Fuzzy match grids in SDL Trados Studio

Has an agency ever asked you for your rates and Trados grid? Or your rates for fuzzy matches? If you don’t use a CAT tool, this post isn’t for you. You can’t give a discount for fuzzy matches in Word. If you do use a CAT tool and need to come up with a fuzzy match rate, read on.

What’s a fuzzy match grid?

A fuzzy match gridTrados grid or weighted word scheme is a method for calculating discounts on fuzzy matches. When you translate a file in a CAT tool, you’ll take less time if you’ve translated a similar file in the past and it’s saved in your translation memory. If the new project is for the same client, you may want to give him or her a discount to reflect the actual time you spend on that project. A fuzzy match grid will help you calculate that discount in an objective way.

What’s a fuzzy match?

A fuzzy match is a segment in a source text that is similar to a segment in a translation memory. It’s only a partial match so it will require editing. The amount of editing needed is reflected in the percentage. A 60% match will probably need complete rewriting, whereas a 99% match may only need to have a comma added.

What isn’t a fuzzy match?

A fuzzy match isn’t how many times a common word is repeated in a file. Matches are calculated on segment content – usually a whole sentence – and only when you use a CAT tool.

What’s a fair discount for a fuzzy match?

It’s up to you to decide whether to apply a discount for fuzzy matches. If you do, the next step is to decide what discount to apply to each fuzzy match range and whether to accept an agency’s scheme. Some agencies will have a fuzzy match grid set in stone: take it or leave it. Others will be open to negotiation.

I usually apply the following fuzzy match grid to my new word rate:

Perfect matches and context matches 20%
100% matches and repetitions 25%
95-99% 50%
85-94% 90%
No match and up to 84% 100%

Note that the percentage isn’t the discount itself. It works the other way round. In the table above, a 100% match gets a 75% discount on my new word rate. That’s equivalent to a weighting factor of 0.25.

Many agencies expect a discount for 75-84% matches. Some will ask for a higher discount for 85-94% matches.

Don’t forget that fuzzy match grids aren’t relevant in many projects. In fact, I only apply my grid in a very, very small proportion of my work.

Ethical issues

As you can imagine, the topic of percentages in fuzzy match grids is a hot potato among translators. Many factors come into play when deciding what is fair and what is not.

  • Will you be using your own translation memory or one supplied by an agency? Having to retranslate poor-quality 100% matches from an agency’s TM at 25% of your usual rate doesn’t make good business sense.
  • Is it ethical /will you be liable for translation errors if an agency proposes locking 100% segments and not paying for them?
  • Even high percentage fuzzy matches may offer meaningless results in short segments. What’s the use of a high fuzzy match when stop words (the, a, if, etc.) are the only matched content?
  • Why should you pass on the benefits of a tool in which you have invested time and money?

Having touched on the ethical issues, let’s go back to the technical aspects.

Fuzzy bands in Studio

Fuzzy bands are the ranges that Studio applies when it analyses a file. The analysis report will show you how many words are in each band. The default bands are:

  • 50-74%
  • 75-84%
  • 84-94%
  • 95-99%

To change these bands for future projects, go to File>Options>Language Pairs>All Language Pairs>Batch Processing>Fuzzy Bands. If you want to add a new file to a current project, and change the fuzzy bands and other settings before you prepare that file, go to Language Pairs>All Language Pairs>Batch Processing>Fuzzy Bands.

Fuzzy bands in Studio

How to perform an analysis in Studio

First, you need to decide what to include in your analysis.

Go to File>Options>Language Pairs>All Language Pairs>Batch Processing>Analyze Files. This is what you’ll see by default:

Cross-file repetitions

Report Cross-file Repetitions gives a separate count for any 100% matches that Studio finds when it compares all the files included in an analysis. This is useful if your files are very similar.

Report Internal Fuzzy Match Leverage sets up a separate category for fuzzy matches that will come up as you translate a file, within that file. They’re not in the translation memory yet.

Report locked segments as a separate category does exactly what it says. If you’re willing to accept a project from an agency that isn’t going to pay you anything for 100% matches (see below), then at least ask them to lock those segments before they set up the project and run the analysis with this setting. Personally, I don’t work under these conditions.

Now you’re ready to run the analysis.

When you prepare a project, Studio will run an analysis on the new file(s) by default.

If you use the “translate a single document” workflow in the Welcome view, Studio won’t analyse the file first, but will open the file straight in the Editor window. You can still analyse the file: save the xliff file (Ctrl+S in the Editor), go to the Files view [target language list], right-click the file and select batch tasks>analyze files.

Analyse a file opened using "Translate a single document"

If you just want an analysis and aren’t going to translate the file immediately, read Paul Filkin’s useful blog post on doing a simple analysis.

Now you know all about fuzzy matches and running an analysis. In my next post, I’ll explain how to export a Studio analysis to get a final price and prepare a quote. Stay tuned!

Image attribution: © waldemarus – Fotolia.com
This entry was posted in 1. The Basics, SDL Trados Studio and tagged , , , , , , , . Bookmark the permalink.