The Issue of Fuzzy Grids - by Philipp Techen

(nur auf Englisch)

The fuzzy grid – “What’s that?”, I hear you ask? A challenging new kind of crossword puzzle? 
A quirky new online trend? Or simply the company spreadsheets left out in the rain?

Translation has always been a challenging task, but the rise of technology has created functional tools to aid translators on their mission to provide content to more people. CAT tools (Computer aided translation) are one such software solution. These tools allow text to be saved in a database called a Translation Memory (usually per sentence or paragraph, but different settings may apply). If the same or a similar piece of text comes up, the Translation Memory can insert the previously translated content and potentially save a good chunk of work. If this is an actual repetition (the exact same text that has been translated before), it does the job, but context is king: the same text in a different context can result in a different translation, especially when it comes to dialogue (imagine how many wildly differing questions could be answered with “I do.”). Now if a segment of text is not a proper repetition and only contains part of the previous text, it is referred to as a fuzzy match. This can certainly make work easier when there are repetitive menu options, item strings in video games, and lists in manuals, but the process of fine-tuning it all is still the translator’s job.
CAT software usually offers plenty of analytics functions to enable accurate planning of workflows and workloads etc. Analysis of a text might, therefore, look like this:

Total: 10,000 words
Repetitions: 1,000 words
99-95% matches: 1,000 words
94-85% matches: 1,000 words
84-75% matches: 1,000 words
74-50% matches: 1,000 words
No match available: 5,000 words

So far so good. The translator purchases this software to provide their services to clients, the translator gets to estimate how much work the text may involve, and thus plans their schedule accordingly. A sensible and functional solution.

Enter the big companies, aka LSPs (Language Service Providers). These companies cover a large portion of the worldwide translation market and since in-house translators require offices and benefits like health insurance, the majority of work is done by external translators. It's neither surprising nor particularly shocking that large companies focus on optimising their earnings. And since large companies can pay a lot more for expensive software than individuals, the CAT tool makers often focus on their business partners. This led us to the age of online collaborative CAT tools. The translation is no longer sent to the translator but instead stored on an LSP’s server to be accessed by as many translators as needed at the same time. In theory, this could also be a tool for good, providing a framework for teamwork (although LSPs often stifle any attempt at communication among translators, which would be a topic for another blog entry by itself), but this isn’t really how it is used in most cases.
Now it's important to know that in many areas translators’ rates have effectively decreased over the last 20 years and have reached a point where LSPs cannot really go any lower. The fact that there should have been a significant increase over the course of the last 20 years is another very sad topic. But what does a smart LSP do when they want to effectively pay their translators less but can't lower the rate directly? They find some kind of workaround – enter the fuzzy grid. (Be honest, you thought we'd never get here.)

The fuzzy grid is simple. The project manager at the company runs an analysis of a text before allocating the job. The job offer is then sent out with a weighted number of words based on how much each word is worth according to the grid.

Looking back at our statistics from before, a typical fuzzy grid might look like this:

Total (raw word count): 10,000 words
Repetitions (worth 10%): 1,000 become 100 words
99-95% matches (worth 30%): 1,000 become 300 words
94-85% matches (worth 50%: 1,000 become 500 words
84-75% matches (worth 70%): 1,000 become 700 words
74-50% matches (worth 100%): 1,000 stay 1,000 words
No match available (worth 100%): 5,000 stay 5,000 words
Total (weighted word count): 7,600 words

In the past, you would have been paid for 10,000 words. Now, all of a sudden, you are only paid for 7,600 words, but are still expected to work on the text as whole, which amounts to 10,000 words. Considering an average rate of 8 cents per word, the translator would have been paid 800 euros in the past. Now it is only 608 euros, effectively lowering their rate by almost 2 cents per word. Obviously, these amounts vary from project to project, but realistically the introduction of these grids can lower a translator’s annual income by thousands!

And efforts to introduce fuzzy grids were successful. For many years, freelancers simply accepted the working conditions of the LSPs and too few external translators complained about them. Networking was actively hindered by the big companies and their failure to credit translators for work. If translators cannot join forces, there is no chance to push back.

And so we have reached a point where LSPs keep adjusting fuzzy grids in order to pay less and less.
50% becomes 30%, repetitions become 0%. The easier tasks in translation should make up for the more challenging elements, such as puns, jokes or transcreation, which in turn are not remunerated at a higher rate. Yet poems, puzzles, and content that requires extensive research, are typical in the field of audiovisual translation.

LSPs argue that the current economic situation is challenging. What they seem to overlook is that we all live in a challenging economy, we all have to deal with inflation, rising rents, energy bills and so much more. Pushing professionals – many of whom hold several university degrees – to work for below the minimum wage is an unsustainable situation.

So if you are working in the field or know someone in the field, don’t just blindly accept the working conditions LSPs want to impose on you. You are a free agent and you can set your own rates. Join your national audiovisual translators’ association and talk to your colleagues to be able to make informed decisions about your business. Because only by standing together can we really make a difference.

Lektorat: Daniel Landes, Christina Milner & Rebekah Smith



Keine Kommentare

Kommentar schreiben

* Diese Felder sind erforderlich