--- In qalam@yahoogroups.com, Marco Cimarosti <marco.cimarosti@...>
wrote:
> Richard Wordingham wrote:
> > [...]
> > > the <DELETE> key would erase the whole "ki" cluster. It is
never
> > > possible to
> > > place the cursor "between" a consonant and its matra.
> >
> > That is horrible.
>
> That's always been also my impression. However, *if* this behavior
is really
> what Indian users want, well...
>
> Thinking about it, this is exactly what happens in Western scripts
with
> "accented" letters, and I don't find it innatural (probably just
because I
> am used to it). Typing in Italian I keep using the wrong accent
all the time
> (my dialect misses the phonetic distinctions marked by accents, so
I only
> rely on memory for which accent should be used), still it doesn't
seem
> strange that the DELETE and BACKSPACE kill both the accent and the
letter
> carrying it.

Interesting. Accented letters feel like single letters to me, but
then most of the accented letters I type in are French, where they
do largely function as extra letters. This may be re-inforced by
the fact that they don't appear piecemeal when I type them. On the
other hand, a combination of Thai letter, diacritic vowel and
tonemark does appear piecemeal as I enter the characters, which I
gather is also the case for the aksharas of Indic languages. (I
hope I've got the right term for the C..CV glyphs - to me the
cognate Thai term means a bare consonant, contrasted with the word
(< Sanskrit _svara_) for vowel.) Furthermore, in Thai the akshara
builds up one glyph at a time - there's almost no glyph replacement,
and only one vowel mark - sara am - causes repositioning

It may just be that I type more slowly in Thai, and therefore each
glyph is a greater investment. While I can almost touch type in the
Latin alphabet, for Thai I always have to look before I presss a
key. (At least I usually know where to look.) The loss of a whole
akshara still irritates me even when I use my own web page to use my
QWERTY-based keying system, which does enable me to type faster
until I consider the effort of cut and paste.

I can't help wondering if the Indian deletion systems was adopted
for simplicity, both of application and of explanation. It would be
confusing to hit backspace and see not the character next to the
cursor, but a preposed (i.e. positioned before the consonant)
character/segment, disappear. That doesn't happen in Thai, because
in input and presentation, preposed and postposed vowels have always
been independent of the consonant. Typed Thai can be thought of as
almost having a non-backtracking direction of writing like, but not
as simple as, Korean:

1) Consonant
2) Subscript vowel, if any
3) Superscript vowel etc., if any
4) Tone mark etc. (above the superscript vowel), if any.

When replacing a letter, I still want to place the replacement
consonant before the old one, and then delete the original
consonant, as I could with the purely glyph based input systems. I
must learn that that won't work - I must copy the vowel and tone
mark before I delete the original consonant and thereby lose vowel
and tonemark.

I wonder what Patrick Chew has in mind for Lanna? It seems a single
akshara can have 2 consonants, 2 vowel marks and 2 tone marks in a
vertical stack, though I don't know how much I can trust the Thai
websites I've been learning from - it seems that they wouldn't
allow /ru:p/ to be written in a single vertical stack, though the CR-
Doitong font allows elegant approximations to the four -u(:){p/y}
glyphs of encoding proposal N1013. I've even seen one site write
one of these rhymes as <subscript u:> above (or rather crossed by)
<subscript y> rather than as <subscript y> above <subscript u:> as
in N1013.

Richard.