Autofail: How Apple’s Autocorrect Teaches Bad English

Robert Stribley   October 7, 2011

Lost in translation. (image via benjamin.krause)

The Breakdown: Apple just released the iPhone 4S, which incorporates voice recognition to intelligently interpret your voice commands. Robert Stribley explains how one of the iPhone’s existing features, however, ain’t so genius.

This week, Apple released the iPhone 4S, which promises, via the wonder of Siri technology, to respond intelligently to voice commands. The innovation may turn out to be ground-breaking, but it was greeted with somewhat muted applause, as Apple’s well-trained audience had been expecting the advent of the iPhone 5. Probably unfairly, the 4S ended up sounding like a way station on the road to the big event. Still, it’ll be interesting to see whether the adoption of Siri’s intelligent assistant feature can mitigate one of the iPhone’s most intermittently annoying features: Autocorrect. Probably not.

I’m hardly the first to notice that Apple’s Autocorrect feature often fails to live up to its name. Many have noted that the program actually proves pretty poor at correcting your spelling, sometimes even inserting an embarrassing substitute for what you intended. There’s a popular blog, which capitalizes on the more amusing instances of this behavior. What I haven’t seen is anyone articulate all the different ways Autocorrect actually performs abysmally. It’s not just that it corrects poorly: In fact, it fails in three key areas. And it fails in ways that arguably teach its users bad English. As a public service then, allow me to codify the ways in which Autocorrect fails.

Autocorrect Substitutes Misspelled Words with Words Which Makes No Sense

The best-known issue with Autocorrect is its sometimes comical tendency to replace misspelled words with something that makes little or no sense or to create a new meaning the writer didn’t intend. The reason this frustrates people so much is that the misspelling is often not too far from the correct spelling. Yet, Autocorrect often manages to suggest something completely different.

Therefore, “making brownies” becomes “making babies,” “Disney” becomes “divorce,” “sinus infection” becomes “dinosaur infection,” and much hilarity and/or awkwardness ensures. And these are some of the tamer examples, you understand.

Autocorrect supposedly works by analyzing the keys near the ones you actually selected to estimate which ones you intended to select. Then it replaces your word with its best guess based on those letters nearby. In order to be improved, Autocorrect would have to incorporate some higher order artificial intelligence, some fuzzy logic, so it would recognize when a word it wants to substitute seems absurd or inappropriate in a particular context. It would need to base its intelligence upon a nuanced, contextual understanding of language, instead of the much more limited contextual understanding of the layout of a Qwerty keyboard. If Apple is able to apply Siri’s semantic capabilities to texting, they could make some great strides in this area.

Autocorrect Teaches Bad Spelling & Punctuation

Yes, perhaps most infuriating of all, Autocorrect sometimes corrects words which are spelled correctly, actually rendering them incorrectly. For instance, if you type “Being simplistic is its problem,” autocorrect will change “its” to “it’s,” which means your sentence now technically reads “Being simplistic is it is problem.” Many college graduates have difficulty distinguishing between these two spellings (one a possessive pronoun, one a contraction) as it is. Now Autocorrect is drumming the exact wrong spelling further into their craniums.

Another example, and this issue is probably often overlooked, but when you use an ellipsis in any sentence Autocorrect automatically capitalizes the first letter of the next word, apparently assuming the last of the three marks to be a period. However, a capital letter does not necessarily have to follow an ellipsis. Ellipses are employed to show that words have been omitted – words, not necessarily sentences – and also, perhaps informally, to show the passage of time. Since Autocorrect cannot know if a capital letter needs to follow an ellipsis, it shouldn’t automatically create one. Otherwise, it may be teaching bad punctuation and capitalization.

Autocorrect Bowlderizes Your Writing

Perhaps it’s inadvertent, but Autocorrect also appears to censor or Bowlderize your writing on occasion. If I write that “Autocorrect doesn’t do a hell of a good job,” the program renders “hell” as “he’ll.”  Oddly enough, however, it doesn’t correct “helluva” as in “helluva good job.” That indicates that if you type in “hell,” Autocorrect assumes you probably meant “he’ll.” That correction, however, takes about six keystrokes to undo. Of course, it’s only one extra touch to dismiss the little bubble that comes up suggesting “he’ll” if you happen to see it, which I seldom do. But the point is, “hell” ain’t a misspelling, so Autocorrect shouldn’t correct it. Think I’m being picky? That, chances are, Autocorrect is generally correct on that one? Well, Autocorrect doesn’t change “shell” to “she’ll,” so it’s not even consistent. And I’m far more likely to say “hell” on any given day than “shell.” But that’s just me.

Additionally, there’s at least two other words Autocorrect doesn’t recognize as spelled correctly despite their being canonical English curse words. I mean, it’s hard to misspell a four letter word. Yet Autocorrect fails to even recognize these words and primly places a neat red line beneath them. Since this is a family publication, I’ll not post the words here. I’ll let you discover them on your own. Of course, I only discovered them myself while thoroughly researching this article.

Let’s also note that Autocorrect highlighted the word “Bowlderizes” above as a misspelling. Leave my writing the he’ll alone, Autocorrect!

Postscript 10/07/11: With the utmost sincerity: Rest in Peace, Steve Jobs, Visionary


6 Responses

  1. lorinda says:

    So is it Autocorrect we should blame for making it a weigh station and not a way station? ;)

  2. I remember when Tom Gruber gave a keynote about Siri at the Semantic Technology Conference in 2009. That was before it was bought by Apple. I wonder how much of the semantic capabilities Apple will will end up using. It would be great if they integrate it throughout the OS, rather than just using the voice recognition aspect.

    Here’s a video of Gruber’s presentation:

  3. Good point, Lorinda, and my goof. We’ve fixed it now. Thanks for pointing it out. Though I can’t blame Apple for my mistake there, I was using an iPad to write this, which also incorporates Autocorrect. Maybe, eventually, the feature will be sophisticated enough to point out my misuse of a homophone like that, via a semantic understanding of the sentence. :)

  4. Katherine says:

    I’m not defending Autocorrect; I’m familiar with its maddening flaws, and I really enjoyed your article. But the reason it highlighted “Bowlderizes” as a misspelling is that it is one. The word is “bowdlerize.” (It needn’t be capitalized; the word’s meaning beyond direct reference to Thomas Bowdler and his editorial (mis)deeds is long- and well-established.)

    Out-of-work editor
    See my forthcoming book, “Autobiography of a Pedant.” ;)

  5. Ah-hah! Good point, Katherine and nice editing chops! “Autocorrect bowdlerizes” works.

  6. Cody says:

    I am on an iPad. I have had issues writing my book. Here in indiana we use well a lot. It keeps trying to correct it to we’ll. I have played around with the keyboard settings and no luck. Maybe apple should get learnt from Microsoft on how to do it. Microsoft word ftw! We’ll leave my words the he’ll alone iPad!!!

Leave a Reply

Razorfish Blogs


  • SXSW Interactive

    March 7 – 11, Austin, TX
    Several of our contributors will be speaking this year. If you’re going, say hi to Rachel, Robert, & Hawk.

  • Confab Minneapolis

    May 7-9, Minneapolis, MN
    The original Confab Event. Rachel will be there doing her Content Modelling workshop with Cleve Gibbon. Get details and we’ll see you there!

  • Intelligent Content Conference Life Sciences & Healthcare

    May 8-9, San Francisco, CA
    Call for Presenters, now open:

  • Confab for Nonprofits

    Jun 16, Chicago, IL
    Another new Confab Event! Early Bird pricing until March 7:

  • Content Strategy Forum

    July 1-3, Frankfurt, Germany
    International Content Strategy workshops & conference: Call for speakers now open!

Search scatter/gather

What is this site, exactly?

Scatter/Gather is a blog about the intersection of content strategy, pop culture and human behavior. Contributors are all practicing Content Strategists at the offices of Razorfish, an international digital design agency.

This blog reflects the views of the individual contributors and not necessarily the views of Razorfish.

What is content strategy?

Oooh, the elevator pitch. Here we go: There is content on the web. You love it. Or you do not love it. Either way, it is out there, and it is growing. Content strategy encompasses the discovery, ideation, implementation and maintenance of all types of digital content—links, tags, metadata, video, whatever. Ultimately, we work closely with information architects and creative types to craft delicious, usable web experiences for our clients.

Why "scatter/gather"?

It’s an iterative data clustering operation that’s designed to enable rich browsing capabilities. “Data clustering” seems rather awesome and relevant to our quest, plus we thought the phrase just sounded really cool.

Privacy Policy | Entries (RSS) |     © Razorfish™ LLC All rights reserved. Company Logo.