Following Sklivvz's advice, I propose here a question I made in Italian Language. Because I am not sure how I should do this, I will just copy/paste the whole lot.

Let's count in Latin from one to twenty:

ūnus/ūna/ūnum, duo/duae/duo, trēs/tria, quattuor, quīnque, sex, septem, octō, novem, decem,

ūndecim, duodecim, tredecim, quattuordecim, quīndecim, sēdecim, septendecim, duodēvīgintī, ūndēvīgintī, vīgintī

As pointed out by symbiotech, "octodecim" and "novemdecim" were also used in Latin, but they didn't survive. On the other hand, as pointed out by martina, "dĕcem (et) sĕptem" was also a common form for "septemdĕcim".

In Attic Greek it was:

ΕΙΣ/ΜΙΑ/ΕΝ (heis/mia/en), ΔΥΟ (dúō), ΤΡΕΙΣ/ΤΡΙΑ (treis/tria), ΤΕΤΤΑΡΕΣ/ΤΕΤΤΑΡΑ (téttares/téttara), ΠΕΝΤΕ (pénte), ΕΞ (héx), ΕΠΤΑ (heptá), ΟΚΤΩ (oktṓ), ΕΝΝΕΑ (ennéa), ΔΕΚΑ (déka),

ΕΝΔΕΚΑ (héndeka), ΔΩΔΕΚΑ (dódeka), ΤΡΕΙΣΚΑΙΔΕΚΑ (treiskaídeka), ΤΕΤΤΑΡΕΣ ΚΑΙ ΔΕΚΑ (téttares kaì déka), ΠΕΝΤΕΚΑΙΔΕΚΑ (pentekaídeka), ΕΚΚΑΙΔΕΚΑ (hekkaídeka), ΕΠΤΑΚΑΙΔΕΚΑ (heptakaídeka), ΟΚΤΩΚΑΙΔΕΚΑ (oktōkaídeka), ΕΝΝΕΑΚΑΙΔΕΚΑ (enneakaídeka), ΕΙΚΟΣΙ(Ν) (eíkosi(n))

Now let's count in Italian

uno, due, tre, quattro, cinque, sei, sette, otto, nove, dieci,

undici, dodici, tredici, quattordici, quindici, sedici, diciassette, diciotto, diciannove, venti.

But numbers from eleven to twenty could also have been, just hypothetically of course (adding accents for clarity's sake):

diciùno, diciaddùe, diciattré, diciacquàttro, diciaccìnque, diciassèi, diciassètte, diciòtto, diciannòve, venti


undici, dodici, tredici, quattordici, quindici, sedici, settèndici, ottòdici, novèndici, venti.

In Spanish it is:

uno, dos, tres, cuatro, cinco, seis, siete, ocho, nueve, diez,

once, doce, trece, catorce, quince, dieciseis, diecisiete, dieciocho, diecinueve, veinte

In Portuguese:

um, dois, três, quarto, cinco, seis, sete, oito, nove, dez,

onze, doze, treze, catorze/quatorze, quinze, dezasseis/dezesseis, dezessete/dezessete, dezoito, dezenove/dezenove, vinte

In French:

un, deux, trois, quatre, cinq, six, sept, huit, neuf, dix,

onze, douze, treize, quatorze, quinze, seize, dix-sept, dix-huit, dix-neuf, vingt

Following martina's hint here are the number words from one to twenty in Romanian:

unu, doi, trei, patru, cinci, şase, şapte, opt, nouă, zece,

unsprezece, doisprezece, treisprezece, paisprezece, cincisprezece, şaisprezece, şaptesprezece, optsprezece, nouăsprezece, douăzeci

I find Latin, Italian, Spanish, Portuguese and French constructions for number words bizarre and inconsistent (whereas in Romanian it seems they are perfectly consistent as well as in Ancient Greek). Is there any academic work on the history of number words in Latin/Italian/Spanish/Portuguese/French where the origin of number words from eleven to nineteen is tracked down, documented, explained, discussed?

Please note the answer by symbiotech in the original post - I am not able to check it, perhaps someone in here can?

  • 4
    In Romanian the two seemingly irregular numbers, 14 (paisprezece vs. patru, "4") and 16 (şaisprezece vs. şase, "6") have perfectly regular variants patrusprezece (14) and şasesprezece (16).
    – Yellow Sky
    Jan 12, 2014 at 14:17
  • 1
    I don't know any work about this, but I think inconsistencies in that number range make sense for every superstrate language. The numbers up to 12 were used so much in daily commerce that everybody knew them correctly. Very large numbers were rarely used, and the people using them mostly knew the language well. But in the medium range, innovations from L2 speakers choosing workarounds (forming numbers according to their L1 patterns, or using ad hoc methods) when they didn't know the L2 patterns, seem most likely to have occurred frequently enough to spread.
    – user4938
    Jul 28, 2014 at 12:41
  • This question has been asked for Italian, Spanish, French, Portuguese. No answer so far explains the inconsistencies. Jul 17, 2015 at 12:10
  • 3
    The inconsistencies are because they're different languages with different speakers. What you have is several colliding and contrasting forms, and no surprise that they work it out differently in different languages and cultures, at different times and places. The 10-20 decade is often irregular, in many languages, so that's not surprising, either. There's the Latin (and Latin-descended) forms like quatorze, quattordici, there's the 'ten-and-seven' pattern, with a conjunction, like dieciseis, the reverse Greek 'seven-and-ten' pattern, and the '20-minus 2' pattern, like duodēvīgintī.
    – jlawler
    Jul 17, 2015 at 17:35
  • 1
    Germanic languages change after 12. Traditional Welsh is even more complicated than Romance: un ar ddeg (one on ten), dau ar ddeg, tri ar ddeg, pedwar ar ddeg, pymtheg (contraction of pymp deg, 5 10), un ar bymtheg (one on fifteen), dau ar bymtheg, deunaw (two nine!), pedwar ar bymtheg, ugain.
    – Colin Fine
    Jul 21, 2015 at 16:41

1 Answer 1


There are forces driving language evolution, and we see two of them at work here.

The first driving force is Regularisation. The irregular pattern of latin (indicated by duodeviginti and undeviginti, showing a counting down from 20 instead of counting up from 10) wasn't able to resist this driving force and all quoted modern Romance languages use "counting up from 10", i.e. a more regular pattern.

The second driving force is Sound Change. At some time and for some descendants of latin, the successors of sedecim and septemdecim became too similar in sound (something like *sedeci and *seddeci), therefore more distinctive forms for the two different numbers were needed.

Different dialects/languages reacted differently to the pressure of theses two forces (evolution is unpredictable, after all). French and Italian made 16 and 17 maximally different (seize vs. dix-sept). Spanish and Portuguese chose a more regular, but still distinct enough pattern (dieciseis and diecisiete). Romanian is the odd Romance language out, where Sound Change didn't put up much pressure and the old latin forms of 16 and 17 are pretty well preserved (şaisprezece and şaptesprezece). Romanian was than able to develop a completely regular pattern up to 19 (nouăsprezece).

ADDITION: In Romanian there is still something left to explain: Romanian has acquired an infix -spre- "to" not present in Latin or the other Romance languages. This exhibits a third driving force to language evolution, the influence of other languages. I daren't decide whether it is the consequence of a Dacian substrat or a Slavonic or Albanian adstrat. At least, Albanian numerals (gjashtëmbëdhjetë "16" with -bë- inserted) and Bulgarian numerals (шестна́десет (šestnádeset) "16" with -ná- inserted) show the same pattern of formation.

  • 1
    You answer is right since it shows the linguistic laws governing the changes. However, it does not answer the original question which asks for an academic work where the phenomenon of irregular numerals have been covered for all listed languages. Also, the bounty says, A detailed canonical answer is required. Would you please consider expanding your answer to make it a more fundamental research? You seem to have a good start so far. Jul 22, 2015 at 11:43
  • 2
    @bytebuster You get what you pay for. This is free. For research in printed books and journals for references, visit a university library and do it yourself (it may take some weeks). I'm sure you will find some references given enough time. Jul 27, 2015 at 13:18
  • 1
    The only currency within StackExchange network is reputation points. Someone who wants to earn money may need to find another network. Yes, a good answer to this question, indeed, worth some weeks of research. Another answer, I mean. Jul 27, 2015 at 15:02
  • 2
    jknappen, I hope I'm not intruding in this thread, I really like your answer regarding numerals. But there's something in Romanian that I need to point out. The infix is "spre" (same pronounciation as "pre", just add "s" in front), which means "to", so "unsprezece" is "one-to-ten", "doisprezece" (or "douăsprezece", the feminine variant) is "two-to-ten", "treisprezece" is "three-to-ten", "patrusprezece" ("paisprezece" tends to be found in informal usage, just like "șasesprezece" versus "șaisprezece") is "four-to-ten" and so on. I'm a native Romanian speaker.
    – user10186
    Jul 27, 2015 at 18:22
  • 5
    AFAIK Romanian unsprezece, doisprezece are a clear influence of the Slavic adstrat. Historically, Slavic 11-19 are constructed as "one-on-ten", "two-on-ten" up to "nine-on-ten" (clearly visible e.g. in Russian: odinnadcať < odin-na-desať or your Bulgarian example šest-na-deset = "six-on-ten"). Romanian seems to have adopted this: unsprezece < unum-super-decem. While Romanian spre changed meaning to "to", it stems from Latin super ("on" or "over").
    – Eleshar
    Apr 8, 2017 at 12:43

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.