Scoping Leveraging and Scoring
The differences in the two WorldServer translation memory (TM) modes are perhaps most apparent when you scope a project. WorldServer 9.x mode and the Studio-aligned mode use very different underlying technologies to break assets (translatable items) into segments, and to break segments into tokens. The Studio-aligned mode produces the same scores as if you scoped the work in Studio itself. SDL Trados Studio users will appreciate this consistency.
We anticipate that users will choose to do all their work in one mode or the other. If you compare the scores that result from scoping the same assets using different TM modes, the scores for the translation matches will differ, producing different leverage levels. If customer cost models depend on leverage levels, the costs associated with the projects could change if you change TM modes.
Language Fallback
In SDL Trados Studio, languages without specified regions (so-called "region-neutral cultures" or "region-neutral languages/locales" such as en or de) are not supported. Instead, users must use a fully region-qualified language (en-US, en-GB, de-CH, …) to obtain the correct date, time, and number patterns. For that reason, WorldServer internally now uses the most likely region-qualified language in case a WorldServer language doesn’t specify one. For example, en-US will be used for en, and fr-FR will be used for fr.
WorldServer and Studio TM Engines May Return Different Translation Units
The Studio-aligned TM mode makes sure that the same translation unit returned by the TM engine gets the same score as it would in SDL Trados Studio. However, the two TM engines may return different translation units (TUs) as the “winning” match for some document (lookup) segments. If that happens, the two returned TUs may be scored differently, which may result in minor differences in the scoping reports if the two TU scores end up in different scoping bands.
This is expected to occur only rarely, and the likelihood is higher in the lower-score fuzzy bands. For higher fuzzy bands, this should occur very infrequently.
In very rare circumstances, it may even be that no TU will be returned by Studio, in which case the document lookup segment will be reported as a “no match” (new translation) against the TM in Studio.