Stupid because-I-say-so prescriptivism in The Atlantic and The New York Times over the word “Gif”. “The war is won, even if the battle ensues. GIF is pronounced with a soft G, and if you think otherwise, you are wrong,” sniffs The Atlantic, its nose hoisted well into the air.
The NYT agrees: “A GIF, pronounced jif, is a compressed image file format invented in 1987,” it declares. And both of them, to support their case, quote the inventor of the Gif, a Mr Wilhite, saying : “The Oxford English Dictionary accepts both pronunciations. They are wrong. It is a soft ‘G,’ pronounced ‘jif.’ End of story.”
Here’s the thing: it doesn’t matter what Mr Wilhite says. It doesn’t matter if, as The Atlantic’s headline declares, “If You Pronounce GIF with a Hard G, You Must Be New to the Internet” (and If You Insist on Capping Up Almost Every Word in your Headline, it Gets Very Annoying Very Quickly). That might be true. But it doesn’t matter. What matters is how people use the word.
The Atlantic article is actually completely right about one thing. The word, when first used among a small cognoscenti, was pronounced with a soft G. But, as the magazine itself says, most people will have come across the word, first, by reading it. And they will assume, reasonably, that it follows the rule that most gi-words, like “gift” or “giddy” or “gills”, are pronounced with a hard G. There are exceptions (“gist” and “gin” being the most obvious), but they’re relatively rare.
This is called change by analogy, and you can see it in formerly irregular verbs, like “help”, becoming regular: it used to be that the past participle of “help”, or “holp” as it was then, was “holpen”, but it became “helped” because users assumed that the rule applied to other verbs applied to “help”. It works the other way: the past participle of “dive” used to be the perfectly regular “dived”, but the term “dove” is becoming more popular, by analogy with “drive” and “drove”. You simply don’t get to say that “helpen” is correct and “helped” is wrong.
This is the mistake my former colleague Simon Heffer makes when he says that “collide” cannot mean “a moving object hitting a stationary object”: “the etymology is strict“, he says, that the Latin “collidere” meant a collision between two moving objects. Now, as it happens, he was wrong about that. “Collidere” was used, among other things, to mean a vessel (moving) hitting a floor (stationary). But even if he had been right, it wouldn’t have mattered. We aren’t speaking Latin, and it is of precisely no use to know what the word’s ancestor meant 2,000 years ago. (As Language Log puts it, “By strict etymology, silly would still mean ‘holy’.” It doesn’t, in case you’re wondering.) Etymology cannot be “strict”.
And for that reason, it also doesn’t matter how “Gif” was pronounced in 1987. What matters is how it is pronounced now, and the hard-G pronunciation, as far as I can tell, is far more common. A soft G isn’t incorrect, it’s just rare, and the OED is well within its rights to list both pronunciations. But from my own experience, users of the word tend to say “G”, and the very fact that defenders of the soft G have to write “it’s pronounced ‘Jif'” tells its own story.
• For the record, the Google search [gif “hard G”] returns twice as many results as [gif “soft G”]. It’s far from perfect, as a scientific study, since the former returns the same Atlantic article I’m discussing. But I think it is a data point that supports my case.