Agreement and coordination in XiTsonga, SeSotho and IsiXhosa: an optimality theoretic perspective

Abstract

This thesis provides a unified Optimality Theoretic analysis of subject-verb agreement with coordinated preverbal subjects in three Southern Bantu languages: Xitsonga (S53), Sesotho (S33), and isiXhosa (S41). This analysis is then used to formulate a typology of agreement resolution strategies and the contexts which trigger them. Although some accounts in the Bantu literature suggest that agreement with coordinate structures is avoided by speakers (e.g. Schadeberg 1992, Voeltz 1971) especially when conjuncts are from different noun classes, I show that there is ample evidence to the contrary, and that the subject marker used is dependent on several factors, including (i) the [-HUMAN] specification on the conjuncts, (ii) whether the conjuncts are singular or plural, (iii) whether or not the conjuncts both carry the same noun class feature, and (iv) the order of the conjuncts. This thesis shows that there are various agreement resolution strategies which can beused: 1) agreement with the [+HUMAN] feature on the conjuncts, 2) agreement with the[-HUMAN] feature on the conjuncts, 3) agreement with the noun class feature on both conjuncts, 4) agreement with the noun class feature on the conjunct closest to the verb, and 5) agreement with the noun class feature on the conjunct furthest from the verb. Not all of these strategies are used by all languages, nor are these strategies interchangeable in the languages which do use them "“ instead, multiple factors conspire to trigger the use of a specific agreement strategy within a specific agreement featural context. I show that these effects can be captured using Optimality Theory (Prince and Smolensky 2004). The analysis makes use of seven constraints: RES#, MAX[+H], MAX[-H], DEP[-H], MAXNC, DEPNC, and AGREECLOSEST. The hierarchical ranking of these constraints not only accounts for the confinement of particular strategies to specific agreement featural contexts within a language, but also accounts for the cross-linguistic differences in the use of these strategies. I end off by examining the typological implications which follow from the OT analysis provided in this thesis.