Let’s say you had a four-year old daughter and she was learning to count.
Let’s further say that you had a deep interest in number language, had read widely on the matter and thought very, very hard about it. Maybe you had even published an academic paper on relationships among number language, quantity and numeration (pdf).
Let’s say that you (among many other people before you) had noticed that while the English teens have a pattern, it’s a fairly obscure one.
Let’s also say that you notice that (unlike some other languages such as Japanese) there is also a relatively obscure pattern for the names of the decades: twenty, thirty, forty, fifty, etc.
Finally, let’s imagine that you noticed that we start counting at one, but we don’t start the twenties at twenty-one nor the thirties at thirty-one.
Don’t you think you’d put all of this together to build a theoretical model of learning to count that includes (1) trouble in the teens, (2) skipping twenty in favor of twenty-one, (3) more success in the twenties than in the teens, and (4) ending the count at or about twenty-nine?
And then, when you saw that theoretical model play out in great detail, don’t you think you’d want to capture it on video?
I know I would.