Welcome, film, and thanks for that pointer. It seems useful stuff mostly, but I take exception to the one-line summary:

> Most of these rules are arbitrary and are learnt only with experience by native speakers.

What a pile of manure this description is - something is either a rule or it's arbitrary. This imho is the ultimate absurdity of the prescriptivist pretence. This sentence actually suggests these things:
1. Native speakers are privileged by some sort of magic ear
2. This allows them to discern required differences in speech patterns despite the apparent random qualities observed in practice
3. This superior performance allows their speech to fit the rules

Balls. Surely what is actually going on is that the accretion of patterns of usage is indeed pretty arbitrary in its variations, and it's only custom that privileges the apparent euphony of one form over another. To attempt to draw a 'rule' given the observable discrepencies is simply perverse. The most that can usefully be said is that certain forms tend to be mostly found doing this job (with notable exceptions) whilst the alternate form is mostly found doing a slightly different job (again, with some notable exceptions). In other words, you can describe what is, but you cannot usefully ascribe the resultant language as the product of some rule.