Localizing for the Future

by Tina Lieu

Microsoft, by most accounts the world's largest software publisher, ships products in some 30 different languages. Microsoft outsources its localization work in part because the company doesn't want to manage the peaks and valleys inherent to the work. David Brooks, Microsoft's senior director of international product strategy, worldwide product group, notes that to fully handle localization for "one product, we may need 50 to 100 people. If we staffed up to meet the peaks, we'd have an enormous staff. So what do we do when the deadline is over?"

Brooks is in charge of developing Microsoft's localization strategy, and finding and recommending vendors to the company's regional groups. Not surprisingly, he has a firm grasp on the Japanese localization market situation, and is unequivocal in his assessment. "There is a chronic shortage of competent localization firms; the demand clearly outstrips the capacity of the top-tier firms."

"It's one thing to localize documentation for 'straight' MS Word," Brooks says. "But [for] networking, telecommunications, and other mission-critical applications - where documents run into the thousands of pages - it's not something an ordinary translator [can handle], no matter how skilled they are. The linguistic aspect is, relatively speaking, a small portion of the challenge. Firms that can do [the job] best are ones with engineers."

Dimitri Kieffer, East Asia/Middle East group manager of the Microsoft International Business Network, concurs with Brooks' assessment. "There is a worldwide shortage of technical translators working into Japanese. And, at this stage, there's no proper memory translation [tool] on the market for Japanese." [See the -Software physician, heal thyself?" sidebar.-Ed.]

Asked about how the vendor market has changed over the past few years, Brooks says, "Vendors have become more technical.... Compared to Europe, [the quality is] not as good, but it's a lot better than it used to be."

Brooks reiterates the shortage of good localization firms and notes that, to help boost the supply of localization services, Microsoft encourages its localization vendors to find other work in order to be financially stable, not minding who else they work for. This is beneficial, Kieffer adds, because by working for a variety of publishers (software developers), "the vendors gain greater expertise and become familiar with terminology used in the whole industry."

Market competition in Japan
Such comments might lead one to think that the top localization firms in Japan are raking in the clients. But the top-tier firms always face competition from vendors in the "lower price/lower quality" slice of the market. Some publishers, because they don't understand the important role that localization plays in a product's life cycle, tend to use a simple, cold cost-comparison to select a localizer. And this puts pressure on top-tier vendors, who need to charge higher prices to cover the extra cost of their high-quality work.

The "costs" of using a lower price/lower quality vendor can be more than just low quality. Time-to-market - cited by localizers as commonly being their clients' number one concern - can drag out.

In today's software industry, getting your product out before the competition can mean a huge difference in the amount of sales. As Sunil Sadhwani, managing director of International Translation & Publishing Japan (ITP Japan), summarizes the situation, "If a hardware manual is late, you can still sell the product. But with software, you have no product to sell until [the localization] is done."

The onus isn't solely on the publishers, however. Carl Kay, CEO of Japanese Language Services*, admits that vendors must make more effort to "help clients understand the added benefits [lower technical support costs, for example] as well as the added costs [such as for spending additional time editing the draft translations] of high-quality localization."

Bob Myers, Bowne Global Solutions' vice president, Asia, argues that the cost of a professional localizer is not that great. "If you count localization as a cost per unit, that cost is going down. Although document sizes are increasing and software is getting cheaper, volume is going up."

"Many view localization as buying toilet paper, one-ply versus two-ply," Myers laments. "But it's not. Localization should be treated as part of the developmental cost, and counted that way."

"One-stop shopping" localization
An overall trend in localization, at least on the vendor side, is a move towards providing "one-stop shopping" localization service. Vendors are offering localization into many languages, and a full range of localization services from consulting during development through to a finished, localized product. ITP Japan's Sadhwani reasons that a lot of the recent mergers among localization firms [enumerated in last month's article] are intended to create vertically integrated companies that can take care of the entire localization process, and so leave developers to concentrate on their core competencies.

Kay of Japanese Language Services elaborates. "Our view is that the governing trend will be towards outsourcing of more and more of the functions related to internationalization and localization, to the point of outsourcing complete departments in some cases. Such outsourcing gives publishers the maximum flexibility to reinvent their companies every year or two, as they must to prosper, without the ship getting too big to turn quickly. And the investment in state-of-the-art knowledge and technology required to be a good localizer is borne on the vendor side."

Although client involvement is necessary to a successful localization job, Tomoyuki Kudou, Bowne's localization department general manager, says that a minimum level of client participation in the localization process is possible. The important point, he says, is that the client liaison "knows the application and documentation thoroughly, so that if we ask a question, they have the answer."

How do the publishers feel? "We'd be ecstatic if a vendor could deliver the quality of work we require without a lot of involvement on our part," says Microsoft's Brooks. "But our commitment is to the highest quality in the shortest amount of time possible. We want the design and feel of the product [to be as if it were] built in Japan. Vendors are still a long way from being able to do this in a turnkey way. Maybe it's because we're demanding, or because the job is complex."

"We like the idea of 'one-stop shopping,' but we won't compromise on quality," he continues. Brooks sees the biggest barrier to complete one-stop shopping as the testing required to catch and resolve software and hardware compatibility issues. "Testing requires a substantial capital investment in equipment and technology."

Future sources of demand

In general, all three localizers interviewed for this article are receiving an increasing number of inquiries from Internet software companies. These companies need localization in such areas as electronic commerce, search engines, security, plug-ins, and video-streaming.

"The website localization boom has just started," says Bowne's Myers. "The next generation of websites are a heterogeneous combination of various types of applications. Localization would involve text, cgi, Java apps, redesigning databases, etc. - and making sure they all work together smoothly. Website localization is a perfect example of why, as long as technology keeps cycling, there will always be demand for good, technically-oriented localizers."

Myers also sees the CD-ROM market growing, although not immediately (given the high cost of customization and content development, and the relatively low sales of CD-ROMs in Japan). But he estimates that, "It's only a matter of time before CD-ROM localization takes off. Depending on the media, it might be about five years down the road."

Kay of Japanese Language Services, meanwhile, predicts increasing demand for localization of ERP (enterprise resource planning) software as well as industry-specific, specialized software. "Software turns up in more and more contexts - cars, medical devices, education - where the awareness of localization for Asia is much lower than in the general, commercial software industry." Although translators and editors who are specialists in each particular field need to be brought in, "the engineering work itself is not necessarily vastly different between consumer/general business software and more specialized applications," says Kay.

As an extension to localization, he hypothesizes that localizers may eventually come to handle international technical support as a means of leveraging the detailed and thorough knowledge of a product that they gain through localizing it. "After all," asks Kay, "why should the publisher invest twice to teach two different vendors the same thing? Both localization and international tech support require a high level of understanding of the product, so it would be a natural next stop to ask the localizer to also offer [foreign language] tech support."

"Say a translator has worked on the security portion of a program; he or she would come to know that section extremely well. And if, later on, a tech support question came up in that area, the question could be routed through the Web to that translator, who would be paid based on the number of questions answered. If he or she eventually compiled a FAQ [frequently asked questions] document, that could be published and sold."

Future growth

On a worldwide scale, the trend to globalize as well as localize is gaining momentum. Localization is the adaptation of software for a foreign market. Globalization encompasses the individual localization efforts as well as all the activities needed to do business in foreign markets.

In its 1995 study, Globalisation, Creating New Markets with Translation Technology, London-based IT consulting firm Ovum Ltd. emphasizes the important role Japan plays in the world software market. "Viewed from a purely national (rather than regional) perspective, Japan is and will continue to be the most important single market" for globalization services.

Further, Ovum estimates the 1995 globalization services market at $1.70 billion with software localization holding a 33% ($561 million) share. Ovum forecasts the market to grow to $6.26 billion by 2000, when software will account for 39% ($2.38 billion). In terms of languages, Japan's share of globalization revenues is expected to increase from 18% in 1994 to 31% in 2000.

The increased total localization costs to publishers (for in-house and outsourced localization work) is an additional growth factor. Ovum forecasts that by 2000, software publishers will be spending $925 million to localize into Japanese, a nearly 480% increase from 1994 ($160 million).

The first challenge for the market will be educating software publishers in order to funnel the flow of work to professional localizers so as stimulate the expansion of the professional software localization industry (to increase its capacity and alleviate the growing shortage). The second challenge will be for localizers to find and train more and more staff to satisfy the increasing demand for localization into Japanese.

In short, the demand for software localization services looks to be unflagging into the future - especially for Japan. The localization market bears close watching in the coming years.

-Software physician, heal thyself?- Demand is not likely to be a problem for localizers going into the next century. But what about the work itself, which is very labor-intensive? "Translation and localization companies are highly dependent on people, - says Michael Shannon, sales manager, Asia, of International Translation & Publishing Japan. -It's like a virtual company: a combination of freelancers, partners, internal staff, and the clients themselves [working] long hours together." Translation memory tools This is where technology enters the picture. Recent translation memory tools have greatly speeded up the localization process and helped to make it more efficient. Simply explained, such tools help a localizer ensure consistency of terminology throughout printed manuals, online help, and user interface - which together may run into millions of words. At the start of a new project, the localizer and publisher compile a glossary of terms and frequently occurring phrases, with their equivalents in the target language. The translation memory tool creates a database containing every term or phrase in all the documentation, then substitutes the translated terms from the glossary. The actual translation work, and filling in the gaps, is still done by a human translator and checked by an editor. Another valuable function of a translation memory tool comes into play when localizing a version upgrade. If significant portions of translated material from a previous version can be reused, the translation memory tool compares the databased text of the two versions and tags the reusable parts. It can also estimate the amount of translation that will be required for the newer version. Translation memory tools have been used for about two years for European languages, but they have only recently started showing up for the Japanese market, where localizers have typically had to develop their own tools or do without. Commercially available translation memory tools for Japanese are still at a prototype stage. The TRADOS solution In the first quarter of 1997, TRADOS, a well-known Stuttgart-based translation memory tool maker, released the first versions of its Translator's Workbench and MultiTerm 95 Plus! translation memory tools that support Japanese as a target language. In the first quarter of this year (1998), the release of a complete Japanese translator's tool suite that supports Japanese as a source and target language is planned. TRADOS got a boost when Microsoft acquired a 20% share in the company last September. This may well push TRADOS in the direction of becoming a de facto industry standard for translation memory tools. According to Microsoft senior director David Brooks, Microsoft's investment in TRADOS -reflects the belief in the importance of translation memory as a technology, and TRADOS is the most successful translation memory tool in Europe. Our investment is specifically intended to make sure that TRADOS works very well with Microsoft products." "In our business model," Brooks continues, -we'll benchmark our products with TRADOS, for example, to determine the amount of reusable text [when localizing a version upgrade] and thus set fees.... Vendors may use any translation memory tools they wish, or none at all, provided they return to us the translation memories in a TRADOS-compatible format. We are localizing the TRADOS UI [user interface] into Japanese, and continue to work with TRADOS to ensure DBCS [double-byte character set] and BIDI [bi-directional languages, such as Arabic and Hebrew] are supported, and that the functions work as well in Japanese as in any other language." Tooling into the future Looking into the future, Bowne vice president Bob Myers says, "I believe that, eventually, a combination of machine translation and translation memory will be 80% of all the value we [localizers] provide. In 8 to 10 years, machine translation will be at a point for Asian languages where, if the phrase to be translated is original, it will create a temporary translation; if it's an exact match, the translation is put in; and if it's a fuzzy match, it will use machine translation-type logic to figure out what goes in there." Carl Kay of Japanese Language Services also sees a future in machine translation - but one that will not significantly affect the localization market for at least the next 10 years. But Kay imagines that -eventually, perhaps within 20 years, there will be a giant global database of all human speech and writings, collected through all kinds of devices and media. Rather than analyze based on rules, massively powerful computers will literally search through the entire archive of human expression for a close match in the target language. So [translation] won't be done by elegant algorithms, but rather by brute force of data collection and matching." As for the effect of tools on the market, Kay sees it fueling a trend towards more global vendors. -The tools will make it more efficient to do localization on a bigger scale, and less efficient to do small, piecemeal projects. Similarly, the cost of entry into the localization business will increase due to the investment of time and money needed to master these tools. This, too, will further the trend towards a fewer number of larger, global vendors." ITP Japan's managing director Sunil Sadhwani agrees. -The third-party vendors that can use these technologies will be on top. Growth will be enormous for these companies because, most importantly, it means you can deliver on time. [Doing localization into] Japanese affects delivery time and the number of products you can localize in a year." So, ironically, the localization into Japanese of localization tools could well have a significant impact on Japan's software localization industry.

Back to the table of contents