LISA There are already a handful of translation memory systems (TMs) on tap, some free and some for payment. Where is there room for a new one?

Yves Champollion I started like most other developers from the point of view of a working translator. The basic idea was to simplify the translator's relationships with the agencies they might work for by removing the learning curve and technical barriers to using existing TMs. A typical translator might be expected to know about SGML and HTML tags, fields, hook-ups, file conversion and file versions, all of which represents a mass on non-core knowledge that is almost impossible to keep updated in time with agency needs. Translation SMEs are increasingly being asked to localize websites. So your working translator on the ground wants to be able to seamlessly and safely translate a page of HTML-formatted text without tears. These were more or less my specs for Wordfast.

I have been using the available tools for many years, and as a translator, came to the conclusion that they were improvable. So I began to develop a tool to interface with agency-sourced translation memories a couple of years ago, and brought out what became Wordfast 1.0 in 1999. I then tested it, rendered it operational and brought out the fully marketable version in 2001.

LISA Why make the move from the personal satisfaction of developing a prototype to the competitive jungle of marketing a product?

Yves Champollion After developing a preliminary version, I made the TM available on my AOL personal home page, and used an informal reviewer group of 50 fellow translators to road-test the software. In fact I designed the whole system around the MS Office environment for the very good reason that this is the word and text processing software most frequently used by translators.

Originally, I had no intention of productizing the system and then trying to market it as a standard piece of software. But by the time I had developed version 2 of Wordfast, drawing on suggestions for improvements made by my beat testers, I began to think more seriously about some sort of ROI. So I signed an agreement with a French localization company called Linguex, whereby they could use WordFast 2.0 for 9 months with an unlimited supply of licenses.. This enabled me to fine tune Wordfast yet again in an agency type environment.

However, industry maneuvers intervened before the end of the first Wordfast contract and Linguex was acquired by Lernout & Hauspie and integrated into the Mendez Group. When I saw the L&H debacle, I decided to just wait and see.

Then Logos offered to web-market version 3 of Wordfast from its own site. To ensure support and development continuity, we set up the Yves Champollion-Wordfast joint venture in June 2001 with Logos as the majority partner. Today the system can be downloaded free of charge.

My core competence is in tool development not marketing so the arrangement suits me fine. And making Wordfast available for free is not part of a longer term plan to seed the market and then slap a price on it. It is pretty clear that most working translators do not upgrade immediately to, say the latest version of Trados tools each time there is an upgrade, and they won't necessarily upgrade to a new version of Wordfast either, even though it is free of charge. So I need to provide ongoing support for all versions.

LISA Continual development and support costs money. What kind of business model will you apply to keep Wordfast in the black?

Yves Champollion Now that I've become a full-time tool developer, I shall certainly have to think about the bottom line. For the time being, we have nearly 1500 registered users and 20% of them participate in a highly active mailing list, all of which provides useful input for further development.

There are, however, several options available for both maintaining a free download while at the same time generating modest revenues from a paying product. I'm currently working on Wordfast version 4 which unlike the previous standalone versions will be a purely web-enabled translation tool.

In this new configuration, the translation memories stay in the translation agency rather than being downloaded to the local translator, and the contents of the memories can be shared across a group of translators working on a given project. I may ask for a modest contribution for making this sort of functionality available to agencies.

Another move would be to develop dedicated versions of Wordfast for specific client companies. For example, a given airline might have all its legacy documents available exclusively in SGML, and due to costs and other internal reasons, does not wish to convert their files to, say HTML. In this case, we would develop a translation memory filter geared specifically to this format, so that local translators could work on the content without having to export the whole lot from SGML into some other format.

A further possibility would be to develop a toolbox to complement the basic Wordfast environment. It might contain special filters for exotic fonts or extra tagging utilities.

LISA Now that the engine is up and running just one nerdy question: What aspects of Wordfast give you most satisfaction as a developer?

Yves Champollion Writing a translation memory engine which operates fast on huge TM files entirely in VBA, the MS Word programming language, without the whole thing slowing down. I could obviously have used another programming language, but this would have spoilt the seamless efficiency of the whole system.

For example, WordFast also works in the Macintosh environment—the one and same version that works with Windows. Not many professional translators in Europe use Macs, but they are plenty in the USA. Since all of Wordfast's macros are pure Word macros, they will operate in the Mac environment.

And we are also TMX compliant, following useful discussions with Yves Savourel, the OSCAR technical chair. Wordfast's native format is in fact plaintext, but we thought it essential to offer users the TMX option, which has the advantage of Unicode compatibility.

Further information

For more on Wordfast, go to http://www.logos.it/Wordfast.

The Wordfast mailing list can be found at http://www.yahoogroups.com/Wordfast.

A collateral descendant of the French-man who deciphered the Egyptian hieroglyphics of the Rosetta Stone, Yves Champollion (yves@champollion.net), the Chairman of Y.Champollion-Word-fast Ltd, is a much-travelled polyglot programmer and sometimes translator based in Paris. One of his more recent languages is Shangana, a Zulu-related language of Southern Mozambique, where he sponsors a secondary school.