I'm analysing a large ttx file against a TM which contains an earlier translation of a similar text. While translating this earlier text, I expanded and shrank a lot of segments. My question is: Will Trados find these matches in the new ttx (where obviously no segments have been expanded or shrunken yet) as well, or will it only check the standard segmentation?
No, only the default segmentation will be used.
So, you'll receive only fuzzies or nothing at all.
Probably the best tool here is memoQ.
Although the first analysis will not take in account the modified segmentation, the pretranslation permits to leverage the manual changes made in the precedent project (the segments are expanded/shrunken automatically), the second analysis will be much closer to the reality.
Generally, it works well although the leverage may not be perfect at 100%.