From: "Hammarstrom" <haha2581@...>

> > vowels are denoted by diacritics on the basic consonant symbols: e.g.
> > Ethiopic, Indic] and alphabets proper [with both vowels and consonants:
> > e.g. Greek]).
>
> How do they justify the division between alphabets proper and abugidas
> "syllabarys"?

The obvious and sufficient justification is that it is convenient.

> Once you have established that e.g the vowels are
> (consistently) diacritics, then why aren't they alphabets?

Why aren't French diacritics (acute, grave, circumflex and diaeresis)
reckoned as letters?

>That seems
> to require that you have some objective difference between a sign
> and a diacritic?

Say practical and you've got it. With Arabic and Hebrew, the vowel marks,
along with which you might wish to include the geminate marker, and some
(all?) instances of hamza over an alif, are optional and don't affect
alphabetic sorting. If you remove these marks, you get an 'abjed'. Now,
even with an abjed, you get some letters that may serve as vowel, though in
the few examples I can think of, such letters can also indicate consonants.

Vowel marking is mandatory in Indian alphabets (abujidas, not true
alphabets), though they don't always distinguish the default vowel from the
absence of a vowel.

In the Indian alphabets (mostly, if not all abujidas), the absence of a
vowel mark implies either the default vowel or, in some alphabets, no vowel.
This ambiguity can arise because final 'a' in Sanskrit has been dropped by
the time a word reaches the language being written. The diacritic can be
before, after, above or below the consonant(s) it follows in speech.
Descriptions of the system then become much easier if one distinguishes
consonants and vowels.

The Indian alphabet I know best is the Thai alphabet. In Thai, words start
with a consonant (e.g. a glottal stop), and the spelling reflects this.
However, there are two exceptions in the Thai alphabet - it is not required
in words that start with 'reflexes' of the Sanskrit syllabic liquids r. and
l. Thais have long known that their alphabet has 44 letters. When two
letters that were utterly redundant (unless you are interested in Tai
etymology) were abolished, the vowel symbols for Sanskrit r. and l. were
recategoriese as consonants to make the numbers back up to 44.

Incidentally, the glottal stop is not very prominent, and can be elided. I
was very confused once when _ca ao_ 'will eat' was elided to _cao_. I had
very little context to guide me and an instant answer was being demanded.

Thai actually has three tiers - consonants, vowels, and tone marks. Vowels
count in alphabeticisation, with a consonant and its vowel marks being
treating as C(V)(V) even when the vowel precedes the consonant. Tone marks
are only brought in as tie breakers. This immediately provides a practical
reason for not regarding the five vowel marks that precede consonants as
consonants.

The next practical issue I can think of is crosswords. In Thai crosswords,
preceding and following vowels get their own squares, while superscript
vowels don't. The square is distorted to a sort of house shape to provide
room for the the superscript vowel. I presume subscript vowels are treated
similarly, but I can't remember any examples. This leaves the three vowels
that follow the consonant. I can't see any essential reason to classify
them as vowels rather than consonants, not even the alphabetic sequence.
However, Thai text processors don't allow them to be followed by another
vowel.

To add to the confusion, there are two consonants (the glottal stop and the
<w>) in Thai that can serve as vowels, but in writing they are treated as
plain consonants whether they are acting as vowels or consonants.
Additionally, Sanskrit /ar/ is written <r><r> and pronounced /a/ or /an/
accoring to the context, e.g. Sanskrit svarga- 'heaven' > Thai sawan.

In short, the differences are practical. You may remember Miguel saying
that he couldn't see how to classify Egyptain hieroglyphics.

Richard.