So if breaking grammar provides certain effects, wouldn't the standardization of such constructs into canon grammar diminish the effect?
Those words are powerful, but not because they form a comma splice. That is, their power does not derive from failing the model's rules. Rather, the model's rules fail to capture that those words are powerful. (Or even allowed.)
Again, grammar is a model. A model that people don't actually use to communicate with one another. Nobody hears words and constructs a parse tree to figure out what they mean. So what's encoded in the model (or not) has little effect on the effect that the words actually have.
Change the model all you want. Nobody will hear any differently.
EDIT: P.S. When I wrote that rhetoric was "breaking grammar for effect," I didn't mean to imply that the effect of rhetoric came from breaking grammar. Rather, I meant that the focus in rhetoric is on the effect and, because grammar fails to admit many effective phrasings, rhetoric necessarily "breaks grammar" from the outset.
That is, there is no "rule set" in reality. Nevertheless, you can create a rule set that describes some of reality. Just don't start thinking it's the real thing.
I offered a sentence in English and made the claims that (1) it was powerful and (2) not admitted by the grammars that govern such constructions. I thought that these were necessary and sufficient to make my point.
Do you believe that either of these claims is false (or somehow depends on what's true or false for Latin)?
Thanks for your help!
Here's one from English, in honor of the upcoming holiday:
Go back to Mississippi, go back to Alabama, go back to Georgia, go back to Louisiana, go back to the slums and ghettos of our northern cities, knowing that somehow this situation can and will be changed.
As far as your claims, (2) is contentious. Some grammars rail against the evils of the comma splice, others acknowledge its place and provide more nuanced guidelines for proper use. It's never been a cut-and-dried issue. I personally disagree with (1) for English while agreeing with it for Latin, but that is neither here nor there.