Cell phone cupertinos

« previous post | next post »

Reader JH's wife texted from the playground

She's so tired though… may come home Zionist

This was not an example of the role of fatigue in political identity formation, but rather a cupertino, created when her iPhone helpfully corrected (some spelling of) "soonest" to "Zionist".

SMS messaging and cellphone email must be a rich source of cupertinos, since autocompletion and spelling correction are always (?) on, and the input methods are very error-prone.


  1. Oskar said,

    August 23, 2010 @ 8:25 am

    This is an especially bothersome problem if you use your phone to communicate in two different languages. I've configured my iPhone so that it has both a Swedish keyboard and an English keyboard, but because the keyboards are so similar (the only difference is that the Swedish keyboard has extra keys for "å", "ä" and "ö"), I often forget to switch between them. But of course, it's not just the keyboards who are different, they use different auto-correcting dictionaries, so when I accidentally start typing an English sentence on the Swedish keyboard, it comes out as a non-sensical string of Swedish words which my phone thinks I intended. It's extremely annoying.

  2. Alex said,

    August 23, 2010 @ 8:31 am

    A while back I read that there was a short ironic trend to use "book" as a substitute for "cool", because the T9 system cupertinoed the 2-6-6-5 texting combination in that way.

  3. Jarek Weckwerth said,

    August 23, 2010 @ 8:45 am

    I'm a die-hard conservative on this one. No autocomplete, thank you.

    The second mobile I owned, years ago, had really useless autocomplete. I ended up switching it off. Same for the next one. Then, the first thing I'd do on a new phone would be switching it off. (So, yeah, I actually don't really know if the systems work any better now…)

    A hardware qwerty keyboard is far better, and not that error-prone at all.

    Also, like Oscar above, I tend to text in both English and Polish. Switching settings etc. would be a chore.

    Plus, at least in Polish, autocomplete gives you the words in full. With Polish words as long as they usually are, you end up going over the 160 character limit more often than you really need to. (I wonder what German speakers do…)

    PS Makes me think of this sketch, which I first saw on Language Log:

    Youtube link

    That's how I learned about Armstrong and Miller. Not a bad show at all.

  4. Ginger Yellow said,

    August 23, 2010 @ 8:46 am

    My favourite is "pint", which (on Nokia phones anyway) is usually autocorrected to "riot". Which can lead to friendly invitations to the pub becoming incitement to violence.

  5. Colin John said,

    August 23, 2010 @ 8:51 am

    I get "shot" for "pint" on my Nokia – so it's still off down the pub!

  6. Amy Reynaldo said,

    August 23, 2010 @ 9:24 am

    On my Droid phone, there's no autocomplete when I use the real QWERTY keyboard. The autofixing appears limited to capitalizing the first letter of a sentence, capitalizing I/I've/I'm/I'd, and interpreting two spaces as "please insert a period and capitalize the word at the start of the next sentence" (alas, my two spaces represent a typo).

    When using the Droid's virtual QWERTY keyboard, typing the second letter of a word (and subsequent letters) generates a list of likely words, but if you keep typing, I think it leaves whatever you've typed.

    I don't know why Apple programmed the iPhone to autocomplete rather than display choices. Those cupertinos are far too common.

  7. Leonardo Boiko said,

    August 23, 2010 @ 9:41 am

    I’ve once called someone’s children “bolinhos” (pastries) instead of “anjinhos” (little angels), thanks to T9.

  8. Cameron said,

    August 23, 2010 @ 9:46 am

    I've been using the word "texto" to describe errors of this sort for years now.

  9. Laura said,

    August 23, 2010 @ 9:58 am

    People did used to use 'book' for 'cool', as Alex says, until some genius had the bright idea of making 'cool' the first suggestion. It only took them about five years. Perhaps they were hoping to get people to read more books, but gave in?

  10. Jen said,

    August 23, 2010 @ 10:19 am

    This looks like it may have been an attempt at "soonish" rather than "soonest" to me.

  11. richard said,

    August 23, 2010 @ 10:25 am

    Pretty sure it would be "soonish" that got autocorrected to "Zionist". At least, I would use soonish in that situation before anything like soonest.

  12. Adrian Bailey said,

    August 23, 2010 @ 10:27 am

    "may come home soonest"? What's that supposed to mean?

  13. Shrikant said,

    August 23, 2010 @ 10:33 am

    My personal favourite is (again, on a Nokia using T9) – 735328.

    Same combination for both 'select' and 'reject'. That is just so awesome. A friend and I had made a list of such 'T9 koans' someplace. Offline sadly; that list remains forever lost, unless we can dredge up the enthu to go at it once more.

  14. Karen said,

    August 23, 2010 @ 10:39 am

    "soonest" means "as soon as possible".

  15. Faldone said,

    August 23, 2010 @ 10:44 am

    Re: soonest = as soon as possible.

    Probably it dates bake to early telegraphy days at least.

    This has been provided as a public service announcement by the Recency Illusion Council.

  16. Faldone said,

    August 23, 2010 @ 10:44 am

    For "bake" read "back"

  17. Doctor Science said,

    August 23, 2010 @ 10:44 am

    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.

  18. Ceiswyn said,

    August 23, 2010 @ 11:08 am

    That would be precisely the reason my friend and I say 'farmhu' when we mean 'darnit' :)

  19. Mfahie said,

    August 23, 2010 @ 11:12 am

    The favourite one that I've read is "eating human for the first time! New Orleans is awesome!!!"
    5 points to the first person who gets the original meal.

  20. Ginger Yellow said,

    August 23, 2010 @ 11:20 am

    The OED has a citation for "soonest" in this sense going back to 1225, so it's pretty old. It does reek of the telegraph age though. I could have sworn it features in Scoop, but Amazon's "search in book" feature says no.

  21. Robert Coren said,

    August 23, 2010 @ 11:25 am

    @Mfahie: Got to be "gumbo", no?

  22. Mfahie said,

    August 23, 2010 @ 11:36 am

    You got it Robert. (for difficulty I should have left out the second sentence). 5 points nevertheless!

  23. h. s. gudnason said,

    August 23, 2010 @ 11:40 am

    Several years ago I tried to text someone about an interesting article in The New Yorker, which my phone insisted on rendering as The New Worker. Much as that might have suited my political leanings, it didn't convey my intended meaning.

    I finally gave up in disgust and decided to leave the article unrecommended.

  24. Jerry Anning said,

    August 23, 2010 @ 11:51 am

    Then there is the case of 'woohoo' Web cartoonist David Willis typoed it as 'wiigii' in a forum post and the term caught on in certain segments of the webcomic community. Early texting spellcheckers unaccountably converted 'woohoo' to 'zonino', which also had a hip vogue and sparked a campaign to get it included in the OED.

  25. Ben Zimmer said,

    August 23, 2010 @ 12:00 pm

    Words using the same key combination in predictive text have been called "textonyms." Earliest cite I've seen is from Nov. 27, 2004 in The New Scientist, credited to Meg Kingston. In his book Txtng, David Crystal also refers to "homonumeric words," but I haven't seen that elsewhere.

  26. mtngirlrunning said,

    August 23, 2010 @ 12:00 pm

    My best one: I was texting my Search and Rescue group to remind them of the upcoming "Mega Evaluation" required for individual re-certifications. As the space on our messaging system is quite limited, I abbreviated it "MEGA EVAL" only to find that I had sent my entire team a message reminding them of the "MEGA FUCK" on Saturday! Gotta love T9…

  27. Cavity Lee said,

    August 23, 2010 @ 12:51 pm

    The best thing about iPhone autocomplete is that it considers sending a message an implicit 'accept this suggestion', the same as pressing space or full stop. Or at least, it used to.

    Shortly after I got an iPhone, I was trying to send my partner a text that said "Remind me how you made those great mashed potatoes." But I totally bungled the first word, and managed to hit Send instead of backspace, as a result of which she got an SMS reading only


  28. Josh said,

    August 23, 2010 @ 1:15 pm

    I sent a friend a text to wish her a happy birthday, and give her a hard time about getting older. I ended it with a "haha" to make it clear I was just joking. My phone corrected it to "Hagar", so instead of sending her a good natured ribbing, I told her she was getting old and then called her an unattractive boy's name.

  29. Sili said,

    August 23, 2010 @ 1:19 pm

    Isn't modern autocomplete/-suggest supposed to be trainable? I'm sure I heard of someone doing adaptive dictionaries a coupla years ago that were supposed to be context aware – working on a word by word basis as well as letter by letter (Markov chains?).

    Don't have a smartphone, myself, though. Do have the option of switching between Danish and English dictionaries.

  30. Jerzy said,

    August 23, 2010 @ 1:40 pm

    I love how my phone will suggest "beastcake" and "beartable" before it actually gets to suggesting "adaptable."

  31. groki said,

    August 23, 2010 @ 2:12 pm

    Jarek Weckwerth Youtube link:
    thanks. "…and a weirdly extensive vocabulary"!

    Cameron "texto":
    I call mis-hearings "earpos" and brain hiccups "thinkos."

    Cavity Lee demons:
    it's just those djinns of the cell phone, giving themselves a shout-out.

  32. Jeff R. said,

    August 23, 2010 @ 3:31 pm

    My own favorite of these is "daughter/fatigues"…

  33. ed k said,

    August 23, 2010 @ 4:46 pm

    Cell phone dictionaries insist that when I attempt to type "ovaries" I actually mean "nubsids". This one really confuses me, since the former is an unremarkable word while the latter doesn't seem to exist at all.

  34. Xmun said,

    August 23, 2010 @ 4:58 pm

    @Ginger Yellow
    Probably you're thinking of Chapter 5, Section 2, in Scoop:

  35. Roger Lustig said,

    August 23, 2010 @ 9:40 pm

    @Josh: Hagar was a boy? Makes the story even stranger.

  36. Stuporman said,

    August 23, 2010 @ 11:00 pm

    My phone will say "Goodmight" and "Homenight" before getting to "Goodnight".

  37. David said,

    August 24, 2010 @ 3:34 am

    In T9 "Smirnoff" = "poisoned"

    My Android phone doesn't have "parsnip" in its dictionary, so it suggests "paramilitary"

  38. bkd69 said,

    August 24, 2010 @ 3:57 am

    I was amused to see that my Droid's autocomplete dictionary contained Lovecraft, and later I learned it didn't have Farnsworth. Which leads me to think there's probably an interesting paper to be had here, 'Cultural Assumptions Found in the Proper Names Contained in Text Autocompletion Dictionaries.'

  39. Samuel Baldwin said,

    August 24, 2010 @ 12:32 pm

    Recently my friend sent me a jabber message on his iPhone which auto-corrected 'melee' (the accepted Angilicisation of the French mêlée) to 'melée', wrong in both languages, as far as I can make out.

  40. Toby said,

    August 24, 2010 @ 2:38 pm

    A friend put together a Nokia Text Map of Britain. It still makes me chuckle. http://40k.org.uk/graphics/textmap.gif

  41. Toby said,

    August 24, 2010 @ 2:42 pm

    I should add that we regularly go walking in the Yorkshire Dales and we still refer to Horton-in-Ribblesdale as Impuno-in-Shackereble — now so frequently that I occasionally forget which is the real place name and which is the texto.

  42. Clarissa at Talk to the Clouds said,

    August 24, 2010 @ 4:21 pm

    @Cavity Lee
    That made me wheeze with laughter. What an alarming and peculiar message to get!

    The iOS system that iPhone and iPod users encounter is not a spellcheck, but a proximity-error (typo) correction service. It only corrects (?) things based on the keys next to them, so it'll correct "ronfyw" "to "tongue," but it won't correct "tonuge" or "tounge." It works pretty well for me because I rarely make spelling errors, but it drives my Japanese friend up the wall when he's trying to type in English. He's only been a frequent English user for about four years, and he makes a lot of spelling errors (understandably). iOS doesn't deal well with that.

    iOS does learn, so if you always click the X in the "Zionist" suggestion to ignore it, it will allow and eventually learn to fill in "soonish" once you've typed "sooni" (mine now fills in things like "Aaaaah!" and "sociolinguistically" and "omgwtfbbq" and "okonomiyaki"). So yes, it is trainable. iOS4 will let you edit the dictionary, apparently, but I haven't bothered to upgrade. It would be useful, though, because it's possible to accidentally train it into misspellings or other errors.

    Still, if you get ominous or peculiar or downright insulting messages, it's good to pause and ask yourself if some sort of "correction" might be at work.

  43. Kragen Javier Sitaker said,

    August 24, 2010 @ 5:03 pm

    The redundancy in natural language both enables error detection and slows text entry on cellphones. The Android and iPhone software allows you to type faster on smaller keyboards by exploiting some of that error-detection capability to reject implausible typos and search for nearby plausible typing streams, much as we humans do when we listen to each other talk. OCR software does the same thing, as do people reading handwriting.

    Taken to its natural limit, of course, such software would convert any possible input sequence into a valid, sensible English sentence. The consequence would be that fewer and fewer typographical errors would be recognizable as such by a human recipient; a sort of Turing-Asymptotic Cupertino Trend will result. The resulting pervasive uncertainty about reality will reveal that Timothy Leary was right when he said, "[O]ur research with psychedelic drugs … was a forecast of[,] or preparation for[,] the personal computer age." Software is psychedelic. Deeply groovy, man.

    On a less speculative note, I think errors like "nubsids" for "ovaries" are the result of phone software with a small dictionary that contains no "ovaries", and indeed no other words that begin with "[mno][tuv][abc][pqrs][hij]" (the only others in Google's teraword corpus stats are "ovarian", "ovariectomized", "ovariectomy", and "otari".) You need about 30 000 words in your dictionary in order of use frequency (in the British National Corpus) before it contains "ovaries".

    So, having guessed "nubs" and received "[hij]", the software might fall back, perhaps, on letter-level models: "si" is roughly twice as common as "sh", and "sj" is around sixty times rarer. ("ri" is more common, but I haven't run the Viterbi algorithm to figure out if "nubsi" is in fact the most plausible alternative among the 324 possibilities based on digram frequencies. My own (horrible Motorola) phone, in these situations, seems to take its guess from before the current letter as given, rather than trying to update it.) However, this hypothesis predicts that "ovaries" would come out as "nubsier", not "nubsids", as does the simpler hypothesis that it might use simple letter frequencies.

  44. Rodger C said,

    August 24, 2010 @ 6:30 pm

    @Cavity Lee: Well, obviously your phone was programmed by Lovecraft.

  45. Killer said,

    August 24, 2010 @ 6:38 pm

    I typed "Hope you're having a good day" without looking till I was done, at which point it read "Hope you're hating a home day." I miss the days when the people whom you'd wish a good day already knew you felt that way, and you didn't have to text it.

    P.S. @Cavity Lee's story takes the cake.

  46. Urbane Legend said,

    August 26, 2010 @ 5:18 am

    I kind of love these cupertinos. My friends and I use "huspag" instead of "hurray". Last night my partner texted me to let me know what was for dinner, and I asked if he could pick up some salad to go with it. His iPhone corrected "OK" to "okra". Would have been a very different meal..

  47. dan said,

    August 26, 2010 @ 6:07 am

    While texting my other half to inform her that we were stocked up with refreshing beverages to numb the pain of another dreadful England world cup football performance, I mistakenly informed her that I had obtained some "rear cheer" for her rather than the "pear cider" I had actually intended.
    Suspecting me of being a pervert, she was a little cautious when she returned home. Oh how we laughed etc.

  48. George said,

    August 26, 2010 @ 9:33 am

    It might be worth noting that the maker of the iPhone is located in – Cupertino, Cal.

    (But, this wouldn't explain Nokia, Droid and others.)

  49. seriously said,

    August 26, 2010 @ 9:43 am

    Why is everyone assuming it's a cupertino? Perhaps the girl ("she") had been a lukewarm supporter of Israel when she went to the playground, but a spirited debate with some radical Palestinians who were already playing there (perhaps a disagreement on the "two sliding-board solution") had strengthened her views.

  50. Bluebird said,

    August 30, 2010 @ 11:42 pm

    Ordinary spellcheckers still come up with this sort of thing: my favorite was my pastor's computer changing "little-faithed" to "little fathead".

  51. Rodger C said,

    September 1, 2010 @ 9:32 pm

    Anne Shelby has a poem about spellcheckers' ignorance of Appalachia: "I ordered soupbeans and it served me subpoenas." (And sure enough, I see a red line under "soupbeans.")

RSS feed for comments on this post