interview
Japan's Expanding Localization Market
Bob Myers is the founder and president of Pacifitech Corporation, a Yokohama-based firm specializing in Japanese software localization. Senior Editor Steven Myers (no relation) talks with him about the evolution of Japan's software localization market.interviewed by Steven Myers
How did you get started in software localization?
Bob Myers: I started this company about nine years ago; initially, it consisted of just me. I was doing both some translation work and some software-related work. There's a lot of history, but basically I found some partners, and for a while we were a hybrid company that did both translation and software development work.
Then one day -- maybe the beginning of 1993 -- we woke up and realized that the skills we had developed from our two separate businesses (translation and software development) were in fact the precise set of skills that are required for localization. At that point, we shut down our other projects and said, "We are localizers." That's what we've since dedicated ourselves to.
How large is Pacifitech now?
Myers: We currently employ about 35 people.
Do you hire dedicated engineers and translators, or do you prefer people who can do a little of both?
Myers: Well, because we do engineering work -- two-byte enablement work and adding features -- we obviously have engineers; but that's a relatively small percentage of our workers, probably less than 10%. Most of the software companies that have their act together are basically doing all development back at their headquarters in the US. That includes two-byte work (which, with the well-run companies, will be in the baseline from day one) and also the job of adding new Japanese features.
Some companies attempt to take the localization process and slice it up into little tiny bits: translators, editors, help builders.... This doesn't work very well. The parts have too many interactions and dependencies, and it's still very far away from an assembly-line type situation.
What we look for, essentially, is a new type of professional that we call a "localization engineer," or "localization specialist." Our goal is to find people who can do as much of the process as possible. Basically three fundamental skills are required. First, these people don't necessarily have to be engineers or programmers, but they should be power users -- able to write macros in MS Word or to pick up tools easily and use them effectively. The second thing, obviously, is that linguistic skills are required; the ideal would be something close to bilinguality. The third skill -- which some people say you're either born with or you're not -- is project management ability. In some ways, this is the most important, since the basic unit of work in localization is the project.
Few such people, though, exist in the outside world. So we hire some people from translation backgrounds, and some from systems orientations, and try to turn them into "localization engineers."
What developments do you see occurring in the Japanese localization market at present?
Myers: Over the past year or so, there have been some obvious changes in localization. One is the emergence of dedicated localization companies. In the past, some companies localized their products in-house, some did it through translation companies, some managers gave it to their friends, some did it in the States -- it was a fairly ad hoc process.
What's happening lately in the industry is that there's a group of companies emerging quite clearly as being professional localizers. Most of these companies are in the $5- to $20-million range, and they're all growing very rapidly (as much as 100% a year).
What are the reasons for this change?
Myers: There are a couple of reasons. One is that typical software companies now are on something approaching an annual product cycle -- putting out all the versions of their software in the summer, for example -- and so this means that the requirements for localization resources have huge peaks and valleys. One of the roles of this new localization industry is to provide a load-leveling mechanism for software companies, so we can do Microsoft's huge product release in the summer, and then in the fall we can do someone else's huge product release.
Another point is that, if you look at the individual parts of localization, there are always going to be some parts that are easier and more accessible, and other things that are harder and more technical. Recently, more and more quite specific technical knowledge is required for localization.
The obvious example is that five years ago, if you talked about localizing documentation, basically it was finding a word processor, typing a bunch of Japanese text, and then taking it to the printer. And three years ago, it became building and testing Windows help systems and other standard online help systems. The vast majority of documentation will soon be provided in the form of some sort of browsing mechanism, or full-text search (more hypertext functionality), and the process of building and translating these help systems is getting more and more complicated. You need more professional technical knowledge.
Another thing that's happening -- and it's also kind of driving the emergence of these localization companies -- is that in the past 2 or 3 years, there has been a major change in the importance of the Japanese market to software companies. Typical companies that we work for might be getting 20% to 30% of their revenue from Japan. So we're talking about a situation now where you can't take chances by trusting localization to non-professionals.
What other trends have you been observing?
Myers: Well, I'd say another trend is that, whereas localization used to be considered on almost a project-by-project basis, now the larger software companies are very conscious of the value of having a relationship with a strong localizer. And strong has to not only mean "does good work on time," but it also has to mean "provides a certain level of capacity to do work." One of the largest foreign software companies in Japan now has annual localization costs well into seven digits. There is a lot of localization going on.
The main problem right now is finding people to do that; they need capacity. They can't get into a situation where they go to someone and say, "Please start working on this," and the company answers, "We're busy right now; we can't start for another three months."
Most of the large software companies are focusing on developing strong, long-term relationships with the localization companies. Typically it includes things that normally one would not imagine might be involved in a localization relationship. For instance, it would involve some level of joint long-range planning, looking at next year's project schedule and fitting products into some sort of time frame.
What about input regarding the content?
Myers: Yes, another trend that I've been noticing lately -- though I think it's only partially starting -- is toward more thorough-going localizations. In particular, localizing the content, even adding content, and localizing and adding the features. That goes not only for the software but also for the manual. In other words, potentially rewriting manuals, at least partially, for the Japanese market. Until now, most localizations have been fairly bare-bones localizations: just translate the user interface, translate the manual, maybe add some minor features like Heisei dates... stuff like that.
The reason is not that Americans are lazy, or stupid, or anything; it's simply the economics of the situation. In the past, software companies just weren't selling enough copies here to make the added expense of a thorough localization worth their while. In a way, the Americans have taken advantage of the Japanese situation by putting all these mechanically localized products on the market. Really, with all these conceptually American applications and documentation, it's amazing that the Japanese have adopted them as readily as they have.
I think in the future, you're going to see a generation of "kinder, gentler" localized applications, those with more of a Japanese spirit. Localizers will be adding to their bag of tricks more cultural sensitivity.
Do you find much demand yet from foreign companies for localizing content on their WWW pages?
Myers: This is something that a lot of companies haven't realized yet: that they're going to have to start localizing their Web content. Right now, I'd say that even limiting it to just software companies that have stuff on the Web -- recent bug reports, patch info, tech notes, what have you -- on the average, much less than 1% of this is localized into any language other than English.
We're designing a system now that will let you create well-integrated multilingual Web presentations. It's based on a language we developed that allows the user to specify how stuff is to be localized. You can say, for example, "All the files in this directory should be localized into German, Spanish, and Japanese." The system notices when pages get out of date and automatically displays an "out-of-date" message. If somebody goes in and edits the English page, the next guy that hits the Japanese page will see a message saying "the English page contains newer information."
Anyway, we certainly believe that the Web is going to be a massive source of localization work.
What are your views about royalty vs. flat-fee payment for localization projects?
Myers: We fundamentally believe that royalties are a good thing, and we believe that they'll become a relatively standard aspect of some localization projects. They have the obvious advantage that the client has a way of kind of hedging his bets and not having to pay enormous localization fees without knowing how well his product is going to sell. At the same time, for the localizers it obviously represents a very attractive upside.
Essentially, the idea of royalties is a combination of two things from a localizer's standpoint: a combination of a bet and also a loan. Let's say the project really is going to cost $50,000, but I'm going to do the project for $25,000 and try to make the rest of it back -- and hopefully a lot more -- through royalties. Essentially, what I've done is loaned $25,000 to the customer, because now I have those costs without any idea of whether or not that money is ever going to come back. The thing is, if you're really sure that the money is going to come back -- maybe because this is the fourth version of the product and you've looked at the sales figures for the last four years, so you can guess very precisely what the income levels are going to be... well, obviously the client also has that information. So why on earth would the client sign a royalty deal with you that's going to pay more money than he would have had to pay with just a flat fee?
Yes, why would they?
Myers: Actually there's a very good reason: the idea of tying the royalty stream to providing ongoing technical support or minor version work. This is probably the most persuasive reason for adopting the royalty model. Of course, royalties are also an excellent motivator for the localizers to do their best work.
Pacifitech has done several royalty-based deals; in fact, we've done some quite large ones. What I've found is that they work very well for both vendor and client. For the client, it can be quite easy to pay a small royalty per unit sold, because they can think of it as a kind of manufacturing cost, like making the box. For small projects or small software companies, though, you're actually talking about a case where the product might literally sell zero units; so you'd think long and hard before doing a royalty arrangement.
What do you see happening in the future in the area of Japanese localization?
Myers: If you're going to talk about the future, then you're going to have to talk about machine translation (MT). In ten years, major parts of the localization process will be done by MT. You may see people start to use it when the next generation of MT systems comes out, in the first part of the next century.
But it amazes me no end to hear people talking about MT today, because they basically know nothing about it. They're pretty much at the level where they read an article in the newspaper, which is basically a PR blurb put there by some company. We have competitors that make it one of their sales points to say that they use machine translation. But we're looking at them, seeing what they can and can't do ...and really, they're total garbage, completely unusable. Anybody who says that they can produce the type of manuals required for one of today's products by MT is either a liar or has no idea of what translation is about.
Another thing I think we'll see in the future is much more sophisticated systems of reuse. In a few years, the original English manual will need to be maintained using not Word or PageMaker, but some new document management scheme that facilitates reuse and localization. The difference between doing something meaningful for resuse and doing everything from scratch is huge. For user interface stuff -- dialog boxes, for example -- its relatively easy to do reuse with standard methods. Resources have IDs that can be matched up, and so forth.
The real problem is the manuals and online documentation, because there's just so much stuff, and it's getting massively bigger all the time. The programmers and engineers in the US have actually gotten quite good in terms of thinking about localization. The writers, on the other hand, apparently never think about localization practices. They just go through the manual making changes, including changes that aren't necessary -- stylistic changes, or reorganizing books and moving chapters around.
The fundamental solution to this problem -- besides saying that the writers in the US should have guidelines about changing and editing -- is that you use some sort of technology. At the moment this technology only exists in bits and pieces, but I think we'll see it develop over the next two to five years.
So, it sounds as though you expect demand for software localization to keep growing.
Myers: Yes, and I think in the future we will see increasingly larger companies that are dedicated solely to localization. If you look at the way all the pieces add up -- the desktop market for localization, the "multimedia" content-oriented product demand for localization, Web-based content that needs to be localized, not to mention the in-house corporate projects that will need to be localized -- well, let me put it this way: I don't think there'll ever be a $500 million dollar company specializing in localization, but I don't have that much trouble imagining a $100 million dollar company specializing in localization. The demand for competent localization services is huge, and will continue to grow.
Contact Information
Pacifictech Corporation
Phone: 045-476-0521
Fax: 045-476-0525
E-mail: rtm@pac.co.jp
