If a rule has an exception it can't be a rule, and the expression 'the exception proves the rule' is almost never used correctly because 'prove' does not mean 'demonstrate beyond question' but 'test to the point of destruction' - as in a proving-ground, or 'the proof of the pudding'. If there's an exception to the rule then the rule is invalidated, and no longer a rule
This is the case with grammar, which Chomsky once described as 'a set of complicated rules to which there are exceptions.' In other words there are no rules of grammar, complicated or otherwise. So what are there?
Grammar is better seen as any frequently-occurring pattern of words that usage settles on. And these are culturally determined, not naturally inherent to any language. Patterns, not rules, are acquired early in our development and finessed in maturity. The top dressing of what non-linguists think of as grammar is just that - a superficial (if, to be sure, important) addition to the long-acquired (or innate, as Chomsky argued) deep structures of the language. To be sure the patterns of grammar are vital, and without them communications would be much harder (although I expect we'd soon compensate in the search for clarity) - but they are not rules.
Correct grammar is hard to define but quite easy to recognise, at least to those of us who are at ease with it. Bad grammar is also easy to recognise, at least by those of us alert to it. Take this startlingly illiterate blurb from the back cover of a new collection of Stan Barstow's short stories (Parthian Press, an imprint of the Library of Wales)
A classic selection of the best of Stan Barstow's stories covering the last five decades of British life.
A group of young tearaways on a night out that begins with horse-play and ends in tragedy; the loneliness of a drunken miner's wife; a war-shocked ex-sailor forced beyond endurance, a widower is brought to grief by a woman outside his real understanding, a factory worker finding his way through the physical world of his marriage - real and involving, Barstow's stories are urgent slices of life, men and women struggling and succeeding to come to terms with The Likes of Us.
This is beyond wrong, although I'm intrigued by 'slices of men and women'. What's alarming is that Barstow (a very good writer indeed) should be so poorly served by the publishers who employ whoever is paid to write this awful stuff - presumably somebody in the Parthian marketing department. The Library of Wales should be ashamed of itself, but libraries, like publishers, have no shame. What's equally alarming is that the person who concocted this atrocious blurb has no idea how wrong they are - they imagine it's punchy and persuasive copy rather than something the cat's sicked up.
Enough already. The subject of this blog is rules for writers, so here are some, from William Safire of the New York Times. He has plenty of chortling fun with the exceptions that prove the rules:
Remember to never split an infinitive.
The passive voice should never be used.
Do not put statements in the negative form.
Verbs have to agree with their subjects.
Proofread carefully to see if you words out.
If you reread your work, you can find on rereading a great deal of repetition can be by rereading and editing.
A writer must not shift your point of view.
And don't start a sentence with a conjunction. (Remember, too, a preposition is a terrible word to end a sentence with.)
Don't overuse exclamation marks!!
Place pronouns as close as possible, especially in long sentences, as of 10 or more words, to their antecedents.
Writing carefully, dangling participles must be avoided.
If any word is improper at the end of a sentence, a linking verb is.
Take the bull by the hand and avoid mixing metaphors.
Avoid trendy locutions that sound flaky.
Everyone should be careful to use a singular pronoun with singular nouns in their writing.
Always pick on the correct idiom.
The adverb always follows the verb.
Last but not least, avoid cliches like the plague; seek viable alternatives.