What the fingers want

« previous post | next post »

Whole-word substitutions are a common type of speech error: "Italy" for "Israel", "competent" for "confident", "restaurant" for "rhapsody", "drink" for "breathe". The substituted word is often associated with the target word or with its context, often starts with sounds similar to the target word, and often has similar syllable counts and stress patterns. An even stronger regularity is the syntactic category rule — the substituted word is almost always the same part of speech as the target word. Thus in the speech-error corpus examined by David Fay and Anne Cutler in their 1977 work "Malapropisms and the structure of the mental lexicon", this syntactic category rule held for 95% of all word-substitution errors.

Therefore substitutions like "They provider very good care" for "They provide very good care", or "He resignation yesterday" for "He resigned yesterday", are quite unlikely — in speech. In typing, in contrast, such slips of the finger are very common. I make errors like this all the time, with -ing or -ed or -s or -er or nothing appearing where one of the other choices would be correct. I haven't counted, but I think that my lapsus digitorum of this kind are an order of magnitude more common than the confident-for-competent variety.

And I see this kind of substitution now and then on the web, e.g. here:

My intuitive impression of what's going on when I do things like this is that my conscious attention has shifted to the following words and phrases, and my fingers — or more properly, the part of my brain that controls my fingers in typing — follows well-worn associative paths that happen not to be the right ones.

But this is more or less what happens with speech errors as well. So why is the mental lexicon for typing apparently organized in a way that doesn't impose the syntactic category rule on substitutions? Apparently the letter-to-letter associations in typing have a power that phoneme-to-phoneme associations in speaking don't.

Probably this is a well-studied issue in the psychology of language. But I don't know the references, and don't have time this morning to look them up, so I'll open the question up to the commentariat.


  1. Dick Margulis said,

    July 30, 2015 @ 6:44 am

    My first thought is that the writer meant adopted, not adapted. So already we're dealing with someone who is thinking faster than he's typing (not a criticism, just a clue). I take it this is someone who writes frequently about healthcare policy, and in that field the word provider is top-of-mind quite often and the word provide is not. He has just written PPO, and he obviously knows that it stands for preferred provider organization, so he is really well primed to type provider. I don't think syntactic category comes into play in this example at all.

  2. Stan Carey said,

    July 30, 2015 @ 6:51 am

    I haven't succumbed to provider for provide, but I do make similar typos pretty regularly.

    Typing that when than was meant is very common even in edited prose (I keep a collection), but I don't think I've ever heard the mistake in speech.

  3. Bean said,

    July 30, 2015 @ 7:27 am

    I think one doesn't type words: once the word is well underway, one types popular combinations of letters. I learned to touch-type (for real! not hunting and pecking) in English and obviously I type in English most of the time. I'm fast, but the kind of typos you point out (different parts of speech) do occur.

    But every once in a while I need to type in Italian and it's a disaster. Most of the words end in vowels, and I simply have trouble getting the last vowels onto the words, I keep having to go back and fix things, the words end "English style" when left unattended as I think ahead of my typing. I might add an n after the actual final vowel, or if the word ends in i, it might turn into ing, or often the final vowel turns into an e because that's about the only vowel that's common at the end of English words.

    Actually, speaking of -ing, I have a friend named Franklin and I almost always type it Frankling first, before correcting it.

  4. Eric P Smith said,

    July 30, 2015 @ 7:33 am

    I wonder if it's evolution. The mental association between what we think and what we say has had 200,000 years to evolve, and the syntactic category rule on substitutions in speech has had 200,000 years to evolve with it. We have only been typing for 200 years.

    my fingers — or more properly, the part of my brain that controls my fingers in typing — follows well-worn associative paths that happen not to be the right ones

    My intuition here is the same as Mark's.
    My personal favourite speech error occurred a few years ago when I was working very late, very tired, on a spreadsheet. I am a verbal thinker, and I may have been subvocalising. I thought: "Nearly done: all I have to sort out now is the Commonwealth. I don't mean the Commonwealth: the column width."

  5. Mark Meckes said,

    July 30, 2015 @ 8:26 am

    Similar to than/that, I've seen lots of is/in/it substitution errors in typed, even edited, prose.

    For some reason I almost always type Christman when I mean Christmas, consistently enough that I automatically go back to delete. And while typing that last sentence I was reminded of another such typo I often make: enought.

  6. Stan Carey said,

    July 30, 2015 @ 8:38 am

    Whenever I type probable I have to go through probably first.

  7. Tim Leonard said,

    July 30, 2015 @ 8:55 am

    In typing, certain spelling fragments (e.g., "sh", "ent", "ion_", "_sch", and whole words such as your name or email address) occur so commonly that the associated finger motions become automatic. The substitution of "provider" for "provide" is likely to be an articulation error—execution of the wrong motor plan—rather than a lexical error. So a fair comparison would be one between the relative frequency of articulation errors in typing to the relative frequency of articulation errors in speech.

  8. Yerushalmi said,

    July 30, 2015 @ 9:14 am

    I discovered a fascinating typing substitution rule quite by accident a few years back. Apparently the direction of the text you are typing also somehow affects your mental lexicon (well, mine anyway – it might be difficult to get a group of people to test this out).

    When I was a teenager, I developed a skill intended to impress friends over the internet: typing backwards. I would place my right hand above the left arrow key, and touch-type with my left hand alone. Between each two keypresses with my left hand, I would hit the left arrow key with my right hand, such that the cursor would move left, to immediately before the key I had just placed. In so doing, the sentences would come out backwards. After some practice, I was able to get a strong rhythm going, alternating fingers very quickly, and I could type out entire paragraphs, backwards, at almost the same speed at which I would type them forwards.

    .ti depyt I tsaf woh ees t'nac uoy ylsuoivbo hguoht ,elpmaxe na si ereH

    When I was 19, long after the last time I had conducted this parlor trick, I moved to Israel and had to learn to type in Hebrew. The different layout of the keyboard, particularly of the punctuation, is something that takes some getting used to. The Hebrew period, for instance, is located on the / key; hitting the English period while typing in Hebrew gives you the letter ץ. I eventually became relatively fluid in Hebrew typing, such that I can now type in Hebrew almost as fast as in English.

    Then one day I tried once again to type backwards, and found that, while my overall skill in it only slightly deteriorated from lack of use, I began experiencing a new difficulty. For some reason, whenever I intend on typing a period or comma while typing backwards, I instinctively press the Hebrew period and comma keys rather than the English ones. Though renewed practice in typing backwards reduced my rate of other errors to that exhibited when I was a teenager, for some reason this period and comma mistake proved impossible to eradicate entirely.

    The strong implication here is that, at least in my case, the different keyboard layouts are called up in my brain not based on the language in which I am typing, but rather on the direction of the text. For some reason, now that I've learned to type in Hebrew, typing from right to left has become so strongly associated with it that my brain calls up the Hebrew keyboard map even if I'm typing from right to left in English. Since none of the English letters appear on the Hebrew keyboard, this does not present a problem for most characters because there's no contradictory information; but any character that the two keyboard layouts have in common but not in the same place – i.e., the period and comma – comes out wrong.

  9. Noni Mausa said,

    July 30, 2015 @ 9:31 am

    For me, "within" generally becomes "withing," and "think" comes out as "thing." The mistake is a physical habit error rather than a language error. Interesting. Spoken language is built into our hardware, but the conversion to text (whether reading or writing) seems to be an add-on app.

    Something similar happens in the kitchen. I might be peeling eggs or carrots, and end up tossing some of the shells or the peels into the cooking pan, or tossing carrot slices into the trash with the peels. I might eat the peels but would have to consciously decide to do that, and eating the eggshells just wouldn't happen. But the cooking pan is an intermediate step that subverts the food/not food decision habit.

  10. Scott Mauldin said,

    July 30, 2015 @ 10:14 am

    9 times out of 10, I will type "Soviety" instead of "Soviet", I imagine by analogy to "society".

  11. MsH said,

    July 30, 2015 @ 10:15 am

    Before I learned to touch-type, I never ever mis-wrote words that sound the same but have different meanings, like "their" and "they're". However, when I learned to touch type, this kind of error started appearing sometimes. My thought was that they must be using partly different mental processes – when writing longhand, the process had been based on meaning, but when typing (which is quite a bit faster) it must be governed much more by sound.

    On top of that, there are now lots of errors introduced by predictive text when writing and editing in various mobile devices. These don't follow the same pattern at all but could easily produce examples like the one above.

  12. Jerry Friedman said,

    July 30, 2015 @ 10:45 am

    I make all these kinds of errors, though I don't think wrong suffixes are among my more common ones. I also repeatedly make some strange errors that don't have explanations that are obvious to me, such as "electrnoic".

  13. bfwebster said,

    July 30, 2015 @ 11:06 am

    I have noticed more and more typing errors like this on my part in the past few years as I have grown older (I'm 62) — for example, typing "development" when I meant to type "developing". It's been puzzling and a little disturbing; keep in mind that I'm a great touch-typist, have an excellent grasp of English (99th percentile on English-skills GRE 40 years ago), and made my living as a full-time writer (information technology, articles and books) for a few years. I wonder if some of it is due to autocorrect (the iOS autocorrect is particularly prone to make unwanted substitutions), but there are other instances where there is no autocorrect in place — just me and my fingers.

    Because I am a touch typist and don't actually think about the individual keys I'm pressing most of the time, I've wondered if there is some short-circuiting or miscommunication that is going on — I think about typing one word, and my wrists and fingers type another. Of course, I have had no such error of that type while typing this comment. :-)

  14. bfwebster said,

    July 30, 2015 @ 11:10 am

    > Noni For me, "within" generally becomes "withing," and "think" comes out as "thing."

    I have some of these as well (the think/thing one in particular has plagued me forever), but I've had those most of my life and am used to catching them as soon as I type them. The recent errors are almost all suffix-related, with the mistyped word being a different form of the intended word (as I mentioned, "development/developing/developed", etc.).

  15. Janet Novak said,

    July 30, 2015 @ 11:26 am

    Within the class of typing errors that represent plausible combinations of letters, I've noticed that adding extra letters at the end of a word is surprisingly common. Examples of this pattern from the other commenters include:

    "provide" typed as "provider"
    "enough" typed as "enought"
    "Soviet" typed as "Soviety"
    "within" typed as "withing" (and I myself often add a "g" onto words ending in "in")

    For me, a very common typo is adding an "s" onto a noun that I'd intended to be singular, whereas I almost never leave off the final "s" of a noun I intend to be plural. In general, adding extra letters is much more frequent for me than leaving off letters at the end of a word, and somewhat more common than making substitutions of plausible letters.

    When I'm conscious of making the "extra letters" error as I type, it feels like my fingers want to execute the longest motor plan possible for the string of letters I've just typed. Assuming this phenomenon is real, I have no idea why there would be a bias towards longer motor plans over shorter ones.

  16. wanda said,

    July 30, 2015 @ 11:38 am

    I offer a different hypothesis. I find that when I'm composing slowly, one of the things that I change frequently is the part of speech of a word. I'm experimenting with sentence structures, such as whether an idea sounds better when the "important part" is part of the noun-phrase or the verb. If I don't go back and edit, sometimes part of the rest of the sentence escapes noun-verb agreement.

  17. Bob Ladd said,

    July 30, 2015 @ 11:50 am

    I think Mark has an important part of the issue when he says "Apparently the letter-to-letter associations in typing have a power than [sic] phoneme-to-phoneme associations in speaking don't." My own illustration of this is that I often type gladd when I am aiming at glad – there's no whole-word or part-of-speech factor here at all, just automated sequence learning from frequent typing of my last name.

    I also regularly have experiences of the sort Bean reports when I type in other languages.

    Mark: Bill Cooper went on to do some work on typing after Cooper & Sorensen. Don't know any other literature on this, though like you I imagine it's out there.

  18. K Chang said,

    July 30, 2015 @ 12:49 pm

    Maybe it's just way-too-aggressive autocorrect… .:D

  19. rosie said,

    July 30, 2015 @ 1:31 pm

    Perhaps some other typos are due to keyboard rollover? I have many times typed thte instead of the, and eset instead of est at the end of a word (e.g. intereset).

  20. Elika said,

    July 30, 2015 @ 3:27 pm

    ooh, looks like this old paper is due for an update, would be interesting to compare results, though probably impossible to find a young adult non touch-typist group:
    (Typing Errors as Clues to Serial Ordering Mechanisms in Language Behaviour, MacNeilage, 1964).
    A quick google scholar search doesn't reveal anything else looking precisely at this in the last few years but I didn't do a very thorough search tbh

  21. Viseguy said,

    July 30, 2015 @ 7:55 pm

    > "Soviet" typed as "Soviety"

    For me, "Judith" always comes out as "Judity". Fortunately, it's the kind of mistake that you can safely AutoCorrect.

  22. Biscia said,

    July 31, 2015 @ 3:12 am

    This is tangential, but I was chatting the other day with fellow translators about how – especially when going fast on a first draft – I sometimes find myself coming out with weird phonetic spellings, i.e., “peeple” or “senter”, things I didn’t even write in fourth grade. It’s as if I were transcribing somebody else talking, and guessing wrong about what word they were starting to say. Apparently I’m not the only one, and for me at least it gets worse when I’ve been watching too many movies in my target (native) language, which is not my language of everyday oral conversation.
    And then there's my tic of always typing "white" as "shite" when I'm tired and annoyed, which makes design and art-related jobs amusing, but I'm pretty sure that comes from elsewhere in my brain.

  23. richardelguru said,

    July 31, 2015 @ 5:51 am

    I find that I often insert a lettfer from the next (or next-ish) word.
    (that one was intentional)

  24. Bean said,

    July 31, 2015 @ 5:55 am

    @Elika: Ha ha, I submit that all young adults are technically non touch-typists… in the "hands on home row and type while looking at copy the whole time" sense. They start out as hunt-and-peck at a very young age, and those patterns and odd fingering choices are then ingrained; I doubt you could train them out of it. Also they need to look at what they're typing; they can't type without looking at the screen. (That is my impression – perhaps I'll quiz our co-op students at coffee time today for a sample of N=3.)

    Another change I've noticed is that we are now so inundated with non-edited text that when we see problems like their/they're/there confusion we've become inured to them. The corollary is that the same typos start to come out of my own fingers as well and they don't look as awful as they used to!

    I think we've collectively hit on something here, that typing errors are controlled differently than speech or handwriting errors, and it's not quite clear what the hierarchy of controls is, and yeah, if the last paper on the subject is from 1964 it's probably time to do another experiment. :P

    Handwriting errors: I make tons of them now because I don't write enough to be good at it, and they are ALWAYS omissions of letters. Again – brain moving faster than pen. I keep thinking I should start a hand-written journal, not because I have anything useful to say, but to keep my handwriting muscles in shape (and those neural connections connected).

  25. Alex said,

    July 31, 2015 @ 1:05 pm

    Since the question is about the difference between errors in speech and typing, does anyone know what types of errors are common in sign language? (I'd imagine more similar to speech, but I don't know.)

  26. Mark F. said,

    July 31, 2015 @ 2:34 pm

    My guess is that speech production, in terms of actual control of the relevant muscles, is probably very tightly integrated with the rest of the language system in the brain. Typing is another step removed.

    Bean talked about how you type clumps of letters, but in principle the vocal system could work that way too. I'm wondering if this tighter integration that I am assuming is what makes the difference.

    I also wonder if people ever get the wrong part of speech on words that take a really long time to say.

  27. Karen said,

    July 31, 2015 @ 6:47 pm

    It is very reassuring that you attribute it to sound-related processing error rather than senescence. I definitely come up with more sound-related typos on a keyboard than I used to back when I used handwriting more and typing was much less fluent. But which is it – time or technology? The ones that bother me the most are when I substitute one of (o, i, u) for another on unstressed syllables, or type "ov" for "of". I don't fret when typing "r" for "t" since that is clearly not forgetting how to spell, but when the error is plausibly ignorant, it can derail the rest of the sentence. Especially without Biscia's job description.

    [(myl) Like you, I seem to produce more of these errors as time goes by. My story about it is that it's the result of accumulated practice in typing, which builds up stronger and stronger chains of habitual associations among the finger gestures for letter sequences.]

  28. Lars said,

    August 1, 2015 @ 6:55 am

    Speaking as someone who learned touch typing (for Danish) on a mechanical typewriter in grade 8 (but never got the diploma because I switched my elective to Latin in grade 9), I recognize all of the above. But I also have a 'motor plan confusion' where check will occasionally come out as chack. It's always e -> a, never the other way around, and almost always in that single word.

    I'm not confused about how to say check (or spell it), but it's possibly relevant that word-initial 'ch' is always followed by a vowel, and 'a' is the only other left-hand vowel so it's the same left-right-left rhythm. (And I want to use the same (middle) finger for 'e' and the second 'c', so chack is actually a few tens of milliseconds faster to type. But that's hardly a conscious consideration).

    Another interesting thing (to me) is that I 'listen' to the motor feedback when typing, so I often know when I made a mistake and can start to correct it before checking the screen. The big surprises when looking at the screen come from 'misregistration' (finding the wrong home row position = 'oisutuib') – that feels like I'm typing the right thing, of course.

  29. Michael P said,

    August 1, 2015 @ 12:04 pm

    On a related note, has anyone noticed an increase in what one might call "typing metathesis?". For instance, writing "applicatino" instead of "application"? The infamous teh is a well known example. A few of us at work have been remarking on this lately; it seems to be happening to all of us more & more often. I think it's related to the ascendancy of autocorrect – since so many errors of this kind are caught and corrected as soon as we make them, we are defaulting to ever-looser motor plans, until autocorrect ends up losing the error-correction arms race.

    Embarrassingly, this has started happening with my own name: It comes out as Mcihael or some such variant. When I'm typing at speed, I can mistype, erase, and retype my name several times before getting it right!

    Also interesting is that it seems like the metastasis very frequently takes the same form: for instance, I'd say applicatino is by far the commonest way I misspell application.

    I'd be interested to hear others' thoughts on this.

  30. Graham Blake said,

    August 1, 2015 @ 9:07 pm

    Echoing some others, I have noticed my typing is prone to errors based on the sound of a word, especially as I get increasingly tired. It's like my brain is dictating to my fingers and my fingers have stopped trying to figure out if they heard my brain correctly, and they just type whatever they think they heard. Really strange, only vaguely homonymic errors, like typing "are" when my brain said "hour". It definitely feels like two completely different mental processing systems, one of which still knows words, but can completely detach from their meaning if not monitored closely.

  31. Chas Belov said,

    August 4, 2015 @ 3:00 am

    When hand-printing, I'll occasionally mix -r and -d as the last letters of words.

RSS feed for comments on this post