Chapter 94Pro-Drop, (Non)configurationality, and the Pronominal-Argument Debate

94.1 Overview

Hokkaido Ainu combines two properties whose joint presence creates genuine analytical tension. The finite verb is clause-final without exception, and every phrase type is head-final: modifier precedes head, postposition follows noun, subordinate clause precedes main clause Nakagawa (2024: 363); Shibatani (1990: 22–24). At the same time, within the preverbal field the relative order of noun phrases is controlled by information structure rather than grammatical function: the same three-argument verb epakasnu 'teach' is attested with its arguments in at least four different preverbal sequences in natural speech from a single speaker, and no case particle indicates which NP is subject and which is object Nakagawa (2024: 363–364); Bugaeva (2022). What fixes grammatical roles is the verb's obligatory person-index system: bound argument-marking affixes appear on every finite predicate, and an overt NP is structurally optional — an elaboration on a participant already indexed, not a required structural position Satō (2025: 359–361); Bugaeva (2012: 472).

These properties place Ainu in the cross-linguistic neighbourhood of head-marking languages such as Mohawk and Lakhota, whose argument NPs have been analysed as adjuncts to pronominal argument positions inside the verb — the Pronominal-Argument Hypothesis (PAH) as reviewed by Bugaeva (2022). Whether that analysis extends to Ainu is a question the accessible literature addresses but does not settle; the present chapter sets out the evidence, the competing positions, and what Ainu unambiguously does and does not exhibit. The verb-finality rule and the head-final NP structure are treated in Chapter 92 (Basic Constituent Order and Head-Final Syntax); the person-affix paradigm, alignment typology, and ditransitive indexing cap belong to Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms) and its sibling chapters.

94.2 Obligatory head-marking and the appositional noun phrase

A finite verb in Ainu marks its arguments by means of bound person prefixes and, for some persons, suffixes; these affixes are obligatory on every predicate. Satō, following the head-marking framework of Nichols (1986), describes the verb as the clausal head that bears the central information about which participants fill its valence frame Satō (2025: 361). A finite verb alone, with no overt NPs at all, is a complete and grammatical clause:

When an overt NP does appear, its relationship to the verb is appositional rather than structurally necessary. In Role and Reference Grammar terms, Satō places the person prefix (PRO) inside the CORE as the actual argument, with any co-occurring NP outside the CORE as an optional, non-obligatory adjunct Satō (2025: 363–365). This is the Lakhota-type structure: "the same logical structure gives rise to morphologically different structures" depending on how many overt NPs are supplied Satō (2025: 365). The overt object NP in the following example holds precisely that adjunct status:

Personal pronouns, being informationally redundant with the person index, are used only for emphasis or contrast and are rarer in Ainu than their Japanese counterparts Bugaeva (2012: 472); Satō (2025: 360). The independent pronoun kani 'I' can co-occur with ku=mina 'I laughed', but without kani the clause still indexes a specific first-person singular subject; the pronoun is emphatic, not anaphoric Bugaeva (2012: 472).

The possessor NP stands in the same appositional relationship to the possessed noun: the possessee is obligatorily head-marked by an agreement prefix, so the possessor NP is normally omitted and expressed only when referentially new or contrastive Bugaeva (2012: 486). The community observation that the bare verb form with no person affix is the third-person form and the citation form of the verb — with the zero-marking implicitly conceding a zero third-person affix — is consistent with Satō's formal argument that such a zero is required by the two-affix cap on finite verbs Satō (2025: 369) (nukopoli, aynu-corpora Discord, 2024-12-25, asserted; aynu-corpora Discord (2023–2026)) ‹contested›. This debate is developed in Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms).

94.3 The free ordering of preverbal noun phrases

With intransitive predicates the single preverbal NP is unambiguously the S argument. With transitives, the evidence is divided, and the ordering question is the analytical crux of this chapter.

Shibatani uses the following minimal pair to argue that linear position determines grammatical function:

Both NPs are third person, zero-marked; role assignment differs only by position. Shibatani concludes that word order "plays as important a role as in English in determining the grammatical functions of certain nouns" Shibatani (1990: 22), allowing object-first order only when context or semantics rules out an incorrect reading. Satō's introductory description is comparably cautious: the role of an NP is "shown by word order or context" Satō (2008: 2). The pedagogical grammar of Ijäs (2023) likewise describes Ainu as SOV and explains apparent flexibility by appeal to the zero third-person marker, such that "if we put the linguists' zero mark (Ø) in our example sentence … the word order indeed is correct." This framing is pedagogically effective but does not engage the evidence for genuine order freedom.

That evidence comes from Nakagawa's Chitose data. Three utterances by the same speaker describing the same scene place the three arguments of epakasnu 'teach' in three different preverbal sequences, all translating as 'my wife taught our daughter women's work' Nakagawa (2024: 363, exx. 700–702); reproduced in English at Nakagawa (2022: exx. 71–73):

A fourth order (DO – A – IO) is attested in a Saru text Nakagawa (2024: 363–364, ex. 703). Nakagawa's conclusion is explicit: "since the noun phrase itself carries no indicator of which is subject and which is object, which one is the subject cannot be grammatically determined"; the disambiguation rests on "common sense" and knowledge of the scene, not on syntactic position Nakagawa (2024: 363). Bugaeva, drawing on the same data, states that Ainu "does not show its subject and object either by case marking or by word order" Nakagawa (2022). This is the ‹contested› heart of the question: whether role is fixed by first-noun default (Shibatani, Satō 2008) or is genuinely free and recoverable only via person-indexing and real-world plausibility (Nakagawa, Bugaeva). No corpus-level count of A-before-O versus O-before-A frequencies in transitive clauses exists in the literature consulted; the Nakagawa triple is the strongest available evidence, but it derives from one speaker in one session ‹contested›.

The topic particle anak(ne) in example 702 is a distinct confound: when a NP is overtly marked as topic it tends to come first regardless of grammatical role Nakagawa (2024: 364); Bugaeva (2012: 472). The fuller treatment of topicalization and left-dislocation belongs to Chapter 140 (Topic and Focus Marking: anak(ne), patek, and ka) and Chapter 142 (Pragmatic Word-Order Permutation, Dislocation, and Argument Ellipsis). The community observation that anak/anakne is used with far less frequency than Japanese は, carrying approximately the weight of 'as regards' rather than a default topic marker, is consistent with the descriptive literature's picture of a device that marks rather than requires left-periphery position (nukopoli, aynu-corpora Discord, 2024-11-09, asserted; aynu-corpora Discord (2023–2026); cf. Shibatani (1990: 38)) ‹corpus-suggested›.

A separate class of word-order evidence comes from right-dislocation: some Ainu texts place the subject or object NP after the verb, attached via the light verb ki 'do', providing participant specification after the predicate has already established the referential relations Bugaeva (2022). This discourse-level device is a processing-driven supplement to verb-final order, not a grammaticalized alternative to it.

94.4 Null arguments and zero anaphora

Null arguments are the default state of Ainu syntax. A verb's person affixes index every participant, so overt NPs appear only when they carry additional discourse work — introducing a referent, contrasting it, or marking it as topic. Nakagawa formalizes this in his sentence template: a NP "need not be present in the sentence as long as it can be supplied meaningfully from context, and the sentence still stands" Nakagawa (2024: 361). Satō, framing the same observation under head-marking, writes that "grammatical relations (subject and object) … are not obligatory elements in Ainu; rather it is the personal affixes that are obligatory for the predicates" Satō (2025: 359–360).

When a participant is contextually available, the standard strategy is zero anaphora: the missing argument triggers no overt substitute, not even an independent pronoun. Bugaeva illustrates this for a demoted oblique: in itanki huraye hine (itanki) ani Ø=i=ko-i-puni 'she washed a bowl and served me food with (that bowl)', the second mention of the bowl "can easily be omitted and the sentence would trigger a zero-anaphora interpretation 'holding (that bowl)'" Bugaeva (2012: 476, ex. 22). Ditransitives extend the same pattern: the language's two-affix cap means a third argument cannot be indexed, so it is expressed as a bare NP whose reference is guaranteed by verbal valency rather than by its own person marking Satō (2025: 368–369). The cap is treated in Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms) and the object-indexing chapters.

The following attested clause illustrates how overt possessed-NPs in narrative serve to reintroduce participants rather than maintain ongoing reference. Both NPs name the narrator's parents, which context would have made recoverable, but appear here to foreground the two co-subjects of the embedding:

Reference continuity in oral narrative depends primarily on the unbroken chain of person affixes across linked clauses, not on overt NP repetition. Re-introduction by NP marks a scene change or new referential ground, a function treated in the reference-tracking chapter (Chapter 139 (Reference Tracking and Argument Continuity)).

94.5 Discontinuous noun phrases

A standard diagnostic for nonconfigurationality is the permissibility of discontinuous NPs: two parts of one noun phrase separated by material that is not part of that phrase. Nakagawa provides an attested Saru instance in which the head noun toma 'mat' and its numeral sinep 'one' are separated by the focus particle ka 'even', yet the predicate isam 'be absent' is an intransitive verb requiring exactly one S argument:

Nakagawa frames this via verbal arity rather than via the nonconfigurationality literature, but structurally it instantiates what that literature treats as quantifier float: the numeral has drifted from its host noun into a position that an independent argument would occupy. The community observation that the pattern N + adverbial-numeral + V (pasuy sinep uk 'takes one chopstick') is "perfectly normal, maybe more frequent than in Japanese" is consistent with quantifier float being productive in Ainu, though frequency claims remain unverified (nukopoli, aynu-corpora Discord, 2024-12-12, proposed; aynu-corpora Discord (2023–2026)) ‹corpus-suggested›. The NP-internal ordering of head and numeral is treated in Chapter 42 (The Structure of the Noun Phrase).

94.6 Configurationality diagnostics assessed

The standard diagnostics for nonconfigurationality, as applied to Ainu in Bugaeva (2022) and Satō (2025), pattern as follows:

Six of the seven diagnostics are positive. The seventh — rigid verb-finality — is a counter-signal. In the best-described nonconfigurational languages (Warlpiri, Jiwarli), the verb can appear in any position among the free NPs. Ainu's verb-finality is absolute: every adverbial, every NP, every negative somo precedes the verb phrase, but the verb phrase itself does not move. Nichols notes that head-marking languages achieve word-order freedom through either verb-initial or verb-final strategies with equal ease (Nichols 1986: 104); Ainu takes the verb-final option. Its preverbal field is pragmatically ordered but bounded by the clause edge.

The combination — verb-final, head-marking, caseless core arguments, free preverbal NP order — is typologically rare. Nakagawa observes that languages with SOV order that mark neither subject nor object by case nor by fixed position "are probably relatively rare"; he cites Nivkh as the nearest geographic neighbour and notes that Nivkh compensates with obligatory immediately-preverbal object position plus a consonant-mutation dependency, while Ainu does neither Nakagawa (2022).

94.7 The Pronominal-Argument Hypothesis and the Ainu counter-evidence

The Pronominal-Argument Hypothesis holds that in head-marking languages the bound affixes are the actual grammatical arguments and the free NPs are adjuncts — lexical specifications of what the pronominal arguments refer to, with no direct argument status of their own. This position is associated with the work of Jelinek and of Baker, as reviewed in Bugaeva (2022) and in Foley & Van Valin (1984). The logical consequence for syntax is that there are no argument positions for NPs to fill, which is why they can appear discontinuously and in any order: they are not in the phrase structure at all. Ainu's profile — obligatory affixes, optional NPs, free preverbal order, null anaphora, discontinuous NPs, no case — fits this description wherever the accessible data bear on it ‹contested›.

Bugaeva presents the hypothesis explicitly and then argues against the pure appositive reading for Ainu:

"There remains an important theoretical issue of whether NPs in head-marking languages are arguments or clause constituents of any kind at all. A number of researchers consider them mere appositives that lexically specify or qualify the actual arguments, which are on the verb. … However, a more widely accepted view is that argumenthood is shared between bound pronouns and coreferential NPs. As to Ainu, the latter view seems to fit better because there is an extra <location> elaboration on the noun …" Bugaeva (2022)

The Ainu-specific counter-argument turns on a subset of nouns with inherent locative semantics. When two third-person arguments of a three-place verb are both zero-marked, the locative-noun class — nouns such as or 'place (of)', sam 'near', ka 'above', corpok 'below', as well as the category of place-role nouns (kim 'mountain-area', pis 'beach-area') — distinguishes the Location argument from the Theme. Nakagawa's Chitose example [ci a=e-p] [saranip or] oma-re 'she put the cooked food into the basket' shows the mechanism: the NP saranip or (basket + locative or) is unambiguously the Goal because it contains a locative noun, leaving the other NP as Theme; without or, the two basket-and-food NPs would be ambiguous Nakagawa (2022). Bugaeva reads this as evidence that the noun carries grammatically relevant information about its own argument role, which a purely appositive / PAH NP would not need to do Bugaeva (2022).

The counter-argument has two acknowledged limits. First, it applies only when one argument is a Location: in the many transitive clauses where neither argument is locative — including the kamuy aynu rayke pair — the disambiguation rests on plausibility and world knowledge, not on lexical noun class. Nakagawa is explicit that in those cases it falls back on "common sense" Nakagawa (2024: 363). Second, the argument rests on a small number of illustrative examples and has not been tested against a quantified corpus of three-argument clauses with two zero-marked third-person NPs ‹contested›.

Satō's analysis, using Role and Reference Grammar, reaches a structurally PAH-like result — person affixes fill the argument positions inside the CORE, NPs sit outside it as optional adjuncts — but within a framework that does not endorse the PAH label, and it explicitly rejects the head-movement analysis of Baker (1988) as unable to handle the attested subject incorporation in Ainu (koy-yanke 'the waves wash [me] ashore', where the A argument is incorporated into the verb) Satō (2025: 366–367); Satō (2016). Noun incorporation and its implications for argument structure are the subject of Chapter 88 (Object (Patient) Noun Incorporation) and Chapter 90 (Oblique Incorporation and the Syntactic-vs-Lexicalist Debate).

The three positions — PAH-type (NPs as adjuncts, affixes as sole arguments), shared argumenthood (Bugaeva), and RRG-structural (Satō) — produce near-identical empirical predictions for the phenomena described in this chapter; they differ in theoretical commitments about what counts as an argument. No Ainu specialist has published a study dedicated to the PAH or the nonconfigurationality hypothesis as such; the framing reaches Ainu through Bugaeva's and Satō's typological discussions. Chapter authors should present the classification as a live framing question rather than an established result Bugaeva (2022) ‹contested›.

94.8 What Ainu does not exhibit

Several properties typical of the best-described nonconfigurational languages are absent from or unattested in Hokkaido Ainu, and these absences limit the scope of any nonconfigurationality claim.

No morphological switch-reference. Warlpiri and many Australian languages deploy dedicated same-subject / different-subject morphology on non-finite verb forms (converbs). Ainu has no converbs and no morphological switch-reference (Chapter 131 (Clause Linkage in Ainu: Typology and Inventory); Chapter 133 (The hine/akusu Contrast and Switch-Reference-Like Linkage)). Reference continuity across linked clauses is maintained by the obligatory person affixes on each finite predicate, by zero anaphora, and by the fourth-person logophoric device — functional substitutes, not formal equivalents of switch-reference morphology Bugaeva (2012: 498).

No wh-movement. Content-question words remain in situ: hunna 'who', hemanta 'what', hunak 'where' occupy the same position they would hold in a corresponding declarative, with no movement to a left-peripheral question position Bugaeva (2012: 498). This is consistent with a head-marking typology but is shared with many language types that are not nonconfigurational.

Verb-finality is rigid. Unlike Warlpiri, where the finite verb participates in the general NP-scrambling freedom, the Ainu verb does not leave clause-final position. This is the one reliable constituent-order fact: every preverbal element — NPs, adverbials, negative somo, topic-marked constituents — is ordered by discourse and pragmatics, but the verb-phrase occupies clause-final position without exception Nakagawa (2024: 363); Shibatani (1990: 22). The verb-final constraint is the one point on which all sources, across the dispute about whether order is otherwise free, agree.

No dedicated person-tracking morphology on the NP. Case suffixes on NPs exist only for obliques (postpositional adverbs and case particles); core S/A/O NPs are bare. There is no obviation morphology on the NP, no "obviative" nominal suffix marking a secondary third-person referent. The fourth-person prefix set on the verb does analogous work for first- and fourth-person participants in discourse, but it is verbal morphology, not nominal, and it targets person, not rank-in-discourse Bugaeva (2012: §15.4.1) ‹contested› (Chapter 67 (The Indefinite/Fourth Person: Forms, Reference, and Agent-Defocusing)).

94.9 Reference tracking across clauses

Reference continuity in Ainu oral narrative is maintained primarily by the verbal person-index system combined with zero anaphora. A participant introduced by an overt NP in one clause is reidentified in later clauses by the recurrence of the same person index on the finite predicate, with no overt NP required. Re-mention by NP signals reintroduction, scene change, or contrastive foregrounding — not ordinary anaphoric continuity.

The fourth-person form-set (Chapter 67 (The Indefinite/Fourth Person: Forms, Reference, and Agent-Defocusing)) performs additional tracking work in oral narrative. It marks the protagonist's viewpoint and first-person reference inside the reported world, distinct from both the narrating speaker's first person (ku=) and the unmarked third (zero). In performance, narrators alternate a= (fourth-person agent) and ku= (first-singular agent) to distinguish the in-story protagonist's speech from the narrator speaking as herself; this alternation has been observed in attested uwepeker performance (thegodofneet, aynu-corpora Discord, 2024-07-04, asserted; aynu-corpora Discord (2023–2026); consistent with Bugaeva (2008: 40–43)) ‹corpus-suggested›. The logophoric analysis of the fourth-person narration device is in Chapter 150 (First-Person Narration, the Logophoric Fourth Person, and Reported Discourse).

For clauses with two zero-marked third-person arguments, the locative-noun class provides a lexical disambiguation device. When one argument is a locative noun or a noun modified by a locative element (or, sam, etc.), it is unambiguously the Location argument, leaving the other NP as Theme or Patient Nakagawa (2022). Outside that subset, disambiguation rests on plausibility. Ainu therefore substitutes a combination of person-index transparency, lexical semantics, and discourse context for the case-marking and fixed-position strategies that other SOV languages use to track grammatical role across clauses Bugaeva (2022).

Verbal number contributes independently in some environments: the transitive plural stems mark object plurality — rayke → ronnu 'kill' — an ergative-like S≈O≠A pattern in the number system that disambiguates S and O for zero-marked third persons when the count of participants differs Bugaeva (2022) (consistent with the community observation by nukopoli, aynu-corpora Discord, 2023-12-12, asserted; aynu-corpora Discord (2023–2026)). The fuller treatment of verbal number as a grammatical resource is in Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa) and Chapter 72 (Tripartite, Nominative-Accusative, and the Central Alignment Debate).

Clause-linkage particles pattern with participant continuity statistically but not morphologically: wa correlates with same-subject sequences in Refsing's Shizunai count, but the correlation is not grammatically enforced and carries no formal same-subject / different-subject encoding Refsing (1986). This is treated in full in Chapter 132 (Sequential wa and Clause Chaining) and Chapter 133 (The hine/akusu Contrast and Switch-Reference-Like Linkage).