What a way to make a livin'

At work I regularly receive graphics files for translation. These are product communication materials (POS, catalogues etc). It's great being able to work directly on the native format (rather than send the translation to the execution agency in a Word document), because it saves a huge amount of time on two fronts: I no longer have to convert PDF files to text, the graphic execution agency no longer has to copy that text into the native file. We reckon it saves about two days. And it practically eradicates errors.

But it's not all roses. This week I received 80 product signs. They happened to be for bikes. The file format is indd (InDesign). Now we work with a translation memory package (essential tool for any translator). It supports InDesign files, but not the standard .indd format. I have to export the text content in a sort of xml file (.inx format). Then I can translate the text, save the translation as an indd file, and the only DTP required is to update image links and check the fonts. Great isn't it?

Only exporting the .inx file can't (to my knowledge) be done as a batch, it has to be done file by file. So I took my bikes signs, double-clicked each file to open them, clicked OK on the message in InDesign that tells me I don't have the latest modules, clicked again on OK at the message that tells me I don't have all the fonts in the file installed on my system, and clicked Don't repair at the message that asks me if I want to update links to missing graphics files. Then I hit Ctrl E to open the Export dialog. The default file type is pdf, so I opened the drop-down list and chose InDesign Interchange. Then I clicked Save. And I did this 80 times. I don't know how many clicks that adds up to. But it's a LOT.

Once I'd finished exporting it was time to create the translation projects. The translation memory software operates with projects. You create a project, import your source files (basically, extract the text), translate in a specific interface, then export the files (a process which recreates the original file with all attendant formatting etc.). So I created my projects. For 80 files. And sometimes, if the source file size is quite big, it doesn't import properly on the first go. So I reimported quite a few times.

So. Now I have my translation projects and I can get cracking. Except that I can't just yet. Translation memories work on the basis of "segments". The translations they save are not single words, but segment entries. A segment is usually a sentence, but most memories will also segment when they detect a line break in a text. Makes sense. Except that in my graphics files some text boxes are on quite narrow columns which means one sentence can have four line breaks in it. The translation memory duly segments the sentence into 4 lines. Only, in the memory database, the translation is saved as one line, not four! Fortunately, I can join together segments before translation, to stick together the sentence and thus obtain a match from the memory. So before running a pretranslate on my files, I joined up the segmented catchlines on 80 product signs.

Can you possibly imagine a collection of tasks with any less added-value than that?

Finally, after probably the best part of an hour mindlessly clicking, I was ready to pretranslate. And that made it all worth while. Because out of just short of 25,000 words, only 3,200 or so remained to translate. So an hour of brain-dulling, wrist-numbing and finger-fatiguing click-clicking is a good trade-off really. Because if I had to translate all that without the memory, it would easily take four days. Now it will only take four or five hours.

That's why the translation memory is such an excellent thing, especially for product communication. Because, with bikes for example, we don't bring out a whole new range every season. So POS and other materials don't radically change every 6 months. There might be a new size added, or a limited edition colour. But the bike still has an SRAM X7 drivetrain and a RockShox fork with remote lockout. And the marvellous translation memory has it all stored in its database.

Sorry for this lengthy and boring insight into my daily routine! Just thought I'd share the highs and lows as it were. But I did exaggerate somewhat about the number of bikes files. There were only 79 really.