A paper presented at the Annual Meeting of the American Anthropological Association in New Orleans, November 21, 1969
This paper is brief report of Kipsigis attitudes toward surrounding tribal groups, and of some of the social and psychological correlates of these attitudes.
The Kipsigis are the largest of the Nandi-speaking or Kalenjin cluster of tribes in the western highlands of Kenya, members of what used to be called the Southern Nilo-Hamites. Precolonial Kipsigis culture (Peristiany 1939; Orchardson 1961) was characterized by:
- a cyclical age-set system with emphasis on the warrior age-grades recruited by adolescent initiations involving circumcision,
- an economy based on cattle herding and female horticulture of finger milled (eleusine) and numerous minor crops,
- local communities of dispersed homesteads related through manifold intermarriages, and
- family organization marked by a high rate of polygyny, with bride-price paid in cattle; leviratic an woman marriages; and inheritance of the house-property type.
Judging from the data it appears that the broad transformations of Kipsigis economy described by Robert Manners (1967, also Pilgrim ) have not had a major effect on the patterns of ethnocentrism described here (space limitations do not permit a discussion of various control variables such as age, education, and work experience included in the full study).
The major tribal groups bordering the Kipsigis, and the general Kipsigis stereotype of each may be summarized as follows:
- To the north, the Nandi,, differing from the Kipsigis only in very minor ways. Said to be brave, strong, even-tempered and peaceful among themselves, trustworthy, sexually strict, and so forth. To the Kipsigis they are "just the same as we are."
- To the south, the Maasai, whose culture in many ways represents extreme developments of Kipsigis tendencies. Described as fearless in battle, arrogant, aggressive, cruel, manly, direct in their dealings with others, wealthy in cattle and stubborn about modern changes. They are as much to be admired as to be feared.
- To the southwest, the Bantu Gusii, hot-headed and quarrelsome, unruly, noisy, dirty, afraid of the ark. In sum, an unsavory bunch nut dangerous because of their large numbers.
- To the Northwest, the Nilotic Luo, meek, cowardly, strong in the use of magic but pushovers in a direct confrontation, sexually rather loose, poor, well-suited for menial, manual labor. To the Kipsigis the Luo are women, or at best children.
How might we explain these stereotypes? Clearly a tribe's success in the precolonial period depended, in part, on having reasonably reliable expectations of their allies and enemies. This reality aspect of intertribal stereotypes underlies the method of "reputational anthropology" suggested by Robert LeVine (1966). It is also clear, however, that the stereotypes Kipsigis hold of their neighbors are not simply objective observations. In this paper I am concerned with the extent to which Kipsigis ethnocentrism can be viewed as a projective system related to the personality structure of the typical Kipsigis adult male.
Kipsigis consider initiation as the necessary and sufficient experience by which an individual acquires both the social status and the personal attributes of manhood (or womanhood). Thus Kipsigis men not only ascribe to Luo men characteristics which they themselves consider feminine, the Kipsigis explain Luo characteristics in terms of the absence of initiations among that tribe. Similarly, Maasai men are seen as being the epitome of many of the virtues associated with manhood because they practice a particularly elaborate set of initiation rituals. The stereotype of the Gusii is rather ambiguous on the matter of masculinity. A similar ambiguity can be seen in the partial accreditation Kipsigis give to Gusii circumcision. The Kipsigis and Maasai operations involve two stages; the Gusii operation corresponds only to the first of these (removal of the ). Although the Kipsigis do not refer to Gusii as uncircumcised, it is claimed that in the past male Gusii captives were often recircumcised in the Kipsigis manner.
In short, Kipsigis ideas about the psychological effects of initiation are used not only to explain the observed characteristics of other groups but also to deduce what the other peoples must be like.
These stereotypes are not neutral characterizations. They are evaluations, charged with sentiment, in the case of some individuals very strongly so. They frequently crop up in conversations about matters internal to the Kipsigis. The stereotypes of the Maasai, and especially of the Luo, are repeatedly invoked during the initiations. The presence or absence of circumcision is typically one of the first questions raised when an unfamiliar group is mentioned. Kipsigis men in general have a great deal of emotional investment in these stereotypes.
Implicit in this concern with the relative masculinity of members of these other tribes is, I believe, a concern with one';s own masculinity. I will just cite here two typical Kipsigis comments which seem suggestive in this respect:
- "Even if you spear a Maasai, as he dies he will say `it's nothing; I was killed by a woman'."
- "When fighting Luos all you had to do was throw your sword in the air and they would run away."
If the ethnocentric attitudes of the typical Kipsigis relates to a concern with his own masculinity, what are the factors which contribute to this situation? And what might explain the variations in the degree of ethnocentrism among different Kipsigis individuals? The theory of sex identity developed by John Whiting and his associates (Whiting et al. 1958; Burton and Whiting 1961), which was of course first developed in the cross-cultural study of male initiations of this type, seems most relevant. If the interpretations of the cross-cultural findings in terms of individual personality functions is correct, it should be possible to demonstrate that variations in the degree of cross-sex anxiety among Kipsigis individuals are due to variations in father salience in childhood, as modified by initiation experiences during adolescence. This led to the following set of hypotheses, summarized in Table 1:
- Those men whose childhoods were characterized by low father salience will score relatively more feminine on measures of sex identity,
- To the extent that this pattern is modified by initiation, those men who experienced more severe initiations will score relatively more masculine. More specifically, the effect of initiation severity will be strongest for men with low father salience. Variations in the severity of initiation experience should have little or no effect on men with father salient childhoods, since they will already have relatively masculine identities.
- The general pattern of responses relating to the relative masculinity of other tribes will be: Maasai considered very masculine. Luo as very feminine, Gusii as intermediate. Kipsigis (or Nandi) will also be considered very masculine.
- Men with cross-sex identity conflicts will make greater distinctions between these tribes along this dimension of relative masculinity than along other dimensions, and in this respect they will differ from other respondents.
The basic hypotheses concerning ethnocentrism are:
Figure 1 SCHEMATIC REPRESENTATION OF HYPOTHESES
To test these hypotheses a standardized interview was designed, and a potential sample drawn from all married men living in a group of related primary communities a few miles from the trading center of Bomet. Among sets of full or half brothers one was selected at random. Because of time limitations the actual sample interviewed fell short of the original expectations, but did come close to being a full sample of the potential subjects in a contiguous area. Only two men, to my knowledge, refused to be interviewed. In all fifty-six men meeting the criteria completed the interview. The interviews averaged one and a half hours each. All questions were asked in Kipsigis by my assistant while I recorded the answers directly in English.
The primary independent variable of father salience was measured in terms of father absence, defined as the total absence of the father (pater) for a year or more during the subject's first ten years. This measure was scored first in the subsequent analysis.
A measure of the relative severity of initiation was constructed by combining (a) the length of time in seclusion relative to the median length of seclusion for each subject's' sage-set (this was done to control for an overall decline in the length of seclusion following the introduction of schools), and (b) the age of the subject at the time of initiation relative to that of his initiation mates (it is generally agreed that younger boys find the operation easier and are less severely hazed by the young men).
Measures of Sex Identity
Three sex identity measures will be discussed. The first was the presence or absence of male pregnancy symptoms. The use of this measure follows the work of Robert L. and Ruth Munroe on the couvade (Munroe and Munroe n.d.; Munroe et al. n.d.). It is postulated that at the time of a wife's pregnancy a man with unconscious sex identity conflicts will experience psychosomatic symptoms, often analogous to female pregnancy symptoms. Thirty-one subjects reported experiencing one or more of the following symptoms in association with one or more of their wives' pregnancies: fatigue or drowsiness, dizzy spells, nausea or vomiting, headaches, and specific food cravings. Detailed data were also collected on the nature and outcome of each pregnancy of each wife. No general relationship has been found between these data and the occurrence of symptoms reported by male subjects.
Two cross-cultural tests of cognition, developed by John Whiting and Marylou Lionells, were also used to measure relative masculinity (Whiting and Lionells 1967). In the first, subjects were presented with sets of three styrofoam objects, sphere and cubes of various sizes, and asked to indicate which object was different from the other two, The sets were arranged so that the differentiation could be either by size or by shape (for example, a set consisting of a large sphere, a large cube, and a small cube). It was hypothesized that the typical feminine response would be differentiation by shape, and for males by size.
The second test, an adaptation of the Franck test, involved six design completion tasks using pieces of felt. Following Franck's method of interpretation based on body imagery, it was hypothesized that designs completed by females would typically show more internal elaboration of the stimulus, and designs by males more extensions and external elaborations.
Both these tests were also administered independently by Jane F. Martin on samples of women, girls, and boys drawn from the sane Kipsigis community. Both test showed highly significant differences between the sexes in the predicted directions.
A Measure of Ethnocentrism
The construction of a measure of the extent to which the individual subject evaluates other groups along a dimension of relative masculinity went as follows: each subject was asked to rank each of the following tribes: Maasai, Kikuyu, Nandi, Gusii, and Luo, on 24 trait pairs such as cruel.not cruel, masculine/feminine, and clever/foolish. For purposes of analysis the rank scores were treated as if they were raw scores, and a rotated factor analysis was done, with the twenty-four trait pairs as the variables. Three main factors emerged (I will mention just one half of each trait pair):
- Factor 1: hot-headed, arrogant, dangerous, cruel, unruly
- Factor 2: brave, strong in war, masculine, wealthy, weak in magic
- Factor 3: leading in economic development, active, clever, strong, clean
- Five trait pairs did not contribute significantly to any of these three factors.
The interpretation was made that Factor 1 being concerned with the potential for aggressive behavior, represents the reality aspects of the stereotypes most clearly, while Factor 2, being concerned with the personal virtues associated with becoming a man through initiation, most nearly represents the relevant projective aspects of the stereotypes. If this interpretation is correct, it would be expected that the subjects with cross-sex anxiety would tend to polarize the differences between Maasai and Luo along the projective dimension, Factor 2, while the relatively more masculine subjects would differentiate Maasai from Luo more strongly on the reality dimension, Factor 1.
The results of the analysis are summarized in Tables 2 and 3. As a whole the sample described the various tribes as very distinctly different types of people. As expected, the Maasai and Luo were described as polar opposites: Maasai being both aggressive and masculine, Luo as both passive and feminine.
For the sample as a whole, father absence is significantly related to relatively feminine responses on each of the three measures of sex identity, and to differentiating between Maasai and Luo primarily along a dimension of relative masculinity as predicted.
Figure 2: SCHEMATIC REPRESENTATION OF RESULTS
Table 1: SUMMARY OF RESULTS
|M-L on F2
M-L on F1
|No. %||No. %||No. %||No. %||No. %|
|FATHER ABSENT||Easy Initiation||16||14 88%||13 81%||13 81%||11 69%|
|FATHER ABSENT||Hard Initiation||10||6 60%||3 30%||6 60%||5 50%|
|FATHER PRESENT||Easy Initiation||18||7 39%||9 50%||9 50%||6 33%|
|FATHER PRESENT||Hard Initiation||12||4 33%||3 25%||4 33%||4 33%|
|AGE (young to old)||.05||** .36||.04||.20|
|FA ABSENCE (abs.pres)#||** .43||* .25||* .27||* .23|
|SEVERITY OF INITIATION#|
ALL SUBJECTS (easy/hard)
|SEVERITY OF INITIATION|
|FA ABSENT SUBJECTS@||.23||* .43||.14||.11|
|FA PRESENT SUBJECTS@||.00||.18||.10||.00|
# = partial correlation coefficients controlling for effects of AGE
@ = phi values calculted from x²
* = p < .05
** = p < .01 all one-tailed tests
Table 2: INTERRELATION OF INDEPENDENT VARIABLES
|A. Age (young to old)||-|
|B. Father Absence (abs/pres)||* .24||-|
|C. Severity of Initiation (easy/hard)||.21||.03||-|
Table 3: INTERRELATION OF DEPENDENT VARIABLES
|A. Male Preg. Symptoms (yes.no)||-|
|B. Select by Shape over Size||* .26||-|
|C. Design Internal over External||* .27||.04||-|
|D. Discrimination by F1 over F2||.12||.05||-.10||-|
* = p < .05 one-tailed test
Also as predicted, the effect of initiation severity is greater for father-absent subjects than for father-present subjects,and is in the predicted direction. However, for three out of four measures the results are not strong enough to be statistically significant, perhaps because of the small numbers involved. It should be noted, in Table 3, that contrary to expectations the measure of ethnocentrism is not directly related to any of the sex-identity measures.
Whether or not these results support the general interpretation of Kipsigis ethnocentrism described here I leave to you.