The Kinship Terminology of the Adi of Arunachal Pradesh (Padam and Minyong)

This paper describes the kinship system of the Adi of Arunachal Pradesh (Padam and Minyong subgroups), focusing on its kin terminology. This system corresponds to the Omaha model defined by Lévi-Strauss, marked by generational skewing and only proscriptive matrimonial rules. It is also characterised by a remarkable profusion of terms of address indicating birth order. A comparison with the Mising system provides an opportunity to challenge some current neo-evolutionist paradigms regarding the transformations of kinship terminologies.

The Kinship Terminology of the Adi of Arunachal Pradesh (Padam and Minyong) Pascal Bouchery 1 The Adi are a major tribal group of the northeast Indian state of Arunachal Pradesh. Their territory encompasses large swathes of mountainous land stretching along the banks of the Siang River, extending southwards to the Himalayan foothills. They are divided into a number of subgroups (Ashing, Bori, Bokar, Karko, Komkar, Minyong, Padam, Pailibo, Panggi, Pasi, Shimong, Ramo, Tangam) which tend to occupy separate geographical sectors, of which the Padam and Minyong are by far the largest. 1 The two groups are spread mostly between Upper Siang, East Siang and Lower Dibang Valley districts, the Minyong living on the western side of the Siang river and the Padam on the eastern side. 2 So close are they linguistically and culturally that they are sometimes grouped together under the name 'Padam-Minyong' in anthropological literature. 3 Padam, Minyong, Pasi and Panggi sections basically speak the same language sometimes labelled 'Lower Adi' (Blench and Post 2014;Post 2015Post , 2017Post and Sun 2017), belonging to the Tani subgroup of the Sino-Tibetan family (Sun 1993;Post and Burling 2017: 223;Post and Sun 2017: 323). However, regional variations within this language have been reported, especially between Lower and Upper Minyong. 4 2 Padam and Minyong kinship systems are very similar with only minor lexical differences, which is why they are considered together here. The kinship terminology structure of other Adi sections, as well as that of the Mising, a non-Adi ethnic group whose members nevertheless speak a language almost identical to 'Lower Adi', will be briefly discussed at the end of this paper. As we will see, some remarkable features can be observed in the kinship vocabulary of Padam and Minyong, not least of which is the profusion of terms of address resulting from a form of genealogical ranking which requires a differentiation of close relatives by birth order in everyday speech. The referential vocabulary already offers a rich set of some thirty-odd kin terms, but this eight to ten generations. Similarly, Tayeng opin is comprised of Kibing,Kirang,Kidang,Sapkom,Sapno,Tonkir and Tonmuk pinmik. 7 Padam-Minyong kinship terms 5 Information regarding Padam and Minyong kin terminology is still limited and sometimes contradictory. A dictionary of the 'Miri' language compiled by a Baptist missionary at the beginning of the 20th century (Lorrain 1907) contains a few Padam kin terms, along with others marked as 'Pasi-Minyong'. In his Aspects of Padam-Minyong Culture, Roy (1960) presents a list of kinship terms collected in the late 1950s with some comments. Although this study is to be welcomed as a valuable contribution providing a first glimpse of an Adi terminological system, the author does not distinguish between terms of address and terms of reference, and fails to grasp the system of age hierarchy by virtue of which one relative may be addressed by three or four distinct terms according to birth rank. Moreover, it is not always clear to which particular dialect, Padam or Minyong, Roy refers or whether his list contains an admixture of both. The Anthropological Survey of India conducted a field survey in the Lower Siang area in 1968-69 and gathered data on Galo and Minyong kinship systems, including a collection of core kin terms (Lal and Das Gupta 1979). Srivastava (1990), for his part, discusses a few Minyong kinship terms. 8 6 Adi kinship terms are independent nouns, mostly bisyllabic, consisting of an affixed root or a compound. As is common in Tani and other Tibeto-Burman languages, core kin terms tend to bear an aprefix (eg abu, ane, apang, aki, amo, ato, ayo), which is also found as the initial syllable of body parts among other semantic types. 9 Plurals are formed by adding post-nominal nouns that function as referential modifiers (Post 2010: 230), typically kidar (/ kɨdar/) or kerar (/ kǝrar/) in Padam and kiding (/ kɨdɨŋ/) in Minyong. Thus 'elder brothers' translates as abiyang kerar in Padam and 'grandchildren' as oten omo koo kidar. Like many Tani lects, Padam and Minyong use the final formatives -bo (/ bo/) ('father/male') and -ne (/ nǝ/) ('mother/female'), both of which primarily serve to differentiate animals' sexes and to denote the semantic gender of several kinship relations such as in magbo (DH, ZH, FZH), yingne (WZ, HZ), belbo/belne (WZH/ HBW) or kimbo/kinne (CSpF/CSpM). 7 The nomenclature presented in the list below is basically from Padam. When the two lects are known to differ, Minyong counterparts (either Lower or Upper) are indicated in footnotes or further down in the description. For the sake of clarity, only the terms of reference are listed here. A list of the terms of address and of reference used for each kin relation is provided in the Appendix. Varieties of Padam and Minyong spoken in the hills along the Siang river have lexical tones (Post 2015: 190, 203) but, generally speaking, 'Lower Adi' as spoken around Pasighat tends not to have any. As I am not qualified to address this language's complex tone system, no tone notations are given here. However, vowel length has been taken into account. Long or doubled vowels are simply represented by doubling the letter. Abbreviations used for kin type primitives are as follows: father [F] [abu-buku] (Post 2015: 203). (Post 2015: 203 The description begins with consanguineal relations from ascending to descending generations and then proceeds to affines.

Consanguineal relations
Grandparents, great-grandparents 9 Generally speaking, in the second and third ascending generation the system distinguishes terms by sex but not by bifurcation. In Padam the term abu-buku (lit 'old father') refers both to father's father and mother's father, and the term ane-neku (lit 'old mother') refers both to father's mother and mother's mother. Minyong equivalents are babu and naane. 18 Terms of reference for great-grandfather and great-grandmother are respectively buku (from bu-, 'father' and ku-, 'old') and neku (from ne-, 'mother' and ku-, 'old'). Adi offers a range of possibilities for addressing lineal ascendants at G+2 and G+3, the most common in Padam and Minyong being mijing [mɨǰiŋ lit 'old (man)'], ejing, ejo (lit 'old woman'), ojo, jokak, jokuk, jinkak, along with various compound words using the same root words: abu-mijing, ane-ojo, ejo-jokak etc. If needed, a further distinction between paternal and maternal grandparents can be made by using compound terms such as ato-abu (FF), ato-ane (FM), ayo-abu (MF) and ayo-ane (MM). Note that /ato/ (or / a:to/) and /ajo/ are common terms for 'grandfather' and 'grandmother' in many western and eastern Tani lects. 19 Parents 10 In Padam and Minyong, the words abu ('father') and ane ('mother') are used exclusively to refer to the biological or socially recognised father and mother, ie they are not extended to any collateral kin. Ane is a purely referential term, whereas abu can also be used as a term of address, along with yaayi. brother (apang), just as the mother (ane, /anǝ/) is from her sister (amo). Parents' siblings of the opposite sex are also identified by separate terms, aki for the maternal uncle and nyaanyi for the paternal aunt. In direct address, all are further differentiated according to birth order. A man whose father has four brothers calls them, from eldest to youngest, paate (eldest brother), paayong (second eldest brother), paatum (third eldest brother) and paayi (youngest brother). If the father has only two brothers, the elder is addressed as paate and the younger as paayi. Similarly, a mother's brothers are addressed as teete (/tǝǝtǝ/) or kite (/kɨtǝ/), yooyong 20 or kiyong (/kɨjoŋ/), yiyi (/jɨjɨ/) or kiyi (/kɨjɨ/), while the mother's sisters are addressed as [teete (or ate), yooyong, and ayi (or ayi-nane)]. If there are more than three maternal uncles, the latter category yiyi can be broken down into yi abiyang, yi bodong, yi aniyang from eldest to youngest. Likewise, for the mother's sister, the category ayi can be broken down into tutung, ayi bodong and ayi aniyang. The corresponding terms for the father's sister are aanyi abiyang (eldest), aanyi raadang (middle) and aanyi aniyang (youngest).
Siblings 12 Biro and birme are generic terms for brother and sister respectively. Siblings address each other according to their relative age and use the same terms for both male and female speakers. The term aabing [a:bɨŋ] is employed by younger siblings to address and to refer to an elder brother. Biibing [bɨɨbɨŋ] is also commonly used to call one's elder brother. More specific terms uttered in a vocative context can further indicate birth order such as bing abiyang (eldest), bing bodong (middle), bing aniyang or bing aniyalum (youngest). Similarly, aami is used by younger siblings to address and to refer to an elder sister, while miim(i) is of common usage in address. The latter term can be further broken down into miim botte, miim bodong, miim andeng and miim aniyang from eldest to youngest. Terms of reference for younger siblings are formed by adding aniyang ('younger') to the generic term, biro aniyang for a younger brother and birme aniyang for a younger sister. In practice, the way siblings address each other may vary from family to family.
Cousins 13 Cousins are distinct from siblings and are also all differentiated from one another. Paternal parallel cousins are treated terminologically as siblings; they also address and refer to one another by the same terms as siblings. Anying is the term of reference for maternal parallel cousins, which covers MZS (anyinge) and MZD (anying birme). In direct address, a male or female speaker calls his/her mother's sister's sons by using appropriate kin terms only if they are older than him or her, otherwise by their names. Various derived terms are employed in vocative contexts to express birth order, eg miim botte, miim bodong, miim andeng, miim aniyang for FBD, or aanyi biibing and aanyi miim(i) for MZS and MZD respectively, although usage may vary from one family to another.
14 No specific term exists for cross cousins. Maternal male cross cousins are equated with maternal uncles (aki) and addressed in a similar way by distinguishing them according to birth order (serially teete, yooyong, yiyi), at least if they are older than the speaker. Kiki and aki de are possible alternatives if the speaker and the addressee are of approximately similar age and enjoy a relationship of friendship. If younger, teete ngaanga, yooyong ngaanga, yiyi ngaanga are more appropriate, but even so they should not be called by name. A single son is usually addressed as yeyeng or yiyi in Minyong. Maternal male cross cousins can be referred to simply as aki or aki oo ('maternal uncleson'), or aki ngaanga ('younger/junior maternal uncle'). In a similar way, maternal female cross cousins are equated to maternal aunt (amo) and addressed by using appropriate kin terms according to birth order: teete (or mote, /mo tǝ/), yooyong (or moyong), ayi (or moyi). 15 Paternal male cross cousins are equated with children and referred to simply as 'sons' (oo, for FZS) and 'daughters' (ome, for FZD). Oo (or o-de) and ome are also appropriate terms of address but only if the paternal cross cousins are older than ego. Alternatively, they can be called oo-biibing and oo-meme (lit 'son-elder brother' and 'son-elder sister'), which can be contracted to biibing and miimi/meme, though the latter are less common than oo and ome and may not be regarded as appropriate by some. Another compound word, aanying ngaanga, applies to a younger male or female cousin.
Children, siblings' children 16 Oo is the term for son, but also by extension for child, and ome is the term for daughter. Parents usually call their children by their personal names or by nicknames. Firstborn sons and daughters are referred to respectively as oo obing and ome obing. Irrespective of the speaker's sex, all siblings' children are equated to children and referred to as oo/ ome. They are addressed by their names or nicknames, as are own children.
Cousins' children 17 Children of paternal parallel and cross cousins (FBSC, FBDC, FZSC, FZDC) are equated to children (oo/ome) and addressed by name only. MBSS, as his father, is referred to as aki or kiki ('maternal uncle') while MBSD, as her mother, is referred to as amo ('maternal aunt'). Appropriate terms of address for them are respectively teete (ngaanga), yooyong (ngaanga), yiyi (ngaanga) and teete (or mote), yooyong (or moyong), ayi (or moyi), according to birth order. Similarly, MBDS, as his mother, is referred to as aanying biro. In direct address he may be called nyiinying biibing if he is older than the speaker and nyiining if he is of the same age group.
Second and third descending generations 18 The term of reference for a grandchild is oten. To specify gender, one has to add oo for a grandson and ome for a granddaughter. To our knowledge, there is no separate kin term for great-grandchildren. Both grandchildren and great-grandchildren are called by their names or by nicknames.

Affinal relations
Spouses 19 Husband and wife commonly call each other using nicknames or they resort to teknonymy by using a child as a linking relative. A woman can also simply call her husband yayi or abu ('father'), whereas he can reciprocate by calling her ayi or ane ('mother'). The Padam words for 'husband' and 'wife' are milo and meeng respectively. The corresponding terms of reference in Minyong are milo or mijing ('husband') and myeng or mime ('wife').
Parents-in-law 20 The Padam and Minyong terms for father-in-law and mother-in-law are ato (SpF) and ayo (SpM). However, in a vocative context, a Padam woman would call her husband's father ato-abu (lit 'father-in-law-father') and her husband's mother ato-ane ('father-inlaw-mother'), or simply abu and ane (ie she addresses parents-in-law in the same way as her husband addresses them).

Spouses of uncles and aunts 21
The paternal uncle's wife (FBW) is referred to and addressed as nane in Padam and as maam or maami in Minyong. 21 The latter term is often prefixed or suffixed by the village of origin, or the subclan/surname of the addressee (eg Mongku maami). The term of reference for a paternal aunt's husband (FZH) or a maternal aunt's husband (MZH) is magbo. Taaram, taaram yaayi, magbo-abu or magbo yaayi are used in direct address, but taaram can be used referentially with proper names (eg 'Taaram Molɨŋ': 'Uncle Moling' said that...) (Post, pers comm). Yaayo is a specific term for the maternal uncle's wife (MBW) and used vocatively or referentially.
Siblings' spouses 22 A male or female ego addresses the wife of an elder brother as mamo and the wife of a younger brother as nyameng (/ ɲamǝŋ/), though both are referred to as nyameng ('daughter-in-law'). A sister's husband is addressed and referred to as magbo by both sexes, irrespective of the brother's or his wife's relative age to ego. But the husband of an elder sister is more appropriately called magbo aabing (aabing, 'elder brother') or magbo yaayi. In Minyong, babi is similarly used to address one's elder sister's husband (real or classificatory) in a respectful manner.
Spouses' siblings 23 All spouse's brothers fall into the category yigo, and all spouse's sisters into the category yingne. But the terms of address are not exactly the same for a male and a female speaker. A married man addresses all his wife's siblings by the same terms she uses. A married woman addresses her husband's sisters by the same terms her husband uses, but not his brothers whom she usually calls 'paternal uncle' (paate, paayong, paatum, paayi according to birth order), as her children would.
Cousins' spouses 24 All female cousins' husbands (FBDH, FZDH, MBDH, MZDH) fall into the category magbo, 'son-in-law', whereas FBSW, FZSW and MZSW fall into the category nyameng, 'daughterin-law'. Only the wife of the mother's brother's daughter is treated separately by being classed under the same kinship label as the mother's brother's wife, yaayo. Magbo, nyameng and yaayo are also used as a term of address, especially if the addressee is older than the speaker. Wives of paternal cousins (FBSW, FZSW) older than ego are called mamo, ('elder brother's wife'), whereas wives of maternal cousins (MBSW, MZSW) are simply addressed as nyameng ('daughter-in-law') irrespective of their relative age to ego.
Children's spouses 25 Magbo and nyameng are generic terms for 'son-in-law' and 'daughter-in-law' respectively. Though both can be used as terms of reference and of address, in everyday speech one may call one's sons-and daughters-in-law simply by their names. Both terms cover a wide range of relatives and even non-kin (see below).
Spouse's siblings' spouses 26 Adi has two co-sibling-in-law self-reciprocal terms of reference. One is belbo, used by men whose wives are sisters. The other is belne, used by women whose husbands are brothers. Both can be broken down in address according to age hierarchy. A man may address his WZH as belbiyang (a contraction of belbo and abiyang, 'eldest') if he is older than him or belniyang (a contraction of belbo and aniyang, 'youngest') if he is younger.
Children's spouse's parents 27 Adi also has two specific terms of reference for co-parents-in-law, ie people whose children marry each other, kimbo (or kumbo, CSpF) and kinne (CSpM). All women married to a person whom an individual calls aki/kaki fall into the yaayo category.

Extension of kin terms
Similarly, the amo category covers all daughters of men whom an ego calls aki/kaki and, in a broader sense, it encompasses all women from the mother's, grandmother's and greatgrandmother's opin.
Any male person whom a man's wife would call 'brother' is his yigo, 'brother-in-law', as is any female person whom a woman's husband regards as a real or classificatory brother.
Any male person who marries a real or classificatory daughter, niece, sister, first cousin or aunt (paternal and maternal) is regarded as magbo. By extension, this category may also cover all non-kin males who are potential marriage partners for women. In a vocational context, a distinction is drawn in Padam between close and not so close affinal kin by means of the pair magbo/taaram: Magbo primarily applies to the real husbands of aunts, sisters and daughters, while taaram is a term of respect for more distant kin. But one may also choose to address one's cousin or even friend as magbo if one wishes to express and underline their close relationship.
Similarly, the term nyameng may be applied to all in-married women of a male individual's opin (own, brothers', sons', paternal real or classificatory cousins' wives). Typically, cross cousins are differentiated from parallel cousins and designated by terms applying to consanguineous kin that cut across generational divisions. Maternal cross cousins are 'raised' a generation and equated to ego's mother's brother (aki) or ego's mother (amo) depending on their sex, while those on the paternal side are skewed downwards a generation and merged with children (oo, ome). Very much like the Omaha nomenclature proper, the aki category primarily denoting the mother's brother extends upward to also include MMB, MMMB and their further male agnatic descendants (MMBS, MMBSS, MMMBSS etc). In other words, all male colineals of an ego's matriline that are traced upward and their male agnatic descendants are lumped together. This also extends on the father's side to cover father's mother's brothers (FMB) and all male descendants of the corresponding patrilines (FMBS, FMBSS etc). 22 Unlike most Omahan systems, however, the Adi system is not bifurcate merging but bifurcate collateral, ie an individual refers to his or her parents by distinct terms and uses separate terms for each kind of parent's sibling. In addition, possibly as a result of the distinction made between all parent's siblings, a separate term (aanying) is used for MZC, therefore only parallel patrilateral cousins are merged with siblings. 30 The classifications of this type of system, like the Adi system, appear to be complicated at first glance. Yet they are simple to native speakers who follow a few basic rules. Two brothers' children call each other 'brothers' and 'sisters', but a brother's and his sister's children call each other differently. A brother's son and daughter automatically become maternal uncle and aunt for his sister's children. Thus, though they are actually first cousins, they are 'nephew' and 'niece' for the brother's children. This principle can be extended to encompass larger units. For example, for a female born Ering but married to a man of the Bamin opin, all Ering individuals, male and female, automatically become her children's uncles and aunts, whereas every child born of a Bamin girl automatically becomes her children's niece/nephew.
Marriage system 31 The matrimonial system of the Padam and Minyong Adi also conforms to the 'Omaha' exchange system as defined by Lévi-Strauss (1965) and is characterised by extensive marriage prohibitions. In the Padam and Minyong Adi terminologies, there is no positive rule that prescribes marriage with a person of a particular kinship position. Instead, both groups state marriage prohibitions that apply to certain kin groups: a man cannot marry a woman from his own opin among the Minyong or among those who are regarded as allied and therefore form one larger exogamous unit, olung. 23 There are also rules prohibiting repeated alliances between affines in both groups. As a rule, marriage with a first cousin is not permissible. Moreover, a man should not take a wife from his mother's section, from his paternal grandmother's section or from his maternal grandmother's section. It is generally acknowledged that for a marriage to be contracted there should be an acceptable distance from the mother's side: five generations according to Borang (2015: 81). 24 In any case, an individual is forbidden to replicate their parents' marriage and is forbidden or at least discouraged from replicating the marriage of their grandparents or great-grandparents. It follows that each time a descent line is selected to provide a spouse, all its members are excluded from the pool of potential spouses for the reference line. This especially entails the exclusion of MB, FMB, MMB and MMMB patrilines for several generations.
32 As women retain their father's name and opin affiliation upon marriage, it is easy in practice for individuals to determine who may or may not marry. For example, a male whose mother is Yirang, whose maternal grandmother is Perme and whose maternal great-grandmother is Saring will regard all members of Yirang, Perme and Saring sections as his aki and amo, according to their sex, regardless of age or generation. He will also regard all members of his paternal grandmother's opin (eg Tayeng) as his aki/ amo and will therefore avoid marrying a girl belonging to any of these four sections. This relation is inherited so his children will consider all members of Yirang, Perme, Saring and Tayeng opin as maternal affines. This system provides the possibility of repeating past alliances once a required number of generations has elapsed but suffices to lead to a general dispersal of alliances at each generation. Note that there is close congruence between the terminology and the ban on the repetition of marriage between certain lines of descent, since the term aki (MB), as seen above, covers members of those patrilines with whom marriage is forbidden. Padam and Minyong terminologies thus contain indirect indications of marriage regulations. 33 One may conjecture that the main purpose of dispersed affinal alliance is to use marital ties to create as many allies as possible. It makes sense in Adi society because individuals can traditionally rely on the support of their maternal relatives. Wife-giver groups as a whole are viewed as being fundamentally protective, as being under the obligation to render assistance in emergencies. Those who are closely related are expected to be more dependable allies. Being the closest relative on the maternal side, the mother's brother is a particularly respected figure among the Adi and acts as guardian for his sisters' children, even allegedly ensuring their physical safety by warding off evil spirits.
34 Once five generations have elapsed, however, the situation changes and distant cross cousins become preferential mates according to Borang (2015: 81). Given that the Adi only allow matrilateral, not patrilateral, cross-cousin marriage, a man is encouraged to marry a female descendant of one of his agnatic ascendants' aki/kaki (maternal uncle).
The system thus appears to combine an Omaha-type dispersal with the feature of replication after a lapse of some generations. 35 Another salient feature of all Adi terminologies is the fairly clear-cut distinction between consanguineal and affinal categories. As a matter of fact, the two sets never overlap. Fourteen elementary terms of reference are used by Padam and Minyong Adi to denote ties of affinity (nos 19-32), none of which include consanguineal relations. Among them, two transgenerational categories -magbo and nyameng -show the distinction between wife-givers and wife-takers. Magbo basically denotes wife-takers (ie men who marry women of a man's agnatic group (opin or pinmik) such as FZH, ZH, DH, FBDH, BDH), whereas nyameng indirectly refers to wife-givers by primarily denoting all women who marry male members of a male ego's patriline (BW, FBSW, FZSW and SW). By extension, the latter category encompasses all in-married women of a male individual's opin. Figure 1: distinction between wife-givers and wife-takers 36 Thus characterised by a general distinction between wife-givers and wife-takers, together with an absolute ban on marriage with FZD and a preferential alliance with distant matrilateral cousins, the Adi matrimonial system can be said to combine dispersed alliance and asymmetric alliance. Asymmetric alliance was initially defined by authors such as Lévi-Strauss (1949) and Needham (1960) as a form of marriage based on a prescriptive rule. 25 However, Löffler (1964Löffler ( , 1966 has shown for the Assam-Burma region that, in most cases, an asymmetric alliance system results from proscription alone (ie prohibiting a man from choosing his wife from groups that are wife-takers) which in itself is sufficient to induce a general orientation of marriages, and that unions with a maternal cross cousin are at best preferential. 26 This evidently applies to the Padam and Minyong systems. Neither prescribe marriage with a mother's brother's daughter -quite the opposite in fact as they explicitly forbid it -, and marriage with a second or third maternal cross cousin is socially frowned upon. Nor do they prescribe marriage with a woman of an established wife-giving descent line, though they may favour marriage with distant maternal kin. But most importantly, they prohibit unions with a father's sister's daughter and relatives from the father's side in general, thus preventing a reversal in the direction of marriage from one generation to another. 27 Birth order 37 Of particular interest is the Adi intricate system of ranking consanguineal kin according to their order of birth, which also exists in some neighbouring languages, but takes a rather complex, albeit irregular form in Padam and Minyong. Kinship terms 38 Although several possibilities are offered for each kin relation to which the age hierarchy applies, none includes them all, which confers on the system an irregular quality (see below). This is in line with what Post (2017: 50) noted for Tangam Adi. The Padam-Minyong system also presents several inconsistencies insofar as kin ordinals and the way they are formed vary to some extent from one relation to another. Most terms denoting the eldest position or order between siblings are formed by using or combining the morpheme temeaning 'big' or 'first' in Padam, while most of the terms denoting the youngest position or order are formed by using -yi meaning 'last' or 'youngest'. Yongor dongare used as suffixes attached to root words to indicate the second rank, and -tum to indicate the third rank.
39 Below are the main birth-order terms of address used by Padam and Minyong speakers. Those marked with an asterisk (*) are specifically Minyong.  40 In both dialects, age hierarchy applies primarily to consanguineals in the three medial generations: children, siblings, uncles and aunts, as well as paternal parallel cousins and maternal cross cousins, all classified and ordered in relation to one another according to their position. It even extends to more distant consanguineals (eg FFZC) provided that the addressee's birth rank is known to the speaker. Another noticeable feature is that the principle of ranking by birth order tends to blur the distinctions based on seniority. A woman, for example, may call her husband's brother paayi, implying that he is the last-born of a group of male siblings, regardless of the husband's relative age to her. Similarly, calling someone paatum, 'third/middle father's brother', does not indicate whether the uncle or aunt is older or younger than the parent of reference. Birth order simply defines the seniority of the kinship position within one terminological class. Contrary to terms denoting relative age, those denoting birth order are not relative ego-centred terms indicating a relationship between two or more persons, but are absolute and permanent. 41 It is significant that age hierarchy is only expressed through terms of address, as is the fact that birth order is also used primarily for addressing someone older than the speaker. In this regard, it is interesting to note that while all collaterals in the parental generation are differentiated according to age hierarchy, this is not the case in ego's generation. Siblings and parallel cousins are ranked, but not paternal cross cousins (FZC). This is because, due to the Omaha terminology's skewing characteristic, the children of one's paternal aunt are equated to children and therefore regarded structurally as juniors. This clearly indicates that an age hierarchy is an element of a social verticality. Being used, for the most part, for relatives older than ego, terms expressing birth order are also honorific, ie linguistic markers that signal respect and, as such, reflect social asymmetries. Their relative extension beyond the range of family and genealogically close kin leads individuals to recognise their rank and responsibilities in relation to one another no matter how distantly related.

Discussion
42 While the tribal societies of Arunachal Pradesh are highly diverse, there has been relatively little anthropological work specifically devoted to their kinship systems over the past four decades and a full investigation of the kin terminologies of the many languages spoken in this part of the Himalayas has yet to be conducted. That being said, Birth order 43 Kin terms denoting birth order are a common feature of many languages of the Sino-Tibetan family, mostly for distinguishing children and/or grandchildren. In Tibeto-Burman languages, this is usually done by means of suffixes or prefixes attached to the root word, most often ordinal numbers, to form kin ordinals or, as Benedict (1945) called them, 'kinship numeratives'. In the westernmost part of Yunnan abutting Arunachal Pradesh, Anong has sets of ten of the same sex to order sons and daughters (Sun and Liu 2009: 66), while Drung (Dulong) similarly has sets of nine, all distinct for male and female (Gros 2012: 293 (Bradley 2007: 61) extend the use of kin ordinals to siblings. Birth-order markers may also constitute an element of personal names, as in Lisu and Nuosu (Ma 2001: 83-84;Vermander 2007: 117), or be used as substitutes for personal names, as in Lamkang (Shethon 1994). In Wancho (all lects), Nocte (all lects) and Tutsa, uncles, aunts, siblings, siblings-in-law and children are distinguished according to their birth order by means of various suffixes (Burling and Wancho 1999;von Fürer-Haimendorf 1969;Sonowal 2014; own data). But Tani certainly exhibits some of the fullest sets among trans-Himalayan languages. The hierarchical structure of kinship terms based on relative age between same-sex siblings of different generations appears to be found in most, if not all, eastern Tani lects and at least some western Tani lects, such as Galo and Pailibo. 29 As is the case with Lower Adi, it primarily applies to parents' siblings and own siblings but, by virtue of their Omaha terminological structures, also extends to maternal cross cousins (since they are equated to maternal uncles and aunts) and paternal parallel cousins (since they are equated to siblings). More distinctively, certain affinal relations may also be ordered along the same principles: in Tangam, the same three basic suffixes -te, -oŋ andji serve to differentiate sisters' husbands according to age hierarchy (Post 2017: 50). In Pailibo, the eldest, second eldest, third eldest and youngest wives of one's brothers and paternal uncles are called nete, nede, neyo and neyi according to Kumar (1979a: 212). In Galo, brothers' wives (nə) and daughter-in laws (ɲaməə) are ranked similarly, although not according to their husband's position in the sibling age hierarchy but according to the order of incorporation into the parental home (Basar 2006: 295;Post 2007: 213-216 (Bouchery 2016). Both societies not only share similar terminological structures based on multiple cross-generational equations, but also similar rules that forbid marriage with close cousins and proscribe structurally identical alliances for same-sex kin, while allowing and sometimes encouraging the renewal of previously contracted alliances after four or five generations. By contrast, most of the other systems of this region that have been labelled 'Omaha' in anthropological literature (Kachin, various Naga and Kuki-Chin groups) exhibit only a minimal set of Omaha-type terminological equivalences such as MB = MBS and FZS = ZS, coupled with an asymmetric alliance pattern based on prescriptive or preferential rules, usually enjoining a man to marry one of his real or classificatory maternal cross cousins. Within the Tani linguistic cluster, the data collected to date is far from comprehensive; still there is a discernable Omahan pattern in the lects spoken in Arunachal Pradesh insofar as maternal cross cousins tend to be 'raised' a generation and equated to ego's mother's brother and sister, while those on the paternal side tend to be skewed downwards a generation and equated to ego's children, sister's children or grandchildren. This is also the case in Milang which is not a Tani language but has been influenced by Padam, and whose speakers regard themselves ethnically as Adi.  45 Several authors report the possibility of, and in some cases even a preference for, matrilateral first cousin marriages (a male marrying his MBD or sometimes MZD) among Bokar, Nyishi, Bangni, Pailibo, Ramo and Tagin (Banerjee 1999: 69, 71;Dhasmana 1979: 96;Duncan 1970;Kumar 1979a: 223, 225;Riddi 2006: 66;Shukla 1965: 133). Although some of this data is not necessarily reliable and further research is needed, it contrasts singularly with the strong disapproval of all first-cousin unions among Apatani and Padam-Minyong Adi and suggests that there is no such thing as a 'pan- Tani' marriage pattern but rather differences from one group to another that must be studied on a case-by-case basis.
46 From a terminological perspective, even though the kinship systems of northern Adi sections are still not fully described, there is evidence to suggest that most if not all nomenclatures of ethnically Adi subgroups (one may also add Galo) are structured along a common set of general principles: If we now look beyond the ethnically defined category 'Adi' and consider more broadly the Tani language cluster, the closest relatives of the Padam, Minyong Pansi and Pangging Adi are the Mising of upper Assam, who also constitute the largest population of Tani language speakers. Like the Adi, the Mising are divided into a number of regional subgroups such as Oying, Shayang, Dambok, Delo and Pagro, the latter being by far the most numerous. It is this main section that we will focus on since we were able to gather kinship terminological data. In Assam, the Pagro Mising mainly inhabit Lakhimpur, Sibsagar, Dhimaji and Majuli districts along the banks of the Brahmaputra River.  neutralising the distinction between paternal and maternal kin, is a characteristic feature of Dravidian systems. It entails typical arrangements in different generations: At G 0 , parallel cousins are equated with siblings, whereas cross cousins are equated with inlaws of ego's generation.
At G +1 , the mother's brother and the father-in-law are grouped together in the same category (akiu/kakv), just as the father's sister and the mother-in-law (nyanyi) are.
At G -1 , for each sex, the children of one's same-sex sibling's children are equated terminologically to one's own children, whereas the children of one's other-sex sibling's children are distinguished by a specific term (C = ♂BC = ♀ZC) ≠ (♂ZC = ♀BC).
51 Pagro Mising terminology also conforms to Lounsbury and Scheffler's genealogical definition of the Dravidian model based on the classification of parents' cross cousins and cross cousins' children as opposed to the Iroquois model: 31 the parents' cross cousins of the same sex as the father or the mother are cross parents, whereas those of opposite sex are siblings. For example, FFZS and FMBS are akiu/kakv, ie cross kin (they would be parallel kin in an Iroquois system). Similarly, the children of a cross cousin of the same sex are always cross children, whereas those of opposite sex are treated terminologically as children (eg ♂MBSS and ♂FZSS (eo/io) are cross kin, but ♀MBSS, ♀FZSS (aao) are parallel kin, the opposite would be true in an Iroquois system).
52 As a logical consequence of merging cross relatives with affines, Pagro Mising terminology contains very few exclusively affinal terms apart from those denoting spouses and parents-in-law. 32 Also typical of the Dravidian pattern of classification is the fact that affinal relatives of affinal relatives are treated terminologically as consanguines, eg spouse's siblings' spouses are equated with siblings and parallel cousins.
53 In fact, the binary character of Pagro Mising terminology remarkably mirrors that of their social structure, a dual organisation characterised by the presence of moieties that divide a society into two major exogamous kin groups, whose membership is determined by patrilineal descent. Marriage is enjoined between the moieties, but strictly banned within the moieties. In a child's own moiety, the male kin at G+1 are all 'fathers' (baabu/abu) and the female kin are 'mothers' (aane), whereas adults of the opposite moiety are simultaneously 'parents-in-law' and 'cross uncles and aunts' (akiu/ kakv, nyanyi). At G0, members of one's own moiety are collectively referred to as 'brothers' (biro/birang), while members of the opposite moiety are referred to as 'brothers-in-law' (igo-magbo), irrespective of their sex. 33 Since marriage between people belonging to different generations is either prohibited or strongly disapproved, both sexes are expected to select their spouse among the children of those people of the opposite moiety whom they call akiu/kakv and nyanyi. However, as a husband is expected to be older than his wife, it is customary for a man to choose his wife in the yigne category comprising all females of the opposite moiety and of his generation, who are younger than him, and for a woman to choose her husband in the magbo category comprising all males of the opposite moiety and of his generation, who are older than her. 34 54 Given the fact that Padam Adi and Mising languages are quite close, or may even be regarded as two varieties of the same language, how do we reconcile the difference between the kin terminology of the former, of Omaha type with marked skewing and no overlapping between affinal and consanguineal categories, and that of the latter which is Dravidian in every sense of the word and merges affinal relatives with consanguineals? The difference mainly lies in the presence of cross-cousin marriage among Pagro Mising, which is absent among Padam-Minyong Adi. The Pagro Mising Dravidian terminology is congruent with their marriage system, insofar as the distinction between cross and parallel kin at G0 also marks the boundary between marriageability and non-marriageability. 55 From a historical perspective, we know that the formation of present-day ethnic Mising is the result of a progressive southward migration of Siang area eastern Tani-speaking groups (mostly ancestors of present-day Adi) along the course of the Brahmaputra River. Virtually all authors, as well as the two groups' oral traditions, agree on this point. Since Adi terminological systems are fairly uniform in their structure, it is relatively safe to assume that the Pagro kin classification evolved from this general model, ie a terminological system in which a basic distinction was made between wifegivers and wife-takers, and affinal relations constituted a separate subset. It should also be noted that no other Tani language, either western or eastern, exhibits a terminological structure similar to that of Pagro Mising. Thus, if the Mising relationship terminology derives from a more stable and widespread 'Adi' pattern, how has it been transformed?
56 First, certain specific terms of affinity attested to in several Adi nomenclatures such as nane or maam(i) (FBW), yaayo (MBW) or yayi (MZH), are no longer present in Mising. This is not surprising because, in an exogamous moiety organisation, relatives of the parental generation tend to be split between consanguines and affines according to their respective moieties, hence the equivalences FBW = MZ and MBW = FZ: the distinction between same-sex parents' siblings (F ≠ FB; M ≠ MZ) becomes blurred and fades, leading to the transformation of a Bifurcate Collateral system into a Bifurcate Merging system. Likewise, Lower Adi terms of reference for parents-in-law, ato and ayo, are absent in Mising, parents-in-law being lumped together with 'maternal uncles' (akiu/kakv) and 'paternal aunts' (nyanyi). This too is expected in an exogamous moiety system where all mother's classificatory brothers are simultaneously one's spouse's classificatory parents. It is also to be noted that the 'son-in-law' category has a much more limited extension than the Adi nyameng category, it being reduced to the focal term and no longer containing affinal relatives of one's patriline.
57 Conversely, at G0 some purely affinal categories in Adi encompass a wider range of relatives in Mising due to the fact that cross cousins have been added, namely: 58 This merging of affinal and consanguineal relations is consistent from a matrimonial point of view, as in the context of an organisation into exogamous moieties where cross cousins necessarily belong to the marriageable category. The following must be emphasised: it is consanguineals who are incorporated into pre-existing affinal categories, not affinal relatives who are lumped together with consanguineals. Indeed, an examination of all Adi kinship lexicons shows that everywhere magbo, nyameng, yigo and yigne are terms of affinity and exclusively so. They have simply been extended in Mising to include consanguineals of ego's generations who belong to the opposite moiety as they are also classificatory brothers-and sisters-in-law. This is in line with Dumont's approach to Dravidian systems in which cross cousins are considered primarily as affines: it would therefore be inaccurate to say that the Mising nomenclature hardly contains affinal categories.
59 Also worth noting is that all transgenerational terms have disappeared in Mising. The maternal uncle's son is no longer treated terminologically as his father but as a 'brother-in-law', which he actually is in a system of exogamous moieties associated with the rule of bilateral marriage. Similarly, the maternal aunt's daughter is no longer a 'mother' but a 'sister', the mother's sister having necessarily married a man belonging to ego's moiety and therefore her daughter being de facto ego's classificatory sister.
60 The comparison between Padam-Minyong Adi and Pagro Mising gives us a glimpse of one of the processes by which an asymmetrical kinship system may evolve into a symmetrical system based on bilateral marriage. This has important theoretical implications since it can call into question one of the most basic and recurrent neoevolutionist paradigms that have flourished over the past two or three decades in opposition to the more conventional structural approach. The latter readily supports the idea that the particular arrangement of the categories of relatives at least partially reflects matrimonial practice, and that any change in it is likely to lead to changes in the way of classifying kin categories through a process of reciprocal adjustments. Universal theories of kinship evolution consider, on the contrary, that terminological systems are independent of social systems. They postulate the existence of a logic inherent to the kinship system itself that makes its transformations not only directional but also irreversible. In particular they claim -and too often take for granted -that all kinship terminologies evolve from a symmetric, Dravidian-like form based on a rule of bilateral cross-cousin marriage. Most importantly, they consider that a transformation in the reverse direction (ie towards Dravidian or Kariera models) is impossible or at least highly unlikely (Allen 1986(Allen , 1989(Allen , 1998(Allen , 2008Godelier 1998Godelier , 2004Kronenfeld 1989;Kryukov 1998;Needham 1967;Parkin 2012;Trautmann 1981Trautmann , 2012. Prominent and much discussed among those theories is Allen's tetradic model (1986), which postulates that all kinship terminologies have evolved via the progressive breaking of certain primordial equations such as alternate generation equivalences or equations pertaining to prescriptive marriage regulations as seen in Dravidian terminologies. According to Allen, the trend regarding the evolution or historical transformation of kinship terminologies is always to create more specific, more descriptive categories and, once primordial equations have been broken, they cannot be restored. 61 Allen's tetradic theory grew out of analyses of Tibeto-Burman kinship terminologies, so one may assume that it is particularly suited to explain the evolution of terminologies within this linguistic family. In a previous paper (Bouchery and Longmailai 2018: 244-245) I showed that the Bodo-Garo terminological corpus does not fit in well with this evolutionary model and even contradicts it. The same can be said regarding eastern Tani. Not only does the Mising kin terminology suggest the possibility of an evolution from the Omaha-type to the Dravidian-type, but it also contradicts one of the main arguments on which Allen's theory is based, namely the impossibility that the introduction of the prescriptive rule of bilateral cousin marriage alone generates the characteristic equations of the Dravidian model. In the tetradic model, a Dravidian-type symmetrical prescriptive terminology is first transformed into an asymmetrical prescriptive terminology by 'breaking' the symmetrical equations, which intervenes in order to allow for the distinction between wife-takers and wife-givers. For example: Here, it is not the symmetrical equations that are broken up but the category of wifetakers: DH = ZH = FZH in Adi is transformed into DH ≠ ZH ≠ FZH in Mising, just as the distinction between paternal and maternal fades into a general distinction between parallels and crosses. In this case, too, the change consists in turning certain close allies into consanguines, by transforming for example the husband of the paternal aunt (yaayo) into a maternal uncle (akiu/kakv), not in creating specific categories of affinity by splitting certain kin categories. In any case, it leads to a reduction or to a rearrangement of pre-existing categories, rather than their expansion. Finally, this process also results in the disappearance of skewing, once again contradicting the evolutionary perspective which systematically considers skewing as the end point or at least a late development of several chains of logically conceivable transformations. Indeed, proponents of the evolutionary approach most often imagine processes which, starting from Dravidian or Kariera systems, lead to Crow-Omaha systems. Here, on the contrary, we observe skewed systems transforming into an unskewed system. Eastern Tani linguistic and ethnographic data thus undercut the idea that Dravidian-like terminologies based on bilateral cross-cousin marriage are historically primordial. They also cast doubt on the supposed irreversibility of kinship transformations on which universal theories of kinship evolution are based.
I would like to express my thanks to all my Adi informants, without whose kind collaboration this work could never have been undertaken: Ratan Apang, Aido Ering Bamin, Leksar Modi, Dong Pertin, Jur Pertin, Une Pertin, Maksam Tayeng, Oken Tayeng. Special

Adi (Padam) Kin Terms
Padam Adi kin terms are listed by generation and kin type.

4.
A distinction should be drawn here between Adi as an ethnic label and Adi as a language, the extent of which is yet to be defined. Though several Adi speech varieties form a chain of mutually intelligible dialects, Tangam, for example, is not mutually intelligible with any other Adi variety (Post, pers. comm.), and Milang belongs to another branch called Siangic Blench 2011, Post andModi 2011).

5.
In the literature these subdivision levels are commonly labelled 'phratries', 'clans' and 'subclans', though this is not entirely accurate from an anthropological perspective. While both clan and lineage are types of unilineal descent groups defined in relation to a founding ancestor, by convention the main criterion for distinguishing them is that only the members of one lineage can trace their descent back to the founding ancestor, whereas in the case of a clan the links back to the founder are not specified or not remembered. As the patronymic linkage system enables individuals to recall the names of all ancestors, neither 'clan' nor 'subclan' adequately describes the basic sections of this society, and for sake of consistency we should instead consider them as 'major' and 'minor' lineage subdivisions. Thus, the largest Adi sections often labelled 'subtribes' in anthropological literature can be better understood as selfproclaimed descent groups resulting from the segmentation process rather than territorial entities or dialect clusters. For a discussion of the relevance of some common anthropological labels applied to Adi subdivisions, see Behera 2010.

6.
It is said that the Tayeng opin ancestor branched off at a higher level. Whatever the case, all Padam claim to be descendants of Bomi, the great-grandson of Pedong Nane. Pertin (2015a: 58) notes that Bomi serves as an alternative autonym for Padam in migration stories.

8.
Two more recent works can be mentioned here insofar as they deal with Tani languages that have been influenced by either Padam or Minyong: Modi's (2017) description of Milang includes a section on kinship terms and a kin terminology (pp44-54). The Milang language is only remotely related to Adi, but Milang are ethnically and culturally Adi. Conversely, while Galo are now ethnically distinct from Adi, their language is closer to Adi than Milang is, and their kinship terminology presents numerous similarities with the one discussed here. On Galo (Lare dialect) core kin terms, see Post 2007: 212-216.

9.
At least this is the case in referential usage. According to M Post, the vocative form in Lower Adi, has a long vowel aa-, not found in the referential form. For example /ŋok abu/ 'my father' vs /aabu ə!/ 'O Father!' (pers comm).

18.
Maam is another term for grandmother in Minyong, used for both address and reference.

19.
Tani languages have been divided by Sun (1993) into two major branches on phonological and lexical grounds. Western Tani includes languages such as Nyishi, Apatani, Hill Miri and Tagin, while eastern Tani includes Mising and most varieties of Adi.

23.
Padam rules appear to be slightly different: according to our informants, marriage used to be tolerated within the same opin between descendants of two brothers provided that there was a four-generation gap. In the early 2000s, however, this rule was abolished because of pressure from representatives of various lineage segments and of an overall desire among Padam to conform to the matrimonial rules of other Adi sections. A comparable situation has been reported for Milang (Modi 2017: 42).
24. Modi (2017: 46) reports similar avoidance of the reduplication of previous alliances for five generations among the Milang. Needham and Elkin (1973: 166): 'a man is not simply prohibited to marry certain categories or types of woman, but he is positively constrained to take a wife from a specific terminological class of persons'.

26.
The author, in opposition to Lévi-Strauss and Needham, argues that the basic function of an asymmetric alliance system is not to establish a generalised exchange but to prohibit restricted exchange (1966: 78).
29. Similar ordinal kin terminologies are also found in languages transitional between eastern and western branches, such as Bokar and Tangam Post and Burling (2017: 214): 'On the other hand, people belonging to different "tribes" can sometimes converse easily (for example, Padam and Mising, which are perhaps no more strongly differentiated than are New York and Midwestern US English)'. Lounsbury (1964a) showed that Dravidian terminologies have the genealogical calculus of cross/parallel distinctions for distant kin, contrary to that of Iroquois (Seneca) terminologies. He was followed by Scheffler (1971) who, besides identifying several more Dravidian systems in Melanesia, established that these were independent of descent mode and did not necessarily take place in the context of a dual organisation. To Scheffler and Lounsbury, the main difference with the Iroquois type lies in the classification of parents' cross cousins at G +1 and cross cousins' children at G -1 .

33.
Despite their androcentric definition, these two categories apply equally to men and women.

34.
Although the spouse is loosely defined in this system, it does not preclude the possibility of preferential, symmetric arrangements between two descent lines or decent groups. Bhandari (1992) conducted in-depth fieldwork among the Pagro-Mising of Majuli island and reported that, though marriage is theoretically possible with any segmentary unit of the opposite moiety, dyadic sets of affines tend to develop and perpetuate close relations of intermarriage. Thus, in the village of Kumarbari, more than 80% of unions were found to have been contracted between the Gajera section of the Doley moiety and the Gupit section of the Pegu moiety. This occurred despite a low endogamy rate, as 133 out of a total of 194 marriages involved partners from one of the 25 surrounding villages. As the author states, '...most of the negotiated marriages take place between those erang [families] which have previous alliance relationship' (1992: 195).

ABSTRACTS
This paper describes the kinship system of the Adi of Arunachal Pradesh (Padam and Minyong subgroups), focusing on its kin terminology. This system corresponds to the Omaha model defined by Lévi-Strauss, marked by generational skewing and only proscriptive matrimonial rules. It is also characterised by a remarkable profusion of terms of address indicating birth order. A comparison with the Mising system provides an opportunity to challenge some current neo-evolutionist paradigms regarding the transformations of kinship terminologies.