Go back to the LSoRB Program Cariban languages are spoken across the northern part of South America, and are solidly of the "OV type" in the sense of Greenbergian correlations: they uniformly have postpositions, the genitive precedes the head noun, they are dominantly suffixing languages, and auxiliary elements are suffixed to the verb or follow the verb. However, there is considerable variation in allowable order of object, subject, and verb. In Panare (Venezuelan), lexical subject expressions are almost rigidly postverbal, though the order of lexical object expressions can vary quite freely. Though it is not completely clear where they got their information, Derbyshire and Pullum provisionally reported Panare as either OVS, or as a verb-initial language.
Based on both elicited and text material, this paper examines the word order and pronominal argument facts of Panare and concludes that the distribution of lexical expressions is partially sensitive to an aspect split in the language. In particular, VSO is dominant in past-perfective aspects, while OVS and VSO are more evenly split in non-past-perfective aspects. The two aspect types have different derivational histories (Gildea, 1992). The non-past-perfective aspects arise from what were historically nominalized clauses, while the perfective aspects apparently reconstruct back to Proto-Cariban as main clause aspects.
Thus, VSO would at first glance appear to be a historically older main clause pattern IN PANARE, even though OVS is almost certainly historically older as a main clause pattern IN THE CARIBAN FAMILY AS A WHOLE.
This seeming paradox between the language-specific facts of Panare and the general language-family facts of Cariban may be partly resolved by looking at the status of bound person-marking forms on the verb. In particular, preverbal lexical object NPs and prefixal object forms on the verb are both ARGUMENTAL in the sense of Jelinek (1984). Postverbal lexical NPs that refer to the object participant are more likely adjuncts, and not as well integrated into the clausal core. Thus, despite the higher frequency of VSO in past-perfective aspects, the lexical postverbal "O" is probably an innovative order pattern, in which the lexical expression merely cross-references the real object ARGUMENT. This real object argument is consistently indicated by a pre-verbal piece, whether it is a lexical object expression or a bound pronominal prefix.
Thus, the paradox disappears in terms of SYNTACTIC ARAGUMENT PLACEMENT. Nevertheless, thefrequency of VSO in both past-perfective and non-past-perfective aspects does suggest a language in the process of word order change.
Dept. of Linguistics
Go back to the LSoRB Program