RE: Web page layouts in different cultures - question from DIWG

Thank you Addison for your comprehensive response, even if some of it is just from memory!

You mentioned:

'... expect the site navigation down the left or across the top, with a search box generally in the upper right corner. Progression (not just text, but also "where next is" and "where previous is" is left to right. Other languages move these elements around...'

This is exactly the kind of issue that DIWG will be interested in. The issues around text may have some bearing on the layouts, but from a higher level we are particularly concerned about the division of the page and its subsequent reordering. The left-to-right and top-to-bottom assumption gives us a built-in idea of "next" and "previous", but in a multi-cultural world this idea does not hold. Therefore, to preserve the idea of next/prev it is probably necessary to represent proximity and order within the authored layout so that when the content is adapted to a new layout the proximity semantics are preserved.

The idea of a call with i18n is a good one. Rhys (the DI chair) is no doubt observing this dialogue and will follow up on the suggestion. I think it would be nice to have a BOF or some such meeting at the next TP so we can go over these ideas. The layout work is only just starting, so there's no immediate urgency. (It's just good to be prepared...)

Thanks again.

---Rotan.

-----Original Message-----
From: Addison Phillips [mailto:addison.phillips@quest.com]
Sent: 22 September 2005 17:30
To: Rotan Hanrahan
Cc: Richard Ishida; Felix Sasaki; www-di@w3.org; member-i18n-core@w3.org
Subject: RE: Web page layouts in different cultures - question from DIWG


Hi Rotan,

I'm responding in a more constrained way for a number of reasons. Some of the information below is from memory and this is a *personal* response.

One thing I'll note: I sense that your WG is working on these issues and may be making assumptions based on the responses you receive. It might be beneficial for DIWG and I18N-Core-WG to discuss what you're working on directly (perhaps a telechat or two) so that the information you solicit and the results fit with what you are working on.

I'll respond to your specific questions inter-linearly below, but I must admit that I feel your broad statement about layouts is not correct. Different cultures, influenced by their writing systems, use different layouts and will organize information differently. When reading this thread my impression is that people are thinking in terms of multiple language versions of the same content. Off the Web, the range of variation is much larger than on it because of historical limitations in the technology... something we don't wish to perpetuate.

Bidi mirroring is only the most obvious example of a change, but it is one that can be handled more-or-less through the markup and rendering process. Other type-handling traditions do not adapt so readily. Minority script's text layout traditions may not be well-understood in a Web context, since they are not at all accommodated by existing user agents.

HTML has historically had an inherent left-to-right, top-to-bottom "page oriented" bias based on Western style typesetting built into it. Effective presentation of content, by contrast, may, even in English, require different layouts.

One of the interesting things about having CSS (and the evolution of XHTML) is that the logical organization of information in a Web page and the presentation of that information can divorced from one another. DIWG is naturally interested in how the author's intention in a particular language can be expressed in different devices. Although we may never entirely remove HTML's internal bias towards Western writing systems, the ability of content authors using other scripts to support or adapt their traditional type handling traditions is something we need to work towards.

Anyway, some specific (personal, of course) responses below to your questions.

Addison

Addison P. Phillips
Globalization Architect, Quest Software
Chair, W3C Internationalization Core Working Group

Internationalization is not a feature.
It is an architecture. 

> -----Original Message-----
> From: www-international-request@w3.org [mailto:www-international-
> request@w3.org] On Behalf Of Rotan Hanrahan
> Sent: 2005?9?22? 3:39
> To: www-international@w3.org; www-di@w3.org; bidi@unicode.org
> Subject: Re: Web page layouts in different cultures - question from DIWG
> 
> 
<snip>
> 
> * Western text (e.g. english) uses bullet lists that have the bullets on
> the left, the text running horizontally and the list growing downward
> (vertically). What is the case for languages that write text vertically?
> Or Right-to-Left (RTL)?
[Addison Phillips] 

RTL languages do the reverse with bullets: the bullet appears on the right side with text running to the left. Vertical text has the bullet on top, with indent being from the top (how common this actually is in text is an open question, but I've seen vertical text with bullets like this).

In general, the bullet itself may vary in size, shape, and so forth in order to better distinguish it from text. Different languages and writing systems prefer different sets of bullets.

Organized lists (numbers, letters, etc.) can use a wide array of different numbering schemes.

> 
> * Western text requires several characters per word. So the information is
> dense vertically, but not so dense horizontally. What is the information
> density for other languages? Idiographics etc? For example, in english,
> the phrase "Home Page" fits into a few pixels high, and many more pixels
> wide, but how does the sample phrase in Chinese compare?
[Addison Phillips] 

Information density is not perhaps a good measure. As a general rule of thumb, English is a fairly succinct language. Most alphabetic languages will expand by (approximately) 30% in terms of horizontal text measurement compared to English. But this is a rough rule of thumb. Some text expands by more and some languages will be even more succinct than English (not many though).

Ideographs tend to convey the same number of words in fewer characters--typically only a couple of characters per word. However, ideographs can be complex and legibility is compromised by small point sizes, so a line of Chinese or Japanese (for example) will tend to be taller, even though fewer overall symbols are on the page.

Whitespace and layout also are done differently (in traditional print media) for other writing systems. Western text uses a baseline, with ascenders and descenders. CJK text uses a center line and may use "boxed" character positioning (with the characters evenly spaced). Other writing systems have similar variations (a look at some of the Indic scripts will give you an idea of where the baseline is in those scripts) Overall, concepts like line height and leading may be different if one considers the script and not just how the script got smashed into a Latin-centric text layout system.

> 
> * Are there any special considerations for word-wrapping of vertical text?
[Addison Phillips] 

There are lots of special considerations for word-wrapping of both horizontal and vertical text. See Unicode Technical Report #14 for more information:  http://www.unicode.org/reports/tr14/

> 
> * Are there special text layout considerations for ruby annotations?
[Addison Phillips] 

Yes, of course there are! See http://www.w3.org/TR/ruby

> 
> * Western text often uses coloured underlining to indicate a hyperlink.
> What is the norm for languages where such lines might not be easy to
> notice? Are there such languages?
[Addison Phillips] 

Lines under the text are a Web norm, but they do conflict with some languages. Underlines may detract from the display of certain scripts or interfere with legibility and content authors may suppress the underline using styling. For example, see http://www.asahi.com (the text is more legible without the underlines, but hovering produces the underline and color change, since users are used to links having underlines).

Generally one would not underline CJK text, for example, and one would use various external markup or variations in size and style (in traditional layout) to indicate emphasis. User agents have never, to my knowledge, supported traditional CJK text emphasis, so content on the Web has adapted (as have users). That doesn't mean that the other tradition doesn't exist. It also might not mean that DIWG can/should do anything about it, per-se (users expect underlines).
> 
> * If users could choose, would they prefer portrait or landscape layouts?
> How would the rules of their written text influence their preference?
[Addison Phillips] 

If users could choose, they would choose different layouts based on personal preferences and the material to be presented. The Internationalization Activity home page, for example, is laid out in a landscape format. The W3C home page is portrait. It may be very hard to say that one or another page orientation is the "preference" of a particular language, as the writing system and culture will inform that choice in different ways. Within any writing system or culture one could probably find examples of "the other choice" being attractive and functional at this level. The problem I see is that the cultural "gestures" associated with a particular culture are usually on a lower level of granularity. I look at different corners of a Web page for specific information based on a cultural bias and Web feedback.

For example, in English I expect the site navigation down the left or across the top, with a search box generally in the upper right corner. Progression (not just text, but also "where next is" and "where previous is" is left to right. Other languages move these elements around, so that navigation on a Japanese Web page may very well be down the right side (for example).

> 
> These are just some of the kinds of issues that DIWG participants are
> considering while we work on the layout concepts. These issues may or may
> not have an impact on the general layout technology we eventually propose.
> Nevertheless, we want to be sure that we have as much information as
> possible while we do this work.
> 
> Thank you for your input to date.
> 
> ---Rotan.
> 
> 
> 
> 
> 
> 
> 
> 
> 
> ____________________________
> Dr Rotan Hanrahan
> Chief Innovations Architect
> Mobileaware Ltd
> 
> 3094 Lake Drive
> Citywest
> Dublin 24, Ireland
> E: rotan.hanrahan@mobileaware.com
> W: www.MobileAware.com

Received on Thursday, 22 September 2005 16:51:08 UTC