@jonn Dynamic goals would make the experience so much better. Though if you are going to take slower action, of the three goals, new word number is I suspect the one most would like control over. You could leave the card number and percent correct alone, then allow users to set their daily goal.
Or an easier kludge would be just give users the option to stop new words for a given section and do as many of their old stack as possible. Do that, and see how users utilize it and you will have some actionable data to decide what to do when you finally let users control their own goals. Make it so that it automatically returns to adding new words after an hour, and users would have to click it again to continue. That way you would still be defaulting to adding new vocabulary.
That was what made DuoLingo a great program for me - I could do as much as I wanted of practice and I would only advance when I felt comfortable. For the smallish vocabulary that I got from Duo, I know that on a more "comprehending" level. That is, instead of El alcalde -> The mayor -> Concept of what a mayor is, it goes directly from El alcalde -> Concept of what a mayor is. It would also let users practice when they are not tip-top intellectually. If I have been at work and used Spanish moderately-extensively, I have noticed that if I practice right before bed that day, I cement words in my brain better. However if I already did 50 cards and ran up against the limit of having to add new vocabulary in order to practice then I don't because it's just too intellectually taxing after a 10 hour shift in the ER.