Non-science starts with just collecting data and trying to make inductive generalizations from it, and it gets absolutely nowhere. You just can't do it.
It's been understood, one of the parts of the modern scientific revolution, so-called Galilean revolution, is you don't even try that. That's why scientists do experiments. In fact, in Galileo's case, kind of thought experiments. Like he didn't drop two balls off the top of the tower of Pisa, that would never have worked. He just had a very elegant thought argument why rate of fall would not be affected by mass. Sometimes he may have done experiments---a lot of them didn't.
But the point is, ever since the 17th century, even before, scientists inquire of the world. They don't just observe it, they inquire of the world. That's called experiment. They concoct situations that might give you some insight, and from them they make some guesses about what the theories might be. And then they try other experiments to test the theories. And ultimately they sort of get back to phenomena, but they don't care very much if they get back to phenomena. Cause in fact, the phenomena themselves are so complex and involve so many variables, that you don't even try to approximate phenomena.
Take the examples that I mentioned: bee scientists don't try to approximate bees swarming. It's just too complicated: the wind's blowing, one of them's changed his mind, whatever. And physicists certainly don't take a look what's going outside the window, and try to draw inductive generalizations from it.
You go back far enough, pre-classical Greece, maybe science looked like that. But this is just mythology, it doesn't happen, and it couldn't happen. Scientists are inquiring about nature. And the same is true of linguistics. If you're a field worker, so you're working some unstudied language in the Amazon, if all you can do is take recordings, okay you take recordings. But you're not going to find much. If you're really doing serious field work, you use the techniques you learned in your field methods course in college. Namely you try to figure out the kinds of questions that will elicit data that might be significant and relevant.
You just take a look at masses of data, you basically get nothing, just noise. It's true that it's a methodological critique, but it's a methodological critique of something that dominates in the human sciences that has absolutely nothing to do with science. That's true of the whole behaviorist tradition, or of what was called behavioral science---1950s all the human sciences were called behavioral science. That makes as much sense as calling physics "meter-reading science." It's true that---take Eddington and others---you can regard physics as, in principle, the study of meter-readings. But it's not meter-reading science. You're using the meter-readings trying to discover something about the world.
Well behavior's just data. Not all the data, incidentally, just some of the data. And selected parts of that data, if you aren't smart enough to figure out which ones, may tell you something about human capacities and the nature of the mind. But to just collect data and organize it somehow is going to get you nowhere. If you can't think of anything else to do, no ideas, then maybe you do that. But it's not the way science is done.
https://www.youtube.com/watch?v=-xerglwYdkE