Internationalizing a Web Site, Part 2
April 19, 2001

by James Mendelsohn

Internationalizing a site poses several technical challenges. These can be divided into two categories, globalization (addressed in Part 1) and localization.

The checklists that follow include information specific to localization, as well as general issues for deploying an internationalized site.

Part 2: Localization

• To present to users the essential functions, content, and design of the site found on the original home site
• To expedite and thereby minimize the costs of translating, most of which are associated with the translation services that businesses must inevitably hire

Both of these purposes may be helped along by the questions in the remaining sections.

• How will users gain access to other language pages of the site?
• Will end users be able to directly access the site in their own languages?
• If not, will the foreign-language site be accessible only from the English home page (assuming English to be the original language of the site)? Or will the foreign-language site be accessible from the different corresponding pages of the home site?
• Is all content on the site localized? If not, does the site alert users in advance to which content isn't localized?
• Will the site link each page of a foreign-language version to the corresponding page of the English version?
• Will the foreign-language pages be included in the same directories as the English Web pages?
• Will the site provide a pull-down menu for users to choose their language? Will that menu be written in each of the various languages it allows users to select?
• Will the site keep all foreign language pages as subpages or will the site acquire new domain names for each country (typically ending with the country-name extension)? It may help to analyze how much content variation there will be from country to country to determine the answer to this question.

• To ease identification by both your company and your translation services, how will files be organized?
  • For example, will file names be prefixed or otherwise noted with a language code? (Prefixing may simplify the creation of subdirectories for each language or language variation.)
  • Will they be further encoded for site versions that are regional language variations or dialects, such as Brazilian Portuguese, American English, or Canadian French? (It may help to use codes that are standardized, such as those from the International Organization for Standardization.)
• How will the files for multimedia elements be organized? Will there be separate directories for global files, for those files modified by language, and for those files particular to one language site?

• What translations (and what foreign language and country-specific pages) do you really need?
  • Where are the principal foreign-language audiences for the site?
  • In what regions are the audiences most likely to grow?
• Does the site need a full-service page for each of the countries it is considering?
• Does it in some cases need only a static, foreign-language page to direct users to local contacts?
• How will content that needs to be translated be directed to the translators? Will the site use software that automatically sends site content to translators?
• What kinds of translation services will the site deploy? Will it use a translation service bureau or a company that specializes in Web content translation?
• Does the translation service employ translators who have the ability and knowledge of a native speaker?
• Is the translation service experienced with translation memory database tools, which enable reusing previously translated text? Using these tools may:
  • Reduce updating costs
  • Help make site vocabulary and style consistent
  • Protect format tags and scripts during translation
• Does the translation service charge by the word? Does it give cost estimates up front?
• Will the site use any automatic translation software? (A cautionary note: While automatic translation is potentially useful, many internationalization experts have urged that humans check automatic translation results to ensure that the translation is not simply transliteration -- that it fits the context and is otherwise accurate.)
• Will each local site be modified as the host site is modified or updated?
• How much time does it take to update and translate content for a local site? Does that affect what content should be posted, rendering some of it obsolete by the time it is ready?
• Will the translator have to localize any scripting? If yes, does the translation service have translators who understand scripting?

• Do each of the translations and each localized version of the Web site reflect the culture and history of the audience? Do the translators adapt content to meet cultural conventions, including the following?
  • Appropriate forms of address and other issues of decorum
  • Voice and tone suitable to address and market to citizens of that country and culture
  • Proper colors and visual symbols (For example, an image of an envelope may be better understood than a mailbox to symbolize mail or e-mail.)
  • Appropriate dialect (the difference, for example, between traditional Chinese, which is used in Taiwan, and "so-called" simplified Chinese, which is used on the mainland of China)
  • Correct understanding of gender roles

• Do the application and Web servers, as well as content management software, support non-English characters? Does the database software? How well do they tolerate international character sets?
• If they support non-English characters, is there adequate documentation for configuring them to make use of their multilingual ability? (This question may be especially important to ask if your site uses a Web-hosting firm.)
• Does the service provider give robust support for non-English characters (and all other aspects of internationalization)?
• Has the site loaded the proper language fonts, especially for Asian and Middle Eastern languages?

• What character-encoding will the site use to ensure users get the proper character sets? How will the site reconcile variations in code pages?
• If you choose between the two most popular encoding schemes -- native encoding or Unicode (most likely UTF-8) -- which will the site use? (Note: Although Unicode claims to be comprehensive, many Asian Web sites are not encoded in it and therefore require additional character sets.) Consider the following before making your decision:
  • Which encoding scheme is most compatible with the site architecture?
  • Which encoding scheme is most compatible with the kinds of browsers the site must support?
• If the choice is native encoding, will the site work with all code pages, and will you be able to anticipate future problems as the site creates other localized versions?
• Does the site specify the character set in each Web page header? (Not doing so may require users to have the technical expertise to adjust their browsers so they can view the page.)
• More generally, does the site correctly display information and functions to users of all local versions? Does it do so from each component of the Web site, including the following?
  • Databases
  • Form elements
  • Component object model (COM) objects
  • JavaScript programming language
  • DHTML
  • ActiveX controls
• Are data sent to and from all forms and databases accurately received and sent?
• Does the source code tell the browser exactly what encoding and character sets to use for foreign languages?
• More specifically, how will the site handle ideographic languages, such as Chinese, Japanese, and Korean? Are the selected character sets common to the most popular browsers? (Note: Although Unicode claims to be comprehensive, many Asian Web sites are not encoded in it and therefore require additional character sets.)
• Will the site provide browsers with instructions for installing the correct character sets and the proper encoding, as well as inputting Asian text and sending e-mails in Chinese, Korean, or Japanese? (This is particularly useful when the visiting browser does not have the character set to match the encoding, in which case it will default to ASCII and produce something unintelligible.)

• Has each country version of the site been adapted to address the customs regulations and regulatory laws for commerce in that country?
• Have the legal parameters of that country been explored, such as:
  • Privacy laws that might restrict your ability to track and collect information on users (as in many European countries)?
  • Acceptable advertisements?
• Do you have local experts who specialize in the legal issues of each country in which your company has localized pages? Will you outsource this expertise?

• How will customer service and response to foreign-language e-mail be provided?
• Will each foreign-language e-mail query generate an automatic response in that language, informing the user that he or she will receive an answer within a certain time period (enough time to perform a translation or secure a native speaker from a translation bureau)?
• Will the site deploy ASP or JavaServer Pages(JSP) technology to respond to a site visitor, once his or her country and language have been identified?
• Will each foreign-language site have native language staff available for immediate response?

Testing and Evaluation
The most efficient use of your company's money and time may be comprehensive testing and evaluation -- from before you begin localizing to the completion and ongoing modifications as the local versions are used. The following questions may help you institute comprehensive testing and evaluation.

• Have you studied successful regional/local sites for a country or language before creating its local versions?
• Have you performed usability studies and reviews with target foreign audiences?

• Will the site have a dedicated tester with native-speaking ability?
• Will each site element be tested during all phases of the localizing project, including testing as a form of quality assurance once the local site is launched?
• Will testing include:
  • Translation -- including grammar, spelling, and cultural propriety?
  • Site design and appearance?
  • Functionality, including all elements of the interface and communication between a browser interface, the server, and databases?
• Before deploying the localized version, will you do test runs to check the interface, site stability, and the ability to further localize?
• Consider including the following kind of testing:
  • Temporarily increasing the word lengths and text sizes, as well as substituting accented characters to see if the localized site can process such changes
  • Creating a prototype to see if translation adversely affects site design
• If creating prototypes or doing test runs, what criteria will you use to select which localized version to test in?
  • The version for the most important customer base
  • The version with the most difficult language to localize
• To save time and money associated with translation, will one phase of testing be done immediately before outsourcing to the translation service, including such issues as:
  • Input and output of foreign-language characters to databases?
  • Appearance of characters in the text on all localized pages and on all elements of the pages, especially controls
  • Ability of those translating to access strings
  • String truncation

• Will the pilot project create a written document -- a kind of template for creating more localized versions of the site? Such a document might:
  • Outline the various stages of localizing
  • Alert local site developers to problems they should notice
  • Provide discrete sets of instructions for the various members of a localizing project -- including engineers, translators, testers, and site administrators -- written in the appropriate jargon for each group
  • Evaluate additional elements in need of localizing for that version and future sites
  • Review localized elements that continue to malfunction and might be globalized or eliminated

• Given the technical and legal complexities and the demand on resources, how much if any of the foreign-language versions should be outsourced to a content delivery service provider?
  • Support services for call centers
  • Legal and regulatory guidance
  • Oversight of the foreign-language version development and management
• If outsourcing, does the provider
  • Test and engineer?
  • Have experience in the kind of localizing you're attempting?
• If outsourcing, how will communication between you and the localizer occur?
• What safeguards exist for containing costs?
• What is the cost of outsourcing completely?
• If outsourcing localization entirely, how will you maintain control over source files? What means will you create for efficiently updating and handling problems on all localized sites?

Promotion
Have the local versions of the site been submitted to international, country-specific, and local search engines and directories? (You might search sites such as Yahoo!, Google, and so on, to locate them.)