[Egyptian] Towards a deterministic sort for Egyptian Unicode characters

[Michael Everson]

Nigel,

At 13:02 +0100 2007-08-16, Nigel Strudwick wrote:

>I confess to not understanding what this is about.

It is arcane, yes. But you're asking good questions that I think 
everyone might be asking.

>Surely one can sort the transliteration characters, but there is 
>little point in sorting glyphs, as only a relatively small subset of 
>them have phonetic values, and some have several? Aren't we getting 
>beyond the basic encoding of hieroglyphs again?

You are right... the transliteration characters will be sorted as 
Latin characters are. (Typically this will mean that the default 
order will not be as Gardiner orders it but using the Levantine ABC 
order, unless a special Egyptological sorting algorithm is applied.)

But there is indeed a point to sorting the glyphs. Because they are 
*characters*, computers will sort them. If we offer no specification, 
then the characters will be sorted by computers as though the 
alphabet were:

r, a, d, b, x, A, m, w, f, D, i, q, n, S, s/z, h, p, T, H, k, g, t, y, X.

That's not going to be very easy for someone to use. Consider for 
instance someone who has named a number of files with Hieroglyphs. If 
the files are listed in r-a-d-b... order they will be pretty hard to 
find in the list. On the other hand, if they are listed in A-i-y-a... 
order it will be a lot friendlier.

A, i, y, a, w, b, p, f, m, n, r, h, H, X, x, s/z, S, q, k, g, t, T, d, D

1) It's true that only a subset of hieroglyphs have phonetic values, 
but the uniliterals, biliterals, and triliterals are very common, and 
if they were sorted in A-i-y-a... order then more words would appear 
in the right relative position in an ordered list. That will help 
people find them.

2) Alphabetical ordering isn't always as precise as we who grew up 
with the alphabet song. The ordering I proposed won't meet your 
expectations in the same way that the ordering of the Latin 
transliteration characters do, but it's not possible to order 
Egyptian **characters** according to that... it has to be done by 
ordering Egyptian words against another field which contains the 
transliteration.

3) A sorting specification *is* part of the basic encoding of any 
script. ISO/IEC 10646 specifies the content of the Universal 
Character Set, and ISO/IEC 14651 specifies the sorting of the UCS. 
And on the Unicode side, there is the Unicode Collation Algorithm.

So there is no way of getting away from the question of ordering. The 
computer will order the characters in r-a-d-b order if we do nothing, 
and, warts and all, ordering the uniliterals, biliterals, and 
triliterals (and possibly some or more than some ideograms) will in 
the long run be more helpful to users of Hieroglyphs on computers 
than r-a-d-b will be.

That puts us back to my questions:

1. Have errors been made?

2. For signs which have more than one reading, have I chosen the 
correct one? (One must be chosen, on whatever arbitrary or 
non-arbitrary basis we have to make it. Options are: (a) which comes 
first in alphabetical order, (b) which is more frequent, (c) visual 
similarity to another sign which has an unambiguous reading, or (d) 
which wins a coin toss.

3. Should characters which are proper ideograms be ordered according 
to their pronunciation?

Best regards,
-- 
Michael Everson * http://www.evertype.com