I've read a similar question here which mainly dealt with why English only has eleven and twelve as unique words with some interesting ideas. But my question is why do different languages have different cut-off points for unique words. Here I am using unique just to mean words which follow a different pattern to simply number + the word for ten, or which are formed differently to the other numbers between 10 and 20. French for example goes up to 16 before saying 10-7, 10-8 etc. and I just found out that Spanish and Portuguese go up to 15.

Checking online I can see that Catalan and Italian follow French, going up to 16, while German, Dutch and Nordic languages follow English and stop at 12 . Greek also acts similarly and has 11 and 12 with different forms to the other numbers. Gaelic, Romanian and Hungarian have no unique words between 10 and 20. Finnish and Czech appear to also have no unique words (or all unique words depending on your outlook: they use a suffix different to the word ten). No offence to it, but Welsh is a mess. Latin itself has 11-17 following one pattern, and 18 and 19 with a different pattern.

So it seems that in Europe only Germanic and Romance languages have unique words between 10 and 20. Modern Germanic languages all have 11 and 12 as unique words, while modern Romance languages are split between 11-15 (Spanish and Portuguese), 11-16 (French, Catalan and Italian) and none at all (Romanian).

I'm mostly curious as to why the numbers 10-20 seem to have more differences between languages than any other set of numbers (after 0-10 of course). And why these words in Romance languages seem to have splintered due to geography while in Germanic languages they haven't?


2 Answers 2


Eleven and twelve are remnants of a duodecimal system which still shows up in words like dozen. There have to be 12 unique base units for duodecimal, just as there have to be 16 for hexadecimal.

They're not always present, however. And the other numbers between 10 and 20 show quite a lot of variation, even in European languages, as you point out for various Romance languages.

Probably the main generalization here is that numbers, and number patterns, often repeat the

  • One - Singular
  • (Two) - (Dual)
  • Many - Plural

pattern found in, for instance

  • Singular (Dual) Plural
    Dual words and morphology are irregular and arbitrary (e.g, Eng glasses, pants, both, pair, couple)
  • 1-syllable Comparative bigger (2-syllable easier/more solid) Polysyllable more interesting
    Disyllabic comparatives are irregular and arbitrary

as well as many other linguistic phenomena, which suggests that duality of one kind or another is semantically unusual, and has some of the irregular and arbitrary characteristics of the numeral one (which is often extended for use as a demonstrative, a pronoun, or an article, and which has unique syntax in every language); as well as some of the regular characteristics of compositional numbers like seventy-three.

Dual is transitional between singular and plural, and it's semantically significant because we are bilaterally symmetric animals and therefore have plenty of reason to refer to things that come in pairs. Plus there are two sexes which mate in pairs, and that's an area rife with irregularity and arbitrariness.

This extends to decades (for 10-based systems; I'll use the term for all recurring sequences, of whatever base) as well as numbers. In any system, the first decade is usually composed of unique base units (though not always; in Ute, for instance, nine means 'almost ten'). The second decade is full of irregularities, but not usually new bases. The third and following decades are usually very regular.

  • Did old Indo-European peoples use a base-12 number system? Or is it ancient European languages, from before the Indo-Europeans coming in? I don't know what modern Basque uses or what it's ancient version did, that could settle the question. I do know old Celts used a base-20, just like ancient Mesoamericans, are numbers 10-20 irregular in modern Celtic languages?
    – Joe Pineda
    Dec 12, 2013 at 15:56
  • 2
    Most cultures use several systems, or one productive system with remnants of other systems showing up as words like dozen, quarter, score, etc. In Yucatec Mayan, for instance, the Spanish numeral system has completely replaced the Mayan numerals, except when talking about the Long Count, when they all come back again.
    – jlawler
    Dec 12, 2013 at 18:23
  • "Remnants of a duodecimal system which still shows up in words like dozen.". Actually, "dozen" is a recent borrowing from French or Italian. It is thus not a remnant of a duodecimal system, but at best a re-introduction of an extinct duodecimal system. "Still" is the wrong adverb.
    – fdb
    Jul 23, 2021 at 10:09

In Chinese number names, and both Korean number systems, the numbers follow strict base 10 representations: the name for any number from 11-19 is the name for 10 plus the name for the unit value.

  • e.g. the Chinese name for 14 - 十四 (Shí sì) - can be literally translated as "10-4".

But this strict regularity doesn't seem to exist in the teens of most European languages:

  • In English and German, the words for 11 and 12 have unpredictable names only historically related to 1 and 2 (from Proto-Germanic *ainalif (one left), and *twalif (two left)).

  • In most Romance languages there is some phonetic representation of 10 in the names for teen numbers (albeit heavily modified in Spanish and French), but this representation 'flips' somewhere along the way (between 15-16 in Spanish and Portuguese, and between 16-17 in Italian and French etc). See below:


English Latin Spanish Portuguese Italian French
zero nihil cero zero zero zéro
one ūnus uno um uno un
two duo dos dois due deux
three trēs tres três tre trois
four quattuor cuatro quatro quattro quatre
five quīnque cinco cinco cinque cinq
six sex seis seis sei six
seven septem siete sete sette sept
eight octō ocho oito otto huit
nine novem nueve nove nove neuf
ten decem diez dez dieci dix


English Latin Spanish Portuguese Italian French
eleven ūndecim once onze undici onze
twelve duodecim doce doze dodici douze
thirteen trēdecim trece treze tredici treize
fourteen quattuordecim catorce catorze quattordici quatorze
fifteen quīndecim quince quinze quindici quinze
sixteen decim diez y seis dezesseis sedici seize
seventeen septendecim diez y siete dezessete diciassette dix-sept
eighteen duodēvīgintī diez y ocho dezoito diciotto dix-huit
nineteen ūndēvīgintī diez y nueve dezenove diciannove dix-neuf
twenty vīgintī veinte vinte venti vingt

(Note: Catalan, Sicilian, Friulian follow the same pattern as French and Italian.)

Classical Latin itself is regular up until 18 and 19, where representations translated as "2 from 20" and "1 from 20" are the norm (if this seems unexpected, remember the roman numeral representation for 19 is XIX, "10 and 1 from 10").

While it might seem strange that the 'flip' in the Romance languages representations wasn't present in Classical Latin, switched forms such as "decem et septem" for 17 can be found in some Latin texts, and was a feature of some rural and lower class Latin speech that was presumably particularly prevalent among the settlers and soldiers, whose speech directly affected modern Romance languages.



Your Answer

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