In recent years, the Bantu object affix that is commonly known as the object marker (OM) has attracted considerable debate regarding its status in generative grammar. One view takes the OM to be an incorporated pronoun (see for example, Bresnan and Mchombo 1987; Bresnan 1993). Their analysis is based mainly on the contrast between object marking and subject marking. The subject marker (SM), they point out, behaves as an agreement marker while the object marker behaves like a bound pronoun, freeing the noun from word order restrictions, permitting contrastive focus like a bound pronoun, and permitting non-local anaphoric relations. The other view takes OM to be an agreement marker which licenses a null object (see for example, Bergvall 1987; Kinyalolo 1991; Ngonyani 1996). In this paper I take the second position and, on the basis of Kiswahili constructions in which the lexical object is not realized, I argue that a null object analysis is consistent with VP ellipsis facts, idiom chunks, and co-occurrence between OM and the lexical object. It is consistent with the general analysis of agreement as instantiation of Spec-Head relation (Chomsky 1986a, Kinyalolo 1991). I demonstrate using the elliptical constructions that the verb moves to an Inf-position.

In recent years, the Bantu object affix that is commonly known as the object marker (OM) has attracted considerable debate regarding its status in generative grammar. One view takes the OM to be an incorporated pronoun (see for example, Bresnan and Mchombo 1987; Bresnan 1993). Their analysis is based mainly on the contrast between object marking and subject marking. The subject marker (SM), they point out, behaves as an agreement marker while the object marker behaves like a bound pronoun, freeing the noun from word order restrictions, permitting contrastive focus like a bound pronoun, and permitting non-local anaphoric relations. The other view takes OM to be an agreement marker which licenses a null object (see for example, Bergvall 1987; Kinyalolo 1991; Ngonyani 1996). In this paper I take the second position and, on the basis of Kiswahili constructions in which the lexical object is not realized, I argue that a null object analysis is consistent with VP ellipsis facts, idiom chunks, and co-occurrence between OM and the lexical object. It is consistent with the general analysis of agreement as instantiation of Spec-Head relation (Chomsky 1986a, Kinyalolo 1991). I demonstrate using the elliptical constructions that the verb moves to an Inf-position.