Jul 18, 2020
7 mins read
For this classification we used as the main source the classification of tribes provided in Murdock (1969). In Murdock, each tribe is associated to a tribe number y within a chapter x so that its code is x:y. In most cases, the chapter is equivalent to a culture family encompassing a number of tribes. When information is available, tribes pertaining to one chapter are subdivided into several clusters (i.e. sub-families).
For Example
The code of the tribe Akyem is 32:16 which means:
> It belongs to the family Twi, described in chapter 32. > Within chapter 32, it pertains to the Akan Cluster, which encompasses the tribes from 32:16 to 32:23.
As a first step, we extracted all the tribes from the index of Murdock with their codes. Thus, we created a data set associating each tribe’s name with at least one chapter and tribe number (i.e. x:y). In some cases, when information about cluster (i.e. sub-family) was available, we additionally add the name for the specific cluster. When the name of the tribe was the same name as the chapter or cluster we indicated.
In most cases, chapters in Murdock’s book refers to a cultural family. However, due to lack of information, some chapters define regions, historical cultures, or simply describe an uninhabited areas. For pointing this out in our dataset, we create a categorical variable that indicates when a chapter is: (i)Family , (ii) Family/Region, Region,Historical, or Uninhabited. A chapter is of the category family if the book leaves no doubt that the tribes in it are described in the same chapter because of their cultural affinity. The next two categories (i.e. family/region and region ) are chosen if the book does not make entirely clear whether it uses cultural or simply geographic criteria to group the tribes. The category historical underlines when a chapter simply describes a historical cultural area which no longer exists and uninhabited is only relevant for those areas which Murdock assigns as uninhabited.
Once the Murdock’s index data set was built, we proceed to merge each tribe from GREG (2010) and Murdock (1956)’s map with a specific tribe code from the index. As expected, all tribes contained at Murdock’s map were easily found in the index data set. In contrast, to match each tribe in GREG’s map to one of tribes or families which exist in Murdock’s index, we needed a more comprehensive approach. In fact, a number of adjustments to the raw data extracted from Murdock were needed to account for different spellings of tribe names and also tribes which were forgotten, misspell or associated with the wrong code in Murdock’s index. As a result, we were able to assign to each tribe from GREG (2010) an equivalent tribe from the Murdock’s index.
Matching GREG tribe to Murdock’s index
We take the main body of text from Murdock as our main reference as it contains a lot of details for each tribe making errors in spelling or association with the chapter much less likely in the text body than in the index which just includes the codification x:y and thus makes for easy typos.
Differences in spelling of tribes’ names often resulted from the use of the singular in one source (mostly Murdock) for tribes such as Konjo or Mbwela and the plural in the other source, Bakonjo and Wambuela, respectively. Ba and Wa are both plural forms in a number of African languages, most importantly in Swahili.
We needed to account for tribes (such as Afrikaners, French, or Indians) which did not show up in Murdock but are relevant in some geographic areas for the time assessed in GREG. In this cases, we combine assign to the most com
At this point, we have for tribe in the map from Murdok and GREG a correspond tribe code (i.e. x:y) with its respective information on chapter, sub-family, among others. Based on this information we proceed to make the classification for each intersection. We classified the intersections as containing similar or different tribes according to the following rule.
To take into account differences in definitions between Murdock and GREG, we made some further adaptations. Firstly, we recognized that GREG understands Algerian Arabs, Tunisian Arabs, Moroccan Arabs, Libyan Arabs, and Arabs of UAR (Egyptians) as generalized terms describing all of the arab and arabized populations of these countries. Thus, wherever an \emph{Arab tribe} or a tribe of Arabized Berbers is replaced by any of the above, we did not consider this a replacement of one population by another but as a reclassification of a specific Arab or Berber tribe as Algerian or similar. Furthermore, if one of the tribes in an intersection, which would have been classified as unknown, was either Arab, Fulbe, or Tuareg, we classified the couple as being different. This appears reasonable because Murdock makes it clear that all three of these are different from the peoples described in the chapters of categories other than Family.
Sharing is caring!