A client asked us:
Wordbee can have many target languages in its TMs but we tend to use single-language TMs as that's the way we've always done it.
I was wondering what impact a multi-language TM will have on analysis performance. As we have large TMs and many-language projects analysis can take a long time.
Wordbee supports any number of languages in single memories. In addition a memory can have more than one source languages (if you want/can to use it for multi-directional translations).
Are word counts and translation finder slower with multilingual memories?
Internally, all your many memories (including project memories) are stored in one big "container" in our systems. The separation into "TM"s is a logical one.
If you have a lot of multilingual projects and consolidate them into multilingual TMs, then the word count operations will actually be faster. For example, if we try to find hits for a specific source text, the system will find hits in the multilingual TM and get all language translations immediately and "for free" with just one search.
With bilingual memories, the same source text needs to be searched language by language. So you will see slower performance.
When to use multilingual memories?
Please use bilingual memories vs multilingual memories as you like. This is a personal preference.
You might look at multilingual TMs if:
* You want to keep the total number of TMs manageable.
* You often have projects where you translate each document into many languages. Consolidation is then simplified: You all translations in one step (since you can consolidate into a multilingual memory). With bilingual memories you need to consolidate each project language one by one.
* Adding a segment with 5 languages into a multilingual TM, will create a single TM record with all the languages nicely aligned.
What are your thoughts?