Read these two made-up words out loud, bouba and kiki.
Now picture two shapes, one all soft curves like a cloud, the other all sharp points like a shard of glass. Match each word to a shape before you read the next line.
You almost certainly gave bouba to the rounded shape and kiki to the spiky one. Nearly everyone does. The point of doing it yourself first is that you just felt something happen that you did not decide. You did not reason your way to the answer. The sounds arrived already carrying a shape.
That reflex has a name. It is sound symbolism, the non-arbitrary link between how a word sounds and what the mind assumes it means. It is the same reflex that decides whether a brand name feels fast or heavy, cheap or premium, in the half-second before a buyer reads a word of your copy. What follows is whether that reflex is real, what it is doing, and how to aim it.
Is it real?
Yes. It is one of the most reliably reproduced findings in the study of language, with a record that runs back almost a hundred years.
In 1929 the linguist Edward Sapir ran the experiment in its plainest form. He gave people the nonsense pair mal and mil and asked which named a larger table and which a smaller one. Across his groups, somewhere between 75 and 96 percent chose mal, the back-vowel word, for the larger object, with 83 percent agreement in his largest group of 124 teenagers. The vowel alone carried the size.
The shape version you just ran comes from the same era. The psychologist Wolfgang Köhler showed people a jagged figure and a rounded one alongside two invented words, takete and maluma, and found the same lopsided agreement about which sound belonged to which form. He had renamed the rounded word by the 1947 edition of his book, and the result still held.
The modern numbers are larger and harder to wave off. Ramachandran and Hubbard reported in 2001 that about 95 percent of people match the spiky shape to kiki and the round one to bouba. In 2022 a cross-cultural study put the effect to its hardest test yet, running it across 25 languages from 9 language families and 10 different writing systems. It held at a 72 percent match rate, far above chance, even among readers whose scripts look nothing like the Roman alphabet.
None of this is a branding myth or a copywriter's flourish. The effect is measured, and it has reproduced for nearly a century, across cultures and writing systems.
What your brain is actually doing
Knowing the effect is real matters less than knowing when it fires. The useful part for a founder is about timing and control. The inference is automatic, and it runs below the level you can switch off.
The clearest evidence comes from a 2004 study by Yorkston and Menon in the Journal of Consumer Research. They invented an ice cream and served it to two groups, changing exactly one thing, the name. One group got Frish, the other got Frosh. The only difference is the vowel, a front one against a back one. The group that ate Frosh rated the same ice cream as smoother, creamier, and richer, scoring it 5.06 against 4.25, and the effect held even when people were told it was a real brand name rather than a made-up one.
The part that matters most is how the inference behaved. Yorkston and Menon found it was automatic, outside awareness, and effortless. Their subjects were not sounding out the name and deducing a flavor. The meaning was already there before they noticed it. And in their study the effect grew stronger when people were distracted, not weaker.
Put that next to the timing. A buyer meets your name before they read your tagline, before the product shots load, before a single claim registers. In that half-second the sounds have already deposited expectations, fast or slow, hard or soft. Your copy then either confirms what the sound promised or argues with it. A name whose sound contradicts its positioning is not neutral. It is spending the buyer's first impression fighting itself.
Dial one, vowel frontness
For practical purposes the science reduces to two dials you can actually turn. The first is vowel frontness.
Vowels sit on a rough front-to-back scale set by where your tongue is when you make them. The ee in seen and the i in sit are front vowels. The o in go and the a in father are back vowels. Sapir's mal and mil told the first half of the story, back vowels read larger, front vowels read smaller. The fuller picture came from the study most people in branding point to as the one that moved sound symbolism into naming, Klink's work in 2000. Klink gave people invented brand names that differed only by vowel, like nidax and nodax, and found the front-vowel versions were judged smaller, lighter, milder, thinner, faster, colder, and more feminine than the back-vowel ones. One vowel, a whole stack of associations.
The result that should change a decision arrived in 2007, when Lowrey and Shrum showed the effect moves real preference, but only when the sound fits the thing. People preferred front-vowel names for small, sharp products like a convertible or a knife, and back-vowel names for large, heavy ones like an SUV or a hammer, by roughly a two-to-one margin. The dial does not make a name better in the abstract. It makes a name fit, or clash with, what you sell.
So the first question to put to a shortlist is plain. What does your product want to feel like, small and quick or large and substantial, and do the dominant vowels in the name point the same way. A featherweight running shoe named with heavy back vowels is turning the dial against itself.
Dial two, consonant voicing
The second dial is consonant voicing, and it separates the hard sounds from the soft ones by a difference you can feel with a finger on your throat.
Say the p in pat, then the b in bat. Your vocal cords stay quiet for the p and switch on for the b. The first is voiceless, the second voiced. In 2020 Pathak, Calvert and Lim tested this split across 25 invented brand names and found a steady pattern. The voiced consonants, b, d, g, z, and v, read as harder and more forceful. The voiceless ones, p, t, k, s, and f, read as milder and lighter. Voicing is the force dial. It sets how heavy or gentle a name hits, while the snap of a plosive and the shape of the vowels set how sharp or round it feels. A name can be light in force and still sharp at the edge, which is exactly what the brand reads below will show.
The most quoted naming decision in history turns on exactly this. George Eastman invented the word Kodak, and he was explicit about why he bookended it with the hard letter K, which he called a strong, incisive sort of letter. He built the name to three rules of his own, that it be short, impossible to mispronounce, and unlike anything else. This is the rare case where the founder put his reasoning on record, and the reasoning came down to the texture of one consonant. It is one man's account of one name, not a law that K means strength. It is also the cleanest founder-level evidence on record that the texture dial is something namers reach for on purpose.
The dials, read off four real names
Dials are easier to trust once you see them on names already on a shelf. Here are four, run through the two dials the way BrandNames reads its whole corpus. Each read follows from the sounds themselves, not from anything the founder has said about the name.
Poppi, the soda brand, is two syllables built on repeated voiceless plosives, the p sounds, opening on a plosive. Those plosives read fast and sharp, and the short clipped shape adds a lightness on top. It is a sound built to feel quick and bright, the right promise for a soda set against heavy legacy cola.
Turn to Quip, the oral-care brand, and you get the textbook small-and-sharp profile. One syllable, two voiceless plosives, a single front vowel. Every dial points the same way, fast and sharp and light, with the front vowel pulling it small, toward something neat and pocket-sized, which is what a slim travel toothbrush is.
Now both dials go the other way. Lalo, a baby-gear brand, has no plosives at all, just two liquid L sounds gliding over two open back vowels. The liquids and the back vowels read large, smooth, and heavy, the soft and substantial texture you want on a stroller a parent is trusting with a newborn. The same family of sounds that makes Poppi feel quick makes Lalo feel slow and solid.
Bombas, the sock brand, shows the back end of both dials together. Two voiced B plosives for force, two back vowels for weight, with a fricative s and a nasal m in the mix. The voicing and the back vowels read heavy, bold, warm, and rich, on a product whose whole pitch is cushioning and bulk. The sound is selling the same thing the socks are.
Four names, two dials, four different settings, each matched to what the product is for. None of it is decoration. It is the sound doing a job the copy has not started yet.
Two gates that decide whether it lands
Two conditions decide whether any of this lands, and a name can nail both dials and still fail on either one.
The first gate is fluency, how easily the name moves through the mouth and the eye. In 2009 Song and Schwarz ran a study that should sober up anyone in love with an exotic spelling. They gave people food additives and amusement-park rides with either easy or hard-to-pronounce names. The hard ones, things like Hnegripitrom, were judged more harmful and more likely to make people sick than the easy ones like Magnalroxate. The difficulty itself read as danger. A sharp sound profile buried in a spelling nobody can say does not read as sophisticated. It reads as a risk, and the buyer backs away without knowing why.
The second gate is vocabulary the buyer already owns. Sound symbolism fires inside words the mind can process, so the effect is only as strong as the name is readable. A front vowel cannot do its quick, light work if the reader stalls trying to decode the word it lives in. The interest has to sit in the connection between familiar sounds, not in reaching for an unfamiliar one.
There is a smaller, softer bonus on top of the two dials, worth a mention with the caveat that its pull is weaker than the vowel and voicing effects. English carries sound clusters that leak meaning of their own, what linguists call phonesthemes. The gl in glow, gleam, and glint hints at light. The fl in flap, fling, and flutter hints at movement. Firth named these in 1930 and Bergen showed in 2004 that they are psychologically real and stretch to brand-new words. Borrow one and a name picks up a faint, free association. Just do not lean on it the way you lean on the dials.
How to run this on your own shortlist
Here is the sequence to run on a shortlist you already have.
Say each name out loud, in a sentence, the way a customer would hear it rather than the way you read it in your head. Argo, Popa and Smith found in 2010 that names with sound repetition, spoken aloud, generate real positive feeling, though they also found you can overdo the repetition until it tips into a gimmick, so listen for both. Then name the one attribute your product most needs to own, fast or substantial, precise or warm, and check the dials against it, front vowels for small and quick and back vowels for large and heavy, voiceless consonants for light and mild and voiced ones for strong and forceful. Last, run the two gates. Can a stranger say it on the first try, and does every part read as something familiar rather than a word they have to decode. A name that passes the dials and clears both gates is not guaranteed to win, but it has stopped fighting itself, which is more than most names on most shortlists can say.
Two dials are not the whole of naming. Whether a name should be polarizing rather than safe is a separate decision and a different lever, and the companion guide on how to name a brand takes that one up.
Frequently asked questions
Is sound symbolism real or pseudoscience? Real, and heavily replicated. The bouba/kiki shape-sound match holds for about 95 percent of people (Ramachandran and Hubbard 2001) and at a 72 percent rate across 25 languages, 9 families, and 10 writing systems (Cwiek et al. 2022). The size effect goes back to Sapir in 1929.
What is sound symbolism in a brand name? The non-arbitrary link between a name's sounds and the meaning a listener infers, so the phonemes pre-load a set of product beliefs before the dictionary meaning is even read.
What is the bouba/kiki effect and why does it matter for branding? People match the rounded shape to bouba and the spiky shape to kiki about 95 percent of the time. For naming it means soft, round sounds read as smooth and approachable while sharp plosives read as precise and edgy, so the sound has to match the feel of the category.
Do front and back vowels really change how a product feels? Yes. Front vowels read small, light, and fast, back vowels large, heavy, and substantial. In brand-name choice the preference runs about two to one when the vowel fits the product (Lowrey and Shrum 2007), and a single vowel swap made an invented ice cream taste creamier (Yorkston and Menon 2004).
Does the effect work if people are not paying attention? It works better. Yorkston and Menon found the inference is automatic and outside awareness, and it got stronger when participants were distracted.
What makes a phonetically smart name backfire? Being hard to say. Difficult names are judged riskier (Song and Schwarz 2009), so a sharp sound profile buried in an unpronounceable spelling reads as a warning instead of a signal.
BrandNames applies sound symbolism and processing-fluency research to founders' real shortlists. The doors open soon. Drop your email to get one note when they do.