]>
