To use an AutoSuggest dictionary, you must first have a TM (Translation Memory) of at least 25,000 units of translation. The good news is that you can create a dictionary from one or several TMs. For example: we have a client who needs a translation on telecommunications. We use the customer’s TM (with, let’s say 2,000 units), we add a general TM of our own authorship (adding 15,000 additional units) while also incorporating a general telecommunications TM (10,000 more units). The more we use to create our AutoSuggest dictionary, the better matches we will get in our work and the more productivity we will increase.
It is not yet proven but apparently, this feature works well with many languages and not so well with some in particular. An example is Finnish, which uses many different suffixes for the same word, so the fragment may not appear in the list of auto suggestion, or there are several forms of a word or passage, which ends up making very slow to use.
If your TM does not have enough units for this function, there are several alternatives:
1. As mentioned above, you can merge two or more TMs on the same subject and in the same pair of languages. This, in Trados 2007, for example, is accomplished by opening each TM from Workbench > File > Export. A .txt file will be created. Once you repeat this procedure with all the desired TMs, you go back to Workbench, create a new TM in the appropriate language pair and then go to File > Import, and you can populate an empty TM. Repeat the procedure to import TMs until you get to 25,000 units of translation to create your AutoSuggest dictionary.
2. There are public use dictionaries available on several websites. The most common are those of SDL itself.
3. Create a TM from scratch, using Trados’s WinAlign tool, aligning source and destination files.
AutoSuggest dictionaries are an additional helpful tool when translating and when used in conjunction with MultiTerm glossaries and AutoText entries, the increase the quality of our translations and speed of our translators, which is directly proportional to the overall productivity of our services. In fact, I would say that this feature is a milestone in the history of CAT tools, perhaps the most important after the introduction of “fuzzy matches”.