Before you decide to actually use one or more TMs for a document, you should answer the question whether this makes sense. In general, using TMs makes sense if this saves time and effort. Usually, this is especially true when you translate technical documents (in the widest sense) or when you have to translate a revised version of source document. Especially in the latter case, you can save an enormous amount of time by using TMs.
However, in some cases, you might prefer not to use TMs. For example, when you translate novels or highly. scientific texts, you might even feel annoyed when you are presented with segments from a TM, for the similar segments can have completely different meanings in different contexts. On the other hand, even in "texts of art" there can be many repetitions (especially in poems).
The question whether to use a TM or not is connected to the question whether you use the automation options: If you de-activate all automation options, you have full manual control over how the TM is used so that you can avoid the "annoying" effects of using a TM. (For more information about automation policy, see TM Automation Policy and Leverage Effects.)
Upshot: As a general rule, you are advised to use a TM, simply because you should not miss any efficiency gains through MetaTexis. However, there might be cases when using TMs can be annoying or superfluous, and you have your preferences, of course. Plus, you can always change as you go along: You can turn off the automatic functions, and turn them on again later. Unlike other CAT tools, MetaTexis gives you complete flexibility.
If you have decided to work with TMs, you should spend some time on thinking about your TM policy: How many TMs do I want to use? What is the best way to organize them? Where do I want to save them? What do I need my TMs for? What are my personal preferences? Which policy is the most efficient one?
As you can see from these questions, you have many possibilities. Below, I will present and examine a few possible strategies:
· Some translators prefer to use only one TM for all translations. This TM contains all the translations ever produced by the translator.
Advantages: It is always clear where all TUs are saved. Copying, making backups and transferring the TM to other computers are easy. Most importantly, the leverage effects are best if all TUs are in one TM and are available all the time.
Disadvantages: The database can become very big so that the saving and search processes can become a bit slower. You can lose overview of the contents.
· Another faily extreme strategy is to have one TM per project. Each TM only contains the TUs of one translation project which might consist of one or of many documents.
Advantages: The TMs remain small and the searching and saving processes are very fast. If you want to search in more than one TM, you can simply add other TMs as secondary databases (see below).
Disadvantages: You can lose overview. You have to be careful where to save the TMs: Together with the project documents? In a special directory? With sub-directories?
· Another strategy is to have a TM for each broad subject. For example, all economic documents are saved in the TM for economic texts.
Advantages: The number of TMs remains small while the size of the TMs does not necessarily become too big.
Disadvantages: Many texts cannot easily be allocated to one single subject. So, in some cases, you might have to save a document in two or three subject TMs. You might lose overview.
The first option seems to guarantee the best translation efficiency, for the TM does not have to be organized! The only purpose of a TM is to provide TUs for re-use, and, for this, no order is required.
Each approach has advantages and disadvantages. In any event, you do not have to stick to only one strategy; you can follow all strategies at the same time! And you can always adapt or change your strategy as you go along.
If you change your strategy, some work is required because you have to re-build your TMs. But this task is made easy through the Batch processing command in MetaTexis (see Batch Processing).