The second life of a Language Log comment
« previous post | next post »
More than four years ago, on Aug. 23, 2010, Doctor Science left the following comment on a post by Mark Liberman, "Cell phone cupertinos":
I'm pretty sure I saw something several years ago about a whole dialect (argot? jargon? slang?) that had developed among young people in Japan (or possibly some other Asian country), based on phone cupertinos. Basically, they used the first suggestion from the autocomplete function *instead* of the original target word, to create an argot that was reasonably opaque to outsiders.
Now that comment has been brought back from the dead, appearing in two different articles about autocorrect.
First up was "The Fasinatng … Frustrating … Fascinating History of Autocorrect" by Gideon Lewis-Kraus (Wired, 7/22/14):
Today the influence of autocorrect is everywhere: A commenter on the Language Log blog recently mentioned hearing of an entire dialect in Asia based on phone cupertinos, where teens used the first suggestion from autocomplete instead of their chosen word, thus creating a slang that others couldn't decode.
Anna North then quoted this passage in her 9/4/14 New York Times Op-Talk piece, "Will the New Autocorrect Steal Your Soul?", bringing the factoid to an even wider audience.
Intrigued, I contacted Doctor Science to see if I could stir up anything more than a vague memory of this "dialect in Asia." The good Doctor then published a puzzled post on Obsidian Wings, "Help me Snopes myself!" Since the Language Log comment was hardly "recent," despite what Lewis-Kraus said, we're dealing with misty recollections:
It's clearly a good story — not only was it picked up at Wired and the NY Times, but when I discussed this at supper last night my family knew what I was talking about, because it's been mentioned in conversation around here. But at this point I honestly have no idea whether it's true, or something my mind put together and told me was true.
I remember making the comment at Language Log, enough to have talked about it to my family over the years. I remember trying to find a link while writing the comment, and gave up when I couldn't come up with one quickly. I have a mental image of the screen where I read it, and a sense that there was an image of people sitting on a train or subway, each looking at their phone — but that illustration might have just been provided by my brain, I don't know. The text might have been about Korea instead of Japan, and I have the persistent feeling it was from the NY Times, even though I'm pretty sure now that it wasn't.
After floating some suggestions about how the seed of the comment was planted, Doctor Science is left to wonder, "Did I make something up, not realize it was fiction, then spread it as fact? Or was I actually talking vaguely about something I had truly heard elsewhere?" If any Language Log readers can help out, please let us know in the comments below!
Max Pinton said,
September 8, 2014 @ 4:10 pm
What seems unlikely about this idea is that autocomplete is going to be correct in many cases, so it wouldn't be terribly opaque to outsiders. It also seems like it would vary from user to user depending on the device used.
However, it does remind me of Gyaru-moji, which is plenty opaque and basically uses non-Japanese glyphs from all over the Unicode spectrum to piece together Japanese characters. It can be extremely bizarre.
Jim said,
September 8, 2014 @ 4:27 pm
I love this story. It's absurd, but believable.
Ex Tex said,
September 8, 2014 @ 4:30 pm
This reminded me of Cockney rhyming slang, so it seemed perfectly plausible.
lolPhonology said,
September 8, 2014 @ 4:46 pm
Before Qwerty keyboards became widespread on smart phones, flip phones used T9 autocomplete software to facilitate texting on numeric keypads, using the mapping of 2 -> ABC, 3 -> DEF etc. T9 would make its best guess from your numeric sequence, so for example it would interpret 262 first as "bob", and if you wanted "cob" or "boo", you would need to use the up/down keys to cycle through your less likely choices. By coincidence, "cool" and "book" are both keyed as 2665 in this system, so "that's so book" is a plausible autocomplete error for someone trying to text "that's so cool". I had students tell me that (for a short time at least) they would use "book" intentionally in place of "cool", both in text and speech. That was 2008 (so the time frame fits), but in California, and with a single lexical item (not an entire dialect or argot).
Ethan said,
September 8, 2014 @ 6:18 pm
Some of us are so un-book that our phones still work like that.
Matt said,
September 8, 2014 @ 7:33 pm
There is definitely a similar thing people do in Japan, but it's more about character puns in the input method than autocorrect.
For example, for "go!", I can type "ike!", and when I hit space the input method will initially give me either "行け！” (go!) or "池！” (pond!) depending on what I typed beforehand. On message boards people deliberately leave the "wrong" selection as a bit of a joke, because everyone has the same input problems.
Input methods are getting much better at contextual word selection than they used to be, so initially incorrect choices are increasingly rare. I'd say that many uses of "wrong" kanji are deliberate selections these days.
Another example is kaeru -> 帰る(go home)/蛙(frog). Modern IM apps like LINE give you a little frog icon when you type "kaeru", so it's just too tempting and amusing to leave it as "I'll (frog) at 7pm".
Max Pinton said,
September 8, 2014 @ 7:43 pm
I love the punning, but I thought a key element of original claim was about a "relatively opaque" argot used by young people. Kaeru as a frog is clever but no (Japanese) parent would be stumped by that.
Matt said,
September 8, 2014 @ 7:52 pm
@Max agreed, especially for the simple examples I gave it doesn't live up to the original claim – but it may have been the seed of it. I have seen more difficult to decipher examples, but can't recall them off the top of my head.
Faldone said,
September 8, 2014 @ 8:03 pm
I visit a forum that used to have a spell checker that would simply pick the next word in alphabetical order if the word you input was not in its dictionary. We contemplated a similar "code" but never carried it out.
John Chew said,
September 8, 2014 @ 8:51 pm
The main effect that my iPhone is having on my language is to train me to pronounce the names of my friends and family the way that its text-to-voice algorithm does, if I want it to recognize their names when I want to phone them. The Japanese names tend to be the worst, but even my wife Kristen's name is understood better by the phone if pronounced with a clear /ɛ/ rather than a more reduced vowel.
Janne said,
September 8, 2014 @ 10:14 pm
Agree with @matt abobe; it sounds very much like the Japanese idea of selecting the wrong homophone when typing. I've never heard of it described as an actual slang or argot, but it's not uncommon as an occasional in-joke. You can get a whole email all with the wrong word selections. Looking at it it's completely incomprehensible, but if you read it out to yourself it makes sense.
AntC said,
September 9, 2014 @ 12:42 am
This kind of idea long predates electronic input methods.
There's a text [**] written (purportedly) by someone who's got a bargain second-hand typewriter. One or two keys are 'stuck', but it's easy to type one instead of el and 9 in place of p. As the story proceeds, more and more letters get stuck, leading to more and more substitutions. The last sentence is all substitutions, but understandable because we've got there one-by-one.
[**] My (unreliable) memory suggests I first saw it on a DEC PDP-11 bulletin board, late 1970's. A quick google isn't finding it.
maidhc said,
September 9, 2014 @ 3:46 am
AntC: The Two Ronnies television show from back in the 1970s used to do sketches about newsreaders dealing with typewriters with stuck keys. Funny stuff. I suppose you could find it on YouTube.
jan said,
September 9, 2014 @ 3:54 am
There is a short British documentary on the issue: http://www.youtube.com/watch?v=6hcoT6yxFoU
jo lumley said,
September 9, 2014 @ 5:38 am
Sorry, I can't help with the sleuthing, but wanted to add that (at least if this is really about Japanese) I find Matt's suggestion very much more plausible than the possibility of literally "use[ing] the first suggestion from the autocomplete function".
For the "first suggestion" substitution to work, it would probably have to be the case that all users are given the same autocomplete suggestion based on input (the first one or two kana in a word). Otherwise it would be a very difficult even for the intended recipient to decipher.
Even in the mid-2000s, which is the first time I used a Japanese cellphone with great frequency, the autocomplete supplied words based on frequencies of what the user usually writes, so two separate people's phones would not give the same suggestions. Indeed, autocomplete suggestions can be quite revelatory of one's personal preoccupations (e.g. does typing ka first give you kareshi 'boyfriend', or kanojo 'girlfriend', or karee 'curry', etc.).
Brian T said,
September 9, 2014 @ 7:30 am
Hey, even if this idea turns out to have been a figment of somebody's imagination THEN, we can still make it start being true NOW, folks.
Hey, even I this I'd turns I to have been a figment of somebody I'm Thelma, we Caroline at make I at being tri NOW, do.
Q. Pheevr said,
September 9, 2014 @ 8:27 am
このアーガイルは有能非圧縮性である。
BZ said,
September 9, 2014 @ 9:13 am
Not sure how this would work. Surely the same autocomplete would appear for multiple intended words, so it's a one-way encoding that can at best be used to verify text that one already possesses.
SamC said,
September 9, 2014 @ 9:53 am
@lolPhonology – That's exactly what I thought of. I was in high school 2002-2006, & my old phone's T9 substituted "Woohoo!" with "Zonino!" It then became an in-joke with my friends, and we'd say "Zonino!" when something was exciting or cool.
So it probably functions best as an in-joke between a small group of people than recognizable slang.
Jongseong Park said,
September 9, 2014 @ 11:20 am
Doctor Science wonders if the story could have been from Korea. I doubt it, not just because I have not heard of this before—I'm not book enough to claim to keep abreast of all the latest slang in Korea anyway.
Rather, I have a hard time picturing this happening in Korea because Korean phone cupertinos are not really a thing. Predictive text input technology is virtually non-existent in Korea for two major reasons. One, the featural aspect of the Korean alphabet makes for fairly intuitive groupings of letters to distribute on the keypad, e.g. the button 1 might stand for ㄱ ㅋ ㄲ (the velar stop series) instead of ABC (following the arbitrary alphabetical order), which also makes it more economical because the more frequent letters tend to require less strokes. Inputting text without predictive text is far less cumbersome for Korean than it is for the roman alphabet. Two, the agglutinative nature of Korean with its endless particles and verb endings makes it fiendishly difficult to set up a database for predictive text.
I haven't seen predictive text attempted for Korean, but even if it existed, the type of slang described would be fairly easy to decode. Especially given the context, using 끼다 kkida for 기타 gita for example wouldn't exactly be impenetrable.
Chris Waters said,
September 9, 2014 @ 1:13 pm
One thing makes me dubious about the story, but I'm not familiar enough with the technology to know if it's a reasonable objection or not. It seems like this would depend on the phones all giving the same correction/option each time. Which would, in turn, depend on them all basically using the same software or software with identical functionality. Which seems to me, as a software developer, fairly implausible, but maybe for Japanese autocompletion, it's really a thing? Do all Japanese phones offer exactly the same choices every time, no matter who the manufacturer/developer?
Walter Burley said,
September 9, 2014 @ 2:29 pm
From David Kahn’s The Codebreakers:
O. Henry penned a sardonically amusing story [in which a newspaper correspondent named Calloway who is covering the 1904-05 Russo-Japanese war] gets a scoop past the censor's eye by using the first half of a newspaper cliché as the codeword for the second half, which forms the plaintext. Thus FOREGONE meant conclusion; DARK, horse; BRUTE, force; BEGGARS, description [and so on]. And—sad to say—the journalists in New York understand it.
http://www.online-literature.com/o_henry/1006/ has the complete text of the story.
Kenny said,
September 9, 2014 @ 6:55 pm
I know that people my age have started bowlderizing the word "fucking" as "ducking" and "shit" as "shot" because that's what the iPhone always tries to correct misspellings of those words to.
pj said,
September 10, 2014 @ 1:16 pm
I seem to remember hearing Stephen Fry mention the "'book'-for-'cool'" thing in a fairly early QI. I can't on a quick search turn up video of any such episode, but here are 'textonyms' being discussed on a QI forum in September 2007, including the (vague) claim that
And here's a Daily Telegraph article from February 2008 reporting 'Predictive text creating secret teen language' – but without any corroborating evidence as far as I can see.
Derwin McGeary said,
September 11, 2014 @ 12:00 pm
I have a cite!
The Streets – Memento Mori, end of verse two.
"I don't really care about the luck and the look
But driving a Ferrari is fucking book"
( http://rap.genius.com/The-streets-memento-mori-lyrics )
In context, it clearly means cool.
Yuval said,
September 13, 2014 @ 1:37 pm
We've looked into the "that's so book" myth and found no evidence for it, as you can read on the Labs' blog.
Chris Kern said,
September 13, 2014 @ 9:21 pm
I'm inclined to believe that this story originated with what Matt is talking about, garbled by people with a poor understanding of the Japanese writing system and the usual "telephone game" manner that stories spread.
There's a good deal of Internet slang that's based on conversion idiosyncrasies. Ones I can immediately think of are 乙, read as "otsu", to represent "otsukaresama" (basically "thanks for doing that" or a similar feeling), うｐ for "upload" (because if you type "up" in the Japanese IME and hit space you get the previous), おｋ for "OK" (same reasoning), and using 厨房 or just 廚 to refer to someone annoying (originating from a similarly pronounced word meaning a middle school student). There's also ようつべ for "Youtube" (just typing "youtube" and hitting the conversion key to get the hiragana) and うぽつ for "up otsu[kare-sama], i.e. "thanks for uploading"".