One of the most well-documented areas of Bantu, a group of African languages forming a subdivision of the Benue-Niger division of the Niger-Congo branch of the Niger-Kordofanian language family, is the moraic status of preconsonantal nasals. In previo ...
One of the most well-documented areas of Bantu, a group of African languages forming a subdivision of the Benue-Niger division of the Niger-Congo branch of the Niger-Kordofanian language family, is the moraic status of preconsonantal nasals. In previous studies, it has been frequently argued that in Bantu prenasalized consonants, derived from a nasal and a consonant, are unitary and so trigger lengthening of the preceeding vowel (Clements 1986). However, some studies (Hubbard 1995a, 1995b; Odden 1996, 2000, Hyman and Ngunga 1997) propose that in fact there are two types of prenasalized consonants: one whose preconsonantal nasal bears a mora in the underlying representation triggering vowel epenthesis and another whose preconsonantal nasal does not do so, and therefore does not trigger vowel lengthening. In spite of this argument, this paper suggests that there is only one underlying moraic nasal of a nasal plus a consonant, a moraic nasal.
Instead of proposing two different types of preconsonantal nasals, it is proposed that, within the framework of Correspondence Theory (McCarthy and Prince 1995, 1999), the different moraic realizations of the nasal in the sequence of the NCs may result from the re-ranking of the universal and violable constraints. The different realizations of NCs in different languages are accounted for by re-ranking the crucial constraints: Identity-μ, Identity-σ, Uniformity, *VV, and *[Cμ. In light of this assumption, this paper proposes three general constraint hierarchies with respect to the prenasalized consonants in Bantu. In the constraint hierarchy in the language where NCs trigger vowel epenthesis since the nasal part of NCs is moraic, Identity-μ, Identity-σ, Uniformity, and *Coda outrank the other constraints. This constraint hierarchy is shown in (24a). In contrast, if a language displays the NCs, which do not trigger vowel lengthening, *VV, *[Cμ, and *Coda should be ranked high; this is illustrated in (24b). If the language displays a syllabic nasal, the constraints Identity-μ, *Coda, Identity-σ, and Uniformity should be ranked over the other constraints. To avoid the occurrence of prenasalized consonants, which are derived from mapping two input segments (i.e., the nasal and the following consonant) into a single correspondent (i.e., the prenasalized consonant, NC) in the output, the string-based constraint Uniformity is proposed. The low ranking of Uniformity ensures the occurrence of the prenasalized consonants in a language.