INSIDER: This is LISA’s second survey about Translation Memory usage. How does this report build on the previous survey?

I think the greatest strength of the two surveys in combination is that we can start to see trends in the industry, trends that we couldn’t quantify otherwise. For example, are people still entering the TM field? Is it a field that can still grow? The results of this survey indicate that the market is growing and maturing, but that there still are plenty of people entering the field, so there is plenty of upside in this market. It also really demonstrates the cumulative nature of TM assets: they keep growing over time. The previous survey had many fewer respondents, so it was hard to know how solid some findings were. The fact that the results of this survey correspond well with the last survey indicates that our analysis then was on the right track.

One new area we added in response to reader requests was an examination of what specific TM tools people are using. Even though I consider myself fairly knowledgeable about the localization industry and the tools that are available, I was really surprised at the variety of tools used, including some that are new or niche tools.

INSIDER: What are the biggest surprises in the survey?

For me, the biggest surprise is one that isn’t new. It’s that there is absolutely no correlation between translation volume and TM usage. Since we view TM usage as the enabler for high-volume localization, I am at a loss to explain how some of the companies with the greatest demand for localization are simply not using TM to any real extent. Part of it may be that these companies are outsourcing their work to partners who use TM and thus obscuring the fact that TM is used for their content, but I don’t know that corporations with large authoring projects would be unaware of TM usage by their service providers any more than those with smaller authoring projects. If anything, I would expect the converse to be true, but instead we find that TM usage peaks among organizations translating 5–10 million words per year. It is higher among those translating less than one million words than among those translating over 50 million words. Since we found the same result in 2002, I think that this is a more or less permanent characteristic of TM usage.

I think that this result may instead point to a problem that Andrew Draheim addresses in his article in this issue: the human factor. If you are working in a small company with, say, ten authors, and one localization service provider, you can fairly easily get everyone to work with the same tools and processes. When you move to a company with hundreds of authors and twenty service providers distributed across four continents with autonomous local units, it is much harder to get everyone to work together and to implement TM.

I also think that this points to a need for tools to reflect the needs of enterprise users. Our industry really grew out of tools designed to help individual translators with their work. These tools are really good for that, but they may not be as effective at addressing the needs of companies. Going from the individual or small business to a multinational corporation requires something other than just beefing up database access times and speed. It requires support for processes and technologies that belong to a different paradigm: content management systems, web portals, etc. Fortunately, recent developments from many tools developers indicate to me that they are aware of this problem and aggressively trying to meet enterprise demands.

On the other hand, it is heartening to see that there is a strong tendency for users to increase usage levels over time. That means that they are seeing real value in TM tools and processes. Users are also using TM for more than short-term cost savings. Some of them tell us that they don’t actually realize short-term savings, but that TM enables them to implement strategies and processes that would otherwise not be possible. That means that TM is not simply a tactical tool, but has become a tool with strategic importance, at least for some companies. It also means that TM tools developers have room to sell their products to higher levels of management within the organization. I’m not saying that’s an easy task, but maybe this report will help both tools developers and users sell the importance of TM technology to upper management.

INSIDER: Have tools improved in recent years? How is this reflected in your results?

The answer is that they have improved in many respects, but that not all users are taking advantage of these improvements. A few years back though, lack of language support was often cited as a problem, but now this concern scarcely shows up in the survey results. Whatever issues may exist with TM, they are not primarily technical in nature.

Tools have also become more scaleable, even if they can still be improved to provide better enterprise support in some cases. We had people report using TM for up to 100,000,000 segments in an average of eleven languages. 100,000,000 segments is an astounding figure that would not have been possible just a few years ago, at least not in any unified way. While the bulk of users are not anywhere near that point (a typical user might have around 300,000 segments), the fact that such volumes have been attained tells us that there is no fundamental limitation to TM that would keep it from working. Just a few years ago, we didn’t really know whether or not there might be some fundamental limit beyond which TM would actually become counter-productive.

INSIDER: What tools are most popular?

At present about 70% of the market is using TRADOS, but SDLX, Déjà Vu and Catalyst are all used by at least 20% of the market (these numbers add up to over 100% because companies use multiple tools), so the market is really quite competitive, and a significant number of users, particularly the largest users, are using in-house tools. Note that most TM users are not limiting themselves to a single tool. This is partially because they need to work with partners that use different tools, but also because there are many specialty tools that do certain tasks very well. Thus, it is not really a question of TRADOS or Catalyst, but rather of how many tools a company needs to meet its demands. And then there is Translation Memory eXchange (TMX), which about a third of users report as a format in which they have TM data. Having been personally involved with OSCAR and TMX for a number of years, it is heartening to see that TMX is being adopted and used in the industry. We clearly have work to do, but TMX is having an impact.

INSIDER: What are your predictions for the TM market and the language industry over the next few years?

I see continued growth, both in terms of volume, and in terms of skill and maturity. I also see an increasingly diverse and competitive market, with developers increasingly trying to appeal to and capture the enterprise market. While the bulk of TM users will still be smaller users, enterprise users represent an opportunity for real market differentiation. This means we can expect to see more distributed TM systems (i.e., TM servers), better support for content management systems, and better support for XML. Although all TM systems I know of support XML to some extent, there is room for improvement and sophistication. I don’t really expect to see any fundamental leaps in technology (although I would love to be proven wrong), but I do expect to see improved integration with other processes, and real improvements in flexibility, particularly for the largest users. TM systems will get better and faster.

If readers really want the details on where all of this is headed, I encourage them to download the report, which is available at no cost. LISA is also running a new survey on the Middle East localization market in its Industry Survey Series. We encourage all of our readers to participate by clicking here.