The [AV] ergative VP
One verbal system, which Franchetto (1990) calls “DeErgative” in Kuikúru, is characterized by a VP which takes as its only internal argument the subject of the transitive verb. This verbal system has only been documented in Kuikúru and Panare, and while the AV verb phrase is the same for both languages, other properties of the system are not.
Most main clauses in Kuikúru belong to the ergative Full Set II system, but within the set of three “interactive moods”, the Intentional, the Hortative and the imperative, there is a person-based split. When A is first or second person and O third person, the verb must occur in what Franchetto has called the DeErgative conjugation.15 The DeErgative is characterized by a verbal proclitic that indicates person of A, the verbal prefix ñ-/]- ‘DErg’, the absence of case-marking on free nouns and the absence of auxiliaries. Due to the nature of the personal hierarchy that conditions use of the De-Ergative in Kuikuro, the De-Ergative system never occurs with a third person A; as a result, we have no examples of the construction with a free (pro)nominal A, but only with the first or second person proclitic on the verb. The O argument can occur freely either before or after the VP.
In Panare the cognate construction is used only when the speaker wishes to place the A/S of a clause in focus, either questioning the identity of A/S in an information question, answering such a question, or in placing emphasis on the identity of A/S (all functions for which subjects are fronted in English). Any person of A or S can occur in this construction. Grammatically, the A proclitic is in alternation with a preverbal A (pro)noun; A can occur in no position in the clause other than immediately preverbally. The verb bears the agent focus prefix n- (cognate to Kuikúru ñ-/]-). The O (pro)noun clearly does not form a phrase with the verb, occurring rigidly in the post-auxiliary position. The optional auxiliary agrees with O. Unlike in Kuikúru, the agent focus prefix n- also occurs with intransitive verbs (at least in elicitation), however the preverbal S (pro)noun appears to be required, as speakers are uncomfortable about reducing S to the proclitic form.
This construction cannot be analyzed as simply a stylistic order variant of some other clause type — the grammar is fundamentally different from that of all the other clause types in both Panare and Kuikuro. This is the only construction in which the transitive subject is indexed with the set of verbal proclitics usually reserved for the absolutive, and in Panare the auxiliary agreement normally reserved for the nominative is claimed by the direct object. In Panare the constituency tests, which in other verbal systems show that the absolutive or the accusative forms a constituent with the following verb, can be applied to this construction to show that A and S form a constituent with the following verb. The configuration of this clause type is thus exactly reversed from that of the other main clause types. Beyond the simple order variation, these configuration facts require explanation.
Like the other innovative verb phrases, this VP comes from a possessed nominalization. As is often the case typologically, the Proto-Carib object nominalization presents a different structure from the action and subject nominalizations. It is formed by adding the prefix *nG- to a subset of the action nominalizations; the possessor then is obligatorily the notional A and the notional O does not occur (cf. English ‘my employee’, where the A possesses the nominalization). The object nominalization did not occur with intransitive verbs (and arguably still does not in any modern Cariban language). Again I illustrate with Makushi examples (Abbott 1991:72). In (18a), the notional A, t-inyo ‘her (own) husband’ precedes and possesses the nominalized verb, which in turn serves as the O of the main verb yanumG ‘pick up’; note that the noun saaro ‘meat’ precedes the possessed nominalization, apparently in a modifying apposition relationship. This same relationship exists between a lexical noun and a modifying derived noun in (18b), but with the linear order reversed: the object nominalization is possessed by the first person possessive proclitic, but it is the noun in apposition to the noun it modifies, pemonkon ‘person’, the O of the main verb.
Obviously, this possessed noun phrase provides an ideal source for the De Ergative VP, right down to the nasal prefix on the verb, as shown in Figure 6. The Proto-Carib object nominalizing prefix *n- occurs on a verb already bearing one of a small subset of the action nominalizing suffixes, and the resultant noun is possessed by the notional A. This possessed noun is then the predicate of a simple equative clause, which takes as its subject the notional O. After reanaly sis, A is internal to the VP, O is external (with all the expected syntactic freedom of any etymological S), and the copular auxiliary (when it occurs) agrees with O. An English parallel would be ‘Ii am his employeei’ > ‘He employs me’.
After reanalysis, A forms a constituent with the V, leaving the O to follow in the slot assigned to it by the auxiliary. In Kuikúru, the construction remains restrict ed to transitive verbs, and becomes the non-ergative part of the person-based split ergative system (cf. the ‘DErg’ gloss in Figure 6). In Panare, this post reanalysis construction becomes the only way to formulate sentences which question the identity of A and answers to such questions; it also occurs in certain additional contexts where the identity of A is in focus (cf. the ‘A-Foc’ gloss in Figure 6). The same prefix (with its preceding possessor cum ergative) then extends to intransitive verbs in Panare, creating a true conflation of A and S.
