Subject and object marking in Bembe

Abstract

Two notable typological characteristics of the Bantu languages are the phenomena of subject and object marking, the cross-referencing of co-referential arguments via verbal morphology. The cross-linguistic variation with respect to the distributional and interpretational properties of Bantu subject and object markers has led to a dichotomy of their roles between being either agreement morphology or (incorporated) pronouns. Despite an increasing number of explanatory attempts of these phenomena, a unifying formal derivation of this inter- and intra-language variation is yet to be found. This thesis is an attempt towards providing a solution by giving a grammatical description of the under-documented Bantu language Bembe (D54) and presenting a novel analysis that employs an Agree-based approach while still accounting for the pronominal properties of Bembe subject and object markers. Subject and object markers in Bembe cannot co-occur with the arguments they cross-reference, unless the latter are dislocated. In addition, subject marking only occurs with preverbal subjects (A’-position) but not with postverbal subjects (A-position). I demonstrate that these and a number of other facts are explained under the assumption that both subject and object markers in Bembe are pronominal elements rather than agreement morphology. In particular, I treat subject and object markers as defective clitics (φPs), which incorporate into their probe whenever their φ-feature set is a subset of the φ-feature set of the probe. Incorporation is seen as Agree-based phenomenon (Roberts 2010a,b), necessary to satisfy the feature-valuation requirements of heads. Besides capturing the subject and object marking facts in Bembe, the presented analysis is able to give a principled explanation as to (a) the difference in interpretation of preverbal and postverbal elements, (b) the constraint on locative constructions and default agreement inversions to appear only with unaccusative and copular verbs, (c) the ungrammaticality of subject-object reversals, and (d) the variation in subject marking in object relative clauses.