Morphosyntactic co-variation in Bantu: Two case studies

The morphosyntactic properties of Bantu languages, in particular of South-East Bantu languages, are often described as being fairly uniform across different languages. For example, the vast majority of Southern and Eastern Bantu languages have an elaborated noun class system with about 15-20 formal distinctions, complex verb morphology encoding agreement, temporal-aspectual distinctions as well as valency and meaning-affecting morpho-lexical operations, and display basic, or underlying SVO word-order, which can, however, be varied according to pragmatic or information structure considerations. However, more detailed studies have revealed a high degree of morphosyntactic variation between different languages within this overall structural theme. For example, typological differences with respect to the marking of objects have been described, in terms of the morphological expression of object marking (Beaudoin-Lietz et al. 2004) as well as in terms of the order and cooccurrence restrictions on object markers and NP objects (see Marten and Kula to appear). Another aspect of variation which has been discussed in a number of papers is variation in locative inversion constructions in different Bantu languages, which differ according to what type of predicate can participate in locative inversion, and how locative inversion constructions relate to information structure and locative agreement more generally (e.g. Demuth and Mmusi 1997, Marten 2006, Buell 2007). In this paper, we are discussing two aspects of variation in more detail based on a larger comparative study of morphosyntactic variation in Bantu which compares ten Bantu languages with respect to 19 variables or ?parameters? of variation (Marten et al. 2007). In the following section we present a short overview of the design and main findings of this study. Against this background, we then present results related to double object constructions on the one hand, and locative agreement on the other, and show that in our sample implicational relation obtain between different parameters, indicating a single underlying source. In the final section we present some conclusions of the study