On because PIE
Text message from hapax to bewilderbeast, December 30, 2008

The mistakes that we make are governed, at least in part, by the technology we use to make them. The point is so obvious as to border on banal, but I still think it's worth considering for a moment.

Typos through the ages

The historian who works with manuscripts, that is to say handwritten documents from before the fifteenth century or so, must deal not only with damaged documents or documents in obscure dialects, but also with documents containing what could be called handwritten typos -- "write-o"s. We've all done it: dotted a U instead of an I, crossed an L instead of a T, looped a G's descender in the wrong direction to make a misplaced Q. Ancient writers were not immune to these problems; if anything, their handwriting was even more difficult to decipher than the worst modern doctor's.

In what palaeographers call minim script, letters that contain no ascenders or descenders can be very easily confused with one another: u, m, n, w, r; occasionally o and c; and even, in some hands, an undotted i can make certain words extremely difficult to tell apart. This is true in some modern typefaces as well; no doubt you've struggled to distinguish the important word porno from the trendy short form pomo. (... Or maybe it's just me that frequently uses both those words.)

In a typewritten or word-processed document, however, minim is not quite as much of a problem, and there is no natural way that a word with an L in it would be mistyped with a T. I don't know if I've ever seen a tore where there was supposed to be a lore, for example. The reason for this is that the T and L keys are nowhere near one another on the keyboard. Nobody, from the most gifted touch-typist down to the most incompetent hunt-and-pecker, would wing one of those keys when she meant to hit the other. Put another way, where the pen might slip, the typing finger might not. (Of course, there are some coincidental overlappings: the U and the I can be confused with one another in minim, and they are also next to one another on the standard QWERTY keyboard.)

What happens more often on a keyboard is that letters are transposed (hence the infamous teh, which has become so ubiquitous that it is now often used deliberately and ironically; I've even used it out loud), or adjacent keys are pressed by mistake (including not just letters, but numb3rs and caPSLOCK as well). As has been noded elsewhere, QWERTY is the most common keyboard layout in the English-speaking world, making many typos predictable and easy to decode, while the minority Dvorak keyboard brings with it its own distinctive brand of error.

The curious phenomenon of the SMS typo

For the first time since the development of the QWERTY keyboard, we have invented a new way to make typos: the cellphone keypad. Cellphone typos generally fall into two categories:

Number of keypresses. On English cellphones, each number key corresponds to three or four letters of the alphabet. If one is creating a text message "by hand," as it were, one needs to cycle through the letter choices and land on the correct one. Thus, making a J requires hitting the 5 key once; K, twice; L, three times; usually the fourth press will get you the number itself (though different cellphone companies have different ways of dealing with this). After that, the cycle usually starts again, though some cellphones expand the loop with various diacritics.

One click too few or too many, and the letter you pick might be the one that immediately precedes or succeeds the letter you want. It is interesting to me that this is the first time in the history of the English language that the standard order of letters in the alphabet is directly related to the errors that appear in written texts. In neither handwritten nor in typewritten texts is there a chance that quit might appear as ruit, but it's a common error in SMS.

Related to this, someone who is text-messaging letter-by-letter might wait too long, or not long enough, between keypresses within a word. This is most common in words that double a letter: if SMSing "free," for example, the texter will need to pause for a couple of seconds so the phone can recognize that the second E is not simply a recycling of the first E. Simply hitting the 3 key four times in a row will give you either "frd" or "fr3."

Predictive text. This is where things get really interesting, and it's the technological "advance" that leads to sublimely garbled messages like the one I used as an epigram for this writeup.

In the T9 system, popular among cellphone providers in North America, the phone tries to understand what word you want by narrowing down key sequences into possibilities for real (dictionary) words. For example, consider the sequence 94373:

9 4 3 7 3
W G D P D
X H E Q E
Y I F R F
      S

WHERE is the only English word that can be made out of these keypresses in this sequence, so most cellphones will suggest that word when you type 94373. You can hit OK, or often even the "spacebar" (0), and move on.

When this system works, it saves a lot of time: in the WHERE example, the user doesn't need to hit 4 twice for an H, 7 three times for an R, and so on. You only have to spend five keypresses for a five-letter word: it's just about as quick as typing.

But a sloppy user might accept the wrong suggestion from the computer, turning her SMS into nonsense. In the example at the beginning of this writeup, I was trying to answer a question about the whereabouts of a friend: "Is she driving you to the airport?" bewilderbeast asked.

The answer I'd intended to send was "No because she is somewhere else." I wasn't paying attention at the beginning, and the computer suggested ON instead of NO (both of which are keyed by the sequence 66).

More weirdly, my phone seems to think that PIE is a more common word than SHE. I did catch this error, but instead of hitting the backspace key so that I could correct it, I hit the send key. Thus bewilderbeast received the, er, bewildering answer: "On because PIE". (Capslock sic.)

My friends have created something of a game out of finding key sequences that make funny word possibilities. (My favourite is the sequence 2625, which can be both COCK and ANAL; on a slightly less vulgar note, archiewood informs me that KISS and LIPS are T9 anagrams.)

And of course, numerous deliberate misspellings, not to mention hilarious obscenities, are not in a cellphone's default dictionary and need to be added. Not long ago, I was trying in an exhausted, drunken haze to add sexing to my phone's dictionary in order to make an extremely important point to a friend in the next room. But the phone kept trying to tell me that I meant to be talking about sewing.