Post-inflectional derivation in Zulu: Further evidence against the split morphology hypothesis

This article adds to the evidence that the Split Morphology Hypothesis (SMH) (Anderson 1982; Scalise 1984; Perlmutter 1988) is incorrect. Scalise (1984) and Anderson (1992) claim that inflectional morphology must always be peripheral with respect to derivation'. Booij (1994) challenges the SMH with evidence from a range of languages, and concludes that while this may hold for contextual inflection (e.g. agreement and case marking), it does not hold true for inherent inflection (e.g. noun number or verb tense). The isiZulu language contains some clear counter-examples to the SMH, namely quantifiers based on adjectives, presentatives and demonstratives. Each of these word-types is analysed to show that they contradict the SMH. Furthermore, they contradict Booij's prediction that the SMH is more likely to be violated when new words are derived from inherently inflected forms than from contextually inflected ones. The unsustainability of the SMH suggests that lexicalist and non-derivational models of grammar are likely to be correct.