Ask most people whether counting is natural and they will say obviously yes. Every culture counts. Children learn numbers early. It feels less like a skill than a faculty.
The evidence says otherwise, and the clearest case comes from a small community in the Brazilian Amazon whose language contains no method of expressing an exact quantity of anything.
Not a limited system. Not a system that stops at three. No exact number words at all, including no word for one.
That is an unusual natural experiment, and what researchers found when they tested it is more interesting than either side of the argument expected. Here is what happened.
What the Language Actually Does
The community concerned numbers a few hundred people living in villages along a tributary of the Amazon, and their language has been studied intensively for decades.
The initial finding, published in 2004, was that the language had words for one, two and many — a limited numerical system of a type reported in a number of foraging societies.
A subsequent MIT-led study revised that substantially, and the method used to revise it is worth explaining because it is a good piece of experimental design.
The earlier work asked speakers to describe sets of objects as items were added, counting upward from one. That produces a consistent-looking result: a word for the first object, a different word when the second arrives, and a third word thereafter.
The later team ran the same procedure in reverse, starting with ten objects and removing them one at a time. If the words meant one, two and many, the same words should appear at the same quantities.
They did not. Counting downward, speakers used the supposed word for two when as many as five or six objects were present, and the supposed word for one for any quantity between one and four.
That is not a counting system. Those words signify relative quantities — something closer to few, some and more — and their meaning depends on the comparison rather than on any fixed amount.
The conclusion was that the language has no linguistic method whatsoever for expressing exact quantity, not even one.
Like our content? Follow us for more.
What the Speakers Can Do
The obvious next question is what happens to numerical ability when the words are absent, and this is where the results become truly surprising.
The strong version of the theory — that language creates the concept of exact quantity, and that without number words a person can only represent approximate amounts — makes a clear prediction. Speakers should be unable to handle exact quantities at all.
That prediction failed. In a matching task where participants were shown a line of objects and asked to lay out the same number, speakers performed accurately, including with large quantities. Faced with a line of up to ten items, they reproduced it correctly.
So the concept of exact quantity is available without the words for it. A person can perceive that a set has a specific size and reproduce it, using nothing but direct comparison.
What They Cannot Do
The failures are equally informative, and they fall into a consistent pattern.
Performance collapsed on tasks requiring memory. When the original set was hidden, or when the match had to be made from recall rather than by direct comparison, accuracy dropped sharply as quantities increased.
The same applied to tasks where the objects could not be lined up alongside one another — where the correspondence had to be tracked mentally rather than laid out physically.
So the boundary is not between approximate and exact. It is between quantities you can currently see and quantities you have to hold in your head.
That distinction is the whole finding, and it produces a substantially better theory than either of the ones being tested.
Number Words as a Technology
The interpretation the researchers offered is that number words are a cognitive technology.
On this account, a number word does not create the concept of a quantity. It creates a durable handle for one — a compact label that can be remembered, carried away from the objects, communicated to somebody who was not present, and compared with a label attached to a different set at a different time.
Seven is not a perception. It is a token that lets a perception survive being looked away from.
That reframes counting as an invention rather than a faculty, comparable to writing. Writing does not create the ability to have thoughts; it creates a way of storing them outside a head. Number words appear to do the same job for quantities.
It also explains the distribution. If number words are a tool, then societies develop them when the tool is needed — for trade, taxation, herding, storage and any activity requiring quantities to be tracked over time or agreed between people. A community whose economy does not require any of that has no particular reason to develop them, which is roughly how one of the researchers characterised the situation: not that the ability is absent, but that the technology has not been adopted because it has no application.
The Systems Everyone Else Built
Set the extreme case aside and the variation among the systems that do exist is substantial, which reinforces the technology reading.
Most languages count in tens, which almost certainly reflects the number of fingers available rather than any mathematical merit. Base ten has no particular advantage; it simply matches the equipment.
Others did not. Base twenty systems, counting fingers and toes together, appear in several parts of the world and leave traces in languages that have since converted. Base twelve has been used, and has real advantages, since twelve divides evenly by two, three, four and six where ten divides only by two and five.
Some systems count in body parts, running up one arm, across the head and down the other side, producing a sequence of named locations rather than abstract numerals. Some use pairs, so quantities are built from twos rather than from a fixed base.
The structure of a system measurably affects its users. Languages where the words for eleven to nineteen are transparently built from ten and a unit produce faster acquisition in children than languages where those words are irregular and have to be memorised individually.
That is a small but real effect, and it makes the general point neatly. If counting were a faculty, the words would be a labelling exercise. Because it is a technology, the design of the tool changes how well it works.
Why the Argument Is Not Over
Some caution is warranted, because this is a small and much-debated literature.
The community is small, the language is extremely difficult for outsiders to learn, and the number of researchers who have worked with it directly is very limited. Findings from a single community do not establish universals.
There is also a real and continuing disagreement. Later work by another researcher, repeating the field experiments of both earlier studies, reported that speakers had difficulty with even simple one-to-one correspondence tasks — which is closer to the stronger claim and harder to reconcile with the matching results.
So the field has produced results pointing in somewhat different directions, using overlapping methods with the same community, and the disagreement has not been resolved.
What is not seriously disputed is the linguistic finding itself. The language has no exact number words. Whether that limits what its speakers can do, and how much, is where the argument continues.
There is a further point that deserves stating plainly. This community has been studied a great deal by outsiders, and the framing of that research has sometimes been questionable. The absence of number words is a fact about a language, not a deficiency in the people who speak it, and the research question is what number words do rather than what anybody lacks.
What It Says About the Rest of Us
The reason this matters beyond one community is what it implies about a system everybody else uses without noticing.
Counting feels like perception. It does not feel like operating a device. But if the ability to hold exact quantities across time and distance depends on having words for them, then everyone reading this is using an inherited piece of equipment that had to be invented, taught and transmitted.
Children spend years learning it, which is a substantial clue. Nobody spends years learning to recognise faces or to walk.
And the equipment is not the same everywhere. Number systems differ enormously — in base, in structure, in how they build large values from small ones — and those differences measurably affect how quickly children acquire them and how easily certain operations are performed.
Which suggests the honest version of the original question. Counting is not a human universal in the way that seeing or hearing is. It is a very old, very widely adopted and extraordinarily useful invention that a small number of communities never had a reason to take up.
Like our content? Follow us for more.
