Mental Subvocalization --"Saying" Words In Your Mind As You Read

by Torello1 min read15th Feb 201478 comments


Personal Blog

I'm curious about how often or to what degree visitors to this site subvocalize as they read.  I was originally interested in reducing subvocalizations as a way to increase reading speed, as the idea is mentioned in multiple pieces I've read about speed reading.

The Wikipedia entry seems to focus on subtle throat and muscle movements, but I'm more interested to know if you "say" or "hear" the words in your head as you read.

Since reading about subvocalization recently, I seem to notice that I "say/hear" what I'm reading quite frequently.  I'm not sure if this is causal (in the way that the command "don't think of pink elephants" obliges you to do so), or if I just notice it more now, or both. 

When I'm very engrossed in a book either I don't notice the subvocalizations or they stop happening, so seems that it could either be a cause or a symptom of distractedness.

In the comments, please describe your mental subvocalizations (or lack of them) and if they are related to how engrossed you are in the book.  Any other comments relevant comments about speed reading or subvocalizations are welcome.

78 comments, sorted by Highlighting new comments since Today at 3:58 PM
New Comment
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

I suspect that I subvocalise only when I'm paying attention to my own thoughts. So it seems like I always subvocalise, just like it seems that the light in the fridge is always on.

1FeepingCreature8yUpon trying to observe myself vocalizing, that is how it seems to work. It feels a bit like the sentence "expands" into speech when I pay attention to it. Also, I always seem to vocalize when writing.
0mindspillage8ySame here, with writing: it generally feels like transcribing an internal monologue.
0Torello8ySo, do you have control over paying attention to your thoughts? To use your metaphor, do you intentionally chose to open the fridge door, or does it sometimes swing open uninvited?
0Oscar_Cunningham8yWhenever I read this thread I notice I'm subvocalising and can't turn it off until I get distracted and am no longer paying attention to whether I'm doing it or not.
0mindspillage8yI suspect this is true for me also, but of course I can't quite manage to observe it...

I've always thought of written language (at least of the alphabetical/syllabary sort) as a kind of learned synesthesia. Certainly the results of the Stroop test on most people with typical neurology would suggest this...

4pianoforte6118yThat's one of the coolest things I've read all week.

Reading these responses I realized that I don't know what subvocalization is. I read at 500 wpm with good comprehension. I certainly hear the words when I read and I can't turn that off even if I go up to 800 wpm. But I can also count out loud while reading with not much effort. And I don't hear the words linearly as though someone was speaking really fast. I hear them sort of all at once in chunks like five words at a time.

Do I subvocalise?

3private_messaging8yI consider sub-vocalization to be what goes on when you read something out loud (as in, reading a story to a child), but without full movement and making noise. But still producing the signals for your mouth and throat. This would be suppressed by counting out loud regardless of how automatic is the counting. Counting is good precisely because it is automatic, something more difficult could interfere on the conscious thought level rather than on subvocalization level. edit: instead of counting you can try singing a song while reading something unrelated...
0pianoforte6118yI think I have a good test of subvocalization. For those of you who subvocalize, is there a difference between reading and saying the words out loud in your mind? Ulcbgurfvf: Vs lbh fhoibpnyvmr, gurer vf ab qvssrerapr (gur dhrfgvba qbrfa'g znxr frafr). Vs lbh qba'g fhoibpnyvmr, gura gurer vf n irel pyrne qvssrerapr.
0Error8yThis is interesting. Can you speak pseudorandom numbers as well, or only count in sequence? (hypothesis: Bayl va frdhrapr. V rkcrpg pbhagvat va frdhrapr gb or zragnyyl nhgbzngvp, yvxr jnyxvat. V qbhog guvaxvat hc "enaqbz" ahzoref vf bcgvzvmrq gung jnl, fb V rkcrpg vg gb erdhver pbafpvbhf nggragvba naq qvfehcg ernqvat.)
0pianoforte6118yIts harder, it cuts my speed in half.

Reading and counting interference poll.

Report the "best" performance of any language that you speak:

Can you read while counting out loud?


(No interference means that you can count while reading without any impact from one on the other)

Can you imagine a sentence read in a specific voice / accent? (e.g. professor Farnsworth, BBC commentary, etc)


Can you do that while counting out loud?


Do the answers differ for different languages?


1bramflakes8yWow, I was doing fine until the counting while reading in a different accent part. It felt so strange, like the gears had jammed in my brain.
0primality8yI speak two languages fluently. I observe that it is slightly easier to count in one language and read in the other. The full ranking is reading without counting > reading English, counting Danish > reading Danish, counting English > reading Danish, counting Danish > reading English, counting English. That counting Danish > counting English is presumably because I don't count in English nearly as often. I used an LW article as my English text, and a physics textbook as my Danish one. I would say these two texts have similar difficulty. Like bramflakes, I was surprised by how difficult it was to count while reading in a different accent.

I don't subvocalize, and read pretty darn quick. As a result, I tend not to think at all about how to pronounce character's names (esp in fantasy series) and treat the name more like a pictogram, and often have no idea how to say them when asked without pulling out the book and looking.

Do you subvocalize when reading in your native language? [pollid:607]

Try to read some text without subvocalizing. [pollid:608]

Try to read some text and subvocalize. [pollid:609]

I suspect that the answers may be more informative of people's different conceptions of what sub-vocalization is, or different ways of measuring it, rather than of the mental processes involved in reading...

E.g. I presume that I can't sub-vocalize while counting out loud, especially counting at a fast speed. I can read, but not write, while counting, which is why I think counting impedes vocalization.

2tgb8yAs an always-subvocalizer, I'm not even sure what it would mean to read something without subvocalizing it. For me, that would be like the Chinese room thought experiment - maybe something is understanding that sentence, but it's not 'me.' (Edit: also my reading speed is ~240wpm when I actively try to read fast, depending, obviously, upon the material. I suspect my typical is more like 200wpm.)
2private_messaging8yCan you read while counting out loud? I wonder if it's similar to the Chinese room for most people - as in, "I don't normally subvocalize, and it would make sense that the Chinese room is not sentient" vs "I always subvocalize and Chinese room makes no sense".
0tgb8yYes, but my comprehension and speed decreases. (Or it feels that way, I haven't actually tested this. This also doesn't feel like the right test to me - I can 'squeeze' counting in between reading, so it's not necessarily the case that one isn't blocking the other, it might just be that I'm alternating between them quickly.) Edit: And I didn't mean anything deep when I used the Chinese Room as an example! It was just meant to illustrate the enormous gulf in my mind between 'comprehending through subvocalization' versus 'comprehending despite not subvocalizing.' I tried several other metaphors before settling on that. I would be surprised to hear that there was a significant correlation between subvocalization and acceptance of the Chinese Room results and did not mean to imply that, though I suppose it is an interesting question regardless. Maybe this metaphor makes my meaning clearer: asking me to read without subvocalizing would be like asking me to look at a painting without ever experiencing the qualia of seeing it. If I shut down that qualia somehow (and I don't think could), then it's still possible that some portions of my mind are becoming aware of what was on that picture and maybe you could discover that through some clever experiments. But that part of my brain that would have learned that doesn't feel like it could be 'me' in the same way that it doesn't feel like it could be 'me' that understands some read text without subvocalization.
5private_messaging8yI still have the qualia of reading, just not the qualia of sound (hence impossible to imagine that i am listening to it being read in an accent, while counting). It's hard to describe. I can talk about an arbitrary subject while reading, and right now I am typing this message while reading what you wrote. I have simultaneously the qualia of "hearing myself think" (or the qualia of talking if i count), and the qualia of reading, basically.
0lmm8yDo you mean those two options the other way around? I don't normally subvocalize but I don't find the Chinese room argument at all compelling.
1ESRogs8yI'm confused as to why the two would be related at all.
3private_messaging8yThere was a hypothesis that people's different opinions on questions involving consciousness have largely to do with personal experiences that differ between people. People who always subvocalize may be unable to imagine that distinct ways of processing the language can exist.
1ESRogs8yThanks for the clarification! This reminds me of Galton's research on mental imagery, and I can see how people who do or don't subvocalize could potentially have different intuitions about various aspects of consciousness. However, I think I'm still not understanding your suggestion. For context, here are the major responses to the Chinese Room thought experiment that I'm aware of (do you know of others?): 1. such a system is not possible (i.e. it wouldn't work -- you couldn't get correct answers to questions this way) 2. such a system is possible, and it would be sentient 3. such a system is possible, and it wouldn't be sentient, nor would any other artificial thinking process 4. such a system is possible, and it wouldn't be sentient, but other artificial thinking processes could be, depending on how they work My own first guess is that if there is a connection between subvocalization and intuitions about the CR, it would be that subvocalizers are more likely to think explicit internal monologue is necessary for consciousness, and so would be more likely to choose #4 over #2, if they are already reductionists. Was your suggestion that those who don't subvocalize would be more inclined to choose 1, 3, or 4 above, or something else? And likewise for those who do subvocalize.
2tut8yI took the last option for the middle question. How it is for me is that I can look at a text without subvocalizing it and for example look for a certain word or passage, or look for the difference between two texts faster than I can read. But I can't understand a text that I haven't read before without hearing it in my head. And if I try to do anything with the text in my head that I couldn't do with an abstract picture I automatically hear it. I would say that mentally subvocalizing a text without seeing it feels to me a lot more like reading than what seeing the same text without hearing it does.
1RowanE8yIt seems I can't turn off subvocalizing right now, but I suspect it's something I only do when prompted, and am too uncertain on how often I'm actually doing it when I read to answer the first question.
0Luke_A_Somers8yI hear it spoken at around 600 wpm, just measured on a not particularly challenging text. No subvocal motions are associated with this. I can count or even perform basic arithmetic while reading or writing.
0Creutzer8yCuriously, I almost never subvocalise when reading texts in my native language, but it is virtually impossible to switch off in English. I wish I could switch it off because I suspect it's one of the factors that depress my reading speed in English.

I read in my native language without subvocalizing, and in English with subvocalizing. I can make an effort and read w/o subvocalization in English, but then I get an unpleasant feeling that I'm reading in a very shallow way, understanding and retaining much less than usual. I don't know for certain that this feeling is actually correct, but the evidence leans that way.

I was shown that I can reach effective reading speeds of something like thousands of words per minute, with decent retention, by quenching the subvocalization reflex and just forcing words to fly past my eyes. The trick is to stop every paragraph and summarize what you just read. Doing this is exhausting and I haven't kept up the skill.

Reading without sub-vocalization is impossible, although you may not notice you are doing it.

Here's the problem with that. You can't read without subvocalization. Carver and Rayner have both found that even the fastest readers all subvocalize. Even skimmers subvocalize key words. This is detectable, even among speed readers who think they don't do it, by the placement of electromagnetic sensors on the throat which pick up the faint nerve impulses sent to the muscles. Our brains just don't seem to be able to completely divorce reading from speaking.

5private_messaging8yThis is a bit ridiculous. I can count out loud while reading (doing both at maximum speed for each separately). This means that the motor areas in the brain which generate signals to control muscles in my tongue, larynx, etc. compute the right nerve impulses for words 1, 2, 3, 4, 5 .... . This means that they are not computing the nerve impulses for vocalization of what ever I am reading. Now, if during reading some little signals are present, that can mean all sorts of things. E.g. in a robot that would mean that the power supply's voltage stabilization is not perfect. Unless they actually decode words from the impulses they can't say if its some sort of noise or subvocalization.
1Douglas_Knight8yThat claim does not seem to appear in the two sources he cites. The word subvocalize does not appear in Rayner [\].pdf) and appears only twice in Carver []. Going by the titles, it would be rather surprising if it appeared in those sources. I'm skeptical enough of the usefulness of the claim that I'm not going to try other sources, but if someone gives me a precise citation, I'd be interested.
0Torello8yI think private messaging makes an interesting point, but I'm more concerned with hearing (or noticing that your hear) the voice in your head saying the words. In other words, even if you are making incredibly small movements with your lips/throat, can you reduce only the subjectively observable mental voice, and how can it affect your reading speed or experience.

I'm more interested to know if you "say" or "hear" the words in your head as you read.

Thought you made a great distinction there, but I think maybe you missed it.

"Saying" is not the same as "hearing", and in the subvocalization business, people don't ordinarily make that distinction. I think I hear, but don't say, when reading.

One way I tried to test this was by humming while reading. If I were saying, I'd expect that to interfere with reading, while hearing would not. Tried the same while literally biting my tongue.

Reading felt the same to me with either intervention.

2WingedViper8yI don't think it was meant as a distinction but as a description of a mental process that might not be exactly the same for everyone. So the dichotomy is between say/hear on the one side and not say/hear on the other.

Try counting out loud while you're reading (idea stolen from Feynman).

I can read perfectly fine while counting, but I can't type while counting. I am typing this message while re-reading your post.

That you can hear words while reading doesn't necessarily indicate that you are sub-vocalizing, as you may be perceiving words at the post-word-parsing stage, with the individual syllables having never been produced by any mental process.

1itaibn08yThat's strange. I thought I wouldn't be able to do it, but I tried and it turns out I can. It's difficult, though, and I'm not counting perfectly (there are repetitions and possibly skips). If I focus more on getting the counting right my reading ability degrades. I've also noticed on earlier occasions that I can read while listening to someone speak, this as well done imperfectly. EDIT: I realised you might have meant to count loud out. I actually just counted in my head. I'm now trying to count to out loud (even while writing this comment 290.. 291... 292..) I can still read (and even write). It's hard, I sometimes have pauses in my counting and I read with noticeably less attention and speed. 380.. 381.. 382
2private_messaging8yPerhaps we can use that as a gauge of sub-vocalization while reading, which would not be subject to different interpretations as much as the question whenever you sub-vocalize at all... i.e. if there is any interference between reading and counting out loud, then you subvocalize to some extent, if there's no interference then you don't subvocalize at all.
0hylleddin8yHuh, same here, it was much easier than I expected. Elsewhere in the comments, buybuydandavis noted a distinction between 'hearing' and 'saying', and I think that's what's going on here, for me it least. I say what I'm counting, but mostly hear what I'm reading. I can't read while listening to someone, so at least somewhat different things are going on between us.
0itaibn08yI tested this some more, and I'm not sure how much I can really do this. A large part of it seems to be reading in the gaps of speech when someone is speaking slowly or reading and not noticing I missed some portion of the speech.

I don't subvocalize when I'm reading a narrative account, and my readings speed is 800-1200wpm. When I read poetry, textbooks on unfamiliar subjects, or in a non-native language, I do subvocalize and my reading speed is reduced to approx. 600-700wpm.

When I stop subvocalizing, the narrative turns into action in my head. I don't know what words I'm reading, but I see the characters playing out the actions the text describes. It's more like a dream than a movie, in that the images are not well fleshed out, and I only see what I am focusing on. Sometimes... (read more)

1Torello8yI have this experience when I read and really get immersed. I have the feeling that a noticeable or persistent voice in my head is more of a symptom or signal of my distractedness than the cause. Also, how did you measure your reading speed, is your comprehension good at the upper end, and did you do anything to purposefully increase the reading speed?
0seez8yI measured it with the first two reading speed websites that came up on my google search here [] and here [] . My reading speed was also tested when I was 16, with very similar results. My comprehension is very good (in the 80-90% range) at the upper end for anything I can visualize, or when the details of the sentence do not change the meaning. For example, my comprehension would be very high for an action scene in a novel, but when I have tried to read that quickly about any non-familiar technical topic, my comprehension has fallen to the 20-50% range. I did nothing to increase my speed. I learned to read at a very late age, and became able to read over the course of approximately 20 minutes (it was quite possibly one of the weirdest experiences of my life), and after that 20 minutes passed I could with far above-average adult speed. If anyone else has ever had this experience, I would love to hear about it.

I don't sub-vocalize, as far as I can tell. I tend to read by just looking at the first letter of a word and then the shape of the word, rather then by reading phonetically; that's just how I taught myself to read, even before I stated learning to read in school. The upside is that I read much faster then most people seem to; the downside is that I'm terrible at spelling, and I tend to be bad at pronouncing words out loud if I've only read them and not heard them.

In fact, I've noticed before that I often think in written English rather then in audible English; I sometimes "see" words when I'm thinking to myself instead of "hearing" them.

0Douglas_Knight8yMost people read by recognizing whole words (mainly using first and last letters and the overall shape), so I don't think that's much evidence about whether you subvocalize. It is also probably not much connected to your spelling ability. That's not to say that I doubt your introspection about not subvocalizing any more than other people's introspection.
0Yosarian28yInteresting. I was under the impression that the way I read was more like how people are taught to do in speedreading classes, and that most people read more phonetically then I do. But I guess I don't really know which is more common.

I subvocalize, mentally, while "hearing" my thoughts, for example.

Is subvocalization is negatively correlated with getting a mental picture when reading fiction?

0hylleddin8yMy single datapoint says no. I almost always subvocalize, but get quite vivid pictures while reading.

Hello, I'm Italian and I apologize in advance if I will be fairly straightforward, but it's because I'm tired .. and the reason is this general paranoia on: "Oh, I can hear my voice pronouncing the words .. Am I normal? Am I not efficient enough?" creating a multitude of anxious people about their own level of efficiency. I remind you that I'm Italian, my country has produced the best scientists, engineers, architects when the problem of subvocalizzazione did not even exist. As for me, I stopped to consider it a problem .. I subvocalize everythi... (read more)

Here are two projects that try to remove subvocalization. It's fun to try at least.

If you aren't sure if you subvocalize while reading, try forcing yourself to imagine the words being read in a specific way - possibly in your friend's voice, or read in a certain easily stereotyped accent. Once you do that, you can see how different that feels from the reading you normally do.

When I try "reading in a Russian accent", my reading speed severely decreases, and the feeling is considerably more auditory than when I am reading with no gimmicks.

I do, but purely mentally. I have in past tried to avoid doing so, because it feels like a bottleneck in my mental speed, but I find I spend more time trying to not think the words than it saves me.

I'll add the weird datapoint that I normally subvocalize, but that I've been aware since age 10 that I can also read prose while singing along to music that I'm familiar with.

I just tried doing this right now and my observations yielded that I appear to not subvocalize when I do this and that I seem to therefore speed up and lose a bit of comprehension/retention. Which is going to be fine in lots of contexts, but is worth being aware of.

I've felt my OCD got worse after I learned touch typing during high school. Now I subvocalize every keystroke.

0[anonymous]8ySometimes that happens to me when typing passwords (even when they do contain pronounceable words) or hard-to-spell words.
0Torello8yFor example, when you type "the" you say "t, h, e" either under your breath or in your mind?
0polymathwannabe8yIt's less regular when I type in English, where not every letter corresponds to every sound, but in my native Spanish I do hear every letter in my mind as I type it.

That's a great question - I didn't realise I did that, and better yet did not realise I could turn it off. Though I am interested to see if there is information on reduced comprehension, as I actively and slowly subvocalize just about all maths texts, and doubt if I there would be much benefit changing that.

I subvocalize. Can't turn it off.

I can change the voice, though. Some phrases are hard to not read in someone else's voice, like I tend to want to hear Feynman when I see "interesting" because I really like the way he says it.

0ntroPi8yFor everyone who thinks he can't change the voice: Picture []
1Strilanc8yThat one didn't really work for me, but a bunch in the google picture results for "reading in my voice" [] do.
[-][anonymous]8y 0

I always do. I can't turn it off. I "hear" the words rather than saying them.

My top reading speed is about 900wpm, so it can't be slowing me down that much. When I read very quickly I only hear bits and pieces of each word though.

1Torello8yDid you measure your WPM online? If so, where? Also, doesn't your reading speed differ from book to book.

I can and often do skip the whole "hearing the text I'm reading" thing, but tend to enjoy slowing down and turning it back on for engaging, complicated, or fun texts. I also have a bad habit of skimming text instead of reading it if it's both boring and I'm not hearing what I read - I still get enough to decide whether or not it's worth remembering, just not enough to always recall it outright.

Augh! Normally I never subvocalize, but after reading this post I've started dong it uncontrollably and it's noticeably slowing down my reading speed. And my typing speed.

edit: well, not exactly never. I think I sometimes do it when I'm reading a challenging text while distracted/tired. It's usually a harbinger of the dreaded "same sentence over and over for 5 minutes", so I usually don't get far with subvocalizing. (Also, thankfully, it only took a few minutes of being distracted from this topic for the effect to go away. Whew.)

edit 2: even when this is in effect, I can still easily read while counting. I guess it's not really subvocalization after all.

I don't think I do much subvocalisation. There are certainly some words that I don't subvocalise: I often (like about once a week or a fortnight or so) have the experience of talking in person about a topic that I've previously only read and written about, and realising that I have never even tried to say key specialist vocabulary out loud, and so have no idea about how you pronounce it.

I don't subvocalise, and when I learned that other people do I was very surprised. A data point for subvocalisation being a limit on reading speed: I read at ~800wpm.

I always subvocalize everything.

I subvocalize parts of words and parts of sentences. Lumping everything from this strategy to hearing complete words and sentences under "subvocalization" might be a bit misleading. I stop noticing it when in deep concentration, but as you said this doesn't mean it's not happening. You stop noticing all sorts of stuff that's definitely happening when you concentrate, that's the point. Switching language from Finnish to English doesn't seem to have any effect.

As far as I can tell, I don't subvocalize while reading English.

I don't usually imagine a voice (my own or someone else's) when reading things. But I experience some kind of abstraction of having heard or said the word without a voice attached. I think out every syllable like this. Only when skimming do I not do this. My couple of experimental attempts to not do this while really getting the point of what I read just now were failures (I tried it on Slashdot news stories).

I always do. Mentally but not muscularly, and I can kind of suppress it if I consciously try. It is indeed the limiting factor on my reading speed.

2trist8yYou wouldn't notice the muscular subvocalizations. The easiest way to detect them is EMGs on the neck. I do get to a point where the external world fades away and (with fiction) I have much stronger auditory and visual sensations. I imagine I stop subvocalization during that, I certainly appear to read faster.