Innocence Seekers: The Black Rose – Numerals in Namari

This post is dedicated to the numeral system of the Namari language. The original concept can be found on the wiki; however, I’m going to make some tweaks and minor overhauls with regards to how numbers work in the language.

But first, a bit of background. Namari is a Japonic language (more precisely, it is more closely related to Japanese than the Ryukyuan languages), meaning it largely shares the same numerals as Japanese. Old Namari was, in many respects, similar to Old Japanese; however, the two languages diverged due to differing influence of Chinese on the two languages (this was because the two languages were spoken on two different planets, linked only by a single Portal Monolith). As such, Modern Namari retains the old numeral system found in ancient versions of Japanese.

Old Namari, like Old Japanese, had a nascent classifier system. More common was the prefixing of numerals directly on to nouns (e.g. yakumo “many clouds”, literally “eight clouds”). In Japanese, thanks to the influence of Chinese, classifiers are far more prevalent, to the point that almost all nouns are “mass nouns”. On the other hand, Namari has a reduced classifier system, and as a result of the combined influence of Guruni (which uses prefixed numerals) and Blylandic (which uses adjectival and nominal numerals), the vast majority of nouns either use the generic classifier or accept direct prefixing; only in a few cases (e.g. waga imouton putain or waga putain imouton “to my two little sisters”) are classifiers used (also note the case agreement; this is also part of my overhaul).

The following are the numerals, in Namari, Japanese (native) and Proto-Japonic (Disclaimer: I am not a historical linguist; take my reconstructions with a grain of salt):

• 1: pito, hito, *pitə
• 2: puta, futa, *puta
• 3: mi, mi, *mi(C)
• 4: yo, yo, *yə(C)
• 5: etu, itsu *itu
• 6: mu, mu, *mu(C)
• 7: nana, nana, *nana(N)/*nanə(N)
• 8: ya, ya, *ya(C)
• 9: kokono, kokono, *kəkənə
• 10: tō, tō, *tə-wə
• 20: pata, hata, *pata
• ×10: -so, -so, *-so?
• 100: momo, momo, *momo
• ×100: -o, -(h)o, *-po?
• 1,000: chi, chi, *ti
• 10,000: yoyozu, yorozu, *yərən-tu

In the reconstructions:

• (C): a possible final consonant (a coronal); may be attested in the ancient languages of Korea
• (N): a possible syllable-final nasal; a form nanən for “seven” is attested in one ancient language in Korea
• I: either */e/ or */i/; I have yet to figure out which
• The two forms for “seven” reflect a possible consequence of Arisaka’s Law, which dictates that /ə/ cannot co-occur with a back vowel in a morpheme. The latter form may be the older form, with the former being unique to Insular Japonic (i.e. Japanese and Ryukyuan, not the supposed Japonic languages that were spoken in Korea).
• I assume that Proto-Japonic */a/ is a back vowel *[ɑ], and that */ə/ is a rounded central vowel *[ɞ] or *[ɵ]. Arisaka’s Law will shift */ə/ back, most likely to *[ɒ] (which merges with /a/), but possibly *[o] (or even *[u]).

The following are the Namari numerals given as root forms and combined with the generic classifier, classifier for people, and classifier for days:

• 1: pito, pitotu, pitoi, pitoka/pitoe/pitofi/pitopi
• 2: puta, putatu, putai, putuka
• 3: mi, mittu, mittai, mikka
• 4: yo, yottu, yottai, yokka
• 5: etu, etutu, etutai, etuka
• 6: mu, muttu, muttai, muika
• 7: nana, nanatu, nanatai, nanoka
• 8: ya, yattu, yattai, yauka
• 9: kokono, kokonotu, kokonotai, kokonoka
• 10: to, tō, totai, tōka
• 11: tōte, tōtetu, tōtetai, tōtoka
• 12: tōta, tōtatu, tōtatai, tōtuka
• 13: tō-mi tō-mittu, totai-mittai, tōka-mikka
• 20: pata, patachi, patatai, patuka
• 21: pata-pito, patachi-pitotu, patatai-pitoi, patuka-pitoka
• 30: miso, misoji, misotai, misoka
• 50: eso, esoji, esotai, eka
• 100: momo, momo, momotai, momoka
• 132: momo miso-puta, momo misoji-putatu, momotai misotai-putai, momoka misoka-putuka
• 200: putao, putao, putaotai, putaoka
• 500: eo, eo, eotai, eoka
• 1,000: chi, chi, chitai, chuka
• 2,000: putachi, putachi, putachitai, putachuka
• 5,000: echi, echi, echitai, echuka
• 10,000: yoyozu, yoyozu, yoyozutai, yoyozuka
• 50,000: eyoyozu, eyoyozu, eyoyozutai, eyoyozuka

The root forms of numerals less than 100 cannot occur alone outside of counting; the numeral must be combined with a noun or classifier. To denote such numerals alone, the generic classifier must be used if the objects referred to are implied to be inanimate or abstract. Non-human animate nouns are quantified using -moni (a third declension suffix, hence a short vocative -mon and a genitive -monga), while humans are quantified using the classifier for people. The three classifiers mentioned, as well as the classifiers for days (used to count days, nights and daily recurring events) and layers (-e), are the only classifiers in the modern language; what seem to be additional classifiers in the language are simply count forms of their corresponding nouns, or measure words used to quantify mass nouns.

On the other hand, the numerals 100 and above can stand alone, and do not take generic classifiers (the only classifiers allowed for large numerals are the classifiers for people, days and layers).

Complex numbers are formed by simple concatenation of entire numeral-noun or numeral-classifier compounds. Note that, particularly for complex numerals, the numerals for 100 and higher may stand alone without a classifier or count form noun, although they will still show agreement. If the number has a zero in the tens place, mä- (a heavily-reduced form of the Old Namari amari) may optionally be used. Examples are shown below:

• 2,483 years: putachi(tose) yoo(tose) yasotose-mitose
• 3,915,307 persons: mio kokonoso-pitoyoyozu(tai) echi(tai) mio(tai) (mä-)nanatai

Note that for complex numbers larger than a million, any freestanding numerals modifying yoyozu are assumed to modify yoyozu directly, rather than the head noun. Additionally, multiples of 100,000, one million and ten million do not take any classifier or count form noun.

Ordinal numbers are formed regularly by adding the suffix -me to a numeral-classifier or numeral-noun compound (e.g. mittume “third”, nanataime “seventh person”, totoseme “tenth year”). If used attributively (as part of a numeral-classifier compound), the ordinal agrees with the head noun in case. Complex ordinals only add -me to the final numeral.
Namari has special words for “first” (pajime(te), patu), “second” (otugi), “former” (ekoto), “latter” (otoi) and “last” (sue), that can be used in place of a regular ordinal.

Distributives are denoted by adding the suffix -zutu to a numeral-classifier or numeral-noun compound, in the same manner as -me.

All three categories of numerals mentioned above work identically to each other. Regardless of the number of numerals, all components agree in case. This includes the head noun, if one exists. This also serves to distinguish the classifiers from classifier-like measure words, as numeral-measure word compounds always use the genitive when modifying nouns.

Old Namari could only denote numbers up to 99,999,999 (which would be rendered as kəkənəchi kəkənəpo kəkənəso-kəkənəyərədu kəkənəchi kəkənəpo kəkənəso-kəkənə). Early in the Middle Namari period, a word sume (borrowed from Old Guruni cumen “hundred thousand”) was used to denote 100,000, and at some point was repurposed to represent 100,000,000. However, by the Modern Namari period, it had fallen into disuse, having been replaced with the Sino-Namari oki (a third declension noun/suffix). However, even this had become archaic by the 17th century (by Earth reckoning), and had been replaced by the modern system for large numbers.

The modern system in Namari prefixes yoyozu with a numeral-classifier compound using the classifier for layers (which is also used to denote exponents). Hence the first ten powers of 10,000 are as follows:

• 10,000: yoyozu
• 100,000,000: putaeyoyozu
• 10¹²: mieyoyozu
• 10¹⁶: yoeyoyozu
• 10²⁰: etueyoyozu
• 10²⁴: mueyoyozu
• 10²⁸: nanaeyoyozu
• 10³²: yaeyoyozu
• 10³⁶: kokonoeyoyozu
• 10⁴⁰: toeyoyozu

The numerals for 100,000,000 and higher do not take numeral prefixes. Instead, their multipliers take the generic classifier for inanimates, regardless of the animacy of the head noun.

In general, numerals larger than mieyoyozu are not found in common speech. In scientific language, the structure (make-)Y-eno (make-)X is used to denote the number X × 10^Y, where the base of the numeral-classifier compound for layers is implied to be ten (the prefix make- simply indicates that the number is negative).

Zero is denoted by the verbal adjective nakai and its inflected forms. However, when reading digits one-by-one, the Sino-Namari rei is used instead to represent zero.

Unlike with integers, numbers with decimal components do not take classifiers. Such numbers are separate noun phrases, and are treated as nominal adjectives (meaning they modify the head noun by using the genitive) or nouns.
Numbers with fractional components are split into their integer (omitted if it is zero) and fractional components. The integer component occurs before the fractional component, and takes the appropriate classifier or count form noun if applicable. The integer component is then declined in the comitative case. The fractional component is formed by the structure (denominator)-waino (numerator). The entire complex is treated as a nominal adjective or a noun. Note that if the fractional component is a half, it is instead denoted as nakaba (e.g. pitotuto nakaba for “one and a half”, as opposed to pitotuto miwaino pito for “one and a third”).
Numbers with decimal components are, like those with fractional components, split into their integer and fractional components. This time, the integer component is suffixed with the count form noun -mayu, unless the integer component is zero, in which case reimayu is used instead. The fractional component is read out digit-by-digit (e.g. 32.7903 is miso-putamayu nana kokono rei mī; note that mi is lengthened to bimoraic mī by phonotactic constraints).

Anyway, that will be it for now.

P.S. Count forms of nouns are usually identical to their standalone counterparts; however, there are exceptions. Examples include:

• toshi (“year”, second declension e-stem): -tose (first declension)
• tume (“claw”, second declension a-stem): -tuma (first declension)
• yoyu (“night”, first declension): -yo (first declension)

And some additional notes on my reconstructions:

• Where I have capital vowel letters (*E, *I, *U), it means I am not sure which phoneme it represents, due to mergers in the attested Japonic languages. A reconstructed *E at the end of a word is either *e or *ay, *I at the beginning or middle is either *e or *i, *U at the beginning or middle is either *o or *u, and *I at the end is either *i, *Uy or *ey.
• I assume that Proto-Japonic had a six-vowel system. The traditional four-vowel reconstruction has long been discredited as it relied too much on Western Japanese, to the exclusion of Eastern Old Japanese and the Ryukyuan languages. While some authors postulate a seventh vowel, the evidence is, in my opinion, too flimsy (the variations the seventh vowel is supposed to explain, namely that *əy becomes both /e/ and /i/ in Japanese, can be explained by dialect mixing; Proto-Japonic *əy consistently became Proto-Ryukyuan *e).
• As a corrolary to the above point, my ability to determine new Namari words is hampered by the lack of a comprehensive, up-to-date and easily accessible Proto-Japonic wordlist. The only comprehensive wordlist I could find on the Internet is not only out of date (it assumes the four-vowel system), but the website on which it is hosted appears to subscribe to fringe historical linguistic theories. As of writing Wiktionary only has 68 Proto-Japonic words. As such, I have to guess whether a word should have /e, o/ or /i, u/, or do my own reconstructions based on Ryukyuan evidence, which may not always be easily accessible.
• For example, the Namari word sue (“last”) was derived not only from Classical Japanese suwe, but also Okinawan shī, which confirms that the /u/ in the Japanese cognate indeed derives from Proto-Japonic */u/, and not */o/ (which would’ve yielded Okinawan *sui from a pre-form **sowe, hence my Proto-Japonic reconstruction is *suwe).
• When reconstructing Proto-Japonic, I opt for the simplest reconstruction. This means a strict adherence to CV(N) syllable structure, with diphthongs only occurring at the end of a word. I don’t take into account Proto-Japonic vowel length, again due to lack of easily-accessible resources, this time on the historical pitch accent of Japanese words.
• My justification for reconstructing */a/ as a back vowel and */ə/ as rounded is due to a combination of the modern reflexes of */ə/ and Arisaka’s Law. I’ve interpreted Arisaka’s Law as a remnant of an earlier Insular Japonic vowel harmony law that distinguished between back and central vowels. Because */a/ is in the same category as the back vowels */o/ and */u/, it makes sense for */a/ to be a back vowel as well. With regards to */ə/, its reflex in Modern Japanese is /o/, and in the Ryukyuan languages it is consistently /u/. By Occam’s Razor, it is best to assume that like its modern reflexes Proto-Japonic */ə/ was rounded. My reconstructions of */ə/ are compatible with both the six-vowel (*[ɞ]) and seven-vowel (*[ɞ] and *[ɵ]) systems.
• Following on from above, Proto-Japonic may have had an additional vowel *[ʉ] or *[y] that may have resulted from the vowel harmony law (alternating with */u/), and that merged with */i/. However, this is simply speculation on my part, and not reflected in my reconstructions. If this were true (I highly doubt it), then Proto-Japonic at one point would’ve had a ten-vowel system with neutral vowels *[e i], “light” vowels *[ɜ ɞ ɵ ʉ], and “heavy” vowels *[ɑ ɒ o u], where *[ɜ], *[ɞ] and *[ɵ] merged to */ə/, *[ʉ] and *[i] merged to */i/, and *[ɑ] and *[ɒ] merged to */a/.

And finally, some notes on my tweaks for Namari:

• The one thing I noticed with all Japonic languages is the sibilantisation of /t/ and /d/ before /u/, and fronting of /u/ after coronal obstruents. I believe I’ve only observed the coronal stops remaining stops before /u/ in the Tosa dialect of Shikoku and neighbouring dialects (and even then, they may be realised as sibilant affricates). Sibilantisation of /t/ and /d/ before /u/ is also widespread in Namari-speaking areas, and once again there is dialectal variation (Chiyohara /tu/ is [t͡sʉ], Yaezora /tu/ is [θy], deriving from earlier [t͡sʉ]).
• As a corrolary to the above, the yotsugana have two pronunciations in the Yaezora dialect (/di/, /zi/ = [ʑi ~ d͡ʑi]; /du/, /zu/ = [d͡zy ~ zy ~ ðy]) and three in the Chiyohara dialect (/di/, /zi/ = [ʑi ~ d͡ʑi]; /du/ = [d͡zʉ ~ ðʉ]; /zu/ = [zʉ])
• The distinction between /u/ and /y/ in the Yaezora dialect is neutralised after coronal obstruents. Hence sui and sū are homophones ([syː]). Likewise, the distinction between /t/ and /θ/ is neutralised before /u/, with both being pronounced [θy]. Chiyohara dialect still distinguishes both, but does not have [t͡s] in free variation with /s/.

Edit (2019-03-06): One thing I forgot: the Namari word for “how many” is now ekutu, derived by combining eku with the generic classifier for inanimates.

Edit (2019-04-29): Slight adjustments