"I think when it comes to traditional categories, we should err on the side of inertia (or conservatism) because we would otherwise be asking people (admittedly not a large group these days) to change a trusted way of thinking. It's not like we are looking at alternative proposals in a vacuum. If the arguments are equal, I would vote for the status quo."I agree. But in this case, I don't think the arguments are at all equal.Otto Jespersen's observations about the behaviour of prepositions and their ability to take not just objects as complements but a wide range of complementation, including no complement at all--just like verbs, just like nouns, just like adjectives--is both simpler and more robust than traditional accounts. It arose about the same time that Einstein, Bohr, Heisenberg, Planck and others were developing the new physics. We still teach Newtonian physics in high schools and universities because it works very well for "medium-sized dry goods". But even when I was in high school, we talked about relativity and quantum physics (at least at the level of wave-particle duality). In contrast, I'd never heard of Jespersen's ideas until I read the Cambridge Grammar of the English Language a few years ago.
Here's another example. The OED started using 'determiner' as a part of speech only in the last few years (and even now has 'det' as a special kind of 'adj'). Yet it was 1924, when Palmer first tried to corral this group of theretofore-heterogeneous English words by adopting the concept from the French analysis and calling them determinatives. It's a much more parsimonious analysis than grouping 'many' together with 'big', but most grammar teachers haven't got a clue what a determiner is.
When Thomas Kuhn wrote about paradigm shifts in science, he argued that people can't operate in two paradigms at once. In arguing this, though, he refuted himself. Obviously in tracing the history of various paradigms, Kuhn demonstrated that it's perfectly possible to understand multiple paradigms and to translate between them. Admittedly, it's difficult when you're looking at a shift from Newtonian physics to something like special relativity, but it's very possible. The Jespersenian view isn't nearly as hard to comprehend as something like string theory with all its extra dimensions. The basics of his argument can be understood by any average adult with an elementary grasp of grammar in a few minutes. That's not to say that you have to buy into it or decide to teach it, but at least you can understand it.
To my mind, the maintenance of the status quo in grammar teaching stems more from laziness and lack of scholarship than anything else. Most (obvioulsy not all) grammar teachers appear to know next to nothing about grammar beyond what they learned in elementary school and picked up in the textbooks they teach from. Few have ever opened a comprehensive grammar. Many come from lit backgrounds, and seem to assume that by virtue of being proficient users of the language, they are competent to explain it.
Again, even though I think the ideas I'm advocating are simpler, more robust, and more logical than other analyses, I'm not trying to force them down anyone's throat. I enjoy hearing counter arguments. It's fun to interact with informed people with a diversity of views instead of just running into the same gaping ignorance over and over.