Some restrictions on Sesotho null noun class prefixes

Abstract

Some Sesotho nouns may be used without a noun prefix (a phenomenon we refer to as null noun class prefix). There are several studies that deal with this phenomenon in relation to language acquisition (Kunene, 1979; Suzman, 1980; Connelly, 1984; Tsonope, 1987; Demuth, 1988 and Ziesler & Demuth, 1995). Machobane (2003) deals with this phenomenon in Sesotho, paying attention to the determiner phrase (DP) structure. She argues that in some Bantu languages the noun prefix is the head of DP, while in others a phonologically empty D is the head. This difference leads to parametric variations in DP structures. This article explores phonological, morphological, semantic, syntactic and discourse factors that contribute to the null prefix phenomenon. It shows that phonologically null prefixes are associated with the consonants [1], [s] and [d], which are [+coronal]. The prefixes with the consonants [m] and [b], which have the feature [-coronal], do not allow null prefixes (except in the case of class 14). Morphologically, the nouns that have an agreement which is identical to the noun prefix, allow a null prefix, while the ones that have an agreement prefix which is not identical with the noun prefix, do not (except for class 2). Syntactically, a null prefix is possible where there is agreement in the form of the subject verb agreement, noun modifier or copulative complement. At a discourse level, a null prefix appears when a noun expresses given/old information or is salient in the discourse context. However, none of the suggested explanations can account for a null prefix on its own.