Small words with long histories
The words for one, two and three are among the oldest and most stable words in many languages. They are used constantly, learned early and rarely replaced, which is why linguists can often trace them back thousands of years. Yet number vocabulary is not frozen. Number words are built from other words, borrowed from neighbours, shortened by habit and sometimes swapped out entirely.
This article gives the overview: the main ways number names come into being and change. Several of the specific stories mentioned below have articles of their own.
Counting starts before the words
Before there were large vocabularies of numerals, people tracked quantity with tallies, fingers and objects. Evidence of prehistoric counting comes from notched bones and similar finds, though what the notches meant is usually uncertain. Spoken number words probably grew alongside such practices, but there is no direct record of the earliest ones, so any claim about the very first number word is speculation.
What we can observe is that many languages have only a few number words, and that others build long series from a short base. The research on languages with very few number words is discussed in its own article; the point here is that a numeral system is a cultural tool that can be extended, not a fixed feature of every language.
Hands, bodies and the base of ten
Why do so many languages count in tens? The usual suggestion is the ten fingers, and the pattern fits the evidence fairly well. In many languages around the world the word for “five” is the same as, or closely related to, the word for “hand”, and “ten” can be expressed as something like “two hands”. Some traditions extend this further, so that body counting runs up the arm, across the face and down the other side, as it does in parts of Papua New Guinea.
Other bases exist. Base twenty counting is found in languages such as Basque and Georgian, which may reflect fingers and toes together. Some cultures favoured twelve, sometimes with a thumb counting the finger joints of one hand, an idea explored under duodecimal counting. And the Mesopotamian sexagesimal system, based on sixty, left traces in how we still divide hours and angles. Whatever the base, the number words usually name a few building blocks and then combine them.
Compounding: building big numbers from small ones
Most numeral systems beyond ten are made by combining words. English “twenty” began as something like “two tens”, and “thirteen” is “three and ten” with the parts worn down. The way the pieces are arranged differs by language. English says “twenty-one”, but German says “one-and-twenty” (einundzwanzig), and older English used that order too.
Sometimes the building blocks are less obvious. English “eleven” and “twelve” look irregular, but they are old compounds meaning roughly “one left over” and “two left over” after ten, a story with its own history. French uses an arithmetic of twenties for eighty, one of the best-known quirks among French number names. Danish does something comparable for some tens. In each case a numeral that looks arbitrary turns out to be a small calculation fossilised in speech.
Shared inheritance: related languages, related numerals
Languages that descend from a common ancestor often share number words. English “two”, Latin “duo”, Greek “duo” and Sanskrit “dva” belong to one family, as do “three”, “tres” and “tri”. This is the evidence that helped establish the Indo-European numerals as cognates from a reconstructed ancestor, and similar comparisons work in other families. Regular sound changes explain why the words differ: where Latin has “pater” and “piscis”, English has “father” and “fish”, and the same shift turns an ancestral p-sound into the f in “five”.
Numerals are useful for this work because they are common, short and resistant to replacement. They are not immune, though. When two related languages have very different numerals, that usually signals a replacement, a borrowing or a long separation.
Borrowing: when a number word travels
Number words do get borrowed, especially larger ones and ones attached to trade and administration. English “million” came from Italian via French, and “zero” and “cipher” both go back to an Arabic word for “empty” that travelled through Italian and Latin along with the Hindu-Arabic digits. Those digits themselves have a long history, from Brahmi numerals in India to the forms used in Europe after medieval mathematicians popularised them, which belongs to the wider story of numeral systems.
Sometimes a whole second set of numerals is borrowed and kept alongside the native one. Japanese has a native series (hitotsu, futatsu, mitsu) and a Chinese-derived series (ichi, ni, san), each used in different contexts, the arrangement behind the Japanese number readings. Korean has a parallel arrangement with native and Sino-Korean sets, and Vietnamese also has numerals of Chinese origin. In such languages the choice of series depends on what is being counted, which is an unusually visible example of how borrowing can reshape an everyday system.
Grammar around the number
Number words do not exist in isolation, and the grammar around them varies a great deal. Some languages mark exactly two of something in the noun itself, a feature described under grammatical number; a few have extended this to three. Others require a counter word between the numeral and the noun, as Chinese, Japanese and Korean do, which is the topic of numeral classifiers. Even the words for “first” and “second” follow rules of their own, as with ordinal numbers. These features do not create the number words, but they shape how people use them.
Drift, wear and surprise meanings
Once in use, number words keep changing. Sounds wear down, so that “thirteen” and “fifteen” have shortened first parts. Meanings drift: “decimate” originally referred to one in ten, and “quarantine” to a period of forty days, a pattern shared by many number-derived words. Fixed phrases can also keep older counts alive, such as “score” for twenty in “four score”, or units like the dozen and the gross that survive in trade as traditional counting units.
There are also differences in how large numbers are named. English “billion” has not always meant the same quantity in every country, because usage has differed between the so-called short and long scales. The word may be shared across languages while the number it names is not.
What can and cannot be concluded
Etymology is careful detective work, and it has limits. Some number words have well-supported origins; others have several competing explanations or none. Reconstructed ancestral forms are scholarly hypotheses, written with an asterisk for that reason. Claims that a particular number word secretly “means” something deep or reveals the nature of the number should be treated cautiously. Such readings belong to the history of beliefs, not to the evidence of how the words were formed.
The general picture is more modest and more interesting. Number words are built from hands and bodies, from combination and subtraction, from inheritance and borrowing. They are worn down by use and sometimes renewed. Following any one of them back through time usually shows how ordinary people solved the practical problem of saying how many.