Chapter 131Clause Linkage in Ainu: Typology and Inventory

131.1 Finite clauses joined by post-predicate particles

Hokkaido Ainu builds complex sentences by chaining fully finite clauses. Each linked predicate keeps the ordinary verbal morphology of its own clause — person indexing, polarity, plural stem choice, aspectual auxiliaries, and any clause-final evidential or nominalizer — and the link between clauses is carried by a separate post-predicate particle such as wa, hine, akusu, kor, kusu, or yakun. There is no finite/non-finite split in the verb and no converb series: the Ainu verb does not oppose a sentence-final form to a special linking form, so a Japanese rendering in 〜て gives no warrant for a converb analysis Nakagawa (2024: 514). The typological profile of the language states the same fact directly — Ainu has no participles, converbs, or other standard non-finite forms Bugaeva (2022). The connectives are therefore particles that join finite clauses, glossed here as conj and not as verbal suffixes.

The particle is looser than a verbal suffix in prosody as well as in structure. Nakagawa, reporting Tamura, notes that a speaker may close the first clause with full sentence-final intonation, pause, and only then produce the connective and the clause that follows it; on this account the connective belongs to neither of the clauses it joins Nakagawa (2024: 514). Refsing records the same delayed continuation for Shizunai, where a pause may fall between the verb expression and the conjunctionalizer Refsing (1986: 238). This autonomy feeds a genuine disagreement over constituency: Nakagawa declines to posit a clause (節) unit or a main-versus-subordinate division for Ainu, holding that the apparatus cannot be carried over from Japanese or English unchanged, while Refsing and the Handbook describe the same surface domain with terms such as conjunctionalizer, adverbial-clause marker, and subordinate clause Nakagawa (2024: 510, 514); Refsing (1986: 238); Bugaeva (2022) ‹contested›.

Two word-classes do the linking work, and keeping them apart resolves most of the descriptive clutter. The post-predicate conjunctive particle (接続助詞) is bound to the predicate it closes; a clause-initial connective adverb such as orowa(no) ‘and then’ or rapok(ke) ‘meanwhile’ is free and can stack after a conjunctive particle. Refsing makes this the organizing cut of her chapter, between bound conjunctionalizers and free conjunctions; Nakagawa keeps only the particles as a 助詞 class and folds the free connectives into the adverb phrase Refsing (1986: 238); Nakagawa (2024: 516).

Several of the particles are segmentally identical to material from other classes, and the grammar resolves the clash by word-class and position rather than by form. Wa is the sequential linker, the ablative case particle, and a sentence-final emphatic; kor is a linker and the transitive verb ‘have’; kusu is a causal/purpose linker, a free ‘therefore’, and part of the intention auxiliary kusu ne; no is a linker and the adverb-forming suffix -no Nakagawa (2024: 340, 353); Ijäs (2023).

131.2 Inventory and routing

Nakagawa organizes the Saru-based conjunctive particles into seven semantic groups — temporal succession, unfolding event, condition, adversative, concession, cause/purpose, and conjecture/manner — together with a parallel-listing use Nakagawa (2024: 337–356). Refsing's Shizunai grammar reaches a near-identical seven-way taxonomy, with local realignments: Shizunai uses kane for the simultaneity that Saru marks with kor, and konno for an inevitable-consequence conditional where Saru again uses kor Refsing (1986: 238–250). The detailed treatment of each domain belongs to the chapters routed below; this chapter sets out the system and the boundaries between them.

Whole-corpus word-form counts confirm that these items carry the descriptive load, with the homonym caveat kept in view. In the aligned Hokkaido corpus wa (≈50,600), kor (≈29,300), and kusu (≈16,000) are among the commonest words, but those totals fold in the ablative and final-particle wa, the verb kor ‘have’, and the future kusu ne, and so overstate the connective. The cleaner items, where few homonyms intrude, still show the shape of the system ‹corpus-suggested›:

131.3 The coordination–subordination continuum

The inventory spans a continuum rather than a clean coordinate/subordinate dichotomy. At one edge, hine enumerates events in temporal order and takes no stance on how they are related; at the other, the complementizing uses of sekor, kuni, kunak, yak, and kusu place an entire clause under a verb of speech, thought, or command Nakagawa (2024: 338, 470–479, 507–513). Most linkage falls between the poles: the medial clause stays finite and keeps its own argument indexes while the chain is read as one larger episode, the configuration that the coordination–subordination typology of (Foley & Van Valin 1984) treats as cosubordination.

Wa binds more tightly. In Tamura's characterization, as transmitted by Nakagawa, wa gathers two situations into one coherent event; Nakagawa divides its readings into result-state, manner or method, and cause-and-effect Nakagawa (2024: 339–342). Satō gives the same three-way description and supplies the diagnostic that separates Ainu wa from a same-subject converb: the two clauses it joins may have different subjects, which a Japanese 〜て clause cannot Satō (2008: 45–46).

Akusu and awa sit at a third point: a realized first event is followed by a discovered, witnessed, or newly unfolding second one, with the causal connection left open. Nakagawa treats the two as semantically equivalent and conditioned by register and dialect — Saru prose favours akusu, Saru verse and several other varieties favour awa Nakagawa (2024: 345). Satō describes the domain as factual condition and ties awa to oral literature Satō (2008: 47–48).

131.4 Reference tracking without switch-reference morphology

Hokkaido Ainu tracks participants across a chain through the obligatory person indexes on each finite verb and through discourse expectation; it has no same-subject/different-subject converb morphology Bugaeva (2022). The connectives carry only statistical tendencies. Refsing finds that the predicates joined by wa share a subject in about ninety percent of her Shizunai material, yet Satō shows the preference is not a constraint, since wa tolerates different subjects Refsing (1986: 240); Satō (2008: 46). The result is a same-subject-preferring chainer with no dedicated morphology.

The no/wa choice is derived from the first predicate rather than from coreference. Nakagawa and Satō tie no to the aspect and polarity of that predicate: stative predicates and negated action predicates take no, with the high-frequency reading ‘without doing X’ under negation Nakagawa (2024: 340–341); Satō (2008: 45–47). Ijäs offers the learner's heuristic that no marks a same-subject negated clause and wa a different-subject one; that framing is hers, and the primary grammars condition the choice on polarity and aspect, not on subject identity ‹contested› Ijäs (2023).

The nearest approach to a switch-reference device is discourse-structural. The Handbook's descriptive grammar reports that the sequence of a perfective a auxiliary and a connective such as korka, wa, or connective p tends to fall where the scene, topic, or actor of a narrative changes, the perfective marking a situation as completed and cut off from the present Bugaeva (2022) ‹speculative›. The exact author and page of that passage have not been confirmed against the print edition, and the device is a clause-chaining boundary marker, not grammaticalized switch-reference. Whether the sequential connectives hine and akusu encode subject continuity in a switch-reference-like opposition or instead encode discourse (un)expectedness is the central open question of the domain ‹contested›; it is adjudicated against a corpus count of subject continuity in Chapter 133 (The hine/akusu Contrast and Switch-Reference-Like Linkage), and the reference-tracking toolkit as a whole is drawn together in Chapter 139 (Reference Tracking and Argument Continuity).

131.5 Condition, concession, cause, and the mood of the matrix clause

The conditional system is organized by realis status and, distinctively, by the sentence type of the matrix clause. Satō's dedicated study finds that Ainu conditionals are mood-constrained but do not privilege the epistemic mood the way Japanese conditionals do: the language has no conditional restricted to truth-judgement, yet it has conditionals that require an imperative or interrogative matrix Satō (2009: 53–54). Yakun is the broadest member, attested with declarative, interrogative, imperative, and volitional or conjectural matrices, and so it bears most of the translator's load across Japanese レバ, タラ, and ナラ Satō (2009: 51–53); Nakagawa (2024: 347). Yak attaches to law-like or warning premises and clusters with an evaluative pirka ‘good’ or wen ‘bad’; ciki and conditional kor are the mood-bound members, ciki introducing a premise for a command or question and kor the general-causation conditional that elsewhere requires a wh-interrogative matrix Satō (2009: 50–51); Nakagawa (2024: 349–351).

Concession and counter-expectation sit alongside the conditionals. Yakka ‘even if, even though’ states that the matrix holds whatever the event of the marked clause, and in the frames yakka pirka / yakka wen it grants or withholds permission; korka ‘but’, transparently kor ‘when’ plus ka ‘even’, marks counter-expectation Nakagawa (2024: 351–352); Satō (2008: 53). Ijäs banks the same set for the learner, and her listing construction illustrates the concessive cleanly:

Cause and purpose are centred on kusu. Plain kusu gives the purpose of the following event, especially with motion verbs, or its reason; wa kusu and hi kusu are reason-only forms that present the first event as actual Nakagawa (2024: 352–353); Satō (2008: 53–54); Refsing (1986: 251–252). The same segmental material enters the intention auxiliary kusu ne, the copula taking no person index, so the detail chapter separates clause-linking kusu from future and intention syntax Chapter 136 (Causal and Purpose Clauses (kusu)).

131.6 Interfaces with evidentiality, complementation, and reported speech

Conjunctive particles attach outside the clause-final evidential frame. A clause may close on a nominalizer-plus-copula evidential such as ruwe ne or hawe ne and then take a linker, giving attested orders ruwe ne akusu, hawe ne yakun, and hawe ne kusu; the evidential sits above clause-internal aspect but below the connective, so kusu ne ruwe ne is well formed while *ruwe ne kusu ne is not Nakagawa (2024: 393). The general order for a linked clause is thus predicate – evidential – connective. The inferential value is treated in Chapter 121 (ruwe ne — the Inferential / Visual-Trace Evidential) and the system as a whole in Chapter 120 (The Nominalization-plus-Copula Evidential Schema). Satō notes one cross-linguistically marked interaction at this seam: akusu can follow a ruwe ne tail, where the Japanese factual と/たら rejects a preceding のだ Satō (2009: 52–53).

Several connectives are formal nouns pressed into linking service, and the same set runs a grammaticalization cline. A formal noun can be a relative-clause head that fills an argument gap, a complement marker that nominalizes a saturated clause, or a connective that is neither subject nor object of the following clause; Nakagawa distinguishes the three by the argument structure around the marker, and places connective p, pe, and hike ‘but, and’ at the connective end of the cline Nakagawa (2024: 507–516). This links clause combining to nominalization, treated in Chapter 102 (Lexical-Head and Formal-Noun Nominalization (kur, uske, ruwe/hawe/siri/humi)) and Chapter 101 (The Nominalizers -p/-pe and -i/-hi (Participant, Event, Place, Fact)).

Sekor is the bridge into reported speech. In Saru and Chitose it frames a direct quote with a verb of speaking or thinking, names entities in the pattern X sekor a=ye ‘call it X’, and frames cries and mimetic sound, its history transparent in se ‘make a sound, say’ ‘say’ plus kor ‘while’ ‘while’ Nakagawa (2024: 470); Ijäs (2023). It is also the one connective that triggers person reassignment: a third party's quoted first person is expressed by fourth-person forms while the narrator's own quoted speech keeps the first person, the mechanism behind the fourth-person voice of oral narrative, which Bugaeva reanalyses as logophoricity Nakagawa (2024: 470–471); Bugaeva (2008: 40–44). The interaction belongs with Chapter 62 (Architecture of the Personal-Affix System: The Four Persons and the S/A/O Paradigms) and is developed in Chapter 105 (Quotative Complementation and Reported Speech (sekor)).

131.7 Dialect and comparative edges

The inventory is dialectally uneven, with Saru and Chitose the best-described baseline in Nakagawa and Satō and Shizunai a full reference grammar in Refsing. Shizunai realigns two cells: kane marks the simultaneity that Saru assigns to kor, and konno the conditional/consequential function Refsing (1986: 248–250); Nakagawa (2024: 342, 347–348).

Tek is the clearest micro-debate. It is not a connective in Saru or Chitose, where it is an auxiliary, yet in most other dialects it links clauses with an ‘immediately upon X’ sense. Nakagawa's synthesis is geographical: western dialects near Saru carry the short-duration restriction, while eastern descriptions dispute whether short duration belongs to the meaning at all, and one analysis makes its core function parallel listing Nakagawa (2024: 343–344) ‹contested›. A related open question is whether the bare verb-plus-verb sequences of Tokachi, Ishikari, and Shiranuka — where Saru requires an intervening connective — constitute a serial verb construction Nakagawa (2024: 518–521) ‹contested›. Sakhalin supplies a contrast at the edge of the system, where sa kusu covers the continuation and result-state range that Hokkaido splits between kor and wa, rather than marking cause or purpose Nakagawa (2024: 344).

131.8 Open points

The foundational descriptive source for the whole domain, Tamura's Saru study of the connective particles, could be reached only through Nakagawa and Refsing; the classic characterization of wa as fusing two events into one, and the formulation that the connective belongs to neither clause, both rest on that intermediated witness ‹speculative›. The Handbook's claims on negation asymmetry and on the perfective a scene-switch device await verification against the print edition Bugaeva (2022) ‹speculative›. The quantitative base is also thin: the available corpus figures are homonym-mixed word-form counts, and the per-function rates that would settle how often connective wa differs from the ablative, or how far hine and akusu correlate with subject continuity, belong to Chapter 133 (The hine/akusu Contrast and Switch-Reference-Like Linkage) and Chapter 138 (Tail-Head Linkage and Narrative Clause Cohesion) ‹corpus-suggested›.