A sample text widget

Etiam pulvinar consectetur dolor sed malesuada. Ut convallis euismod dolor nec pretium. Nunc ut tristique massa.

Nam sodales mi vitae dolor ullamcorper et vulputate enim accumsan. Morbi orci magna, tincidunt vitae molestie nec, molestie at mi. Nulla nulla lorem, suscipit in posuere in, interdum non magna.

Context matches in Déjà Vu X2 Professional

Context matches (called “Guaranteed matches” in Déjà Vu) have been supported since Mid 2003, with the introduction of Déjà Vu X, but only in the Workgroup edition, which used to cost about €2500. The feature has been made part of the cheaper (current list price: €595) Professional edition since the introduction of Déjà Vu X2 in 2010. Like most other tools (eg. Heartsome, memoQ) except SDL Trados Studio (in which same context means same preceding segment, but not necessarily same following segment), Déjà Vu defines context as identical preceding and following segments. Déjà Vu goes even further: it is the only tool in which the size of the context (number of identical preceding and following segments) can be defined. A context size of 3 means identical preceding segment, actual segment to be translated and identical following segment, while a size of 5 means two identical preceding segments, actual segment to be translated and two identical following segments. Guaranteed matches can be used via the pre-translate function, and they can also be locked.

Introduction video:
Context matches in six CAT tools

Review of context matches in other tools:
Heartsome Translation Studio Ultimate edition
CafeTran Espresso 2013
memoQ translator pro
Swordfish III
SDL Trados Studio

Other relevant post:
101% matches explained, with memoQ and Olifant

3 comments to Context matches in Déjà Vu X2 Professional

  • Interesting video Dominique although I think I disagree with you over the “stricter” requirements for a Context Match than Studio. As far as I am aware DVX says a “Guaranteed” match is an “Exact match” where the surrounding segments have been checked as you stated. An exact match being one where there the “source segment” is the same. So my reading of this is that DVX does not compare the target.
    Studio compares the source and target for a “Context match” which I think is not only stricter but more sensible because if there is a difference in the translation due to gender differences for example then you will only see this when you compare the target too.
    So I think a Context match in Studio would be more reliable irrespective of the number of segments either side of the source you check. What do you think?
    Paul

  • Dominique

    Hi Paul, I’m not sure I follow you here: how can Studio look at target segments when evaluating context at pre-translation time? By definition, the translation of new text is not known at that time, since it’s what we are trying to do using the existing TM (and the context information it contains).
    One situation where the “stricter” context can be useful is with the first segment of a document. In my example, “Using our services” would be considered a context match by Studio even if the next segment is different. I can think of many situations where the first segment (typically, a title) wouldn’t always be translated in the same way. For instance, “Table of content” could be “Table des matières” or “Sommaire” in French. If the second (source) segment is identical, that would provide me with more certainty that the context is the same.
    I think in the particular examples I used in the video, Studio’s Perfect Matches would be more appropriate than context matches. Unfortunately (for us freelancers), it’s only available in the Professional edition.
    Dominique

  • Hi Dominique, of course it can’t for new content. But it can during interactive translation and it can also differentiate between 100% and Context Match for same source/different target with things that have been translated before. So if you have four segments like this for example:

    Mathematics can be tricky.
    Go and ask the teacher.
    Asking the teacher is sometimes the best way to resolve the problem.
    It can save you lots of time in the long run.

    Now translate this into German when you know the teacher is a male and you get one context. Do it again when the teacher is female and you have another as “lehrer” becomes “lehrerin”. Studio will differentiate between these whereas DVX will not. So I guess you would see “guaranteed” matches that are incorrect (based on my understanding so far).

    I guess it’s easy to find examples of where any of the tools fall down, but in general the point I was making is that Studio takes into consideration the target context as well so it’s not really apples with apples and I therefore wouldn’t say DVX is stricter. Perhaps it is in one way, but not in another… apples and oranges really (I like my fruit)

    I also think you are starting to mix comparisons too. “Guaranteed” matches looks to me a little as though they’re straying into the Perfect Match / x-translate area so you have switched something on which may not be a true comparison and then minimised the range to reflect something that looks like a Context Match or 101%.

    Paul

Leave a Reply

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>