Smartcat decided to create an open access translation memory based on the materials about the coronavirus
The same goes for Smartcat. It has become complicated these days to write a post without mentioning the name of this ill-fated virus.
Taking into account the numerous texts about the coronavirus, epidemics, etc. being hastily translated in all language directions, Smartcat decided to do a good deed collecting them in a translation memory open to general use—CovidTM, and invites all translation companies to participate in its filling. Once the appropriate materials appear, it will be uploaded as an open access multilingual TMX file. Smartcat promises to fill it with new translations regularly.
In the time of the pandemic, many developers are trying to help their clients
Starting from April 1 and version 9.3.5, QA Distiller (also known as QAD), a semi-automated translation quality assessment program, has become free. No, it’s not a joke.
QAD is a competitor of Xbench, Verifika, and other similar programs. It has been offered in three versions so far: Freelance, Professional, and Enterprise. Now, there is only one left—Professional,—and you don’t have to pay for it anymore.
Every language has its specific features, but some of them are quite a challenge for translators
Among the vast diversity of sounds used by the human to construct words, the so-called click consonants, or clicks, which occur in the Khoisan languages, stand out rather distinctly. They are spoken by the peoples who live in southern Africa near the Kalahari desert. The best-known nation of the Khoisan-speaking indigenous groups is the San people, also known to the rest of the world as the Bushmen, and mostly thanks to the great movie The Gods Must Be Crazy and its sequels.
Those of you who weren’t lucky enough to be born into a San family will find it impossible to pronounce these sounds without training (the task gets even more challenging as there are as many as five kinds of these sounds). Here is what they are like.
When explaining how such extraordinary sounds had found their way into the language, the popular modern anthropologist Stanislav Drobyshevsky introduces an assumption that they assist during hunting: animals don’t associate them with the human speech and consider them to be natural sounds allowing hunters to get closer.
So here comes the issue of writing down the click sounds as there are no actual letters for them. Linguists in many countries have come up with different solutions.
For example, the name of the Namibian bush farmer who starred in the mentioned The Gods Must Be Crazy movie can be written in English as Nǃxau ǂToma or Gcao Tekene Çoma. There have been multiple systems developed, and alternative spellings are still used. Written versions look rather scary, but that’s the best we have so far.
Usually, this error doesn’t prevent you from translating. Nevertheless, you can get rid of it.
First, let’s figure out what causes this error.
When you create a package for translation in Trados Studio, each file you add to it gets converted into the SDLXLIFF format, and the translation is performed in this new file.
The name of the SDLXLIFF file is generated in the simplest way possible: by adding the SDLXLIFF extension to the name of the original file. Thus, a file with the name, for instance, Text_to_translate_asap.docx is transformed into the file Text_to_translate_asap.docx.sdlxliff.
By the way, it means that with the file’s name you can reconstruct the name and type of the source file used for its creation very easily. Apparently, a Trados Studio file with the name Translate_right_now!.xlsx.sdlxliff is created from the Excel fileTranslate_right_now!.xlsx.
Once the translation is completed, you face a reverse task: you need to recreate the original file replacing the source text there with the translated one. This operation is called “Clean” and stands for cleaning the source text off the file. To perform this operation, Trados Studio has to store the source file somewhere.
If the source file is small, it is stored right inside the SDLXLIFF file. If it is big, Trados Studio remembers its location to return to it when needed later.
What file’s size is considered small and what is recognized as big is determined by the parameter in Trados Studio that is hidden here:
It equals 20 MB by default. Looking ahead, we need to say that it’s worth increasing this value.
If the work on the project is performed at the same computer where it has been created, no problems usually occur as Trados Studio knows where all original files are stored. But if the files have been moved or if the package is being sent to another computer, Trados Studio will have no access to them.
If Trados Studio gives out the “Dependency file not found” message, it means that it needs the original files, but it has “lost” them. That’s why it will wonder, “Would you like to browse for this file?”—i.e., ask for specifying the path to them.
It will be great if you have the original files. Then you say “Yes” to the question and just specify the path to them. The error disappears, Trados Studio calms down and proceeds with operation as usual.
If you don’t have the original files or if you decided not to specify the path to them, answer “No” as there is no other option actually. You’ll be able to keep working on the text, but some operations with files won’t be available (in particular, Save Target As, Preview, Generate Target Translations, etc.). In most cases, you won’t need them anyway.
Generally, this error doesn’t prevent you from continuing translating: you’ll be able to finish the translation, create the return package, and send the translation to the customer. Since the package was created on their computer, such a problem is not going to puzzle them.
But there are situations when Trados Studio refuses to work because of this error. In this case, the most important is to save the translation memory you’ve been using. It keeps all the segments you’ve translated if you had been confirming them while translating. You’ll be able to use them in the fixed Trados Studio package.
The topic of machine translation capacity is being quite hot these days
Machine translation systems often amaze us with the quality of their product, both brilliant and horrible. Mr. Translator by Tencent, for instance, made a fool of itself.
As of today, the biggest issue about machine translation lies in reproducing the syntax of the source language in the sentence in the target language. But people do it in another way. We decode the phrase meaning in the source language first and then encode the meaning using the target language.
Generally speaking, the underlying issue of machine translation systems in their current state is failing to work with the meaning. When they give a wonderful text, it is no sign of their excellence in translation. It is the sign of their excellence in imitation. Hence, there is a question of whether it is just to recognize an imitation system as artificial intelligence.
While using old versions of WordFast, you may get an attribute error
If you use WordFast Pro 5.6 or another version older than Wordfast Pro 5.12 and try opening TXLF files generated in the new version, you may get the following error message:
Attribute ‘gs4tr:uuid’ is not allowed to appear in element ‘file’
It takes place because earlier WF versions can’t recognize this attribute. To solve this problem, you can download and install a new 5.12 version. But there is another bypass solution.
Open the TXLF file in the text editor Notepad++, use the search function (CTRL+F) to find “gs4tr:uuid,” and delete the attribute together with its number and the redundant space as shown in the screenshot. Click the Save button and close the file tab.
After this, the file opens in WordFast 5.6, and no error messages appear.