Terminology Verification – I had forgotten this

I have been working on a larger project that has a lot of 100 % and high fuzzy matches from the client’s TM. Unfortunately, it looks like everyone and their uncle have been translating the previous versions of the manual and consequently several key terms have been translated very inconsistently. While pulling my hair out trying to fix the terminology, I remembered Studio’s terminology verification feature. I have never really used it for my own work, so I wanted to try it out and see if it could help me with this project.

Studio offers three different types of terminology verification. One of them is under the QA Checker and the other two are part of the Terminology Verifier tool:

1. QA Checker: Word List

Here you can create a target language word list of words or phrases that should not be used. The list also needs to include the “correct form” of the word. However, the correct form is not going to be used to automatically fix the incorrect word but it’s included in the error message to remind you what the correct form should be. It’s an obligatory field but it doesn’t really matter what you type into it. In addition, there are also the “Ignore case” and “Search whole words only” options for fine-tuning the search.

2. Terminology Verifier: Check for possible non-usage of the target terms

When this is selected, source segments are checked for terms that are in the default termbase. If a term is found, then the terminology used in the target segment is compared to the terminology in the default termbase and if it doesn’t match, you get a notification (error, warning or note).

You can adjust the minimum match value (fuzziness) for the source term but unfortunately not for the target. That makes the feature much less useful for Finnish because even a slight variation in the Finnish target term makes it unrecognizable and triggers a notification. This happens very often because of the various endings used in Finnish.

This feature can also produce a lot of false notifications because every occurrence of every term that’s in the default termbase is checked, and if you haven’t used the same exact term in your translation, this discrepancy is flagged. Obviously, this works much better with a very specific terminology and in languages where words don’t change much.

3. Terminology Verifier: Check for terms which may have been set as forbidden

This takes a bit more setup work because you have to define a picklist field in the termbase and then enter the forbidden terms to that field for those terms where needed. Adding a new field to a termbase is actually very simple in MultiTerm 2009. You just go to the Catalog view, right click Definition and select Edit. This opens the Termbase Wizard where you can then add the new field. Note that the field must be a picklist field.

This produces much less false notifications than the “Check for non-usage” feature because it only checks for those terms you have specially entered as forbidden in the termbase. However, for Finnish, this has the same problem as the “Check for non-usage” feature because it doesn’t allow enough fuzziness in the target term.

Conclusions and some Finnish grammar

I found the first method to be the most useful in my case, mostly because I only had about ten terms I was concerned about – so it was easy to enter them into the list – and because the other two methods produced too many false notifications. I just entered the main part/root of the term and unselected the “Search whole words only” option, and the QA Checker was able to find the incorrect terms regardless of their conjugation. For those of you who don’t know what I’m talking about, here’s an example. The Finnish word ‘nappula’ (=button) can be conjugated in tens of different ways depending on its usage in a sentence, such as nappulat, nappulan, nappuloita, nappuloissakin, nappulattomille, etc. Only the nominative case ‘nappula’ and very similar conjugations (such as ‘nappulat’ and ‘nappulan’) would be recognized by the Terminology Verifier (features 2 and 3 above). Anything else would be too different. However, the QA Checker Word List function would recognize all these if I typed the root ‘nappul’ as the “wrong form” in the word list (and unselected the “whole words only” option).

Anyhow, Finnish or not, these are good features to remember. You probably don’t need them all or all the time, but in certain situations they can come in very handy.