Language features

Bi-di languages are languages such as Arabic and Hebrew whose main direction is right to left ("RTL"). Unlike the case for European languages, providing a localized application for these languages means that your application should be able to handle text manipulation, cursor movement, printing, etc. in bi-directional mode (i.e. both right to left and left to right). Arabic is the primary language of more than 20 countries in the Middle East with a total population of 250 million. What is more, the Middle East offers one of the world's fastest-growing economies. According to Microsoft's Executive Vice President Steve Ballmer, sales in the Middle East increased 40% during fiscal year 1998, and he expects the region to offer the fastest sales growth for the company in the next four to five years. The following Arabic features are the biggest challenges for designing and/or localizing any application to Arabic:

1. Arabic characters are laid out in RTL order, with the exception of numbers, which are laid out left to right ("LTR") as in English. Text is right-aligned on the page, and written from top to bottom (like English).
2. Letters change shape depending on context. Each letter has up to four forms: the initial form, where it is the first letter in a word; the final form, where it is the last letter in a word; the isolated form, and the medial form. This means that as a word is typed, the shape of previously entered letters changes as well as a new letter being added.
3. Arabic and Hebrew both include diacritics. These are marks placed above or below letters which typically represent vowel sounds or other modifiers. These are primarily used for children's and religious texts, but they have to be supported.
4. Arabic includes ligatures, i.e. two letters printed together are replaced by a single new character.
5. Numbers are handled differently in different Arabic-speaking countries. Some use Hindi digits, while others use Arabic digits. However, numbers are displayed from left to right in all bi-di countries.

The figure above shows bi-lingual text (Arabic-English) from the Arabic localized version of Word 97.

The localization process

Although many of the above features might seem very complicated, supporting bi-di is not a major problem in all the necessary localization steps. The following steps will explain briefly, the necessary steps and tools in order to support Arabic (or Hebrew) in a LTR application:

1. Enabling the product

In this step, the application needs to be Arabic-aware (bi-directional). With Arabic OSs such as Win95 and Win98, enabling the product is a matter of calling the right APIs. In this case, many of the above-mentioned features are supported (flipping menus to RTL, bi-di cursor movement control, text display and font support). However, required changes in the original application depend on the complexity level of the supported text manipulation task(s).

2. UI localization

In this area, a UI localization tool is required to translate and flip dialogs, menus and strings. Although many tools making the localization process easy for other languages (e.g. Corel Catalyst) are now available on the market, I have not seen a tool that supports bi-di features (e.g. automatic dialog flipping). The only solution to cope with the lack of tools in this area is either to handle the problem manually or to use an in-house tool.

3. On-line help and documentation localization

Given the existence of many Arabic tools in this area (PageMaker 5.5, Word 97, many DTP applications on the Mac), RTF and Word document localization is fully supported. However, the real challenge is HTML localization. To support bi-di in HTML, certain codes have to be inserted to flip the existing text. Nowadays, with the expansion of the Internet and Arabic-aware explorers, many applications support online documentation in HTML format. However, for bi-di languages, an HTML localization tool still does not exist in the market.

Conclusion

Only a few users today are willing to accept a product that doesn't speak their language. With the Arabic market as fertile as it is today, there is an ongoing demand for Arabic localization. As explained above, not many tools today support bi-di to help to make the Arabic localization process easier and faster. However, I hope that the existence of Arabic OSs (such as Win 95, Win 98, Mac) and Arabic localized applications (such as MS Office and Adobe PageMaker) will encourage other tools manufacturers to support bi-di features.

---

Wafaa Mohiy El-Din
Localization Consultant
Saudisoft Co. Ltd.
Cairo
Egypt
Tel.: +2 (02) 303-2037
Fax: +2 (02) 303-2036
E-mail: Wafaam@Saudisoft.com