15 October 2009

AutoSuggest: translating in a higher gear

SDL has just released Service Pack 1 for SDL Trados Studio 2009. The update will be welcomed with a sigh of relief by Studio's early adopters, who suffered from some stability issues. I've been testing the service pack's beta, and can only confirm that these issues have now been fixed.

After playing (and, er, working) with Studio for a few months, I can barely conceal my enthusiasm for the innovative AutoSuggest technology. Hmmm, "AutoSuggest" - sounds more like some esotheric method to quit smoking, doesn't it? Anyway, this feature turns out to be a tremendous productivity enhancer for the translator. In fact, this may be the single most useful CAT-feature since the invention of fuzzy matching.

So how does this work? Well, the simplicity is almost too good to be true. While you translate in Studio's editor, tooltip-style suggestions will appear as soon as you type the first character(s) of a word. If the suggested term or phrase is appropriate, you simply press the Enter key to insert it. If you don't like the suggestion, you ignore it and keep typing. No buttons to click, no Alt+Ctrl+whatever. Just Enter.

The best part about AutoSuggest is its intelligence: you won't get hundreds of suggestions, but only those that make sense within the context of the current sentence. This distinguishes Studio's technology from the predictive "dictionary" option for texting on mobile phones. Just have a look at the examples below.

Screenshot of first segment with AutoSuggest in SDL Trados Studio 2009
Figure 1: In this sentence, as soon as I type "r", Trados Studio suggests réchauffement planétaire as a French translation for the source term "global warming".

Screenshot of second segment with AutoSuggest in SDL Trados Studio 2009
Figure 2: In the next sentence, when I type "r" once more, a different suggestion appears. Because of the new source-sentence context, Trados Studio now suggests réduction des émissions (emission reduction).

The secret behind this intelligence is my so-called "AutoSuggest dictionary". Studio allows me to generate such a dictionary from any translation memory or TMX file, provided that it contains at least 25,000 translation units (like the EU's Acquis Communautaire in TMX format). This approach ensures that AutoSuggest will only propose translations that are relevant for my subject field. Additionally, termbase terms and AutoText entries can be inserted in the same way (type the first characters and press Enter).

AutoSuggest is simply the cleverest implementation of sub-segment matching I've seen so far. The translator stays in full control, but benefits from a true time saver and quality enhancer. Want to try this yourself? You can find freely downloadable AutoSuggest dictionaries on TranslationZone and Xenotext.

Bookmark and Share