1It is a privilege to have The owners of kinship reviewed with such passion. Despite his critical opinion of the book, Allard has obviously been taken by it: he dissects it, wrestles with it, at times admires it, but ultimately cannot endorse it. The first part of his review is largely positive, praising what he understands to be the book’s fascinating ethnographic contributions. But, in a surprising turn, the second part reviews an entirely different book, one that dismisses ethnography in favour of a naïve linguistics. This is no trivial methodological transgression, and I thank the editors of the Journal de la Société des américanistes for affording me the opportunity to reply.

What words do in an ethnography

3It is true that my ethnography includes an investigation of the Kanamari words which circumscribe and traverse the kinship processes. Allard dismisses these as being based on “dubious etymologies,” but there is only a single timid (and ultimately abandoned) attempt at reconstructing an etymology in my book (p. 76-80). Etymologies point to histories of lexicalization and changing meanings, and concern linguistic relationships that speakers of a language may not explicitly (or even intuitively) be aware of. With the exception of -warah, discussed below, my analyses of the Kanamari language concern the content of conventional compositional expressions that function as lexemes. Such expressions are composed of non-compositional lexical and morphological elements that contribute meaningful information. Any Kanamari man or woman can reflect on how these elements are combined to form standardized expressions that describe social practices, relations and processes (e.g. p. 26-27). Sometimes these explanations can involve translation, but just as often they involve paraphrases internal to the Kanamari language (e.g. p. 149-150). I don’t deal with etymologies, but reveal conscious meanings of words and how these meanings relate to social practices.

4The lexemes I analyse and their semantic ranges map an ethnographic space: they are benchmarks around which I weave my ethnography of how the Kanamari transform predation into kinship via feeding and the ownership relation it generates. No conclusions are ever drawn from analyses of the lexemes. Rather, linguistic analyses invite the reader to follow me in unveiling the practices that occur in the vicinity of a lexical-semantic field. Allard questions their accuracy, but the reader is never asked to believe them on their own. Proof lies in the accompanying ethnography. Lexical-semantic description does not substitute for the ethnography, but draws the reader into Kanamari social praxis by revealing the idiom in which social relations and practices are spoken about.

7Allard surely knows this, which might account for his uneasiness with my reference to Evans-Pritchard’s perceptive but uncontroversial claim that ethnographies depend on our understanding of “a few key words.” After all, he does not want to criticize him for, say, “draw[ing] very hypothetical conclusions from the analysis of linguistic constructions.” So he recommends that we understand Evans-Pritchard’s statement in context. (I suppose it is easier to accommodate context where renowned anthropologists are concerned.) He may have written that, but he was also interested in practice. Well, of course he was. We all are. This did not prevent him from opening Nuer religion with a chapter on the semantics of the Nuer word kwoth, which he translates as “God” (Evans-Pritchard 1956). The first pages of that book compare kwoth with words from Latin, Greek and Hebrew, and consider how kwoth behaves in different grammatical environments, semantic roles and in conjunction with a range of qualifiers. Evans-Pritchard explains how the Nuer associate ‘God’ with other words and how ‘God’ is refracted through experience. This method is repeated throughout Nuer religion for different words in the Nuer language, their usages and implications. Semantic analysis of key words makes ethnography compelling and ethnography galvanizes analyses of language.

(-)Warah does not mean ‘body-owner’

11There are four senses of (-)warah that I discuss in my book:

1. -Warah is a possessive noun phrase (PNP) that describes a salient pattern of relations that are inferred from and construed as immanent to the world. It defines the singular in relation to multiple, autonomy in relation to dependency, the whole in relation to parts, owner in relation to owned, and source in relation to derivation (p. 57). Grammatically, in a construction such as X-warah, -warah can be defined as that which the argument (‘X’) is a part of, is derived from, and/or belongs to (p. 60). It applies to contexts in which something emerges as singular in relation to the many parts that compose it (e.g. trunk in relation to rest of tree, main channel of a river system in relation to its tributaries and headwaters), or as composite but autonomous in relation to dependent elements, which may themselves be composite or simple (e.g. a parent tree in relation to derived species, a major river basin in relation to connected minor river basin). I demonstrate this usage of -warah through examples on p. 60-62, 140-148 (in the latter section in terms of hydrology, always relating sense 1 of -warah to settlement pattern and political structure, as in sense 2 below). I translate this sense of -warah as ‘body-owner.’
2. In some cases, the connotations of -warah are interpreted as emerging from a feeding relation, and the effects of magnification and dependency that feeding generates in the parties involved. In these cases, in which -warah is grammatically identical to (but ethnographically different from) sense 1, I also translate it as ‘body-owner.’ I provide ethnographic accounts of many of the different relations in which ‘body-owner’ and ‘dependent’ are equivalent to, respectively, ‘feeder’ and ‘fed’: owner and pet (p. 30-41, 48-58); shaman and familiar spirit (41-58); mother and child (chapter 3); chiefs and followers (p. 137-168); the Brazilian State and the Kanamari (p. 168-173; Costa 2017b). Such relations are not always interpreted as instances of -warah; they are so interpreted when the speaker wishes to draw attention to the supplementary productive and distributive capacity of one party in relation to the other (p. 28-29), for which I provide examples and case studies (e.g. p. 70-74).7
3. While (1) and (2) are PNP realizations of -warah, the word warah also occurs as a verb (I here write it without the hyphen, indicating it is no longer a bound lexeme; see p. xx). In these cases, it expresses predicative possession, and has the general meaning of ‘to have,’ which I tentatively propose is a denominal verb derived from senses 1 and 2 (p. 75-78). I show that the verb warah is mostly restricted to contexts in which what one has or owns is of non-Indigenous origin (p. 76-80). In brief, one ‘has’ (warah) Western merchandise, whereas one is an X-warah as a consequence of ‘body-owning’ pets, dependent kinspeople, shamanic familiar spirits and so forth.
4. The word warah also occurs as a non-predicative noun, in which case it refers to ‘Western merchandise.’ Very briefly, western merchandise as a bulk concept is called warah because it is posited as a mysterious manifestation of the power of the whites, that which makes them into X-warah (in sense 2) of (specific items of) merchandise (p. 78-80). I describe that both senses 3 and 4 are linked to Kanamari interpretations of and relations with the whites, and provide ample historical and ethnographical analyses of how these senses of warah differ from and relate to senses 1 and 2 (p. 80-96).

The concept of the ‘body-owner’

13Allard really dislikes my translation of X-warah (senses 1 and 2 above) as ‘body-owner.’ Here are some of his reasons why:

1. He is “convinced [that] with the evidence [I] provide it is impossible to claim that: ‘“body,” “owner,” and “chief” are imperfect glosses for what, in the Kanamari language, is one concept.’” This is in part because of the heterogeneous meanings of (-)warah and because some of the instances of -warah or warah that I discuss might be cases of homophony and I was duped.
2. His imaginary pragmatics of -warah (frustratingly without recourse to the evidence I provide, which had convinced him of my error) assures him that the word must be translated according to context.
3. “Rendering some of the semantic range of ‘-warah’ as ‘body’ is problematic because [body] has connotations of individuality and wholeness in European languages, whereas [owner] is essentially a relational concept.”
4. Having claimed that I shouldn’t look for a stable translation to -warah, Allard (somewhat confusingly) then proposes that I “could perhaps have better conveyed [my] point by talking of the ‘trunk-owner’, insofar as ‘trunk’ implicitly always implies a contrast with the branches (of a tree) or limbs (of a body).” According to him, translating -warah as ‘body-owner’ is like translating the English word ‘chair’ into the French président-chaise, since ‘chair’ can sometimes mean ‘a seat of office and authority’ and at other times ‘a seat, especially for one person, with four legs for support, a rest for the back and (often) rests for the arm.’

14Regarding the first critique, Allard relies exclusively on the use of the word ‘concept’ in p. 2 of my book. The word appears there as part of a vignette I use to introduce the reader to the ethnography and to acclimatize her to the sorts of issues I will be addressing. I recount that, in the early months of fieldwork, I tried to obtain from my Kanamari grandfather, Poroya, a statement on the bodies of ancient chiefs, whom I had been told were beautiful and incorruptible. I put together a phrase using the word -warah for ‘chief’ and the word boroh for ‘body.’ It turns out, however, that the latter word actually means ‘corpse.’ Poroya explained to me that the Kanamari word for ‘body’ is -warah, which left me dumbfounded. Perceiving my incomprehension, Poroya told me in Portuguese that “our body is our owner and our chief.” This is a story of how, early on in my research, I learned one of the words that would accompany my ethnography. It is as part of this introductory exchange that I write that “‘body’, ‘owner,’ and ‘chief’ are imperfect glosses for what, in the Kanamari language, is one concept” (p. 2).

15Allard accepts that Poroya’s explanation “is indeed a remarkable fact,” but protests that it “was intended to solve a problem of vocabulary rather than as a comment on the bodily and political order of things.” I do not claim otherwise. The evidence for the non-semantic implications of -warah does not, of course, come from this exchange, which only drew my attention to usages and practices associated with (-)warah. The evidence for its non-semantic implications—its role in the “bodily and political order of things”—is presented throughout the book (specially Chapter 4). Poroya’s explanation nonetheless provides us with a native affirmation of the semantic unity of -warah, which allows us to discard the question of whether we are dealing with homophony, polysemy or whatever.

Fetishizing translation

18In his second objection, Allard insists that the word -warah must have contextual translations. I agree with him:

Portuguese-speaking Kanamari admit their difficulty in translating -warah [senses 1 and 2] in a consistent manner. Instead, they provide a number of contextual glosses that take into account the nature of the argument. These include the Portuguese for “trunk” (tronco), “chief” (chefe, cacique, tuxaua), “big” (grande), and “the place of” (o lugar de, i.e., the origin of something or someone). However, the most common and conceptually most wide-ranging and flexible glosses are “body” (corpo) and “owner” (dono), as Poroya first revealed to me when correcting my clumsy question (see Introduction). In most cases when I asked bilingual Kanamari to translate -warah, I was told that it means either “body” or “owner” or that both translations would be correct. Although contextual glosses are necessary in pragmatic circumstances, I submit that, in the Kanamari language, they make little sense. When it occurs in a noun phrase, -warah always designates an asymmetrical relation that can be synthesized as either the “body of” or “owner of,” but also as “chief of,” “main tributary of,” and so forth. For the sake of brevity, I shall render -warah as “body-owner [of],” a composite of the two most common translations for -warah. (p. 62)

21But since Allard ignores this work and builds his critique on a translation, it is worth asking: why translate native words at all? Translation of foreign words serves a number of purposes in ethnographic writing. It can be a shorthand, as it were, which tries to be as effective as possible in terms of ethnographic yield and literary intelligibility, but which ultimately only provides stand-ins for words the meanings of which are irreducible to a given translation. It would simply be too cumbersome for the reader if I retained Kanamari words throughout. Since meaning furthermore depends on extra-lexical factors, it would also require that the reader keep in mind grammatical relations of the Kanamari language. This would not only demand too much from the reader, it would also be a sign of laziness on my part. But translation has another advantage: it draws the reader into a field of meanings that relates a chosen stand-in to words in the target language or in anthropological jargon. This is why I translated -warah as ‘body-owner’: frequency and consistency of occurrence of ‘body’ and ‘owner’ as translations, and the fact that this allows me to interpret and discuss -warah in light of Amazonian theories of the Amerindian ‘body’ and the ‘owner’, both of which are pertinent to an understanding of -warah, and which are themselves affected by the data that -warah contributes.

What is a body?

22This brings me to Allard’s baffling third objection. I find it hard to see how an Amazonianist—or an anthropologist, for that matter—could object to my use of ‘body’ in the composite ‘body-owner’ on the grounds that ‘body’ connotes individuality and wholeness in European languages. There are so many prejudices involved in this claim that it is difficult to know where to start. For one, it simply isn’t true. In English, ‘body’ derives from an Old English word that denoted not only the trunk of humans and animals, but also more generally the main or principal part of a thing (and hence not ‘wholeness’, but always the mainstay of a ‘whole’). Contemporary ‘body’ can refer to a collectivity or corporation of persons, things or ideas, as much as to an individual (as when Allard uses the expression “growing body of scholarship” in his review). The same is true of the Portuguese word corpo, which derives from the Latin corpus. The idea of a ‘body politic’ dates from at least the middle ages and frames a typically European political philosophy (see Kantorowicz 1957 for a classic discussion). Even when ‘body’ is coterminous with ‘individual’, it denotes a composite set made up of multiple parts and their interrelations (which is what a “whole” is), and hence a singularity composed of relations internal to it. In European languages, ‘body’ is as much a “relational concept” as ‘ownership.’

23Secondly, for anthropology, as for all of the humanities and sciences, the idea that the body precludes its relational constitution, or that it can be isolated from its environment (including other bodies), goes against all available empirical evidence. It would be impossible to discuss here the very many (ancient and venerable, as well as modern and cutting-edge) approaches to the body as dynamic and relational, rather than discrete and individualized. I am afraid that convincing my critic of this would take me well beyond the confines of this reply (see Farquhar and Lock [2007] for an introductory reader).

24Finally, it is surprising to read that my use of the English word ‘body’ must have connotations which are ostensibly typical of “European languages”—as if by writing in a European language I were unable to convey Indigenous Amazonian meanings, a state of affairs that would render anthropology impossible. As Allard must know, a recognition of the centrality of the ‘body’ for native Amazonian people during the 1970s is Amazonian anthropology’s coming-of-age moment, its emancipation from Africanist models of ‘primitive’ society, North American culturalist models of acculturation and the like. Pioneering statements of the meanings and usages of the body in Amazonia include Overing Kaplan (1977) and Seeger, DaMatta, and Viveiros de Castro (1979), but it’s no exaggeration to say that pretty much all subsequent anthropological studies of Amazonian societies have dealt with the body and processes related to it in one way or another. The body is the inescapable theme in Amazonian anthropology, whether it emerges as an index of and target for kinship relations (e.g. Gow 1991; McCallum 2001; Overing Kaplan 2003), as the seat of knowledge and capacities (e.g. Seeger 1981; Kensinger 1995), as a focus of controlled or uncontrolled transformation (e.g. Taylor 1996; Vilaça 2005), as a bundle of affects that defines a perspective (e.g. Viveiros de Castro 1996), and so forth. Though Allard may intuitively believe that in European languages the words for ‘body’ do not imply relations, he would have to deny all scholarship on lowland South America to argue the same for concepts of the body in the region. Surely, a positivist, even a “vulgar” one, has to take this evidence into account.

26For the Biá River Katukina, who speak a dialect of the same language as the Kanamari (p. 10-11), the word wara denotes: the living body; a family of spirits linked to plants and animals; the origin of something; and ‘fruit’, i.e. what is derived from a tree (Deturche 2009, p. 296). When the suffix -hi is added to wara, it denotes the ‘possessor of something’ or the ‘owner’ of a ritual (waik-wara-hi, ‘owner of song’) (Deturche 2009, p. 294). For the Kanamari, the owner of rituals, who is usually also a ‘subgroup chief’ (p. 154), is called nohman, though he can also be called by a -warah PNP that takes the name of the relevant genre of songs as its argument (p. 216). In Biá River Katukina, in contrast, nohman or nohmanhi designate the ‘chief’ (which in Kanamari is -warah) or the non-Indigenous ‘boss’ rather than the ritual sponsor (Deturche 2009, p. 111, 159). There is an evident ethnographic field that is common to the Kanamari-Katukina, involving ownership, the body, source-derivation and ascendancy over ritual functions and village life. This field is linguistically mapped by words that are related to the Kanamari -warah, warah and nohman, and to the Biá River wara, -warahi and nohman(hi). Far from being an ethnographic anomaly, this sort of pattern and the relations it communicates are common to a number of people in the Juruá-Purus region, speaking languages from at least three different families. Much of Allard’s “surprise” at the implications of my analyses probably stems from his lack of familiarity with the pertinent ethnographies.

27Perhaps I was overly pessimistic about the possibility of carrying out etymological analyses. Itaquaí Kanamari (which I studied) and Biá River Katukina are dialect extremes of Katukina-Kanamari, situated at the western and eastern limits of the region in which the language is spoken. A superficial analysis suggests that the Kanamari -warah synthesizes semantic aspects of the Biá River wara and -warahi, as well as being a lexical compromise between the two. Furthermore, Biá River wara exists in Kanamari as the verb ‘to be born’ (see ‘source-derivation’) (p. 105, n. 4), and -warahi exists in Kanamari in the form -mowarahi, which refers to ‘nameless ancestors’ (-mo is a locative for fixity; on the meaning of -mowarahi see p. 166-167, and Costa 2007, p. 80, 102-13). Furthermore, the Kanamari word warapikom means ‘wild fruit’, but it is rarely used in the Itaquai (where the name of specific fruits takes precedence), though it is very common among the Kanamari living to the east, where it also denotes a ritual complex occurring during the fruiting season (Labiak 2007) (p. 215-216). There may be further clues in the dialect variants spoken in-between the Itaquaí and the Biá. The role of nohman as a further variable may be key here, particularly since the word is also found in some Panoan languages (though all specialists agree it is a borrowing from Katukina languages), where it either designates ancestral spirits or is a vocative term which spirits use to refer to the souls of humans who participate in ritual (Erikson 2004; Matos 2014, p. 151).

Président-chaise and trunks

28Finally, translating -warah as ‘body-owner’ is obviously not the same as translating the English ‘chair’ into the French président-chaise. In the former case we have two predicative nouns, whereas in the latter meaning differs with valency: one must be président of something, but a chaise is not necessarily of anything. Furthermore, the term ‘chair’ obviously has the primary historical and conceptual meaning of ‘seat of a certain type’ and it is only through metonymy (a common process of semantic change) that it has come to refer to a président, i.e. one who sits in the chair, or the ‘chairperson’. The gender-neutral ‘chair(person)’ (for président) is actually quite recent: until the 1980s, ‘chairman’ or ‘chairwoman’ were more common. Similar difficulties apply to the meanings of ‘head’ as ‘part of body’ and as ‘leader’, the latter meaning borrowing from the ascendant place of the former in the structure of the body. The translation of -warah as ‘body-owner’ captures the synergy of the meanings of ‘body’ and ‘owner’ in Amazonia: both imply relations, and neither meaning is anterior (as far as we can know).

29‘Trunk-owner’ presents another problem. ‘Trunk’ is a possible translation of -warah where the salience of the trunk or torso vis-à-vis the rest of the tree, human or animal is referred to (sense 1), but it is an insufficient translation because -warah can also denote a parent tree in relation to trees derived from it, or a person in relation to humans, animals and spirits dependent on him or her (chief, mother, owner of pet, shaman) (p. 60-65). Furthermore, the encompassing level eclipses the encompassed, so that a woman is the -warah of her pet but no one refers to the trunk/torso of that pet as the -warah of its appendages (p. 63-64, n. 5). In other ethnographic contexts, however, the word for ‘chief’ means something like ‘mainstay’, and hence maintains a semantic relation to ‘trunk.’ The idea of the chief as a ‘mainstay of people’ has been brilliantly described by Guerreiro (2015) among the Kalapalo of the Xingu (226). Indeed, the association between an arboreal and a chiefly idiom (a “botanical and sociopolitical relation”) is as common in the Xingu as in the Juruá-Purus, despite their immense ethnographic differences (Maizza 2014; Fausto 2017; Shiratori 2018; Aparício 2019). Although Allard has doubts, it in fact recurs throughout Amazonia (e.g. Rival 1993; Kohn 2013).