Another View | Multilingual Web sites

Agencies don't need to blow their entire budget to create multilingual Web sites

Niki Clayton

Contributed

As recently as 15 years ago, multilingual electronic
transactions were relatively uncommon in the United States. Today a
trip to your local grocery offers the option to do business in at
least English and Spanish. But that doesn't begin to address
the growing need for multiple languages in certain
circumstances.


I still get asked why local, state and federal government
agencies need multilingual Web sites. Isn't English the de
facto national language? Perhaps, but an increasing number of
Americans do not speak or read English proficiently.


According to the Census Bureau, approximately 4.8 percent of
American households are linguistically isolated. Moreover, only an
estimated 26.7 percent of those households speak Spanish. Public
schools such as those in Montgomery County, Md., are increasingly
developing multilingual Web sites to communicate with parents who
speak Chinese, Korean and Vietnamese.


Furthermore, the inability to communicate with this segment of
the population in an emergency, such as a pandemic flu outbreak,
puts everyone at risk. The Department of Health and Human Services
has begun to address this concern with its pandemicflu.gov
site.


The motivation to provide public information in multiple
languages extends beyond emergency management. More than 20 federal
agencies have overseas charters that require outreach to a
community in a local language. Topping the list is the CIA, FBI,
and the Agency for International Development. However, it also
includes the Food and Drug Administration which now translates food
safety documents into many languages to help other governments and
businesses understand U.S. safety and import regulations.


Agencies also have a legal obligation ' well, sort of
' to publish their content in multiple languages. An
Executive Order signed by President Clinton in 2000 directs federal
agencies and recipients of federal funds to make their content
accessible to people with limited English proficiency. However, the
same order directs agencies to follow Justice Department guidelines
that say the agency does not need to translate the entire Web site,
but rather should prioritize and translate content on a
document-by-document basis.


Undoubtedly, this guidance is meant to reduce cost and ensure
responsible use of taxpayer dollars. Unfortunately, this patchwork
approach to translation is contrary to software globalization best
practices and can end up costing the government more money in the
long run.


The cost of bad architecture


What happens when a few dozen translated PDF documents or HTML
pages become hundreds or thousands? What happens when two
translated languages grow to five?


Assume a content manager must edit 1,000 documents individually,
embedding the proper fonts, graphics and layout elements for each
language. Let's say an average document takes 20 minutes to
create. That's 333 hours just for one language!


However, the preparation time for multilingual documents can be
cut in half by loading language specific text to a database and
allowing the application code to create documents on-the-fly,
applying formatting.


The guiding principle of software internationalization is this:
Build a single application code base capable of dynamically
displaying any language -- even if your organization isn't
planning to go global any time soon.


Failure to build a flexible architecture may lead to massive
data conversion efforts from ACSII or an ISO-LATIN encoding into
UTF-8 when your organization does decide to expand language
support. The argument that English characters map directly to UTF-8
with no conversion is nonsense. Microsoft Word documents default to
an odd version of ISO-LATIN in which double quotes and dashes will
turn into question marks, square boxes or other gibberish if not
handled by special conversion routines.


Leilani Martinez, bilingual Web content manager for
GobiernoUSA.gov, says although the English 'USA.gov'
web site is larger and more feature-rich than its Spanish
counterpart, multilingual support is now folded into purchase
decisions so that functionality can be added to GobiernoUSA in the
future.


Eight cost-saving tips


The good news is government agencies do not need to blow their
entire budget to create multilingual Web sites. The initial cash
outlay may be greater to do it right, but the long-term savings in
maintenance will be significant. Here is a sampling of ways to
save:



  • Build a single application code base to render multiple
    languages rather than building separate Web sites for each language
    or creating static content.
  • Create a UTF-8 character encoded database using a single
    storage schema to represent data for any language. Although this
    may make initial planning more complex, the infrastructure and
    resource costs over the long run are much lower. Use one database
    to provide redundancy rather than a separate database for each
    language.
  • Develop Extract Transform Load scripts to automate the upload
    of translated text into a database.
  • Localize the look-and-feel of the website by using different
    cascading style sheets for different languages. Do not bake in the
    interface design by constructing divergent front-ends.
  • Overestimate the length of your data fields; it will save you
    money. If you underestimate, you may have a huge data conversion
    effort on your hand when you discover multi-byte characters, such
    as those in Chinese and Japanese, take up more space than you
    allocated.
  • Generate HTML, PDFs and other documents on the fly based on
    content stored in a database.
  • Government agencies should consider reducing implementation
    costs by delivering their individual content through the parent
    agency's portal. For example, many of the agencies in the
    Department of the Interior deliver content through the
    department's Web site.
  • Consider purchasing tools such as Unipad, which make it easier
    for translators to type content text by using the correct UTF-8
    encoding from the beginning. And remember, technology is not the
    only challenge.

The National Library of Medicine's Paula Kitendaugh warns
that translation quality may be as important to launching a
successful multilingual Web site as providing sound technical
architecture.

When the NLM launched the Spanish version of MedlinePlus.com in
2000, focus groups revealed that bilingual speakers often judge the
quality of the information provided by the quality of the
translation, toggling between the English and Spanish sites to
compare content. Kitendaugh says 'because we are providing
medical information, it is very important to us to maintain our
commitment to quality; so we review our content every six months.
That is why we only recently launched the Multilingual Content
Center in over 40 languages. We had to take a look at our manpower
and make sure we could provide the same quality.'


Finding people with the right skill sets is also challenging.
According to Martinez, agencies frequently rely upon translation
staff, rather than bilingual Web developers, to create multilingual
sites. Content management systems with a publishing mechanism to
translate could go a long way to reducing the time spent by
precious human resources.


Whether you are trying to reach all your constituents, or trying
to leverage worldwide collaboration for your organization, building
in the flexibility for languages from the start is the most
cost-effective practice in Web site development for those who wish
to truly reach their intended audience.


Niki Clayton (nclayton@acumensolutions.com) is senior manager
in the public sector practice of Acumen Solutions, a business and
technology consulting firm serving eight cabinet-level agencies in
the U.S. government.



inside gcn

  • ARL seeks private cloud to modernize IT infrastructure

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group