Cross-Linguistic Typology of Numeral Systems (WALS Chapters 53--56, 131) #
@cite{aikhenvald-2000} @cite{greenberg-1978} @cite{stolz-veselinova-2013}
Typological data on four dimensions of numeral morphology and syntax, drawn from WALS chapters by Stolz & Veselinova (Ch 53) and Gil (Ch 54--56). The data covers ordinal numeral formation, distributive numerals, numeral classifiers, and the relationship between conjunctions and universal quantification.
Ordinal Numerals (Ch 53, Stolz & Veselinova): How ordinal numbers are formed from cardinals. The dominant cross-linguistic pattern is for "first" to be suppletive (not derived from "one") while higher ordinals are regular derivations from cardinals. This asymmetry reflects the cognitive and discourse-functional salience of first position.
Distributive Numerals (Ch 54, Gil): Whether a language has a dedicated morphological form for distributive numerals ("two each", "three apiece"). Reduplication of the cardinal is the most common strategy among languages that mark distributives (e.g., Turkish iki-ser 'two each', Tagalog dalawa- dalawa 'two-two'). Many languages lack a dedicated form entirely.
Numeral Classifiers (Ch 55, Gil): Core types and distribution data live in
Phenomena.Classifiers.Typology. This file importsClassifierStatusfrom there and uses it inNumeralProfilefor cross-dimensional generalizations (Sanches-Slobin, areal concentration).Conjunctions and Universal Quantifiers (Ch 56, Gil): Whether a language uses the same morpheme for 'and' (conjunction) and 'all' / 'every' (universal quantification). Many languages show identity between these forms, suggesting a deep semantic connection between conjunction and universal quantification (both involve exhaustive predication over a set).
How a language forms ordinal numerals from cardinal ones (WALS Ch 53).
The key typological variable is whether "first" is suppletive (not derived from "one") and how many of the lowest ordinals show suppletion or irregularity. The pattern "first" suppletive + higher ordinals regular is overwhelmingly dominant.
- firstSuppletion : OrdinalFormation
- firstSecondSuppletion : OrdinalFormation
- allFromCardinals : OrdinalFormation
- various : OrdinalFormation
- noOrdinals : OrdinalFormation
Instances For
Equations
- Phenomena.Numerals.Typology.instBEqOrdinalFormation.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- d.total = d.firstSuppletion + d.firstSecondSuppletion + d.allFromCardinals + d.various + d.noOrdinals
Instances For
Actual WALS Ch 53 counts.
Equations
- Phenomena.Numerals.Typology.ch53Distribution = { firstSuppletion := 99, firstSecondSuppletion := 45, allFromCardinals := 28, various := 83, noOrdinals := 66 }
Instances For
Whether and how a language marks distributive numerals (WALS Ch 54).
Distributive numerals express "N each" or "N apiece" meanings. Languages vary in whether they have a dedicated morphological strategy and, if so, what kind. Reduplication of the cardinal numeral is the most widespread dedicated strategy (e.g., Japanese hito-ri hito-ri, Hungarian két-két).
- noDistributive : DistributiveNumeral
- markedByReduplication : DistributiveNumeral
- markedBySuffix : DistributiveNumeral
- markedByPrefix : DistributiveNumeral
- markedByOtherMeans : DistributiveNumeral
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- d.total = d.noDistributive + d.reduplication + d.suffixCount + d.prefixCount + d.otherMeans
Instances For
Actual WALS Ch 54 counts.
Equations
- Phenomena.Numerals.Typology.ch54Distribution = { noDistributive := 63, reduplication := 85, suffixCount := 34, prefixCount := 19, otherMeans := 50 }
Instances For
The relationship between 'and' and 'all/every' in a language (WALS Ch 56).
Gil distinguishes languages where the conjunction marker and the universal quantifier share the same form (identity) from those where they are morphologically distinct (differentiation). Identity between 'and' and 'all' reflects a deep connection between conjunction (exhaustive pairing) and universal quantification (exhaustive predication).
- identity : ConjunctionQuantifier
- differentiation : ConjunctionQuantifier
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
WALS Chapter 56 distribution: language counts per conjunction-quantifier type. Total: 220 languages.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- d.total = d.identity + d.differentiation
Instances For
Actual WALS Ch 56 counts.
Equations
- Phenomena.Numerals.Typology.ch56Distribution = { identity := 43, differentiation := 177 }
Instances For
Equations
- Phenomena.Numerals.Typology.instBEqRegion.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Whether a language has obligatory grammatical plural marking on common nouns. Used for the Sanches-Slobin generalization relating classifiers and plural.
- obligatory : PluralMarking
- optional : PluralMarking
- none : PluralMarking
Instances For
Equations
- Phenomena.Numerals.Typology.instBEqPluralMarking.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
The base of a language's numeral system (WALS Ch 131, Comrie).
Most languages worldwide use a decimal (base-10) system. Vigesimal (base-20) systems are the most common alternative, concentrated in Mesoamerica, West Africa, and the Caucasus. Many "vigesimal" systems are actually hybrid, using base-20 for higher numerals and base-10 within each score.
- decimal : NumeralBase
- vigesimal : NumeralBase
- hybridVigesimalDecimal : NumeralBase
- otherBase : NumeralBase
- bodyPartSystem : NumeralBase
- restricted : NumeralBase
Instances For
Equations
- Phenomena.Numerals.Typology.instBEqNumeralBase.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
A language's numeral typology profile across all four WALS dimensions.
- language : String
- iso : String
ISO 639-3 code
- ordinal : OrdinalFormation
Ch 53: Ordinal numeral formation
- distributive : DistributiveNumeral
Ch 54: Distributive numeral marking
- classifier : Classifiers.Typology.ClassifierStatus
Ch 55: Numeral classifier status
- conjQuant : ConjunctionQuantifier
Ch 56: Conjunction-quantifier relationship
- region : Region
Areal region (for areal generalizations)
- pluralMarking : PluralMarking
Plural marking on common nouns (for Sanches-Slobin)
- numeralBase : Option NumeralBase
Ch 131: Numeral base (optional; not all languages surveyed).
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Equations
- One or more equations did not get rendered due to their size.
- Phenomena.Numerals.Typology.instBEqNumeralProfile.beq x✝¹ x✝ = false
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
English: "first" is suppletive (not *one-th), "second" is suppletive (not *two-th), higher ordinals regular (-th suffix). No morphological distributive numerals (*two-each), no numeral classifiers, and conjunction 'and' differs from universal quantifier 'all'. Obligatory plural on nouns.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Mandarin Chinese: ordinals formed regularly with di- prefix (di-yi 'first', di-er 'second', etc.) --- no suppletion. No morphological distributive. Obligatory numeral classifiers (san ge ren 'three CL person'). Conjunction 'he' and universal 'dou' are different morphemes, but 'dou' shows quantificational force in both contexts. No grammatical plural on nouns.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Japanese: ordinals formed with -banme suffix (ichi-banme 'first', ni-banme 'second'). No suppletion. Distributive by reduplication (hitori-hitori 'one by one'). Obligatory numeral classifiers (san-nin 'three-CL.person', ni-hon 'two-CL.long.thing'). Conjunction 'to' differs from universal 'subete'. No grammatical plural on common nouns.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Thai: ordinals formed with thi- prefix (thi-nung 'first', thi-song 'second') --- all regular from cardinals. No morphological distributive. Obligatory numeral classifiers (maa sam tua 'dog three CL.animal'). No grammatical plural on nouns.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Vietnamese: ordinals formed with thu prefix (thu nhat 'first', thu hai 'second') --- regular from cardinals. No morphological distributive. Obligatory classifiers (ba con meo 'three CL cat'). No grammatical plural on nouns.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Turkish: "birinci" 'first' derived regularly from "bir" 'one' via -inci suffix (all ordinals formed by -inci or -unci suffix). Distributive numerals by suffix -er or -ar (iki-ser 'two each'). No obligatory numeral classifiers, but optional classifiers exist (iki tane kitap 'two CL book'). 'Ve' (and) differs from 'hepsi' or 'butun' (all). Obligatory plural with -ler or -lar.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Korean: ordinals use native Korean numerals + -jjae suffix (cheot-jjae 'first' partially suppletive). Distributive by suffix -ssik (du-ssik 'two each'). Optional numeral classifiers (se myeong-ui haksaeng 'three CL student'). 'Gwa/wa' (and) differs from 'modu' (all). Optional plural marking with -deul.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Hindi: ordinals show "first" (pehla) suppletive, higher ordinals regular with -vam suffix (dusra 'second' also partially suppletive). Distributive by reduplication (do-do 'two-two'). No numeral classifiers. 'Aur' (and) differs from 'sab' (all). Obligatory plural marking.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Bengali: ordinals formed with -tho suffix, but "first" (prothom) is suppletive. Distributive by reduplication. Optional classifiers (tin-ta boi 'three-CL book', but bare tin boi also grammatical). 'Ebong' (and) differs from 'sob' (all). Optional plural.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Burmese: ordinals formed regularly with prefix (pa-hta-ma 'first' from inherited Pali, but modern ordinals use prefix tha-). Numeral classifiers obligatory (lu thon yauk 'person three CL'). No morphological distributive form. No grammatical plural on nouns.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Hungarian: "first" (elso) is suppletive, higher ordinals regular with -dik suffix (masodik 'second', harmadik 'third'). Distributive by reduplication (ket-ket 'two-two'). No numeral classifiers. 'Es' (and) differs from 'minden' (all/every). Obligatory plural with -k.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Russian: "first" (pervyj) is suppletive, "second" (vtoroj) also suppletive, higher ordinals from cardinals with suffix (tretij 'third', chetvertyj 'fourth'). No morphological distributive (uses prepositional phrase 'po + dative'). No numeral classifiers. 'I' (and) differs from 'vse' (all). Obligatory plural marking.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Yoruba: ordinals formed with -(i)keji prefix system, varying patterns across the paradigm. No morphological distributive. No numeral classifiers. Conjunction 'ati' and universal quantifier 'gbogbo' are distinct. Plural marked optionally (awon).
Equations
- One or more equations did not get rendered due to their size.
Instances For
Swahili: ordinals formed with -a prefix + cardinal (wa-kwanza 'first' has distinct root, but -a-pili 'second' etc. are regular). No morphological distributive. No numeral classifiers (noun class system serves a different function). 'Na' (and) differs from '-ote' (all). Obligatory plural via noun class prefixes.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Tagalog: ordinals with pang- prefix (pang-una 'first' from una 'precede', pang-alawa 'second'). Distributive by reduplication (dalawa-dalawa 'two-two'). No obligatory numeral classifiers (linkers na/ng serve different function). 'At' (and) and 'lahat' (all) are differentiated. Optional plural (mga).
Equations
- One or more equations did not get rendered due to their size.
Instances For
Georgian: ordinals formed with me-...-e circumfix (me-or-e 'second', me-sam-e 'third'). "First" (p'irveli) is suppletive. Distributive by suffix -agan (or-agan 'two each'). No numeral classifiers. 'Da' (and) differs from 'q'vela' (all). Obligatory plural.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Finnish: ordinals formed regularly with -(n)s suffix (ensimmainen 'first' is suppletive, toinen 'second' from eri 'different', kolmas 'third' etc. regular). No morphological distributive. No classifiers. 'Ja' (and) differs from 'kaikki' (all). Obligatory plural with -t.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Tzeltal (Mayan, Mesoamerica): ordinals not productively formed. Numeral classifiers obligatory (distinct from Mayan noun classifiers). No morphological distributive. Conjunction and universal quantifier are differentiated. No obligatory plural on nouns.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Malay/Indonesian: ordinals with ke- prefix (ke-satu 'first' regular, pertama 'first' from Skt also used). No morphological distributive. Obligatory numeral classifiers (tiga orang murid 'three CL student'). 'Dan' (and) differs from 'semua' (all). Optional plural by reduplication.
Equations
- One or more equations did not get rendered due to their size.
Instances For
All language profiles in our sample.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Suppletive "first" is the dominant ordinal formation strategy (WALS Ch 53). Languages with suppletive "first" (alone or with suppletive "second") outnumber languages where all ordinals derive regularly from cardinals.
"First" suppletion alone is the single most common ordinal pattern.
Languages with some form of ordinal formation (regular or suppletive) outnumber languages lacking ordinals entirely. Most languages have ordinals.
Languages with dedicated distributive numeral forms outnumber those without, but neither is a negligible minority.
Reduplication is the single most common distributive strategy, outnumbering any other individual morphological means.
Differentiation between 'and' and 'all' is the dominant pattern (WALS Ch 56). Most languages use distinct morphemes for conjunction and universal quantification.
Differentiation accounts for more than three-quarters of the sample.
Identity between 'and' and 'all' is a non-negligible minority pattern, attested in roughly a fifth of languages (43 out of 220).
Does a language have obligatory numeral classifiers?
Equations
Instances For
Does a language have any numeral classifiers (obligatory or optional)?
Equations
Instances For
Does a language have obligatory plural marking on common nouns?
Equations
Instances For
Does a language form "first" by suppletion?
Equations
Instances For
Does a language have a morphological distributive numeral form?
Equations
Instances For
Is a language in the East/Southeast Asian region?
Equations
Instances For
Sanches-Slobin generalization: Classifier languages tend to lack obligatory plural marking on nouns. In our sample, every language with obligatory classifiers lacks obligatory plural.
The theoretical explanation: classifiers already individuate entities for counting, making obligatory plural morphology redundant.
Converse of Sanches-Slobin: every language in our sample with obligatory plural marking lacks obligatory classifiers. (This is the contrapositive, independently verifiable on the data.)
Numeral classifiers concentrate in East and Southeast Asia in our sample: every obligatory-classifier language in our sample is in East Asia, Southeast Asia, or Mesoamerica.
In our sample, East/Southeast Asian languages with obligatory classifiers all form ordinals regularly from cardinals (no suppletion of "first"). This is consistent with the observation that agglutinative numeral morphology in these languages extends uniformly to ordinals.
European languages in our sample all show suppletive "first" (either alone or with suppletive "second"). No European language in our sample derives "first" regularly from "one".
Japanese is a notable exception to any strict complementarity between obligatory classifiers and distributive morphology: it has both obligatory classifiers and distributive-by-reduplication (hito-ri hito-ri).
In our sample, the majority of obligatory-classifier languages lack morphological distributive forms. Japanese is the sole exception.
@cite{greenberg-1978}'s implicational universal for ordinal suppletion: if a language has a suppletive ordinal for numeral N, then it has suppletive ordinals for all numerals less than N. Equivalently: suppletion cuts off at some point in the sequence 1st, 2nd, 3rd,... and all ordinals above the cutoff are regular.
The WALS data captures the coarsest version: suppletion is most likely for "first", less likely for "second", and rare beyond that. Our inductive type captures three attested cutoff points (none, first-only, first+second).
- none : SuppletionCutoff
- first : SuppletionCutoff
- firstAndSecond : SuppletionCutoff
Instances For
Equations
- Phenomena.Numerals.Typology.instBEqSuppletionCutoff.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Numeric rank for the suppletion cutoff (higher = more suppletion).
Equations
Instances For
Map ordinal formation type to suppletion cutoff. Languages with 'various' or 'no ordinals' patterns are excluded from the hierarchy.
Equations
- Phenomena.Numerals.Typology.OrdinalFormation.allFromCardinals.suppletionCutoff = some Phenomena.Numerals.Typology.SuppletionCutoff.none
- Phenomena.Numerals.Typology.OrdinalFormation.firstSuppletion.suppletionCutoff = some Phenomena.Numerals.Typology.SuppletionCutoff.first
- Phenomena.Numerals.Typology.OrdinalFormation.firstSecondSuppletion.suppletionCutoff = some Phenomena.Numerals.Typology.SuppletionCutoff.firstAndSecond
- Phenomena.Numerals.Typology.OrdinalFormation.various.suppletionCutoff = none
- Phenomena.Numerals.Typology.OrdinalFormation.noOrdinals.suppletionCutoff = none
Instances For
The hierarchy is consistent: rank of each cutoff increases monotonically.
In our sample, the WALS aggregate confirms the hierarchy: languages with "first"-only suppletion are more numerous than those with "first+second" suppletion, which in turn are more numerous than those with no suppletion at all. This reflects the implicational scale: suppletion at higher numerals is rarer.
Ch 53 distribution from WALS data: firstSuppletion count (WALS "First, two-th, three-th" + "First, two, three" both map to firstSuppletion).
Ch 53: Mapped categories are exhaustive — all 321 languages are accounted for.
Ch 54 distribution from WALS data: reduplication is the most common strategy.
Ch 54: Mapped categories are exhaustive — all 251 languages are accounted for.
Ch 55 distribution from WALS data: all three categories match exactly.
Ch 131: Decimal is the dominant numeral base worldwide.
Ch 131: Pure vigesimal and hybrid vigesimal-decimal together.