Word-Order Typology (@cite{dryer-haspelmath-2013} / WALS) #
@cite{dryer-1992} @cite{dryer-haspelmath-2013} @cite{gibson-2025} @cite{greenberg-1963}
WALS data from @cite{gibson-2025}: cross-linguistic counts of harmonic vs disharmonic word-order pairings. @cite{dryer-1992} documents that languages overwhelmingly prefer consistent head direction across construction types (the head-direction generalization, @cite{greenberg-1963}).
Data #
Three cross-tabulations from WALS:
- Table 1: VO × Adposition order (981 languages)
- Table 2: VO × Subordinator order (456 languages)
- Table 3: VO × Relative clause order (665 languages)
Each table is a 2×2 contingency table: VO/OV × head-initial/head-final for the second construction. Harmonic cells (both HI or both HF) dominate.
Single-Word Exceptions (Table 4) #
Some constructions show disharmonic tendencies cross-linguistically: adjective-noun, demonstrative-noun, intensifier-adjective, negator-verb. @cite{gibson-2025} argues these are cases where the dependent is a single word (no recursive subtree), so head direction is irrelevant to DLM.
A single cell in a 2×2 alignment table.
dir1 is the direction for the first construction (verb-object order),
dir2 is the direction for the second construction.
- dir1 : HeadDirection
- dir2 : HeadDirection
- count : Nat
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
Instances For
Whether an alignment cell represents a harmonic (consistent-direction) pair.
Equations
- c.isHarmonic = (c.dir1 == c.dir2)
Instances For
A 2×2 cross-tabulation of two construction types.
- name : String
- construction1 : String
- construction2 : String
- hihi : AlignmentCell
- hihf : AlignmentCell
- hfhi : AlignmentCell
- hfhf : AlignmentCell
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Total number of languages in the table.
Equations
Instances For
Whether harmonic pairings are the majority.
Equations
Instances For
Table 1: Verb-Object order × Adposition order (@cite{dryer-haspelmath-2013}, WALS). @cite{gibson-2025} Table 1. 981 languages.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Table 2: Verb-Object order × Subordinator order (@cite{dryer-haspelmath-2013}, WALS). @cite{gibson-2025} Table 2. 456 languages.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Table 3: Verb-Object order × Relative clause order (@cite{dryer-haspelmath-2013}, WALS). @cite{gibson-2025} Table 3. 665 languages.
Equations
- One or more equations did not get rendered due to their size.
Instances For
All three cross-tabulations from @cite{gibson-2025}.
Equations
Instances For
Table 1: harmonic (926) > disharmonic (55).
Table 2: harmonic (393) > disharmonic (63).
Table 3: harmonic (547) > disharmonic (118).
The head-direction generalization: across all three construction pairs, harmonic word-order pairings dominate.
This is the core empirical observation that @cite{gibson-2025} argues DLM explains: consistent head direction keeps recursive spine dependencies local.
Harmonic cells have matching directions.
Disharmonic cells have mismatched directions.
Construction types where disharmonic order is common (Gibson Table 4).
These are cases where the dependent is typically a single word (no recursive subtree), so head direction doesn't affect DLM. Gibson's argument: DLM only cares about direction when subtrees intervene between head and dependent.
Data here is qualitative — WALS counts not provided in Gibson for these. Marked as data TBD for future WALS lookup.
- adjN : SingleWordException
- demN : SingleWordException
- intensAdj : SingleWordException
- negVerb : SingleWordException
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
All single-word exceptions from Gibson Table 4.
Equations
- One or more equations did not get rendered due to their size.
Instances For
These exceptions all involve dependents that are typically single words (leaves in the dependency tree), not recursive phrases.
Equations
- Phenomena.WordOrder.Typology.isSingleWordDependent Phenomena.WordOrder.Typology.SingleWordException.adjN = true
- Phenomena.WordOrder.Typology.isSingleWordDependent Phenomena.WordOrder.Typology.SingleWordException.demN = true
- Phenomena.WordOrder.Typology.isSingleWordDependent Phenomena.WordOrder.Typology.SingleWordException.intensAdj = true
- Phenomena.WordOrder.Typology.isSingleWordDependent Phenomena.WordOrder.Typology.SingleWordException.negVerb = true
Instances For
Ch 81 total: 1376 languages.
Ch 82 total: 1496 languages.
Ch 83 total: 1518 languages.
WALS Ch 81: The six-way classification of basic constituent order.
The "basic" order is determined by the dominant order in pragmatically neutral, declarative clauses with full NP arguments. Languages where no single order clearly dominates (e.g., free word-order languages like Warlpiri, or V2 languages like German where underlying order is debated) are classified as "no dominant order."
Key finding: SOV and SVO together account for >76% of languages; object-initial orders (OVS + OSV) are vanishingly rare. This asymmetry is one of the most robust typological generalizations: subjects overwhelmingly precede objects.
- sov : BasicOrder
- svo : BasicOrder
- vso : BasicOrder
- vos : BasicOrder
- ovs : BasicOrder
- osv : BasicOrder
- noDominant : BasicOrder
Instances For
Equations
- Phenomena.WordOrder.Typology.instBEqBasicOrder.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
- Phenomena.WordOrder.Typology.instBEqSVOrder.beq x✝ y✝ = (x✝.ctorIdx == y✝.ctorIdx)
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
WALS Ch 83: Binary classification of O-V order.
The most theoretically significant binary parameter: whether the object precedes the verb (OV = head-final VP) or follows it (VO = head-initial VP). This single parameter correlates with adposition order, genitive order, relative clause order, and subordinator order — the head-direction generalization formalized in the cross-tabulations above.
Instances For
Equations
- Phenomena.WordOrder.Typology.instBEqOVOrder.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 basic word-order profile across WALS Chapters 81--83.
Pairing a language's 6-way basic order (Ch 81) with its binary S-V (Ch 82) and O-V (Ch 83) classifications enables cross-chapter consistency checks: e.g., an SOV language should have SV order and OV order.
- language : String
Language name.
- iso : String
ISO 639-3 code.
- basicOrder : BasicOrder
Ch 81: basic order of S, O, V.
- svOrder : SVOrder
Ch 82: order of subject and verb.
- ovOrder : OVOrder
Ch 83: order of object and verb.
- notes : String
Notes on the word-order system.
Instances For
Equations
- One or more equations did not get rendered due to their size.
Instances For
Japanese: rigid SOV with postpositions, relative clause before noun, genitive before noun. The canonical head-final language.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Turkish: SOV with postpositions. Relatively free word order for pragmatic effects, but SOV is strongly dominant.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Hindi-Urdu: SOV with postpositions. Word order is relatively flexible due to case marking, but SOV is the default neutral order.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Korean: rigid SOV with postpositions, relative clause before noun. Very similar head-final profile to Japanese.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Quechua (Southern): rigid SOV with postpositions, suffixal morphology. One of the most consistently head-final languages.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Basque: SOV with postpositions. Ergative-absolutive case system. Word order is flexible but SOV is the pragmatically neutral order.
Equations
- One or more equations did not get rendered due to their size.
Instances For
English: rigid SVO with prepositions and relative clause after noun. The canonical head-initial language (among non-V-initial types).
Equations
- One or more equations did not get rendered due to their size.
Instances For
Mandarin Chinese: SVO with prepositions (mostly). Some constructions are head-final (relative clause before noun, some postpositions), making Mandarin a mixed-headedness language.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Russian: SVO as the pragmatically neutral order, but word order is relatively free due to rich case morphology.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Swahili: SVO with prepositions. Bantu language with rich verbal agreement that cross-references subject and object.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Indonesian: SVO with prepositions. Relatively rigid word order for an Austronesian language.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Arabic (Modern Standard): VSO as the classical/formal basic order, though SVO is increasingly common in spoken varieties.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Irish: VSO with prepositions, relative clause after noun. Celtic languages are the canonical European VSO family.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Tagalog: VSO / VOS (verb-initial with flexible S/O ordering). WALS classifies as VSO based on pragmatically neutral clauses. Philippine-type voice system complicates the S vs O distinction.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Welsh: VSO with prepositions, relative clause after noun. Another Celtic VSO language.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Malagasy: VOS with prepositions. The best-known VOS language, extensively studied in formal syntax.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Tzotzil (Mayan): VOS basic order, common in the Mayan family.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Hixkaryana (Cariban; Brazil): the first language for which OVS basic order was convincingly demonstrated. This discovery refuted the earlier claim that OVS and OSV orders were unattested.
Equations
- One or more equations did not get rendered due to their size.
Instances For
German: V2 in main clauses, SOV in embedded clauses. WALS classifies as SVO, but many syntacticians analyze the underlying order as SOV with V2 movement. The tension makes "no dominant order" defensible (though WALS itself codes German as SVO).
Equations
- One or more equations did not get rendered due to their size.
Instances For
Warlpiri (Pama-Nyungan; Australia): radically free word order, with all six permutations of S, O, V attested in natural discourse. The canonical non-configurational language.
Equations
- One or more equations did not get rendered due to their size.
Instances For
All basic-order language profiles in the sample.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Count of languages in a profile list with a given basic order.
Equations
- Phenomena.WordOrder.Typology.countByBasicOrder langs o = (List.filter (fun (x : Phenomena.WordOrder.Typology.BasicOrderProfile) => x.basicOrder == o) langs).length
Instances For
Count of languages in a profile list with a given S-V order.
Equations
- Phenomena.WordOrder.Typology.countBySVOrder langs o = (List.filter (fun (x : Phenomena.WordOrder.Typology.BasicOrderProfile) => x.svOrder == o) langs).length
Instances For
Count of languages in a profile list with a given O-V order.
Equations
- Phenomena.WordOrder.Typology.countByOVOrder langs o = (List.filter (fun (x : Phenomena.WordOrder.Typology.BasicOrderProfile) => x.ovOrder == o) langs).length
Instances For
Generalization 1: SOV is the most common basic order.
Generalization 2: SOV + SVO together exceed 75% of all sampled languages.
Generalization 3: In Ch 83, OV slightly outnumbers VO.
Generalization 4: Subject-first orders (SOV + SVO) far outnumber verb-first orders (VSO + VOS).
Generalization 5: Object-initial orders (OVS + OSV) are extremely rare — less than 2% of the sample.
Generalization 6 (Greenberg's Universal 1): In declarative sentences with nominal subject and object, the subject almost always precedes the object. SOV + SVO + VSO (subject before object) vastly outnumber VOS + OVS + OSV (object before subject).
Generalization 7: SV overwhelmingly dominates VS in Ch 82. SV languages outnumber VS languages by more than 5 to 1.
Generalization 8: Cross-chapter consistency — all SOV languages in our sample have OV order (Ch 83) and SV order (Ch 82).
Cross-chapter consistency — all SVO languages in our sample have SV order (Ch 82) and VO order (Ch 83).
Cross-chapter consistency — all VSO languages in our sample have VS order (Ch 82) and VO order (Ch 83).
Cross-chapter consistency — all VOS languages in our sample have VS order (Ch 82) and VO order (Ch 83).
Generalization 9: All OV languages in our sample have basic order SOV or OVS (the two S-containing OV orders).
All VO languages in our sample have basic order SVO, VSO, or VOS.
Number of languages in our basic-order profile sample.
Distribution of basic orders in our sample.
For each profile whose language is in WALS Ch 81A, prove its basic order matches the WALS data. Languages not in the WALS sample (Malagasy, Quechua) are excluded — their profiles are ungrounded assertions.
Ch 84 total: 500 languages.
Ch 85 total: 1184 languages.
Ch 86 total: 1249 languages.
Ch 87 total: 1367 languages.
Ch 88 total: 1225 languages.
Ch 89 total: 1154 languages.
Ch 90 total: 824 languages.
Ch 91 total: 481 languages.
Ch 94 total: 659 languages.
Ch 95 total: 1142 languages.
Ch 96 total: 879 languages.
Ch 97 total: 1316 languages.
Ch 81B total: 67 languages.
Ch 90B total: 191 languages.
Ch 90C total: 620 languages.
Ch 90D total: 63 languages.
Ch 90E total: 23 languages.
Ch 90F total: 10 languages.
Ch 90G total: 5 languages.
Ch 60 total: 138 languages.
Ch 61 total: 124 languages.