> From: suzmccarth [mailto:suzmccarth@...]
> Sent: Monday, July 05, 2004 3:08 PM

> I have read many Tamil lists and have read that they have not
> accepted typing in order of phonetic sequence...

It is certainly true that many Tamils, as well as users of other scripts
of India, have strong preferences for visually-ordered entry rather than
phonetically-ordered entry. It is a mistake to think it is a universal
preference, though: there are users who are quite happy with
phonetically-ordered entry as provided by simply input methods.

Of course, that doesn't imply that input methods for visually-ordered
entry shouldn't be provided for those that prefer that paradigm for
keying of text.

The problem for Microsoft, or any other software vendor, is that a bunch
of money can get spent on providing different input methods but you may
still get just as many complaints: there may be lots of keyboard layouts
with varying numbers of adherents -- as you observed,

> Maybe it isn't the best input method but there is no
> consensus right now for Tamil input...

-- and latching onto one that hasn't been determined to be used by a
large proportion of users is prone to leave the level of discontent
about the same. Some might then suggest that they should simply
distribute *lots* of keyboard layouts for e.g. Tamil, but that still has
problems: each one requires resources to develop (with one of the larger
costs being the work to establish and verify the spec), and it is
actually *unhelpful* to the average user to overwhelm a given user with
lots of layouts all but one of which she doesn't care about.

So, when someone from a company like MS says that an IME isn't needed
for Tamil, the fuller story isn't that the person doesn't know what
they're talking about or doesn't care about Tamil users; rather, it's
that the person knows all too well how complex an issue it can be to
provide input methods that users are happy with, that there's a very
limited amount that can be promised given the already-overburdened
resources, and that a simple, phonetically-based layout does provide the
means to enter any text elements that may be needed and is acceptable to
at least one segment of Tamil users.

> FOR ME, typology and input method ARE related.

OK, if you want a typology of input methods, it will not be directly
related to a typology of scripts and writing systems. Rather, it will be
based on factors such as the following:

- the physical mode for input (keyboard, tablet, etc.)

- the number of characters or character sequences that must be
genereated

- whether input is done directly, or involves some intermediate stage

- when there is an intermediate stage in entry, then there are several
sub-factors: the nature of the intermediate stage, the UI involved, the
kinds of processing involved... (both dead keys and IMEs involve an
intermediate stage, but the similarities end there)

- the relationship between the order in which units are entered and the
order in which corresponding units are stored

Except insofar as different types of script tend to involve different
magnitudes in the numbers of characters involved (alphabets and abjads,
in the tens; syllabaries and abugidas, in the hundreds; logosyllabaries,
in the thousands), a typology of input methods would not particularly be
correlated with a typology for scripts.

> The way I think
> about it, I see Tamil as having syllabic characteristcs and then I
> can look for the syllabic IME...

> It is painfully obvious to me that FOR EVERYONE ELSE, typology and
> input are NOT related. So we disagree. Does this mean that I am
> wrong? My way of thinking correctly predicted the existance of the
> syllabic IME.

Yes, I think you are wrong. It is not the significance of the syllable
unit to Indic scripts that makes some users want a different type of
input method. Rather, it's the fact that there is a mismatch between
visual order and encoded order of text elements. That could, in
principle, occur with different types of writing system. For instance,
if English text were encoded in terms of phonemes, then there'd be a
mismatch between visual and encoded elements for words with final "e"
("tale" encoded as t + a-e + l), and users wanting visually-ordered
entry would need a different input method from those that used
phonetically-ordered entry.



Peter Constable