Language Patterns
Recurring patterns across languages, space, and time.
Pattern

Counting Things

What goes with a number to count a thing?

English puts a number straight in front of most nouns — three books, three dogs — and only reaches for an extra word when the noun resists counting: three sheets of paper, two head of cattle. Across East and Southeast Asia that extra word is required for ordinary countable things, and which one you pick depends on what the thing is. Japanese counts books with 冊, sheets with 枚 and pencils with 本. Bantu languages do a comparable job from the other end, making the numeral agree with the class the noun already belongs to. Most of these counting words are nouns caught partway through a change: Indonesian ekor is a tail, Vietnamese cây a tree, Thai khon a person.

Strategies

A classifier picked by the thing

NUM + CLF + N / N + NUM + CLF

A counting word joins the numeral, and which word it is depends on the thing counted — its shape, whether it is alive, what it is for. Japanese, Korean, Mandarin, Thai, Vietnamese, Burmese and Yucatec Maya all work this way. Mandarin, Vietnamese and Yucatec Maya put the numeral and classifier before the noun; Thai, Burmese and Korean put them after it; Japanese does both, 三冊の本 before and 本を三冊 after. Where the noun already names a unit — a year, a baht, a cupful — no sortal classifier joins it, since the noun is doing that work itself.

Japanese · 本を三冊 hon o san-satsuKorean · 책 세 권 chaek se gwonMandarin Chinese · 三本書 sān běn shūThai · หนังสือสามเล่ม nǎngsɯ̌ɯ sǎam lêmVietnamese · ba quyển sách Burmese · လူ သုံး ယောက် lù thóuñ yauʔYucatec Maya · óox-túul / óox-p’éel Bengali · তিনটি বই tinṭi boi

The noun as its own classifier

N + NUM + N

When no dedicated classifier covers a noun, the noun itself fills the slot, so it appears twice in one phrase. Burmese does this readily. Something has to go in the slot, and where the inventory has no word for this kind of thing, the noun serves.

Burmese · မြို့ သုံး မြို့ myo̰ thóuñ myo̰

Classifiers you may drop

NUM (+ CLF) + N

The counting words are picked by the kind of thing, exactly as in the languages above, but the grammar does not require them. Indonesian has orang for people, ekor for animals and buah for things, and everyday speech often leaves them out where a formal register keeps them. Obligatoriness runs as a scale rather than a line, and this is a point along it.

Indonesian · tiga (buah) buku

A unit word only where needed

NUM + UNIT of N

A word that cuts a countable unit out of something that comes in no natural pieces. English needs sheets of paper, loaves of bread, head of cattle; German has drei Blatt Papier. The choice still follows the substance — a loaf of paper is nothing — but the word supplies the unit rather than sorting a thing that already had one. Languages with sortal classifiers use measure words too: 三杯の水, 三杯水, tiga gelas air.

English · three sheets of paper German · drei Blatt Papier

Agreement instead

CLASS-N + CLASS-NUM

The noun already belongs to a class marked on the noun itself, and the numeral picks up that class as agreement. Swahili vitabu vitatu “three books” and watu watatu “three people” carry the prefix on both words. Nothing is inserted, because the sorting is already in the noun. A noun class differs from a classifier in reach: it runs as concord through the whole clause, where a classifier is a word chosen for one counting phrase.

Swahili · vitabu vitatu Zulu · izincwadi ezintathu

Nothing in between

NUM + N

The numeral sits directly against the noun, with no classifier and no agreement to speak of. This is the ordinary case across Europe and much of the world.

English · three books German · drei Bücher Russian · три книги tri knigiFrench · trois livres

Geographic distribution

Pick one thing, and each dot is how that language says it, coloured by the strategy behind the verb. Click a dot for the form.

A classifier picked by the thing Classifiers you may drop Agreement instead Nothing in between

Marker positions are approximate cultural centres — they are not territorial claims. Tiles: OpenFreeMap · © OpenStreetMap contributors · Dark tiles: © CARTO.

Language comparison

LanguageStrategyExpressionNote
Japanese
Japonic
A classifier picked by the thing本を三冊 hon o san-satsu
the numeral and classifier form one word, and stand apart from the noun
Korean
Koreanic
A classifier picked by the thing책 세 권 chaek se gwon
Mandarin Chinese
Sino-Tibetan › Sinitic
A classifier picked by the thing三本書 sān běn shū
Thai
Tai-Kadai › Tai
A classifier picked by the thingหนังสือสามเล่ม nǎngsɯ̌ɯ sǎam lêm
the noun comes first, then the numeral, then the classifier
Vietnamese
Austroasiatic › Vietic
A classifier picked by the thingba quyển sách
Burmese
Sino-Tibetan › Burmish
A classifier picked by the thingလူ သုံး ယောက် lù thóuñ yauʔ
noun, then numeral, then classifier. Books take အုပ်, which is also the second half of the word for book, so this row uses people instead
Yucatec Maya
Mayan
A classifier picked by the thingóox-túul / óox-p’éel
the classifier attaches to the numeral and splits animate from inanimate
Bengali
Indo-European › Indo-Aryan
A classifier picked by the thingতিনটি বই tinṭi boi
-ṭi and -ṭa are enclitics on the numeral rather than separate words, and on a noun they mark it as definite: boi-ṭi “the book”. South Asian classifiers work this way generally
Burmese
Sino-Tibetan › Burmish
The noun as its own classifierမြို့ သုံး မြို့ myo̰ thóuñ myo̰
“city three city” — the noun repeats where no separate classifier serves
Indonesian
Austronesian › Malayic
Classifiers you may droptiga (buah) buku
orang for people, ekor for animals, buah for things; casual speech drops them
English
Indo-European › Germanic
A unit word only where neededthree sheets of paper
and two head of cattle, where the unit word survives in an old invariant form
[citation needed]
English
Indo-European › Germanic
Nothing in betweenthree books
German
Indo-European › Germanic
A unit word only where neededdrei Blatt Papier
Blatt stays singular after a numeral, as do Stück and Mann in the same use; drei Blätter Papier also occurs
[citation needed]
German
Indo-European › Germanic
Nothing in betweendrei Bücher
Russian
Indo-European › Slavic
Nothing in betweenтри книги tri knigi
the numeral governs the case of the noun, but adds no word
[citation needed]
French
Indo-European › Romance
Nothing in betweentrois livres
Swahili
Atlantic-Congo › Bantu
Agreement insteadvitabu vitatu
class 8 vi- on both the noun and the numeral; animals take class 2 agreement whatever prefix they carry themselves
Zulu
Atlantic-Congo › Bantu
Agreement insteadizincwadi ezintathu
class 10 agreement on the numeral

Which counting word for which thing

Read across a row for one language counting seven things; read down a column for one thing across languages. A classifier language uses several different words down its row, while English, German, French and Russian use the numeral alone and Swahili marks the numeral to agree with the noun. Cells in the classifier languages give the numeral and its classifier without the noun; elsewhere the whole phrase is short enough to show. A blank cell means no form is recorded here yet, not that the language lacks one.

Language
books
Bound volumes — often their own class.
sheets of paper
Flat, thin things. The one English also has to mark.
pencils
Long and thin, the commonest shape category.
dogs
An animal, where animacy usually decides.
people
Humans, almost always set apart from everything else.
cars
A machine, and a late arrival to every system here.
houses
Buildings, counted by their own word in most of these languages.
Japanese
Japonic
三冊 san-satsu
A classifier picked by the thing
bound volumes
三枚 san-mai
A classifier picked by the thing
flat thin things: paper, plates, shirts
三本 san-bon
A classifier picked by the thing
long thin things
三匹 san-biki
A classifier picked by the thing
small animals; large ones take 頭
三人 san-nin
A classifier picked by the thing
三台 san-dai
A classifier picked by the thing
machines and vehicles
三軒 san-gen
A classifier picked by the thing
houses and shops
Korean
Koreanic
세 권 se gwon
A classifier picked by the thing
세 장 se jang
A classifier picked by the thing
flat sheets
세 자루 se jaru
A classifier picked by the thing
things with a shaft or handle
세 마리 se mari
A classifier picked by the thing
animals of any size
세 명 se myeong
A classifier picked by the thing
세 대 se dae
A classifier picked by the thing
세 채 se chae
A classifier picked by the thing
Mandarin Chinese
Sino-Tibetan › Sinitic
三本 sān běn
A classifier picked by the thing
三張 sān zhāng
A classifier picked by the thing
flat spreadable things: paper, tables, beds
三支 sān zhī
A classifier picked by the thing
三隻 sān zhī
A classifier picked by the thing
homophonous with 支 but a different word
三個 sān gè
A classifier picked by the thing
個 is the general classifier; 位 is the polite one for people
三輛 sān liàng
A classifier picked by the thing
wheeled vehicles
三棟 sān dòng
A classifier picked by the thing
buildings; 間 counts rooms
Thai
Tai-Kadai › Tai
สามเล่ม sǎam lêm
A classifier picked by the thing
books, and also knives and candles. Thai puts the whole phrase after the noun: หนังสือสามเล่ม
สามแผ่น sǎam phàen
A classifier picked by the thing
สามแท่ง sǎam thâeng
A classifier picked by the thing
ด้าม is used for pens and other things held by a handle
สามตัว sǎam tua
A classifier picked by the thing
animals, and clothing with a body shape
สามคน sǎam khon
A classifier picked by the thing
the same word as the noun “person”
สามคัน sǎam khan
A classifier picked by the thing
สามหลัง sǎam lǎng
A classifier picked by the thing
Vietnamese
Austroasiatic › Vietic
ba quyển
A classifier picked by the thing
cuốn in the south
ba tờ
A classifier picked by the thing
ba cây
A classifier picked by the thing
cây is also the noun “tree”
ba con
A classifier picked by the thing
con covers animals and marks animacy broadly
ba người
A classifier picked by the thing
the same word as the noun “person”
ba chiếc
A classifier picked by the thing
ba ngôi
A classifier picked by the thing
căn is also used
Burmese
Sino-Tibetan › Burmish
သုံး အုပ် thóuñ ouʔ
A classifier picked by the thing
သုံး ကောင် thóuñ kaùñ
A classifier picked by the thing
animals
သုံး ယောက် thóuñ yauʔ
A classifier picked by the thing
Indonesian
Austronesian › Malayic
tiga (buah) buku
Classifiers you may drop
tiga lembar kertas
Classifiers you may drop
tiga batang pensil
Classifiers you may drop
batang is also the noun “stem, trunk”
tiga ekor anjing
Classifiers you may drop
ekor is the noun “tail”
tiga orang
Classifiers you may drop
orang is the noun “person”
Bengali
Indo-European › Indo-Aryan
তিনটি বই tinṭi boi
A classifier picked by the thing
-ṭi attaches to the numeral; -jon is used for people
তিনজন tinjon
A classifier picked by the thing
Yucatec Maya
Mayan
óox-p’éel
A classifier picked by the thing
the inanimate classifier
óox-túul
A classifier picked by the thing
the animate classifier
óox-túul
A classifier picked by the thing
Swahili
Atlantic-Congo › Bantu
vitabu vitatu
Agreement instead
class 8 on the noun and on the numeral
mbwa watatu
Agreement instead
animals take class 2 agreement whatever their own prefix
watu watatu
Agreement instead
class 2
English
Indo-European › Germanic
three books
Nothing in between
three sheets of paper
A unit word only where needed
paper cannot be counted directly
three pencils
Nothing in between
three dogs
Nothing in between
three people
Nothing in between
three cars
Nothing in between
three houses
Nothing in between
German
Indo-European › Germanic
drei Bücher
Nothing in between
drei Blatt Papier
A unit word only where needed
Blatt keeps its singular form after a numeral
drei Bleistifte
Nothing in between
drei Hunde
Nothing in between
drei Menschen
Nothing in between
Russian
Indo-European › Slavic
три книги tri knigi
Nothing in between
the numeral puts the noun in a particular case, but adds no word
три человека tri cheloveka
Nothing in between
French
Indo-European › Romance
trois livres
Nothing in between
trois feuilles de papier
A unit word only where needed
trois personnes
Nothing in between

How the columns cluster

Two columns count as close when the same language uses one word for both. Branches joining further right are the pairs more languages merge; the column order in the grid and on the map follows this tree, so neighbours behave alike.

bookssheets of paperpencilsdogspeoplecarshouses
22% dogs · people 2 of 9 — Yucatec Maya, Swahili

Books sits at the far end. No language here uses its verb for sheets of paper, pencils, dogs, cars, people or houses.

Examples

Toggle between Natural / Literal / Gloss to see how each language conceptualises the same idea.

The same three books

One phrase in five languages: a classifier word in Japanese, Mandarin and Thai, class agreement on the numeral in Swahili, and nothing at all in English.

Japanese · 日本語

本を三冊買った。

hon o san-satsu katta.

Natural
I bought three books.
Mandarin Chinese · 普通话

我買了三本書。

wǒ mǎi le sān běn shū.

Natural
I bought three books.
Thai · ภาษาไทย

หนังสือสามเล่ม

nǎngsɯ̌ɯ sǎam lêm

Natural
Three books.
Swahili · Kiswahili

vitabu vitatu

Natural
Three books.
English

three books

Natural
Three books.

References

  1. Aikhenvald 2000 01
    Aikhenvald, Alexandra Y. (2000).
    Classifiers: A Typology of Noun Categorization Devices.
    Oxford Studies in Typology and Linguistic Theory. Oxford University Press.
  2. Ashton 1944 02
    Ashton, E. O. (1944).
    Swahili Grammar (Including Intonation).
    Longmans, Green & Co..
  3. Bisang 1999 03
    Bisang, Walter (1999).
    Classifiers in East and Southeast Asian languages: Counting and beyond.
    In Gvozdanović, Jadranka (eds.), Numeral Types and Changes Worldwide. Mouton de Gruyter. 113–185. doi:10.1515/9783110811193.113
  4. Blaha Pfeiler & Skopeteas 2024 04
    Blaha Pfeiler, Barbara; Skopeteas, Stavros (2024).
    Numeral classifiers in Yucatec Maya: Microvariation and syntactic change.
    Journal of Historical Syntax. 8(6). doi:10.18148/hs/2024.v8i6.219
  5. Chao 1968 05
    Chao, Yuen Ren (1968).
    A Grammar of Spoken Chinese.
    University of California Press.
  6. Downing 1996 06
    Downing, Pamela A. (1996).
    Numeral Classifier Systems: The Case of Japanese.
    Studies in Discourse and Grammar 4. John Benjamins. doi:10.1075/sidag.4
  7. Gil 2013 07
    Gil, David (2013).
    Numeral Classifiers.
    In Dryer, Matthew S.; Haspelmath, Martin (eds.), The World Atlas of Language Structures Online, chapter 55. Max Planck Institute for Evolutionary Anthropology. https://wals.info/chapter/55
  8. Grinevald 2000 08
    Grinevald, Colette (2000).
    A morphosyntactic typology of classifiers.
    In Senft, Gunter (eds.), Systems of Nominal Classification. Cambridge University Press. 50–92.
  9. Lucy 1992 09
    Lucy, John A. (1992).
    Grammatical Categories and Cognition: A Case Study of the Linguistic Relativity Hypothesis.
    Studies in the Social and Cultural Foundations of Language 13. Cambridge University Press.
  10. Nguyễn Đình-Hoà 1957 10
    Nguyễn Đình-Hoà (1957).
    Classifiers in Vietnamese.
    Word. 13(1): 124–152.
  11. Okell 1969 11
    Okell, John (1969).
    A Reference Grammar of Colloquial Burmese.
    Oxford University Press.
  12. Poulos & Msimang 1998 12
    Poulos, George; Msimang, Christian T. (1998).
    A Linguistic Analysis of Zulu.
    Via Afrika.
  13. Senft 2000 13
    Senft, Gunter (2000).
    Systems of Nominal Classification.
    Language, Culture and Cognition 4. Cambridge University Press.

Related patterns

Corrections

Any of this can be wrong. Dispute what the page says, add a language it is missing, or edit the file it was rendered from.