subphonemic orthographies [was: Re: Rotokas (was: California Cheeseburger)]

On Jun 17, 2004, at 8:11 AM, Mark P. Line wrote:

> I can't think of any examples right now in modern, phonologically
> engineered orthographies. Can you remind me of some?
The Fort Hall/ISU orthography developed for North(west)ern Shoshoni
might qualify. Shoshoni has 12 consonant phonemes
/p,t,c,k,k_w,?,m,n,s,h,y,w/ and 6 vowel phonemes /i,1,u,o,a,e/ plus
distinctive vowel and consonant length. There is a great deal of
consonant allophony; voiceless stops are realized as voiced fricatives
between vowels, as voiced stops following nasals, and as voiceless
fricatives following /h/. In addition, alveolar stridents are realized
as alveopalatal stridents following front vowels, and plain alveolar
obstruents are realized as dental obstruents following front vowels.

In the late 60's, Wick Miller developed a practical orthography for
(Western) Shoshoni which was strictly phonemic. The consonants were
mapped to the graphemes {p,t,ts,k,kw,',m,n,s,h,y,w}, and the vowels
were mapped to the graphemes {i,e,u,o,a,ai}. Long segments are
represented as sequences of short segments (e.g. {moppo} [mop:o]
'mosquito', {anii} [ani:] 'beaver') There are two important points to
note. First, there are digraphs in the practical orthography which
correspond to unit phonemes {ts}, {kw}, and {ai}; second, the mapping
of /e/ to {ai} is potentially ambiguous, since there is also a vowel
cluster /ai/. In a large number of words, the two pronunciations are in
free variation; e.g., /maison/ 'cricket' which can be pronounced
[meSo~] or [maiSo~]. However, in other words the value of {ai} does not
vary; so {haintseh} is always [haiJtS1_0] and never *[heJtS1_0], and
{kai} is always [ke] and never *[kai]. But there are no minimal pairs
involving [ai]/[e], so the choice of {ai} for /e/ is justified.

The Fort Hall orthography was developed in cooperation by a Shoshoni
language teacher from the Fort Hall Reservation and a linguist at Idaho
State University. It departs from the strictly phonemic Miller
orthography in positing different symbols for long and short stop
consonants and in representing in full the allophony of /c/ and /s/.
Short stop consonants are represented by {b,d,dz,g,gw} while long
consonants are represented by {p,t,ts,k,kw}. Between vowels, /c/ is
realized as [z] and following nasals it is realized as [dz]. If the
preceding vowel is a front vowel, /c/ is realized as [Z]
intervocalically and [dZ] following a nasal. /s/ is realized as /S/
following a front vowel. These predictable variants are represented in
the orthography:

[z] {z}
[dz] {dz}
[Z] {zh}
[dZ] {j}
[s] {s}
[S] {sh}

Here are some head-to-head comparisons:

'older sister' Miller: {patsi}; Ft Hall {bazi}
'antelope (fawn)' Miller: {wantsi}; Ft Hall {wandzi}
'breast' Miller: {pitsi}; Ft Hall {bizhi}
'friend' Miller: {haintsi}; Ft Hall {hainji}
'wing' Miller: {kasa}; Ft Hall {gasa}
'almost' Miller: {peaise}; Ft Hall {beaishe}

These deviations from the strictly phonemic Miller orthography were
undertaken in the belief that they would make the language more
accessible to native English speakers who are interested in learning to
read and write Shoshoni. Initial voiceless stops are unaspirated and
therefore sound like English "voiced" stops, so using {b,d,g ...} to
represent them is a good move. Often the front vowel which triggers
palatalization of the sibilants is deleted (though not always), so
showing the allophony of the sibilants is also a good move. So in that
sense the orthography is "phonologically engineered", since it takes
the phonological habits of Shoshoni and English speakers into account
in a principled way.

So while this does not directly address the issue of Rotokas's
orthography, it is an example of a modern, phonologically engineered
orthography with built-in subphonemic distinctions.

Dirk
--
Dirk Elzinga
Dirk_Elzinga@byu.edu

"I believe that phonology is superior to music. It is more variable and
its pecuniary possibilities are far greater." - Erik Satie