Knowledge base · Part XI — Valency, Voice, and Noun Incorporation · this topic as a network
87. Middle, Anticausative, and the (Non-)Passive Question
How does the language encode middle, anticausative, and passive-like meanings?
Intransitive verbs can be derived from transitive verb stems through prefixes including anticausative he- and ho-, antipassive i-, reflexive yay- and si-, and reciprocal u-. (Nakagawa 2022: 17.4)
The ci- middle prefix
How does the ci- prefix derive middle voice, and what is its semantics?
The verbal prefix ci- functions as a middle or object-resultative marker. (Bugaeva & Kobayashi 2022: 15.4) The prefix chi- functions as a middle voice marker on certain transitive motion verbs (e.g., chi-maknaraye 'go back', chi-sanasanke 'step forward', chi-riknapuni 'go up', chi-ranaranke 'go down'), interchangeable with reflexive shi-. (Kindaichi & Chiri 1936) The prefix ci- marks middle voice, expressing an action affecting something from the viewpoint of the affected entity, and the derived verbs are often translated into Japanese as intransitives or passives. (Nakagawa 2024: 189) In Refsing's Shizunai dialect material, neither a- nor ci- expresses the middle voice. (Refsing 1986: 31) The prefix ci- forms middle-voice expressions, as exemplified by ci-maknaraye 'go backward' and ci-riknapuni 'go upward'. (Shibatani 1990) Affixation of ci- produces attributive forms with passive force. (Shibatani 1990) To modify a noun directly, transitive verbs must be converted into the middle (spontaneous) or passive voice with the prefix ci- or a-. (Shibatani 1990: 40)
Person restriction and the ci= / ci- homophony
What person restrictions does the middle ci- prefix have, and how are first-person affected participants expressed?
In Ishikari dialect, enci= and unci= are used in passive-like expressions meaning 'I was / we were (done to)'. (Nakagawa 2024: 172) The forms enci= and unci=, used for fourth-person subject with first-person object as passive expressions in Ishikari and Tokachi dialects, are also attested, though rarely, in Kushiro and Shizunai dialects. (Nakagawa 2024: 173) Intransitive verbs derived with ci- are rarely used in persons other than the third person, and personal prefixes are rarely attached to them. (Nakagawa 2024: 189) In the Chitose dialect, to express 'I was startled/flustered', *ku=cimontapire is ungrammatical according to Nabe Shirasawa, and a=en=montapire must be used instead. (reported in Nakagawa 2024: 189) In Saru and Chitose dialects, middle ci-, like the personal affix ci=, often drops its i before vowels a, u, e, o to become ca, cu, ce, co. (Nakagawa 2024: 190)
Community analysis: ci- as a productive middle
What evidence suggests ci- is a productive middle/anticausative prefix beyond published examples?
In verse, the middle prefix ci- can combine with a causative suffix to form a periphrastic 5-syllable expression whose meaning is identical to the basic root (e.g., ciranaranke from ran). (Nakagawa 2024: 430)
Anticausative semantics and the si- overlap
How does the reflexive si- overlap with anticausative semantics?
The prefix si- indicates a spontaneous action or an event occurring on its own without a conscious agent. (Refsing 1986: 31) The intransitive suffix -ke functions as an anticausative marker, alternating with the causative suffix -V on bound stems. (Bugaeva 2012: 488–489) Direct causatives formed with -V can undergo anticausativization with -ke, which deletes the original causer subject. (Bugaeva 2014) The transitive prefix a- is used as a marker for the spontaneous construction, deriving intransitive verbs from transitive verbs. (Shibatani 1990: 56) Kindaichi (1931) calls the spontaneous construction formed with a- "middle passive". (reported in Shibatani 1990: 56)
Body-region prefixes he- and ho-
How do the body-region prefixes he- and ho- reduce verb valency?
The indefinite-agent construction: a= as a functional passive
How does the indefinite-agent construction with a=/an= function, and does it promote the patient to subject?
The ablative particles orowa and orowano can be used to mark the agent in passive constructions. (Kindaichi & Chiri 1936, Shibatani 1990: 34) In the passive construction, the patient is indexed on the verb by transitive object affixes. (Shibatani 1990: 56, 58) The prefix a- can express middle voice, where the verb has a transitive form but an intransitive meaning (e.g., hawehe a-nu korka netopake somo a-nukar). (Kindaichi & Chiri 1936) The passive voice in Ainu is formed by an incorporation construction in which the indefinite person is the agent. (Kindaichi & Chiri 1936) In the language of kamuy yukar, a-un- is used for first person singular and plural in passive inflection. (Kindaichi & Chiri 1936) In the indefinite-person passive-like construction, an overt agent can be introduced marked by or(o) wa 'from'. (Nakagawa 2024: 429) While the passive-like construction with a= allows an overt agent to be marked with or(o) wa, the middle-voice prefix ci- does not permit expressing an agent in this way. (Nakagawa 2024: 430) The indefinite prefix a= cannot form the periphrastic reflexive-causative construction found with ci-; in oral literature, a=ranaranke means the entity designated by a= is necessarily distinct from the entity that descends. (Nakagawa 2024: 430) Orwa may be used to mark the agent in a passive construction. (Refsing 1986) The indefinite person prefix can be used to create a passive construction. (Refsing 1986) Passive clauses in Ainu are formed as indefinite-person clauses using the prefix a-, with the agent marked by the postpositional phrase oro wa (or or wa) 'from the place of'. (Satō 2008: 205) When a personal affix is attached to mark the patient in a passive clause, an object personal affix such as en- 'me' is used rather than a subject affix. (Satō 2008: 206) In the passive construction, the transitive version of the personal affix is invariably used. (Shibatani 1990: 56) Ainu passive clauses typically lack an overt surface agentive nominal. (Shibatani 1990: 58) When an agent is overtly encoded in a passive clause, it is marked as an oblique ablative by the postposition or(o)wa 'from'. (Shibatani 1990: 58) The passive transitive subject prefix on the verb does not agree in person and number with an overt agentive nominal. (Shibatani 1990: 58) In passive clauses, the patient nominal typically appears at the beginning of the sentence or before the ablative agent nominal, rather than immediately before the verb. (Shibatani 1990: 58) Plural verb forms can be used honorifically in passive clauses to express respect toward the referent of the patient nominal. (Shibatani 1990: 58) Passive-like expressions in Ainu are formed by attaching 'oro-wa after the word denoting the actual agent and marking the subject of the transitive verb with the indefinite person, while the undergoer of the action stands in the objective case. (Tamura 1972: 34)
Does Ainu have a true passive?
Does the language have a subject-promoting passive, and if not, why?
Ainu lacks a true passive construction in which the patient or undergoer of the action is promoted to subject position. (Nakagawa 2024: 428, Tamura 1972: 34) The Ainu passive construction represents a historical syntactic change from a transitive indefinite-person construction to an intransitive passive construction. (Shibatani 1990: 58, 60) The ability to express the demoted Actor of an impersonal passive using an oblique allative phrase may be due to the influence of the Japanese passive. (proposed in Bugaeva 2012) The adverb 'anno marks the passive ('be -ed'). (Murasaki 2025: 106) In an Ainu passive clause, the patient remains syntactically an object rather than a subject, as shown by the requirement of nominalization with hi (or i) plus ki rusuy when expressing a desire with rusuy. (Satō 2008: 205) The Ainu passive has diachronically developed from the indefinite-person construction (or ultimately the first-person plural inclusive transitive construction), but has severed its syntactic and semantic connection with it. (Shibatani 1990: 60) In Classical Ainu, the passive prefix a- cannot be analyzed as an indefinite-person prefix because the passive agent is not indefinite or arbitrary, being typically clear from the context or an overt orowa-phrase. (Shibatani 1990: 60)