11 autosegmental study. Nevertheless, the remainder of this section will summarize some of
the available works.
2.1 Tonal inventories
There are a number of published phonologies and grammars of Gur languages that include at least a minimal pre-autosegmental description of tone. Dagaari Kennedy
1966; Bodomo 1997; Antilla Bodomo 2000, Frafra Schaefer 1975, Hanga Hunt 1981, Vagla Crouch 1985, and Dagbani Olawsky 1996, among others, are described
as having two tones and exhibiting downstep. Konkomba Steele and Weed 1966 and Kar Wichser 1994 are described as having three tones but neither of these works
discusses downstep or uses autosegmental theory, so it is possible that the mid tone would be better analyzed as a downstepped high. More thorough non-autosegmental
analyses of three-tone languages include Supyire Carlson 1994, Ditammari Nicole rcher 1983, Moba Russell 1985 and Mbelime Rietkerk 2000; Neukom 2005.
More thorough autosegmental analyses exist in a smaller number of languages. Three such languages are K
nni Cahill 1992, 2007a, Kabiye Roberts 2002, and Samoma
edreb ogo 1997. All three have two underlying tones and exhibit downstep. To my knowledge, there is only one thorough autosegmental analysis of a three-
tone Gur language: Sucite Garber 1987. Sucite is a Senufo language, a sub-group of Gur languages, and is spoken in Burkina Faso. Garber posits four tonal processes: low
raising, subregister high tone deletion, low tone spreading, and high delinking. She uses Clements
’ 1981 model of tone features in her analysis, which has one tone feature
12 dependent on the other. However, there is now a consensus in the literature that each of
the two features needs to be able to spread independently, so this analysis of Sucite would likely be improved by more recent work on tone features such as Snider 1999 or Yip
2002. In particular, the analysis of Mid and Mid-low contours would be much improved. Garber 1987 devotes a lot of space to the analysis of words with these two
tone patterns, and discusses several possible featural specifications in great detail before coming to a final analysis. So many possibilities are discussed that it is difficult for the
reader to discern the best solution. Given the state of tone feature theory at the time when thi di ertatioꜜ wa publi hed, Garber’ aꜜaly i i quite iꜛpre ive. However, giveꜜ
more current theory, the analysis could probably be shortened and clarified. Garber speculates that Sucite may have somewhat recently evolved into a three tone system from
the splitting of L in a two tone system based on several facts. First, there are many more words both nouns and verbs with H tone than with M or L. Secondly, two processes
low raising, subregister high deletion reduce tonal contrasts in certain environments Garber 1981:61-67.
2.2 Grammatical function of tone