Monday 3 June 2013

Cupertino effect - the dangers of spellcheckers and autocorrect

Six months ago I learned that misheard lyrics are known as mondegreens and wrote a blog posting on the subject. This week, listening to an archived podcast of the wonderful RadioLab, I learned that there is a term for the tendency of a spellchecker or autocorrect facility to come up with inappropriate words to replace words that are mis-spelled, or at least are not in its dictionary. This is the Cupertino effect.

The effect was named by writers and translators for the European Union who found that early spellcheckers could not recognize the word cooperation unless it was hyphenated. Instead they would routinely replace the word with Cupertino, the name of a Californian city which happens to be where Apple Inc is headquartered. A problem arose when an author would run an automated spellcheck on a document then fail to proof-read what the spellchecker may have 'corrected'. Even now you can come across archived official documents that contain strange phrases such as:

The Cupertino with our Italian comrades proved to be very fruitful.

South Asian Association for Regional Cupertino

...stimulating cross-border Cupertino.

Mistakes can also occur as a result of the spellchecker failing to correct a word because they recognize it from a different context. That's the premise of a lovely little spellchecker poem written a good few years ago now by Janet Minor, who describes herself as an 'internet poet'.

I have a spelling checker
It came with my PC;
It plainly marks four my revue
Mistakes I cannot sea.
I've run this poem threw it,
I'm sure your pleased too no,
Its letter perfect in it's weigh,
My checker tolled me sew.

Though spellcheckers tend to be a little more sophisticated these days it is still dangerously easy to get into difficulties if you are not careful about how your spellchecker corrects common words that you may have slightly mis-spelled. Examples I have seen quoted include:

definitely mis-spelled as definately corrected to defiantly

acquainted mis-spelled as aquainted corrected to aquatinted.

Foreign expressions and names can cause a problem for English language spellcheckers. A lawyer using the Latin phrase sua sponte ('of one's own accord') found the phrase corrected to sea sponge. A Reuters report referring to Pakistan's Muttahida Quami Movement was changed by a spellchecker to read Muttonhead Quail Movement.

One of the best examples of automated name changes I've seen comes from a student yearbook published by a high school in Middletown, Pennsylvania. The student register should have included these real names:

Max Zupanovic
Kathy Carbaugh
Alessandra Ippolito
William and Elizabeth Givler
Cameron Bendgen
Courtney and Kayla Hrobak

But those students would have looked in vain for their names. The spellchecker had 'corrected' them to:

Max Supernova
Kathy Airbag
Alexandria Impolite
William and Elizabeth Giver
Cameron Bandage
Courtney and Kayla Throwback

Sometimes individuals and organizations create their own correction problems by customising their spellcheck software. To introduce the first example, let me ask you to make the connection between these two images:



Reuters of London's stylebook instructs its journalists reporting on the monarch of England always to use her full name 'Queen Elizabeth' rather than 'the Queen'. As a reminder or reinforcement Reuter's spellcheck software is customised to autocorrect to the default convention. A problem occurred with a news release in October 2006 about the genetic code of the honey bee. No-one spotted before publication that the article included the following odd phrases:

With its highly evolved social structure of tens of thousands of worker bees commanded by Queen Elizabeth, the honey bee genome could also improve the search for genes linked to social behaviour.

Queen Elizabeth has 10 times the lifespan of workers and lays up to 2,000 eggs a day.

Even more amusing, in my view, is the 2008 headline that read:

Homosexual eases into 100m final at Olympic trials

This was a delicious error perpetrated by the far right fundamentalist Christian group, the American Family Association (AFA).  Their website re-posts news of interest, but heavily censored for content. This group's autocorrect facility is set to alter the word 'gay' to 'homosexual', apparently because 'gay' is not a word they wish to see associated with a practice they regard as abhorrent.

The article in question, however, was about the American sprinter Tyson Gay, who easily won his semi-final at the Olympic trials. According to The AFA version of the article:

Tyson Homosexual was a blur in blue, sprinting 100 meters faster than anyone ever has... "It means a lot to me," the 25-year-old Homosexual said: "I'm glad my body could do it, because now I know I have it in me."

The news director later told a reporter: "We took the filter out for that word" after Tyson Homosexual surfaced on the site. "We don't object to the word gay except when it refers to people who practise a homosexual lifestyle."

That's all right, then.

I began by saying that the Cupertino effect as named was a problem for early spellcheckers, but the general danger still exists not only for PC users but perhaps especially for modern texters whose fingers sometimes work faster than their brains. Predictive text and Smartphones with dictionary supported keyboards that can automatically replace 'mistakes' may be a help or a hindrance. I'll leave you with a link to Damn You Autocorrect and its list of 25 Funniest Autocorrects. They are a caution.

Now I'd better proof-read this carefully before posting - you just can't trust those damn spellcheckers.






8 comments:

  1. My eyes catch more than the spellchecker, and my brain is smarter. And I definitely don't use find and replace, unless I'm monitoring the stupid computer every step of the way!

    Lauren

    ReplyDelete
  2. I agree, Lauren. I sometimes use it to check individual words but never the whole document and I avoid the Replace All option as if it were an unexploded bomb.

    ReplyDelete
  3. I used to always say to writers turn off the spellchecker. Since my stroke I've amended it to say, "turn off the auto correct" because I can't spell worth a hoot.

    ReplyDelete
  4. Thanks for the comment JL - and not a spelling error in sight.

    ReplyDelete
    Replies
    1. David that's because of the red squiggly lines and the google dictionary.

      Delete
  5. Oh, but don't you just hate those red squiggly lines? And the green ones are worse - this sentence, for example, would get the green treatment for starting with an 'And'; shocking. I turn em all off.

    ReplyDelete
  6. I'm very late to this party but only just saw this post.

    I wonder how many others working in education are, like me, repeatedly infuriated by Microsoft's insistence on correcting "Your A Levels..." to "You're a levels...".

    ReplyDelete
    Replies
    1. I know a good few of this blog's followers work in education so I imagine they are nodding their heads about that one.

      Delete