I set up a project with 2 xlsx files. The count says they have 34.932 untranslated words (one smaller, the other with 33.347 words), including 22100 repetitions.
The same project in memoQ or Trados 2014 (simple settings, without any of the well-known tricks) has about 20.000 words including 7000 repetitions.
How can this be? Is there anything I can change to get correct numbers (hey, I offered a quote on the CT count).
Can the diff be caused by numbers? Especially the gross word count of memoQ and CafeTran are very similar.
Isn't it so that e.g. memoQ already in its default analysis settings leaves all reps out? You can create an extract that only contains the unique segments in the MQXLIFF to compare. THOSE differences can be gigantic.
This is a never-ending source of frustration and annoyance to me too. All programs offer different counts and God only knows what’s really going on. I usually try a few, pick the best, and hope the client agrees
Okay, I tested it again yesterday with the very basic settings of memoQ (I do not believe in SDL counting features) and the problem remained. OmegaT counted even only 17.500 words ...
I sent the bigger, problematic file to Igor, maybe it helps (and anyway, the client accepted the count, I will see what I do...).
I tested your Excel file as follows:
1. Task > Transfer source segments to target segments.
2. Project > Convert project > To Text File.
3. Open the resulting text file in LibreOffice Writer.
4, Check the word count in LibreOffice.
The word count is exactly the same as in CafeTran.
Igor is right.
This would mean that memoQ, Trados Studio and even omegaT omit some cells or fail to import the whole file (it has eight tabs).
The difference is net about 5 K weighted words (and obviously the client hat the same result) - very nice.