Hugh Kenner, in The Pound Era
"If we no longer think, with Swift and Johnson, that languages ought to be stabilized, we still feel that their proper condition is stability. The admission of ain’t to a large American dictionary provoked newspaper hysteria in 1961-62. That in Canto 53 the same emperor appears indifferently as Tcheou Kong and Chao Kong causes many readers uneasiness outweighing the instruction the Canto affords, and a scholarly convention in citing the word ideogramic is to tag it [sic], meaning “not so in my dictionary.” Words, since the 18th century, have seemed fixed upon a rigid and authorized grid, each little violation of which incites the Great Anarch.
Behind such feelings lies the notion of a stable shared world in which all men’s senses participate and the features of which have been labeled by agreement, though different agreements obtain in Italy and in Sweden. Gatto, say the Italians for some reason, and katt the Swedes; it would be simpler if they said the same thing, but anyhow cats are cats. The linguistic contracts, being arbitrary, are fragile, and the only code book, Webster’s or Larousse’s, wards off unspeakable disorder. An alternative notion, that names should be left in place because they are somehow right, is traceable in theory to Plato’s Cratylus but in practice to costive notions of correctness. Both positions were still seriously defended in the early 19th century. Both linger in the average literate psyche. Both were rendered obsolescent by the slow discovery of language, a complex coherent organism that is no more the sum of its constituent words than a rhinoceros is the sum of its constituent cells, an organism that can maintain its identity as it grows and evolves in time; that can remember, that can anticipate, that can mutate. Latin is not a dead language; everyone in Paris speaks it, everyone in Rome, everyone in Madrid. The poetic of our time grows with this discovery."
Borrowed from this Languagehat post. It bears repeating.