Learnstream Fruit.page

I want to try to break the perspective of lecture-structured courses by switching away from calculus/physics for now and thinking about language learning instead. The new thing at AJATT is "massive clozed deletion." I'll just steal a description (of the cards that would be added to an SRS):

> Having multiple sentence cards with a single or multiple grammar parts taken out. different sentence multiple parts removed for each card. Must input correct grammar point to pass card.
> Simple example:
> Card 1:
> 私＿Cranksです。
> Card 2:
> 私はCranks__。
> It's supposed to train you to think about the grammar you need. I guess it's more production based/active learning.

http://forum.koohii.com/viewtopic.php?id=7070

The point is having a lot of context for whatever item you are learning. And also removing that item so you have to produce it yourself.

Now let's consider this with the questions we are facing with Learnstream: now that we have components, when do we present the pure description or a lecture or an example? How do we support scaffolded problem solving during exercises? 

The answer is MCDs... for everything!

LS Fruit (the name is fairly arbitrary) is a spinoff of LS Atomic that revamps the idea of "Lessons" and "Exercises" by introducing a continuous stream for learning (a learnstream if you will) along with what I call "quiz on demand", which will resemble using an SRS with MCD cards.

From the students perspective, they will hit "Play" for a course and start hopping from one thing to the next until they are done rather than going back and forth between exercises and lessons like before. 

First, the system looks at what, if anything, needs review and pulls up an appropriate lesson, and it displays only the quizzes that contain components needing review. Think of "lesson" here as often just one problem broken up into many small embedded quizzes---in fact, let's call them clips instead of lessons. (Essentially, this is hacking back in the problem-steps stuff we took out before, and it help alleviate problems caused by tagging too many components to a problem.)  So the review is basically an MCD where the context is the rest of the problem, whatever is going on in the lecture, etc.

So what happens when the reviews are done? The exact same thing! Some new components become unlocked (we'll have to introduce an ordering) and appropriate clips are presented.  

The interface will take some experimentation, but I was thinking of allowing a choice when something new is to be learned: Learn, Challenge, or Skip, where "Learn" will just let you read or watch the material, "Challenge" is the same as a review, and "Skip" drops it from the review schedules. In addition, there may be a few clips that cover the same components, so we could present a choice from among a few of them. 

I was hoping to do some kind of MS Paint/Powerpoint mashup type presentation of the UI, but I didn't find any software that didn't suck. (Opportunity!) (You can see what I did manage to make in Google Docs R&D if you really want)

LS Fruit: An apple a day... 
LS Fruit: Keep your studying fresh