GET A FREE CONSULTATION OR SAMPLE TO GET YOUR PROJECT GOING.

Yes, I want my consultaion

QA Distiller: An Old Pal

Checking translation with QA DistillerWe’ve been using QA Distiller to run quality assurance checks for years. QAD meets most of our needs. If we decide to switch to something different, it should be something that’s clearly supreme to QAD. A few weeks ago, I came across a program called Verifika that might be our (and yours!) next QA tool of choice. I came to know about it in the same way that I had learned about QAD: a client sent us a check report generated by analyzing our translation with Verifika. This might be a sign that change is in order.

Verifika is a product developed by a Russian translation agency Palex. Just as QAD, it’s a standalone program running multiple types of checks on various bilingual files such as SDLXLIFF or TMX. Before actually reviewing Verifika, I want to put the comparison of Verifika and QAD in perspective by giving you a recap of what I believe are QAD’s advantages and areas for improvement. Let’s start by reviewing some of QAD’s advantages.

Basic advantages

  1. QAD includes almost every type of checks expected by a translation professional these days. (I guess this is why Yamagata Europe, the translation company behind the program, did not release any new versions in the last couple of years.) The program supports checking for various kinds of omissions, inconsistencies, formatting, or terminology errors.
  2. QAD is built for high efficiency. It displays all errors in a single, easy-to-navigate list, allowing you to move between the errors by pressing up or down arrows. You can open a translation unit in a bilingual file to correct an error by simply pressing Enter. Almost all functions have keyboard shortcuts assigned to them.
  3. Floating license. While QAD isn’t cheap (the cheapest version is €249), its TCO isn’t that high because this license type makes it possible to acquire just one license and use it across as many PCs as you want. Of course, there’s a natural limit to this since if you purchase just one license, you won’t be able to use QAD simultaneously with a colleague. One of you will have to stay on the sidelines until the other one is done.

Specific advantage: Stem-based terminology check

I love this feature because it makes a huge difference in our workflow, saving at least a few hours per week. Here’s what it does:

For instance, I have an English to Russian translation glossary with this term:

Source term: Console

Translated term: Пульт

I need to make sure this term is consistent across the entire translation. This means I want to detect those translations where it was incorrectly translated as, say, “консоль.” To do so, I need to add my glossary to QAD, enable terminology check, and disable the Whole words only checkbox both for the source and the target in QAD’s settings.

As a result, QAD will check that there is one “пульт” in the target for each “console” in the source. For example:

Source: Clean the consoles.

Translation: Очистите пульты.

QAD’s judgment: Correct

Source: Remove the cover from the console.

Translation: Снимите крышку с пульта.

QAD’s judgment: Correct

Source: Handle the console carefully.

Translation: Необходимо осторожно обращаться с пультом.

QAD’s judgment: Correct

Even though each “пульт” in the translation had a different ending appended to it (-ы, -а, -ом), QAD was able to match all of them with the term in the glossary.

So far, so good. But what if I have a term where the ending is not appended like the ones above, but changes instead? This rarely happens in English, but happens in other languages such as Russian all the time. Here is an example:

Source term: tab

Translated term: вкладка

Other forms include:

вкладки

вкладку

вкладке

Or even: вкладок

Or what if you have a multiple-word term like the one below?

Source term: Working console

Translated term: Рабочий пульт

If you leave these terms in a glossary as they are, this is what QAD is going to find:

Source: The tab opens.

Translation: Откроется вкладка.

QAD’s judgment: Correct

Source: Open the tab.

Translation: Откройте вкладку.

QAD’s judgment: Incorrect, expected: вкладка

Or:

Source: Clean the console.

Translation: Очистите рабочий пульт.

QAD’s judgment: Correct

Source: Dismantling the console.

Translation: Демонтаж рабочего пульта.

QAD’s judgment: Incorrect, expected: рабочий пульт

In other words, QAD can find just the exact match of the glossary term whether it occurs as a standalone word (e.g. “пульт”) or as a part of a larger word (e.g. “пультов). QAD fails to detect this term in all those numerous occurrences where the case or the form of a term is different from the one in the glossary; that is, the ending is different. As a result, it generates a plethora of false positive errors that take countless minutes to check and discard.

This is where the stem-based search comes into play. Instead of having a full form of a term in a glossary, I can enter just the shortest possible stem. The goal is to enter a stem that will match all the possible forms of a given word. This applies to single words and word combinations alike.

In the example above, here’s how I address this issue in a glossary:

Source term: tab

Translated term: вклад instead of вкладка

This term will match all the occurrences below:

Вкладка закрыта

Открыто 5 вкладок

Работа с вкладками

Even though the endings are different, the stems are the same and match what’s in the glossary. As a result, I can still check the entire translation for consistency properly, but this time, I won’t get any false positive errors.

The same goes for multiple-word terms. Each constituent word should be a stem only. Here’s an example:

Source term: Working console

Translated term: Рабоч пульт instead of рабочий пульт

This term will match all the occurrences below:

Обслуживание рабочего пульта.

Запрещается контакт с рабочим пультом.

For the translators like us who work into the languages where a word can take 4 or 7 forms with different endings, this is nothing short of indispensable. We can have just one form in the glossary, and it will match all the occurrences. Otherwise, we would’ve either to include all the forms in a glossary or check an avalanche of false positive errors in a QA report. Both options are insane from the efficiency perspective.

Summary

This is an non-exhaustive list of what makes QAD indispensable in our translation workflow. It’s powerful. It’s relatively cheap. And it lets us be very efficient when it comes to checking terminology.

Stay tuned for the next article where I’ll review QAD’s areas for improvement. In the meantime, you can read more about automating the terminology checks.

Add comment


About the Author

Roman Mironov
Roman Mironov
CEO & Founder

As the founder of Velior, Roman has had the privilege of being able to turn his passion for languages into a business. He has over 15 years of experience in the translation industry. Roman has helped dozens of clients increase sales by making their products appealing for speakers of other languages.