Compounding and tonal non-transfer in Bantu languages 1
In this paper I discuss a source of reduplicative (non-)correspondence common to Bantu tone languages which has not received attention in the formal literature. In many Bantu languages, the entire RED+Base complex forms a single domain for realisation of the Base stem tone, leading to the occurrence of non-corresponding marked tone in both the RED and the Base. I propose that this pattern of tone realisation in reduplicated forms falls out if the reduplicative complex is a compound verb stem, as compounds are frequently treated as single domains for the realisation of prosodic properties. The length condition on tonal transfer in reduplication which has proven a challenge for previous work on this topic (Myers & Carlton 1996, Hyman & Mtenje 1999) is also shown to follow from the compound structure of the reduplicative complex. This analysis confirms that the morphological status of the reduplicative complex plays an important role in determining the realisation of reduplicative morphemes.
1 I would like to thank the organisers and the audiences at the Typology of African Prosodic Systems conference in Bielefeld, the Journées de Tonologie in Toulouse, and colloquia at Humboldt-Universität, UBC, Potsdam, SOAS and ULCL for the opportunity to present this work as it developed and benefit from their feedback. I am also grateful to the editors, an anonymous associate editor and three anonymous reviewers for comments which greatly improved both the substance and the presentation of this paper. I owe a debt of gratitude to my many language consultants over the years for their help in understanding tone and reduplication in their languages. Any errors of fact and interpretation are, of course, my responsibility.