Chapter 13Person marking and alignment

13.1 The person-marking paradigm

Table 1 sets out the person markers of the Sakhalin verb. It combines Chiri (1942: §37), Murasaki (1979: 49–50), Satō (1985), Sakaguchi (2024: 29–34) and Dal Corso (2025a: 200–201); where the coasts diverge, the western form is marked W and the eastern E.

First person singular ku- prefixes to intransitive and transitive verbs alike. It is the everyday first person of conversation and of quoted dialogue in Asai Take's tales:

Second person singular e- and plural eci- are invariant for role; Chiri records esi- for 2PL at Tarantomari (Chiri 1942: §37). Sentoku Tarōji (千徳太郎治) writes e= to Piłsudski:

Third person is unmarked in all roles and numbers. With Chiri we take the absence of an affix on a finite verb to be itself interpretable — a zero exponent read as third person wherever no overt marker appears (Chiri 1942: §23, §37); plurality of a third person argument may be signalled by -hci (§13.7).

The CI-series splits by transitivity: -as suffixes to intransitives, ci- prefixes to transitives and the copula. It is productive on the east coast, while in Rayciska it is marginal — Hattori's consultants knew cioka as the speech of elders or of the southern west coast, and Murasaki's Rayciska corpus contains a single possessive token Hattori (1961: 5), (Sakaguchi 2024: 27). In the east-coast folklore the CI-series regularly refers to a single speaker: it is grammatically plural (it selects plural verb stems) but semantically ‘I’ (Satō 1985), (Sakaguchi 2024: 27). We keep the conventional gloss 1PL.EXCL while flagging this mismatch wherever it matters.

The series do not mix freely. Satō (1985) formulated a personal agreement rule: within one clause (pronoun included), markers of the AN-series do not co-occur with markers of the KU- or CI-series, while KU- and CI-forms do co-occur with each other — Dobrotvorsky already records cookay ku-eraman ‘I understand’ with the CI-pronoun and the KU-prefix (Dobrotvorsky 1875: 150), (Sakaguchi 2024: 30). The rule is what justifies sorting the markers into series in the first place, and §13.4 puts it to work.

13.2 The indefinite person

The set -an (S), an- (A), i- (O), with the pronoun anoka ‘I, we’ ~ anokay, is the workhorse of Sakhalin first-person reference. Its attested functions are: a general first person plural ‘we’, with no exclusive–inclusive opposition of the Hokkaidō colloquial kind; an ordinary first person singular ‘I’; an indefinite or unspecified agent; and the agent slot of the impersonal passive construction (Chiri (1942: §37); Sakaguchi (2024: 27–28); Dal Corso (2025a: 196–197)). The honorific second-person use of Hokkaidō dialects is unattested in Sakhalin (Dal Corso 2025a: 196). Chiri observed that Sakhalin uses one and the same paradigm in literary and colloquial registers, essentially the epic paradigm of Hokkaidō (Chiri 1942: §37).

The indefinite is grammatically plural regardless of its reference: where a verb has suppletive number stems, the AN-series selects the plural stem even for a single referent (§13.7). In Sentoku's letters it is the unmarked way of saying both ‘I’ and ‘we’:

In the same letter Sentoku writes an=omante ‘I send (it)’ of himself alone and addresses Piłsudski as an=koro nispa ‘my master’. The indefinite is also the classic narrative first person of folklore: the protagonist of a tale, often a god or hero, is cross-referenced by AN-forms throughout, a logophoric device that keeps the narrator formally distinct from the character whose words the whole tale constitutes (Dal Corso 2025a: 198–199). Piłsudski himself glossed such forms simply as first person singular in his notes (e.g. sapan ‘I went down’, (Piłsudski 1912: 37–39)):

The indefinite dominates Sakhalin first-person text frequency. In the three Sentoku letters we count 23 tokens of an= against 17 of ci=, 19 of e=, and not a single ku=; in Yamabe Yasunosuke's (山辺安之助) book-length autobiography the AN-series carries the whole narration, the KU-series appearing once and the CI-series twice (Sakaguchi 2024: 46). The corpus texts recorded by Dobrotvorsky in 1867–1872 use fourth-person narration as their convention, while Asai Take's twentieth- century tuytah ‘folktale’ are narrated in the third person with the reportative manu and use ku- freely in dialogue — so the choice between ku- and the indefinite is a matter of register and genre, not of grammar. Within quoted speech both occur, subject to the personal agreement rule of §13.1.

13.3 Object marking

Overt object prefixes exist only for first person and the indefinite; second person objects are indexed by the same e-/eci- as subjects, and third person objects are zero. The first person object form divides the island: the west coast has en-, as in Hokkaidō, while the east coast generalises in- for both the KU- and the CI-series (Satō (1985); Sakaguchi (2024: 33–39)). Earlier work treated i- and in- as free variants (Chiri (1942: §37)), but Satō showed from the personal agreement rule that in- belongs with the KU/CI-series and i- with the AN-series (Satō 1985). For the exclusive object, Hattori recorded west-coast un-, cognate with Hokkaidō un-, though even his Rayciska consultants used the CI-series only rarely (Hattori 1961: 16); Dal Corso notes that east-coast in- cannot be derived from it by any regular correspondence, so its origin remains open (Dal Corso 2025a: 205).

Sentoku likewise writes in=konte ‘give me’ in his third letter (Tangiku & Ogihara 2001). The indefinite object i- appears wherever an AN-series referent is acted on, including the narrative first person:

This prefix is formally identical to the antipassive i- (Chapter 14), from which it historically derives (Dal Corso 2025a: 196); synchronically the two are kept apart by the referential (first-person or logophoric) reading of the agreement use. The east coast also has a dedicated first-person passive prefix inci-, used when a definite agent acts on the speaker; it serves both the KU- and the CI-series, and corresponds to enci-/unci- of north-eastern Hokkaidō (Sakaguchi 2024: 35–37):

13.4 First–second person interaction forms

When first and second person meet on one transitive verb, the marking is not the simple sum of an A-prefix and an O-prefix: no Ainu variety allows the expected *ku-e-, *ci-e-, *ku-eci- or *e-en- (Sakaguchi 2024: 32). In Sakhalin the verb has in effect a single prefixal person slot; the object is what gets expressed there, and the subject is either recoverable from context or indexed by a suffix (Dal Corso 2025a: 207–208). Table 2 gives the attested forms; KU- and CI-series behave identically throughout and are collapsed as ‘1’ (Sakaguchi (2024: 53–55); Murasaki (1979: 50); Hattori (1961: 14–16)).

The portmanteau prefixes are best seen in a minimal pair that Dobrotvorsky recorded from nineteenth-century speech — identical word order, the prefix alone carrying the directionality:

The 1→2SG form eci- is segmentally identical to the 2PL marker but is a distinct portmanteau; Sentoku uses it to his single addressee:

The AN-series, by contrast, combines its prefixes compositionally: an-e- for INDEF→2SG and e-i- for 2SG→INDEF. Kindaichi reported precisely this “strict” combinability as the hallmark of Sakhalin (Sakaguchi 2024: 32), though his tables also contain regularised forms (e-in-, eci-in-, an-eci-) for which no textual attestation has been found (Sakaguchi 2024: 38, 51) — a caution against filling paradigms by analogy.

One tale recorded twice on the east coast supplies the same insult in both series, the neatest demonstration of the personal agreement rule: KU-series subject with in-, AN-series subject with e-i-, never crossed (Sakaguchi 2024: 38):

Where the second person is plural, the suffix -(y)an is added to the portmanteau: eci-…-yan for 1→2PL, in-…-an (west coast en-…-yan, Hattori (1961: 16)) for 2PL→1. Murasaki stated the generalisation for Rayciska — -yan appears whenever a first person and a plural second person interact, in either direction (Murasaki 1979: 50) — and Sakaguchi confirms it for the east coast: the suffix marks the plurality of the second person, whatever its role (Sakaguchi 2024: 53–54), an analysis Tangiku also adopts for the contemporary description (Tangiku 2022: 358, in (Bugaeva 2022)). It is surely related to the plural imperative particle yan (Murasaki (1979: 30); Chapter 12).

Piłsudski took inranupan here to stand “instead of” inranu (Piłsudski 1912: 92); the two tokens in the same tale show rather that the suffix tracks the number of the second-person subject (Sakaguchi 2024: 40–41). In the AN-series the same surface string eci-…-(y)an is compositional — object eci- plus the ordinary indefinite subject suffix -an — as shown by tokens co-occurring with the pronoun anokay and prefix an- in Yamabe's narrative (Sakaguchi 2024: 49–51). In the corpus, Dobrotvorsky's conversational texts attest the AN-series 1→2SG combination directly: an=e=nukara ‘I met you’ (Dobrotvorsky 1875), (Sakaguchi 2021).

The empty cells of Table 2, finally, are not all gaps of the same standing. The configurations in which agent and object are the same speech-act participant — a first person acting on a first person, a second on a second — can never be filled. Such a clause is necessarily coreferential, and coreference between the two core slots is not built by stacking an A-prefix on an O-prefix: it is routed through the reflexive prefix yay- (marginally si-; Chapter 14), which withdraws the object from the person paradigm, so that the derived verb carries a single person marker. East-coast yay-reske-an ‘I brought myself up’ shows the reflexive stem taking plain intransitive agreement where reske ‘raise’ takes prefixes (Dal Corso 2024: 73–74), and Dobrotvorsky's conversational texts attest the configuration with the lexicalised yayko-:

The 1→1 (and 2→2) dashes are thus theorems of the grammar, not accidents of the record, and no portmanteau should be reconstructed for them: reflexivisation empties this corner of the person paradigm in head-marking languages quite generally. The cells pairing the AN-series with the KU- or CI-series follow in turn from the personal agreement rule of §13.1, which bars the two series from a single clause; where a genuinely indefinite agent acts on the speaker — the one non-coreferential reading — the east coast resorts to the passive inci- (§13.3). Only what remains once these principles have done their work counts as a documentary gap in the strict sense, and in a closed corpus it must stay one.

13.5 Morphological alignment

Read off the paradigm, the synchronic alignment of Sakhalin Ainu person marking is split by person, and we state it as follows. First person singular is nominative–accusative: one form for S and A (ku-), another for O (en-/in-). The first person plural exclusive and the indefinite distinguish all three roles — -as/ci-/in- and -an/an-/i- — a tripartite arrangement, with the S-form suffixed and the A-form prefixed. Second persons are neutral (one invariant form each), and third person is neutral by zero. This is Sakaguchi's classification of the east-coast system (Sakaguchi 2024: 31), and it holds for the west coast with the lexical substitutions noted above.

Dal Corso (2025a) frames the same facts differently. Treating -as/ci- and -an/an- each as a single marker whose suffixal and prefixal realisations continue one historical item — Tamura's old observation that -an behaves prosodically like the verb an ‘exist’ it descends from, reported and extended at Dal Corso (2025a: 190–191) — he dispenses with the tripartite component and describes the basic system as mixed nominative–neutral: nominative for first person and the indefinite, neutral for second and third. The difference is one of segmentation, not of substance: once the distribution (suffix on intransitives, prefix on transitives, distinct O-forms) is stated, nothing empirical hangs on whether S- and A-forms are one morpheme or two. We use the role-explicit glosses INDEF.S, INDEF.A, INDEF.O precisely so that the description does not prejudge this.

The substantive part of Dal Corso's proposal concerns the interaction forms of §13.4 and their history. He reconstructs Proto Sakhalin Ainu with only ku-, en- and e-, plus a collective verb *asti that grammaticalised into the plural markers and, fused with the person prefixes, yielded ci-, -as and eci- (Dal Corso 2025a: 203–205). A single prefix slot, accent shift and syllable weight then conspired against double prefixation in first–second person contexts, leaving object-only portmanteaux (eci-, en-/in-) with subject syncretism (Dal Corso 2025a: 207–208). Finally, the indefinite — already used for a low-referentiality agent acting on a more specific patient, as in impersonal passives — was recruited to disambiguate subject number. The pivot of this last step lies outside the speech-act domain altogether, in 3→3 clauses whose agent is low in referentiality. There an- already does inverse-like work: the patient outranks the unexpressed agent, and the clause hovers between an indefinite-agent and a passive reading:

Extending this device into the first–second person domain gave the discontinuous eci-…-(y)an, en-…-yan and in-…-an of §13.4: the prefix goes on indexing the object, while the suffix, in origin the mark of a backgrounded agent, comes to spell out the subject whose person the portmanteau leaves syncretic. Dal Corso reads the result as an emerging, non-prototypical inverse — special marking wherever the object outranks the subject in discourse salience — and draws a typological parallel with Algonquian and Athabaskan, where impersonal-agent forms (Meskwaki -ekoː, Potawatomi -ikoː) were likewise regrammaticalised as inverse markers (Dal Corso 2025a: 209–213). He is explicit that the system is not uniformly inverse — 1→2SG marking is hierarchical rather than inverse, and 2→1 forms encode object role transparently — and that most reconstructed stages are unattested (Dal Corso 2025a: 202, 212–213).

As a diachronic account we find this persuasive: it derives the portmanteaux from independently motivated phonology, connects the interaction suffix to the indefinite, and explains why the object is always the overtly indexed argument. Synchronically, however, we do not describe Sakhalin Ainu as inverse-aligned. The decisive datum is Sakaguchi's demonstration that in the KU/CI-series the suffix -(y)an co-varies with the number of the second person in both directions (1→2PL and 2PL→1), while in the AN-series the same string is the ordinary indefinite subject suffix (Sakaguchi 2024: 53–55). A marker whose function is second-person plurality is not an inverse marker. The two readings are nonetheless not flatly incompatible: they approach the same closed corpus from opposite ends, Sakaguchi from the synchronic distribution of the attested cells, Dal Corso from the trajectory that produced them, and the elicitation that would force a choice can no longer be carried out. Where the inverse analysis has genuine synchronic purchase is in the 3→3 indefinite-agent clauses above — a live textual pattern, not a reconstruction — and a reader who follows Dal Corso (2025a) may prefer to gloss such tokens of an- as inverse where we keep INDEF. Our synchronic statement is therefore: core alignment nominative–accusative for first person and the indefinite (with role-split tripartite exponence in the CI- and AN-series), neutral for second and third persons; and, overlaid on it, a closed set of portmanteau prefixes and circumfixes for first–second person interactions in which the object controls the prefix slot — a hierarchical trait whose inverse-like asymmetries are the fossil of the development Dal Corso reconstructs.

13.6 Person marking on nouns

The transitive A-prefixes double as possessor indexes on nouns, which then take the possessive (“belonging”) suffix — ku-matak-ihi ‘my younger sister’ in the following, with the 1→2 portmanteau on the verbs beside it:

Alongside direct prefixation, possession is freely expressed by a relative construction with the verb koro ‘have’: Sentoku's an=koro nispa ‘my master’ and e=koro kotan ‘your village’, the hero's an-koro nay ‘the river I own’ in §13.2, and the compounded ci-koro-cise-he ‘the house I own’ in §13.4. Chiri adds that dative (recipient) persons are built on the object forms with an applicative: in-ko- ‘to me/us’, e-ko- ‘to you’ (Chiri 1942: §37). The morphology and semantics of possessive forms, including the nominal plural -ahcin, are treated in Chapter 7.

13.7 Person and verbal number

Person marking interlocks with verbal number at two points (Chapter 8, Chapter 12). First, the suffix -hci (east-coast realisations also -(a)hci, -si, Dal Corso (2024: 62–63)) marks plurality or collectivity of a core argument. Murasaki listed it among the person affixes as an optional third person plural (Murasaki 1979: 49–51), but since it co-occurs with overt person markers of any person — Sentoku's an=e-hci ‘we eat it’ above, Murasaki's own eci-itak-ahci ‘you all speak’ — we treat it with Dal Corso (2025a: 202) and Sakaguchi (2024) as a number marker, not person agreement. With unmarked third persons it is what signals plurality:

The controller need not be the subject: Piłsudski already noted of an-kuyra-si ‘I stole up on them’ that the plural sign is there “not on account of the subject (‘I’) but because the object (the gods) is in the pl.” (Piłsudski 1912: 40–41). Second, suppletive stem pairs (oman ‘go (sg)’ ~ paye ‘go (pl)’, an ‘exist (sg)’ ~ okay ‘exist (pl)’) agree with the grammatical number of the person series: the CI- and AN-series select plural stems even for a single referent — asip-as, paye-as, makap-an, ariki-an, okay-an in the examples above — while the KU-series takes singular stems (ku-san, e=oman) (Sakaguchi 2024: 27–29). This grammatical plurality is the reason we segment the indefinite and the CI-series as plural categories whatever they refer to.