The Partial Set II system
The Partial Set II System is characterized by absolutive verb proclitics (6a), (6d), postverbal auxiliaries that agree with the nominative for animacy (6a–b), (6d), and no case-marking on any (pro)nominal arguments. Configurationally, the absolutive proclitics are in alternation with a preverbal free absolutive argument (6a–b), (6c–d); the explicit nominative argument shows a fairly rigid post auxiliary position (Gildea 1993). To date, the Partial Set II system is clearly described only in Panare (but cf. Gildea 1998 [ch. 10] for other possible attestations).
Although a preverbal S can alternate with the absolutive proclitic (as in 6c–d), in texts the S is almost exclusively encountered postverbally, and in elicitation it is frequently disallowed preverbally (Gildea 1993).8 The single argument of the intransitive verb appears to have two configurational slots, one as the preverbal argument which is internal to the verb phrase (as in (6c), the same slot occupied by the O nominal or proclitic in transitive examples (6a–b)) and the other as the postverbal nominative subject which controls auxiliary agreement (as in (6d), the same slot occupied by the A nominal in transitive examples (6a–b)). As both appear to entail constituency relations, both the preverbal and postverbal position of S require reconstruction; also in need of explanation are the strong synchronic preference for a proclitic in the preverbal S slot and a (pro)noun in the postverbal S slot.
Etymologically, the Partial Set II system clearly has the same VP nucleus as the Full Set II system: a possessed nominalized verb. However, A and S are treated differently — not only is there no ergative agent phrase in the Partial Set II System, but the copular auxiliary agrees with A and S, forming a nominative pattern. Thus, our source construction must be different from the pleonastic pivot described in Figure 1. In order to assure auxiliary agreement with S/A, we must reconstruct a nominative pivot. To arrive at the appropriate pivot in the source construction, the predicate must either be an equative with a predicate NP coreferential with S/A (Figures 2 and 3) or an adverbial/locative phrase, which can either attribute some property to A/S (Figure 4) or locate A/S in some action (Figure 5, Section 2.2).
In Figure 2, the subject of the superordinate clause is identified as coreferential with the possessed agent nominalization in the predicate, cf. English ‘Shei is a makeri of beer / shei is (the) beer’s makeri.’ After reanalysis, the possessor is O, the possessed nominalization is the ‘Present’ inflected verb,9 and the subject and copula of the superordinate predicate are, respectively, the subject and auxiliary of the new clause.
The resultant clause is morphologically indistinguishable from an agent nominalization in a predicate nominal clause: the linear string of words in (7a–b), with their hierarchical structure, could as easily be given a nominal as a verbal translation.10 The structure is also exactly what one would expect of main clauses: the A is clearly identifiable as a subject, with the O in a well-behaved [OV] verb phrase.
As the Proto-Carib agent nominalization only occurs on transitive verbs, this exact source is not available to intransitive clauses. In Panare, the intransitive half of this inflection comes from a distinct source, the cognate to the action nominalization seen in Figure 1 as the source of the Full Set II system. However, as seen in Figure 3, the source construction in Panare utilizes the S — rather than a pleonastic subject — as the pivot. The action nominalization is possessed by the notional S, forming a constituent that can be reanalyzed as a VP; the superordinate S of the copula is coreferential with the S of the nominalized verb in the predicate, creating a second syntactic slot for the same semantic participant. The result, after reanalysis, is that the S is represented both as an absolutive prefix and as a postverbal noun that controls agreement with the copular auxiliary. An English parallel would be something like ‘Hei is hisi singingi’ = ‘he sings / is singing / will sing.’
Example (8a) (repetition of (6d) above) illustrates both etymological S slots, neither of which has been reanalyzed away, in a simple nonspecific aspect intransitive clause, again morphologically indistinguishable from a possessed action nominalization in a predicate nominal clause. Example (8b) (from (6c) above) shows the parallel [SV] clause (which looks identical to the Full Set II SV clause).
A second means of obtaining an S/A pivot is to place a nominalization into an attributive predicate, cf. Figure 4. The nominalized verb is possessed by its notional absolutive, then this NP is the object of the attributivizing postposition *pe, and the resulting derived attributive phrase is the predicate of the copular clause. The copula agrees with its subject, which is the notional A/S of the nominalized verb. English approximations might be ‘Hei is like [a (potential) maker of beer]i’= ‘he will make beer’and ‘hei is like [hisi (potential) singing]i’ = ‘he will sing’.
After reanalysis, the nominalizer bonds with the postposition to form a single complex future tense marker and the S/O forms a VP with the following V; similarly, the S/A pattern together in controlling auxiliary agreement and in being assigned a rigid syntactic slot after the auxiliary. For illustration, consider the Panare future,-séhpa. In (9a–b) the verbal proclitics indicate person of S and O, respectively, while the postverbal first person pronoun indicates the identity of S and A, respectively. The nominal analysis is no longer available for this form in Panare, as no simple nominalizing reflex remains of the source nominalizer, *-cetG.
This scenario explains not only why the Partial Set II A behaves differently from the Full Set II A, but also why the S partakes in both the absolutive and the nominative behaviors. It is not unexpected that, given a choice of two etymological syntactic slots, the free form would prefer the slot associated with the main clause predication (as subject of the copula) and the obligatory expression of the S inside the VP would be simply the proclitic, now functioning primarily as a verbal agreement prefix.
