Zompist Bboard Again

This topic came up on the cldr-users list for the Unicode Common Locale Data Repository.

Units of measure are usually nouns, and the forms of the preceding number can depend on the gender of the noun. Are there any examples of SI prefixes affecting the gender of the unit?

As a motivation, consider the Roman pace as a measure of distance. The basic unit, passus, is masculine. However, in the plural, the 'kilopace' is milia passuum, a neuter (plural) noun.

A conceivable structure would be for prefixes to be treated as number nouns with the unit of measure qualifying the number noun. Then, if gender were assigned to the prefixes, the number might agree (or, in the case of Semitic, disagree!) with the prefix in gender.

I think SI names act as part of derivational morphology, and thus have no gender. But there's a lot of languages, maybe some act differently!

One thing to check might be kilo, used in several languages as an abbreviation (thus not really SI) for "kilogram". In French, kilo, gramme, and kilogramme all happen to be masculine. Ooh, here's one: in Russian кило is neuter and килогра́мм is masculine.

Huh, interesting, the Russian word is one of those indeclinable o-stems. For some sort of international studies requirement, I took a Russian class ages ago, and remember thinking that was really odd, since there was an entire class of o-nouns that seemed to decline quite normally.

I thought Bantu might work with (kilo)metre, with kilometre being sucked into the ki/vi - concord class, but though I can find vilo being the plural of kilo, I can't find the data to nail it down. Swahili sticks them both in the foreign word class (n), so numbers get the same form for both.

Richard W wrote: ↑Mon Mar 01, 2021 4:15 pmAs a motivation, consider the Roman pace as a measure of distance. The basic unit, passus, is masculine. However, in the plural, the 'kilopace' is milia passuum, a neuter (plural) noun.

It'd be good to clarify that in mīlia passuum ("thousand.PL.NOM[NEUT] steps.PL.GEN[MASC]") lit. 'thousands of steps', meaning '(the/some) Roman paces' (in the plural), the whole noun phrase is neuter-gender because the noun isn't the head, the number is the head. Unlike in trēs passūs ("three steps.PL.NOM") 'three steps', where 'steps' is the head. Mīlle 'thousand' has the odd property of being an indeclinable adjective in the "singular" (a thousand Xs) but a neuter plural noun in the "plural" (two thousand Xs, five thousand Xs, thousands of Xs).

Similarly, in Standard Arabic, the numbers 1 and 2 are adjectives, but most other numbers are the head of their noun phrase, as if saying literally "three of men", "fifteen to man", for three men and fifteen men. I've heard Russian has something similar.

Relevantly derived nouns sometimes get a different gender for some reason... An example would be Latin hordeum (neuter) 'barley', derived diminutive hordeolus (masculine) 'stye, a type of barley-grain-shaped eye infection'. But here the significant change in meaning is surely involved in some way, much like those Romance gender pairs mentioned in the Language Construction Kits, like punto 'dot, point' and punta 'tip (of spear, pen, etc.)', puerto 'dock' and puerta 'door', cerco 'tall or thick fence' and cerca 'fence' (more examples).

Well, some of the SI prefixes are augmentative or diminutive prefixes for rather large changes in size, so that could change the gender - compare the German diminutive suffix -chen, which yields neuter nouns. Another mechanism is that the shape of words can determine what gender they adopt when borrowed, which is why I've been looking at Bantu languages.

In German at least, prefixes never change the gender of the noun, though suffixes often do. I guess it is the same with most other European languages (I am pretty sure that it is so in Latin, for instance). So how should an SI prefix change the gender in these languages?

WeepingElf wrote: ↑Tue Mar 02, 2021 2:50 pm In German at least, prefixes never change the gender of the noun, though suffixes often do. I guess it is the same with most other European languages (I am pretty sure that it is so in Latin, for instance). So how should an SI prefix change the gender in these languages?

Latin for 'three paces' is TRESPASSVS but Latin for 'three thousand paces' is TRIAMILIAPASUUM.

If the SI prefix were interpreted as a noun, and the base unit as a dependent genitive, then the prefix would determine the gender of the whole.

I didn't expect European examples. However, if gender-related information goes at the start of the word, prefixes could easily switch the gender of the word.

Richard W wrote: ↑Tue Mar 02, 2021 4:48 pm Latin for 'three paces' is TRESPASSVS but Latin for 'three thousand paces' is TRIAMILIAPASUUM.

If the SI prefix were interpreted as a noun, and the base unit as a dependent genitive, then the prefix would determine the gender of the whole.

I didn't expect European examples. However, if gender-related information goes at the start of the word, prefixes could easily switch the gender of the word.

A similar construction exists in French: une seconde ~ un million de secondes. SI prefixes don't change gender though: une kiloseconde, un kilomètre.

Indeed, I don't think there are any IE examples, for the reasons WeepingElf exposed.

As you said, Bantu looks like it could do just that! So I checked 'kilometer' in Swahili, Lingala, Kikongo and Zulu.
Disappointingly, all of these treat 'ki' as part of the root, and none as a class prefix. In Swahili and Lingala, it's invariable. Kikongo places it in its -/ba- class, so kilometers is bakilometre. Zulu adds a noun class prefix: sg. ikhilomitha, pl. amakhilomitha

If anyone wants to keep on researching this, I'd like to share this wonderful tool, though: https://fr.glosbe.com/fr/kg/kilom%C3%A8tre
The best part is the 'example' section; scroll down for that.

I thought I'd got Swahili cornered with wati 'watt'. According to the grammars, the only possible plural should be *nyati (u-/n- classes), for in the n-/n- class the stem wati would yield the form *mbati. However, from the examples I can see, the speakers haven't read the textbooks, and both singular and plural are _wati_. Wiktionary had got my hopes up by saying that Swahili kilo could have the plural vilo, but I can't find any evidence of it. I've even found one example of vilogramu, but a single example might be a typo.

Rounin Ryuuji wrote: ↑Mon Mar 01, 2021 6:35 pm Huh, interesting, the Russian word is one of those indeclinable o-stems. For some sort of international studies requirement, I took a Russian class ages ago, and remember thinking that was really odd, since there was an entire class of o-nouns that seemed to decline quite normally.

The indeclinable ones I know are all loans (kilo, kino "cinema", pal'to "coat", metro "subway"); all have stress on the final syllable. But at least pal'tó gets a declined plural in colloquial Russian (pól'ta), with the shift to stem-based stress in the plural typical for this noun class. But this pluralisation is not accepted in the formal standard.