Number Complexity

[Richard W]

Earlier this year there was some disagreement as to the complexity of numbers. I've just had a lovely example of the ambiguity of Thai numbers. Asked how much I had just sent by way of repayment, I replied /mɯːn sɔːŋ ba:t/ lit. 10,000 2 BAHT, as I thought /mɯːn sɔːŋ pʰan ba:t/ ˈ10,000 2 1,000 BAHT' would be too pernickety. Thai numbers, for the most part, are uttered non-zero digit by non-zero digit in the careful form <digit><name for power of 10>, except that the word for 'one' is omitted. I thought my answer was only ambiguous between 12,000 and 10,002. It turns out that it was also ambiguous with 10,200!

[Kuchigakatai]

That's a pretty amusing story.

Another example of ambiguity: in much of Latin American Spanish, doce '12' is phonetically very close to dos '2' (among most speakers in Spain, the difference is at least also reinforced by the /θ/-/s/ contrast). So you either catch that unstressed [e], or you're left in ambiguity. It does happen quite often that someone asks the time, and the exchange goes:

A: ¿Qué hora es? 'What time is it?'
B: Las doce. '12.'
A: ¿Las doce o las dos? 'Was that 12 or 2?' (overpronouncing the word doce as [ˈdo ˈse])

And it gets worse when you add the minutes, because they're added with the conjunction y 'and', and phonetically it is even more prone to be affected by lazy articulation or environmental noise. Las doce y veinte '12:20' ~ Las dos y veinte '2:20'.

In some dialects, the problem may be solved by pronouncing dos as [doh] at least in some phonological contexts, which then remains different from doce [ˈdose]~[ˈdos^e]. But in many others it is a problem, as in my dialect where /s/ is so very often [s] before a pause (e.g. los raspones [loh rahˈpones], las aspas [lah ˈahpas].

[Travis B.]

I often find myself having a hard time telling apart fifty/fifteen and sixty/sixteen...

[Richard W]

I often find myself having a hard time telling apart fifty/fifteen and sixty/sixteen...

Whence the spoken numbers three oh etc., which I suspect really complicate the grammar of English numbers.

[Linguoboy]

In German, zwei "two" and drei "three" are easy enough to confuse that the standard language has borrowed the variant zwo (originally a feminine form) from southern varieties for use in lossy environments, such as over a telephone connexion.

[vegfarandi]

Icelandic numerals are a total nightmare for second-language learners. The cardinal numerals up to four decline for case and gender and are almost completely irregular.

(nom, acc, dat, gen × masc. / fem. / neut.)
einn, einn, einum, eins / ein, eina, einni, einnar / eitt, eitt, einu, eins
tveir, tvo, tveimur, tveggja / tvær, tvær, tveimur, tveggja / tvö, tvö, tveimur, tveggja
þrír, þrjá, þremur, þriggja / þrjár, þrjár, þremur, þriggja / þrjú, þrjú, þremur, þriggja*
fjórir, fjóra, fjórum, fjögurra / fjórar, fjórar, fjórum, fjögurra / fjögur, fjögur, fjórum, fjögurra

In addition, there's a special form of these numbers for use with pluralia tantum which is plural in form, but these numerals actually also have singular forms that are used in certain esoteric situations. These are a bit more regular; for the most part declining like most strong adjectives.

einir, eina, einum, einna / einar, einar, einum, einna / ein, ein, einum, einna
tvennur, tvennan, tvennum, tvenns / tvenn, tvenna, tvenni, tvennar / tvennt, tvennt, tvennu, tvenns // tvennir, tvenna, tvennum, tvennra / tvennar, tvennar, tvennum, tvennra / tvenn, tvenn, tvennum, tvennra
þrennur, þrennan etc. like tvennur
fern, fernan, fernum etc. like tvennur other than the masc. sg.

So in order to ask for less than five of anything, you have to know 132 forms (though it helps that the dative and genitive plural are always the same across three genders, so if we subtract the fem./neut. dat./gen. pl. you're down to just 104). My dad's coworker who was an immigrant from US spoke really good Icelandic, had an almost undetectable accent, but she could never get these right so whenever she went to a bakery she'd ask for at least fimm 'five' of whatever she needed. No need for declension.

*For extra spice, the forms dat. forms tveimur and þremur have alternates tve(i)m and þrem which in certain usages (though not all!) have certain classist connotations and sound undereducated. The trick is to use them just right.

[sasasha]

My dad's coworker who was an immigrant from US spoke really good Icelandic, had an almost undetectable accent, but she could never get these right so whenever she went to a bakery she'd ask for at least fimm 'five' of whatever she needed. No need for declension.

That's really funny!

When asked "how many x are there?" in Icelandic, and there are fewer than five, and you answer just with a number, must the answer agree with x?

[vegfarandi]

My dad's coworker who was an immigrant from US spoke really good Icelandic, had an almost undetectable accent, but she could never get these right so whenever she went to a bakery she'd ask for at least fimm 'five' of whatever she needed. No need for declension.

That's really funny!

When asked "how many x are there?" in Icelandic, and there are fewer than five, and you answer just with a number, must the answer agree with x?

Yes:

Hversu marga banana? “Bananas”
Þrjá.

Hversu margar kerrur? “Wagons”
Þrjár.

Hversu mörg oststykki? “Pieces of cheese”
Þrjú.

And with pluralia tanta:

Hversu marga sokka? “Socks”
Þrenna.

Hversu margar buxur? “Pants/trousers”
Þrennar.

Hversu mörg stígvéli? “Boots”
Þrenn.

[Nachtswalbe]

Wouldn't it make sense to use some sort of decimal place marker to reduce this ambiguity in spoken language?

[Richard W]

Wouldn't it make sense to use some sort of decimal place marker to reduce this ambiguity in spoken language?

See the opening post for why I didn't use the unambiguous four word form.

[Richard W]

It's just occurred to me that the unexpected ambiguity in Thai /mɯːn sɔːŋ ba:t/, literally 10,000 2 BAHT, arose because I am an English speaker, and I have been warned that some people understand my Thai by translating it into English and then translating that English into Thai. The interpretation may then have gone:

mɯːn > ten thousand = 10 × 1,000
sɔːŋ > two, or 0.2 of the previous unit.
ba:t > baht

Taking 'thousand' as the previous unit, as in English, would yield ฿10,200. Thai goes up in powers of 10 to a million, then multipliers of a million seem to go up to 999,999, unlike US English which goes up in powers of 10 to a thousand, then up in powers of a thousand.

[hwhatting]

In Russian, 50 (пятьдесят pyat'desyát) and 60 (шестьдесят shest'desyát) both have the stress on the final syllable and can be confused in what linguoboy calls lossy environments, so people often substitute полсотни polsótni "half-hundred" for 50.