Man Vol. 16, Number 1, March 1981, pgs. 135-138


The authority of ancestors

C.J. Calhoun's criticism (Man (N.S.) 15, 310-13) of my interpretation of ancestor cults in Africa requires a reply on several points.
First, some matters of ethnography. It is one of Calhoun's strictures that 'Fortes also offers a term for ancestor and a distinct one for elder (yaab and kpeem), translations Kopytoff does not consider' (Calhoun 1980: 312). In a sense, this sums up a central issue in my disagreement with previous critics as well as Calhoun: how cavalier can one be with translations? Fortes does not, in fact, offer the translations Calhoun say he does. What Fortes does at various places is to gloss the English words 'ancestor' and 'lineage head' by Tallensi terms. But to gloss is not to translate.
The Tallensi term kpeem does not, pace Calhoun, mean 'elder'. In certain contexts, Fortes glosses it by the following English meanings: lineage head, head of lineage section, senior male member of the lineage and his age and generational equals, senior, elder (Fortes 1940: 251; 1945: 219, 224). In one instance, Fortes ventures to say that 'grandfather' (ba-kpeem; ba = 'father') is 'literally "older father"' (Fortes 1949: 236). However, Fortes states elsewhere: 'Seniority (kpem[partialdiff]t) is defined by four criteria, singly or in combination according to the situation and the group involved. These are age, generation, social maturity, and status' (1945: 225). The last two criteria take the term out of the lineage sphere and out of the age and kinship spheres. Indeed, we are told that 'a chief is always kpeem irrespective of his age or generation' (1945: 22).
What, then, does kpeem mean? I have no access to a Tallensi dictionary (if one exists). Cognates, however do exist in related languages of the Central Gur group (see Rapp 1966: 188-90 and Alexandre 1953: ii, 206, 213). In the most closely related Gurunsi (so-called Frafra), kpeem means someone who is stronger, more powerful, greater than oneself; kwenne means strong man, a man of power and energy; kpeengo means strength, might, force, authority. In More (Mossi), the radicals kwem and kyem- refer to being solid, hard, larger, more important. In brief, the root term has to do with power, hierarchy, authority (and it can be applied to one whose authority derives from age). Rather than seniority, it denotes superiority by virtue of status, age, maturity,

strength, office or whatever. (This conforms with the semantics in Bantu languages which, as my analysis showed, often draw upon a radical best rendered by the French les grands to indicate what anthropologists call 'ancestors' and 'elders'.) To translate kpeem by 'elder' is to mistranslate it and push it into an entirely different semantic field.
What of the other Tallensi term, yaab, which Calhoun says I ignore and which he claims means 'ancestor'? Once again, the translation is careless. Fortes uses yaab at various times to gloss the following: ancestor, founding ancestor, grandparent and grandparent's sibling, any patrilineal ancestor beyond and including grandfather (Fortes 1945: 54, 79, 201; 1949: 146-7, 236). The semantic range thus includes (in addition to other extensions, probably) both the living and the dead lineage elders above the father's generation. This conforms with the meaning of yab in Gurunsi and yâba in More (Rapp 1966: 230; Alexandre 1953: i, 3II; ii, 458).
There is another Tallensi term relevant here and which Calhoun does not mention. This is ba-. Fortes uses it to gloss the following: father, founding ancestor, forefathers (1945: 201; 1949: 174, 236, 329; see also Rattray 1932: 356). The semantic range thus includes one's father and the forefathers beyond. It is not clear whether it extends laterally from the direct line of ascendants (for example, to father's father's brother) as ba- does in More (Alexandre 1953: ii, 10). Two differences between the two terms emerge (and there must be others): ba- includes the father while yaab stops with the grandfather, and yaab includes women while ba- appears to apply only to males.
Tallensi, then, use terms for lineage elders that disregard the dividing line between those who are alive and those who are dead. In this, they operate very much within the semantic universe of the African societies that I examined in my article (1971: 134-6). This should lay to rest Calhoun's statement that 'Kopytoff's argument is apparently based on the assumption that all Africans are the same, for he continually speaks of 'African ancestor cults' as a unit, while unthinkingly criticising Forte's analysis of the Tallensi on the basis of his own material on the Suku' (Calhoun 1980: 311). The semantic are indeed very much the same across the sub-Saharan continent. As to

my venturing into generalisations, I was indeed extending Fortes's own masterly generalisations about 'African ancestor worship' and my references to the Tallensi were as peripheral as those of Fortes. Calhoun misunderstands, I think, the premisses of my argument. Nowhere, for example, have I said that living elders are not different from dead elders. Indeed, I discussed the implications of this difference for the dominance of formal status in the relationship with dead elders (1971: 138).
Let me briefly restate the argument. In African conception, the 'horizontal' boundary between living and dead kinsmen and particularly between living and dead lineage members is secondary tot he 'vertical' boundary between one's lineage and outsiders, between one's elders (living-and-dead) and the other living-and-dead. This conception - so often commented on in ethnographies - is reflected, I argue, in the semantics of the nomenclatures used. In some cases (though not in all), the term for elders actually spans the living-dead divide (as it does with the Suku and the Tallensi). Even more significantly, a comparative linguistic-semantic analysis shows that the terms for what we call 'living elders' and 'dead ancestors' are often cognate and belong to the same semantic field. African languages seem to draw these terms from the same semantic pool, within which we see evidence of semantic drift as we move from society to society. This is a fact of what might be called pan-African cultural 'macrosemantics' and I suggest - nay, insist - that we should incorporate it into our anthropological theorising rather than cling to the semantics of Western modern nomenclatures in which the living-dead divide has far greater significance than it does in Africa.
What I am saying is, of course, hardly new, and I must confess to certain discomfort in having to restate it in 1980. Before the term 'cultural semantics' was born, Boas was raising a similar issue in 1889, in his 'On alternating sounds, ' with respect to inter-cultural phonetic apperception - namely, that what is a single sound in one language can be heard as two sounds by a listener whose language makes different phonemic 'cuts'. The point is that the ethnographer's job is to understand the semantics of the culture at hand, as it is the linguist's job to understand the phonemics of the language at hand. Only secondarily is the ethnographer concerned with whether English (or French or Chinese) words turn out to convey that culture's concepts - more or less, and by push, pull, and puff. I other words, we try to write a Suku-English or a Tallensi-English dictionary, rather than an English-Suku or an English-Tallensi dictionary.
This is at the heart of the disagreement. Calhoun, for example, approvingly quotes James Brain's critique of my analysis: 'that
Bantu languages have no word for ancestral spirit is patently absurd' (Brain 1973: 126). This is the sort of semantic naivete on which abridged dictionaries thrive. Of course one can find a Bantu word 'for' ancestral spirit, as one can find one 'for' God, cousin, uncle, king, slave, family, and so on. Early explorers had no problems with this and early anthropologists erected many of their conclusions on such commonsensical translations.
But as Evans-Pritchard (1936) showed us long ago, the fact that there is a Zande word 'for' God has little to do with what mbori means in Zande, and to render mbori by the English god results in a subtle and radical misunderstanding of Zande thought. I would claim that this also goes for the English term ancestor, no matter how commonsensical it appears to be. (Nobody, by now, argues that this does not go for English cousin or uncle!). It should be noted that after his dismissal of the problem as absurd, Brain supplies the Luguru word for 'ancestral spirit'. But alas, the word is also revealed, in the course of his discussion, to stand for the following package: spirits in general, burial and sacrificial places, mysterious and awesome places, high places, and even uncommonly curious baboons (Brain 1973: 131-32).
To be sure, behind my argument there lie some fundamental theoretical issues. Are linguistic and cultural categories important? To take an extra example, if one subscribes to the position that the reality is in the social structure and that culture, with its semantic categories, is but a symbolic reflection of that reality, then my concerns are indeed irrelevant. Why waste time on what are in any case imperfect reflections of reality? Ancestors are real, and so are family, descent, slavery and so on. Let us study 'them' and gloss them occasionally with the imperfect cultural nomenclature. On the other hand, if one thinks that the social structure is not determinant but is in some interactive and dialectical relationship with culture, then cultural semantics are worth more time than Calhoun is willing to give them.
Igor Kopytoff
University of Pennsylvania
Alexandre, R.P. 1953. La langue möré. Dakar: IFAN.
Boas, Franz 1889. On alternating sounds. Am. Anthrop.2, 47-53.
Brain, James L. 1973. Ancestors as elders in Africa - further thoughts. Man (N.S.) 43, 122-39.
Calhoun, C.J. 1980. The authority of ancestors: a sociological reconsideration of Forte's Tallensi in response to Fortes's critics. Man (N.S.) 15, 304-19.
Evans-Pritchard, E.E. 1936. Zande theology. Reprinted in Social anthropology and other essays. New York: Free Press.

Fortes, Meyer 1940. The political system of the Tallensi of the Northern Territories of the Gold Coast. In African political systems (eds) M. Fortes & E.E. Evans-Pritchard. London: Oxford Univ. Press.
- 1945. The dynamics of clanship among the Tallensi. London: Oxford Univ. Press.
- 1949. The web of kinship among the Tallensi. London: Oxford Univ. Press.
Kopytoff, Igor 1971. Ancestors as elders in Africa. Africa41, 129-42.
Rapp, Eugen Ludwig 1966. Die Gurenne-Sprachein Nordghana. Leipzig: Verlag Enzyklopädie. (Lehbücher für das Studium der Orientalischen und Afrikanischen Sprachen. Band XI).
Rattray, R.S. 1932. The tribes of the Ashanti hinteland, Vol. 2. Oxford: Clarendon Press.

Professor Kopytoff offers us a variety of observations in what he calls 'pan-African cultural "macrosemantics".' Claiming Boas as a distinguished ancestor, he argues that 'the ethnographer's job is to understand the semantics of the culture at hand, as it is the linguist's job to understand the phonemics of the language at hand'. Though Kopytoff devotes most of his letter to the translation of specific terms, it is, as I think he knows, in this conception of task that we crucially disagree. I hold that ethnographers ought, at least in part, to study society.
It is the profound importance of the sociological side of Forte's great ethnography of the Tallensi which I sought to demonstrate in my recent article. Fortes's critics, I contended, repeatedly fall into a culturalism which obscures their vision of key social issues, and which correspondingly prevents them from understanding the social anthropology they propose to criticise. I have little doubt, in any case, that 'social structure ... is in some interactive and dialectical relationship with culture'. My argument is that semantics cannot decide the sociological issues under consideration.
Kopytoff's self-defence is somewhat disingenuous. In restating his earlier argument, he leaves out most of the substance, and thus most of the specific errors. I shall note here just one part of the crucial difference between ancestors and elders. Kopytoff writes now: 'In African conception, the "horizontal" boundary between living and dead kinsmen and particularly between living and dead lineage members is secondary to the "vertical" boundary between one's lineage and outsiders, between one's elders (living-and-dead) and the other living-and-dead.'
Let us accept this as plausible, though noting that this does not oblige us to accept the alleged African conception. Let us also grant, for purposes of argument, that 'the term for elders' actually spans the living-dead divide.' Now, does this justify Kopytoff's (1971) argument that the category of ancestors can readily be scrapped as ethnocentric, and the social significance of ancestral authority be understood through the category of elders? I think not. Perhaps most obviously, the genealogical specificity of ancestors distinguishes them from the continuum of relative age which defines elders. An ancestor can genealogically define a lineage; an elder as such cannot.
We see the sociological poverty of 'pan-African cultural "macrosemantics"' when we look at empirical cases. He holds the following description of Suku practice to be 'congruent with most ethnographic descriptions of African "ancestor cults",' including, apparently, the Tallensi: In short, to those on the outside, a lineage is represented by the eldest member present. Within the lineage, the lineage is represented to any one member by an older member present, and collectively, by all older members living and dead (1971: 133).
The notion that this is true of Forte's description of the Tallensi is nonsense (see 1945: 31; 1965: 129). Genealogical seniority is usually correlated with age, but need not be. More importantly, perhaps, Kopytoff fail to grasp that Tale lineages are not static units, but evanescent ones with segments or wholes called into action by specific events. The definition of the relevant lineages among the parties to any interaction depends on establishing unity (where possible) in the nearest common patrilineal ancestors, and/or distinction in the most distant separate patrilineal, or in certain cases matrilateral, ancestors. For such reckoning, ancestors must be genealogically specific. Ancestors must be older than descendants, to be sure, but that is not enough; they must also be at least the putative progenitors of those descendants. Because of the importance of descent as such, an older brother is a very different sort of elder from a father.
There are a variety of other instances of Kopytoff's inattention to the sociological dimension of analysis, due perhaps to an assumption that semantic analysis covers the same territory better. In original essay and recent letter alike, he ignores the contrast between matriliny and patriliny, taking the existence of some cognate terms to justify generalisation. Closer to the preceding example, Kopytoff thinks that Fortes (1945: 225) means by 'social maturity' and 'status' Tale phenomena which have nothing to do with age, kinship or lineages!
I shall not belabour the point. Fortes deserves to be read carefully, and kinship

deserves to be studied as a social, as well as cultural, phenomenon.
C.J. Calhoun
University of North Carolina - Chapel Hill
Fortes, M. 1945. The dynamics of clanship Among the Tallensi. London: Oxford Univ. Press.
- 1965. Some reflections on ancestor worship in Africa. In African systems of thought (eds) M. Fortes & G. Dieterlen. London: OXford Univ. Press.
Kopytoff, I. 1971. Ancestors as elders in Africa. Africa41, 129-42.

Go to the main Ancestors in Africa page Go to the main Experience Rich Anthropology page Go to the CSAC Anthropology Pages - online