--- In qalam@yahoogroups.com, Michael Everson <everson@...> wrote:
> At 12:20 +0000 2005-08-11, Richard Wordingham wrote:

> I will develop a QWERTY-based keyboard layout, because that is what
> they will have on their hardware, and as they are all familiar with
> the Latin alphabet (English being the official language of Liberia).

I'm not sure what that means for a syllabic script. Does it mean
using (approximately) alphabetic input of consonant plus vowel, as for
example with SIL's Vai keyboard?

One argument for 'glyph'-based input, as opposed to phonetically based
input, is the changes in the (primary) sound values of the symbols.
Examples (all but SIL taken from the Vai proposal):

1. In both Massaquoi and Dalby, syllable O (I use the proposed Unicode
names for the characters) has the sound /wO/. In Kandakai &
Hutchison, it is explicitly shown as having both the sounds /wO/ and
/O/. The other now-independent glyph variant, HO, has the sound /O/
in Massaquoi! Massquoi used the double dot diacritic for /hO/. There
is some variation in how this was distinguished from MGBO - compare
the contrasts in Massaquoi (Propsoal Figure 4) and Jensen (Proposal
Figure 9).

2. From Dalby, Jensen and Kandakai & Hutchison, the primary value of
EN seems to have been /hE~/ rather than /E~/. HE seems to have
spawned two new syllables - E (glyph variant?) and HEN (by double dot
diacritic). Massaquoi, supported only by Jensen, gives an entirely
different symbol for /E/ that seems to have vanished. I wonder if
this was a deliberate modification of syllable EE.

3. From Dalby, Jensen and Kandakai & Hutchison, the primary value of
AN was /Na/ (initial velar nasal). NGAN (why is NG the only
completely nasal onset after which nasalisation is shown in the
Unicode scheme?) has been formed from it be a double dot diacritic,
and there is variation in how it is shown - sometimes a vertical pair,
sometimes a horizontal pair.

Unfortunately, Massaquoi's 1911 paper has no entry for initial /N/; I
have a horrible feeling the editor deleted its section because it
appeared to contradict the section for initial /Ng/. It's not
impossible - SIL's Vai keyboard mapping confounds them! According to
Tucker, NGAN, or at least the variant with vertical dots, represents
/Nga~/ - a mistake for /Na/?

4. Syllable ON, which has a pair of horizontally arranged dots,
presents even greater variation. From Dalby it seems that the primary
sound was /NO/, but that it also represented /O~/ and /hO~/. Indeed,
in Massaquoi it appears as 'nwo', presumably /wO~/.

In the standard script, NGON differs from ON by the two dots being
arranged vertically. HON is a dotless square version, in which the
two now-squared semi-circular halves are joined by a bar. Apart from
Figure 1, it only appears in Jensen's table.

In the 'Teach Yourself' example (Proposal Figure 2), which gives no
symbol for /O~/, ON represents /hO~/. In Kandakai & Hutchison and the
SIL font (if the documentation of the keyboard is to be believed),
which likewise lack a symbol for /O~/, ON represents /NO/. In the SIL
font, it seems that NGON represents /hO~/; Kandakai & Hutchison has no
symbol for /hO~/.

That's the end of the intended examples.

It's possible that the SIL keyboarding manual simply reflects an
incorrect specification. (It is an accurate description of the
keyboard in these matters.) One reason for concern is that the column
on phonetics shows [h] as the initial consonant for syllables with a
nasal vowel and no initial consonant (as implied by the keying
sequence). The oddities are:

~i gets HIN, but then there is no IN. Moreover, Tucker gives HIN for
/i~/!

~a gets HAN, but then the keying of AN implies its original sound /Na/.

h~u gets NGAN (with vertical dots, not horizontal). A later table
gives unkeyable HUN as /hu~/ so I suspect the key sequence should have
been h~a.

xg~a gets AN. The sequence 'xg~' is a systematic 'feature' for 'x'.
It is currently tolerable because the SIL font does not provide NGGEN;
NGGE is keyed as 'xgq'. ('q' is used for /E/; 'e' for /e/.)

The partial permutation of ON, NGON and HON *might* be an error. It
is a partial permutation because SIL gives no symbol for /O~/ and does
not support HON.

Has anyone here investigated the status of these matters - local
variation or software 'feature'?

> >Does anyone know how standard the reduced set is?
>
> Come on, people. Look at the syllabary charts from the children's
> primers in the proposal document and do the counting yourselves.

Which are the primers? Tucker (Figure 3) and the teach-yourself book
(Figure 2)? And do they reflect reality, or are they just the
'proper' way of writing? Will they reflect the reality when another
generation has died off?

Richard.