That Silent Inner Voice.

Keely Savoie reports on a phenomenon that’s always fascinated me:

That inner voice that enunciates the written words you read comes in many different forms. Some say it sounds like the spoken voice. Some say it sings. And others say it is someone else’s voice entirely. Whatever the voice sounds like, it performs an important function in interpreting the written word.

Mara Breen, assistant professor of psychology at Mount Holyoke College, studies the relationship between the inner voice and its musical rhythms—known as implicit prosody—and how we mentally process the written word. She recently received a grant from the James F. McDonald Foundation to study the role of the inner voice in reading fluency and comprehension.

“What we are specifically interested in is what is the experience of prosody when we are reading silently,” Breen said. “In other words, what is the little voice in your head doing while you are reading? How does implicit prosody support reading comprehension?”

Breen and colleagues at Haskins Laboratories and the University of Connecticut have completed research that demonstrates a relationship between prosodic fluency and reading comprehension in high school students, even when accounting for IQ differences.

There’s some very interesting stuff about reading limericks and “what happens when the written word violates the brain’s expectation”; I look forward to seeing more such results. (Thanks, Sven!)


  1. Fascinating read, thank you! I have a question about this, though:

    The brilliant abstract the best ideas from the things they read.

    Most people will read “brilliant abstract” as an adjective preceding a noun, a superlative review of a scholarly overview (brilliant AB-stract). But it quickly becomes clear that “brilliant” is a noun—a class of people—and “abstract” is a verb with a different pronunciation (ab-STRACT). The reader has to start fresh to make the correction.

    I read it as referring to brilliant people on first glance, and I’m pretty much the antithesis of brilliant. Do “most people” really not read it that way?

  2. I don’t hear a voice in my head unless I am reading poetry, in which case I am slowed from my standard 4000-wpm speeds to spoken-word speed. Of course, when I am reading out loud, I hear the voice from outside my head.

  3. I read “brilliant” as an adjective at first, but that means nothing. I’m guessing the majority thing is part of the study.

  4. La Horde Listener says

    Christ Jesus! All my reciting/pronouncing voices sound like those from A Charlie Brown Christmas and South Park: stammering little Butters, the whole ” C R A B P E O P L E ” mantra, and the underpants gnomes, included. Suddenly it’s very dark in here, *shudder*…

  5. Sometimes when I’m singing a song without accompaniment, the voice in my head will outpace me and trip me up.

  6. “If a sentence has a particular rhythmic structure, people will tend to predict that strong syllables will occur at specific time points,” Breen said.

    Motor theory of speech perception? Not mentioned, perhaps outdated. Could be extended to reading perception. Also not mentioned is the old issue of when silent reading began historically. (“Midway through the Confessions, St. Augustine recalls how he used to marvel at the way Ambrose, the bishop of Milan, read his manuscripts: ‘His eyes traveled across the pages and his heart searched out the meaning, but his voice and tongue stayed still.'”

    I hear a voice when I read slowly (the only way I read), a voice not my own.

  7. marie-lucie says

    I have mentioned earlier (perhaps more than once) that in reading a language that I don’t know very well and am trying to learn, I read it aloud. This seems to silence or at least diminish the voice in my brain which wants to translate the text into English or French. It also helps me concentrate on the prosody. I find Spanish ideal because the spelling indicates stress, either by default or by a specific mark. My Italian is not quite as good because its spelling only indicates stress “under duress”, if I may use this phrase – only if it really, really has to. I would like to brush up on my limited Russian using the same method but I have the same trouble as with Italian (in addition to the much larger amount of non-cognate vocabulary compared with the Romance languages).

  8. the written word violates the brain’s expectation?

    I remember a story about a first time Russian poet who brought for review a poem entitled “Obelisk”, with this word repeated from stanza to stanza … except the poor chap stressed the MIDDLE syllable every single time. A mind-tearing experience

  9. Here

    Я говорил: «Бляхин, почему ты пишешь о кабаках, ведь ты воевал, пиши о том, что видел, знаешь…» Как-то Бляхин приходит: «Ну, старший лейтенант, оставляю вас на неделю, мне рулон обоев принесли, так что сажусь за поэму»…

    Perhaps my inner voice is seriously rhythmical because I couldn’t get the silly example out of memory…

  10. Those of us writing fiction have a particular problem presenting regional accents to readers. How do we encourage the readers’ inner voices to hear/speak the dialogue with the intended accent? In my view (as a reader) using phonetic spelling seldom works and leads to much stumbling and re-reading. (Such writers also seems to assume, unreasonably, that the reader will share a core accent with them, so that variations in spelling will suggest the same sounds to the reader as to the writer.)
    Conversations with other writers suggests general agreement that the authors with the best chance of achieving some authenticity in characters voices tend instead to match word order and rhythm to imply the pattern of a character’s speech rather than the sound of it, (with perhaps a few recognisably regional phrases thrown in).
    My question to the readers/writers here however would not be about writing accents but ‘what does the reader’s inner voice do with the accents suggested on the page’?

  11. I think I tend to pick up on such things pretty accurately, but of course I’m an extreme outlier (not only having been exposed to all manner of accents, dialects, and languages but having a degree in linguistics). It’s an interesting question.

  12. From the FAQ in Leo Rosten’s introduction to The Return of H*Y*M*A*N K*A*P*L*A*N (an early example of a FAQ, being published in 1963):

    “Is dialect harder to write than other forms of humor?”

    Much. It is also more risky, more tricky, more perplexing, and more dangerous.

    Humor is the affectionate communication of insight. (Satire is focused bitterness, and burlesque the skewing of proportions.)

    Humor is, I think, the subtlest and chanciest of literary forms. It is surely not accidental that there are a thousand novelists, essayists, poets, journalists for each humorist. It is a long, long time between James Thurbers.

    Comic dialect is humor plus anthropology. Dialect must seduce the eye to reach the ear and be orchestrated in the brain. It must tantalize without irritating, and defer without frustrating. It must carry a visual promise to the reader that what he does not instantly recognize can be deciphered with ease and will be rewarded by pleasure. The reader must be cued into making what be thinks is his own special and private discovery — a discovery of delight which, he suspects, neither the character nor the author fully appreciates.

    Dialect is not transcription. Nothing is more depressing than a passage of broken English exactly transcribed from the spoken. The “accurate ear” for which a writer is praised is as inventive as it is accurate. It is creative, not literal, for the writer transforms that which he hears into that which you could not. There is a magic in dialect which can liberate us from the prisons of the familiar.

    In the antic freedom of phonetics, mortals say “ship” when they mean the source of wool, or “sheep” when they mean vessel. Others throw “bet” around with the abandon of a gambler — to mean either “bat,” “bet,” “bad,” or “bed.”

    The writer must therefore create exact, if camouflaged contexts within which the reader’s responses are firmly controlled. He must convey without explaining. Mr. Kaplan may say “fond mine fife fit don,” which suggests a devoted shaft of flutes fit for an Oxford tutor. But if the clues have been properly structured, and the channels of association properly cut, the reader will have to do no heavy lifting to know that Mr. Kaplan, having lost something, found it five feet down.

    Or take Mrs. Moskowitz. Poor, dear Mrs. Moskowitz. She says, if I were to record it with absolute rigor, “I hate the brat.” But that is not at all what she means. If I made her say, instead, “I ate the brat,” which is closer to her message, Mrs. Moskowitz would be tainted with cannibalism, which is absurd. (There are certain animals Mrs. Moskowitz would not dream of digesting.) I would have to violate truth, in the service of truth, and write Mrs. Moskowitz’s perfectly innocent thought as “I ate the brad.” (Why doesn’t she say “I ate the bread”? The day she does, I’ll promote her to Miss Higby’s grade.)

    The problem is immensely complicated when the characters in a story write as well as speak. For they write quite differently from the way they speak. Each, indeed, uses two separate vocabularies. Irate Hawkshaws used to write me complaining that on page such-and-such Mr. Kaplan wrote “was,” whereas on page tit-for-tat he said “vas.” They missed the point: beginners learn to write “was” without the slightest anguish, but many pronounce it “vas” until threatened with execution.

    There are more, and more mischievous, jokers in this enchanting deck. Take Mr. Kaplan’s name. It is a simple name. Anyone in the class can spell it correctly. But notice: Mr. Kaplan refers to himself as “Keplen,” Mr. Blattberg calls him “Kaplen,” Mr. Plonsky always bellows “Keplan,” Mr. Matsoukas mutters “Koplen,” and Olga Tarnova, who could wring lurid overtones from a telephone number, moans “Koplan.”

    Apart from driving my proofreaders crazy (and any departures from the code above are clearly their doing and not mine), these variations seem to me to add richness to the characterizations, and texture and nuance to the language which Mr. Parkhill’s disciples keep transforming to suit their own neurotic needs.

    If, at this point, you wonder whether it is really possible for so many to mangle so much with so little let me assure you that they do, they do. (See Arthur Winsome Platt’s admirable study, Humor, Dialect and Catharsis, published by the Rutgers Press.) *

    * On second thought, don’t try. Professor Platt taught at Rutgers for years, but was asked to resign when they discovered he wasn’t real. I just made him up to add the weight of authority to my argument. I hope this teaches you not to be impressed by footnotes.

  13. A fine passage (I hope people still read Rosten), and I love “There is a magic in dialect which can liberate us from the prisons of the familiar.”

  14. My favorite bit is “It is a long, long time between James Thurbers.” Which reminds me that John O’Hara wrote the following letter to Thurber and his wife Helen:

    Dear Thurbs: By the way, what does a thurber do? What is thurbing? “I think I’ll go out and thurb the nasturtiums.” “Shall we go thurbing this afternoon?” “That goddam thurbing son of a bitch Ross.” “Father, the greeve needs a new thurber.” All these years, and I never had the philological curiousity [sic] to ask a simple question.

    In the third Kaplan book (a heavily edited combination of the first two, known to Kaplanists as the Revised Version), he adds an example to the “Dialect is not transcription” paragraph: “Any yokel in the Ozarks can try to be a thigh-slapper by writing ‘We wuz shure suprized by Maw’s coolinary conkokshun.’ The sounds are authentic enough, and enough to make me wince.”

    Also “unneurotic” above is a typo (not mine) for “own neurotic”. Pkease fix.

  15. No voice in my head either. In fact, it’s the other way around: when I hear words, I normally see them written down on some inner page. How does it work? Do you hear it when, say, you see a road sign or a restaurant name, or just when you read text?

  16. Hey, Lameen, you might have something to say about this.

  17. The voice in my head is more or less my own, unless it’s a character I’ve seen portrayed in a movie/tv show, or I have an actor whose voice fits the story well.

    I went through a stage in grade school where I saw words as people spoke, especially when I was in church, but often everywhere.

  18. Natively, it’s like an abstraction of a voice, not as concrete as a memory or imagination of any real voice. Interesting to hear how it is that concrete for some people.

    I don’t hear it as my own voice, though it’s fairly similar. A male voice, I’m fairly sure. I suppose it’s not inconsistent with my own voice. I can make it be more concrete, but that’s volitionally imagining some voice I choose.

    There’s not an inner voice at all when I’m reading normally, but I use it to read poetry or other text that benefits from the stronger perception of the sound. Actually, I hadn’t thought about this, but if I slow down reading to a speech-like pace, I get the voice, willy nilly.

  19. having a degree in linguistics Ha! It comes out. I always thought that.

    Humor is the affectionate communication of insight. That is golden. Golden, golden, golden.

  20. Feynman on how people count mentally. Short but much less amusing version: Some, like Feynman and me, subvocalize the numbers, so they can read and count at the same time, but not talk and count at the same time. (This wouldn’t work for people who subvocalize while reading, so I deduce that Feynman didn’t any more than I do; apparently he never tried to read poetry and count at the same time.) Others, like John Tukey (of the Cooley-Tukey FFT algorithm), mentally watch a tape unspooling with the numbers on it, so they can talk and count at the same time, but not read and count. (Feynman also had synaesthesia and I don’t, but we both share the experience that “thinking is just talking to yourself inside”.)

  21. Trond Engen says

    I subvocalize when I read. It’s been annoying me, since I’m pretty sure it slows me down, but when I force myself not to, I soon discover that my brain can’t keep up with my eyes. I also have a dialogue with myself when I think. I’ve been thinking of that as the same process. The exception is maths, or maybe rather physics,, which I believe I can handle as geometry without words.

  22. Trond Engen says

    The Norwegian master of writing dialect without eye dialect is Knut Hamsun. He wrote in a conservative Danified Riksmål, but his stories are thoroughly Northern Norwegian, and his characters even more so.

  23. I, on the other hand, don’t have no degrees in nothings. I am self-edumacated.

  24. David Marjanović says

    we both share the experience that “thinking is just talking to yourself inside”

    The next sentence mentions that there’s plenty of languageless thinking, too, and indeed I’ve found I do both.

    I type fastest when I copy something by hand without sending it through language processing. This way, I don’t imagine the pronunciation, and I miss large parts of the content – I don’t notice at least half of what the text means.

    I can recognize words faster than language processing sets in. But that’s also faster than spatial perception sets in: I can accidentally read a word or two before noticing where on the page they are – so then I have to look for them if they were interesting. Normal reading does go with imagining a pronunciation and intonation. The voice for this, as for everything else I think in language, is an abstraction of my own; on average higher and with no volume to it whatsoever. (In movies, unspoken thoughts are conventionally represented as the actor’s voice with a lot of echo added to it; that’s exactly the inverse of what it’s like for me.)

    I also often imagine words I hear or speak or think as written black on white. Russian is in Cyrillic, Mandarin is in Pīnyīn. I don’t actually remember being illiterate, but I do remember figuring out how to read.

    I count in words.

  25. David Eddyshaw says

    Like John Cowan and Lameen, I have no such voice at all. As a child I assumed this was the norm, until a conversation I had with a music teacher about the great difficulty I had sight-reading. “But when you read, don’t you hear your own voice saying the words?” Nope, and nobody else’s either.

    It occurs to me that I don’t hear a voice when I read French either, though I am certainly aware of the sound of the words *somehow*, in particular because of the considerable mismatch with the spelling. Unfortunately introspection as to how this apparent contradiction can be resolved has brought me no enlightenment at all.

    In both languages, I am perfectly aware of homophones , but not by hearing any inner voices. It’s just now occurred to me that this is odd …

    Maybe the failure of introspection is that I don’t *realise* I’m hearing a voice. Sort of internal blindsight. Blinddeafness. Deafhearing. Whatever.


  1. […] Hat notes an interesting-sounding study of the nature of inner […]

Speak Your Mind