State-save for the quiz is a useful feature. However, personally I do not see a *need* for more than one. It will complicate the interface (a separate menu item to select which save to resume, a save dialog window to choose the name for the save, etc.) Study lists are usually of manageable size. What would be the practical purpose of switching between quizzes back and forth?
In my case, this would be useful because I'm studying Japanese, Spanish, and German. So let's say I start a very long quiz in Japanese on Monday. Of course, I cannot complete it because the study list contains more than 300 words. The next day, if I feel like learning German, I will lose the state of the Japanese quiz as soon as I start a new quiz. I wish I could resume the Japanese quiz on Wednesday. And why not, resume the German quiz on Thursday...
That's the use case.
I'm not sure the loading/saving the selection itself is useful, but loading/saving quizzes could be useful (at least, it would be, for me ) to satisfy this use case.
Actually, to make things simpler, maybe I could have one saved quiz state per language pair. So if I start a quiz with fr_de (French -> German), and quit. The quiz would be saved and resumable only if the language indicators are fr_de. If I start a quiz in en_ja (English -> Japanese) and then later, select Resume Quiz, the latest quiz in en_ja would be restored. That could be an interesting option that would not modify the interface at all and suits my short terms needs. What I don't like about this is that the handling of saved quizzes is not explicit to the user. This behavior will have to be explained in the documentation. The advantage though is that it won't bother people that are learning just one language, and I suspect that this is the majority of toMOTko's users. That seems like a good compromise.
EDIT: I implemented this option yesterday. It works like a charm. If you just learn one language, you will not even notice it's there. And if you learn more than one language, you will be able the resume a quiz for each learnt language.
However, I think keeping all sorts of statistics would be very useful (numerical, graphs?). Seeing how well/bad you are doing creates psychological incentives, and helps you to identify the areas that need improvement. For example, recently I wanted to know how many words I had learned so far in Japanese. The only way I knew how to do it without counting words in each study list that I have learned was to select all these lists and go through the combined quiz. At the end, there was a notice that I had learned four hundred something words. However, I know that some words are 100% memorized but I have trouble with others.
I began to address this a little bit. I've introduced a Properties panel on which I display the count of words, glossaries and folders. I will probably add other statistics in there in time. If I can, I would like to use graphs as I think it could be fun to do and to see. But to tell the truth, I have not given much thought about this aspect.
I think that if the algorithm that picks the next word to be asked was able to consider these statistics that I won't need to load/save the words selection so I probably don't need that.