Suzanne MCarthy <suzmccarth at yahoo dot com> wrote:

>> Script typology isn't primarily for "encoders." It's mostly for (and
>> by) scholars of writing systems. Modern character coding systems do
>> generally try to encode a script according to its inherent structure,
>> though.
>
> So Cree and Tamil have some similarity in the way they are coded?

In a sense. Cree is encoded as a syllabary, which is what it is, and
Tamil is encoded as an abugida, which is what it is.

> If encoders think they have a system, why shouldn't those working
> on, or trying to choose, input methods have a system?

I didn't say they shouldn't have a system, and I don't think anyone else
said so either. Keyboard and input-method design is important and
deserves a lot of thought.

>> You think the fact that Tamil displays in non-linear sequence is what
>> makes it an abugida?
>
> No, non-linear sequence somehow distinguishes it from an alphabet.

Not true at all:

* An alphabet has symbols for both consonants and vowels, and all are
mandatory.
* An abjad is similar, but symbols for most vowels are optional.
* An abugida has the concept of an inherent vowel, which must be
explicitly overridden if not present.
* A syllabary has a separate symbol for each syllable. The shape of
these symbols may be systematic (Ethiopic) or not (hiragana).

Notice that linear and non-linear sequence was not mentioned in this
taxonomy. In fact, there's really no reason why an alphabet couldn't
have reordrant letters, though I can't think of such an alphabet
offhand.

-Doug Ewell
Fullerton, California
http://users.adelphia.net/~dewell/