In the course of writing a Professional Development module for teachers, and I've spent hours downloading papers about algorithms that kids invent for division of fractions.
Here's a summary of a not-so-common, but very useful method for dividing fractions and the representation that facilitates its discovery. I'll call this the scaling algorithm because it is born out of considering areas of rectangles and how the ratio of area to length of a side is preserved under scaling. The algorithm is:
To answer this, let's make seven copies of the rectangle -- Why? Because integers are easier to deal with and we all know that 7x(2/7)= 2. The question has now been reduced to "What is the width of a rectangle whose area is 2 and whose length is 7x 3 fourths?"
Since 7x3=21, we have 21 fourths as one side of the rectangle. We have exchanged ? x (3/4) = 2/7 for ?x (21/4) =2. We may also notice that scaling each factor by the same number preserves the answer!
Let's scale again!!! Take four copies of the big rectangle, which gives you a total area of 8, and a total side length of 21.
We have now replaced all fractions with integers and traded the original question in for the simpler sounding:
"What is the width of a rectangle whose area is 8 and whose length is 21?" At this point, we see that the answer is 8 divided by 21 or 8/21! Woohoo!!!
Do kids invent this scaling algorithm? Indeed they do if they are given the area context and their brains are not stuffed with "this is how you do it" algorithms. Want to read a paper that talks about this? Jaehoon Yim from South Korea has a 2009 paper that's fascinating -- Children’s strategies for division by fractions in the context of the area of a rectangle -- I'm not sure if you'll be able to access it for free. But, the general gist is that ten and eleven year old students were able to "make the width equal to 1, make the area equal to 1, and change both area and width to natural numbers". The strategy outlined above is of the last variety. She even researched whether children could formalize their pictorial drawings to create a numeric algorithm -- and surprise, surprise -- they could! Granted, these were students who were picked because they had a "positive attitude towards mathematics".
Incidentally, if the numbers are nice like 6/20 divided by 3/4, then they can use the area model to see that the answer is 2/5 since 6 divided 3 is 2 and 20 divided by 4 is 5. In other words,