Chapter 92Basic Constituent Order and Head-Final Syntax

92.1 The head-final profile

Hokkaido Ainu is verb-final. Every non-verbal element of the clause — nominals, adverbials, negators, subordinate clauses — precedes the verb phrase, and the verb phrase occupies the rightmost position in the sentence. This verb-finality holds without exception: no grammatical process moves a finite verb to a non-final position Nakagawa (2024: 363). The same head-final logic ramifies down into the noun phrase — modifiers, relative clauses, and property verbs precede the head noun — and upward into clause-level relations — subordinate clauses precede main clauses, and markers that relate an NP to the predicate follow the NP as postpositions. Bugaeva gives the canonical summary: "the basic constituent order in Ainu is SV/AOV. Modifiers are prepositive, and subordinate clauses come before main clauses. All particles are postpositional. Topical NPs tend to come first" Bugaeva (2012: 472).

The deep typological fact behind these orders is head-marking. Satō frames the profile through Nichols's (Nichols 1986) head-marking / dependent-marking distinction: the verb, as clausal head, bears obligatory person-indexing suffixes and prefixes that already satisfy all argument slots; overt NPs are accordingly optional, and when present they are appositional to those indexes rather than positional carriers of grammatical role Satō (2025: 359–361). The surface resemblance to Japanese — both verb-final, both postpositional — therefore conceals a fundamental typological difference: Japanese is dependent-marking, with case particles on the NP encoding grammatical relations; Ainu is head-marking, with person markers on the verb doing that work Satō (2025: 359–360); Bugaeva (2012: 466).

For the person-indexing apparatus that underpins this system, see Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms). For the question of whether optional and freely ordered overt NPs license a nonconfigurational / pronominal-argument analysis of Ainu, and for the Ainu-internal counter-evidence from the 〈location〉-noun class, see Chapter 94 (Pro-Drop, (Non)configurationality, and the Pronominal-Argument Debate) and Chapter 93 (Argument Realization: Affix Indexing and Free NP Order). For the overall typological placement of Hokkaido Ainu, see Chapter 2 (A Typological Profile of Hokkaido Ainu).

92.2 Verb-finality and the sentence template

Nakagawa (2024: 361, §10.1) schematizes the Ainu sentence as a left-to-right sequence:

[adverbials 0…x]   [NPs 0…n]   [verb phrase (n)]   [(evidential frame)]   (sentence-final particle)

Within the preverbal field, "there is no fixed rule for the order of NP–NP and NP–adverbial; both need only be before the verb phrase" Nakagawa (2024: 361). An evidential frame (ruwe ne, hawe ne, and the rest: see Chapter 120 (The Nominalization-plus-Copula Evidential Schema)) attaches to the right of the verb phrase; clause-linking particles (wa, hine: see Chapter 131 (Clause Linkage in Ainu: Typology and Inventory)) extend further right. Only a connective that introduces the next sentence sits outside the sentence boundary.

Head-finality applies at every level of structure. The following table — adapted from the Shibatani (1990: 23–24) inventory — shows the major construction types: Shibatani (1990: §3.2)

(Shibatani 1990: 22) notes one structural feature that sits awkwardly with the "ideal SOV" characterization: Ainu is simultaneously verb-final and heavily prefixing, with person-marking affixes, applicative prefixes, and incorporated nouns attaching to the left of the verb root. In the Greenbergian correlates of SOV order, postpositions, Rel–N, and Gen–N are harmonic, but prefixing morphology is not Shibatani (1990: 22). The prefixing pattern follows directly from head-marking: the clausal head hosts its argument indexes and derivational material to the left of the root, independently of the clause's surface word order ‹contested›.

92.3 Preverbal NP order: free or fixed?

Whether the ordering of agent (A) and object (O) NPs among the preverbal positions is grammatically fixed is the central word-order debate in Ainu studies. Two readings of the same data have been sustained in the literature.

Shibatani (1990: 22) argues that word order encodes grammatical role: "Since there are no case inflections on nouns, word order plays as important a role as in English in determining the grammatical functions of certain nouns." He presents the following pair as a minimal role-contrast Shibatani (1990: §3.2) ‹contested›:

Nakagawa counters with the observation that the same speaker, in the same session, produces three distinct orderings of the three arguments of the four-place verb epakasnu 'teach', all describing the same scene: 'my wife taught our daughter women's work' Nakagawa (2024: 363); reproduced in English in (Nakagawa 2022):

A further ordering (DO – A – IO) appears in a Saru instance by the same speaker Nakagawa (2024: 363–364). Nakagawa concludes that "which one is the subject cannot be grammatically determined" from word order or NP morphology; common sense — the plausibility that a wife teaches a daughter, not the reverse — is what rules out the inverted reading Nakagawa (2024: 363). Bugaeva independently states the same finding: Ainu "does not show the syntactic relations subject and object either by case marking or by word order" Nakagawa (2022) ‹contested›.

Satō (2008: 2) holds a middle ground: roles are signalled by "語順 or 文脈 (word order or context)", allowing that either mechanism may operate without assigning priority to either.

Refsing's Shizunai data present a third position (Refsing 1986), §2.4, reporting Chiri (1936: 167): a default SOV order when both subject and object are full NPs, with systematic deviation to O–S–V when the subject is a prefixal person marker (the object NP then precedes the prefix-bearing verb). Three-place clauses show "flexible" ordering, with IO before DO as the most frequent pattern when the subject is a prefix (§13.4.3). These regularities match Nakagawa's Chitose data as statistical tendencies, not as categorical rules ‹contested›.

The zero-marking of third-person arguments — the bare verb stem simultaneously serves as the third-person and citation form — means that when both A and O are third persons, no index disambiguates the roles, and interpretation falls entirely to common sense or to the 〈location〉-noun class of one of the NPs Satō (2025: 369). The third person bears no overt affix in the standard Saru–Chitose paradigm; a possessive construction already concedes that third-person indexing is zero-marked (nukopoli, aynu-corpora Discord (2023–2026) 2024-12-25, asserted, consistent with Satō (2025: 369)). See Chapter 67 (The Indefinite/Fourth Person: Forms, Reference, and Agent-Defocusing) for the fourth-person form-set and its contrast with the zero-marked third person; Chapter 49 (Personal, Reflexive, and Emphatic Pronouns) for the free pronouns.

92.4 Modifier order inside the NP

All prenominal modification is head-final: property verbs, relative clauses, demonstratives, genitives, and numerals all precede the head noun. Nakagawa (2024: §10.5) builds the NP compositionally, and the following table summarises the major modifier types:

The two N + N types differ in syntactic status. When both nouns are in the concept (non-possessed) form, the construction is a compound and the two elements cannot be separated: *sar tanne cironnup 'tail that-long fox' is ungrammatical. When the first noun is in the concept form and the second is in the possessed (affiliative) form, the construction is an appositive possessor–possessee pair — two NPs — and the possessor can be displaced by relativization or topicalization: saraha tanne cironnup 'the tail that is long (of the fox)' is well formed Nakagawa (2024: 380–382). This appositive possession structure, where the possessee is obligatorily head-marked and the possessor NP is optionally present or absent, is discussed in cross-linguistic perspective by (Bugaeva 2021). See Chapter 39 (Noun–Noun Compounding: Headedness, Linking, and Semantics) for compound nouns and Chapter 42 (The Structure of the Noun Phrase) for the full NP typology.

Relativization is prenominal and gap-based: the head noun is placed at the right edge of the modifying clause, and the verb's valency identifies which argument is missing. The following example shows the Rel–N order in Chitose:

The full prenominal relative-clause system — including case-retention inside the relative, headless relatives, and the continuum with nominalized complements — is treated in Chapter 103 (Relative Clauses: The Gap Strategy and the Accessibility Hierarchy).

Numerals occupy two alternative slots: a numeral-adnoun precedes the noun (Num–N, e.g., sine aynu 'one person') and a numeral-noun follows it (N–Num, e.g., ponmenoko sinep 'one young woman'); the two patterns are largely interchangeable for common numerals Nakagawa (2024: 383–384). In addition, a numeral-noun can float away from its head noun, separated from it by a focus particle, while both together still fill a single argument slot:

For the numeral and classifier system in full, see Chapter 55 (Numeral Classifiers, Ordinals, and Quantifier Syntax).

92.5 Postpositions and the marking of obliques

Core arguments — the S/A subject and the O object — are never marked by a postposition or case particle; their roles are encoded by person-marking on the verb Bugaeva (2012: 472, 476); Shibatani (1990: 22). Postpositions mark only adjunct and oblique relations: locative, ablative, goal, instrumental, and comitative. All follow their NP, consistently maintaining the N–Postp order.

Nakagawa (2024: 386–389, §10.6) and Bugaeva (2012: 476) distinguish two morphological classes. Postpositional adverbs can stand alone without a preceding NP and some can take person affixes; many derive from verbs still undergoing grammaticalization, such as ani ‘with, using’ (from 'hold') and tura ‘together with’. Postpositional particles (格助詞) cannot stand alone, cannot be person-indexed, and function as purely relational markers: the locative ta ‘at, in, on’ and un ‘to, at (inanimate goal)’, the ablative wa ‘from’, and the perlative peka ‘across, along’ belong to this class. The translative particle ne ‘as, becoming’ (homophonous with the copula) marks the NP-as-role ('as', 'in the role of'). The classes overlap in translation; the diagnostic is whether the element can form an adverbial phrase independently Nakagawa (2024: 388–389). Full treatment is in Chapter 53 (Postpositions, Local Case, Motion Events, Comitative, and Instrumental).

92.6 Negation, adjuncts, and topic fronting

The standard sentential negator somo ‘not’ and the prohibitive iteki ‘don't’ appear before the verb phrase. This is preverbal placement but not a special word-order fact: both are adverbs and take no person marker, so their preverbal position follows the general Adv–V rule directly Nakagawa (2024: 365); Shibatani (1990: 24).

Scope interactions of somo with TAM, evidentiality, and evidential-outer-frame negation (ruwe somo ne) are treated in Chapter 126 (Standard Clausal Negation with somo).

The preverbal field as a whole is freely ordered with respect to adjuncts and oblique phrases. Time adverbials, manner adverbs, and postpositional phrases can appear in any sequence before the verb phrase. Refsing (§2.4) declines to formulate a rule for adjunct ordering in Shizunai, finding it "very free" (Refsing 1986).

Topical NPs gravitate to sentence-initial position regardless of grammatical role, and they are often — though not obligatorily — overtly marked by the topic particle anakne ‘as for’:

Community discussion confirms that anak/anakne is used more sparingly than Japanese は: a referent already established in discourse need not be re-marked, and the particle carries the weight of '～に関しては (as regards ~)' rather than a routine topic reset (nukopoli, aynu-corpora Discord (2023–2026) 2024-11-09, asserted, consistent with Shibatani (1990: 38)). The full distribution of anakne — including its contrast with bare topic-initial placement, its scope, and the diachronic origin from an-yak — is in Chapter 140 (Topic and Focus Marking: anak(ne), patek, and ka). Pragmatic reordering beyond topic fronting, including right-dislocation and the afterthought construction with light verb ki, is treated in Chapter 142 (Pragmatic Word-Order Permutation, Dislocation, and Argument Ellipsis).

Certain adjunct-ordering details show sensitivity to language-contact pressure. The traditional sequence ani patek (instrumental ani before the focus adverb patek 'only') is in retreat in modern use in favour of the reversed patek ani. Community analysis argues that patek is an adverb and cannot be nominalized as Japanese だけ can, so placing it first — matching Japanese particle ordering — represents Japanese-influenced reordering (nukopoli, aynu-corpora Discord (2023–2026) 2024-12-16, proposed ‹contested›).

Questions do not disturb the basic constituent order. Content-question words — hunna ‘who’, hemanta ‘what’, hunak ‘where’ — remain in situ at the position of the corresponding non-question constituent; there is no wh-movement to a pre-clausal position Bugaeva (2012: 498). Polar questions are signalled by the sentence-final particle ya or by rising intonation. The full interrogative system is in Chapter 144 (Interrogative Strategies, Question Particles, and Evidential Questions). For focus-related constructions including cleft-nominalization, see Chapter 141 (Cleft, Pseudocleft, and Nominalization-Based Focus).