[Localization] XOs for Cambodia (was...)

Edward Cherlin echerlin at gmail.com
Mon Feb 25 03:58:13 EST 2008


On Sun, Feb 24, 2008 at 6:14 PM, Javier SOLA <javier at khmeros.info> wrote:
> Walter Bender wrote
>
> >> There is a government standard Unicode keyboard...
>  >
>  > Can you please send me a pointer to this? (I am by default using the
>  > standard layout defined by the XKB symbol table that is standard with
>  > Linux, but as you note, it seems suboptimal.)
>  >
>  I am afraid it is the same. There are vowels that Unicode refused to
>  include in the table, because they could be constructed with two code
>  points, but they are quite used standard vowels, and it does not make
>  sense not to have them in the keyboard.

This decision on the part of character set experts, native speakers of
many languages, and linguists in the Unicode Consortium was quite
correct. Unicode encodes characters, not glyphs, and certainly not
keystrokes. It is not the fault of Unicode that almost all software
developers used to work in broken software models for character
handling, including one byte per character and one character per
keystroke.

It is of course possible to type correct Khmer on
one-code-point-per-keystroke keyboards, just as it is possible to type
accented European letters on keyboards that do not have all of them.
Thus compose, `, o creates ò on the US International keyboard for
Linux and on other layouts such as Dvorak. ['Ò' is used in Kreyòl
(Haitian Creole French), which we are currently localizing to.]
Possible, but not always appropriate.

> Technology has to solve the problem.

Exactly right.

The correct solution for Khmer is to create a Khmer IME for SCIM
(Smart Common Input Method), which will be standard on XOs. IMEs can
take in more than one keystroke to choose a character, as for syllabic
writing systems (Ethiopian, Japanese kana, and others) or logographic
writing systems (Chinese, Yi, Mayan, and others), and can generate
more than one Unicode code point per keystroke, as is required by
widely-used keyboards for Yoruba and other languages.

See http://wiki.laptop.org/go/SCIM for information on available SCIM
IMEs, how to install SCIM in current laptop builds, and a link to
instructions for creating SCIM IMEs. I have put in a bug for the lack
of a Khmer IME that satisfies local preferences. Feel free to add
comments and design guidelines, or to take part in creating the IME.

Walter, are you up for creating a table-driven Khmer IME in addition
to your keyboard work?

https://dev.laptop.org/ticket/6568

>  http://sourceforge.net/project/showfiles.php?group_id=133361&package_id=164533
>
> >
>  >> If you think that this is not necessary, then why is
>  >> XO doing localization?
>  >>
>  >
>  > Necessary no. But of course it is better to work with the local
>  > language, so of course we work with local experts to do so.
>  >
>  This is a strong opinion. It disagrees with most educators in the world
>  and their experience. You need a lot of data (evaluations) to back up
>  such opinion.

A strong but empirically verified conclusion (not opinion).

It is of course perfectly logical and appropriate that educators would
observe that children cannot learn by reading textbooks in a foreign
language. (There are a few very carefully designed foreign-language
teaching books using visual cues to the meaning of all words used. I
began learning German as a child from a book called German Through
Pictures, but that is not what we are talking about here.) However,
textbooks are no longer the appropriate context for all educational
questions. Electronic teaching materials do not have to be textbooks.
They can make use of a wide range of built-in interactive software
capabilities that provide active positive feedback to the inquisitive
child.

Illiterate preschoolers are routinely observed discovering how to use
an XO without instruction, particularly the Paint, TamTamJam (music),
Record (camera), and Memorize (concentration game) activities, which
are almost entirely visual. A great deal of work has gone into
minimizing the use of written language on the XO. It does not have
conventional menus, but is icon-driven as much as possible. I count 17
words in the user interface in the Paint activity, and there are
visual cues to the meaning of all of them.

The earliest model for all of this on computerns is Omar Khayyam
Moore's Edison Talking Typewriter,
http://www.winwenger.com/archives/part26.htm, which he programmed to
teach two-year-olds to read and write. A more recent example is the
Indian Hole-in-the-Wall-Computer experiment, in which school dropouts
discovered and taught each other how to use a computer with no
instruction. See
http://www.pbs.org/frontlineworld/stories/india/kids.html, among many
other reports. But the Montessori Method materials, teaching
arithmetic with Cuisenaire rods, teaching music "by ear", and many
other non-verbal teaching techniques predate computers.

>  I have spent the last five years of my life in Cambodia, doing
>  localization, and designing tools like Pootle o the WordForge
>  localization Editor, because I know it NOT to be true,

"It isn't what you don't know that hurts you, it's what you do know
that just ain't so."--Mark Twain, or possibly one of the other
19th-century humorists to whom this has also been attributed.

Anyway, your work on Pootle and WordForge is certainly not wasted, and
we use them, too. We are inviting you and your colleagues to join us,
as we would like to join you, to make both the XO and KhmerOS
successful in helping Cambodians out of poverty and the other problems
they face.

> because I did try
>  at the beginning to teach the use of computers in English to Cambodian
>  students... and it just doesn't work. If this was not true, the
>  Cambodian Ministry of Education would have never changed to Open Source.

Well, this turns out not to be the case, through no fault of yours.
Some forms of education, obviously including the ones you tried,
absolutely require localization, and some forms that the XO team has
been working on can manage without it because they are not dependent
on any language. There is no either/or.

>  Javier

-- 
Edward Cherlin
End Poverty at a Profit by teaching children business
http://www.EarthTreasury.org/
"The best way to predict the future is to invent it."--Alan Kay


More information about the Devel mailing list