Morrus said:
Sure - but if you extend that principle, you don't have a language any more.
I could just type giurkjhfd and claim that it was English because that's how I use it. But it's not, and never should be.
There are reasons for rules of language and grammar - very good reasons. They help us communicate. If dialects start deviating too much, they'll be different languages. We won't understand each other (in fact, that's actually already happened!)
It's not about unique made-up usages, it's about common usages that are understood throughout the language community, which are despised and derided solely on the whim of William Safire, E.B. White, or Joe Blow's 9th-grade English teacher. Can you seriously argue that people don't understand what, say, the long-standing and regularly generated dialect form "anyways" means vs. the ostensibly more standard "anyway"? I myself don't find that to be a great intellectual challege.
Not but a few generations ago, there was a commonly promulgated rule, with utterly no basis in historical usage, that "shall" was the correct alternative to "will" for 1st person agreement. Even today you'll find a few troglodytes still trying to beat that little gem into their students, and it's just complete malarkey. No significant speech community can be demonstrated to have ever had such a rule as part of its grammar. It may have been introduced because it was felt by the authors of the rule to be more polite to apply shall, which archaically had (and sometimes still does) a more obligatory sense, to oneself than will, which archaically had a more voluntary sense. It's the same bizarre logic that leads to the insistence on "you and me" vs. "me and you". Who decided it's more polite? God? Jonathan Swift? Mrs. Grundy? I don't feel offended by "me and you." I think you'd have to be at least mildly insane to take that as an offense.
There certainly are perceptible rules to English grammar that are generally agreed upon among the various dialects, but the problem is that the people who write the rule books often seem more bent on expressing their idiosyncratic personal preferences than on elucidating the unspoken rules of usage that prevail in the real world. Sadly these books are then often slavishly bowed to by persons who apparently respect authority more than the evidence of their own ears. "Yes, master, I shall not split an infinitive, for it doth offend thine ears, even though my peers and ancestors have been doing it for the last 800 years <mindless drool>." Anybody can invent any little peeve like that they feel like. Expecting others to really go around changing their entire speech patterns to humor one's affected peeves is juvenile and ridiculous though. The difference with nabobs like Safire & co. is that they actually have some influence with which to make their peevishness felt. Marion Zimmer Bradley was refreshingly honest about it when she pointed out that ultimately her editorial rules were
the rules because it was her magazine. Most of those people act like English grammar was handed to them on stone tablets from Mt. Sinai.
The rules you read in your high school English class are nice and all, but in the real world, they often make your words sound pretty damn stupid if you insist on using them all the time. Scientific writers might as well just throw S&W out, because it's so totally unrealistic with regard to accepted professional style in the sciences, not to mention horribly dated in many aspects of general style.
I have to disagree about dialect divergence as well. What you say may have been true at one time, but practically speaking extreme dialect divergence is dead and it's not coming back. Just in the last 50 years, the convergence of regional dialects in the United States has been distinctly perceptible. Sure, there's still mild drift, but unintelligibility only develops when the speaking community, consciously or unconsciously, intends for it to develop (as with teen or Black slang).
Historically, dialect divergence within cohesive political or cultural units seldom if ever led to true mutual unintelligibility. Even disparate or unrelated languages when bound within some political geographic or political unit (forming a Sprachbund, as it's called) natually tend to come to resemble each other. This is why German and the Romance languages have a number of uncanny similarities which are demonstrably not the result of their distant common heritage.