Guarani kinship terms with a few twists

Termos de parentesco Guarani com algumas torções

Abstract

This article seeks to organize, analyze, and elucidate longstanding gaps and controversies concerning Guarani kinship terminologies. It begins by cross-examining the earliest vocabulary records from the seventeenth century and Lafone-Quevedo’s translation and analyzing these sources, identifying major discrepancies in the latter’s rendition of the originals. This is followed by a review of current research, in which Guarani populations in Argentina, Brazil, and Paraguay are generally divided into distinct subgroups: Kaiowa, Mbya, and Nhandeva. The comparison of their respective terminologies offers a map of missing kintypes, overarching similarities among the three variants, as well as important divergences in nomenclatures identified as pertaining to the same subgroup. In an attempt to track some of the missing pieces to this kinship puzzle, new data on the Mbya terminology, acquired by combining ethnographic research and computational methods, establishes a complete set of kintypes, including those applied to distant genealogical positions. This terminology reveals generational skews that connect it to the seventeenth-century Guarani vocabulary, recuperating historical ties that were previously untraced, and elucidates divergences in the Kaiowa nomenclature. The article concludes that the Mbya and Kaiowa share the same semantic kinship structure, undeniably of the Iroquois type, challenging the prevailing tripartite divide.

Keywords
Kinship; Terminology; Guarani; Anthropology and computational methods

Resumo

O artigo procura organizar, analisar e elucidar lacunas e controvérsias concernentes às terminologias de parentesco Guarani. Inicia-se com o exame dos primeiros registros de vocabulário do século dezessete, apontando discrepâncias no trabalho realizado por Lafone Quevedo de tradução e análise dessas fontes. Em seguida, apresenta-se uma revisão de pesquisas contemporâneas, em que os Guarani no Brasil, na Argentina e no Paraguai são divididos em subgrupos distintos: Kaiowa, Mbya e Nhandeva. A comparação das respectivas terminologias traça um mapa das posições de parentesco que faltam ser esclarecidas, com semelhanças que conectam as três variantes e importantes divergências nas nomenclaturas de um mesmo subgrupo. Na tentativa de encontrar peças que faltam nesse quebra-cabeça, novos dados da terminologia Mbya, adquiridos por meio de pesquisa etnográfica e métodos computacionais, permitem identificar o conjunto completo de posições de parentesco, inclusive aquelas aplicadas a parentes com conexões genealógicas mais distantes. Essa terminologia inclui torções geracionais que revelam conexões com o vocabulário Guarani do século dezessete, recuperando laços históricos anteriormente ignorados. Essas obliquidades também elucidam divergências na nomenclatura Kaiowa. O artigo conclui que os Mbya e Kaiowa compartilham a mesma estrutura semântica de parentesco, de tipo iroquês, oferecendo, deste ponto de vista, evidência contrária à divisão tripartite dos Guarani.

Keywords
Parentesco; Terminologia; Guarani; Antropologia e métodos computacionais

Guarani kinship terminology has been an object of inquiry for nearly four centuries. Long before the heated dispute between McLennan (1865, 1876 [1876]) and Morgan (1871, 1982 [1877]) inaugurated the anthropological study of kinship, missionaries had already compiled extensive lists of Guarani kinship terms. From the Jesuit’s compilations to contemporary studies, Guarani kinship vocabulary has been a substantial component in debates on kinship in Lowland South America, especially in cross-cultural discussions concerning Tupi-Guarani systems (Birchall et al., 2019; Hornborg, 1988; Laraia, 1986 [1972]; MacDonald, 1965; Wagley & Galvão, 1946, etc.). However, the study of Guarani terminology is a tale of twists and turns with controversies and persistent lacunas that still beg clarification.

In an effort to contribute to this long-standing brew of mystery and dispute this article examines the earliest reports of Guarani vocabulary (Montoya, 1639, 1640, 1876a [1640], 1876b [1640], 2002 [1640]) and shows discrepancies in Lafone Quevedo’s (1919) English translation of these records. Although the author was a pioneer in the systematic study and comparison of South American indigenous languages and undeniably a key link in the dissemination of anthropological research throughout the region (Villar, 2021, pp. 24-25), the article demonstrates that many of the author’s misconceptions concerning marriage practices and descent stem from a flawed analysis of historic sources.

In contrast with colonial texts that refer to the Guarani in general, from the second quarter of the twentieth century on, Guarani-speaking populations in Argentina, Brazil and Paraguay have been classified in three distinct subgroups: Kaiowa, Mbya and Nhandeva. Studies on kinship have followed suit with scarce comparative research to bridge the divide. However, the review of contemporary terminologies presented in this paper reveals that a cross-examination of the three variants is indispensable in gaining a clear understanding of the conceptual framework of Guarani kinship, thus identifying voids and discrepancies that need to be addressed in order to properly sustain or challenge the tripartite classification.

In this direction, the article presents a Mbya terminology, produced by combining ethnographic research and computational methods, that uncovers and explains generational skews widely absent from other reports on Mbya nomenclature. These skews not only unveil significant connections with the vocabulary reported for the seventeenth-century Guarani, but also account for important points of divergence in disputing Kaiowa terminologies. In addition to demonstrating that the Guarani terminologies are compatible with Iroquois-type crossness¹, the results of this research indicate that the Kaiowa and Mbya share the same semantic structure of kinship. Thus, despite the need for further elucidation of the Nhandeva terminology, there is strong evidence that in terms of this fundamental aspect of culture and language the prevailing tripartite division of the Guarani is questionable.

EARLY RECORDS OF GUARANI TERMINOLOGY AND THEIR TWISTED TRANSLATION

Just over a century ago, Lafone Quevedo (1919), published the article “Guarani kinship terms as index of social organization” in ‘American Anthropologist’. His description of Guarani vocabulary, based on works written by the seventeenth-century Jesuit priest Antonio Ruiz de Montoya, was a response to Lowie’s (1915) appeal, launched in an issue of the same journal, in a review dedicated to River’s ‘Kinship and Social Organization’:

But at present I only respond to Dr. R. H. Lowie’s appeal: “In the first place, it is to be hoped that carefully prepared records of kinship nomenclature will begin to come in from all parts of the world. . . . The correlation of these terminologies with concomitant social customs and organization will henceforth become a duty. . . and may lead to a definite knowledge of the geographical distribution of distinctive features in terminology. . . Finally, the intensive consideration of particular systems must bring to light many points of psychological interest, while comparison with the systems of culturally and linguistically related tribes will show what differences in nomenclature persist where differences in social organization are eliminated and linguistic differences minimized”

(Lafone Quevedo, 1919, p. 435)².

In fact, the Guarani terminology collected by Montoya and translated into English by Lafone Quevedo, as well as an earlier essay published in Spanish (Lafone Quevedo, 1917), would not go unnoticed by Lowie (1968 [1929], p. 48), as it offered a first perspective into the social organization of this important South American population, whose kinship system Lowie (1968 [1929], p. 48) described as a classic example of the ‘usual classificatory type’.

Considered to be the first work to correlate the kinship vocabulary and social organization of a Tupi-Guarani people, Lafone Quevedo’s (1919) article later became the target of severe criticism as more information became available concerning contemporary Guarani populations (V. Watson, 1944; J. Watson, 1952) and other Tupi-Guarani groups, such as the Tapirapé (Baldus, 1935; Wagley & Galvão, 1946) and Guajajara (Wagley, 1943). Philipson (1946), for instance, found Lafone Quevedo’s (1919) analysis to be based on inadmissible linguistic speculations devoid of empirical sociological evidence.

As a matter of fact, Lafone Quevedo did not conduct ethnographic research among the Guarani, thus his analysis of Guarani kinship vocabulary in this article relied mainly on Montoya’s records, more specifically Platzmann’s 1872 editions of the “Tesoro de la lengua guaraní” (originally published in 1639) and the “Vocabulario de la lengua guaraní” (originally published in 1640), as well as the first edition of the “Catecismo de la lengua guaraní” (1640), which contains an index of kinship vocabulary for use in the Jesuit missions³. His assumptions about marriage practices and other topics of social organization discuss references concerning other Tupi groups, such as the Tupinamba and Tamoyo, based on Souza’s (1587 [1851]) ‘Roteiro do Brasil’, but mostly derive from etymological speculations that go as far as tracing roots and explanations among Carib, Arawak and even North-American indigenous languages, such as Lakota.

For those willing to look past the author’s linguistic and anthropological misconceptions and merely seeking a faithful reproduction of seventeenth-century Guarani terminology, as reported by Montoya and largely inaccessible until recent advances in the digitalization of historic archives, a review of Lafone Quevedo’s documental sources reveals that the author was also imprecise in this respect. Table 1 below highlights disparities in Lafone Quevedo’s and Montoya’s records⁴. It should be noted that insignificant spelling differences are neglected.

Table 1
Comparison of Montoya’s and Lafone Quevedo’s terminologies.

Position5	Sex of Ego	Montoya	Lafone Quevedo
MZD	M	yetipé6; cĭĭ mêmbĭ cuña; teyndĭ (C, T)	yetipé; çĭĭ mê’mbĭ cuñâ’
FZD	M	yayché mêmbĭ cuña; Yaiché; yetipé; teyndĭ (C, T)	yaichemê’mbĭ cuñâ’; Iaiché; yetipé
MZD	F	cĭĭ mêmbĭ cuña; quĭ pĭĭ (T)	çĭĭ mê’mbĭ cuñâ’
MZS	F	cĭĭ mêmbĭ cuimbaé; quĭbĭ (C, T)	çĭĭ mê’mbĭ cuimbaé
FBD	F	tubĭ rajĭ; quĭ pĭĭ (T)	tubĭ rajĭ
FZD	F	yayché mêmbĭ cuñã; yayché (T); quĭ pĭĭ (T)	yaiche mê’mbĭ cuñâ’
FZS	F	yayché mêmbĭ cuimbaé; quĭbĭ (C, T)	yaiche mê’mbĭ cuimbaé
D	F	mêmbĭ (V); mêmbĭ cuña (T)	-------
BD	F	quĭ pĭi7; pẽng (V)	quĭ pĭi
FZDS	M/F	yaiché mêmbĭ (T)	------
FZDD	M/F	yaiché mêmbĭ mêmbĭ réra; yaiché mêmbĭ (T)	yaiché mê’mbĭ mê’mbĭ réra
MBSS	M	taĭ; tutĭ raĭ raĭréra; tutĭ raĭ raĭré yoaĭré (C)	taĭ; tutĭ raĭ raĭréra
MBSS	F	tutĭ raĭ raĭréra; tutĭ raĭ raĭré yoaĭré (C)	tutĭ raĭ raĭréra
SS/SD/DS/DD	M	temỹmŷnó (T, V); tamỹ (V)	----------

Puzzling as these discrepancies may seem, a closer look at Lafone Quevedo’s analysis of terminology in connection with his assumptions concerning marriage practices and other aspects of social organization reveals that if some of the terms overlooked had been included, they would have clearly contradicted the author’s conclusions concerning marriage and descent. In this respect, one of the main points of divergence is the classification of cousins as siblings. Following Montoya’s texts, in addition to a set of descriptive terms for cousins, such as tubĭ rajĭ (literally: father’s brother’s daughter), we find that the classificatory terms for female ego’s younger sister (quĭ pĭĭ)⁸ and brother (quĭbĭ) also apply to her cousins, and that male ego’s term for sister (teyndĭ) extends to all opposite-sex cousins, though he does not mention that the terms for elder brother (tĭqueĭ) or younger brother (tĭbĭ) extend to male ego’s same-sex cousins. However, this important aspect of Montoya’s terminology was partially neglected by Lafone Quevedo, possibly in favor of the latter’s theories on descent and moiety marriage exchange. For instance, in the following passage Lafone Quevedo (1919, p. 427) acknowledges that both men and women classify FBCh of MBCh as siblings: “That cousins, children of uncles, whether paternal or maternal, may not intermarry, because ‘uncles’ are equivalent to ‘fathers’, and ‘cousins’, their offspring, to brothers and sisters, because [sic] ‘children”.

However, he claims that this does not apply to MZCh and FZCh, whom he classifies as ‘marriageable cousins’, due to a strange interpretation of the fact that, while men distinguish the sex of their offspring using two different terms, women address their offspring of both sexes as mêmbĭ, adding the terms cuña and cuimbaé, to differentiate daughters and sons, respectively. As we see in the following excerpt, he mistakenly interpreted that, since cuña means ‘female’ and cuimbaé means ‘male’, these automatically translated as ‘wife’ and ‘husband’, meaning that the children of aunts would never be classified by ego as ‘sister’ and ‘brother’.

It is very remarkable that in the former case the children of ‘uncles’ were called ‘son’ or ‘daughter’ according to their sex, which of course converted them at once into ‘brother’ or ‘sister’ reciprocally; whereas in the present instance the children of ‘aunts’ become at once ‘woman’s-child male’ and ‘woman’s child female’, that is to say, ‘tribal (possible) husband and wife’, since the words cuimbaé and cuña are, sexually speaking, ‘husband’ and ‘wife’ respectively, or say, ‘male’ and ‘female’; not ‘son’ or ‘daughter’

(Lafone Quevedo, 1919, p. 429).

The author’s firm belief in marriage exchange between two patrilineal moieties reinforces the exclusion of MZCh as siblings to ego, but also, in contradiction to the interpretations above, also rules out MBCh⁹, as he mentions in the next passage:

This confusion between the ideas of ‘father’ and ‘uncle’ must have much to do with an exogamous custom between two nations, in the one, patrilineal, all the men would be ‘fathers’ and ‘uncles’ of all the offspring and their children nationally ‘brothers’ and ‘ sisters’ for they all would be of U-blood; the women as outsiders’ would not be of any blood in particular, hence the rule that while paternal cousins could not intermarry those of maternal origin could, because their blood was of no account

(Lafone Quevedo, 1919, p. 434).

The existence of moieties and marital exchange between them, which we find no trace of in Montoya or other texts relating to the Guarani, are explained in greater detail in Lafone Quevedo’s (1917, p. 7) earlier article:

Each moiety provided the other with women, so that between the two moieties kinsmen were uncles, cousins and nephews, resulting in a curious lexicon by which they knew who they could or couldn’t marry without violating the rules of their social organization: in Guaira¹⁰ the offspring of t-Ub-y uncles (in one moiety) and that of Tut-y uncles (in the other moiety) were all siblings and not cousins within their respective group or moiety [added emphasis]¹¹.

As we see, the author claims that the system was divided in patrilateral and matrilateral kin, respectively represented by the terms attributed to the paternal uncle (tuby) and the maternal uncle (tuty). Within this division, the offspring of paternal uncles and aunts would be classified as ego’s siblings, whereas the children of maternal uncles and aunts would be classified as cousins and, therefore, marriageable. However, the literature clearly demonstrates that the Guarani kinship system is cognatic and that marriage among consanguines is extremely rare. Although V. Watson (1944) and J. Watson (1952) suggest that avuncular marriages may have existed among the Guarani Kaiowa in the past, neither found any ethnographic evidence of this. According to my own analysis of 542 Guarani marriages dating from 1889 to 2016, only 6 (1.1%) represent consanguineal marriages, all of which correspond to marriages between cousins. However, only four marriages corroborate Lafone Quevedo’s model, where a man marries a matrilateral cousin.

In turning our attention to other discrepancies in Lafone Quevedo’s rendering of Montoya’s records, we find that the terms lacking for female ego’s FZD (yaiché) and both male and female ego’s FZDD (yaiché mêmbĭ) result from the author’s failure to apply a skew recorded by Montoya, by which first-degree and second-degree female cousins, related to ego by means of his/her paternal aunt can be classified a generation up. This means that the same term can be used for FZ and FZD¹². The same applies to the next generation, whereby FZDD, ego’s paternal aunt’s granddaughter, is classified as yaiché membĭ (literally, father’s sister’s daughter), but the skew stops here, since all relatives in the second descending generation are classified as ego’s grandchildren.

Other missing terms may be attributed to the formidable task of searching for kinship vocabulary among thousands of pages of lexical notes and religious texts written in Guarani and seventeenth-century Spanish, even more challenging considering that Lafone Quevedo had no prior knowledge of the Guarani language, which must have hindered his ability to direct his search and securely scrutinize the extensive compilations of native vocabulary.

Though meticulously recorded during more than two decades of firsthand experience among Guarani-speaking populations, we must bear in mind that Montoya’s lists of kinship vocabulary were originally composed to aid missionaries in the religious conversion of the indigenous as were his other studies of the Guarani language and culture. In fact, his texts predate in over two centuries the publication of Morgan’s (1871) “Systems of consanguinity and affinity of the human family”, a work that inaugurated the anthropological study of kinship pointing to differences between descriptive and classificatory terminologies (Trautmann, 1987). Needless to say, Montoya’s records also predate Lounsbury’s (1968 [1964]) important distinction of Iroquois and Dravidian vocabularies based on the examination of terms assigned to relatives in the first descending generation.

It is therefore unsurprising that Montoya’s works do not address relations beyond four genealogical connections, therefore excluding terms applied to cousins’ children, with a few noteworthy exceptions. Considering a male ego, Montoya presents classificatory terms only for the sons of male cousins (FBSS, FZSS, MZSS, MBSS), who are all classified as sons of ego, just like the term used for ego’s brother’s son. On the other hand, no terms are given for the daughters of male ego’s same-sex cousins or for the sons of opposite-sex cousins, with the exception of a descriptive term found in ‘Tesoro’ for FZDS. For the daughters of such female cousins, we are presented with a list of descriptive terms, for example: FBDD is described as tubĭ rayĭ mêmbĭ réra (literally, father’s brother’s daughter’s daughter). Even less information is given on the vocabulary used by women to address their cousins’ children, for whom no classificatory terms are found. In addition to this, descriptive terminology is presented only for two categories of male cousins’ sons (FBSS and MBSS) and all categories of female cousins’ daughters (FBDD, MZDD, FZDD and MBDD). However, nothing is said of these female cousins’ sons, with the exception of FZDS, for whom we find a descriptive term in the ‘Tesoro’.

Considering that the author reports both classificatory and descriptive terms for all relatives of men and women in ego’s own generation, with the exception of male ego’s FBS and MBS, for whom we find only descriptive terms¹³, we might be tempted to say that Montoya used descriptive terminology as a quick fix to fill in relations for which he lacked specific information. Nonetheless, Montoya’s notes on terminology reveal skews that, though also reported by Lafone Quevedo, have since then widely disappeared from descriptions of Guarani kinship vocabularies¹⁴. In addition to the forementioned generational skew by which FZ = FZD, we find other examples summarized in Table 2.

Table 2
Skews reported by Montoya. ^* = Where Montoya does not specify if the term applies to cross or parallel kin, the information has been inferred according to other data provided by the author. For instance, where the author states that the term for female ego’s eZ also applies to ‘niece’, without specifying if this refers to BD or ZD, it is inferred to pertain to ZD (a parallel relative). This is consistent with information provided for male ego, where the author clearly reports that the same generational merge applies to children of same-sex siblings, thus parallel relatives.

Sex of Ego	Generations	Positions	Term	Reference
M/F	G+1, G0	FZ=FZD	yaiché	C (p. 319)
M/F	G+1, G0	MB=MBS	Tutĭ	T (p. 404)
F	G0, G-1	eZ=ZeD*	Tĭq	C (p. 323) V (p. 366)
F	G0, G-1	yZ=ZyD*	quĭ pĭí	C (p. 321)
M	G0, G-1	eB=BeS	Tĭqueĭ	T (p. 392)
M	G0, G-1	FZD=ZD*	Yetipé	C (p. 319)
M	G0, G-1	FZS=ZS*	Yĭra	C (p. 319)

Comparing this information with the Guarani Mbya terminology presented below, it becomes clear that some of these skews not only persist, but also help elucidate an important and unexplained controversy in the vocabulary of the Guarani Kaiowa. But first, a few notes on the classification of contemporary Guarani populations and how they have been divided into different subgroups.

CURRENT GUARANI TERMINOLOGIES DIVIDED BY THREE

The Mbya and the Kaiowa, along with the Nhandeva, comprise three Guarani subgroups currently found in Brazil, Paraguay and Argentina. However, in 1612, when Montoya began working in the Guairá missions, the indigenous populations that spoke the same language or similar dialects and shared many cultural traits were loosely identified as Guarani (Melià et al., 1987, p. 56). Classifications grew even murkier in the following centuries as ethnonyms proliferated and were haphazardly attributed to Guarani-speaking groups in various regions (Schaden, 1974 [1954], pp. 1-2). It wasn’t until 1928 that Müller (1928, 1934-1935; Melià, 1997) classified the Guarani in Paraguay in three different subgroups: Pañ (Paĩ), Chiripá and Mbyá, based on cultural and linguistic differences. A few decades later, Schaden (1974 [1954], 1963) confirmed this information for the subgroups in Brazil, with the difference that the Pañ (Paĩ) in Brazil are called Kayová (or Kaiowa) and the Chiripá, while identified by other Guarani groups as such, call themselves Ñandéva (or Nhandeva).

Akin to the mystery of the holy trinity: father, son and holy ghost being simultaneously three and also one, the Guarani tripartite classification persists, but not without generating a great deal of perplexity. In other words, current ethnographies rely on this classification despite a growing sense of its inadequacy, since determining where oneness prevails or the division begins has often proved tricky. After all, the three subgroups not only speak interintelligible variations of the Guarani language and have many cultural similarities, but frequently also live in the same villages and may intermarry. For instance, my data on 542 marriages, though mostly focused on the Mbya, includes 25 marriages among people from different subgroups.

With regard to kinship terminology, Dietrich (2014) compares the Mbya kinship terms recorded in the “Atlas lingüístico Guaraní-Románico” (Dietrich et al., 2015) with the Kaiowa terms presented by Pereira (1999) and indicates that the two vocabularies are very similar, if not identical. The author also suggests that these similarities might reveal one underlying structure of social organization. If this is so and provided that the same applies to the Nhandeva, we would have evidence that, at least with respect to this very important dimension of language and culture, the division of the Guarani in three distinct subgroups proves unjustified.

Unfortunately, this comparison would have to wait, since, aside from lack of data on the Nhandeva terminology, which was not included in Dietrich’s study, we also find that the Mbya vocabulary that he analyzes does not provide information concerning the terms used for cousins’ children. This void has also prevailed in other descriptions pertaining to Mbya kinship (Ladeira, 2007; Macedo, 2009; Mello, 2006; Pissolato, 2006) and, although Vasconcelos (2011) includes these positions, she does not perceive important structural variations (skews) in men’s and women’s usage of kin classifications, which allow for a more precise comparison with the Kaiowa terminology, as will be discussed later on.

Similar lacunas are detected in the study of the Nhandeva vocabulary. For instance, Nimuendaju (2013 [1908]) presented a genealogy in his pioneer study of this group, but did not provide information about terminology. In a recent attempt to address innovations in Guarani kinship nomenclature, Chamorro (2017) compares contemporary Nhandeva vocabulary with Jesuit accounts of terms used by the Guarani in the seventeenth-century, prior, therefore, to any identification of specific subgroups. She also includes in this comparison the Mbya terminology registered by Dietrich et al. (2015) as well as the Kaiowa vocabularies recorded by V. Watson (1944), Melià et al. (1976) and Pereira (1999). However, Chamorro’s own data concerning the Nhandeva gives no account of terms used for cousins in ego’s generation or their children in G-1. In addition to this, all of the collateral positions in G+1 (father’s and mother’s siblings) are actually Spanish terms, leading us to believe that the Nhandeva have incorporated European classifications into their kinship vocabulary, but the article does not address possible structural and semantic implications of these substitutions.

Chamorro’s findings diverge from those presented by Mello (2006), E. Silva (2007), Albernaz (2009) and Vasconcelos (2011), also concerning the Nhandeva, though it should be noted that the four latter collected data among groups that currently live in villages with Mbya presence. In addition to not finding any European substitutions, Mello (2006), E. Silva (2007) and Vasconcelos (2011) report that the Mbya and Nhandeva terminologies are identical, except in a few aspects, which differ according to each author.

Mello (2006) observed a single distinction in the Mbya term for mother (aiy) and the Nhandeva term (tchi or cy). This distinction is ignored by E. Silva (2007), Albernaz (2009) and Vasconcelos (2011), probably due to the fact that terminologies across the Guarani spectrum identify both aiy (also ha’i) and tchi/cy as terms for mother, being that the former is frequently used as an allocutive term (Symeonidis, 2015, p. 276).

Vasconcelos (2011) also noted only one distinction pertaining to the term used for mother’s sister (MZ) by ego of both sexes. In Mbya, MZ would be chy’y, while the corresponding Nhandeva term would be djaitché. Such a difference in the two terminologies stands out for several reasons. In the first place, the term djaitché has been documented in reference to father’s sister (FZ), in opposition to chy’y (MZ), in all Guarani terminologies from Montoya’s seventeenth-century records to the present. This would mean that, contrary to other records in which the term for MZ (chy’y) is registered as deriving from the term for mother (chy), thus, merging these two categories in opposition to FZ, in Vasconcelos’s (2011) description the collateral positions of FZ and MZ form a ‘superclass’ of ‘aunts’. However, with respect to father’s brother (ruwy) and mother’s brother (tuty), the author’s model is in accordance with all other cited research (with exception of Chamorro’s), which approximate FB to the category for father (ru) in opposition to MB. In other words, the author presents a rather tricky nomenclature, where the classification for MZ and FZ would be typical of an Eskimo system (collateral relatives in G+1 are grouped in the same class), while FB and MB would maintain a bifurcate merge typical of Iroquois and Dravidian terminologies (opposition between parallel and cross relatives in G+1).

E. Silva (2007, p. 215), in turn, indicates other minor variations, which do not, however, result in differences in the semantic structures of the Mbya and Nhandeva vocabularies. According to the author, the Mbya term for female ego’s son (pia) corresponds to memby kuimba’e in Nhandeva, and the Nhandeva term for female ego’s daughter adds the word kunha to the Mbya term memby. Other differences refer to terms for affines: wife’s father (Nhandeva: ratyu; Mbya: ra’yxiru¹⁵) and female ego’s ‘sister-in-law’ (Nhandeva: ucke’i; Mbya: rovaja), though the author does not specify if the position refers to HZ or BW. For wife, we find the Nhandeva term rembireko and the Mbya ra’y xy, though both are usually described in the Mbya terminologies as alternatives that, in the first case, refers to a wife before the conception of a child, whereas the latter literally translates as ‘Son’s Mother’. What is really intriguing, however, in E. Silva’s (2007) description of the Nhandeva vocabulary are the terms used by egos of both sexes for kin in the first descending generation. That is, what Chamorro (2017) does for the first ascending generation, classifying all collateral kin as ‘uncles and aunts’, irrespective of crossness, E. Silva (2007) does for kin in G-1, referring to all siblings’ and cousins’ children as ‘nieces and nephews’, although E. Silva (2007) does this without the use of Spanish terms. More specifically, male ego classifies son as ray and daughter as rajy, but uses the terms ray kyrĩ (male alter) and rajy kyrĩ / jaxipe (female alter) for all collateral relatives in G-1. Female ego classifies son as memby kuimba’e and daughter as memby kunha, but applies memby kyrĩ kuimba’e (male alter) and memby kyrĩ kunha (female alter) to all collateral kin in G-1. Thus, the word kyrĩ is used to distinguish nephews/nieces from sons/daughters. The only position that finds exception is an optional skew for female ego’s MBDD, who can be classified as ‘niece’ or ‘younger sister’.

The Nhandeva terminology recorded by Albernaz (2009) adds even more mystery and controversy, since contrary to all other Guarani terminologies thus reviewed, the author notes no distinction among terms used by men and women for kin in ego’s generation or the descending generation, with the exception of different terms used by males and females to refer to their daughters and sons. In addition to this, all terms applied to cousins in ego’s generation¹⁶ are indiscriminately extended to these cousins’ children and also to siblings’ children in G-1. The terms assigned to kin in G+1 also present unusual equations, where the author states that the terms ruvy (normally assigned to FB in other Guarani terminologies) and tuty (usually applicable only to MB) are both used to designate FB and MB. The aunts (MZ and FZ), as is also the case in Vasconcelos’s (2011) description, share only one common term: chaine. In sum, Albernaz (2009) provides a Nhandeva terminology that is compatible with the Eskimo type, presenting a system that translates neatly into English kinship nomenclature, if not for the fact that siblings’ children are classified as cousins, rather than ‘nieces’ and ‘nephews’.

Surprisingly, the authors of the forementioned Nhandeva vocabularies offer no comments accounting for the distinctive aspects of their data with respect to each other’s work. In fact, no systematic attempt has been made to cross analyze these Nhandeva kinship vocabularies. However, judging from the brief comparison made here, we are confronted with five accounts that present different semantic structures for what the authors claim to be populations that comprise the same Guarani subgroup.

This constellation of gaps and undiscussed controversies in research on the Mbya and Nhandeva kinship vocabularies is remarkable, considering the impressive volume of studies concerning these indigenous populations, in addition to the recurring affirmation of the importance of kinship in connection with other themes that have been given a great amount of attention in Guarani studies, such as religion, politics, territoriality, spatial mobility and the formation of villages (tekoa).

Turning once again to Dietrich’s (2014) hypothesis concerning the identity between the Mbya and Kaiowa terminologies, an additional problem comes to surface. The author overlooked the fact that Pereira’s (1999) analysis of Kaiowa kinship, used as a basis for comparison with the Mbya vocabulary, included a cross-examination of data collected previously by V. Watson (1944) and Melià et al. (1976). What appears to be a mere detail is actually of fundamental importance in the elucidation of the semantic structure of the Guarani kinship system, considering that there are three points of divergence in the terminologies presented by V. Watson (1944), on the one hand, and Melià et al. (1976), on the other hand, the latter of which is corroborated by Pereira’s (1999).

In the first place, V. Watson’s (1944) terminology presents a neutralization of the opposition between parallel and cross relatives in ego’s generation. Hence, siblings, parallel cousins and cross cousins are all identical. Melià et al. (1976) and Pereira (1999), on the contrary, maintain the distinction between parallel and cross relatives in ego’s generation, so that siblings and parallel cousins are grouped together in one class opposing cross cousins¹⁷.

However, Pereira (1999, p. 78) adds a remark to his data that seems to attenuate this discrepancy in the Kaiowa terminologies: “The terms for cross cousins are only used by people who don’t consider themselves closely related in social terms (if they don’t live together and aren’t part of the same political and/or religious composition)”.

The second point of dispute between the two models corresponds to the classification of relatives in the first descending generation and derives precisely from the cited divergence in the classification of parallel and cross cousins. According to V. Watson (1944), children of siblings, parallel and cross cousins are classified based solely on the sex of the connecting relative. For instance, a woman classifies her female cross and parallel cousins’ children in the exact same manner as she does her sister’s and her own children, in opposition to the children of male siblings, cross and parallel cousins, who are grouped together in another class. The same logic applies to male ego. Such a system is perfectly compatible with the Iroquois type (Lounsbury, 1968 [1964], fn. 4).

In contrast, Melià et al. (1976) and Pereira (1999) claim that relatives in G-1 are classified according to crossness in G+1, so that siblings’ and parallel cousins’ children are identical to ego’s own children, in opposition to his sisters’ and cross cousins’ children. However, this only applies to male ego, since Melià et al. (1976) provide no data concerning crossness in G-1 for female ego¹⁸. Pereira (1999), in turn, classifies all female ego’s relatives in G-1 as crossed, with the exception of ego’s sisters’ children and her own offspring.

A third discrepancy refers to the terms used for relatives in the second descending generation (grandchildren). While Melià et al. (1976) and Pereira (1999) record the term remiarirõ for male and female ego’s grandchildren of both sexes, V. Watson (1944) presents the term ramianinó for the same positions. If we compare this information with Montoya’s (1639, p. 378) Guarani terminology in the ‘Tesoro’, we find that both terms (with slight orthographical differences) are recorded, but the first applies to female ego, whereas the latter to male ego. The same usage is found among present-day Nhandeva (Mello, 2006; E. Silva, 2007; Albernaz, 2009; Vasconcelos, 2011) and Mbya (see data below). In addition to this, J. Watson (1952, p. 37), V. Watson’s (1944) husband, speculates that in the past the Kaiowa may also have used both terms for grandchildren according to the sex of the speaker.

To put matters simply, the analysis of the Kaiowa terminologies produced to date reveals two very different portrayals of how this group organizes kinship relations. In a wider perspective, these discrepancies are representative of the whole, as we have seen with respect to the controversies that abound in the the Mbya and Nhandeva vocabularies. In other words, to the despair of those who might hope that the plethora and centuries of studies on Guarani kinship terminologies may have matured into a consolidated corpus with eventual variations that have been adequately identified and discussed, the review presented here shows that there are still far too many grey areas that beg clarification.

THE MBYA TERMINOLOGY: REVISITING THE GENEALOGICAL METHOD IN THE DIGITAL ERA

Adding to the brew of unresolved and often unspoken disputes concerning Guarani kinship terminologies, we find that Vasconcelos’s (2011) rendering of the Mbya nomenclature, unique in its reporting of terms for cousins’ children, brings another spark to reignite longstanding controversies. The author documents a vocabulary that is identical to V. Watson’s (1944) Kaiowa terminology in flagrant contrast with the data presented by Melià et al. (1976, 2008) and Pereira (1999). It should, however, be noted that Vasconcelos (2011) compares her data with Pereira’s (1999), but mentions nothing of V. Watson’s (1944) work.

Eager to join this puzzling debate, I went back to the field to review and complete the information I had gathered during previous stages of ethnographic research. Fortunately, at the time, I was already collecting genealogical data in several Mbya villages in order to compose a database that would aid in the study of how kinship, politics and religion intersect among the Mbya. Working with genealogies was obviously a great opportunity to discuss kinship terminology, and, initially, the terminologies I gathered and checked with multiple people in villages located in the Brazilian States of São Paulo and Espírito Santo all seemed to corroborate Vasconcelos’s (2011) and V. Watson’s (1944) findings. However, as the database grew and I was able to trace more distant connections, things got tricker, or should I say, more twisted, as skews began to appear in the terminologies.

Such was the case when I sat down with an elder couple in the Três Palmeiras Village (Espírito Santo, Brazil), who were also the village’s main shamans. The wife and husband were delighted to see that I was able to accurately identify their connections to close and distant kin in villages that spread from Argentina and Paraguay to the South and South East regions of Brazil, but they were even more amused by my growing perplexity when either of them would take a term used to address kin in one generation and apply it to a relative in another generation. They were skewing what up until then had seemed to be a perfectly untwisted terminology, but I was still unable to pin-point the rules subjacent to their calculations.

During the following months, I completed the database with genealogical information concerning 1,781 individuals from over 70 villages, covering 2,382 ties of filiation (1,291 maternal ties and 1,091 de paternal ties) and 542 marriages spanning from the last decade of the nineteenth century to 2016. I ran this data on the MaqPar (Máquina do Parentesco) computer program, a software tool that processes consanguineal and marriage ties in order to identify all individuals genealogically connected, specify the various kinship ties among them and map out the different types of marriage alliances found within the network. Although the tool is primarily designed to study matrimonial rings by identifying chains of consanguineal relatives that connect one or more marriages, I used a function of the program that specifies all consanguineal ties linking any given pair of individuals in the network¹⁹.

With the aid of this computer tool, I was able to design surveys for specific informants, which contained the names of relatives in each corresponding genealogical position. In other words, the use of this program basically allowed me to apply a computerized version of the genealogical method (Rivers, 1910). It should be noted that for cousins and cousins’ children, I tested several different relatives, in order to avoid eventual ambiguities, since these categories are more likely to contain individuals with whom a person may be related to in more than one way. For instance, due to the recurrence of matrimonial alliances in which a pair of siblings marries another pair of siblings, it is very common to find someone who is simultaneously ego’s MZS and FBS or other similar variations of parallel and cross cousins. These surveys were applied in the Tekoa Pyau village (São Paulo, Brazil) in interviews with a 57 year-old man and his 29 year-old daughter, the latter of which was accompanied by her mother, who was keen to verify if the information given by her daughter was correct.

The data collected during this research provided the nomenclature corresponding to all genealogical positions, therefore allowing a complete description of the semantic structure of the Mbya kinship terminology. The structure is presented below in the form of Figures 1-4, beginning with terms used by male ego and followed by the vocabulary for female ego.

Figure 1
Male ego: consanguineal relatives.

Figure 2
Male ego: affinal relatives.

Figure 3
Female ego: consanguineal relatives.

Figure 4
Female ego: affinal relatives.

The analysis of this semantic structure presents a system with the following characteristics: the presence of five generations (G+2, G+1, G0, G-1, G-2), distinct sets of terms for consanguineal and affinal relatives, differences in terms according to ego’s sex, alter’s sex and the sex of connecting relative. In general, the Mbya terminology corresponds to what Lowie (1968 [1929], p. 42) described as a ‘bifurcate merging system’ and Murdock (1949, pp. 223-224) classified as ‘Iroquois terminology’.

With reference to consanguineal kin, there are no differences in terms used by male and female speakers for relatives in the two ascending generations, whereas, from ego’s generation down, male and female speakers use completely distinct terms. In G+2, we have one term for all male kin (ramoĩ) and another for female kin (jaryi), while in the first ascending generation we observe a bifurcated merge and a distinction between cross and parallel relatives. That is, the terms for mother (xy²⁰) and her sister (xy’y) merge to form a superclass in opposition to father’s sister (jaixe). The same occurs with males in this generation, where father (ru) identifies with his brother (ruvy) opposing mother’s brother (tuty).

In ego’s generation, we observe a complete neutralization of the opposition between parallel and cross kin for female speakers, where all relatives are regarded as brothers or sisters, distinguished by relative age to ego. This is compatible with the variant 3 of the Iroquois terminology or ‘Type B crossness’ (Trautmann & Barnes, 1998, pp. 30-34). The same neutralization occurs for male ego with reference to kin who are related to him by up to four genealogical connections (FBCh, MZCh, FZCh and MBCh) but, from six connections onward, cross cousins (ex: FMBDD) can optionally be classified in the same manner as relatives in the first descending generation (as children of opposite-sex cousins). In such cases, the opposition between parallel and cross relatives becomes relevant, corresponding to what Trautmann and Barnes (1998) typified as variant 1 of an Iroquois terminology/Type B crossness.

In the first descending generation, the distinction between parallel and cross kin based on the sex of connecting relative is operative, so that both men and women classify the children of same-sex siblings and same-sex cousins as they do their own offspring with the addition of the word kyrĩ, which translates as a diminutive²¹. For the children of opposite-sex siblings and opposite-sex cousins men use two terms, according to the sex of alter (ri’y for male relatives and jaxipe for female relatives), while women apply one term indiscriminately to relatives of both sexes (peĩ). However, the vocabulary for female ego presents an optional neutralization of the generational distinction between relatives in G0 and G-1, so that the children of opposite-sex cousins can also be classified as yB = kvyy kyrĩ, in the case of male alter, and yZ = kypy’y, for female alter. From the next generation down (G+2), male ego applies one term for all relatives regardless of sex (ramymino), as does female ego (remearirõ).

The vocabulary for relatives connected by marriage, covers three generations and is distinct for male and female speakers, with the exception of the term rovaja. This term is used by male ego for all affines of his own generation, while female speakers apply the term only to men, reserving the term ke’i for women. Still with reference to ego’s generation, a woman refers to her husband using the term mẽ, which is also applied to form descriptive terms in the ascending and descending generations, used in reference to husband’s father (mẽru) and husband’s mother (mẽxy), as well as daughter’s husband (memby mẽ). Male ego uses the same term within a nominal compound for his daughter’s husband (rajy mẽ), while referring to his wife’s father as ratyu and his wife’s mother as raixo, both classificatory terms. Men refer to their wives as rembireko, which is substituted by ra’y xy (lit. son’s mother), following the birth of a common offspring, regardless of the child’s sex. This same term is also used by male and female speakers to refer to their son’s wife, respectively ra’y ra’yxy and pi’a ra’yxy.

DISCUSSION AND CONCLUDING REMARKS

The data presented above is identical to the Mbya nomenclature registered by Vasconcelos (2011) for up to four genealogical connections, but from that point on the terminology gets twisted, with the discovery of skews for both male and female egos. It should, however, be noted that the skews identified are optional and, therefore, may not be mentioned at all by informants. In addition to this, the generational merges used by male speakers are only traceable when studying categories applied to more distant kin, that is, second-degree cousins with six genealogical connections to ego. In other words, the skews detected in this research would probably have gone unnoticed if not for the aid of a computational tool that provided the means to study each position with constant reference to a large database of genealogical ties. An even more important divergence resides in the fact that Vasconcelos (2011) does not recognize the Mbya terminology as pertaining to the Iroquois type, suggesting it to be congruent with Dravidian systems (Vasconcelos, 2011, p. 114), the same opinion held by Macedo (2009, p. 97). However, a close examination of their data and analysis indicates that the three of us have actually been dealing with an Iroquois vocabulary all along, since there are no traces of cross-cousin marriage equations in the terminology.

With respect to the skews found in the terminology, a survey of the Mbya nomenclature collected by other authors reveals that such a phenomenon has previously been noted only by Ladeira (2007, p. 179) and E. Silva (2007, p. 212). However, contrary to my findings, Ladeira (2007) states that both male and female egos can classify relatives in G-1 as siblings, if ego is younger than such relatives. According to her, the rule applies to the children of siblings and cousins, regardless of the sex of the connecting relative. After conducting tests to verify the author’s claims, I found that this generational merge does not always follow age criteria²², only applies to female ego and, even so, strictly in situations where the connecting relative is a cousin of the opposite sex. On the other hand, the skew by which male ego classifies distant cross cousins in his generation in the same manner as the children of opposite-sex siblings and opposite-sex cousins was not detected by Ladeira (2007). With regard to E. Silva’s (2007), the author only observed a skew applied by female ego in reference to MBDD, who can be classified as ‘niece’ or younger sister. This isolated case of skewing, however, remains unexplained by the author.

Moving toward a comparison with other Guarani terminologies, we find that my data partially reflects Montoya’s seventeenth-century Guarani nomenclature. Both recognize male speakers’ optional use of the terms for ZD (Guarani: yetipe / Mbya: jaxipe) and ZS (Guarani: yĭra / Mbya: ri’y) for cross relatives in ego’s generation. However, in Montoya’s case these terms are applied to FZD and FZS, who are ego’s first cousins, therefore connected to him by only four genealogical ties. My data, on the other hand, shows that this skew only appears from six genealogical connections on, thus skipping first-degree cousins. In addition to this, Montoya describes this skew exclusively for patrilateral relatives, whereas, I have found it to be bilateral, thus including matrilateral second cousins, such as MFZSD and MFZSS.

Montoya also records women’s use of the term for yZ (quĭ pĭí/kypy’y) in reference to female relatives in the first descending generation. However, from his data it can be inferred that this applies to the younger daughter of ego’s sister, whereas I have found it only to refer to the daughters of opposite-sex cousins (regardless of age). While I also describe a corresponding skew in reference to the sons of opposite-sex cousins, thus classified as yB (kvyy kyrĩ), Montoya’s data ignores this application. The author, however, draws attention to another skew, stating that the terms used to identify elder siblings can be applied to an elder daughter of female ego’s sister and an elder son of male ego’s brother²³. In short, the generational merge I identified for female speakers, Montoya describes for egos of both sexes and applies it to parallel relatives, according to the relative age of such kin, not in relation to ego, as does Ladeira (2007), but in relation to alter’s own siblings.

Two additional generational merges found in Montoya’s records also diverge from the Mbya data. According to Montoya, the terms used by male and female speakers for MB (tutĭ) and FZ (yaiché) can be extended to MBS and FZD, respectively. It should be noted that the inclusion of these optional skews renders the skewing of the terms yĭra (FZS, ZS) and yetipe (FZD, ZD) reciprocally consistent. Reciprocity which is not found in my own data or in Montoya’s description of the merging of relatives in ego’s generation and the descending generation, as specified in the paragraph above. In such cases, we must recall that this lack of reciprocity does not deem the system inconsistent, since all skews identified by Montoya and myself are optional variations within the conceptual grid of kinship. As such, they represent the most unstable parts of the semantic structure, meaning that they may be more susceptible to change over time. In this sense, we can only speculate that perhaps in the past the Mbya and Guarani in general applied additional skews that allowed for greater reciprocity, but that they have gradually fallen into disuse.

Nonetheless, it is clear that some of the skews documented by the seventeenth-century author concerning Guarani-speaking populations in Paraguay have withstood the test of time and space, since they are also identified among current-day Mbya living along the Brazilian shore. The same could be said of the Kaiowa in Mato Grosso do Sul (Brazil), since it is precisely the generational merge operated by male speakers that helps to partially reconcile the nomenclature described by V. Watson (1944) with those of Melià et al. (1976, 2008) and Pereira (1999). As we will recall, while V. Watson (1944) maintains that all relatives in ego’s generation are classified as siblings, the latter authors pose an opposition between cross and parallel cousins. But it just so happens that in their descriptions the terms applied by male speakers to female and male cross cousins, atipe and ri’y, are identical to the terms for cross relatives in the first descending generation, thus demonstrating a generational skew.

In addition to this, if we take into consideration Pereira’s (1999) comments that the distinction between cross and parallel relatives in ego’s generation only applies to kin who do not consider themselves “closely related in social terms (if they don’t live together or aren’t part of the same political and/or religious composition)” (Pereira, 1999, p. 78) we can infer, from his discussion of the role of close kinship ties in the formation of local groups, that the social and spatial distance he refers to tends to correlate with genealogical distance. In this sense, it is likely that the generational merge described for the Kaiowa operates in the same way as the one detected for the Mbya. That is, male speakers apply terms used for cross relatives in G-1 to distinguish cross cousins from parallel relatives in their own generation, as long as they are genealogically distant enough (with a minimum of six connections). Having to account for more distant connections without the use of computer tools may explain why V. Watson (1944) did not detect this skew and why Melià et al. (1976, 2008) were uncertain of it, attributing question marks to all of the corresponding terms.

It should also be recalled that while Melià et al. (1976, 2008) do not offer any information concerning female ego’s cross relatives in her own generation and in the first descending generation, Pereira (1999) replicates for female speakers the generational skew used by male speakers, by which cross cousins in ego’s generation receive the same term applied to relatives in the first descending generation. However, in contrast to male speakers, this term (pê) would be used for all parallel and cross relatives in the first descending generation, with the exception of ego’s own children and her sister’s children. Like V. Watson’s (1944) description for the Kaiowa, I too did not find such a skew in the Mbya terminology for female speakers. In fact, excluding the merging rule for the Mbya by which female ego can classify opposite-sex cousins’ children as siblings, my data for both male and female egos on the classification of relatives in the first descending generation is identical to V. Watson’s (1944), in contrast to that of Melià et al. (1976, 2008) and Pereira (1999), as detailed in a previous section.

With regard to the third point of discrepancy in the Kaiowa terminologies under review, the terms used for grandchildren, the Mbya vocabulary once again seems to present a point of reconciliation between the two contrasting models. While Melià et al. (1976, 2008) and Pereira (1999) claim that the term remiarirõ is used by speakers of both sexes in reference to all relatives in the second descending generation, V. Watson (1944) presents the term ramianinó for the same positions. However, as discussed previously, in the Mbya vocabulary, as well as in Montoya’s notes for the seventeenth-century Guarani, both terms apply, the first for female ego and the second for male ego.

With respect to terms used for affines, we find that the Mbya and Kaiowa vocabularies are identical for female speakers, with the exception of terms that refer to daughter’s husband (Mbya: memby mẽ; Kaiowa: peú²⁴) and son’s wife (Mbya: pi’a ra’y xy; Kaiowa: guaxã²⁵). There is also a slight dialectical variation in the terms used to address the husband’s father, but with the same semantic value (Mbya: mẽru; Kaiowa: mendúa/mendúvy). As for male speakers, the terminologies are equivalent in all three generations, with the exception of the term used for son’s wife (Mbya: ra’y ra’y xy; Kaiowa: wãchá/guaxã²⁶). It should also be noted that V. Watson (1944), in contrast with the Mbya vocabulary and other Kaiowa nomenclatures, identifies the terms ahẽ for wife’s father and haí for wife’s mother, in the place of ratyu and raixo.

Through this comparison, we find that the Mbya terminology not only accounts for and elucidates important differences in the three concurring Kaiowa descriptions, but also find that the remaining differences between the Mbya and Kaiowa nomenclatures are minor dialectical variations which result in the same semantical structure²⁷.

These findings not only confirm Dietrich’s (2014) hypothesis, with data that was up until now unavailable for an adequate cross-examination of the Kaiowa and Mbya terminologies, but also bring us one step closer to demonstrating that, from the perspective of this fundamental aspect of Guarani culture and language, a stank tripartite division of the Guarani subgroups is not only questionable, but unsustainable. Admittedly, this cannot be fully established without the elucidation of the Nhandeva kinship vocabulary, which begs further research. However, this article offers some important resources in this direction. In addition to presenting an assessment of current studies that pinpoints unresolved areas of the Nhandeva terminology that need to be addressed, the article provides a consolidated vocabulary of a Guarani variant for comparison and indicates a computational tool that can be of great methodological assistance.

Last but not least, the Mbya terminology presented here not only recuperates skews described almost four centuries ago, but also explains the genealogical arithmetic involved in these generational twists. In contrast with Lafone Quevedo’s rendering of Guarani kinship nomenclature, this terminology also affords a precise account of the vocabulary of a contemporary Guarani population without losing sight of the historic records that served as the author’s main source of reference. As a result, we can finally confront and clarify ambiguities and unfounded speculations that obscured the proper description of the Guarani kinship vocabulary. And, in light of the progress made in kinship theory since the publication of Lafone Quevedo’s important, however flawed, contribution, we can ascertain that the Guarani (Mbya and Kaiowa) kinship terminology, skews and all, is undeniably of the Iroquois type.

• 1
  In this article, I also refer to Eskimo and Dravidian systems, as they appear in some of the texts reviewed. In brief, Eskimo terminologies are defined by a principle of lineality, by which lineal relatives are distinguished from collateral kin (Godelier et al., 1998, p. 8), thus differentiating father and mother from maternal and paternal aunts and uncles, as is the case in English or Portuguese kinship. In contrast, Dravidian and Iroquois terminologies, identified as bifurcate merging systems (Lowie, 1968 [1929]), merge kin in the parents’ generation by applying a principle of crossness, based on the same-sex sibling rule, so that father is classed with his brother, as is mother with her sister (F=FB≠MB; M=MZ≠FZ). However, as Trautmann (2012, p. 36) points out, Dravidian classifications, on the contrary of Iroquois, are consistent with a rule of cross-cousin marriage, which also accounts for the merging of affinal and consanguineal terms (for example, MB=WF), whereas Iroquois terminologies present a separate set of terms for affines. The absence of the cross-cousin marriage rule in Iroquois terminologies also explains the classification of all same-sex cousins’ children as parallel and opposite-sex cousins’ children as cross, irrespective of the crossness of these children’s parents in relation to ego. Dravidian vocabularies, on the other hand, classify relatives in G-1, according to whether their parents (cousins in ego’s generation) are cross or parallel, so that the children of same-sex parallel cousins and opposite-sex cross cousins are classified as parallel, while the children of opposite-sex parallel cousins and same-sex cross cousins are cross. For a more complete review of the differences between Iroquois and Dravidian systems, see M. Silva (2010).
• 2
  Ellipses in the original quotation.
• 3
  Although Lafone Quevedo corresponded with Doroteo Giannecchini and had access to the missionary’s studies concerning the Guarani-speaking Chiriguano in Bolivia (Villar, 2021, pp. 24-25), there is no mention of Giannecchini’s reports on kinship vocabulary in his article.
• 4
  For this review the following works were consulted: the first edition of the “Catecismo de la lengua guaraní” (C), which is the same edition studied by Lafone Quevedo, the first edition and the 1876 edition of the “Tesoro de la lengua guaraní” (T), and the 1876 and 2002 editions of the “Vocabulario de la lengua guaraní” (V).
• 5
  Standard kintype notation is used throughout this article: M = mother; F = father; Z = sister; B = brother; D = daughter; S = son; Ch = children; W = wife; H = husband; e = elder relative; y = younger relative.
• 6
  Both Montoya and Lafone Quevedo describe the term as applying to the ‘aunt’s daughter’, without specifying if the aunt is FZ or MZ.
• 7
  Both Montoya and Lafone Quevedo state that the term applies to a younger niece, without specifying if she is BD or ZD.
• 8
  According to Montoya (1876a [1640]), the distinction of age only applies to ego’s same-sex siblings, in that women would use different terms for their older and younger sisters, but would address their brothers using only one term, regardless of age difference. The same principle applies to men. This is also true of contemporary Guarani kinship vocabularies, with the occasional addition of the term kyrĩ to distinguish a younger opposite-sex sibling from the elder, as is the case with the Mbya terminology. It should also be noted that Montoya reports that female ego’s term for yZ applies to younger same-sex cousins, but mentions nothing about the term for eZ extending to older same-sex cousins.
• 9
  Since Lafone Quevedo ambiguously agrees and later disagrees with Montoya’s classification of MBCh as ego’s siblings, these terms were not included in the preceding table.
• 10
  Guaira was the province in Paraguay where Montoya worked among Guarani-speaking populations.
• 11
  This citation and others originally published in Spanish or Portuguese have been translated.
• 12
  Lafone Quevedo only applies this rule to male ego’s FZD.
• 13
  Lafone Quevedo, on the other hand, only recorded Montoya’s use of classificatory terms for opposite-sex cousins, that is, male ego applied the term sister to all of his female cousins, but would use descriptive terms for his male cousins, the inverse being true for female ego. This is strange not only from a logical point of view, but especially considering that both classificatory and descriptive terms are given by Montoya in the works Lafone Quevedo studied.
• 14
  In exception to most contemporary Guarani terminologies, Melià et al. (1976, 2008), Ladeira (2007), Pereira (1999), E. Silva (2007) and Albernaz (2009) present some generational skews, but with reference to varying positions, as will be discussed further on.
• 15
  In a diagram E. Silva (2007, p. 213), uses the word ruyxyru for wife’s father.
• 16
  Siblings, identified according to their sex and relative age (older or younger than ego), are distinguished from cousins, who are identified by two different terms according to their sex: pe’i for men and membykary for women.
• 17
  It should, however, be noted that Melià et al. (1976) do not provide the terms for a woman’s cross cousins. In addition to this, the terms they identify for a man’s cross cousins are followed by question marks, indicating that the authors were not completely certain of the opposition between parallel and cross relatives in G0.
• 18
  The positions for male ego’s relatives in G-1 are followed by question marks, indicating uncertainty. However, in a more recent text, Melià et al. (2008) remove the question marks, reaffirming the system of crossness identified by Pereira (1999) contra V. Watson (1944).
• 19
  This free software tool was originally created by Dal Poz and Silva (2009). A new version of the program, developed by Álvaro Junio Pereira Franco, is available at: http://maqpar.ufsc.br
• 20
  The term ha’i is also occasionally used as a vocative for mother.
• 21
  Unlike other Guarani terminologies, Mbya women apply distinct terms to sons (pi’a) and daughters (memby), however, the latter term is used to refer to parallel relatives of both sexes in the first descending generation.
• 22
  Though relatives in this case are classified as yZ or yB, they are not necessarily younger than ego.
• 23
  Montoya does not record the use of the term for male ego’s yB (tĭbĭ) in reference to relatives in G-1.
• 24
  According to V. Watson (1944) and Melià et al. (1976, 2008), since Pereira (1999) does not address this position.
• 25
  According to Melià et al. (1976, 2008), because Pereira (1999) and V. Watson (1944) do not address this position.
• 26
  Pereira (1999) does not record a term for this position, but presents the term tovañain for male ego’s brother’s wife, whereas the other authors record the same term used by the Mbya for this position: rovaja.
• 27
  The same basic structure is applied by the Chané of the Western Chaco. According to Morando’s (2021) data, the terminology of this Guarani-speaking population of Arawak origin is identical to the unskewed Mbya vocabulary for consanguines (with minor dialectical variations). The same applies to female ego’s vocabulary for affines, while male ego merges affines in G+1 and G0. Villar (personal communication, 17/08/2021) reports that the Chané of both sexes use the same skews presented in the Mbya nomeclature for male ego, that is, cross relatives in G0 can be classified in the same manner as cross relatives in G-1, though it is still to be elucidated if this applies to cross relatives connected to ego by at least six genealogical connections, as is the Mbya case.

ACKNOWLEDGEMENTS

I would like to thank the Guarani for guiding me through the study of their terminology, especially Poty Poran, Sebastião Karai Tataendy, Alisio Tupã Mirim, Maria Lucia Tatax?, Patricia Jaxuka, Antonio Carvalho Wera Kwaray, Vanda de Lima Carvalho, Divino Karai Xondaro, Ivanilda and Leila Rocha. I am very grateful to Marcio Silva for his assistance every step of the way and to Diego Villar, Lorena Córdoba, Pablo Sendón, Fernanda Noronha, Antonio Guerreiro and the anonymous reviewers of this article for their comments and suggestions. I would also like to thank Maria Inês Ladeira, Fabio Nogueira da Silva, Levi Pereira, Bartomeu Melià, Lauriene Seraguza, Márnio Teixeira-Pinto, João Dal Poz and Elisabeth Fernandes de Sousa for sharing their experience and advice. I am also grateful to Carlos Ferreira, Álvaro Junio Franco and Marcelo Vendramin for their assistance with the computational tools. Research for this article was carried out with the support of the São Paulo Research Foundation (FAPESP), grant nº 2015/16902-9.

• Testa, A. Q. (2022). Guarani kinship terms with a few twists. Boletim do Museu Paraense Emílio Goeldi. Ciências Humanas, 17(3), e20210100. doi: 10.1590/2178-2547-BGOELDI-2021-0100

• ERRATA

  Na página 13, onde se lia:
  

  
  
  Leia-se:
  

  
  

REFERENCES

• Albernaz, A. C. R. (2009). Antropologia, histórias e temporalidades entre os Ava-Guarani de Oco’y (PR) [Doctoral thesis, Universidade Federal de Santa Catarina, Florianópolis].
• Baldus, H. (1935). Designativos de parentesco Tapirapé [Manuscript].
• Birchall, J., Oliveira, L. H., & Jordan, F. M. (2019). Nota sobre o sistema de parentesco em Proto-Tupí-Guaraní. Boletim do Museu Paraense Emílio Goeldi. Ciências Humanas, 14(1), 79-99. https://doi.org/10.1590/1981.81222019000100006
  » https://doi.org/10.1590/1981.81222019000100006
• Chamorro, G. (2017). Terminología de parentesco: de los jesuitas a los indígenas guaraní hablantes contemporáneos. Revista Brasileira de Linguística Antropológica, 9(2), 225-259.
• Dal Poz, J., & Silva, M. F. (2009). MaqPar: a Homemade tool for the study of kinship networks. VIBRANT, 6(2), 29-51.
• Dietrich, W. (2014). Conservação e inovação no campo léxico do parentesco: o caso do Mbyá e do Guaraní paraguaio (Tupí-Guaraní). Revista Brasileira de Linguística Antropológica, 6(1), 195-216. https://doi.org/10.26512/rbla.v6i1.21066
  » https://doi.org/10.26512/rbla.v6i1.21066
• Dietrich, W., Kallfell, G., & Thun, H. (2015). Atlas lingüístico Guaraní-Románico (ALGR) (Tom. II, Léxico del parentesco). Westensee-Verlag.
• Godelier, M., Trautmann, T. R., & Tjon Sie Fat, F. E. (1998). Introduction. In M. Godelier, T. R. Trautmann & F. E. Tjon Sie Fat (Eds.), Transformations of kinship (pp. 1-26). Smithsonian Institution Press.
• Hornborg, A. (1988). Dualism and hierarchy in Lowland South America: trajectories of indigenous social organization (Uppsala Studies in Cultural Anthropology, No. 9). Almqvist & Wiksell International.
• Ladeira, M. I. (2007). O caminhar sob a luz: o território mbya à beira do oceano Editora da UNESP.
• Lafone Quevedo, S. A. (1917). Los términos de parentesco en la organización social sud-americana. Revista de la Universidad de Buenos Aires, 14(37), 5-39.
• Lafone Quevedo, S. A. (1919). Guarani kinship terms as index of social organization. American Anthropologist, 21(4), 421-440.
• Laraia, R. B. (1986 [1972]). Tupi: índios do Brasil atual (Antropologia, 11). FFLCH.
• Lounsbury, F. G. (1968 [1964]). The structural analysis of kinship semantics. In P. Bohannan & J. Middleton (Eds.), Kinship and social organization (pp. 125-148). The Natural History Press.
• Lowie, R. H. (1915). Methods and principles: ‘Kinship and social organization’ by W. H. R. Rivers. American Anthropologist, (17), 329-340.
• Lowie, R. H. (1968 [1929]). Relationship terms. In P. Bohannan & J. Middleton (Eds.), Kinship and social organization (pp. 39-59). Natural History Press.
• MacDonald, J. F. (1965). Some considerations about Tupi-Guarani kinship structures. Boletim do Museu Paraense Emílio Goeldi, Nova Série Antropologia, (26), 1-20.
• Macedo, V. M. (2009). Nexos da diferença: cultura e afecção em uma aldeia guarani na Serra do Mar [Doctoral thesis, Universidade de São Paulo, São Paulo].
• McLennan, J. F. (1865). Primitive marriage MacMillan & Co.
• McLennan, J. F. (1886 [1876]). Studies in ancient history comprising a reprint of Primitive Marriage MacMillan & Co.
• Melià, B., Grünberg, G., & Grünberg, F. (1976). Etnografía Guaraní del Paraguay contemporáneo: Los Pai-Tavyterã [Suplemento Antropológico]. Revista del Ateneo Paraguayo, 9(1-2), 151-295.
• Melià, B., Saul, M. V. A., & Muraro, V. F. (1987). O Guarani: uma bibliografia etnológica Fundação Missioneira de Ensino Superior.
• Melià, B. (1997). Antropólogos y antropología en el Paraguay. Horizontes Antropológicos, 3(7), 24-35. https://doi.org/10.1590/S0104-71831997000300003
  » https://doi.org/10.1590/S0104-71831997000300003
• Melià, B., Grünberg, G., & Grünberg, F. (2008). Los Paĩ Tavyterã: Etnografía Guaraní del Paraguay Contemporáneo CEPAG.
• Mello, F. C. (2006). Aetchá Nhanderukuery Karai Retarã. Entre deuses e animais: xamanismo, parentesco e transformação entre os Chiripá e Mbyá Guarani [Doctoral thesis, Universidade Federal de Santa Catarina, Florianópolis].
• Montoya, A. R. (1639). Tesoro de la lengua guaraní Juan Sanchez. https://digital.bbm.usp.br/handle/bbm/4922
  » https://digital.bbm.usp.br/handle/bbm/4922
• Montoya, A. R. (1640). Catecismo de la lengua guaraní Edição Diego Diaz de la Carrera. https://digital.bbm.usp.br/handle/bbm/4392
  » https://digital.bbm.usp.br/handle/bbm/4392
• Montoya, A. R. (1876a [1640]). Arte, vocabulario y tesoro de la lengua Guaraní (ó mas bien Tupi) – Parte Primera: Vocabulario Maisonneuve y Cia.
• Montoya, A. R. (1876b [1640]). Arte, vocabulario y tesoro de la lengua Guaraní (ó mas bien Tupi) – Parte Segunda: Tesoro Maisonneuve y Cia.
• Montoya, A. R. (2002 [1640]). Vocabulario de la lengua guaraní CEPAG.
• Morando, M. A. (2021). Ñande ñee jekove: Lengua y praxis social entre los chanés del Noroeste Argentino (Colección Scripta Autochtona, No. 26). Instituto de Misionología.
• Morgan, L. H. (1871). Systems of consanguinity and affinity of the human family (Smithsonian Contributions to Knowledge, No. 17). The Smithsonian Institution.
• Morgan, L. H. (1982 [1877]). Ancient society K. P. Bagchi & Co.
• Müller, F. (1928). Drogen und medikamente der Guaraní (Mbyá, Pai und Chiripá) indianer im östlichen waldgebiete von Paraguay Festschrift P. W. Schmidt.
• Müller, F. (1934-1935). Beiträge zur ethnographie der Guaraní-Indianer im östlichen waldgebiet von Paraguay. Anthropos, 29(1-2), 177-208.
• Murdock, G. (1949). Social structure MacMillan.
• Nimuendaju, C. U. (2013 [1908]). Apontamentos sobre os Guarani. Tellus, 13(24), 311-360.
• Pereira, L. M. (1999). Parentesco e organização social Kaiowá [Masters thesis, Universidade Estadual de Campinas, Campinas].
• Philipson, J. J. (1946). Nota sôbre a interpretação sociológica de alguns designativos de parentesco do Tupi-Guarani. Separata do Boletim LVI da Faculdade de Filosofia, Ciências e Letras da Universidade de São Paulo, (9), 5-34.
• Pissolato, E. P. (2006). A duração da pessoa: mobilidade, parentesco e xamanismo mbya (guarani) [Doctoral thesis, Universidade Federal do Rio de Janeiro, Museu Nacional, Rio de Janeiro].
• Rivers, W. H. R. (1910). The genealogical method of anthropological inquiry. The Sociological Review, 3(1), 1-12. https://doi.org/10.1111/j.1467-954X.1910.tb02078.x
  » https://doi.org/10.1111/j.1467-954X.1910.tb02078.x
• Schaden, E. (1963). Caracteres específicos da cultura Mbüá-Guarani. Revista de Antropologia, 11(1-2), 83-94.
• Schaden, E. (1974 [1954]). Aspectos fundamentais da cultura Guarani EDUSP.
• Silva, E. M. (2007). Folhas ao vento a micromobilidade de grupos Mbya e Nhandéva (Guarani) na Tríplice Fronteira [Doctoral thesis, Universidade Federal do Rio de Janeiro, Museu Nacional, Rio de Janeiro].
• Silva, M. F. (2010). Um pequeno, mas espinhoso, problema do parentesco. Ilha, 12(2), 164-207. https://doi.org/10.5007/2175-8034.2010v12n1-2p165
  » https://doi.org/10.5007/2175-8034.2010v12n1-2p165
• Souza, G. S. (1587 [1851]). Roteiro geral com largas informações de toda a costa do Brasil. In F. A. Varnhagen (Ed.), Tratado descriptivo do Brazil em 1587 (pp. 13-365). Typographia Universal de Laemmert.
• Symeonidis, H. (2015). The development of kinship terms as an indicator of the Romanization of the Guaranitic area. In E. P. Pavlakis, H. Symeonidis, D. Drosos & A. Papageorgíou (Eds.), Estudios y homenajes hispanoamericanos III (pp. 269-285). Ediciones del Orto.
• Trautmann, T. R. (1987). Lewis Henry Morgan and the invention of kinship University of California Press.
• Trautmann, T. R., & Barnes, R. H. (1998). “Dravidian”, “Iroquois” and “Crow-Omaha” in North American perspective. In M. Godelier, T. R. Trautmann & F. E. Tjon Sie Fat (Eds.), Transformations of kinship (pp. 27-58). Smithsonian Institution Press.
• Trautmann, T. R. (2012). Crossness and crow-Omaha. In T. R. Trautmann & P. M. Whiteley (Eds.), Crow-Omaha: new light on a classic problem of kinship analysis (pp. 31-50). The University of Arizona Press.
• Vasconcelos, V. C. C. (2011). Tramando redes: parentesco e circulação de crianças guarani no litoral do estado de Santa Catarina [Masters thesis, Universidade Federal de Santa Catarina, Florianópolis].
• Villar, D. (2021). Estudio introductorio. La fe y la evidencia: Vida y la obra de Nicolás Armentia. In N. Armentia, Relación histórica de las misiones franciscanas de Apolobamba y Descripción del territorio de las misiones franciscanas de Apolobamba por otro nombre Frontera de Caupolicán (pp. 13-48, Biblioteca del Bicentenario de Bolivia, No. 4, Historias y Geografías). Vicepresidencia del Estado Plurinacional.
• Wagley, C. (1943). Notas sobre aculturação entre os Guajajara (Tenetehara). Boletim do Museu Nacional, Nova Série, Antropologia, (2), 1-11.
• Wagley, C., & Galvão, E. (1946). O parentesco Tupi-Guarani. Boletim do Museu Nacional, Nova Série, Antropologia, (6), 1-24.
• Watson, J. B. (1952). Cayuá culture change: a study in acculturation and methodology. American Anthropologist, 54(2), 3-137.
• Watson, V. D. (1944). Notas sobre o sistema de parentesco dos índios cayuá. Sociologia, 1(1), 31-48.

Responsabilidade editorial: Adam Singerman

Publication Dates

• Publication in this collection
  09 Dec 2022
• Date of issue
  2022

History

• Received
  25 Sept 2021
• Accepted
  05 July 2022