Chapter 72Tripartite, Nominative-Accusative, and the Central Alignment Debate

72.1 The alignment question

Hokkaido Ainu encodes the S/A/O distinction through person indexes that appear on the finite verb — head-marking in the sense of (Nichols 1986). The architecture of the four formal persons and the combination rules are set out in Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms); the present chapter evaluates the typological labels that have been applied to the paradigm and assesses how far hierarchical and inverse analyses bear on the Hokkaido data. The split-intransitive question — whether the S position subdivides by agentivity — belongs to Chapter 71 (The Split-Intransitive (Active–Stative) Core). Free-NP realization and the relationship between indexing and nominal argument structure are taken up in Chapter 93 (Argument Realization: Affix Indexing and Free NP Order).

72.2 Row-wise alignment

Reading the person-indexing paradigm cell by cell, the morphological alignment is mixed. The standard Saru–Chitose forms, cited directly in Bugaeva, Nakagawa, Tamura, and Satō, yield four distinct row patterns Bugaeva (2012: 471–472); Nakagawa (2024: 165–169); Tamura (1972: 17–39); Satō (2008: 31–32).

Bugaeva's characterisation follows from the table: 1SG is nominative–accusative (S=A, distinct from O); second and third person are neutral (all three roles carry the same form); the first plural exclusive and the indefinite/fourth-person rows are tripartite, because each uses a distinct form for S, A, and O. She summarises the language as "mixed (but basically tripartite)" Bugaeva (2012: 472, 498); Bugaeva (2008: 32).

The tripartite character of those two rows rests on a positional split: the 1PL.EXCL A form ci= and the fourth-person A form a= are prefixes on the transitive stem, while the S counterparts =as and =an are suffixes on the intransitive stem; the O forms un= and i= are prefixal but formally distinct from both Nakagawa (2024: 166–167); Bugaeva (2012: 472). The Tokachi example places the 1PL.EXCL A and S exponents within a single clause chain ‹consensus›:

The fourth-person O prefix i= indexes the undergoer in oral narrative — including the narrating protagonist — and together with a= for A and =an for S makes the tripartite character of that row verifiable in a single utterance Nakagawa (2024: 169):

Bugaeva's three-way gradation of the indexes by prosodic and separability tests — word-like =an and =as, clitic-like a= and eci=, full prefixes such as ku= and en= — and Nakagawa's independent identification of =an as separable from the verb by particles reflect a cline of morphological tightness that is orthogonal to the alignment classification Bugaeva (2012: 472–473); Nakagawa (2024: 49, 183); Sakaguchi (2024: ix, 62–64, 77–85) ‹contested›. The ordering diagnostics belong to Chapter 69 (The Personal-Affix Template: Position Classes and Affix Ordering).

72.3 Catalogue of analyses

The literature assigns different labels because it addresses different questions: synchronic Hokkaido exponence, historical reconstruction, Sakhalin comparison, and the referential structure of the fourth-person set. The table below collects the principal positions; attributions via Dal Corso are second-hand for this grammar.

Dal Corso's mixed nominative–neutral–inverse account is formulated for ‹SA› Sakhalin. He dismisses the tripartite component for Sakhalin by arguing that the S and A forms sharing common origin belong to a nominative sub-system rather than tripartite cells, while the fourth-person or participant-referentiality-mismatch dimension is analysed as inverse Dal Corso (2025: 187, 207, 213). He is explicit that this diachrony may not extend to Ainu as a whole and that the hierarchical analysis may not apply to Hokkaido Dal Corso (2025: 188, 200) ‹contested›. The Hokkaido description accordingly retains Bugaeva's cell-based tripartite label for the two rows in which S, A, and O carry distinct forms and distinct sides of attachment.

72.4 The active attribution

An "active" or split-intransitive characterisation of Ainu person marking appears in the secondary literature and is commonly attributed to Shibatani. The text accessible for this grammar is the 1988 LSJ abstract, which treats noun incorporation and argues that the incorporable nominal corresponds to the patient subject of an intransitive clause or to a transitive object — explicitly invoking an unaccusative analysis Shibatani (1988: 210–211). The abstract assigns no alignment label to the person-indexing paradigm, In the 1990 monograph Shibatani does operate with an active/inactive classification of intransitives — his applicative discussion tests Baker's claim that no inactive intransitive clause feeds applicative formation, and falsifies it with Ainu data — but that is a verb-class distinction inside the incorporation argument, not an alignment label assigned to the person-indexing paradigm Shibatani (1990) ‹contested›.

The person-indexing paradigm itself does not partition the intransitive subject by agentivity. Agentive intransitives and patientive or change-of-state intransitives take the same S-form set once the verb is intransitive; the conditioning factor is valency, not semantic role Nakagawa (2024: 126–127). Satō's 2025 characterisation places Ainu with head-marking languages in the sense of (Nichols 1986) and develops the argument-indexing system without assigning an alignment label Satō (2025: 361, 373–374) ‹speculative›. The split that the literature pursues under the active or split-intransitive name surfaces in verb classes, incorporation, and valency-changing morphology, and is treated in Chapter 71 (The Split-Intransitive (Active–Stative) Core).

72.5 Hierarchical effects in person combinations

Two aspects of the paradigm are better described in hierarchical or referential terms than in a flat S/A/O frame: the speech-act-participant combination cells and the referential range of the fourth-person set.

The second-plural prefix eci= serves as the portmanteau for all combinations in which a first-person A acts on a second-person O — 1SG.A→2SG.O, 1SG.A→2PL.O, 1PL.EXCL.A→2SG.O, and 1PL.EXCL.A→2PL.O — using a single prefix where a sequence of first- and second-person indexes might be expected Tamura (1972: 27–28); Bugaeva (2008: 38); Bugaeva (2012: 474); Nakagawa (2024: 170, 173). The cell and its ambiguity with the plain 2PL marker are developed in Chapter 65 (The eci= Portmanteau (1A→2O) and Second-Person Plural Syncretism).

The 2→1 direction is irregular in a different way. Most dialectal combination tables carry a second-person A index followed by the first-person object form, but Saru and Chitose suppress the expected second-person subject prefix and use bare en= for 2SG→1SG; Chitose extends the bare form to 2PL→1SG as well Nakagawa (2024: 170, 173); Satō (2008: 146–147).

The fourth-person set adds a second hierarchical dimension through its referential range. Five uses cluster in the same form-set: the inclusive first plural, the honorific address to a respected second person, the indefinite or agent-defocused transitive subject, the narrative first person in oral literature, and logophoric reference inside a quotation Nakagawa (2024: 174–179); Tamura (1972: 18–20, 28, 34); Bugaeva (2008: 40–43). The honorific use brings the hierarchy most plainly into view: one addressee is indexed with the fourth-person suffix on a plural verb stem.

The fourth-person A prefix defocuses the transitive agent while the patient remains the grammatical object; Ainu has no subject-promoting passive. The same prefix participates in the indefinite construction that Japanese or English translations often render passively Tamura (1972: 34); Satō (2008: 203–205); Nakagawa (2024: 176–177).

Bugaeva analyses the inclusive and quotation uses as unified by a logophoric function: the set organises participant reference from a discourse perspective, with the singular pronoun asinuma glossed by native speakers as 'oneself' rather than 'I' Bugaeva (2008: 40–43); Bugaeva (2012: 474–476). Dal Corso's participant-referentiality-mismatch framing for ‹SA› Sakhalin treats the same dimension as a referential hierarchy of speech-act participants, but explicitly limits that analysis to Sakhalin data Dal Corso (2025: 196–200) ‹contested›.

The Hokkaido evidence supports a synthesis: person indexing is row-wise mixed in morphological alignment, while the fourth-person set and the eci= portmanteau and 2→1 irregular cells display a referential-participant hierarchy overlaid on the S/A/O frame. This overlay does not replace the tripartite exponence of the 1PL.EXCL and fourth-person rows; both dimensions are operative in the same paradigm ‹speculative›. The basis is the five-use analysis of the fourth person and the combination data cited here (Nakagawa (2024: 174–179); Tamura (1972: 27–28, 34); Bugaeva (2008: 38, 40–43)); the synthesis is offered as an integrative label for an existing pattern, not as a new theoretical commitment.

72.6 Third person and the two-affix cap

The third-person row is formally uniform — Ø for A, S, and O — but the interpretation is contested. Nakagawa treats third person as the absence of any affix: a bare noun phrase satisfies the verb's valency position without triggering a marker Nakagawa (2024: 169). Satō argues for a zero third-person affix, required to account for the cap of two personal indexes per verb and the ungrammaticality of three-index combinations in causative and ditransitive configurations Satō (2023: 39); Satō (2025: 369). Bugaeva's paradigm table writes Ø in the third row, describing the surface form without adjudicating between the two analyses Bugaeva (2012: 471) ‹contested›.

Ijäs's learner grammar presents the same operational generalization for L2 learners: a finite verb obligatorily indexes its non-third participants while third-person arguments surface as bare noun phrases Ijäs (2023). Her presentation is compatible with both sides of the zero-versus-absent debate.

Ijäs's fixed-order statement for ditransitives — indirect object between subject and object — is a teaching simplification. Nakagawa's three-place-verb examples show that the two object phrases and the subject are not locked in a single linear order Nakagawa (2024: 363–365); Ijäs (2023).

The ditransitive evidence bears directly on alignment. Satō argues that the object index in three-place verbs such as kore 'give' ordinarily identifies the human recipient while the theme stays unindexed — a secundative pattern in the indexing slot Satō (2008: 147); Satō (2023: 40, 47). Nakagawa describes the same verbs through valency, with two bare object phrases when both are third person Nakagawa (2024: 363–365). Both accounts converge on the generalization that a verb hosts at most one subject index and one object index; the difference is whether the unfilled third-person position contains a zero marker ‹contested›.

The full object-indexing system and its relationship to valency classes are treated in Chapter 93 (Argument Realization: Affix Indexing and Free NP Order).

72.7 Synthesis

The most defensible synchronic label for Hokkaido Ainu person-indexing is person-conditioned mixed morphological alignment. Row by row: 1SG is nominative–accusative (S=A≠O), second and third persons are neutral (S=A=O), and the 1PL.EXCL and fourth-person rows are tripartite (S≠A≠O, with side of attachment realising the S/A distinction) ‹consensus›. A language-level label such as active, ergative, or inverse characterises at most one row accurately: the active label is unverified for the agreement paradigm Shibatani (1988: 210–211); the ergative analysis is a diachronic reconstruction for Proto-Ainu; and the inverse label is Dal Corso's analysis of ‹SA› Sakhalin, explicitly limited by its author to that variety Dal Corso (2025: 187–188, 213).

Above the row-wise pattern, a hierarchical overlay is operative in the fourth-person set and in the eci= portmanteau and 2→1 irregular cells. This overlay encodes referential distinctions — inclusive versus exclusive, honorific versus plain, logophoric versus direct report — that fall outside the binary S/A/O frame. The overlay is grounded in the five-use analysis of the fourth person and the combination data cited above, and is presented as an integrative account of the competing labels rather than a new theoretical proposal ‹speculative›.

Several questions remain open. Whether the positional contrast that grounds the tripartite label (prefix for A, suffix for S) warrants a tripartite classification or is better captured diachronically, as Dal Corso proposes for ‹SA›, is a live disagreement Dal Corso (2025: 207, 213) ‹contested›. The origin of second-person neutrality — whether it reflects an original non-distinction or a lost S/A≠O pattern — is unresolved; Satō (2024, as reported by Dal Corso) argues for the original account, while most earlier work assumes a later merger Dal Corso (2025: 194–195) ‹contested›. The zero-versus-absent debate for third person remains open between Nakagawa and Satō ‹contested›.