The noncausal/causal alternation in Swahili

Haspelmath (1993) investigates types of noncausal/causal alternation and examines 31 verb pairs in 21 languages, including Swahili, a Bantu language spoken in East Africa. He divides the alternation patterns into causative, anticausative, and non-directed alternations. His data suggest that Swahili is not clearly either a causative or anticausative type of language, since 11 out of these 31 pairs exhibit the causative alternation, while 11 other pairs exhibit the anticausative alternation. However, his analysis does not consider spirantization, a diachronic phonological change where consonants became fricatives in front of high vowels or glides, such as k + y > sh. If such diachronic phonological changes are taken into account when analysing noncausal/causal verb pairs in Swahili, 18 of the 31 pairs can be classified as belonging to the causative correspondence type. This demonstrates a clear causative tendency in Swahili, even more so than in other Bantu languages, such as Matengo and Herero. Comrie (2006) concludes that Swahili is an exception to his hypothesis on the markedness of noncausal/causal distribution, which states that the more spontaneous a noncausal verb, the more unmarked the noncausal verb form. In contrast, the less spontaneous a noncausal verb form, the more unmarked the causal verb form. This study, however, finds that Swahili is not an exception to, but rather a confirmation of Comrie’s hypothesis.