When speakers of a language coin words for one, two, three, four, etc., for the first time, where do they come up with the forms? Are there any common methods used across language families?

Pirahã has words for "few" and "many". Numerous other languages have only "one" and "many", possibly adding "two", especially among isolated hunter-gatherer tribes. If they do express numbers higher than that, it's by counting on their fingers in front of someone else. (See chapters 1 and 2 of The Number Concept, written by Levi Leonhard Conant in 1896.) But once people invent agriculture and trade, languages tend to come to have some system for counting and expressing, say, the number of animals in a flock or the number of days' travel to some other village.

If you can get past the use of "savage" and "rude" to refer to hunter-gatherers and "Aryan" for Indo-European, Conant continues to describe a common strategy for "five" through "ten". The word five is often some variant on "having finished one hand"; this is true of PIE as well where *pénkʷe (five) is related to "finger". Ten can be "having finished both hands" or (in the case of shod cultures) "one man", and six through nine "one more [than five]" through "four more"; nine is just as often "one less [than ten]". Barefoot cultures often adopt base 20 ("one man") with sub-base 5.

Some languages borrow numbers when their speakers borrow other cultural concepts. For example, most European languages borrowed "zero" from Arabic, and Swahili appears to have borrowed six through ten from the same language. Proponents of the Nostratic hypothesis will cite similarity between "six" and "seven" in IE and Semitic as evidence of common descent, if those words weren't also borrowed.

Conant lamented that the names of one through four appeared opaque in the vast majority of languages studied. Have the past 120 years produced any more clarity on trends in number word etymology among several language families? I tried Google basic numbers etymology but most results appeared irrelevant, most simply tracing English back to PIE or reiterating that zero was borrowed from Arabic.

  • Some progress must have been made since 1896. I just googled "language number system" and got 362 million hits. Here's an interesting looking article from SIL: www-01.sil.org/sil/news/2010/4000-number-systems.htm (SIL has sponsored a vast amount of work on the lesser known languages of the world).
    – Greg Lee
    Commented Feb 8, 2015 at 20:48
  • 2
    @GregLee The news article from SIL links to Eugene S. L. Chan's website which links to an article "Typology of Numeral Systems" by Bernard Comrie. Comrie's article largely agrees with Conant on strategies used for building names of larger numbers but doesn't touch on where languages get their words for, say, 1-4. Did I miss some related material on Chan's site? Commented Feb 9, 2015 at 1:36
  • I haven't looked at Chan's website or Comrie's article.
    – Greg Lee
    Commented Feb 9, 2015 at 2:05
  • Are you interested only in number words in PIE? There has been a great amount of work on numbers, counting, and enumeration in the world's languages. Do you have a specific question?
    – Teusz
    Commented Feb 9, 2015 at 10:08
  • 1
    Is *pénkʷe in any way related to "pinky"?
    – Zgialor
    Commented Feb 21, 2015 at 21:08

4 Answers 4


Piotr Gąsiorowski has very recently put forth a hypothesis on the origin of Indo-European kwetwor "four."

'Four: A Map'

He argues that it comes from an archaic word for "pair" or "to pair." The semantic evidence is good; unfortunately, there are not many branches with a non-"four" reflex of kwet in them.

  • 3
    "there are not many branches with a non-"four" reflex of kwet in them" - hmm Russian чета "cheta" meaning a spouse pair is a clear reflex of PIE q̆eta̯ "pair". Russian has other reflexes as well: сочетание "composition", четный "even", счёт "count", читать "to read", число "number" (< PIE q̆et(s)tlom). Also evidently the Latin clitic -que is related
    – Anixx
    Commented May 28, 2015 at 22:55

This is a common cross-linguistic trend. I remember reading a section of a book about it - in fact, I was searching for that book when I found this thread - but alas, I can't seem to find it.

As a non-Indo-European example, Yu'pik does something similar, but a little more thorough. Very roughly:

  1. "that which is indivisible"
  2. "following unit"
  3. "addendum"
  4. "spread out"
  5. "arm"
  6. "one having crossed over"
  7. "one having two" (?)
  8. "one having three" (?)
  9. "not quite ten"
  10. "above"
  11. "it goes down"
  12. "ten two"
  13. "ten three"
  14. "not quite fifteen"
  15. "other side"
  16. "fifteen one"
  17. "fifteen two"
  18. "fifteen three"
  19. "not quite twenty"
  20. "whole person"

When a number is used in the gloss, obviously, it can be broken down to its own constituents, so 12 is "following unit above", etc. The only exception is that the "one" in 6-8 is a pronoun, similar to "something".

http://www.languagesandnumbers.com/how-to-count-in-yupik/en/ypk-esu/ (this source translates 7 and 8 as "two having crossed over" and "three having crossed over", but this appears to be in error) https://books.google.ca/books?isbn=311027857X (pp. 398-401)


I find it interesting that Indo-European numerals from 1 to 4 are declinable and have ablaut nature; meanwhile numerals 5 and higher are indeclinable "stumps". The only explanation is that numerals 5 and higher were used in simple sequantial counting: 5, 6, 7, 8, perhaps in trade with non-Indo-European tribes. *septm being an obvious borrowing suggests that all the numerals from 5 to 10 could be loanwords from some sort of ancient pidgin, and Indo-Europeans themselves casually counted only up to 4.

  • In Sanskrit all the numbers are declinable.
    – fdb
    Commented Sep 5, 2016 at 19:03

If one puts oneself into the situation of an early age where the invention of words for numbers begins there is only one way to do it, by paraphrasing. You can express five by saying as many as fingers on your hand. Two: the number of your arms or feet, etc. Four: the legs of a cow or a dog One: the head of a man or head/tail of an animal

If signs and geometrical signs have been developed it is possible to say: Three: the corners of a triangle Four: the corners of a rectangle or square. Latin quattuor (four) is very similar to quadratum (rectangle) or quadrus/a/um (rectangular).

If words for one, two, three have been developed it is easy to express more numbers: Ten: two hands - Behind Latin decem (ten) might be two hands. Seven: a hand + two. Nine: two hands minus one. Eight: two fours.

The further development of paraphrases will be shortening and contraction till a word form that is short and easy to speak comes into being.

Probably the first stage was by means of gestures, such as showing the hand, pointing to the nose or to the legs. Maybe conventions developed for addition and subtraction of two simple numbers.

  • 6
    "Latin quattuor (four) is very similar to quadratum (rectangle) or quadrus/a/um (rectangular)." Rectangle is called so because it has 4 sides, not other way around!
    – Anixx
    Commented Feb 10, 2015 at 15:05
  • 4
    This answer appears to be pure speculation. -1
    – Colin Fine
    Commented Mar 24, 2015 at 11:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.