To what degree should a CAT tool cater for human errors and sloppiness?
FI-Schutzschalter isn't recognised in the glossary, here.
You actually request a fantastic feature called "Correct human errors" ;)
Nope, Igor, this is just what's mean to be the start of a philosophical discussion ;). I'm sure that MB will tune in, as Home & Garden Philosophy is one of his stick horses.
Seriously though, the number of human errors in typing is almost infinitive for all languages. One solution would be to allow automatic search and replace in the source segments with the option for the user to decide what to search and replace. Now, it is quite easy to correct such errors by search and replace in source segment scope before you start the translation.
Yes, I've thought about that too. I'd have to run some tests with memoQ and Studio first, to make up my mind: do these changes at the source-side also imply that the final projects in memoQ and Studio get changed? Meaning: will my memoQ and Studio get 'falsified' projects delivered, that won't result in 100% pretranslations when they use the project (TM from the delivered project) to re-translate the same segments?
This needs to be investigated.