Translation Issues 3-99 (September 1999) by Wolfgang Abele

I work as a freelance translator, mostly with Word and html files. I also regularly organize localization projects involving translations into the main European languages. When I looked around at the translation memory systems on the market today, I found them not only seriously overpriced but also laden down with so many features that I'd never expect to use in a month of Sundays.

OK. So it's nice having Unicode support (just in case I eventually get round to learning Japanese). And yes, it's neat to be able to link your glossaries to pictures and sounds, but I'm not much interested in a multimedia database, either. Text will do just fine, thank you.

I also got the impression that these systems were just a tad lacking in user-friendliness. For instance, in one program, I tried several times to export a TM database. I just couldn't figure it out at all. Just when I was putting it down to my own incompetence, I asked two users who had just completed a two-day training course. Know what? They couldn't do it either!

This kind of complexity - in a tool whose goal in life is to save me time - strikes me as somewhat counterproductive.

On the other hand, none of the main vendors have implemented what I would consider smart features, namely any sort of syntactic or morphological analysis. While they have fixed bugs, added filters for DTP programs, introduced Unicode etc., the basic approach has hardly changed at all in the last five years. The developers have put a great deal of engineering effort into handling the intricacies of different file formats, but from an algorithmic or linguistic point of view, I don't see much improvement.

So I decided to write my own translation memory system, which I call King Memo, partly as a programming exercise and partly because I reckoned that a simple and inexpensive program might well be useful to other translators with needs similar to mine. Since I made the first beta available for free download, I have been encouraging other users to get involved in the development process. Although I'm always grateful for bug reports and suggestions for improvement, the wish-list does seem to have grown considerably, particularly in the last couple of weeks.

King Memo is a translation memory system for Windows and is designed for working with Word 97/2000 and html files. It consists of two parts: the main program (also called King Memo), and a collection of Word macros called Segment. King Memo allows you to create a glossary (= translation memory) from previously translated Word and html files and to perform a fuzzy search in that glossary. Segment can be used to create a glossary during the actual translation in Word. Some of the comments I have received have been useful in tracking down some of the initial bugs. For that reason, the current Beta 1.1.4 now seems to be pretty reliable, whereas Segment is still an alpha version.

The main benefits of King Memo are:

• Because it uses only csv text files, it does not produce yet another special file format. This means that glossaries can easily be exchanged and read by other translators using different programs. Incidentally, the widely used Microsoft glossaries also come as csv text files.
• It's fast and efficient. Although all users have so far only experimented with small glossaries, King Memo is quite capable of handling very large glossaries. A fuzzy search in MS's WIN 98 glossary (5 MB+ and 43,000 entries) takes less than half a second on a Pentium 166. And yet it requires only 2 MB of RAM in search mode. So, as you use your wordprocessor, it runs inconspicuously in the background.
• It's easy to use. King Memo is compact, with less than 1000 lines of code and no bells and whistles. I maintain that, after reading the brief instructions, even a CAT newbie can get by. If he or she has any questions, I'm happy to answer them via email.

Future development of King Memo is still wide open. I'm currently working on improving and extending the html-related functions and on developing the documentation. In addition, the Saarbrücken-based Institute for Applied Information Science (IAI) has just become involved in the project and we are going to be carrying out some computer-linguistic experiments based on King Memo and a UNIX program. Version 1.0 of King Memo is due for release in October/November 99. I have also considered putting the source code of Segment under the GNU license. Everyone interested in VBA, who would like to play around with the macros, is welcome to get in touch with me.

May I close with a request? I'm still searching for the ideal icon for King Memo. What I have in mind is something like a little fat barefooted man with a crown on his head. So far, my little King has proved quite elusive. If anyone can provide me with a suitable picture, I would be most grateful.

A beta version of King Memo can be downloaded from
http://home.t-online.de/home/Wolfgang.Abele/ [broken link as of 06 Nov 2001 - AvO].  

"transline Deutschland - Übersetzungsdienst für technische Übersetzung"