Re: Glyph shapes and locale (was Unicode CJK Language Myth)

From: Martin J Duerst (
Date: Fri May 24 1996 - 09:11:48 EDT

Ed Hart wrote:

>>From the discussions, users of ideogramic characters have stated a strong
>requirement to see the unified ideogramic characters in 10646/Unicode
>displayed or printed using the shapes they prefer. If this is what
>customers are demanding, developers need to listen to the users.
Definitely true.

>I would
>compare a speaker of English reading text in a fancy script font or Gothic
>font to someone from Japan reading simplified Chinese ideographs: Although
>the text can be read, the person reads much more slowly because the
>characters are harder to discern (read) in these fonts.

The following two things should be carefully distinguished, in order
to not add to the CJK myth:

- Japanese text fully converted to Chinese simplified ideographs.
- Japanese text encoded in Unicode and rendered with a Chinese-style
        Unicode font.

The former is definitely very hard to read. The later may be a little
bit harder than plain Japanese, but in most cases, the difference
is much smaller than between Roman and Gothic styles, esp. for
certain fonts such as Mincho, and in particular on screens (as
opposed to high-resolution printing).

>Moving back to the first requirement, the question is how to specify the
>expected shapes for printing or displaying the unified ideogramic characters
>of 10646/Unicode. For the "correct" or "expected" display and printing of
>unified ideogramic characters in 10646/Unicode, I think that the "locale"
>needs an ideogramic shapes parameter to set the default shapes, whether
>simplified or classical Chinese, Japanese, Korean, or with the addition of
>Vietnam to the WG 2/IRG, Vietnamese. The ideogramic shape parameter would
>apply to "plain text" (coded text without formatting information). The
>ideogramic shape parameter would be in addition to parameters for the
>default language, country, etc. The language, country and ideograph shape
>parameters need to be checked for consistency in a locale. For example, if
>the locale specified the language as Chinese, the default ideogramic shape
>should be either simplified or classical Chinese. Clearly if the computer
>does not have access to a CJKV font or does not have an ideogramic font that
>corresponds to the ideogramic shape parameter, the user is not going to see
>the expected shapes for CJK characters.

I have to say that I am not very fond of the rather rigid Locale mechanism,
but for some cases, it may be very adequate.

However, I don't see the requirement of adding another parameter to the
locale name. If you want to solve this with the locale mechanism, it would
be much better to introduce a new locale *category*. A programmer or
user that indeed wanted Japanese message texts combined with
let's say French dates, German number representation and Vietnamese
glyph shapes, could then easily specify the respective locales for the
appropriate categories. Creating a locale "Japanese for Japan but with
Vietnames glyph shapes" as such does not make much sense.

Another way is of course to do this with the selection of the font(s)
that should be used by the program. Many programs already
provide the possibility to change the font(s) used. Currently, there
are still some problems with the close association of fonts
and encodings, and the limitations for font composition, but
this will hopefully change.

Regards, Martin.

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:31 EDT