By 
Before you go any further, try something. Read the next paragraph out loud, in your actual voice, at whatever volume your surroundings allow.
I ask because I do this with my own drafts. When I compare the pieces I've read out loud with the ones I haven't, the difference is noticeable. Some of the silent ones read like a bunch of information. The ones I've spoken tend to lead even me to reflect.
I have to admit — I talk to myself out loud too, in a mix of Georgian and English. Which language comes out depends on who I have in mind, so the talking is really an imagined conversation with a particular listener.
That's my experience, and I can't measure it. Memory, though, has been measured. In a study of 95 people by Noah Forrin and Colin MacLeod at the University of Waterloo, material people read aloud to themselves stuck better than material they read silently, heard someone else read, or heard in a recording of their own voice.
Psychologists call it the production effect.
So you have just talked to yourself. The research on what that does is more interesting than the old line that it's the first sign of madness.
A habit noticed in the supermarket aisle
In 2012, Gary Lupyan, a psychologist at the University of Wisconsin–Madison, and Daniel Swingley of the University of Pennsylvania published a paper that started from something everyone has seen. "While searching specific objects, people often repeat the name of the object," they wrote. The press release for the study described someone muttering while hunting for a jar of peanut butter on a supermarket shelf.
Does the muttering do anything? In the first of three experiments, 24 students searched a screen of drawings for one target, such as a banana or a zebra. Before each search, they either said the target's name aloud or read it silently. Speaking made them faster by about 50 milliseconds, on searches that took about 1.4 seconds on average, and slightly more accurate.
Fifty milliseconds is a twentieth of a second: reliable, and small. A LiveScience report at the time put the gain at 50 to 100 milliseconds, more generous than the paper.
When saying the name gets in the way
In the second experiment, 12 students kept repeating the target's name while they searched. Overall, they were no faster: slower at first, quicker only in the last three of eight rounds, though slightly more accurate overall. Speaking helped with familiar objects and with pictures that matched the image the word called up. For pictures that matched the word less well, it slowed people down.
The third experiment came closest to everyday life. Twenty-two students did a "virtual shopping" task using photos the researchers had taken on Philadelphia supermarket shelves, finding all three images of a product among 13 distractors. Saying the name brought no overall speed gain, and people missed slightly more items when they spoke (a 97.9 percent hit rate, against 99.1 percent). Speaking helped with familiar, easy-to-picture products, especially ones with a typical colour, like bananas or Cheerios. It hurt for categories whose items look different from one another.
And nobody in any of the experiments had lost anything. They were looking at pictures on a screen.
What a spoken word does to your eyes
Lupyan's explanation is called the label feedback hypothesis. In the paper's words, "actually hearing 'chair' compared to simply thinking about a chair can temporarily make the visual system a better 'chair detector'." On this account, saying a word helps "activate and/or keep active" the visual features typical of that category. The word calls up something like the most typical banana you know, and your eyes check what they see against it.
If the thing you want looks typical, the word points you right at it. If it looks unusual, the word points you at the wrong picture. The authors gave an example from airport security:
Repeating the word "knife" may "help an airport baggage screener spot typical knives, but actually make it more difficult to find less prototypical knives." (Lupyan and Swingley, 2012)
The authors were upfront about what they couldn't tell: whether speaking helped people locate the target, or only recognise it once their eyes landed on it. In 2021, Katherine Hebert, Stephen Goldinger and Stephen Walenchok at Arizona State University tracked people's eyes to find out. Across three experiments, saying the target's name helped, while saying nonwords or irrelevant names did not. The eye movements showed that speaking didn't steer people's gaze toward the target. It helped them recognise the target and dismiss the distractors.
Not everyone hears a voice in their head
Part of why talking aloud can look odd is the assumption that everyone has a running inner monologue. When Christopher Heavey and Russell Hurlburt beeped 30 people at random moments and asked what was in their minds, inner speech showed up in 26 percent of the samples, ranging from none at all to 75 percent between individuals.
In 2024, Lupyan and Johanne Nedergaard, then at the University of Copenhagen, gave the low end a name: anendophasia, the lack of inner speech. They compared 46 adults who reported little inner speech with 47 who reported a lot. The low group did worse at remembering lists of words and at judging whether the names of two pictures rhymed. Then came the detail I keep returning to: the differences disappeared for people who said things out loud while they worked. Talking aloud seemed to do the job the inner voice usually does.
Would people with little inner voice gain more or less from saying a name aloud in those 2012 experiments? Here's what Lupyan shared with me:
"I would expect the effect of talking to be larger in people with less inner speech. We actually found this in the rhyme-judgment and verbal memory tasks in that 2024 paper. People who had less inner speech but who reported talking out loud during the task looked just like the people with more inner speech. For people with more inner speech, talking out loud didn't make much difference."
He described talking aloud as a "compensatory" mechanism for people with less spontaneous inner speech.
Do people with little inner voice lean on talking aloud in everyday life? Nedergaard answered me with what people have told her:
"I have received many informal reports from people with anendophasia to suggest that they do indeed 'externalise' (talk out loud or write things down) to reap some of the benefits that people with inner speech get from talking to themselves in their heads," she told me, adding that this is "precisely one of the things we plan to study next".
And in the 2024 study, people with and without an inner voice were just as likely to talk aloud. "There were no differences in how likely the participants in the two groups were to do so - this was completely on their own initiative," she told me. So for now the everyday habit comes from people's own reports, and measuring it is one of the things she plans to study next.
Still, it turns out the field is working out where the low end sits. In a 2025 commentary, Andreas Lind of Lund University argues there is "no convincing evidence" that people without a speech condition lack inner speech "completely". The debate is over whether the low end reaches zero; that people differ is common ground.
For some people, the low end is daily life. The copywriter Jenny Lucas has written that she has no inner monologue: "If I want to hear my thoughts I have to say them out loud, which explains why I talk to myself so much."
She told me where it happens most: "I talk to myself most in the supermarket, when I'm planning our meals for the week; in the car, when I'm driving and thinking about where I need to go; in the kitchen, when I'm following a recipe; and at work, when I'm deciding what to write."
Lost things are her exception. "If I lose something at home, I tend to picture it more than say it," she said. "I try to imagine it in the last place I saw it."
If Lupyan is right that a spoken word works by calling up an image of the thing, Jenny, who has also written about her unusually vivid mental imagery, seems to skip the word and go straight to the picture.
Chris Turner, who helps people recover lost rings and jewellery at The Ring Finders, sees a search from the other side. "I don't notice so much that people are talking to themselves," he told me, "because I ask a ton of questions on where they believe they may have lost the ring in order for me to find it." Some people know exactly where, he said, and others aren't sure. His search for something truly lost starts with where.
Lupyan starts from the other end, with what the thing looks like:
"So saying 'keys' would help to activate visual knowledge of keys (presumably, your keys because the meaning of 'keys' is always contextualized)."
He doesn't think the search has to start with remembering where it was:
"You don't have to remember where it was (if you did, you wouldn't need to search in the first place), but the effect of self-produced speech should very much depend on how well the person knows what the thing they are searching for looks like (according to our results, anyway)."
His answer is reasoning from the 2012 results, where speaking helped with familiar objects, not a test of searching for lost things at home.
My own muttering to an imagined listener fits an old idea from the Soviet psychologist Lev Vygotsky. As a 2015 review by Ben Alderson-Day and Charles Fernyhough sums up his model, a child's exchanges with other people turn into "an internalized 'conversation' with the self."
Talking to a cupboard, scientifically speaking
Talking to yourself is common, and in these studies it often did useful work. Saying a familiar object's name often helped people spot it, and saying words aloud filled in for a quiet inner voice. Researchers at the University of Texas at Austin who recorded 333 older adults in short snippets over several days caught 918 moments of solo speech among 109,609 snippets: under 1 percent. Ordinary, and far from constant.
Still, I'm not saying any of this can work as a trick for finding lost keys. The 2012 speed gain was a twentieth of a second, and when the word didn't fit the object, speaking made things worse. This research doesn’t diagnose anything, in either direction. Talking aloud is one tool among several, and it works best when you know exactly what you're looking for.
I've just moved to Germany, into a home that came full of other people's stuff: their mugs, their drawer logic, their firm opinions about where scissors live. There are hundreds of spices I have no idea how to use and plenty of objects in the drawers I can't even name. Nothing is where I'd put it, mostly because I haven't worked out where I'd put it. So I've started narrating. "Tape. Tape goes… here? No. Not here."
At this point, it feels absurd — a grown person negotiating with a cupboard.
But if the research is right, saying the name of the thing should help me find it, and saying where it goes might help me remember.
We'll see. I'll report back, possibly in two languages, depending on who I imagine is listening.
