Although translation memory (TM) tools have been around for quite some time, there are always new features being developed. Some of them seem like minor improvements to existing functionalities, whereas others open up new possibilities. With examples from two of the most popular TM systems on the market (memoQ 2015 and SDL Trados Studio 2015), I would like to draw attention to some of the new or enhanced functionalities.
I focus primarily on features that both tools have to a greater or lesser extent, as I would like to show where our translation tools have been going over the last two years. Each tool has individual and wonderful functionalities, not all of which I mention here, but the point is to look at where the tools are going rather than giving all the details on individual features.
Let us start with the look and feel of the tools. We see a strong influence from the web with their simplified and clean icons as well as from Microsoft’s move toward ribbons instead of pull-down menus. Personally, I still think the pull-down menus are more structured and it is easier to find what one is looking for, but that might just be me. I am also not so much in favor of icons being faint in color and looking very similar — this makes them harder to find.
More and more, the user can customize the look and feel of the tool. Maybe developers should think about letting the user customize the icons as well; this is just a thought and I know that the developers will hate me for that — or maybe at some point in the future, we will get some nicely structured pull-down menus back? Or at least a chance to customize the user interface in such a way that one can define one’s own menus? OK, I’ll stop here; these things are minor compared to functionalities that improve quality, file format support, speed and so on.
The ribbon menus try to bring together the most-used functionalities or functionalities that are used in a certain situation (like reviewing a document). Icons are complemented by text, pop-up information and shortcuts. And here I would like to ask the developers to not change the icons with every new version of the tool. It is hard enough to remember all the icons and colors of the features one wishes to use. Just because the icons look cooler with less color or a different image does not mean they are more usable.
When looking at the ribbons some might say that they contain everything they need and are easy to use; others will complain that the one or two functions they always use are not on the ribbon or are to be found on a different ribbon than they would expect. Well, if you can’t make everybody happy, let them customize the ribbons themselves (Figures 1 and 2). Which is what the tools allow now — heaven for those of us who like to play around with these things, hell for anybody who needs to support users of the tools by telling them where to click.
File types and filters
New features are added regularly to the supported file types in both tools. Here we can see a move toward more interoperability, as SDL Trados Studio has added a default filter for memoQ XLIFF files. memoQ has been focused on data interchange from the start and offers import filters for bilingual files from other tools (SDLXLIFF, Wordfast TXML) as well as packages (SDL Trados Studio, STAR Transit, TIPP and WorldServer). It is also becoming easier and easier to export data from a tool to process it outside of the tool with another application and then import the results. With XLIFF-based formats and Word tables, both tools offer a good way to exchange translation data.
With evermore complex file formats or the possibility to embed text of one file format into another (like HTML within XML or Excel), the filters of the tools also needed to become more complex and combinable. memoQ uses cascading filters, where one can use an XML filter for the file itself, add an HTML filter for content formatted in HTML-style, and then add a third filter to markup placeholders as tags with the Regex Tagger filter. These filter combinations can be used for almost any combination of file formats in memoQ, whereas SDL Trados Studio at the moment restricts the use of embedded text to the most commonly used file formats such as XML and Excel.
One of the most misused file formats is definitely Excel — and it is always amazing what kind of content one can have in an Excel document. Often, the files need to be bilingual or multilingual after translation. memoQ has a very clever way for project managers to deal with multilingual Excel files. The file is imported into a multilingual project (for that a project management license is required; translator licenses don’t allow more than one target language per project). One defines which column should hold which target language and memoQ outputs several separate files for translation. In the end, memoQ combines the columns into one single file again. SDL Trados Studio offers something similar, but at the moment it looks like the filter works for bilingual Excel files only, not multilingual ones. As with memoQ, information contained in other columns can be added as a comment or context information about the segments.
PDF is another file format that unfortunately is sent to translation with increasing frequency (or, at least, is used for analysis of word count statistics).
The translation tools are adding more and more capabilities to make it possible to work with these file formats, although we all know that this should not be the default setting for our projects.
Both memoQ and SDL Trados Studio offer conversions of PDF files that were created out of regular text documents. Studio has now added optical character recognition capability to be able to also convert documents that were scanned or consist only of images. This is a nice development, but it still does not help if the PDF file is copy- or password-protected and produces only garbled segments when one tries to convert it. So try to get the original file whenever you can; PDF is just not the way to go.
Together with the discussion on quality (what it is and how it can be checked and measured), the tools are adding more functionalities to their quality assurance (QA) and evaluation features.
Kilgray calls it linguistic quality assurance (LQA) and implemented it in memoQ 2013; SDL calls it translation quality assessment (TQA) and implemented it in SDL Trados Studio 2015. Both use a setup where one can define categories (plus subcategories) of errors, such as terminology errors, punctuation errors or number errors, and then assign severities like critical, major or minor, all with penalty points. This kind of QA step is a manual step, although memoQ offers a possibility to map LQA error categories automatically to the regular QA messages that the QA checker produces, provided that a suitable category such as the use of a forbidden term is available. In other cases, where a translation is not correct because of a misunderstanding on the part of the translator, the error category has to be set manually.
In addition to the customized settings, memoQ also offers a list of pre-set LQA models based on the LISA QA model, TAUS and J2450 translation quality metrics.
In the editor, memoQ uses a red highlight to mark segments with LQA errors. For each segment, the review pane shows the list of messages, which messages have been set manually by the reviewer, and which have been converted to LQA errors (Figure 3).
In previous versions, one had to attach the LQA model to the document during import of the document. In the latest version of 2015, the LQA model can also be attached later, after the documents have been imported into the project.
In SDL Trados Studio, the TQA module can be edited either in the options or the project settings. At this point there are no predefined models; one has to set up a model from scratch.
After some feedback from Richard Sikes, who did a review on SDL Trados Studio for this issue of Multilingual and who had seen predefined TQA models in an online session with SDL, I went to search for these models. Unfortunately they are not mentioned in the online help under TQA, TAUS, LISA or similar search words. Here is where I finally found them: In the setup of SDL Trados Studio some predefined TQA templates can be imported. After that they will appear in the list of templates when creating a project. It would be nice to have an easier way to find them, especially in the online help as they are a valuable resource. The predefined templates use the TAUS, LISA, J2450 and MQM quality models. The models need to be in the project setup when importing documents. Attaching a TQA model to already imported files does not seem to be possible.
In the editor, SDL Trados Studio shows a comprehensive list of manually created TQA messages for the entire file. Within the segment, the area that the TQA message belongs to is marked with a yellow highlight (Figure 4). It does not seem to be possible to add a TQA message without marking something in the segment (which is hard to do if one wants to mark a missing punctuation mark).
SDL Trados Studio introduced projects and project templates with SDL Trados Studio 2009. The user could either base a project on the settings of an existing project, or could create a new and separate template out of a choice of available project settings. And the project setup was accompanied by the possibility to run several (batch) tasks on the files in the project at once, like pre-translation, analysis and so on.
memoQ has always been project-based, but the templates and further automation was added only recently in version 2014. The automation is connected with the template-based projects only, but offers automated tasks at several points in the project.
Before importing the document, when the file is still outside the memoQ project, a script could be run on the file — to save a text-based file to Unicode encoding, for example. Then, after the import of the documents into the project, memoQ could run an analysis, a pre-translation and export the files to a bilingual document format for further processing with other tools. With the Wrap Up step, one could tell memoQ to export the translated documents, update the Master TMs and delete the Working TMs.
This will probably be an area where more and more enhancements will show up as users request automation of their favorite tasks or task sequences.
Plug-ins and apps
As no product usually has all the functionalities all possible users would want, our translation memory tools offer ways to use plug-ins or additional apps to enhance them.
On SDL’s OpenExchange App pages
(www.translationzone.com/openexchange), users can find many little helpers created by different developers. Some are free; some have to be purchased. And, every so often, an app that has become very popular makes its way into SDL Trados Studio, like “AnyTM,” which allows the use of TMs that do not have the exact language variation or have the opposite language direction required for a project.
In SDL Trados Studio, the area above the navigation shows the link to the App Store as well as any apps one has installed — some of them are already pre-installed (Figure 5).
Access to MT systems can be achieved through the Language Cloud services.
memoQ works with plug-ins that the user can activate or deactivate. At the moment there are plug-ins for accessing TaaS online terminology, several MT systems (for which you will need a registration key or possibly a purchased license, depending on the system) as well as online TMs such as TAUS or MyMemory. These plug-ins come with memoQ and depend on what Kilgray implements there.
In addition, the user can add favorite online dictionaries for terminology lookup or select one from the extensive list that is already set up (Figure 6).
Implementing changes from a reviewed target language file
The review of a translation should happen in the bilingual file so that an update of the translation memory is easily done. But what if the review is done in the final target language file? How does one get these changes back into the translation memory so that they are available next time around? Previously this was a step where someone had to insert the changes manually into either the bilingual file or the translation memory directly — which oftentimes did not happen at all, because of the popular “no-time-for-this-at-the-moment” or “we-will-do-this-after-the-project-is-finished” syndromes.
memoQ made a start with this process in memoQ 2013 when it introduced the import of the monolingual review file. Basically, the segments of the reviewed file are compared to the target language segments in the bilingual file of the project and get displayed in an alignment-like editor. The user can connect previous translations with current (reviewed) translation segments. These segments overwrite the old translations. From there the segments can go into the TM.
SDL Trados Studio offers a similar feature, called Retrofit, new as of Studio 2015. The segments from the reviewed target language file are aligned with the target segments of the bilingual file. One can even skip the review of the alignment, if the structure of the file has not changed.
The only time this kind of functionality does not work too well in both tools is if the reviewed file contains new text that has no counterpart in the translated, bilingual file, or when paragraphs have been moved around in the document.
With the customization of the ribbons, memoQ also added more customizability of shortcuts. One can create one’s own shortcuts for memoQ functionalities, but also to insert special characters. For example, one could define the shortcut CTRL-ALT-Y to add the yen (¥) symbol to texts (Figure 7).
Studio allows one to add any number of custom characters to the QuickInsert list, but this will have to be done by file format in the project settings or options. Unfortunately, it is not possible to add one’s own shortcut to get at all these characters, but only for the first seven characters, which can be placed on the tag icons on the toolbar (Figure 8).
Help and support
Translators need to type a lot and any help a tool can give here is greatly appreciated. memoQ calls it predictive typing and shows content from term bases, non-translatables lists, auto-translation rules, concordance hits and muses (for SDL Trados users, this is similar to an AutoSuggest dictionary) after a few letters have been typed.
In SDL Trados Studio, the AutoSuggest feature calls on the AutoSuggest dictionaries (lists of phrase pairs, created out of TM content), the TM (for exact and fuzzy matches as well as for concordance hits), on the term bases, on the AutoCorrect lists and on any MT provider that is attached to the project. The results are nicely marked with different colored icons to show where they come from.
In addition to online help articles there have always been other ways for us to find solutions to our questions, be it a forum or the official knowledge base of support cases from the tools manufacturers. With the newer versions of the tools, we can access these helpful resources more easily from within our translation tool. Studio provides a separate ribbon where one will find a lot of information ranging from the regular online help to the knowledge base (http://kb.sdl.com), from access to your own SDL account on the SDL home page, from blogs to events to product activation, and for checking for updates.
memoQ uses a similar approach and offers online help, access to videos and guides on their website as well as recorded webinars. What I miss here is the direct access to the memoQ knowledgebase at http://kb.kilgray.com.
And here are some things that only exist in either memoQ or SDL Trados Studio, but which would be really nice to have in the other tool, too.
Support for embedded objects in Office files: One thing that as far as I know only STAR Transit has been doing is the import of text from files embedded in other files, as with Excel tables in Word files. memoQ added this functionality in memoQ 2014, so that one can now import the content of the first layer of embedded files (not text from files embedded in embedded files), when one is dealing with Office files. It is essential that the files are embedded and not only referenced.
Influencing the word count: SDL Trados Studio now allows the user to select in the setup of a TM if a word that is written with a hyphen should be counted as two words (which would be the default setting for SDL Trados Studio TMs) or as one word.
Adding abbreviations to the segmentation rules on the fly: memoQ can now send an abbreviation to the abbreviations list in the segmentation rules from within the translation editor — and it can also resegment the file with this setting while one is still in the translation editor. The document does not have to be imported again with the new settings.
Bookmarks: The new bookmark feature in SDL Trados Studio 2015 is also something that I find very useful: a way to mark segments with bookmarks in addition to the previously-available feature of setting comments on segments.
Correction of segments with the help of known elements: memoQ 2015 introduces MatchPatch, which can adjust matches from the TM with other known elements like terms from the term base or non-translatables. This can improve the match values and although I was skeptical at first whether this really would be feasible, I have to confess that it was a great help during the translation of my memoQ training manual.
Regular expressions: Granted, using regular expressions is not something that many users feel comfortable with. But for those of us who use them regularly, they are a blessing. Most prominently, I would use them for searching through the list of segments. Some examples: searching for all segments that contain a number; all segments that have a certain character at the beginning or end; all segments that contain an expression that contains at least three capital letters; all segments that have a dot at the end with only a few preceding letters (to find abbreviations); and so on. The list of use cases is endless and every one of us has our own special things that we search for.
Both tools allow the use of regex for file filters and in their QA component. But Studio also allows it in the search dialog or the display filters (when filtering the list of segments in the translation editor), memoQ up to now does not. I hear that it is on the list of things to be implemented and I would again most politely ask that this gets a higher priority.
Connected with this would be my wish to be able to search within tags (in both tools) and to, for example, filter for all segments where a tag contains a certain attribute. OK, this goes a little beyond the everyday use for translators, but more and more project managers and language engineers need to manipulate or check files for certain items and this would help a lot.
Something that has come up repeatedly in my work in the past years was the question on statistics for terminology. How many terms from a term base appear in a document? How many of these terms already have a translation; how many need to be updated with a translation? Which or how many terms in the list of terminology extraction candidates already appear in a term base? These kinds of statistics would be very useful when it comes to estimating how much effort needs to be put into terminology work.
But before I go into a full wish list of things that I would like to see in our tools, let me summarize: TM tools have steadily developed, and I am sure will continue to develop, new features that can be very useful. With manifold feedback channels, we are empowered to give input to the tools developers. And it looks like they closely watch each other’s developments, too.