> No, that's false. The sole purpose of language is to communicate, and that only requires two people. If two very intelligent youngsters create their own private language, and if the language produces communication, how can that not be called a legitimate use of language?
It's an entirely legitimate use of languages, and both speakers of that language can understand each other. Doesn't mean those two speakers won't disagree on the meanings of certain words. Add a third speaker and the language will develop more nuances.
Yes, language exists to communicate. Like it or not, some attempts at communication will be perceived as correct, and some will not. Assuming you speak or write in an attempt to communicate, and not just to hear yourself talk, you need to care about that perception.
You talk about communication between two people; that's already sufficient to raise this issue. One person may speak or write, and the other person may hear or read, and while the second person's very flexible parsers may correctly interpret meaning, those same parsers will also generate signals like "this person seems quite intelligent" or "this person writes like a lazy teenager on a cell phone". Useful language guides will tell you not just the meanings of words, but also how those words will be perceived by your audience.
When I referred to "consensus", I simply meant this kind of information about perception, scaled up to the entire population of speakers of a language, and made statistical. There's a fuzzy scale from "wrong" to "colloquial" to "dialect" to "variation" to unqualified "correct".
A dictionary that said "there" is a possessive pronoun also written as "their", without commenting on correct and incorrect usage, would not be nearly as useful as a dictionary that does.