New perspectives on grammatical relations, 213-256. [Typological Studies in Language 123]. Witzlack-Makarevich, Alena and Balthasar Bickel (eds). Amsterdam/Philadelphia: Benjamins. 1 Several research institutions and programmes have made possible the research on which this article is based: My home institution SeDyL (CNRS-INALCO-IRD); the programme Investissements d'Avenir overseen by the French National Research Agency, ANR-10-LABX-0083 (Labex EFL/GD1); the ANR-funded project Cortypo (ANR-12-BSH2-0011). The data were collected during the Movima documentation project (DOBES, Volkswagen Foundation, Az. II-81914/54349) and the project Referential Hierarchies in Morphosyntax of the EuroBABEL/EuroCores programme (Deutsche Forschungsgemeinschaft HA-5910/1-1). The specific annotation of three-participant events in the Movima corpus, which shed new light on the potential argument status of an oblique-marked RP, was carried out thanks to the DOBES project The expression of three-participant events in a cross-linguistic perspective (Volkswagen Foundation, Az. II-86740). The colloquium of Area B of the CRC 1252 Prominence in Language provided insights into the role of the agency in argument encoding. I wish to thank the Movima speakers who provided the data presented here for their willingness to share their knowledge with me. The editors of this volume are thanked for their detailed, helpful, and sometimes challenging remarks on a previous version of this paper. Needless to say, I alone am responsible for all remaining errors and shortcomings.respect to which argument can be deleted (5.2), embedded clauses, whose arguments are not retrieved from the main clause (5.3), and quantifier floating, which seems to show a bias towards S and P, but is not restricted to these relations (5.4). Conclusions are drawn in Section 6.The annotated corpus on which the present study is based consists of approximately 130,000 words (30 hours) collected in the field from approximately 50 speakers between 2001 and 2012. 2
The basic clause and its components 2.1 The direct-inverse systemMovima is a language with a direct-inverse system: Bivalent verbs receive morphological marking -either "direct" or "inverse" -indicating the semantic roles A vs. P/G/T of the nominal arguments. There is neither person-indexing morphology on the verb nor case marking distinguishing the two arguments of transitive clauses. Movima argument encoding is best described in terms of constituency: One argument of the transitive clause is encoded by a constituent internal to the predicate phrase and the other one is encoded by a constituent external to the predicate phrase. The predicate phrase occupies the first position in a pragmatically unmarked (i.e. "basic") clause. A first illustration of transitive clauses and the direct/inverse alternation is given in (1), with square brackets indicating the constituent structure. Here, the internal constituent is a cliticized pronoun and the external constituent is a referential phrase (RP), consisting of a determiner and a con...