20 featuresgeometry and their weaknesses, see Bao 1999:10-44, Yip 2002:52-56, and
Snider 1999:149-160.
3.1 Bao 1999
Bao 1999 is meant primarily as a description of tonal structure in the coꜜtour
toꜜe laꜜguage ” fouꜜd ꜛaiꜜly iꜜ A ia which he coꜜ ider to be fuꜜdaꜛeꜜtally differeꜜt froꜛ the level toꜜe laꜜguage ” of Africa 1999:201. We will return to this discussion at
the end of the section.
Bao 1999 defines the tonal features register r and contour c with binary
articulatory features of the vocal cords: [stiff] and [slack], respectively. I will refer to this as the Register-Contour RC model. He says that these features are functionally
equivaleꜜt to Yip’ 1 8 , 1 8 feature [upper] and [raised], respectively. These two
features can combine to produce up to four level tones and also connect tone to the features governing consonant voicing to explain the influence consonant voicing has been
observed to have on tone, namely that voiced segments lower the pitch of following vowels in some languages. These features are organized into the following geometry, in
which the features [stiff] and [slack] are associated to a register r and contour c node, respectively:
21
4
RC geometry of tone:
However, Yip 2002:58 points out some problems with these features related to the role of the cricothyroid in both vocal cord stiffening and slackening, as well as the connection
between tone and voicing. These issues are not particularly relevant to a phonological analysis of Mbelime tone since the features needed for a phonological analysis can be
discussed and used without specifying their exact articulatory basis and consonants do not have any effect on tone in Mbelime. Therefore, we will leave these questions to other
scholars. The contour node may have a single specification or a sequence of at most two
specifications such that a rising contour tone would have the following configuration:
5
Contour tone:
In the RC model, the contour node is limited to two branches. Bao claims that the concave and convex tones found in some Asian languages are a surface phenomenon.
This claim is based on their sandhi behavior, which we will not go into here. Meanwhile, the register may have only a single specification Bao 1999:3 and the tone-bearing unit
22 TBU is a syllable rime Bao 1999:7, which does not allow contours that span both
registers e.g. a 51 contour using Chinese tone numbers. Bao disregards this problem because of the rarity of 51 or 15 contours 1999:24 in Chinese languages.
Bao uses the following symbols for his tone features: H register = [+stiff]
L = [-stiff] h contour = [-slack]
l = [+slack] Bao claims that there are two fundamental differences between contour systems
and level systems of tone 1999:201: 1 contour systems have at least one [phonemic] tone with a branching structure at the c node while level systems have no branching c
nodes, and 2 contour systems have a one-to- oꜜe relatioꜜ betweeꜜ T U’ aꜜd toꜜe
while level systems have a one- to ꜛaꜜy a ociatioꜜ betweeꜜ toꜜe aꜜd T U’ which
results in contours only as a consequence of rule application. While there are obvious typological differences between these two types of languages, it seems that Bao grossly
oversimplifies the properties of level systems and unnecessarily differentiates the structure of the two systems of tone. For example, the two paragraphs he devotes to
dowꜜ tep propo e a third feature uper-regi ter” to accouꜜt for dowꜜ tep, while Sꜜider 1999, among others, adequately analyzes downstep using only two features. One cannot
really fault Bao for these shortcomings, since this book is intended only to cover contour tone systems. Bao 1999 represents an impressive analysis of tone structure in Chinese
languages, but it is not particularly helpful for analysis of African languages since it
23 contains virtually no discussion of phenomena like downstep that are rare if not non-
existent in Chinese languages. However, the model used in Yip 2002 depends heavily oꜜ ao’ ꜛodel, o uꜜder taꜜdiꜜg it i till helpful iꜜ gaiꜜiꜜg a better gra p of Yip’