We are all capable of "hearing" shapes and sizes and perhaps even "tasting" sounds, according to researchers.
This blending of sensory experiences, or synaesthesia, they say, influences our perception and helps us make sense of a jumble of simultaneous sensations.
Oxford University scientists found that people associate lower-pitched sounds with larger and more rounded shapes.
One of the team is now working with chef Heston Blumenthal to incorporate words into a new dining experience.
Synaesthesia itself is a rare and unusual condition thought to affect less than 1% of the population.
It can takes many different forms - some people may "see sounds", in that certain sounds trigger them to see particular colours. Others might experience colours while reading those words in simple black text.
But according to Charles Spence, a professor of experimental psychology at Oxford University, we are all "synaesthetes" up to a point.
He and his co-author, Cesare Parise, tested 12 volunteers in trials during which an image flashed up on a screen at a slightly different time to one of two tones - one low-pitched and one high-pitched - being played.
There were two sets of image: a large and a small black dot, or an angular and a very rounded shape.
The volunteers had to say which came first - the image or the tone.
The larger dots and more rounded shapes are perceived to "match" the lower-pitched sounds. And this matching and mismatching affected how well people performed on the task.
"People are better at discriminating which came first when the sound and shape don't match," explained Professor Spence.
"When the sound and the image didn't match, people found it easier to keep them separate," he said. "Whereas with a congruent pairing - a small dot and a high-pitched sound - the participant's brain seemed to bind them together more."
The team also looked at volunteers' "spatial recognition". They played sounds either the left or right of the image, and discovered that people found it easier to work out which side the sound came from if it did not "match" the image.
It seems our brains may use these synaesthetic associations, says Professor Spence, "to combine all of the different sensory cues that are hitting our receptors at any one time".
"If there are lots of other visual events at the same time, for example, if I'm at a noisy party, how do I know which face goes with which voice?" he asked.
"We can match sights and sounds that come from the same position, or those that happen at the same time, but there are problems with this.
"If you think about thunder and lightning - as things get further away from us, the sound of the thunder gets separated from the visual event of the lightning.
"And if I move my head but not my eyes, or move my eyes but not my head, that's going to introduce some misalignment between my ears and eyes - between hearing and vision.
"So this synaesthetic correspondence is a third thing that the brain can use."
The idea of a particular word "sounding" sharp or soft - is not new. But this is the first time it has been shown to directly affect the perception of "non-synaesthetes".
Eat your words
Something that all synaesthetes have in common is that particular tones or words will always elicit precisely the same colours or tastes.
And Professor Spence thinks he can use this to enhance our experience of flavour.
The concept of sharp- and soft-sounding words was introduced in 1929, when Estonian psychologist Wolfgang Kohler designed an experiment that asked people to choose which of two shapes was named "bouba" and which was "kiki".
The vast majority of people choose kiki for the orange angular shape and bouba for the purple rounded shape.
Professor Spence thinks this strange language can influence our taste buds.
Working with world-renowned chef Heston Blumenthal, he is trying to directly combine an auditory experience into a dish.
"We've been giving people dishes and asking them questions about them, including is that food more of a 'bouba' or a 'kiki'? Or is it a 'maluma' or 'takete'?" he told BBC News.
He said that two of the best examples are brie, which is "very maluma", whereas cranberries are "very takete".
"The idea is that you get people to take part in the experiment by giving them two plates of food, and saying 'one of these is a takete and one is a maluma,' but not tell them which is which until they've eaten it.
The team may also, he says, make up a few new tasty-sounding words for dishes at Mr Blumenthal's restaurant. "We haven't decided which ones to use yet."