the university of chicago the phonology and ... - SIL International

classification listed above, but it has the benefit of being supported by a explicitly stated
methodology and analysis.
C.5 Discussion
The study of how Niger-Congo classification has progressed brings up two
important issues: (1) the completeness of the documentation and availability of data for
supporting claims which are made, and (2) the role and validity of the different
methodologies which are used in making the classifications.
C.5.1 Data and documentation
With respect to data and documentation, I will look at three works to exemplify
the issues involved: Greenberg (1970), Bennett & Sterk (1977), and Bennett (1983a). All
three works provide brief overviews of the type of data used, but fail to provide complete
references as to the source of the data. Bennett & Sterk (1977) state: “It would be
impossible to list all the language sources consulted.” This leaves the reader with no
means within the published literature of checking and verifying the claims made in the
paper. It also means that the reader has no way of assessing the integrity of the data.
Regarding Greenberg, Fodor (1969) states, “There are many controversial, ambiguous or,
to be candid, incorrect data in the material of Greenberg...To avoid misunderstandings it
would have been more fitting to indicate in each case the sources Greenberg relied on.”
Let me illustrate the point. Bennett (1983a: 27) makes passing reference to the
fact that he had access to the dictionary of Banda by Tisserant (1931). I can presume
from my knowledge of published material at the time that this was also the source of
Banda data for Greenberg. Now, though Tisserant’s dictionary is certainly a valuable
resource, it has serious flaws, especially in its completeness regarding dialectal variation
and its accuracy concerning vowel quality and tone. From my own library and field
research on Banda, I have access to more recent data on the language family, both
264