Chapter 111The Perfective/Anterior Particle a (a, a…a, aan)

111.1 Form, etymology, and the number alternation

Two auxiliaries of anterior or perfective value form a closely related pair in the Hokkaido Ainu auxiliary system: a ‘sit (SG)’ (singular) and rok ‘sit (PL)’ (plural). Both are grammaticalized forms of the suppletive intransitive verb pair a / rok 'sit (SG/PL)', and they retain the number-sensitivity of that source pair: a appears when the clause subject is singular, rok when it is plural Nakagawa (2024: 307); Yoshikawa (2022); Bugaeva (2012: 493). Ijäs's Lesson 9 catalogues the a / rok suppletive pair as a morphological special form alongside ek / arki 'come' and an / oka 'exist', without connecting the pair directly to the grammaticalized auxiliary Ijäs (2023). The suppletive pair and its role in verbal number are treated in Chapter 58 (Verbal Number: Suppletive Stems and the Pluractional -pa).

Structurally, a and rok belong to the auxiliary-verb frame: the shape is V AUX, with no conjunctive particle between the lexical verb and the auxiliary. Person marking falls on the preceding lexical verb; the auxiliary carries none. This single-indexed structure sets the construction apart from the pair-verb frames (kor an, wa an), where each predicate bears its own person index because each constitutes a finite clause in its own right Nakagawa (2024: 306, 439–443); Bugaeva (2012: 492–495). The overall periphrastic architecture is set out in Chapter 108 (Tense, Aspect, Mood, and Modality: Overview); the existential-aspect pair verbs in Chapter 110 (The Existential-Aspect System: kor an Progressive and wa an Resultative-Perfect).

When the subject is plural, rok replaces a. The plural verb stem and the plural auxiliary track the same subject-number value independently:

111.2 Five competing analyses of a

What a marks is genuinely disputed, and the five positions in circulation represent substantive rather than merely terminological disagreements Nakagawa (2024: 306–307); Bugaeva (2012: 493); Yoshikawa (2022) ‹contested›.

The perfective / past and perfect readings (positions 1 and 2) capture the most frequent translation equivalence. Their limitation surfaces in what Yoshikawa calls "exceptional a": time-invariant constants and self-evident truths on which a appears but a past reading is unmotivated, pointing toward a non-temporal account Yoshikawa (2022).

The actional-perfect restriction (position 3) carries the sharpest distributional claim. Satō argues that a marks a completed action as still affecting the current situation, and negating the action removes that link; hence a cannot scope over a negated predicate. The stative or resultative perfect — 'V-ed (and is still in that state)' — is expressed by wa an instead Satō (2006: 63); Bugaeva (2012: 493). The stative-resultative frame is treated in Chapter 110 (The Existential-Aspect System: kor an Progressive and wa an Resultative-Perfect).

The background-event analysis (position 4), following Yoshikawa 2015, draws on discourse distribution. Clause-final a recurrently precedes a reversal connector — korka 'but/although', sequential wa, or the nominalizer p in an adversative frame — marking the preceding event as contrasted background. Nakagawa takes this as evidence that a is primarily discourse-structural, and any past-tense reading a contextual by-product Nakagawa (2024: 306–307) ‹contested›.

Yoshikawa's 2022 Handbook analysis (position 5) integrates the distributional and discourse observations: a combines perfective aspect with direct evidentiality and witnessed-reminiscence modal value. The "exceptional a" on time-invariant truths is reanalysed as direct evidentiality marking the speaker's certain knowledge. Yoshikawa explicitly acknowledges that the analysis rests heavily on fourth-person narrative data, with few third-person-subject examples, and he marks the evidential dimension as warranting further investigation Yoshikawa (2022) ‹contested›.

Positions 4 and 5 are partly complementary: the background-event account is a synchronic-discourse claim about the typical environments of a, while the direct-evidential account is a semantic one about its grammatical value. An event marked as the speaker's directly witnessed background is precisely the configuration that both accounts predict Nakagawa (2024: 306–307); Yoshikawa (2022) ‹speculative›.

111.3 Scene-shift in narrative

Cutting across the semantic debate, there is broad agreement on a's function as a narrative scene-shift signal. The pattern [event] a p (perfective + nominalizer) or [event] a korka recurrently marks the point at which the scene, topic, or acting participant changes Yoshikawa (2022); Nakagawa (2024: 307). The genre-level distribution of a across narrative, ceremonial, and conversational texts is treated in Chapter 155 (Narrative TAM and Evidential Patterning by Genre); the temporal reference context in Chapter 109 (Tenselessness and the Time-Reference System).

111.4 Iterative reduplication: the ~a ~a / ~rok ~rok pattern

The auxiliary may reduplicate to express an iterative or prolonged-continuity reading — 'went on V-ing, V-ed over and over'. The pattern repeats the verb-plus-auxiliary unit: V a V a (singular) or V rok V rok (plural). With a fourth-person subject, the person index may stand alone in the second occurrence (=an a =an a), abbreviating the repeated verb base Nakagawa (2024: 307–308); Bugaeva (2012: 493).

Bugaeva labels the pattern "emphatic continuative" Bugaeva (2012: 493). Ijäs records an a … a pattern glossed 'verb and verb (prolonged continuity)' but does not connect it to the auxiliary Ijäs (2023); the connection is established by Nakagawa's glossed corpus examples ‹corpus-confirmed›. The iterative and habitual periphery shared with other aspect forms is treated in Chapter 113 (Phasal, Iterative, and Habitual Aspect).

111.5 nisa: the recent-perfect companion

The auxiliary nisa ‘just-finished; recent perfect’ marks an event as just completed — closer to a terminative or hot-news perfect than the general anterior value of a, and rendered in Japanese as '〜てしまった' with a freshness nuance Nakagawa (2024: 312, 436); Bugaeva (2012: 493). Like a, nisa belongs to the auxiliary-verb frame and takes no person marking; unlike a it shows no number alternation. The two forms overlap in signalling a past event but differ in aspectual profile: nisa foregrounds the just-completed endpoint, while a in its primary use foregrounds the contrast with what follows.

111.6 The admirative aan and rok(o)ka

111.7 The a / aan opposition: aspect, evidentiality, and the Classical Japanese parallel

Yoshikawa's most productive contribution is treating a and aan as a paradigmatic pair along two simultaneous parameters Yoshikawa (2018); Yoshikawa (2022):

The structural parallel between a+an and Classical Japanese き+あり > けり provides an independent argument for the [a+an] etymology: both pairs decompose the indirect-past into a direct-past base plus an existential, and both the direct and indirect members share a perfective component Yoshikawa (2018); Yoshikawa (2022). The same direct / indirect past opposition is cross-linguistically attested and is placed in typological perspective by Bugaeva (2012: 493) ‹contested›.

The grammaticalization source a / rok 'sit' fits a cross-linguistic cline in which posture verbs grammaticalize into aspect markers via a resultant-state intermediate (Bybee et al. 1994). The path 'sit (in the position resulting from settling)' > 'be in a completed-state position' > anterior / perfective auxiliary is typologically plausible ‹speculative›. The broader diachronic framing of the Hokkaido TAM system is in Chapter 109 (Tenselessness and the Time-Reference System).

111.8 aan stacked with ruwe ne

A frequent narrative pattern places aan immediately before the inferential evidential ruwe ne: the speaker first signals an after-the-fact discovery (aan), then packages that discovery as established knowledge (ruwe ne). Hirosawa analyses the sequence as a territory-of-information shift: aan places the content outside the speaker's territory (the surprising, indirectly-accessed event), while ruwe ne reclaims it inside as fact the speaker now owns Hirosawa (2026: 90). Both corpus examples in the preceding section — oripak kamuy poho a=ne aan ruwe ne and mokor=an aan ruwe ne — show this stacking. The inferential evidential is treated in Chapter 121 (ruwe ne — the Inferential / Visual-Trace Evidential); the system-level account of evidential stacking and mirativity in Chapter 125 (Mirativity, Evidential Scope, and Grammaticalization to Sentence-Final Particles).